13C-detected protonless NMR spectroscopy of proteins in solution

13C-detected protonless NMR spectroscopy of proteins in solution

Wolfgang Bermel a, Ivano Bertini b,*, Isabella C. Felli b, Mario Piccioli b, Roberta Pierattelli b

a Bruker BioSpin GmbH, Rheinstetten, Germanyb Magnetic Resonance Center and Department of Chemistry, University of Florence, Via Luigi Sacconi 6, 50019 Sesto Fiorentino, Italy

Received 2 August 2005

Available online 20 December 2005

Keywords: 13C-detected NMR; Protonless NMR; Sequential assignment; Spin-state selection; Paramagnetic proteins

Contents

1. Why the need of hetero detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2. Instrumental aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3. The problem of the 13C–13C coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

4. A protocol for the assignment of backbone and side chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

5. Detection of resonances in paramagnetic proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

1. Why the need of hetero detection

The NMR determination of the structure of large biological

macromolecules in solution is primarily limited by the fast

transverse relaxation that broadens lines and reduces spectral

resolution. Several steps ahead have been made recently to

overcome this limitation. In particular, the constructive use of

cross-correlated relaxation phenomena enables a reduction of

the effective transverse relaxation rates of specific spins, such

as backbone NH groups [1] and aromatic CH groups [2]. More

recently, selected cross-correlation rates were exploited to

obtain line narrowing for methyl [3] and methylene [4] groups.

The other very efficient way to reduce transverse relaxation

rates consists in 2H isotopic enrichment [5–8]. The lower

gyromagnetic ratio of 2H compared to that of 1H contributes to

a drastic reduction of dipole–dipole interactions thus providing

less efficient relaxation mechanisms. Selective isotope label-

ling also simplifies crowded NMR spectra and allows the

selection of specific resonances [9–11].

0079-6565/$ - see front matter q 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.pnmrs.2005.09.002

* Corresponding author. Fax: C39 055 4574271.

E-mail address: [email protected] (I. Bertini).

Direct-detection of heteronuclei, and of 13C in particular,

offers a valuable alternative to 1H detection. The idea of 13C

direct-detection has its roots in the early days of NMR and has

been used for decades to study small molecules such as organic

and inorganic compounds [12] or metabolites [13]. However,13C direct-detection has not been widely applied to study

biological macromolecules. Thanks its large 1H gyromagnetic

ratio, the 1H sensitivity is indeed intrinsically much higher than

that of 13C, and proton-detected experiments are usually

convenient. Moreover, most of the efforts to improve

instrument technology have been devoted to increase the

sensitivity of 1H detection rather than that of the heteronuclei.

The availability of increasingly high magnetic fields and the

advent of cryogenically cooled probeheads [14] has now

moved the sensitivity of NMR spectroscopy into regions that

were unforeseeable less than a decade ago. As a consequence,13C sensitivity has been dramatically increased up to a level

suitable to turn 13C detected experiments on enriched samples

into routine methods for biomolecular NMR applications.

The features of 13C NMR spectra can be realized from

Fig. 1, which shows a 13C–15N correlation experiment recorded

on 13C,15N labeled reduced monomeric superoxide dismutase

(SOD, 15 kDa) at 14.1 T. The map shows all expected cross

peaks with excellent resolution, including those of Pro residues

as well as those of Asn and Gln side-chains. The spectrum

Progress in Nuclear Magnetic Resonance Spectroscopy 48 (2006) 25–45

www.elsevier.com/locate/pnmrs

http://www.elsevier.com/locate/pnmrs

Fig. 1. The 2D CON-IPAP spectrum acquired on a 1.5 mM sample of 13C,15N

labeled reduced monomeric SOD with a 14.1 T Bruker Avance spectrometer

equipped with a cryogenically cooled probehead optimized for 13C detection at

298 K. The pulse sequence used is reported in Appendix.

Fig. 2. Transverse relaxation rates of C 0, Ca and Ha nuclei calculated as a

function of the overall rotational correlation time at two different magnetic

fields (11.7 T, dashed lines, and 16.4 T, continuous lines). Ca relaxation rates

were calculated both for a protonated (Ca(H)) or a deuterated (Ca(D)) protein.

The relaxation contributions considered were: chemical shift anisotropy [98,

99], one-bond dipole–dipole interactions [23], and various dipole–dipole

interactions with 1H (or 2H) to mimic the average proton density in a protein.

W. Bermel et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 48 (2006) 25–4526

shows 161 cross peaks out of the expected 163 while 159 cross

peaks out of the 169 expected are observed in the 1H–15N

HSQC spectrum. The usual drawbacks of conventional 1H–15N

HSQC experiments (e.g. reduced intensity for those NH in fast

exchange, and solvent-suppression dependent artefacts) do not

affect signal intensity in a C–N experiment.

The interest in 13C NMR lies in the fact that the line

narrowing due to the observation of a low-g nucleus allows

systems with large molecular mass to be investigated. Other

scenarios can be envisaged in which 1H may be difficult to

detect and 13C detection can be helpful. This is the case for

protein regions characterized by chemical/conformational

exchange [15], regions that in many cases are important for

the protein function, where the backbone NH protons could be

difficult to detect, interrupting sequence specific assignment.

Unfolded systems, where the chemical shift dispersion of 1H

resonances may be unfavourable, can efficiently be character-

ized using 13C NMR. This is also true for paramagnetic

systems, where the contributions to relaxation arising from the

paramagnetic centre are automatically reduced by more than a

factor of 10 by shifting from 1H to 13C thanks to the smaller

gyromagnetic ratio.

The transverse relaxation rates as well as coupling topologies

constitute the main aspects to be taken into account in the design

of 13C direct-detection experiments. Through 13C direct-

detection, 13C spins in 2H labeled proteins as well as quaternary

carbons can be directly monitored without requiring several

transfer steps as would be the case for NH detected experiments,

opening more possibilities for designing experiments and also

avoiding the need to suppress the solvent signal. The 13C nuclear

spins with the most promising transverse relaxation properties at

high fields and on systems of high molecular mass are aliphatic

nuclei in deuterated molecules [16]. Indeed the dominant

contribution to relaxation is given by dipole–dipole interactions,

which are only marginally field dependent, and thus give the

opportunity to best exploit the advantages in terms of sensitivity

of high magnetic fields [17]. As an example of the line

narrowing that can be obtained by shifting from 1H to 13C,

transverse relaxation rates of Ha and of Ca are shown in Fig. 2 as

a function of the rotational correlation time tr. Upon 2H

labeling, Ca transverse relaxation rates become very small also

for large values of tr. The disadvantage of aliphatic nuclei is that

the carbon–carbon coupling topology yields complex multiplets

in the direct acquisition dimension. The basic building blocks

needed to fully rely on 13Ca detection, which is recommended

for deuterated proteins and for large molecular weights, are

presented.

Carbonyl nuclei (designated C 0) are characterized by a large

chemical shift anisotropy that constitutes the dominant effect

on transverse relaxation, that significantly increases with

increasing magnetic field (Fig. 2) [18–20]. However, they

present a simpler coupling pattern, with just ONE strong

carbon–carbon one-bond scalar coupling. A consistent set of

experiments to perform complete resonance assignment can be

done relying mainly on the detection of 13C 0 nuclei. These

experiments can also be useful for assigning signals from non-

deuterated proteins of medium size.

The less efficient relaxation pathways for 13C relaxation

compared to those of protons also cause a reduction of 13C

longitudinal relaxation rates, especially in the case of 2H

labeling. This effect, that on one-hand causes longer recovery

times (i.e. an increase in the overall duration of experiments),

on the other hand can also be used constructively in 13C–13C

NOESY experiments.

In the next sections we will briefly review the recent efforts

to improve the technology of probes and the methods currently

available for removing the signal splitting due to the presence

of large 13C–13C couplings. Provided these two problems are

tackled and dedicated pulse sequences are developed, high

quality data can be obtained and a sequence-specific assign-

ment strategy based only on heteronuclei can be proposed [21].

