An Introduction to Hydrogen Bond Scalar Couplings

Transcription

1 An Introduction to Hydrogen Bond Scalar Couplings ANDREW J DINGLEY,,2 FLORENCE CORDIER, 3 STEPHAN GRZESIEK 3 Institute of Physical Biology, Heinrich-Heine-Universitat, Dusseldorf, Germany 2 Institute of Structural Biology, IBI-2, Forschungszentrum Julich, Julich, Germany 3 Department of Structural Biology, Biozentrum, University of Basel, CH-4056 Basel, Switzerland ABSTRACT: The hydrogen bond ( H-bond) has been recognized in science for more than 80 years as a concept to explain situations where a hydrogen atom is simultaneously binding to two other atoms Due to the moderate energies necessary for their formation and rupture, hydrogen bonds play a fundamental role in many chemical reactions and most, if not all, interactions involving biological macromolecules For both proteins and nucleic acids, H-bonds are the essential element in the formation of secondary structures and often they also participate in the stabilization of tertiary structures Many properties of H-bonds have been studied by a large variety of experimental methods, including NMR spectroscopy Recently, electron-mediated scalar couplings have been observed which connect magnetic nuclei on both sides of the hydrogen bridge In contrast to earlier NMR observables, these couplings can be used to see all partners of the hydrogen bond, the donor, the proton, and the acceptor in a single COSY experiment In addition, the size of the coupling constant can be related to hydrogen bond distances and angles This article should serve as an introduction to these findings and illustrate their use by various examples 200 John Wiley & Sons, Inc Concepts Magn Reson 3: 0327, 200 KEY WORDS: DNA; hydrogen bond; nucleic acid; protein; RNA; scalar coupling INTRODUCTION Since the first definitive recognition of hydrogen Ž bonding at the beginning of the last century, Received 6 July 2000; revised 26 October 2000; accepted 26 October 2000 Correspondence to: Stephan Grzesiek; Stephan Grzesiek@unibasch Contract grant sponsor: DFG Contract grant number: GR683- Contract grant sponsor: SNF Contract grant number: Ž Ž Concepts in Magnetic Resonance, Vol John Wiley & Sons, Inc 2, the literature on hydrogen bonds Ž H-bonds has become simply too large to be cited in any adequate way Jeffrey and Saenger estimated in the early 990s that a paper on H-bonds would be published every half hour Ž 3 Besides a number of monographs Ž 3 and review articles Ž 28, the chapter on H-bonds in Linus Pauling s book The Nature of the Chemical Bond Ž 9 may serve as an excellent introduction to this very large scientific field A simple succinct definition to hydrogen bonding is the weak attraction between a hydrogen atom attached to an electronegative donor atom D and an electronegative acceptor atom A This 03

2 04 DINGLEY, CORDIER, AND GRZESIEK force is largely attributed to the electrostatic attraction between a partial positive charge located at the position of the hydrogen atom in the D H group and a partial negative charge on the acceptor atom A The predominant types of H-bonds in biomacromolecules are O H O, O H N, N H O, and N H N, although weaker H-bonds involving C H groups as donors are also recognized Ž 2027 For these H-bonds, the site of attraction is commonly identified as an electronic lone pair on the oxygen or nitrogen acceptor atom However, the electrons of aromatic systems can also act as acceptors, and H-bonds involving sulfur groups or metallic cofactors are also known The overwhelming importance of H-bonds in biology and chemistry stems from the very moderate energies needed for their formation and rupture Ž 3 This makes it possible that H-bonds play an essential role in many common enzymatic Ž3, 9, 28, 29 and chemical reactions occurring in many common solvents, such as water, at ambient temperatures The formation of the H-bonded secondary structures of proteins Ž 30 and nucleic acids Ž 3 is probably the most important example for such a reaction In contrast to covalent bonds which are of the order of kj mol in bond energy, the energies of the usual weak H-bonds are approximately an order of magnitude smaller Žie, 30 kj mol Only the so-called very strong H-bonds have bond energies which are similar in size to their covalent counterparts An example of such a very strong H-bond is the bifluoride ion Ž HF with an energy of 63 4 kj mol Ž 32 2 In proteins and nucleic acids, the bond energies of individual H-bonds are often determined through deletion or site-directed mutagenesis Ž 3339, with estimated values ranging between 2 and 20 kj mol However, as the substitution or deletion of entire residues will also affect other properties of the biomacromolecule, the values reported from such studies are usually considered as upper limits Clear-cut correlations between H-bond energy and the H-bond geometry are currently missing On the contrary, it is a very important property of the weak hydrogen bond that the possible bond distances and bond angles cover a rather wide range Ž 3, 40 Apparently, the energy minimum is relatively shallow with respect to a variation of bond distances and angles Therefore a variety of different H-bond geometries can be realized with relative ease This property enables the H-bonds to act as a very adaptive glue and contrasts with covalent bonds that display distinct preferences with respect to bond distances and angles Ž 3 In the geometric characterization of a hydrogen bond, main parameters are the hydrogen bond distances H A and D A, as well as several bond and torsion angles such as the hydrogen bond angle D H A Generally, an H-bond is assumed to be present when the distance RDA is less than or equal to the sum of the van der Waals radii of the two electronegative atoms For example, in the structure of ice, the R distance of 275 A Ž 4, 42 O O is much smaller than the sum of the van der Waals radii of 304 A of the two oxygen atoms Ž 43 The presence of H-bonds in biomacromolecular structures is usually inferred from the spatial arrangement of the donor and acceptor groups once the structure of a biomacromolecule has been solved by either X-ray crystallography, NMR spectroscopy, or even neutron diffraction Due to the weak scattering density of the hydrogen atom, it is particularly difficult to obtain precise information from X-ray diffraction on its position within the H-bond Only for the highest resolution X-ray structures Žie, A resolution is it possible to ascribe individual spatial positions to the hydrogen atoms which are independent of the use of standard covalent geometries Recent technical innovations for neutron diffractometry Ž 4446 bear some promise to provide more precise information on the position of either hydrogen or deuterium nuclei within the H-bonds of biomacromolecules Ž 4750 High resolution NMR has contributed significantly to the understanding of H-bonds For example, a large number of different NMR observables Ž 6, such as chemical shifts, provide indirect evidence for hydrogen bonds However, in recent years direct evidence has been found by the presence of scalar couplings between magnetically active nuclei on both sides of the hydrogen bridge Such couplings have been observed both in biomacromolecules Ž 576 and in smaller chemical compounds Ž 7780 The existence of such scalar couplings makes it possible to correlate the frequencies of nuclei on both sides of the hydrogen bridge by high resolution NMR experiments Thus, in favorable cases, complete H-bond networks in proteins and nucleic acids can be established by COSY-type experiments Besides this practical aspect, the size of the couplings has been shown to be influenced by the geometric arrangement of the nuclei involved in the H- bonds This opens the possibility to derive geo-

3 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS 05 metrical information about the H-bond from the analysis of the coupling size Initially, it was surprising to observe such couplings across hydrogen bonds since we usually associate scalar couplings with the presence of purely covalent bonds However, it is now well established that the H-bond scalar couplings follow the same electron-mediated polarization mechanism as their covalent counterparts Ž53, 59, 78, 8, 82 Therefore the same NMR experimental concepts apply for their detection and quantification This introductory review aims to Ž briefly summarize the various NMR observables which provide indirect information for individual H-bonds; Ž 2 present a basic theoretical description of scalar couplings including the H-bond couplings; Ž 3 illustrate how we originally detected the cross hydrogen bond coupling in nucleic acids by a systematic investigation of magnetization losses; Ž 4 discuss the basic experimental procedures for measuring various H-bond couplings in biomacromolecules The article concludes with a summary of possible applications for the H-bond coupling effect INDIRECT NMR OBSERVATIONS OF THE HYDROGEN BOND Reduced Hydrogen Exchange Rates Many polar hydrogen atoms, in particular the amide, imino, and hydroxyl group hydrogens in biomacromolecules continuously exchange with solvent hydrogens This continous exchange is slowed down significantly when the hydrogen participates in an H-bond, even when this bond is at the surface of a biomacromolecule Ž 8386 Therefore, hydrogen exchange from an H-bond is generally assumed to require H-bond opening The hydrogen exchange is then modelled as a two-step mechanism where the first step is the equilibrium reaction between the closed and open H-bond, whereas the second step is the exchange reaction from the open state to the solvent Ž84, 85 The latter step usually requires the catalysis by acids or bases Thus, the hydrogen exchange rates depend on a number of factors such as ph, the pk of the donor group and of the base or acid catalysistž s, the catalyst concentration, temperature, solvent accessibility, as well as the H-bond opening rate itself Ž 84, 85, 87 Proton NMR spectroscopy has quantified many of the underlying reactions at specific H-bonding sites both in nucleic acids Ž 8789 and polypep- tides Ž 9092 Due to the rather low pk of the imino group, around 9, the exchange rates of H-bonded imino-protons in nucleic acids can be increased by buffer catalysts with similar pk values, such as ammonia Ž 87 Thisincreaseinex- change can be used to extrapolate to the situation where the base pair opening is the limiting step in hydrogen exchange Intrinsic opening times for imino H-bonds in WatsonCrick base pairs determined by this method are on the order of milliseconds Ž 8789 In contrast, catalysis by buffers is ineffective for amide protons in proteins due to the high pk Ž ca 85 of the backbone amide group Ž 86 Therefore, at moderate ph values, the hydrogen exchange rates in proteins are usually limited by the slow exchange from the open state of the H-bond to the solvent, and intrinsic opening rates of protein backbone H-bonds cannot be derived Ž 87, 93 Exchange times for amide hydrogens in short peptides or at the surface of proteins are on the timescale of milliseconds to seconds Ž 9, 92 In contrast, H-bonded amide hydrogens located in secondary structure elements have been reported to have exchange times 6 ranging between 0 to 0 min Ž 90 Clearly, the very slow exchange rates of amide hydrogens observed in the interior of proteins may not only be due to H-bonding, but also may be caused by limited solvent accessibility Ž 90 Chemical Shift H-bond formation usually results in chemical shift changes for all the nuclei involved in the H-bond Ž 6 As the chemical shift is intrinsically related to the local electronic environment, these chemical shift perturbations indicate a redistribution of the electron density upon H-bond formation For H-bonding to an electronegative acceptor atom such as oxygen or nitrogen, there is always a change in the isotropic chemical shift of the H- bonded hydrogen nucleus to higher frequencies Ž downfield shift This downfield shift is a result of a number of not yet fully understood and partially competing factors, including a decrease in the electron density around the hydrogen nucleus and deshielding effects from the electronic currents of the acceptor atom A large number of examples can be given for such proton downfield shifts on H-bond formation with oxygen or nitrogen acceptors; eg, downfield shifts of the amide proton in proteins have long been recognized to be correlated to shorter H-bond lengths Ž 94, 98 Like-

4 06 DINGLEY, CORDIER, AND GRZESIEK wise, the formation of H-bonds in nucleic acid base pairs results in downfield shifts for the imino and amino protons Ž 99, 00 In certain systems, proton downfield shifts upon H-bond formation have been observed to be as large as 20 ppm Ž 4 For example, the extreme chemical shift Ž 6 ppm of histidine imidazole protons is used as an indicator for the presence of low-barrier hydrogen bonds in catalytic processes Ž 0 Conversely, H-bonding to electrons in aromatic rings usually results in a change of the isotropic H chemical shift to lower frequency Ž upfield shift This upfield shift is caused by the aromatic ring current effects outweighing the deshielding effects from the formation of the H-bond itself Ž 6 General trends are also observed for the chemical shifts of donor and acceptor nuclei Invariably, in the formation of an H-bond involving an electronegative acceptor, the donor nucleus experiences a deshielding effect, whereas the acceptor nucleus chemical shift moves to a lower frequency due to an overall increase in electronic shielding Examples for these effects are donor and acceptor N chemical shift changes on H- bond formation in nucleotides Ž 02, nucleic acid oligomers Ž 0305, and histidine side chains Ž 06, 07 Similarly, ab initio calculations indicate that backbone H-bond formation in proteins leads to a deshielding of the amide N nucleus with a corresponding shift to higher frequencies by approximately 3 ppm Ž 08 Primary and Secondary Isotope Shifts from Substitutions of H by 2 H and 3 H The substitution of one isotope for another Žeg, 2 H for H is known to affect the isotropic chemical shift of the substituted nucleus Žprimary iso- 3 tope shift and of magnetic nuclei, such as C and N, one or more covalent bonds away from the substituted nucleus Ž secondary isotope shifts Whereas the primary isotope shift for the substitution of hydrogen by either deuterium or tritium is usually small in weak H-bonds Ž 09, significant effects are observed for strong H-bonds Ž0, Such primary isotope shifts have been used to characterize a number of H-bond systems and provide information on the shape and symmetry of potential energy wells for the H-bonded hydrogen nucleus Ž 4, 03 Secondary isotope effects have also been used to characterize the presence and properties of the weak H-bonds in biomacromolecules Ž 09, 4 Equilibrium 2 H / H Isotope Fractionation Factor The 2 H H fractionation factor,, is a quantity which is defined as the relative occupancy of deuterons versus protons at an exchangeable hydrogen position of a solute, as compared with the corresponding relative concentration of labile hydrogens within a solvent of mixed proton and deuteron content, eg, HO 2 H O: 2 2 Ž Ž 2 D HD H 2 H H solute solvent Thus, a value of less than one corresponds to an enrichment of protium relative to solvent content, whereas a value greater than unity reflects a preference for deuterium relative to solvent content Experimental results suggest a correlation between the fractionation factor for an H-bonded hydrogen and the strength of the H-bond, where a lower value corresponds to a stronger H- bond Although is not an NMR parameter, NMR usually provides the most suitable method for its determination in small chemical molecules Ž 4 and biomacromolecules values for proteins Ž, 6 vary over a wide range Žeg, 03 in staphylococcal nuclease Ž 7, and have provided insights into the presence and strength of H-bonds in secondary structures Ž 720 and catalytic processes Ž 226 Covalent Scalar Couplings As scalar couplings are mediated by the electrons within chemical bonds Ž see below, the redistribution of electron densities upon H-bond formation also gives rise to observable changes in scalar couplings between nuclei associated with the H- bond Changes of one-bond couplings between hydrogen bonded proton and donor nuclei have been observed in a number of small chemical compounds in various organic solvents, where changes in J and NH JCH couplings were measured upon H-bond formation Ž 2730 In particular, for an adenosineuridine base pair analogue in chloroform, base pair formation resulted in an increase of the imino group JNH from 93 to 875 Hz Ž 3 A similar observation has been made for the JNH couplings of imino groups in different nucleic acid base pairs of a DNA triplex Ž 59 This study also suggested that an increase in the J coupling was associated NH

5 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS 07 with a decrease in the donoracceptor distance and thus with an increase in the strength of the H-bond In proteins, H-bonding to the carbonyl oxygen atom Ž O C leads to a strengthening in the sequential one-bond JC N coupling constant within the peptide bond, whereas H-bonding of the amide hydrogen results in a weakening of the J coupling Ž 32, 33 C N This observation was used to correlate the measured JC N coupling constants with the various types of secondary structure elements in a protein 2 H Quadrupolar Coupling Constant Since the quadrupole coupling constant Ž QCC is proportional to the size of the electric field gradient at the position of the nucleus, the QCC of a deuteron within an H-bond is very sensitive to the asymmetry of the charge distribution within the H-bond A shorter H-bond usually corresponds to an increased symmetry and thus results in weaker electric field gradients and smaller 2 H QCC values In proteins, 2 H QCC values have been derived from the relaxation times of the deuterons within H-bonds Ž 34, 35 In particular for protein backbone N H O C H-bonds, a correlation to the H-bond length RH O and the N H Ž O H-bond angle 20 could be established as QCCŽ 2 H khz cos R Ž 35 H O HYDROGEN BOND SCALAR COUPLINGS Basic Principles of Scalar Coupling Before discussing the observation in biomacromolecules of scalar couplings between nuclei separated by a hydrogen bond, let us briefly discuss the characteristics of the principal interactions between two magnetic nuclei in any substance, ie, of the dipolar and the scalar couplings The dipolar interaction is transmitted through space by the magnetic dipolar fields surrounding the nuclei Although this dipolar interaction can be on the order of tens of kilohertz, in isotropic solutions and gases, the dipolar interaction averages exactly to zero due to the isotropic reorientation of the molecules In contrast to the dipolar interaction, the scalar interaction is transmitted via the electron cloud of the molecules Therefore, this interaction is only observed between nuclei that have electrons between them, ie, which are connected by some kind of chemical bond The scalar interaction is also called J-coupling, and is typically detected up to a distance of three to four bonds In solution and gas phases, the J-couplings give rise to the familiar resonance splittings that can be used to decipher the chemical structure of molecules The strength of the scalar interaction is measured by the size of the resonance splitting in hertz Ž Hz This splitting defines the scalar coupling constant, n J ij, in which, n designates the number of bonds separating the two nuclei i and j The value of the J-coupling constant can be either positive or negative Let us describe the mechanism of scalar interactions between two nuclei A and X in more detail If nucleus A is a spin-2 nucleus Žeg, H, then there are two possible states for the z component of its spin vector A Ž A, A, A x y z when A is placed into a static magnetic field B Ž 0, 0, B 0 0 which is directed along the z axis These states are frequently referred to as the and or spin-up and spin-down states Alternatively, the two states can be labeled by the quantum mechanical expectation values for A z Ž 2 and 2 which is usually called the magnetic quantum number m A The magnetic moment of the nuclear spin is proportional to the A spin vector and is given as Ž A, A, A x y z AA, where A is the gyromagnetic ratio The interaction of the magnetic moment with a magnetic field is called Zeeman interaction and the corresponding Hamiltonian is given by H Zeeman A B A B A 0 A 0 A B 2 A z 0 The fact that A z can have two different values Ž 2 and 2 leads to two different energy levels which are separated by an energy gap E AB 0 In nuclear magnetic resonance, we observe electromagnetic transitions between these states corresponding to electromagnetic radiation with the resonance Ž Larmor frequency A AB 0 For a single nuclear species A, we detect a single resonance line at position A Note that for a real chemical substance this picture is slightly incorrect: the nucleus A is surrounded by electrons which slightly shield the external magnetic field Therefore the resonance frequency A is shifted by a tiny relative amount which is called the chemical shift The description of the interaction with the magnetic field is identical for the second nucleus X which for the sake of simplicity is also assumed

6 08 DINGLEY, CORDIER, AND GRZESIEK to be a spin-2 nucleus In the absence of any interaction between the two nuclei, the total Hamilitonian of the two spin system is simply the sum of the two Hamiltonians: Zeeman HAX A A z B0 XX z B0 3 As the two spins are independent of each other, the energy levels for one spin do not depend on the state of the other spin and one simply observes two independent electromagnetic transitions with Larmor frequencies A and X The situation changes when there is an interaction between the two nuclei Then the energy levels of one nucleus do depend on the state of the second nucleus and the total Hamiltonian for such a system contains an additional term which describes the cross-talk between the two nuclei Scalar interactions are caused by a two-step mechanism First, the magnetic moment of one nucleus magnetically polarizes the molecular electron cloud Then, the induced magnetic field of the electron cloud interacts with the magnetic moment of the second nucleus As a consequence, the scalar interaction must be proportional to the magnetic moments of both nuclei A and X The most general Hamiltonian for such an interaction is of the form Ý H J A K X A K X AX, x, y, z Ý 2 A K X 4 A X, x, y, z where K is a 3x3 matrix with elements K The fact that K is not a simple constant, but a matrix, expresses the possibility that the nucleus electron nucleus polarization transfer might depend on the orientation of the molecule However, in the isotropic situation of liquids and gases the orientation dependent parts of K average to zero As a result, Eq 4 can be replaced by Ý H J 2 K A X AX A X x, y, z 2 K A X 2JA X 5 A X where the constant K represents the orientation independent part of K, ie, its trace, and the scalar coupling constant J is given as K2 Equation A X 5 has the practical consequence that, neglecting small additional isotopic effects, the replacement of one nucleus by an isotopic nucleus results in a scaling of the scalar interactions by the ratio of the respective gyromagnetic constants For example, the one- 3 2 bond scalar coupling in a C H bond is usually about a factor of 65 Ž H Ž 2 H smaller 3 than the corresponding coupling in a C H bond A further simplification for Eq 5 can be found when the difference in the Larmor frequencies for nucleus A and X is large compared to the scalar coupling constant J; ie, A X 2 J This limit is called weak coupling In this case, the transverse parts of the scalar coupling Hamiltonian which do not commute with the Zeeman Hamiltonian are neglected: J,weak HAX 2JA zxz 6 A pictorial explanation for this approximation can be given by the fact that the transverse x and y components of both spins rotate very quickly against each other in the external field when the difference in Larmor frequencies is large Therefore the transverse components of the scalar coupling quickly average to zero In any case, we see from the form of the scalar coupling in Eq 5 or 6 that the energy levels of one nucleus are influenced by the state of the second nucleus If we assume that the z components of the nuclear spins A and X are parallel Ž m m 2 or m m 2 X A X A, then according to the weak coupling Hamiltonian, the energy of the system will be increased by the amount J2 If both spins are antiparallel Ž m m 2 or m m 2 X A X A, the energy will be lowered by the same amount A slightly more complicated description needs to be applied for the strong coupling case in Eq 5 There is yet another way to look at this: if we assume that nucleus X is in a state where the z component of its spin is given by mx then we can express the total Hamiltonian for spin A Žweak coupling as H total A B 2JA m A A z 0 z x Ž B 2 Jm A 7 0 X A A z This means that the action of nucleus X on nucleus A looks like an additional magnetic field along the z axis of size 2 Jm XA at the position of A Apparently this field corresponds to the magnetic field that was induced in the electron cloud by the magnetic field produced by spin X The action of this induced field makes the nuclear spin A rotate faster or slower depending

7 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS 09 on the sign of the J-coupling and the direction of the z component of the spin X At thermal equilibrium in a solution containing many AX spin systems, the two states mx 2 and 2 of the spin X are almost equally populated Therefore, the resonance line for nucleus A is split into two lines of identical intensity which are separated by the angular frequency difference 2 J In the analogous situation where the frequency of the X nucleus is detected, an identical splitting of the spectral line is observed From this description of the scalar coupling effect it is apparent that it leads to a correlation in the movement of the two interacting spins The most fruitful use of this correlation is made in the transmission of nuclear frequency information along chemical bonds in COSY Ž correlation spectroscopy experiments Electronic Interactions and the Transfer of Magnetization across the Hydrogen Bond So far, we have considered the size of the scalar coupling constant as a phenomenological parameter A theoretical determination of the value of J involves the quantum-mechanical description of the electronic orbitals in a substance Originally such a formulation of the spinspin coupling was put forward by Ramsey Ž 36 who identified three basic mechanisms which give rise to scalar interactions, all mediated by valence electrons Ž The nuclear magnetic moments interact via the magnetic field produced by the orbital motion of the surrounding electrons Ž electron orbital term Ž 2 The nuclear magnetic moments interact via the magnetic field produced by the spins of the surrounding electrons Ž electron spin term Ž 3 Due to the finite size of the nucleus, the nuclear magnetic moments interact directly with electrons that have a non-zero-probability density at the position of the nucleus Ž Fermi contact term In hydrogen-like atomic orbitals, only s-electrons fulfill this condition The scalar coupling constant is given by the sum of all three terms However, for couplings involving the hydrogen atom, such as the H-bond scalar couplings, the Fermi contact term is dominant and the other two terms can usually be neglected Ž 59, 8 With the current increase of available computational power and the advent of reliable numerical descriptions for the electronic orbitals of non-trivial chemical compounds such as GAUSSIAN Ž 37, it has become feasible to carry out accurate calculations of coupling constants based on these first principles of electrodynamics for systems containing several tens of atoms and a few hundred wave functions For example, very good agreement between experimentally determined H-bond scalar coupling constants and theoretical simulations were found for hydrogen bonds in nucleic acid base pairs and in the backbone of proteins Ž59, 8 In these and other systems, such calculations have become a very powerful analytical tool that complements the experimental measurement of scalar couplings Using this theoretical framework, we might visualize the transmission of nuclear polarization across the hydrogen bond between a donor nucleus D and an acceptor nucleus A in the following way Ž Fig : the nuclear magnetic moment of the nucleus D gives rise to a magnetic field which orients an electron in its vicinity The polarization of this electron is transmitted via electronelectron interactions, ie, electric forces and the Pauli exclusion principle, to a second electron in the vicinity of nucleus A The magnetic field produced by this second electron polarizes the magnetic nucleus of the acceptor A In this situation, a two-bond scalar coupling Ž h2 J DA is observed between both nuclei ŽNote that for n-bond scalar couplings between two nuclei i and j across H- bonds, Pervushin et al Ž 54 introduced the symbol hn Jij where the superscript h should indicate that one of the n bonds is actually an H-bond An important consequence follows from the observation of such an H-bond coupling: as the Fermi contact term is dominant in the transfer mechanism one can conclude that electrons with non-zero density at the location of the nuclei on both sides of the H-bridge must be involved in the transmission This means that the movement Figure Schematic representation of the scalar coupling mechanism across H-bonds A two-bond coupling of type h2 JDA connects the nucleus of the donor atom D with the nucleus of the acceptor atom A The electrons Ž e transmit the magnetic polarization between the two nuclei