Fig. 3. The specified signal-to-noise ratio of 60% benzene-d6 in 40% p-dioxane

W. Bermel et al. / Progress in Nuclear Magnetic Resonance Spectroscopy 48 (2006) 25–45 27

This alternative strategy, that does not actively exploit protons

and for this reason we called protonless NMR, offers an

alternative approach to the standard triple resonance assign-

ment strategy. Such an approach does not aim at replacing

methods relying on NH-detected TROSY experiments, which

combine favourable relaxation properties due to 2H labelling,

constructive use of relaxation interferences and the sensitivity

of 1H direct-detection [22]. However, the fact that protonless

sequential assignment of proteins is nowadays feasible

constitutes a breakthrough per se.
(the so-called ASTM sample) for 13C-observe probes plotted as a function of
the year when a probe was first introduced at a particular magnetic field

(expressed as 1H operating frequency in MHz). Black dots denote the 13C

sensitivity of conventional 13C-observe probes. For the cryogenically-cooled

probes, the 13C sensitivity of 13C-detection probes are indicated with triangles,

the 13C sensitivity of the triple-resonance probes with both cryogenically

enhanced 1H and 13C capacities are denoted by diamonds and the four-nuclei

heteronuclear-observe probe is indicated with square (adapted with permission

from [14]).

2. Instrumental aspects

The sensitivity in NMR experiments is given by Eq. (1)

S=NzAgexcg3=2obsB

3=20 N1=2

scan (1)

were gexc and gobs are the gyromagnetic ratios of the excited

and of the observed nuclei, respectively, A is the number of

spins, Nscan is the number of scans and B0 is the strength of the

applied magnetic field [23]. Therefore, neglecting relaxation,1H observed experiments have a sensitivity gain of a factor of

32 (i.e. (gH/gC)5/2) compared to the corresponding experiments

starting and ending at 13C nuclei (i.e. HCCH COSY or HCACO

[24] vs 13C–13C COSY or CACO experiments [25]). Recent

developments in probe technology and the availability of NMR

spectrometers operating at high magnetic fields hold promise

for dramatic improvements in sensitivity, not only for 1H but

for 13C as well. 13C sensitivity, at least in NMR spectrometers

devoted to biomolecular NMR applications, has so far been

sacrificed in favour of 1H sensitivity. Indeed, as far as probe

design is concerned, it is generally necessary to achieve a

compromise between the performance of different nuclei in

terms of sensitivity and pulse lengths. Generally, optimal

sensitivity for a specific nucleus is achieved for a 2-coil probe

by using the inner coil for that nucleus, at the expense of the

performance of the other nuclei being on the outer coil.

Probeheads optimised for 13C are nowadays available, such as

dual probeheads, with the inner coil optimised for 13C and with

the possibility to pulse on 1H, and triple-resonance probeheads,

where the inner coil is used to pulse on 13C and 15N. The

evolution of 13C signal-to-noise (S/N) ratio for these probe-

heads over the years is shown in Fig. 3.

Changing the focus from 1H to 13C has prompted

manufacturers to concentrate their efforts in obtaining a new

generation of probeheads in which the main goal is optimized

sensitivity on 13C, coupled with the capability to excite and/or

decouple 1H, 2H and 15N. Indeed, a prototype room-

temperature 13C-observe triple-resonance probehead has been

built yielding a substantial increase in 13C sensitivity and has

greatly contributed to the expansion of the set of experiments

available [21,26,27].

A very efficient way to increase sensitivity consists in

reducing the thermal noise that contributes to the observed

signal, leading to the so-called cryogenically-cooled probe-

heads. This has recently been extensively exploited for 1H

allowing an increase in proton sensitivity by a factor of about

3–4 [14]. The improvement in the sensitivity is mainly due to

the reduced thermal noise achieved by decreasing the operating

temperature of the coils and of the 1H preamplifier. Inclusion of

a cryo-cooled preamplifier also for 13C without changing the

probe design of a standard probe for triple-resonance

experiments has given a significant increase in 13C sensitivity

without needing to make any compromise on 1H. The

behaviour of these probes in terms of S/N is shown in Fig. 3

(diamonds). Dual cryogenically-cooled probes optimised for13C sensitivity (Fig. 3, triangles) are also available but as yet

they do not allow pulsing on 15N and, therefore, are mainly

used to study small molecules. Obviously, a combination of the

cryo-technology with the design of triple-resonance probes

with improved 13C sensitivity is in principle possible and

would contribute to further increase the application of 13C

direct-detection experiments in biomolecular NMR. Further-

more, 13C detected experiments are less prone to sensitivity

losses due to high salt concentrations, particularly relevant in

cryogenically-cooled probeheads. It has been shown that 13C

detection significantly extends the range of sample conditions

under which 13C detection becomes competitive with proton

detection [28].

In order to perform a triple-resonance experiment using 13C

detection, we should also consider aspects other than

sensitivity. The minimal hardware requirements in terms of

transmitters and channels are the same as for standard triple-

resonance experiments (1H, 13C, 15N and 2H in the case of

deuterated proteins). In addition, filters may be required to

prevent cross talk between the 13C and 15N channel. This may

become an issue when 15N is decoupled during 13C detection.

These filters could be part of a band-selective preamplifier or

external devices. Additional hardware may also be needed to

perform band-selective 13C homodecoupling.

Band-selective homodecoupling (BH) of 13C 0 from 13Ca and

vice versa may be desirable in some experiments, but requires

some more consideration. During acquisition, data points are

sampled at a speed given by the amount of oversampling.

Fig. 4. Dwell cycle showing the acquisition status, the pulse and the dwell

window on the time axis. The durations of the decoupling pulse tp, of the

hdduty and of the dwell time are indicated. Time periods D are switching

delays, which are predetermined fractions of the dwell time.


In any case the receiver is unblanked during the whole period.

In order to achieve homodecoupling, radiofrequency has to be

irradiated during a fraction of the nominal dwell time, which in

turn is defined as the reciprocal of the spectral width of the

spectrum. This fraction is called hdduty, the homodecoupling

on/off ratio given as a percentage of the nominal dwell time.

During hdduty as well as some short delays for switching from

observe to transmit mode and vice versa, the receiver system is

blanked (Fig. 4). This results in a reduction in S/N with respect

to the theoretical factor of 2 according to [29]:

S=NðBHÞZ 2S=N

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1K

hdduty

100

r(2)

A short value of hdduty would thus provide a higher S/N.

However, a smaller hdduty value implies higher RF power,

which may cause unwanted probe ringdown and sidebands. All

recent reports have used a 20% hdduty, which corresponds to a

maximum theoretical S/N of 1.79 [29–31]. Experimentally

measured S/N values rarely achieve such an ideal value and

show a significant frequency dependent variability. This

prevents a quantitative comparison of peak intensities within

the same spectrum, but does not prevent the use of

homodecoupling for R1 or R2 experiments, or any other

quantitative experiment. Following the convolution theorem in

Fourier transformation, an irradiation period spaced by a time

interval much longer than irradiation produces sidebands

spaced by n/r, where n is an integer and r is the dwell time,

(in the case where small hdduty values are used). Therefore, the

sweep width and the value of hdduty should be chosen to avoid

overlap of the signals of interest with sidebands arising from

homodecoupling. To perform homodecoupling, adiabatic chirp

pulses, with 10 ms pulse length, 25% smoothing and sweep

from low to high field have been used [32].

3. The problem of the 13C–13C coupling

In 13C direct-detection experiments we need to face the

problem of the large homonuclear one-bond carbon–carbon

couplings that evolve during the acquisition delay.

The presence of these couplings is, of course, beneficial for

coherence transfer efficiency but is detrimental to resolution in

the acquisition dimension. In addition to the large one-bond13C homonuclear scalar couplings (1JCCZ35–55 Hz), there are

also a variety of smaller homonuclear scalar couplings, such as2JCC, 3JCC, and 4JCC, that are generally within the resolution of

the experiments. For this reason, we will focus hereafter on the

one-bond scalar coupling constants, i.e. the 1JC 0Ca and the1JCaCb. In the case of experiments with detection of the 13C 0

nuclei, the splitting is quite uniform and leads to doublets

whose lines are separated by about 55 Hz. The same coupling

is responsible for the primary splitting of the Ca carbon signal,

to which the 35 Hz splitting of the Cb-coupling is added. In the

recent literature, several methods have been proposed to

collapse 1JCC splittings from the spectra [21,29,33,34].

Band-selective homodecoupling is the oldest method of

collapsing homonuclear couplings in the direct acquisition

dimension. In principle, the removal of the splitting due to two

coupled spins with BH allows an increase of the S/N of a factor

of about 1.8 (see Section 2), although this is seldom achieved.