8 0 DINGLEY, CORDIER, AND GRZESIEK Ž of electrons with s-electron character on both sides of the H-bond must be correlated Hydrogen Bond Scalar Couplings in Nucleic Acids A Quantitatie Trace of Unexpected Magnetization Losses Although a large body of observations existed about couplings between nuclei that were not connected by usual covalent bonds, eg, the so-called through space couplings Ž 38 or even some observations of cross H-bond couplings Ž5, 52, until the late 990s the common idea about a scalar coupling in a biological macromolecule was associated with a covalent bond This is easily explained because the common H-bond couplings in 3 C and N isotope labeled proteins are relatively small, ie, Hz, and the observation of the larger H-bond couplings in nucleic acids Ž up to Hz depended on the availability of N labeled nucleic acids Efficient labeling strategies for the latter and the concomitant application of heteronuclear experiments became widespread only after such techniques had been developed for proteins Ž 39, 40 In fact, our observation of such couplings in nucleic acids resulted from the attempt to adapt heteronuclear protein experiments for the use in nucleic acids Unexpected losses of magnetization in such experiments revealed the presence of substantial couplings across the H-bonds Since the method to detect such losses is very generally applicable for the debugging of new pulse sequences, some of this evidence is presented here The original experiment was designed as an HNCO sequence where the imino H and N resonances of uridine and guanosine Ž Fig 2 should be correlated via a one-bond coupling to the carbonyl 3 C nuclei next to the imino N nucleus, ie, 3 C4 or 3 C2 in uridine and 3 C6 in guanosine Experience with proteins shows that such an experiment is usually very sensitive, because only relatively large scalar couplings are used in the magnetization transfer pathways Typical values of the imino JNH couplings are around 90 Hz, whereas typical JNC couplings range from 0 to Hz This means that the total time for an INEPT transfer ŽInsensitive Nuclei Enhanced by Polarization Transfer Ž 4 between the imino H and the N nuclei should be set to values of about 55 ms Ž2 J NH and to values of about 33 ms Ž2 J NC for the transfer between the N nucleus and the adjacent Figure 2 Hydrogen bond configuration and nomenclature for Ž A WatsonCrick UA, Ž B WatsonCrick GC, and Ž C UG wobble base pairs Approximate isotropic chemical shift values Ž in ppm for N nuclei are indicated Dark shaded nuclei are involved in h2 J NN scalar couplings 3 C-carbonyl nucleus Neglecting numerical factors and the chemical shift evolution periods, the magnetization path in such a HNCO experiment is given by 55 ms 33 ms y z y z z y H H N H N C 33 ms 55 ms y z x H N H 8 During this sequence, magnetization evolves for ms as transverse proton terms and 66 ms as transverse nitrogen terms Therefore, relaxation losses can be described by a factor expž mst Ž H * expž66 mst Ž N, where T Ž H and T Ž N are the relaxation times for the trans- 2

9 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS verse H and N magnetization terms in Eq 8 These relaxation times can be determined in independent experiments Assuming that the efficiency of the scalar magnetization transfer is close to and that the losses due to pulse imperfections are small, the expected signal intensity of the whole experiment can be estimated from the signal intensity in a simple one-dimensional proton spectrum multiplied by the relaxation loss factor It turned out that the signal intensity of the HNCO experiment was much lower than expected In order to determine the source of such a magnetization loss, it is often a practical way to break down a complete pulse sequence into single parts that still lead to observable magnetization One such part is depicted in Fig 3Ž A It is an H N heteronuclear single-quantum correlation Ž HSQC pulse sequence where the nitrogen evolution period has been replaced by the sequence 80 Ž N y It is recognized that the sequence would yield identical relaxation losses as the HNCO sequence if the delay is set to a value of 33 ms A) H 90 x Preparation Evolution Refocusing Acquisition δ 80 y δ 90 y 90 x δ 80 y δ N a b c d 80 y 90 x 80 y 90 x 80 y Δ Δ Decouple U-A G-C G-U Non-sel 80 B) 2 ppm 2 ppm Δ Δ Δ Δ Δ = ms Δ = 30 ms x 025 C) 2 ppm Δ Sel 80 Δ Δ = 30 ms D) 2 ppm Δ Δ Δ Δ ppm Δ = 30 ms ppm Imino H Figure 3 Original evidence of the h2 J H-bond coupling effect in WatsonCrick base NN pairs Spectra were recorded at 25 C on a 600 MHz Bruker DMX spectrometer using a 3 Ž 6 mm sample of uniformly N C-labeled PSTVd T RNA domain 69 nucleotides, 00 mm NaCl, 0 mm sodium phosphate, 95% H O5% D O, ph N spin-echo modifications of a H N HSQC pulse sequence and resulting imino proton spectra Žsee text Pulse flip angles and phases are indicated above each pulse Unless noted otherwise, Ž Ž carrier positions are H O H and 2 ppm N 2 RF field strength of all non-selective H and N pulses were 29 and 63 khz, respectively N-WALTZ decoupling ŽNB khz was applied during data acquisition For clarity C decoupling pulses are not shown The delay was set to 225 ms A total of 32 scans per spectrum was collected The selective N 80 pulses applied during the 2 period in Ž C and Ž D have a G3 Ž 3 amplitude profile corresponding to an excitation bandwidth of 20 ppm

10 2 DINGLEY, CORDIER, AND GRZESIEK The sequence in Fig 3 is easily understood in terms of the product operator formalism Ž 4244 Neglecting relaxation losses, the initial INEPT period transfers longitudinal proton magnetization of type H z present at point a into antiphase nitrogen magnetization of type 2H z N y at point b If this type of magnetization is also present at point c, the reverse INEPT step will transfer this operator product into observable proton magnetization of the type H x at point d Other types of magnetization at point c should not lead to observable magnetization at the end of the pulse sequence To a good approximation, this is true for the depicted pulse sequence; if a further suppression of the unwanted magnetization pathways is necessary, additional phase cycling steps and pulse field gradients can be applied The interval between point b and c can be used to probe the fate of the operator product 2HzNy under different conditions For example, inserting the 80 Ž N y sequence leads to a nitrogen spin-echo experiment with proton excitation and detection The N 80 y pulse in the middle of the nitrogen spin echo serves to refocus the N chemical shift evolution, presented by a Hamiltonian of the form N Ž N z assuming a convention where N 80N N z y N N z z y z y 2H N 2H N 9 The N 80 pulse also decouples or refo- cuses any kind of heteronuclear scalar coupling to nuclei such as 3 C or H, presented for example by a Hamiltonian of the type 2 J NX: Ž y NX z z 2 J N X 80N NX z z y 2 JNXNzX z z y 2HzNy 2H N 0 In both cases, the 80 Ž N y pulse reverses the direction of the nitrogen x-magnetization in the middle of the echo delay, and as a consequence all the magnetization refocuses at point c Therefore neither chemical shift nor heteronuclear couplings should modulate the intensity of the proton signal which is detected at point d Figure 3 shows the results of this experiment for the base-paired imino proton resonances of the 69-nucleotide potato spindle tuber viroid Ž PSTVd T domain The experiment for a total spin echo 2 delay of 2 ms is illustrated in Fig 3Ž A Clearly observable are resonances downfield of about 2 ppm which correspond to the imino protons for the WatsonCrick uridineadenosine Ž UA and guanosinecytidine Ž GC base pairs Upfield of 2 ppm, we see imino proton resonances which belong to non-watsoncrick base pairs, such as the GU base pairs in Fig 2Ž C When the delay 2 is increased to 60 ms ŽFig 3Ž B, the imino resonances from WatsonCrick base pairs are reduced in intensity to a point where they are no longer detected In contrast, the intensity of the non-watsoncrick GU imino resonances is attenuated to approximately 25% such that these protons are still clearly visible During the interval 2, a considerable signal loss is expected from the relaxation of the 2H z N y magnetization term An independent T experi- ment using N spin-lock fields with a radio frequency strength of approximately 2 khz revealed that the N transverse relaxation time for most imino groups is approximately 45 ms at the 298 K temperature used for the experiments in Fig 3 Neglecting additional losses due to Hlongitudinal relaxation, this N transverse relaxation time leads to an expected decay of expž for the 2HzNy magnetization term during the 60 ms interval Such a loss is close to the reduction in signal which is observed for the non-watson Crick imino resonances However, it cannot explain the much stronger signal loss for the WatsonCrick imino groups, considering that both transverse and longitudinal relaxation rates seemed to be rather uniform for all imino groups in the respective relaxation experiments Apparently, the additional reduction in signal for the WatsonCrick base pairs is caused by an another mechanism which is neither relaxation nor a heteronuclear scalar coupling Such a mechanism could be a homonuclear scalar coupling Ž J NN between the imino N nucleus and another N nucleus If the radio fre- quency strength of the N 80 pulse in Fig 3Ž B is comparable to or larger than the frequency offset of the N coupling partner, then this homonuclear coupling would not be refocused In the case of the experiments of Figs 3Ž A and 3Ž B, an RF field strengthdefined as the inverse of the 360 pulse lengthof 63 khz was used Therefore, such a N 80 pulse would effectively recouple homonuclear N N couplings to partners within a range of approximately 00 ppm