Furthermore, even in the case of Ca carbon decoupling while

acquiring C 0, care should be taken in defining experimental

parameters such as hdduty and spectral width in order to avoid

irradiation of the region of interest by sidebands of the band

selective homodecoupling. Besides causing a Bloch-Siegert

shift [35] BH induces decoupling sidebands around the signals

of interest, since it uses composite pulse decoupling with cyclic

schemes [29,36]. The intensity of the induced sidebands

increases with the strength of the decoupling RF field and,

therefore, it is desirable to use the minimum necessary

decoupling power. Changing the phase of the sidebands in a

controlled manner by starting the decoupling sequence at

different times prior to acquisition permits one to cancel out to

a large extemt those that are due to the cyclic scheme [29,36].

An elegant solution for ‘virtual’ decoupling is the use of

spin-state selective schemes, such as in-phase anti-phase

(IPAP) [37–39] and spin-state selective excitation (S3E)

[39,40], widely used for the measurement of heteronuclear J

couplings [41,42], and recently applied for removal of 13C–13C

J-coupling in solid-state samples and in liquids [21,43].

In the IPAP approach (Fig. 5A,B), the removal of the

splitting is accomplished by recording two FIDs for each

increment, one for the anti-phase and one for the in-phase

components. The IPAP scheme relies on complete interconver-

sion between in-phase and anti-phase coherences and thus has

duration of 1/(2JCC), where JCC is the relevant coupling. For C 0

acquisition, the shortest possible duration of the IPAP block is

thus 9 ms. The in-phase and anti-phase components are stored

separately and are combined to yield the two multiplet

components. These are then shifted to the centre of the original

multiplet (by JC 0Ca/2 Hz) and summed to obtain a singlet

[21,44], as shown in Fig. 5C.

With the S3E scheme (Fig. 5D) two different experiments

are acquired and store separately in which one component is

absortive while the other is dispersive and which differ in

the sign of one of the two components [21,45]. The sum and

the difference of the two FIDs stored separately gives the

Fig. 5. The IPAP and S3E approaches for C 0 direct-detection to remove the large Ca–C 0 splitting in the direct acquisition dimension. They are illustrated for the

simple case of 1D experiments, but can be implemented in any experiment based on C 0 direct-detection, as discussed in the text. Band-selective 13C pulses are

denoted by shapes (narrow and wide ones represent 90 and 1808 pulses, respectively). Panels A and B report the two variants of the pulse scheme for the IPAP

approach to acquire C 0 (A for the in-phase and B for the anti-phase components, respectively). The results of the two experiments reported in panels A and B on a set-

up sample of 13C, 15N labeled alanine in D2O are shown in panel C and indicated with IP and AP, respectively. For 13C 90 and 1808 pulses, Q5 (or time reversed Q5)

and Q3 shapes were used [100]. Decoupling of 1H and 15N was applied with waltz-16 [101] and garp-4 [102], respectively. The 13C 0, 13Ca, 1H and 15N pulses were

applied at 175, 55, 4.7 and 118 ppm, respectively. Panel C also shows schematically the approach employed to treat the data. Panel D shows the pulse scheme for the

S3E approach to acquire C 0. The two FIDs necessary to separate the in-phase and anti-phase components can be obtained by changing the phases of the pulses as

indicated below. The data are treated as described in the text. The delay D is 1/(2JCaC 0) (9 ms). The phases are: (A) fIPAPZx,Kx and frecZx,Kx; (B) fIPAPZKy, y

and frecZx,Kx; D) fS3E(1)Z458, 458; f1Zx, y; f2(1)Zx, y; frecZx,Kx and fS3E(2)Z458, 2258; f1Zx, y; f2(2)ZKx,Ky; frecZx,Kx where (1) and (2) are the

two experiments required to separate the in-phase and anti-phase components. It is worth noting that the S3E requires half the time with respect to the IPAP approach.


absortive and dispersive component, respectively. The

dispersive line in then phase-shifted by 908 and both

components are frequency shifted to the center of the

original multiplet [21,44]. The important difference with the

IPAP approach is that the overall duration of the building

block amounts to 1/(4JCC), where JCC is the relevant

coupling to suppress, and thus is half of that of the IPAP

approach. For C 0 acquisition, the block is 4.5 ms long.

These building blocks (IPAP and S3E) can be implemented in

any experiment based on C 0 direct-detection. In many of them

they can be embedded within the last coherence transfer element

already present in the experiment at no cost in terms of extra

relaxation (i.e. it is not necessary to extend the overall duration

of the experiment). In cases where the coupling to a 15N nucleus

needs to be refocussed (in order to decouple 15N during

acquisition) both building blocks can in principle be used. But

here the use of an S3E element would not shorten the refocussing

delay. So the IPAP approach is the preferred choice, since it is

more robust than S3E with respect to small variations in the

values of 1JCC [46]. Both versions can be implemented in those

experiments that end with the refocusing of the C 0–Ca anti-

phase signals (CACO, CBCACO, CCCO), or those in which the

blocks necessary for spin-state are added at the end of the

experiment prior to acquisition (NOESY, TOCSY). The S3E

approach, due to its shorter duration, may have some advantages

in terms of minimizing relaxation losses at high magnetic fields

and with high molecular mass systems [45].

Along the same lines, Pervushin and coworkers proposed

the acquisition of only the C 0 anti-phase component [47]. The

absolute value of the anti-phase component then looks like an

in-phase component and can be treated in an analogous way to

that described for the IPAP approach (see Fig. 5). A possible

drawback of this method is that the positive and negative

components of different signals in overlap will cancel out

causing loss of information for systems characterized by

extensive resonance overlap. On the other hand, as pointed out

by Dotsch et al., acquiring the anti-phase component allows the

removal of the last delays necessary to refocus it to in-phase

that inevitably causes some relaxation losses [28].

Along the same lines as TROSY, an additional approach to

select one of the two multiplet components in the direct

acquisition dimension of a C 0–Ca correlation experiment

(named COCAINE), has recently been proposed, and provides

an alternative way for ‘virtual’ decoupling of C 0 from Ca [48].

Other methods rely on signal processing algorithms to

deconvolute the C 0–Ca splitting. As an example, deconvolution

using maximum entropy reconstruction has been proposed for

this purpose [33,49]. These processing algorithms can be

applied to spectra where either the in-phase or the anti-phase C 0

signal component [28,33] is acquired.

Direct-detection of Ca is more complicated due to coupling to

C0 and Cb with large one-bond scalar-coupling constants (1JC0Ca,1JCaCb). Therefore, if compared to C0 direct-detection, two large

couplings should be quenched in order to simplify the Ca line and

obtain high-resolution. This can be achieved in principle with the

same approaches discussed for C0, i.e. band-selective homo-

decoupling, spin-state selection and deconvolution.

Pervushin and coworkers proposed triple band-selective

homodecoupling where the three regions irradiated during Ca

acquisition are the C 0 and the two Cb regions [31]. The

performance of this approach is satisfactory in terms of

collapsing the multiplet to a singlet, but the gain in S/N


achieved is much less than predicted. The loss of signal is

attributed to heteronuclear interference coming from simul-

taneous 15N and 2H decoupling [31].

Spin-state selective methods for Ca, such as the described

IPAP and S3E approaches, are sufficient only for those nuclei that

are singly split (e.g. Ca of Gly), but for the removal of the

additional large one bond coupling to Cb we must rely on a

scheme where the spin-state selection is applied repeatedly [21].

Two IPAP blocks, one specific for the C0–Ca coupling and one for

the Ca–Cb coupling can be combined, as suggested for solid state

applications [39]. Actually, the two IPAP blocks can be

concatenated into a shorter block in which the total duration is

determined by the value of the smaller coupling to consider. Fig. 6

shows the double IPAP implementation (DIPAP hereafter) that

has been proposed for Ca [21]. For each increment, four

experiments are acquired that differ in the selectivity of the Ca

pulse at the centre of the delay that determines the effective

evolution of the Ca–Cb coupling during this block and the

position of the C 0 pulses that determine the effective evolution of

the C0–Ca coupling during this delay. These four experiments,

which yield the four possible Ca multiplets, are stored separately,

combined through linear combinations and then shifted to

the centre of the original multiplet in order to remove the primary

splittings. The total length of the block is determined by 1/

(2JCaCb) and, despite compacting the two IPAP blocks into one,

the duration is rather long (14 ms). However, relaxation rates of

Ca nuclei are drastically reduced upon 2H labelling so that this

approach becomes convenient for large proteins at high magnetic

fields. Implementation of a ‘double’ S3E approach to Ca direct-

detection is also feasible.