11 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS 3 Ž B 4 T 0 around the carrier frequency Assuming that the 80 N pulse would affect both nitrogen nuclei, the evolution under a homonuclear coupling during the interval 2 is described by 2H z N y 2 J N N 80N,80N NN z z y y 2 J N N NN z z 2H NcosŽ 2 J z y NN 4H NNsinŽ 2 J z x z NN where N represents the Nnucleusofthecoupling partner Only the first of the two terms at the end of the spin-echo period is transferred into observable magnetization by the reverse INEPT step between time points c and d Apparently, a signal resulting from this term will be modulated by the factor cosž 2 J NN The spin-echo time of 60 ms in Fig 3Ž B results in a near complete loss of magnetization for the WatsonCrick imino groups If a homonuclear coupling is responsible for this loss, then it would be due to the first zero crossing of the cosž 2 J NN factor, because no other zero crossings were observed at shorter delay times Therefore, such a coupling must have an approximate strength of Ž 260 ms 8 Hz To test this hypothesis further, the non-selec- tive N 80 pulse ŽFig 3Ž B was replaced by a selective N 80 pulse ŽFig 3Ž C with an approximate excitation range of 2 20 ppm that covers the frequencies of the guanosine and uri- dine imino N resonances ŽFig 2Ž A,B This selective N 80 pulse not only refocuses any heteronuclear scalar couplings, but also refocuses homonuclear scalar couplings to N nuclei which are outside of the excitation window of the pulse Apparently, the application of the selective pulse for a spin-echo delay of 60 ms in Fig 3Ž C results in the recovery of the magnetization for the WatsonCrick resonances such that their intensities are equal to the intensities of the non-watson Crick GU base pairs This finding unambiguously establishes that the loss of magnetization observed with the non-selective N 80 pulse in Fig 3Ž B is caused by a homonuclear N N scalar coupling To further validate this result and to determine the frequencies of the Ncouplingpartners, the non-selective N 80 pulse was placed back into the center of the spin echo delay However, two further selective decoupling pulses Ž20 20 ppm were applied in the middle of the two delays ŽFig 3Ž D This has the effect that homonuclear couplings to nuclei in this frequency range would be refocused at the end of the first and second delay A similar recovery of magnetization is observed ŽFig 3Ž D as for the single selective pulse centered at 2 ppm ŽFig 3Ž C Therefore, it is obvious that the coupling partners must have resonance frequencies in the range of ppm A Direct Obseration of the N Splitting At this point, one may ask whether it is not possible to observe the couplings directly as a splitting of the N resonances Of course, it is, but this is not as entertaining as doing the Fourier transform of cosž 2 J NN in one s head The splitting can be detected using a 2D H N HSQC experiment where the N evolution period is made sufficiently long, ie, significantly longer than 60 ms, such that the JNN coupling can be resolved Such an HSQC sequence is similar to the pulse scheme illustrated in Fig 3 with the spin-echo delay being replaced by the familiar t-evolution period with a H 80 pulse at its center This proton pulse refocuses the J -coupling during the evolution NH period Similarly, couplings to the 3 C nuclei of the 3 C N-labelled RNA are removed by appropriate 3 C 80 pulses Since the following arguments do not depend on any effects of couplings to 3 C, these pulses are not shown in the figures and the description of such effects has been left out for clarity The product operator description during the t -evolution is then very similar to Eq, except that the chemical shift precession is not refocused Ž t N N z, and thus N frequency labeling is achieved: 2H z N y t N 2 J t N N N z NN z z 2H NcosŽ J t cosž t z y NN N Ž Ž 4H NNsin J t cos t z x z NN N Ž Ž 2H Ncos J t sin t z x NN N Ž Ž 4H NNsin J t sin t z y z NN N 2

12 4 DINGLEY, CORDIER, AND GRZESIEK Only the first term in Eq 2 converges to observable magnetization at the end of the pulse sequence We note that for quadrature detection in the t-domain, usually a sine-modulated term for frequency labeling needs to be recorded This is easily done by recording a second FID where the phase of the first N 90 pulse is changed from x to y Ž 45 Complex Fourier transform of the t-dimension in such an HSQC experiment gives rise to resonances which are split due to the active JNN coupling Figure 4 depicts a part of such a N highly-resolved H NHSQCexperiment containing the uridine imino resonances for the UA base pairs in the T domain of PSTVd As anticipated, these imino resonances show an in-phase splitting of approximately 8 Hz in the N dimension which is due to the N N scalar coupling Although this experiment provides information regarding the magnitude of the coupling, it does not identify the coupling partner Figure 4 Uridine imino region of a highly resolved 2D H N HSQC spectrum recorded on the PSTVd T RNA domain The data matrix consisted of 300* Ž N 024* Ž H data points Žwhere n* refers to complex points with acquisition times of 0 ms Ž N and 77 ms Ž H Total experimental time was 4 h Other experimental conditions as in Fig 3 Resonances are labeled with assignment information Each uridine imino N3 resonance is split according to a h2 J NN coupling of 8 Hz which correlates the uridine N3 and the adenosine N nuclei through the H-bonds in the respective UA WatsonCrick base pairs Simultaneous Detection of Coupling Partners and Quantification of JNN Couplings The Quantitatie HNN-COSY Both the chemical shift of the coupling partners and the magnitude of the n J NN couplings can be determined effectively by using a single quantitative, N-homonuclear COSY experiment In this experiment, we can observe the magnetization which is transferred to the partner N nucleus Ž or nuclei which we could not observe in the 2D H N HSQC approach described above As the rate of magnetization transfer in COSY experiments is given by the magnitude of the coupling constant, the intensity of the cross-peaks presents a measure for the size of the coupling Ž 46 Figure 5Ž A depicts the basic HNN-COSY pulse scheme employed to observe and quantify such n JNN couplings in nucleic acid base pairs During the first INEPT part of this sequence, from time point a to time point b, magnetization is transferred from the imino proton of one base via the covalent bond to the imino nitrogen nucleus to give the familiar antiphase magnetization term 2HzN y During the following first N N COSY delay from point b to point c, part of this magnetization is transferred onto the N coupling partner nucleus as an operator term 4H NNsinŽ 2 J z x z NN, while another part remains on the imino N nucleus as an operator term 2H NcosŽ 2 J z y NN, just as described previously by Eq To ensure maximum excitation of both N nuclei and thus transfer of magnetization via the scalar coupling, the frequency of the N 80 y pulse in the middle of the COSY period is centered approximately at the midpoint between the chemical shift of the imino nuclei and the chemi- cal shift of the coupling partner Ž ie, 85 ppm Apart from its purpose to refocus the chemical shift evolution of the imino N nucleus and to recouple the second N nucleus, this pulse also decouples the N imino nuclei from any heteronuclear coupling partners The following N 90 y COSY mixing pulse at time point c does not affect the nitrogen in-phase 2H N term, but rotates both nitrogen magnetizations in the nitrogen anti-phase term 4H NN : z x z Ž Ž 2H Ncos 2 J 4H NNsin 2 J z y NN z x z NN 90N,90N y y 2H NcosŽ 2 J z y NN Ž 4H NN sin 2 J 3 z z x NN During the evolution period t, between time points d and e, both N and N nuclei precess z y

13 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS Figure 5 Ž A Basic HNN-COSY pulse sequence Ž 53 Narrow and wide pulses correspond to flip angles of 90 and 80, respectively Carrier positions are H O Ž H 2 and 85 ppm Ž N All regular H and N pulses are applied at an RF field strength of 29 and 58 khz, respectively Low-power Ž water flip-back 90 H pulses Ž illustrated as smaller narrow pulses are applied at a field strength of 200 Hz For clarity 3 C decoupling pulses are not shown Delays: 225 ms, ms, a 25 ms, b 025 ms, c 225 ms, d 05 ms Unless indicated, the phases of all pulses are applied along the x axis Gradients are sine-bell shaped, with an absolute amplitude at their center and durations Ž polarities of G 25 Ž, 2 Ž, 35 Ž, 235 Ž, 02 Ž, 04 Ž, and 00 ms Ž,2,3,4,5,6,7 Ž B Two-dimensional HNN-COSY spectrum recorded on the PSTVd T RNA domain The data matrix consisted of 250* Ž N 024* Ž H data points Žwhere n* refers to complex points with acquisition times of 40 Ž N and 77 ms Ž H Total experimental time was 23 h Other experimental conditions as in Fig 3 Positive contours depict the diagonal resonances for the G and U imino groups Negative contours correspond to cross peaks resulting from scalar N N magnetization transfer between the donor imino N nucleus and the acceptor N nucleus on the opposing base in WatsonCrick base pairs The diagonal peak marked by U7 represents a uridine imino group in a GU wobble base pair ŽFig 2Ž C, see text Resonances are labeled with assignment information The insert illustrates the definition of the scalar h2 J correlation via the H-bond NN

14 6 DINGLEY, CORDIER, AND GRZESIEK according to their chemical shifts: Ž Ž 2H Ncos 2 J 4H NN sin 2 J z y NN z z x NN N t N N z N z t 2H NcosŽ 2 J cosž t z y NN N 2H NcosŽ 2 J sinž t z x NN N Ž Ž 4H NN sin 2 J cos t z z x NN N Ž Ž 4H NN sin 2 J sin t z z y NN N 4 Note that during the t-evolution, also the homonuclear N N couplings and the heteronuclear JNH coupling is active Due to their relatively small size, the homonuclear N N couplings are not resolved during the maximal acquisition time for the t-evolution of approximately 40 ms The effect of the heteronuclear JNH coupling is used in the TROSY-type detection at the end of the pulse scheme Ž see below For a reverse-inept step at the end of the pulse sequence, a H 80 pulse would be necessary in the middle of the t-period in order to remove the splitting of the resonance line Of the product terms in Eq 4, the second and the fourth either do not refocus into observable magnetization at the end of the sequence or are suppressed by appropriate phase cycling The fate of the first and third terms after the second N 90 y COSY pulse at time point e and the COSY refocusing period until point f is the following: 2H NcosŽ 2 J cosž t z y NN N 4H NN sinž 2 J cosž t z z x NN N 90N,90N y y 2H NcosŽ 2 J cosž t z y NN N Ž Ž 4H NNsin 2 J cos t z x z NN N 80N,80N 2 JNNNzNz y y 2 JNNNzNz 2 Ž Ž 2H Ncos 2 J cos t z y NN N Ž Ž Ž 4H NNcos 2 J sin 2 J cos t z x z NN NN N Ž Ž Ž 4H NNsin 2 J cos 2 J cos t z x z NN NN N 2 Ž Ž 2H N sin 2 J cos t z y NN N As in the case of the normal HSQC, only the first and the last product operator terms, both of type 2HzN, y are converted into observable imino proton magnetization during the final TROSY Ž 47 part of the pulse sequence from point f to point g The TROSY-type detection scheme will not be discussed any further here, except to point out that a reverse-inept pulse train would also lead to observable magnetization However, it is advantageous to use the TROSY scheme, because it only detects the narrow component of the H N doublet, thereby increasing the sensitivity of the experiment for larger molecules It is apparent from equation Eq that the amplitudes of the two 2HzNy terms contain the factors cos 2 Ž 2 J cosž t and sin 2 Ž 2 J NN N NN cosž t N Fourier transform with respect to t yields resonances in the f frequency domain at position N and N The intensities of these diagonal and cross-peaks are proportional to cos 2 Ž 2 J and sin 2 Ž 2 J NN NN, respectively We also note that for quadrature detection using a complex Fourier transform, it is necessary to record magnetization terms which are proportional to cos 2 Ž 2 J sinž t NN N and sin 2 Ž 2 J sinž t NN N, respectively These sine-modulated terms can be obtained as observable magnetization at the end of the pulse sequence if the phases of both 90 N-pulses at positions b and c are rotated by 90 The result of such a quantitative HNN-COSY experiment for the T domain of PSTVd is shown in Fig 5Ž B For each WatsonCrick base pair we observe a diagonal and a cross-peak The diagonal peaks ŽFig 5Ž B, top correspond to the imino H N and H3 N3 frequencies of guanosine and uridine, respectively The cross- peaks ŽFig 5Ž B, bottom correlate the imino H frequencies to N resonances around 97 ppm in the case of guanosine and to resonances around 222 ppm in the case of uridine To which N nuclei do these cross-peaks belong? Based on the N chemical shifts of the uridine cross peaks, we can exclude the uridine N nuclei with typical resonance frequencies of approximately 45 ppm ŽFig 2Ž A The N nucleus would be the only other nitrogen nucleus Žbesides N3 within the covalent structure of uridine Within the H- bonded structure of a WatsonCrick UA base pair ŽFig 2Ž A, the adenosine hydrogen bond N acceptor atom is the next closest to the uridine imino N3 atom The chemical shift of the adenosine N nucleus has typical values of approximately 222 ppm Therefore its chemical shift matches the cross peaks observed for the uridine imino resonances Such an assignment can be validated by other experiments, eg, a NOESY which connects the H3 proton of uridine and the H2 proton of adenosine and a long-range H N HSQC which correlates the H2 and N frequencies of adenosine Ž 53 Therefore,

15 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS 7 we can conclude that we have identified a twobond scalar coupling across the H-bond which is active between the uridine N3 and the adenosine N nuclei in a Watson-Crick UA base pair Following the notation introduced earlier we term this coupling h2 J NN The assignment of the guanosine cross peaks in Fig 5Ž B is completely analogous Within a Watson-Crick GC base pair, the hydrogen bond acceptor for the imino group of guanosine is the N3 atom of cytidine The observed chemical shift range of ppm for the guanosine crosspeaks in Fig 5Ž B matches closely to typical chemical shifts of approximately 97 ppm for the Ž Ž N3 nuclei in cytidine Fig 2 B Therefore, also in the case of the GC base pairs, the cross-peaks for the H N imino resonances of guanosine correspond to a h2 JNN correlation with the H-bond cytidine N3 acceptor nucleus The magnitude of these h2 JNN couplings can be determined from the ratio of the cross-peak Ž I c to diagonal peak Ž I d intensities observed in the HNN-COSY experiment ŽEq The ratio of these intensities is given by 2 Ž h2 2 Ž h2 I I sin 2 J cos 2 J c d NN NN 2 Ž h2 tan 2 J 6 NN It should be kept in mind that only the absolute value of h2 JNN can be determined by this 2 Ž h2 method, because both sin 2 J NN and 2 Ž h2 cos 2 J NN terms are not sensitive to a sign change of h2 J NN A simple rearrangement of equation Eq 6 allows the straightforward calculation of this value: h2 2 NN Ž c d J arctan Ž I I Ž 2 7 As calculated from the intensities of the HNN-COSY of Fig 5Ž B, the absolute sizes of the h2 JNN couplings are 78 Hz These are in agreement with the values measured from the line splitting observed in the H N HSQC Ž Fig 4 and the estimate derived from the zero crossing of the intensities in Fig 3 Interestingly, the imino group of one uridine Ž U7 in Fig 5Ž B has no H-bond correlation This uridine is part of a UG base pair ŽFig 2Ž C For such a UG base pair, the acceptor atom for the uridine imino proton is an oxygen atom rather than a nitrogen atom Therefore no h2 JNN correlation can be seen in the HNN-COSY experiment This returns us to the question posed earlier on why the magnetization for the WatsonCrick H-bonded imino resonances decayed at different rates in Fig 3Ž B as compared to the resonances which are upfield of 2 ppm The answer is straightforward: as all these imino groups are part of non-watsoncrick base pairs with oxygen acceptor atoms, the magnetization decays only by relaxation and not due to the N N scalar coupling mechanism Other Types of H-Bond Couplings in Nucleic Acids Besides the observation of h2 JNN couplings in WatsonCrick base pairs, H-bond scalar interactions between imino as well as amino N nuclei and the N nuclei of aromatic nitrogen H-bond acceptors have been observed in a number of non-watsoncrick base pairs Ž 59, 6, 64, 65 In the case for H-bond scalar interactions between N amino donor and aromatic acceptor nuclei, the frequency separation between the amino and aromatic N resonances Ž ie, 00 ppm is too large to be covered effectively by the limited strength of the currently available non-selective radio frequency pulses Žeg, 67 khz and a magnetic field strength of 4 Tesla This problem is overcome in the pseudo-heteronuclear HŽ N N- COSY experiments where band-selective N pulses Ž 6, 65 excite the N amino donor and aromatic acceptor resonances separately Further modifications of the HNN-COSY scheme have been proposed for the observation of h2 JNN correlations where the hydrogen nucleus in the hydrogen bond is unobservable due to intermediate conformational exchange of hydrogen bonding amino groups or due to exchange of the proton with the solvent Ž 62, 66, 72 In this situation, nearby carbon-bound protons can be used as starting and end points for the magnetization pathway In almost all cases, values of h2 JNN are in the range of approximately 60 Hz, such that there is no indication of any fundamental difference in the mechanism of magnetization transfer across H-bonds from either imino or amino groups to aromatic nitrogen acceptors In addition to the h2 JNN couplings observed in N H N systems, H-bond scalar couplings of the type h JHN have also been observed in a number of base pairs where these one-bond couplings connect the imino protons to the N nuclei of nitrogen H-bond acceptors Ž 54, 59, 76 The size of these h JHN couplings ranges between 2 and 4 Hz Due to their small size and the faster relaxation times of the imino proton, the detection of the h JHN couplings is more challenging than for the h2 J couplings NN

16 8 DINGLEY, CORDIER, AND GRZESIEK Apart from the aromatic nitrogen atoms, many nucleic acid base pairs involve oxygen atoms of carbonyl groups as hydrogen bond acceptors Žie, N H O C, for example, the GU base pair ŽFig 2Ž C Although the size of scalar couplings across the hydrogen bond to the magnetically active oxygen isotope 7 O is likely to be on the same order as the couplings to a nitrogen N acceptor, the fast relaxation of this quadrupolar nucleus Ž I 52 would prevent the observation of such a coupling Instead hydrogen bond couplings to the next possible nucleus of the carbonyl acceptor group, the carbonyl 3 C carbon, have been observed in guanosine quartets Ž 65 The size of the observed three-bond couplings Ž h3 J NC from the imino N donor nucleus to the carbonyl 3 C acceptor nucleus is approximately 02 Hz Ž 65 as measured by a long-range HNCO experiment Ž see protein section below using selective car- 3 bonyl C pulses Four-bond couplings Ž h4 J NN of approximately 04 Hz have also been observed in a similar G-quartet system Ž 67 where the scalar couplings connect the imino N nuclei in guanosine N H O6 C6 N hydrogen bond networks Thus, these h3 J and h4 J NC NN couplings are much smaller than the h2 JNN couplings that we described above Apparently, the larger number of intervening bonds and possibly also the deviations from a linear hydrogen bond geometry Ž 65 reduce the efficiency of magnetization transfer across the electron cloud Hydrogen Bond Scalar Couplings in Proteins N H O C Hydrogen Bonds The predominant hydrogen bond in proteins connects the backbone amide proton of one amino acid to the carbonyl oxygen atom of a second amino acid Ž Fig 6, insert Protein secondary structures are synonymous with the description of the backbone N H O C hydrogen bond networks The canonical patterns are H i O k, Oi H k, H i-2 O k2, Oi-2 H k2 for anti-parallel - sheets, H i O k, Oi H k2, H i2 O k2, Oi2 H k4 for parallel -sheets, Oi H i4 for -helices, and Oi H i3 for 30-helices where H i and O k stand for the backbone amide proton and oxygen atoms of the hydrogen bonded residues i and k, respectively 3 C α H α N H O O 3 C' residue i h3 J NC' N H 3 C' 3 C α residue j V70 I44 R42 H68 i V7 s L43 s E8 R42 V70 s L C I3 V5 s K6 F45 K48 I3 L 75 L67 F4 V5 I3 s V5 K6 L67 s T55 L50 L43 S57 P9 H68 I44 i I44 s F45 T7 K I23 R54 76 E64 Q ppm HN Figure 6 Selected region of the 2D HŽ N CO spectrum recorded on a sample of 6 mm uniformly 3 C N-labeled ubiquitin, 95% H 2O5% D2O, ph 46 The data matrix consisted of 65* Ž 3 C 52* Ž H data points Ž where n* refers to complex points with acquisition times of 39 Ž 3 C and 53 ms Ž H Data were recorded at 45 C on a 600 MHz Bruker DMX spectrometer Ž 2 h total experimental time Cross-peaks marked as h3 Resi Res j are due to JNiCj H-bond scalar couplings between the N nucleus of residue 3 i and C nucleus of residue j Ž see insert for definition Residue names marked by the superscript s denote incompletely suppressed sequential correlations between the N nucleus of residue i and 3 C nucleus of residue i This incomplete suppression of J NC correlations is a result of the variation in the size of JNC couplings in the range of 3 to 3 7 Hz Ž 32 The superscript i demarks intraresidue two-bond N C correlations i i

17 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS 9 As in the case of nucleic acid hydrogen bonds involving carbonyl groups as acceptors, the carbonyl oxygen nucleus in proteins is not accessible for detection of scalar couplings across the H- bond However, in an analogous way, h3 JNC H- bond couplings can be detected involving the 3 C nucleus of the carbonyl moiety and the N nucleus of the H-bond donor Ž 55, 58 The detech3 tion of these JNC couplings in N H O C hydrogen bonds in proteins is achieved by a straightforward modification of the HNCO experiment Ž 48, 49 used for the sequential assign- ment of protein backbones In the conventional HNCO experiment magnetization follows an out and back path according to H Ni N i 3 C i N i H Ni In order to achieve the transfer between the amide N nucleus of one amino acid and the carbonyl 3 C nucleus of the preceding amino acid, the delays for the N i y 2N i C i INEPT step are usually set to values x z slightly shorter than Ž2 J NC With typical JNC values in the range of 3 to 7 Hz, this corresponds to transfer periods of approximately 30 ms In an HNCO experiment suitable for magnetization transfer by h3 J NC, the INEPT delays are set to 33 ms 2 J Ž 55, 58 NC such that the one-bond transfer from the amide N nu- Ž i 3 cleus of residue i N y to the carbonyl C nucleus Ž i i- of residue i- 2N C is approximately rex z Ž 2 Ž focused ie, sin J T NC 0, where T s However, transfer by N C couplings with coupling constant values different from Hz will still occur As a consequence, the resulting HNCO spectrum does not contain the i i 3 normal H N i C cross-peaks, but cross-peaks resulting from eventual long-range correlations, such as the h3 JNiC j couplings between the amide N nucleus of residue i and the carbonyl 3 C nucleus of the hydrogen bonded residue j Ž Fig 6, insert Figure 6 shows the result of such a modified HŽ N CO experiment where magnetization was transferred by long-range JNC scalar interactions between the amide N and carbonyl 3 C nuclei in human ubiquitin As only amide H N and carbonyl 3 C frequencies were detected in this twodimensional version of the experiment, the J NC N 3 interactions are apparent as H C crosspeaks Many of these correlations correspond to H-bond h3 JNC interactions within the ubiquitin secondary structure For example, two cross-peaks are visible between the backbone amide and carbonyl groups of residues isoleucine 44 and histidine 68 which are part of an antiparallel -sheet A summary of these interactions is shown by the arrows in the secondary structure diagram of ubiquitin in Fig 7 Quantification of the h3 J NC coupling constants can be achieved by compari- α-helix β 3 β 4 β 5 β β 2 G35 I36 P37 P38 D39 Q40 E34 K33 D32 Q3 I30 K29 A28 K27 V26 N25 E24 I23 T22 D52 E5 D2 L50 Q49 K48 G47 A46 Q4 R42 L43 I44 F45 Q62 S20 P9 E8 I6 K63 L7 V70 L69 H68 L67 T66 S65 E64 N60 G53 R54 T55 L56 S57 D58 COO Figure 7 Secondary structure topology of ubiquitin Arrows mark the h3 J correlations NC which were observed in the long-range HNCO experiment of Fig 6 Y59 L8 T7 K6 V5 F4 I3 Q2 M T9 G0 K T2 I3 T4 L E6 V7