The methods discussed for C 0 and Ca acquisition can in

principle be applied also to other carbon nuclei. However, the

diversity of the side-chains of aminoacids is responsible for the

large chemical shift dispersion of Cb nuclei, the partial overlap

IP-IP

IP-AP

A

C

B

D

13C´ 4

4

4

4

4

4

4

4

decoupling.1H/ N15

13C´

decoupling.1H/ N15

13C´

.1H/ N15

13C´

d.1H/ N15

FID

FID

4 4

4

4

φrec

13Cα

13CαφDIPAP

φDIPAPφDIPAP

φDIPAP

∆ ∆ ∆ ∆

∆ ∆ ∆ ∆

13Cα/β

13Cα/β13Cα/β

φrec13Cα

13Cα/β

13Cα

∆+ζ ∆

∆+ζ ∆

Fig. 6. The DIPAP method for Ca direct-detection to remove the two large Ca–C 0 a

simple case of 1D experiments and it can be implemented in any experiment based o

and wide ones represent 90 and 1808 pulses, respectively). The two lines labeled wi

unselective with respect to Cb, or both Ca and Cb spins. For 13C 908 and 1808 pulses,

applied with waltz-16 [101] and garp-4 [102], respectively. The 13C 0, 13Ca, 13Ca/b,

Panels A–D report the four variants of the pulse scheme to acquire and store sepa

respectively. The results of the four experiments performed on the set-up sample of 1

(14 ms) and z is 1/(2JCaC 0) (9 ms). The phases are: fDIPAP(B)Zx,Kx; fDIPAP(A)Z

with other carbon nuclei of the side-chain and the different

coupling topologies. Therefore, a general method of detecting

side-chain nuclei by removing large one-bond carbon–carbon

couplings in a single spectrum is not feasible. Partial solutions can

be employed for specific sets of spins. Multiple band-selective

homodecoupling has been proposed [31]. Also the DIPAP

approach has been implemented in the experiment to correlate Cb

and Cg in aromatic systems [50]. Similar approaches can be

designed for methyl groups, or adapted to specific aminoacid

types, as for example was done with the MUSIC approach [51].

Spin-state selective methods retain the information on the

effective C 0–Ca splitting. The latter contains a contribution

arising from partial orientation of the molecule that provides

precious structural information. These experiments are,

therefore, suitable to accurately measure 13C–13C residual

dipolar couplings, also when protons are not detectable or

absent due to perdeuteration. The possibility of determining13C–13C residual dipolar couplings from fully coupled13C–13C TOCSY spectra has recently been described by

Vogeli et al. [52].

4. A protocol for the assignment of backbone

and side chains

The methods described above to detect selected 13C nuclear

spins whilst collapsing the large one-bond homonuclear

carbon–carbon scalar couplings, as well as the availability of

probeheads with improved sensitivity for 13C [14,53], have

opened the way for the use of 13C direct-detection experiments

in biomolecular NMR. We summarize here a set of

experiments based exclusively on heteronuclei that allows

one to perform a complete sequence specific assignment of a13C,15N labeled protein without using 1H excitation and

detection. These experiments are based on the most efficient

AP-AP

AP-IPE

decoupling

ecoupling

FID

FIDIP-IP

AP-IP

IP-AP

AP-AP

596061 D ( C)13

4 4

4

4

φrec

φrec

δ

∆+ζ–ζ ∆–ζ

∆+ζ–ζ ∆–ζ

nd Ca–Cb splittings in the direct acquisition dimension. It is illustrated for the

n Ca direct-detection. Band selective 13C pulses are denoted by shapes (narrow

th Ca and Ca/b actually indicate band-selective pulses that only affect Ca spins,

Q5 and Q3 shapes were used, respectively [100]. Decoupling of 1H and 15N was1H and 15N pulses were applied at 175, 55, 39, 4.7 and 118 ppm, respectively.

rately the four components indicated with IP–IP, AP–IP, IP–AP and AP-AP,3C, 15N labeled alanine in D2O are shown in panel E. The delay D is 1/(2JCaCb’)

y,Ky; fDIPAP(C)ZKy, y; fDIPAP(D)Zx,Kx; frecZx,Kx.

Table 1

Selection of the protonless NMR experiments available, with the correlations observed in each experiment and an estimate of the number of scans necessary to

acquire them compared to a CACO acquired with N number of scans

Experiment Correlations observed Scans Reference

2D

CACO–IPAP/S3E Cai KC0

i N [58,83]

CBCACO–IPAP/S3E Cbi KC0

i; Cai KC0

iN [21,58]

CCCO–IPAP/S3E Cb;g;d;3i KC0

i; Cai KC0

i2N [58]

CON–IPAP NiKC0iK1 2N [58,80]

CANCO–IPAP Cai KC0

iK1 CaiK1KC0

iK1 16N [26,27]

CBCANCO–IPAP Cai KC0

iK1; CaiK1KC0

iK1; Cbi KC0

iK1; CbiK1KC0

iK116N [58]

3D

CBCACO–IPAP/S3E Cbi KCa

i KC0i; Ca

i KCai KC0

iN [21]

CCCO–IPAP/S3E Cb;g;d;3i KCa

i KC0i; Ca

i KCai KC0

i2N [58]

CANCO–IPAP Cai KN0

iKC0iK1; Ca

iK1KNiKC0iK1 16N [26,27]

CBCACON–IPAP CaiK1KNiKC0

iK1; CbiK1KNiKC0

iK12N [58]

CCCON–IPAP CaiK1KNiKC0

iK1; Cb;g;d;3iK1 KNiKC0

iK14N [58]

CBCANCO–IPAP Cai KNiKC0

iK1 CaiK1KNiKC0

iK1; Cbi KNiKC0

iK1; CbiK1KNiKC0

iK116N [58]

Fig. 7. A schematic representation of sequence specific assignment through

protonless NMR spectroscopy. In each panel, the key correlations necessary to

perform spin system identification and sequence specific assignment are shown.

The inset reports a schematic representation of the backbone of a protein and

the magnetization transfer pathways responsible for the observed correlations.

The experiments based on C 0 acquisition are reported on the left side. The

CBCACO and the CCCO can also be acquired in the 3D mode evolving Ca in

the third dimension in order to remove ambiguities deriving from C 0

degeneracy. The CANCO has been designed as a 3D with N evolution. The

CON experiment is reported in an oblique way to indicate that it can be

combined with the other C 0-based experiments by including N evolution and

obtain the corresponding 3D experiment. The TOCSY experiment with Ca

acquisition is reported on the right.


building blocks of the conventional triple-resonance experi-

ments [54–57] and are described in detail in the Appendix. In

Table 1 the list of the experiments, including a summary of the

correlations observed and the relative sensitivity, is reported.

This heteronuclear assignment strategy is then critically

discussed in terms of advantages and drawbacks with respect

to the conventional sequence specific assignment strategy and

is compared to other proposed approaches that partly rely on

heteronuclear direct-detection. The exclusively heteronuclear

assignment strategy starts with acquisition and analysis of the

most sensitive experiments based on C 0 acquisition and on the

exploitation of the large one-bond coupling constants for

coherence transfer (JC 0Ca, JCaCb, JCC and JNC 0, Fig. 7).

The CACO experiment [29] yields the intraresidue C0i–Ca

i

correlation (n correlations for a protein ofn aminoacids) as well as

the correlations of carbonyl/carboxylate spins and the attached

carbon of Asp/Asn ðCbi –C

gi Þ, Glu/Gln ðC

gi –Cd

i Þ. The C0i–Ca

i

correlation can be detected through a variety of different schemes,

by acquiring the in-phase or the anti-phase C 0 components [28]

and eventually using signal processing algorithms to deconvolute

the splitting [33,47], by using one of the techniques described in

Section 2 to eliminate the JC0Ca coupling in the direct acquisition

dimension (BH [29], IPAP [21], S3E [21], COCAINE [48]), by

evolving single-quantum or double-quantum coherences (or

both). Some of the different experimental schemes used to detect

the C0i–Ca

i correlation are outlined in Fig. 8.