18 20 DINGLEY, CORDIER, AND GRZESIEK son of the cross-peak intensities from this experiment to the intensities of sequential amide carbonyl correlations measured in a second reference experiment Typical values of h3 JNC coupling constants range from 02 to 09 Hz Ž55, 57, 58 Althoughtheabsolutesizeofthesecou- plings is small, it is possible to trace out complete hydrogen bond networks in smaller, nondeuterated proteins Ž 0 kda by using this long-range HNCO method For larger proteins, the sensitivity of the long-range HNCO rapidly decreases due to N transverse relaxation losses during the extended magnetization transfer periods An improvement in sensitivity can be obtained by the TROSY approach Ž 47, deuteration, and the use of larger magnetic field strengths Such an HNCO-TROSY experiment was successful in observing h3 JNC couplings in the perdeuterated 30 kda ribosome inactivating protein MAP30 Ž 63 Another source of improvement in sensitivity to the long-range HNCO experiment has been shown by quenching scalar coupling mediated relaxation by using adiabatic or composite-pulse decoupling on the aliphatic carbons Ž 68, 69 In this study, a 50% sensitivity enhancement over the standard long-range HNCO experiment performed on a 2 kda FK506 binding protein was observed In addition to the h3 JNC couplings, H-bond h2 couplings of the type J Ž 56, 73 HC and of the h3 type J Ž 74 HC have also been observed between the amide proton of the donor amino acid and the 3 C carbonyl or 3 C nuclei of the accepting amino acid With typical values in the range of 0 Hz, these h2 J and h3 HC JHC coupling constants are similar in absolute size to the h3 J NC coupling values Other Hydrogen Bonds A number of H-bond scalar correlations have also been observed in proteins for H-bonds that are not of the predominant N H O C type For example, the earliest observations of H-bond couplings in proteins Ž 5, 52 involved scalar interactions between amide protons and the nucleus of a cysteine-coordinated metal atom Ž 3 Cd or 99 Hg in rubredoxin Apparently these J-couplings of up to 4 Hz were mediated via the hydrogen bonds of type N H S Me from the backbone amides to the S atoms of the cysteine residues H-bonds of type N H N have also been detected by scalar couplings in a hydrogen bonded pair of histidine imidazole rings in apomyoglobin Ž 60 In this case, h2 JNN couplings in the range of 8 to Hz connect the protonated N 2 nucleus of one histidine to the unprotonated acceptor N 2 nucleus of the second histidine H-bond couplings involving protein amide groups as donors and phosphate groups as acceptors have been described in a molecular complex of Ras p2 and GDP 75 In these N H Ž O P H-bonds, couplings involve either the amide N nucleus Ž h3 J or the amide proton Ž h2 J NP HP of the donor amino acid and 3 P nucleus of the acceptor phosphate group In complete analogy to the HNCO experiment, the h3 JNC correlations can be detected in an HNPO experiment, the only difference being the replacement of the 3 C carbonyl 3 h3 h2 pulses with P pulses Ž 75 With JNP and JHP coupling constants of up to 5 Hz in size, the detection of such correlations is particularly important because long-range information on the position of phosphorus in biomolecular complexes is very hard to obtain by traditional high resolution NMR methods Correlation of H-Bond Couplings and Proton Chemical Shifts Strong correlations of the isotropic chemical shift of the H-bonded proton and the H-bond scalar coupling constants have been observed for both nucleic acids Ž 59 and proteins Ž 55 Irrespective of base pair type, in nucleic acid N H N H-bonds, both h2 J and NN JHN increase linearly with increasing imino proton chemical shifts A similar correlation is observed in protein h3 N H O C H-bonds, where JNC values decrease Ž h3 J 0 NC linearly with increasing Ž N N amide proton H chemical shifts Such H downfield shifts have been correlated with decreasing H-bond lengths Ž 9498 Thus, the observed correlations between H-bond coupling constants and proton chemical shifts indicate that both parameters are mainly determined by the spatial arrangement of the donor and acceptor groups, and that shorter H-bond lengths correspond to proton downfield shifts and stronger h2 h3 Ž absolute size JNN and JNC coupling constants both in nucleic acids and proteins Dependence of H-Bond Couplings on H-Bond Lengths and Angles The comparison of H-bond coupling constants to the exact H-bond geometry is currently hampered by the limited availability of very high resolution structures and by the limited precision of hydrogen atom coordinates which are determined by X-ray diffraction In particular, there are no X-ray

19 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS 2 diffraction data available for most of the nucleic acids for which H-bond couplings have been measured Statistical analysis of a number of other high resolution nucleic acid structures showed that the variation of the N N3 distances in WatsonCrick base pairs is limited Ž 59 In these DNA and RNA structures, the average N N3 distances are A and A for GC and AT Ž AU base pairs The shorter N Ž N3 distances in AT AU base pairs as compared to GC base pairs coincide with an increase in the value of the h2 JNN coupling constants from approximately 67 Hz Ž GC to approximately 78 Hz Ž AT, AU Ž 53, 54, 59 Density functional theory simulations Ž 0 indicate that the different chemical nature of the donor and acceptor groups in WatsonCrick GC and AT Ž AU and in Hoogsteen A T base pairs have a very limited influence on the h2 J NN coupling constants 02 Hz, for N Ž N distances from 27 to 40 A Therefore, observed differences in h2 JNN values for GC, AT, AU, and A T base pairs should be largely due to the differences in donoracceptor distances Neglecting angular variations, it follows that at N N distances of 28 to 29 A, a change in the value of h2 JNN by Hz corresponds to a change in donoracceptor distance of A Decreases of h2 JNN coupling constants by approximately Hz have been observed at the ends of helical stems and have been interpreted as a corresponding increase in the ensemble average of the donoracceptor distance Ž 59 Compared to the geometry of N H N H-bonds in nucleic acid base pairs, there is far greater variation in the geometry of N H O C H-bonds in proteins A statistical analysis of a number of crystallographic structures Ž 40 showed that the average values for N O and H O distances and for N H O and H O C angles are A, A, 7, 47 9 in the case of helices and A, A, 60 0, 2 in the case of -sheet conformations, respectively Thus, the measured average N O and H O distances are about 0 A shorter in -sheets as compared to -helices This coincides with an increase in the average strength of the observed h3 JNC coupling from Hz for -helical conformations to Hz for -sheets in ubiquitin Ž 55 An exponential correlation between the h3 J NC coupling constant and the N O distance was h3 described for protein G Ž 57 where JNC is 3 given as 590 Hz* exp4r A Ne- NO glecting again angular dependencies, it follows from this relation that for a typical -helical or -sheet conformation, a variation in the value of h3 JNC by 0 Hz corresponds to a change in the N O distance of 007 or 005 A, respectively Detailed information about the angular dependencies of the trans-hydrogen bond coupling constants is currently limited Density functional theory simulations indicate that the h3 J NC value has a maximum for an N H O angle of 80 and decreases in value by approximately 20 to 30% for a decrease in this angle to 40 Ž 8 The relation between the size of h3 JNC and the H O C angle is less well understood A recent study showed that for similar H-bond distances the h3 JNC couplings in a nucleic acid G- tetrad were weaker than the h3 J couplings NC observed in proteins Ž 65 The main difference in the geometry of the two systems is that typical H O C angles in proteins have values of approximately 0 whereas the average H O C angle in the G-tetrad is 25 5 This points towards a similar angular dependency as for the N H O angle, namely, that a deviation from a straight H O C conformation yields weaker h3 JNC couplings Support for this interpretation comes from very recent observah3 tions of JNP couplings in N H O P H- h3 bonds Ž 75 In this study a JNP coupling of 46 Hz was observed for an H O P angle of 73, but the values of h3 JNP dropped below 035 Hz for conformations where the H O P angle was smaller than 26 CONCLUSION This review should serve as an introduction to the phenomenon of trans-h-bond scalar couplings in nucleic acids and proteins In particular, we focused on the observation of h2 JNN couplings in WatsonCrick base pairs by illustrating how the couplings were initially detected and characterized Since the H-bond scalar couplings follow the same electron-mediated polarization mechanism as the covalent scalar couplings, the NMR concepts for their description and the experimental protocols for their detection are identical For example, the detection of h3 JNC couplings in nucleic acids and proteins can be achieved by a straightforward modification of the conventional HNCO experiment In contrast to the strong H- bond couplings involving nitrogen acceptor atoms in many of the nucleic acid base pairs, the size of H-bond couplings involving oxygen acceptor

20 22 DINGLEY, CORDIER, AND GRZESIEK atoms, such as the h3 JNC couplings in proteins, is typically only a fraction of a Hertz Nevertheless, the use of carefully optimized experiments makes many of these interactions detectable in small to medium-sized biomacromolecules Clearly, most of the applications for the H-bond couplings will be directed towards the unambiguous establishment of H-bond networks in the secondary and tertiary structures of proteins and nucleic acids and their complexes However, the size of the H-bond scalar coupling could also be applied to provide previously inaccessible information in other cases; eg, Ž as the H-bond coupling constants are correlated to the H-bond distance and angles, they could be used as a precise measure for the H-bond geometry Ž 2 Changes in the size of the coupling constants should provide an indication for tiny changes in H-bond geometry when biomacromolecules are subjected to different physicochemical conditions, such as changes in temperature, ph, chemical denaturation, or ligand binding Such reproducible changes could indeed be found and rationalized by an induced fit mechanism in a study of the formation of a complex between an SH3 domain and a small peptide Ž Theobserva- tion of H-bond couplings is also possible during protein folding In this case the size of the coupling constants provides a quantitative measure for the population of the closed and open states of the hydrogen bonds A recent application on the TFE-induced helix formation of the RNase A S-peptide shows that such data can be interpreted in the context of common coil to helix transition theories Ž 2 Ž 4 The coupling constants could be used to characterize H-bond networks in the active centers of enzymes and ribozymes In particular, the proposed low-barrier H-bonds should give rise to very strong coupling values ACKNOWLEDGMENTS We thank Professors Juli Feigon and Michael Barfield for very fruitful discussions, and Professors Detlev Riesner and Georg Buldt for continuous support AJD is a recipient of an Australian National Health and Medical Research Council CJ Martin Postdoctoral Fellowship FC is a recipient of an A v Humboldt fellowship REFERENCES Latimer WM, Rodebush WH Polarity and ionization from the standpoint of the Lewis theory of valence J Am Chem Soc 920; 42: Huggins ML Electronic structures of atoms, J Phys Chem 922; 26: Jeffrey GA, Saenger W Hydrogen bonding in biological structures Berlin: Springer-Verlag; 99 4 Pimentel GC, McClellan AL The hydrogen bond San Francisco: W H Freeman; Hamilton WC, Ibers J Hydrogen bonding in solids New York: W A Benjamin; Vinogradov SN, Linnel RH Hydrogen bonding New York: Van Nostrand Reinhold; 97 7 Joesten MD, Schaad LJ Hydrogen bonding New York: Marcel Dekker; Schuster P, Zundel G, Sandorfy C The hydrogen bond: recent developments in theory and experiments Amsterdam: North Holland; Jeffrey GA An introduction to hydrogen bonding New York: Oxford University Press; Scheiner S Hydrogen bonding: a theoretical perspective New York: Oxford University Press; 997 Desiraju GR, Steiner T The weak hydrogen bond in structural chemistry and biology Oxford: Oxford University Press; Pimentel GC, McClellan AL Hydrogen bonding Ann Rev Phys Chem 97; 22: Kollman PA, Allen LC The theory of hydrogen bonds Chem Rev 972; 72: Hibbert F, Emsley J Hydrogen bonding and chemical reactivity Adv Phys Org Chem 990; 26: Scheiner S Ab initio studies of hydrogen bonds: the water dimer paradigm Annu Rev Phys Chem 994; 45: Becker ED Hydrogen bonding In: Grant DM, Harris RK, editors Encyclopedia of nuclear magnetic resonance New York: John Wiley; 996 vol 4, p Perrin CL, Nielson JB Strong hydrogen bonds in chemistry and biology Annu Rev Phys Chem 997; 48:544 8 Schuster S, Mikenda W Hydrogen bond research New York: Springer-Verlag; 999 vol 9 Pauling L The nature of the chemical bond Ithaca: Cornell University Press; Steiner T, Saenger W Role of C H O hydrogen bonds in the coordination of water molecules Analysis of neutron diffraction data J Am Chem Soc 993; : Derewenda ZS, Derewenda U, Kobos PM Ž His C epsilon- H O C hydrogen bond in the active sites of serine hydrolases J Mol Biol 994; 24: Derewenda ZS, Lee L, Derewenda U The occurrence of C H O hydrogen bonds in proteins J Mol Biol 995; 252: Bella J, Berman HM Crystallographic evidence for C alpha- H O C hydrogen bonds in a collagen triple helix J Mol Biol 996; 264:734742