The next experiment to consider is the CBCACO sequence

[21] (Appendix, panel 1 for the IPAP version and panel 2 for

the S3E version). This exploits the Ca–Cb coupling (JCaCb)

through in COSY-type approach and yields for each residue, in

addition to the C0i–Ca

i , also the C0i–Cb

i correlation. The Cb

chemical shift dispersion, which is much larger than that of Ca,

allows one to reduce the overlap and to identify the aminoacid

on the basis of the Cb chemical shift. The remaining carbons of

the side chains of non-aromatic residues can be identified

through the CCCO experiment [58] (Appendix, panels 3 and 4

for the IPAP and S3E versions, respectively) where C 0 is

correlated, through Ca and by means of an additional TOCSY

step, to the rest of the side chain. By changing the spin-lock

time, the relative intensities of the correlations observed can be

modulated to detect all expected correlations. It is worth noting

that the analysis of these experiments also yields in a

straightforward way the assignment of acidic and amidic

side-chain resonances as each aliphatic carbon is correlated to

the backbone carbonyl and to the side chain carbonyl/

carboxylate nuclei, making the assignment procedure very

simple. When overlap is severe, the CBCACO and the CCCO

experiments can be extended in a third dimension by evolving

the chemical shift of Ca to resolve ambiguities due to C 0

resonance overlap [58]. For aromatic residues, the correlation

with the aromatic ring carbons ðCbi –C

gi Þ can be detected

y

PFG

13C´

y 13C

2 2

decoupling.D1H/15N

2

2

decoupling.

BH

13C

13C´

2-t

2-t

1H15N

A

B

C

D

E

Dec..

BH

FID

FID

PFG

13C´

y

13C 2 2

decoupling.D1H/15N

IP

2

2

AP

13C´

13C 4 44 4

decoupling.

FID

PFG

13C´

13C 2 2

decoupling.D1

1 1

H/15N

4

4

decoupling.FID

PFG

13C´

y

13C 2 2

decoupling.D decoupling.D1H/15N

FID

3

φ1 φ2

φ43φS3E φ φrec

φ3

φ1 φ21α/β

α/β

∆− t1∆− t ∆ ∆

φ1 1 1

φ2

3φ

∆ ∆ φrec

φrec

φIPAP

φ11 1

φ2∆+ t ∆−t

φIPAP

α/β

α/β

∆

∆ ∆

∆∆ ∆

φrec

∆+t ∆− t ∆ ∆

α/β

α/β

φ11 1

φ2∆+t ∆− t

φrecφ

Fig. 8. Some of the different experiments to detect the C 0–Ca correlation: (A)

with multiple quantum evolution during the indirect dimension and

bandselective homodecoupling; (B–D) with single quantum evolution during

the indirect dimension and various approaches to remove the Ca–C 0 splitting

(BH, IPAP, and S3E, respectively); and (E) with detection of the anti-phase

component. Band selective 13C pulses are denoted by shapes (narrow and wide

ones represent 90 and 1808 pulses, respectively). Two different panels indicate

the variants necessary to separate the in-phase and anti-phase components for

the CACO–IPAP experiment. Experiments were tested on different spec-

trometers. Shaped pulses, 1H and 15N decouplings, position of carriers and data

treatment for IPAP and S3E experiments have been described in the caption of

Fig. 5. The delay D is 1/(2JC 0Ca) (9 ms). In (B) and (E) the delay D can be

shortened to reduce relaxation losses with paramagnetic or large systems at the

expense of transfer efficiency. The phases are: (A) f1Zx,Kx; f2Zx; f3Z2(x),

2(Kx); frecZx, 2(Kx), x; (B) f1Zx,Kx; f2Z4(x), 4(y); f3Z2(x), 2(Kx);

frecZx, 2(Kx), x,Kx, 2(i),Kx; (C) f1Zx,Kx; f2Z4(x), 4(y); fIPAP(IP)Z2(x), 2(Kx); fIPAP(AP)Z2(Ky), 2(y); frecZx, 2(Kx), x,Kx, 2(x),Kx; (D)

f1Zx,Kx; f2Z4(x), 4(y); fS3E (1)Z4(458); fS3E (2)Z2(458), 2(2258); f3Z2(x), 2(y); f4Z2(x), 2(y); frecZx, 2(Kx), x,Kx, 2(x),Kx; (E) f1Zx,Kx; f2Z4(x), 4(y); f3Z2(x), 2(Kx); frecZx, 2(Kx), x,Kx, 2(x),Kx.


through the CGCB–DIPAP experiment that employs the

DIPAP approach for Cb spins to eliminate the splitting due to

coupling with Cg and Ca spins [50] (Appendix, panel 5).

The identification of 13C nuclei within each aminoacid is

thus complete. Correlation with backbone nitrogen nuclei is

performed using the CON-IPAP experiment [21,58] through

the one bond C 0–N coupling (Appendix, panel 6). The

experiment yields the sequential C0i–NiC1 correlation (nK1

correlations for a protein of n aminoacids) and the correlations

for side chains of Asn ðCbi –N

gi Þ and Gln ðC

gi –Nd

i Þ residues. This

experiment is quite sensitive (the most sensitive one involving

N nuclei) and characterized by a very good chemical shift

dispersion for both N and C 0 (Fig. 1). Therefore, it can be

combined with the other most sensitive blocks (CBCACO,

CCCO) to yield 3D experiments (CBCACON-IPAP, CCCON-

IPAP) [58] with C 0 in the direct acquisition dimension, Ca (or

Ca/b, or Ca/b/g.) in the second dimension and N in the third

(Appendix, panels 7 and 8).

With the above mentioned set of experiments, complete

identification of spin-systems can be achieved and each residue

can be correlated with the following backbone amide nitrogen.

However, in order to perform complete sequence-specific

assignment an additional correlation through the backbone is

necessary. This correlation can be detected by using the

CANCO-IPAP experiment [26,58]. In this experiment (Appen-

dix, panel 9) magnetization is transferred from Cai and Ca

iC1 to

NiC1 and then to C0i giving the two correlations (Ca

i NiC1–C0i and

CaiC1–NiC1–C0

i) necessary for sequence specific assignment

(Fig. 7). Actually the experiment is optimised to detect the

CaiC1–NiC1–C0

i peak, which is the ‘sequential one’, with most

sensitivity [26]. Two available variants of the experiment allow

one to discriminate between the two correlations (Appendix,

panel 10, for the most sensitive) [27]. A further variant of the

CANCO-IPAP experiment includes the transfer to the Cb,

allowing the exploitation of Cb chemical shift information as

an aid for spin-system identification (Appendix, panel 11).

Therefore, complete sequence specific assignment using

protonless NMR can be achieved through C 0 direct-detection

[58]. An example of the sequence-specific assignment strategy

employing these experiments is shown in Fig. 9. The 3D

experiments are necessary when dealing with large systems in

order to increase the resolution.

As mentioned in the previous paragraphs, methods used to

collapse the large one-bond couplings in the direct acquisition

dimension for Ca spins allow the design of experiments based

on Ca direct-detection [21]. Indeed the C0i–Ca

i correlation can

also be detected in the COCA experiment employing the

DIPAP approach in the acquisition dimension (Appendix,

panel 12). The Ca-TOCSY [21], a TOCSY experiment where

DIPAP has been implemented prior to acquisition (Appendix,

panel 13), yields the correlations of Cai with all the aliphatic

13C nuclei in the side chain (Ca=b=g.i ), providing an additional

way to assign all 13C spins in non-aromatic aminoacid side-

chains. Sequential correlations can be detected through the

CAN-DIPAP experiment in which each Cai is correlated to the

one and two bonds distant backbone nitrogen nucleus (Cai –Ni,

CaiC1–Ni) (Appendix, panel 14).

Fig. 9. The assignment strategy based only on 13C direct-detection 3D experiments is shown for a fragment Gly 73-Pro 74-Lys 75-Asp 76 in 13C,15N labeled

monomeric SOD. The portions of Ca/ali–C 0 planes of 3D spectra reported in the Fig. are taken from the following experiments recorded at 14.1 T: (A) 3D CANCO–

IPAP, (B) 3D CBCACO–IPAP, (C) 3D CCCO–IPAP, (D) 3D CBCACON–IPAP, (E) 3D CCCON–IPAP. For each residue, the figure shows in panel A the region of

the Ca–C 0 plane of 3D CANCO at the Ni chemical shift (129.2 ppm for Pro 74, 115.8 ppm for Lys 75), in panels B and D a portion of the Cali–C 0 plane of the

CBCACO and CCCO at the Cai chemical shift (63.7 ppm for Pro 74, 54.9 ppm for Lys 75), and in panels C and E the portions of the Cali–C 0 plane of the CBCACON

and CCCON at the NiC1 chemical shift (115.8 ppm for Lys 75, 121.5 ppm for Asp 76). All experiments were acquired using the IPAP approach that allows one to

remove the effect of the large C 0–Ca scalar coupling in the direct dimension.