21 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS Fabiola GF, Krishnaswamy S, Nagarajan V, Pattabhi V C H O hydrogen bonds in betasheets Acta Cryst 997; D53: Chakrabarti P, Chakrabarti S C H O Hydrogen bond involving proline residues in alphahelices J Mol Biol 998; 284: Mandel-Gutfreund Y, Margalit H, Jernigan RL, Zhurkin VB A role for CH O interactions in protein-dna recognition J Mol Biol 998; 277: Vargas R, Garza J, Dixon A, Hay BP How strong is the CA H O C hydrogen bond? J Am Chem Soc 2000; 22: Fersht A Enzyme structure and mechanism, 2nd ed New York: W H Freeman; Fersht AR Structure and mechanism in protein science A guide to enzyme catalysis and protein folding New York: W H Freeman; Pauling L, Corey RB, Branson HR The structure of proteins: two hydrogen-bonded helical configurations of the polypeptide chain Proc Natl Acad Sci USA 95; 37: Watson JD, Crick FHC Molecular structure of nucleic acids A structure for deoxyribose nucleic acid Nature 953; 7: Larson JW, McMahon TB Gas-phase bifluoride ion An ion cyclotron resonance determination of the hydrogen bond energy in FHF from gasphase fluoride transfer equilibrium measurements J Am Chem Soc 982; 04: Morgan BP, Scholtz JM, Ballinger MD, Zipkin ID, Bartlett PA Differential binding energy: a detailed evaluation of the influence of hydrogen bonding and hydrophobic groups on the inhibition of thermolysin by phosphorus-containing inhibitors J Am Chem Soc 99; 3: Sampson NS, Knowles JR Segmental motion in catalysis: investigation of a hydrogen bond critical for loop closure in the reaction of triosephosphate isomerase Biochemistry 992; 3: Serrano L, Kellis JTJ, Cann P, Matouschek A, Fersht AR The folding of an enzyme II Substructure of barnase and the contribution of different interactions to protein stability J Mol Biol 992; 224: Baca M, Kent SB Catalytic contribution of flapsubstrate hydrogen bonds in HIV- protease explored by chemical synthesis Proc Natl Acad Sci USA 993; 90: Thorson JS, Chapman E, Schultz PG Analysis of hydrogen bonding strengths in proteins using unnatural amino acids J Am Chem Soc 995; 7: Klumb LA, Chu V, Stayton PS Energetic roles of hydrogen bonds at the ureido oxygen binding pocket in the streptavidin-biotin complex Biochemistry 998; 37: Fernandez-Recio J, Romero A, Sancho J Energetics of a hydrogen bond Ž charged and neutral and of a cation-pi interaction in apoflavodoxin J Mol Biol 999; 290: Baker EN, Hubbard RE Hydrogen bonding in globular proteins Prog Biophys Molec Biol 984; 44: Brill R, Tippe A Gitterparameter von eis I bei tiefen temperaturen Acta Cryst 967; 23: Kuhs WF, Lehmann MS The structure of ice Ih by neutron diffraction J Phys Chem 983; 87: Bondi A van der Waals volumes and radii J Phys Chem 964; 68: Cipriani F, Castagna J-C, Lehmann MS, Wilkinson C A large image-plate detector for neutrons Physica B 995; 23 & 24: Myles DAA, Bon C, Langan P, Cipriani F, Castagna J-C, Lehmann MS, Wilkinson C Neutron Laue diffraction in marcomolecular crystallography Physica B 998; 24243: Niimura N Neutrons expand the field of structural biology Curr Opin Struct Biol 999; 9: Niimura N, Minezaki Y, Nonaka T, Castagna JC, Cipriani F, Hoghoj P, Lehmann MS, Wilkinson C Neutron Laue diffractometry with an imaging plate provides an effective data collection regime for neutron protein crystallography Nat Struct Biol 997; 4: Bon C, Lehmann MS, Wilkinson C Quasi-Laue neutron-diffraction study of the water arrangement in crystals of triclinic hen egg-white lysozyme Acta Cryst D 999; 55: Langan P, Lehmann M, Wilkinson C, Jogl G, Kratky C Neutron Laue diffraction studies of coenzyme cobž II alamin Acta Cryst D 999; 55:59 50 Shu F, Ramakrishnan V, Schoenborn BP Enhanced visibility of hydrogen atoms by neutron crystallography on fully deuterated myoglobin Proc Natl Acad Sci USA 2000; 97: Blake PR, Lee B, Summers MF, Adams MW, Park JB, Zhou ZH, Bax A Quantitative measurement of small through-hydrogen-bond and through-space H- 3 Cd and H- 99 Hg J couplings in metal-substituted rubredoxin from Pyrococcus furiosus J Biomol NMR 992; 2: Blake PR, Park J-B, Adams MW, Summers MF Novel observation of NH Ž S Cys hydrogenbond-mediated scalar coupling in 3 Cd-substituted rubredoxin from Pyrococcus furiosus J Am Chem Soc 992; 4: Dingley AJ, Grzesiek S Direct observation of hydrogen bonds in nucleic acid base pairs by internucleotide 2 JNN couplings J Am Chem Soc 998; 20: Pervushin K, Ono A, Fernandez C, Szyperski T, Kainosho M, Wuthrich K NMR scalar couplings across WatsonCrick base pair hydrogen bonds in DNA observed by transverse relaxation-optimized spectroscopy Proc Natl Acad Sci USA 998; 95:4474

22 24 DINGLEY, CORDIER, AND GRZESIEK 55 Cordier F, Grzesiek S Direct observation of hydrogen bonds in proteins by interresidue 3h J NC scalar couplings J Am Chem Soc 999; 2: Cordier F, Rogowski M, Grzesiek S, Bax A Observation of through-hydrogen-bond 2h JHC in a perdeuterated protein J Magn Reson 999; 40: Cornilescu G, Hu J-S, Bax A Identification of the hydrogen bonding network in a protein by scalar couplings J Am Chem Soc 999; 2: Cornilescu G, Ramirez BE, Frank MK, Clore GM, Gronenborr AM, Bax A Correlation between 3h JNC and hydrogen bond length in proteins J Am Chem Soc 999; 2: Dingley AJ, Masse JE, Peterson RD, Barfield M, Feigon J, Grzesiek S Internucleotide scalar couplings across hydrogen bonds in WatsonCrick and Hoogsteen base pairs of a DNA triplex J Am Chem Soc 999; 2: Hennig M, Geierstanger BH Direct detection of a histidine-histidine side chain hydrogen bond important for folding of apomyoglobin J Am Chem Soc 999; 2: Majumdar A, Kettani A, Skripkin E Observation and measurement of internucleotide 2 JNN coupling constants between N nuclei with widely separated chemical shifts J Biomol NMR 999; 4: Majumdar A, Kettani A, Skripkin E, Patel DJ Observation of internucleotide N H N hydrogen bonds in the absence of directly detectable protons J Biomol NMR 999; : Wang YX, Jacob J, Cordier F, Wingfield P, Stahl SJ, Lee-Huang S, Torchia D, Grzesiek S, Bax A Measurement of 3h JNC connectivities across hydrogen bonds in a 30 kda protein J Biomol NMR 999; 4: Wohnert J, Dingley AJ, Stoldt M, Gorlach M, Grzesiek S, Brown LR Direct identification of NH N hydrogen bonds in non-canonical base pairs of RNA by NMR spectroscopy Nucleic Acids Res 999; 27: Dingley AJ, Masse JE, Feigon J, Grzesiek S Characterization of the hydrogen bond network in guanosine quartets by internucleotide 3h JNC and 2h JNN scalar couplings J Biomol NMR 2000; 6: Hennig M, Williamson JR Detection of N H N hydrogen bonding in RNA via scalar couplings in the absence of observable imino proton resonances Nucleic Acids Res 2000; 28: Liu A, Majumdar A, Hu W, Kettani A, Skripkin E, Patel DJ NMR detection of N H O C hydrogen bonds in 3 C, N-labeled nucleic acids J Am Chem Soc 2000; 2: Liu A, Hu W, Majumdar A, Rosen MK, Patel DJ Detection of very weak side chain-main chain hydrogen bonding interactions in medium-size 3CN-labeled proteins by sensitivity-enhanced NMR spectroscopy J Biomol NMR 2000; 7: Liu A, Hu W, Qamar S, Majumdar A Sensitivity enhanced NMR spectroscopy by quenching scalar coupling mediated relaxation: application to the direct observation of hydrogen bonds in 3CN-labeled proteins J Biomol NMR 2000; 7: Liu A, Hu W, Majumdar A, Rosen MK, Patel DJ NMR detection of side chainside chain hydrogen bonding interactions in 3CN-labeled proteins J Biomol NMR 2000; 7: Lohr F, Mayhew SG, Ruterjans H Detection of scalar couplings across NH OP and OH OP hydrogen bonds in a flavoprotein J Am Chem Soc 2000; 22: Luy B, Marino JP Direct evidence for Watson Crick base pairs in a dynamic region of RNA structure J Am Chem Soc 2000; 22: Meissner A, Sørensen OW New techniques for the measurement of CN and CH N J coupling constants across hydrogen bonds in proteins J Magn Reson 2000; 43: Meissner A, Sørensen OW 3h J coupling between C and H N across hydrogen bonds in proteins J Magn Reson 2000; 43: Mishima M, Hatanaka M, Yokoyama S, Ikegami T, Walchli M, Ito Y, Shirakawa M Intermolecular 3 3 P N and P H scalar couplings across hydrogen bonds formed between a protein and a nucleotide J Am Chem Soc 2000; 22: Pervushin K, Fernandez C, Riek R Ono A, Kainosho M, Wuthrich K Determination of h2 J and h JHN NN coupling constants across WatsonCrick base pairs in the antennapedia homeodomain- DNA complex using TROSY J Biomol NMR 2000; 6: Kwon O, Danishefsky SJ Synthesis of asialo GM New insights in the application of sulfonamidoglycosylation in oligosaccharide assembly: subtle proximity effects in the stereochemical governance of glycosidation J Am Chem Soc 998; 20: Shenderovich IG, Smirnov SN, Denisov GS, Gindin VA, Golubey NS, Dunger A, Reibke R, Kirpekar S, Malkina OL, Limbach H-H Ber Bunsenges Phys Chem 998; 02: Golubev NS, Shenderovich IG, Smirnov SN, Denisov GS, Limbach H-H Nuclear scalar spinspin coupling reveals novel properties of low-barrier hydrogen bonds in a polar environment Chem Eur J 999; 5: Benedict H, Shenderovich IG, Malkina OL, Malkin VG, Denisov GS, Golubev NS, Limbach H-H Nuclear scalar spin-spin couplings and geometries of hydrogen bonds J Am Chem Soc 2000; 22: Scheurer C, Bruschweiler R Quantum-chemical characterization of nuclear spin-spin couplings