The relative performance of experiments based on C 0 and Ca

direct-detection, once the problem of the homonuclear coupling

in the direct acquisition dimension is solved, strongly depends on

the relaxation properties of the system, both in terms of

longitudinal and transverse relaxation and on the chemical shift

dispersion. Indeed, for medium sized proteins at intermediate

fields and for non-2H labelled samples, the C0-based acquisition

experiments are the preferred choice as transverse relaxation rates

of C 0 are still tolerable. Actually these can be the preferred

choice for unfolded systems to exploit the residual chemical

shift dispersion of 13C 0, especially in combination with that of 15N

[59]. At high fields, with 2H labelled proteins, the relaxation

properties of Ca and of aliphatic nuclei in general are favourable

and thus experiments based on Ca detection offer valuable

alternatives. Indeed, by focusing mainly on Ca or on other

aliphatic 13C nuclei, experiments can be designed to minimize

relaxation losses due to C 0 magnetization in the plane.

A combination of the above-mentioned experiments, namely

those based on C 0 and on Ca direct-detection, can also be used to

provide redundant information useful in studying systems

characterized by crowded spectra.

A comment is due about alternative possibilities to the

CANCO-IPAP [58] and CAN-DIPAP experiments used to obtain

the sequential correlations in the protonless NMR sequence

specific assignment strategy. Indeed in both cases described, the

key correlations used to link aminoacid spin systems in a

sequence specific manner are provided by the Ca–N scalar

couplings. However, even if much smaller, there are also

homonuclear C 0–C0 (about 3 Hz) and Ca–Ca (about 2 Hz)

coupling constants that can be exploited for sequence specific

assignment by correlating a C 0 or a Ca of a specific residue with

the previous and with the following one through a spin-lock on a

very narrow bandwidth (C0 or Ca) that thus requires a fairly low

RF power [60].

The exploitation of 13C–13C cross-relaxation rates has been

proposed as an alternative approach to detect correlations

between directly bound 13C nuclei for large molecules using

inverse-detected experiments [61]. With the increase in

molecular mass and with the increase in percentage of 2H

isotopic enrichment, it is actually more convenient to opt for

experiments that start and end on 13C [62]. Indeed, the cross

relaxation rate sCC increases, while the longitudinal relaxation

rates decrease with increasing molecular mass. This makes

dipolar-coupling based transfer competitive with scalar-

coupling based transfer [61].

IPAP and DIPAP versions of the NOESY experiment

have been implemented to increase resolution by removing

complex multiplet structures in C 0 [43] and in Ca regions,

respectively [63] (Appendix, panels 15 and 16). The one-

bond correlations are indeed observed with good sensitivity.

With long NOESY mixing times, two and three bonds

correlations have also been identified and explained by spin-

diffusion [43]. This effect can be exploited as an alternative

approach used to extend the assignment to side-chains in

large macromolecules. The identification of 13C–13C corre-

lations in NOESY spectra between nuclei not directly bound

and not mediated by spin diffusion would mean a break-

through in solution structure determination of large

macromolecules by providing distance constraints. An

extra leap in sensitivity for 13C is required to be able to


detect truly long-range correlations to obtain 13C–13C

distance constraints.

Summarizing, the protonless NMR strategy in biomole-

cules, that was pioneered by Markley in 1988 [25,64,65], has

been expanded to a large set of multidimentional experiments

characterized by high sensitivity and simplified 13C patterns in

the acquisition dimension thanks to spin-state selective

methods and to homodecoupling.

A hybrid approach, that combines 13C direct-detection with

starting on 1H has been proposed by Dotsch and co-workers

[53]. On the grounds that 13C has several advantages they

proposed to ‘cut’ out-and-back versions of existing triple

resonance experiments into ‘out-and-stay’ versions where the

nuclei used to acquire are either C 0 (as in the example of

HACACO [49]) or aliphatic nuclei (as in the case of HCC

[49]). Improved versions of these experiments were later

proposed by Pervushin and co-workers [66,67]. These

additional experiments can also be added to the conventional

triple-resonance assignment strategy. However, the drawback

is the re-introduction of 1H transverse relaxation in one of the

building blocks of the sequences.

5. Detection of resonances in paramagnetic proteins

Paramagnetic systems are characterized by additional

contributions to chemical shifts and nuclear relaxation

arising from the so-called hyperfine interaction, i.e. the

interaction between the nuclear spin I and the electron spin

S [68]. This interaction occurs via two different mechanisms,

a through-bond interaction, which depends on the amount of

unpaired electron spin-density delocalized onto the investi-

gated nuclear spin (contact interaction) [69,70] and a

through-space interaction [71,72], which depends on the

dipole–dipole interaction between nuclear and electron spins.

Both effects contribute to chemical shifts and nuclear

relaxation. The interplay between the two different contri-

butions and their absolute and relative magnitudes depends

Table 2

Expected behaviour of different metal ions in terms of paramagnetism induced rela

Metal ion S, J g, gJ ts (10–12 s)

Fe(III) HS 5/2 2 100

Fe(III) LS 1/2 2 1

Fe(II) 2 2 1

Co(II) HStetra 3/2 2 10

Co(II) HSesa 3/2 2 1

Cu(II) 1/2 2 3000

Gd(III) 7/2 2 10000

Ce(III) 5/2 6/7 0.1

Dy(III) 15/2 4/3 0.5

Tb(III) 6 3/2 0.3

Yb(III) 7/2 8/7 0.3

Tm(III) 6 7/6 0.5

Distances reported in columns 5 and 6 are the calculated distances corresponding to

contributions are taken into consideration, so the calculated values do not include

by through-bond effects. Column 7 reports the calculated R1 values corresponding t

10K8 s, TZ300 K and B0Z16.4 T (700 MHz for 1H, or 176 MHz for 13C). Electro

very much on the nature of the paramagnet and on its

coordination sphere. The peculiarities of NMR spectroscopy

of paramagnetic systems have been extensively discussed in

the literature [73–76] and do not need to be re-introduced in

this review. The essential feature that concerns us here is the

signal broadening caused by paramagnetism-induced relax-

ation. Thus, the NMR signals of nuclei in the surrounding of

the paramagnetic centre can be broadened beyond detectable

limits.

Independently of the mechanism that it originates from,

paramagnetic relaxation depends on the square of the

magnetogyric ratio. This is not the case for paramagnetism-

induced chemical shifts. Therefore, 13C NMR studies of

paramagnetic molecules provide in principle the same

content of information as 1H studies, but the loss of

information due to paramagnetism-induced line broadening

will be much less effective. Actually, the applications on

paramagnetic systems were the ones that promoted the

revival of heteronuclear NMR in the recent years [75,77–81].

The keynote finding is that pseudocontact shifts and

hyperfine relaxation provide metal-nucleus distance restraints

[76]. These two observables, based on 13C nuclei, may

actually provide structural information in regions of a protein

where 1H resonances are broadened beyond detectable limits.

Therefore, synergism between 1H and 13C NMR spec-

troscopy concerns not only the assignments, but also

obtaining the structural constraints.

Table 2 gives the contribution to relaxation rates due to the

hyperfine interaction for 1H and 13C spins in the presence of

different metal chromophores. It can be noted that the different

metal ions present a variety of situations. While for some metal

ions like Gd(III) or Dy(III), 13C signals may be still broad

beyond detection in a sphere as large as 8 A from the metal ion,

in other cases, like low spin Fe(III) or Ce(III) and Co(II), all13C signals can in principle be detected. In the latter case,

nuclei that are 2–3 s bonds away from the metal centre can also

be affected by the contact contribution to relaxation and thus

xation

d (A) R21H at

1000 sK1

d (A) R213C at

1000 sK1

13C R1 sK1

7.5 4.8 430

3.0 1.9 130

6.0 3.7 30

5.2 3.3 190

5.1 3.2 60

7.0 4.5 170

13.3 8.4 45

3.8 2.4 29

9.9 6.3 15

9.3 5.9 15

5.6 3.6 25

7.9 5.0 20

an R2 value of 1000 sK1 for 1H and 13C, respectively. Only dipolar and Curie

the contributions provided by the contact interaction, i.e. the effects induced

o the distances of column 6. The values have been calculated considering trZnic properties of the various metal ions are also reported.

Fig. 10. (A) CACO–IPAP and (B) CACO–AP spectra of 13C,15N labeled

CaTmCb recorded at 16.4 T. The additional peaks observed in (B) are indicated

by arrows. The CACO–IPAP experiment has been recorded using standard

experimental conditions for the detection of diamagnetic resonances using the

pulse sequence shown in Fig. 8C. The CACO–AP spectrum was acquired using

the pulse sequence reported in Fig. 8E, with Ca–C 0 transfer delays shortened

from 9.0 to 5.5 ms and a recycle period shortened from 1.0 to 0.3 s. Both

spectra were acquired using a 13C-optimized triple-resonance probehead at

16.4 T and 300 K.


may become undetectable (this contribution is not included in

the calculations as it depends on the molecular structure and on

the electron spin delocalisation and thus is difficult to account

for in a general way [68,75]).

Direct-detection of 13C intrinsically offers a way to detect

resonances close to the metal ion where 1H resonances are too

broad to be detected. However, the 13C-based approach can be

further optimised for paramagnetic systems by selecting the

most efficient coherence transfer pathways and identifying

those that are least affected by fast relaxation. In principle, the

most sensitive experiments are those based on coherence

transfer mechanisms mediated by large scalar couplings;

however, when transverse relaxation is much faster than

longitudinal relaxation, dipolar-based transfers can be usefully

exploited, especially at higher fields.

The 13C–13C COSY-based experiments were the first to be

used for paramagnetic systems [77–79,81,82]. As an alterna-

tive, several other experiments based on coherence transfer via

the C 0–Ca scalar couplings, and also via the smaller N–C 0

scalar coupling, exploiting either the single quantum or

multiple quantum coherence transfer can be used [29,80,81].

Signal losses due to fast relaxation can be reduced by

shortening the coherence transfer delays [29] and even further

by completely removing the building block in which the anti-

phase C 0–Ca coherence is refocused and detecting directly the

anti-phase component [28,83]. Several CACO schemes have

been compared and the most suitable experiment for

paramagnetic systems turned out to be the single quantum

CACO experiment used without refocusing prior to C 0

detection (CACO–AP) [83], outlined in Fig. 8E. Fig. 10

shows the comparison of a standard CACO–IPAP experiment

vs the single-quantum CACO–AP experiment for the Tm(III)

substituted calcium binding protein Calbindin D9k [84].

Several additional peaks were observed in the experiment

which was tailored for paramagnetic signals. This result is due

both to the different pulse sequence used and to the choice of

experimental parameters, such as optimized coherence-transfer

and recycle delays [84].

As pointed out for large molecules, the 13C–13C NOESY

experiment can be useful when the limiting factor consists in

fast transverse relaxation [61]. Indeed in many paramagnetic

systems the longitudinal relaxation rates are influenced to a

smaller extent than the transverse relaxation rates. Table 2,

column 6, reports the longitudinal relaxation rate for a given13C resonance with R2Z1000 sK1. The large variability of the

values in column 6 provides an estimate of the potential interest

in the use of homonuclear 13C–13C NOE-based transfer. For

example, all lanthanides, including Gd(III) but especially the

more far-shifting ones like Dy(III) or Tb(III), have a relatively

small contribution to R1 arising from paramagnetic relaxation.

For the above cases, the use of 13C–13C NOE can be a useful

alternative to overcome the quench of scalar coupling based

transfer. Of course, this will be particularly true on increasing

the size of the molecule, as the magnitude of the NOE effect

depends on tr.

An interesting application of using 13C direct-detection to

get closer to a paramagnetic centre is for those metalloproteins

in which the metal ion is coordinated via carboxylate ligands.

Indeed, for bound Asp or Glu residues, the carboxyl carbon

atom is only two bonds away from the metal centre, while the

closest 1H spin is four bonds away. Therefore, 13C–13C COSY

has been successfully used in lanthanide substituted calcium

binding proteins [78]. Furthermore, the intrinsic asymmetry of

an homonuclear 13C–13C COSY experiment provides a clever

way of identifying the coordinating residues of a paramagnetic

metal ion and to discriminate between monodentate and

bidentate carboxylate ligands [85]. It is the only NMR method

available to distinguish between monodentate and bidentate

coordinating side-chain CO groups. Finally, lanthanide

substituted calcium binding proteins can also be used as

examples to monitor the different performances of various type

of experiments. This is graphically summarized in Fig. 11 for

the Tm(III) substituted derivative of calbindin D9k [84]. The

use of low g nuclei direct-detection provided the identification

of all residues while only about 50% of aminoacids were

Fig. 11. Radii of the spheres within which signals cannot be observed in the case of

CaTmCb. Different spheres are reported for each different set of NMR

experiments. Radii are as follows: 17.5 A for standard 1H detected NMR

experiments, 15 A for 1H experiments optimized to fast relaxing signals, 14 A for

standard 13C detected experiments, 9 A for 13C detected experiments optimized to

fast relaxing signals, 5 A from 13C 1D spectra, 4.2 A for 1D 15N spectra.


assigned with standard 1H detected experiments. In this case,15N 1D spectra were also used to identify previously

unobserved resonances. Indeed, the development of 15N

optimized probes, characterized by a 15N selective inner coil,

would certainly contribute to increase the S/N achievable for

this nucleus, thus providing better quality 15N direct-detected

experiments.

6. Conclusions

We have shown that the combination of upgraded

instrumental performances and the design of novel pulse

sequences allows 13C direct-detected protonless NMR spec-

troscopy to be used to conveniently perform the complete13C,15N assignment for proteins. Development of the exclu-

sively heteronuclear NMR experiments summarized in this

review has been stimulated by the need for a different approach

to study paramagnetic systems, for which the main problems

are fast transverse relaxation and difficulties with 1H detection.

These two features, however, are common to large proteins, as

well as to exchanging and partially unfolded systems.

Heteronuclear protonless NMR can thus be useful for making

assignments in a wide range of applications.

When one aims at extending this strategy to protein

structural determination, 13C-based restraints are needed.

The present state of the art allows one to fully exploit chemical

shift values [86] as Ca, Cb, and CO chemical shifts are

available from the present approach. Of course, the wealth of13C data available can be further exploited as a source of 13C-

chemical shift structural restraints.13C–13C residual dipolar couplings can in principle be

accurately determined through IPAP-type experiments and

therefore residual dipolar coupling restraints may be

available. Cross-correlation rates and J-values can also be

determined using 13C direct-detection experiments, providing

some dihedral angle restraints. Selected protonated amino

acids can provide 1H–1H NOEs, which are most valuable

within this strategy [11]. Of course, the measurement of

long-range 13C–13C NOEs would be a significant

breakthrough.

In the case of paramagnetic systems, nuclear relaxation

can be related to metal-nucleus distances [71]. This

approach has already been exploited for 13C nuclei [81].

However, reports are available in the literature, which point

to a break-down in the relaxation–distance relationship for

heteronuclei far from the metal ion [87,88], presumably due

to the occurrence of other phenomena not yet fully

understood. Paramagnetic probes and spin-labels were

recently used to exploit the paramagnetism-derived structural

information in natively diamagnetic proteins [89–97]. The

availability of further data via protonless NMR may provide

additional information at shorter distances than those

obtainable through 1H NMR.

Pseudocontact shifts are also very valuable restraints [30].

Here a problem can be the determination of the diamagnetic

reference to be subtracted from the paramagnetic chemical

shift. One should check that the structural features of the

two systems are absolutely the same, as the diamagnetic

chemical shift of carbon nuclei is highly sensitive to

dihedral angle variations. Another paramagnetic-based

restraint is the cross-correlation between Curie-relaxation

and 13C–13C dipolar relaxation. Further studies are needed to

implement their use. The combined use of paramagnetic and

diamagnetic-based restraints is expected to provide a good

method for determination of the structure of proteins in

solution.

Technological advancements in 13C solid-state spec-

troscopy, where 13C is the nucleus of detection, have been

beneficial to the development of direct-detection of hetero-

nuclei in solution and it is likely that the present development

will have a fall-out in solid-state NMR. Optimization of the

hardware and the development of dedicated cryogenically-

cooled probeheads may further establish heteronuclear pro-

tonless NMR as a routine spectroscopic approach for studying

biomolecules in solution.

Acknowledgements

This work has been supported in part by the EC (Contract

QLG2-CT-2002-00988) and by the Italian Ministero per la

Universita e la Ricerca (COFIN 2003).