23 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS 25 across hydrogen bonds J Am Chem Soc 999; 2: Perera SJ, Bartlett RJ Predicted NMR coupling constants across hydrogen bonds: a fingerprint for specifying hydrogen bond type? J Am Chem Soc 2000; 22: Hvidt A, Nielsen SO Hydrogen exchange in proteins Adv Prot Chem 966; 2: Teitelbaum H, Englander SW Open states in native polynucleotides II Hydrogen-exchange study of cytosine-containing double helices J Mol Biol 975; 92: Teitelbaum H, Englander SW Open states in native polynucleotides I Hydrogen-exchange study of adenine-containing double helices J Mol Biol 975; 92: Englander SW, Kallenbach NR Hydrogen exchange and structural dynamics of proteins and nucleic acids Q Rev Biophys 983; 6: Gueron M, Leroy J-L Studies of base pair kinetics by NMR measurement of proton exchange Methods Enzymol 995; 26: Pardi A, Tinoco, Jr I Kinetics for exchange of imino protons in deoxyribonucleic acid, ribonucleic acid, and hybrid oligonucleotide helices Biochemistry 982; 2: Patel DJ, Pardi A, Itakura K DNA conformation, dynamics, and interactions in solution Science 982; 26: Wagner G, Wuthrich K Amide protein exchange and surface conformation of the basic pancreatic trypsin inhibitor in solution Studies with twodimensional nuclear magnetic resonance J Mol Biol 982; 60: Wagner G Characterization of the distribution of internal motions in the basic pancreatic trypsin inhibitor using a large number of internal NMR probes Q Rev Biophys 983; 6:7 92 Bai Y, Milne J, Mayne L, Englander S Primary structure effects on peptide group hydrogen exchange Proteins: Structure, Function, and Genetics 993; 7: Englander S, Mayne L Protein folding studied using hydrogen-exchange labeling and two-dimensional NMR Ann Rev Biophys Biomol Struct 992; 2: Pardi A, Wagner G, Wuethrich K Protein conformation and proton nuclear magnetic resonance chemical shifts Eur J Biochem 983; 37: Wagner G, Pardi A, Wuethrich K Hydrogen bond length and proton NMR chemical shifts in proteins J Am Chem Soc 983; 05: Kuntz ID, Kosen PA, Craig EC Amide chemical shifts in many helices in peptides and proteins are periodic J Am Chem Soc 99; 3: Wishart DS, Sykes BD, Richards FM Relationship between nuclear magnetic resonance chemical shift and protein secondary structure J Mol Biol 99; 222: Zhou NE, Zhu B-Y, Sykes BD, Hodges RS Relationship between amide proton chemical shifts and hydrogen bonding in amphipathic -helical peptides J Am Chem Soc 992; 4: Shoup RR, Miles HT, Becker ED NMR evidence of specific base-pairing between purines and pyrimidines Biochem Biophy Res Commun 966; 23: Griffey RH, Poulter CD Detection of the imino hydrogen bond in G C pairs by H and N nuclear magnetic resonance spectroscopy Tetrahedron Lett 983; 24: Frey PA, Whitt SA, Tobin JB A low-barrier hydrogen bond in the catalytic triad of serine proteases Science 994; 264: Markowski V, Sullivan GR, Roberts JD Nitrogen- nuclear magnetic resonance spectroscopy of some nucleosides and nucleotides J Am Chem Soc 977; 99: Griffey RH, Davis DR, Yamaizumi Z, Nishimura S, Hawkins BL, Poulter CD N-labeled trna Identification of 4-thiouridine in Escherichia coli trnaser and trnatyr2 by H- N two-dimensional NMR spectroscopy J Biol Chem 986; 26: Gaffney BL, Kung P-P, Wang C, Jones RA Nitrogen--labeled oligonucleotides 8 Use of N NMR to probe Hoogsteen hydrogen bonding at guanine and adenine N7 atoms of a DNA triplex J Am Chem Soc 995; 7: Zhang X, Gaffney BL, Jones RA N NMR of fragments containing specifically labeled GU and GC pairs J Am Chem Soc 998; 20: Bachovchin WW NNMRspectroscopyofhydrogen-bonding interactions in the active site of serine proteases: evidence for a moving histidine mechanism Biochemistry 986; 25: Ash EL, Sudmeier JL, De Fabo EC, Bachovchin WW A low-barrier hydrogen bond in the catalytic triad of serine proteases? Theory versus experiment Science 997; 278: de Dios AC, Pearson JG, Oldfield E Secondary and tertiary structural effects on protein NMR chemical shifts: an ab initio approach Science 993; 260: Hansen PE Isotope effects in nuclear shielding Prog Nucl Magn Reson Spectrosc 988; 20: Gunnarsson G, Wennerstrom H, Ega W, Forsen S Proton and deuterium NMR of hydrogen bonds: relationship between isotope effects and the hydrogen bond potential Chem Phys Lett 976; 38:9699 Altman LJ, Laungani D, Gunnarsson G, Wennerstrom H, Forsen S Proton, deuterium, and tritium nuclear magnetic resonance of intramolecular hydrogen bonds Isotope effects and the shape of the potential energy function J Am Chem Soc 978; 00:

24 26 DINGLEY, CORDIER, AND GRZESIEK 2 Chan SI, Lin L, Clutter D, Dea P The anomalous deuterium isotope effect on the chemical shift of the bridge hydrogen in the enol tautomer of 2,4- pentanedione Proc Natl Acad Sci USA 970; 65: Smirnov SN, Golubev NS, Denisov GS, Benedict H, Schah-Mohammedi P, Limbach H-H Hydrogen deuterium isotope effects on the NMR chemical shifts and geometries of intermolecular low-barrier hydrogen-bonded complexes J Am Chem Soc 996; 8: Xia B, Wilkens SJ, Westler WM, Markley JL Amplification of one-bond H 2 H isotope effects on N chemical shifts in Clostridium pasteurianum rubredoxin by Fermi-contact effects through hydrogen bonds J Am Chem Soc 998; 20: Loh SN, Markley JL Measurement of amide hydrogen DH fractionation factors in proteins by NMR spectroscopy In: R Angeletti, editor Techniques in Protein Chemistry IV San Diego: Academic Press; 993 p LiWang AC, Bax A Equilibrium protiumdeuterium fractionation of backbone amides in U- 3 C N labeled human ubiquitin by triple resonance NMR J Am Chem Soc 996; 8: Loh SN, Markley JL Hydrogen bonding in proteins as studied by amide hydrogen DH fractionation factors: application to staphylococcal nuclease Biochemistry 994; 33: Bowers PM, Klevit RE Hydrogen bonding and equilibrium isotope enrichment in histidine-containing proteins Nat Struct Biol 996; 3: Khare D, Alexander P, Orban J Hydrogen bonding and equilibrium protium-deuterium fractionation factors in the immunoglobulin G binding domain of protein G Biochemistry 999; 38: Bowers PM, Klevit RE Hydrogen bond geometry and 2HH fractionation in proteins J Am Chem Soc 2000; 22: Halkides CJ, Wu YQ, Murray CJ A low-barrier hydrogen bond in subtilisin: H and N NMR studies with peptidyl trifluoromethyl ketones Biochemistry 996; 35: Markley JL, Westler WM Protonation-state dependence of hydrogen bond strengths and exchange rates in a serine protease catalytic triad: bovine chymotrypsinogen A Biochemistry 996; 35: Harris TK, Abeygunawardana C, Mildvan AS NMR studies of the role of hydrogen bonding in the mechanism of triosephosphate isomerase Biochemistry 997; 36: Zhao Q, Abeygunawardana C, Gittis AG, Mildvan AS Hydrogen bonding at the active site of delta 5-3-ketosteroid isomerase Biochemistry 997; 36: Lin J, Westler WM, Cleland WW, Markley JL, Frey PA Fractionation factors and activation energies for exchange of the low barrier hydrogen bonding proton in peptidyl trifluoromethyl ketone complexes of chymotrypsin Proc Natl Acad Sci USA 998; 95: Harris TK, Mildvan AS High-precision measurement of hydrogen bond lengths in proteins by nuclear magnetic resonance methods Proteins: Structure, Function, and Genetics 999; 35: Evans DF Solvent shifts of nuclear spin coupling constants due to hydrogen bonding J Chem Soc 963; 5: Paolillo L, Becker ED The effect of solvent interactions and hydrogen bonding on N chemical shifts and N-H coupling constants J Magn Reson 970; 2: Axenrod T, Wieder MJ Nitrogen- magnetic resonance spectroscopy Solvent effects on JŽ NH and hydrogen bonding in ortho-substituted anilines J Am Chem Soc 97; 93: Axenrod T, Pregosin PS, Wieder MJ, Becker ED, Bradley RB, Milne GWA Nitrogen- nuclear magnetic resonance spectroscopy substituent effects on N-H coupling constants and nitrogen chemical shifts in aniline derivatives J Am Chem Soc 97; 93: Poulter CD, Livingston CL 3- N-2,3,5-Tri- O-benzoyluridine Detection of hydrogen bonding in A-U base pairs by N NMR Tetrahedron Lett 979; 9: Juranic N, Ilich PK, Macura S Hydrogen bonding networks in proteins as revealed by the amide JNC coupling constant J Am Chem Soc 995; 7: Juranic N, Likic VA, Prendergast FG, Macura S Protein-solvent hydrogen bonding studied by NMR JNC coupling constant determination and molecular dynamics simulations J Am Chem Soc 996; 8: Boyd J, Mal TK, Soffe N, Campbell ID The influence of a scalar-coupled deuterium upon the relaxation of a N nucleus and its possible exploitation as a probe for side-chain interactions in proteins J Magn Reson 997; 24:67 35 LiWang AC, Bax A Solution NMR characterization of hydrogen bonds in a protein by indirect measurement of deuterium quadrupole couplings J Magn Reson 997; 27: Ramsey NF Electron coupled interactions between nuclear spins in molecules Phys Rev 953; 9: Frisch MJ et al Gaussian, Inc, Pittsburgh PA, 995

25 AN INTRODUCTION TO HYDROGEN BOND SCALAR COUPLINGS Petrakis L, Sederholm CH NMR fluorine-fluorine coupling constants in saturated organic compounds J Chem Phys 96; 35: Nikonowicz EP, Sirr A, Legault P, Jucker FM, Baer LM, Pardi A Preparation of 3 C and N labelled RNAs for heteronuclear multi-dimensional NMR studies Nucleic Acids Res 992; 20: Batey RT, Inada M, Kujawinski E, Puglisi JD, Williamson JR Preparation of isotopically labeled ribonucleotides for multidimensional NMR spectroscopy of RNA Nucleic Acids Res 992; 20: Morris GA, Freeman R Enhancement of nuclear magnetic resonance signals by polarization transfer J Am Chem Soc 979; 0: Abragam A The principles of nuclear magnetism Oxford: Clarendon Press; Sorensen OW, Eich GW, Levitt MH, Bodenhausen G, Ernst RR Product operator formalism for the description of NMR pulse experiments Progr NMR Spectrosc 983; 6: Ernst RR, Bodenhausen G, Wokaun A Principles of nuclear magnetic resonance in one and two dimensions Oxford: Clarendon Press; Marion D, Ikura M, Tschudin R, Bax A Rapid recording of 2D NMR spectra without phase cycling: application to the study of hydrogen exchange in proteins J Magn Reson 989; 85: Bax A, Vuister GW, Grzesiek S, Delaglio F, Wang AC, Tschudin R, Zhu G Measurement of homoand heteronuclear J couplings from quantitative J correlation Methods Enzymol 994; 239: Pervushin K, Riek R, Wider G, Wuthrich K Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution Proc Natl Acad Sci USA 997; 94: Kay LE, Ikura M, Tschudin R, Bax A Threedimensional triple-resonance NMR spectroscopy of isotopically enriched proteins J Magn Reson 990; 89: Grzesiek S, Bax A Improved 3D triple-resonance NMR techniques applied to a 3 kda protein J Magn Reson 992; 96: Barfield M, Dingley AJ, Feigon J, Grzesiek S A DFT study of the interresidue dependencies of scalar J-coupling and magnetic shielding in the hydrogen bonding region of a DNA triplex Žsub- mitted Cordier F, Wang C, Grzesiek S, Nicholson LK Ligand-induced strain in hydrogen bonds of the c-src SH3 domain detected by NMR J Mol Biol 2000; 304: Jaravine V, Alexandrescu AT, Grzesiek S Observation of the formation of individual hydrogen bonds during peptide folding Ž submitted 3 Emsley L, Bodenhausen G Gaussian pulse cascades: new analytical functions for rectangular selective inversion and in-phase excitation in NMR Chem Phys Lett 990; 65: Andrew J Dingley completed his undergraduate studies at The University of Sydney in 99, majoring in biochemistry He continued with a PhD under the supervision of Glenn F King developing techniques to measure the translational diffusion coefficient of macromolecules using pulsed-field-gradient NMR In 997 he joined the laboratory of Stephan Grzesiek at the Forschungszentrum Julich as a postdoctoral fellow, where his research has focussed on the development of heteronuclear, multidimensional NMR techniques for structural studies of nucleic acids Florence Cordier completed her undergraduate studies in physics at the University Joseph Fourier of Grenoble in 994 In 998, she obtained a PhD for methodological developments in high resolution NMR and dynamical studies of proteins working in the laboratory of Dominique Marion in Grenoble Since then she has been a postdoctoral fellow in the group of Stephan Grzesiek at the Forschungszentrum Julich, Germany, and at the Biozentrum Basel, Switzerland Current research interests include the interpretation of scalar couplings across hydrogen bonds in terms of hydrogen bond geometry, the influence of physicochemical parameters on the hydrogen bond network in proteins, as well as general structural and dynamical properties of proteins in solution Stephan Grzesiek received his PhD in physics from the Free University of Berlin in 988 for optical studies of proton release in bacteriorhodopsin During a postdoctoral stay at the Roche pharamaceutical company in Basel, Switzerland, he was introduced to high resolution NMR and continued from 99 as a visiting fellow in the laboratory of Ad Bax at the National Institutes of Health, USA In 996, he was appointed to Professor for Biological NMR Spectroscopy at the University of Dusseldorf, Germany In 999, he joined the Biozentrum of the University of Basel, where he is presently a professor in the Department of Structural Biology His work focuses on the development and application of NMR methods for the determination of structure and function of biological macromolecules