Appendix

The pulse sequences used to acquire the experiments discussed in the review are described in detail here. In particular we

report the experiments based on C 0 direct-detection with IPAP and/or S3E spin state selective methods to remove the effects

of the C 0–Ca coupling in the direct acquisition dimension (CBCACO–IPAP, CBCACO–S3E, CCCO–IPAP, CCCO–S3E,

CBCACON–IPAP, CCCON–IPAP, CON–IPAP, CANCO–IPAP, CANCO–IPAP selective) and the experiments based on Ca

direct-detection with the DIPAP approach to remove the C 0–Ca and Ca–Cb couplings in the direct acquisition dimension

(COCA–DIPAP, Ca–TOCSY–DIPAP, CAN–DIPAP). We also report the NOESY experiment with implementation of spin-

state selective approaches to remove the one bond splittings in the direct acquisition dimension (CACO–NOESY–IPAP and

Ca–NOESY–DIPAP) and the CGCB–DIPAP experiment for connecting the aromatic side-chains to the backbone. These

experiments were tested using different instruments and on several proteins. We report here the parameters used with a 14.1 T

Bruker Avance instrument, equipped with a cryogenically-cooled probehead optimized for 13C sensitivity, on 13C,15N labelled

reduced monomeric superoxide dismutase. When acquiring experiments on a deuterated protein, 2H decoupling should be

applied as indicated for 1H. In all the figures (unless otherwise specified), band selective 13C pulses are denoted by shapes.

For 13C bandselective 908 and 1808 pulses Q5 (or time reversed Q5) and Q3 shapes [100] were used with durations of 320

and 256 ms, respectively except for the 1808 pulse indicated in grey (Q3, 1.0 ms) and for the 1808 pulse indicated in crossed

stripes (adiabatic inversion pulse over the C 0 and Ca regions, smoothed chirp 500 ms, sweep width 60 KHz, 20% smoothing

[103]). The rectangular wide and narrow pulses correspond to 180 and 908 hard pulses. Pulse field gradients (PFG line) are

also indicated by shapes. All the gradients, used for purging and not for coherence selection, have a duration of 1.0 ms and a

sine-shape. The 1H and 15N carriers were placed at 4.7 and 118 ppm, respectively. The change in the position of the 13C

carrier (39 ppm for Cali, 55 ppm for Ca and 173 ppm for C 0) is indicated by vertical arrows. The RF power used for the 13C

FLOPSY16 spin-lock was 10 kHz (applied for durations ranging from 10 to 22 ms in the 2D versions and 22 ms in the 3D

version). Decoupling of 1H and 15N was applied with 2.9 kHz (waltz-16) [101] and 1.0 kHz (garp-4)[102] respectively. For

experiments that employ the IPAP approach to suppress the C 0–Ca coupling, the in-phase (IP) and anti-phase (AP)

components are acquired and stored separately using the pulse schemes illustrated that differ only for the two panels indicated

with IP and AP respectively. For experiments that employ the S3E approach to suppress the effect of the C 0–Ca coupling, the

two components that need to be acquired and stored separately differ by the phase fS3E and by a p increment of phase f4 in

the CCCO-S3E experiment and f5 in the CBCACO-S3E experiment. For experiments that employ the DIPAP approach to

suppress the C 0–Ca and the Ca–Cb couplings, the four variants of the experiment that should be acquired and stored

separately are shown in the four panels indicated with IP-IP, AP-IP, IP-AP, AP-AP respectively. The phase cycle, the method

used for quadrature detection and the durations of the delays shown in the pulse sequences are reported case-by-case.

Panel 1

CBCACO–IPAP: The delays are:DZ9 ms,D1Z8 ms. The phase cycle is:f1Zx,Kx;f2Z8x,8(Kx);f3Z2y,2(Ky);fIPAP(IP)Z4(x),

4(Kx); fIPAP(AP)Z4(Ky),4(y); frecZx,(Kx),(Kx),x,(Kx),x,x,(Kx). Quadrature detection in the F1 and F2 dimensions is obtained by

incrementing f1 and f3 in a States-TPPI manner.

Transfer pathway: F1(Ca/b, t1)/F1(Ca, t2)/F1(C 0,t3)

Correlations observed : Cbi –Ca

i –C 0i ; Ca

i –Cai –C 0

i

Panel 2

CBCACO–S3E; The delays are: DZ9 ms, D1Z8 ms and 3Z4 ms. The phase cycle is: f1Zx,Kx; f2Z8x, 8(Kx); f3Z2y, 2(Ky); f4Z4x, 4(y); f5Z4x, 4(y); fS3E(1)Z4(458), 4(2258); fS3E(2)Z8(458); frecZx, (Kx), (Kx), x, (Kx), x, x, (Kx). Quadrature

detection in the F1 and F2 dimensions is obtained by incrementing f1 and f3 in a States-TPPI manner.


Transfer pathway: F1(Ca/b, t1)/F1(Ca, t2)/F1(C 0,t3)


i –C 0i ; Ca

i –Cai –C 0

i

Panel 3

CCCO–IPAP; The delays are: DZ9 ms, 3Zt1(0). The phase cycle is: f1Zx,Kx; f2Z2x, 2(Kx); fIPAP(IP)Z4x, 4(Kx);

fIPAP(AP)Z4(Ky),4y; frecZx, (Kx), (Kx), x, (Kx), x, x, (Kx). Quadrature detection in the F1 and F2 dimensions is obtained by

incrementing f1 and f2, respectively, in a States-TPPI manner.

Transfer pathway: F1(Cali, t1)/F1(Ca, t2)/F1(C 0,t3)

Correlations observed : Calii –Ca

i –C 0i ; Ca

i –Cai –C 0

i

Panel 4

CCCO–S3E: The delays are: DZ9 ms, 31Zt1(0) and 3Z4 ms. The phase cycle is: f1Zx,Kx; f2Z4x, 4(Kx); f3Z2x, 2(y); f4Z2x, 2(y); fS3E(1)Z4(458); fS3E(2)Z2(458), 2(2258); frecZx, (Kx), (Kx), x, (Kx), x, x, (Kx). Quadrature detection in the F1 and F2

dimensions is obtained by incrementing f1 and f2, respectively, in a States-TPPI manner.

Transfer pathway: F1(Cali, t1)/F1(Ca, t2)/F1(C 0, t3)


i –C 0i ; Ca

i –Cai –C 0

i


Panel 5

CGCB–DIPAP: The experiment to correlate the Cb of aromatic aminoacids with the quaternary Cg carbon that employs

the DIPAP approach to obtain singlets in the direct acquisition dimension. Pulses on Cg and on Cb and Ca/b were centered

at 130, and 33 ppm, respectively. The phase cycle is: f1Zx,Kx; f2Z8(x), 8(y), 8(Kx), 8(Ky); f3Z4(y), 4(Ky); frecZx,

(Kx),(Kx), x,(Kx), x, x,(Kx), (Kx), x, x,(Kx), x,K(x),(Kx), x; fDIPAP(IP-IP)Z2(x), 2(Kx); fDIPAP(AP-IP, IP-AP)Z2(Ky), 2(y); fDIPAP(AP-AP)Z2(Kx), 2(x). Quadrature detection is achieved by incrementing f1 in a States-TPPI manner.1H and 15N decoupling is applied during all the experiment, except during the relaxation delay at 7 ppm in the first part

and at 3.5 ppm in the second part (1H) and at 197 ppm (15N). The duration of the delays is: DZ4 ms, D1Z14.4 ms, dZ9 ms.

Transfer pathway: F1(Cg, t1)/F1(Cb,t2)

Correlations observed : Cgi –Cb

i

Panel 6

CON–IPAP: The delays are: DZ9 ms, D1Z25 ms, 3Zt1(0). The phase cycle is: f1Zx,Kx; f2Z2x, 2(Kx); f3Z4x, 4(Kx);

fIPAP(IP)Zx; fIPAP(AP)ZKy; frecZx, (Kx), x, (Kx), (Kx), x, (Kx), x. Quadrature detection in the F1 dimension is obtained by

incrementing f1 in a States-TPPI manner.

Transfer pathway: F1(C 0)/F3(N, t1)/F1(C 0, t2)

Correlations observed : Ni–C0iK1

Panel 7

CBCACON–IPAP: The delays are: DZ9 ms, D1Z25 ms, D2Z8 ms and 3Zt2(0). The phase cycle is: f1Zx,Kx; f2ZKy;

f3Z2x, 2(Kx); f4Z4x, 4(Kx); f5Z8x, 8(Kx); fIPAP(IP)Zx; fIPAP(AP)ZKy; frecZ2(x, (Kx), (Kx), x), 2((Kx), x, x, (Kx))

. Quadrature detection in the F1 and F2 dimensions is obtained by incrementing f1 and f3, respectively, in a States-TPPI

manner.

Transfer pathway: F1(Ca/b, t1)/F1(Ca)/F1(C 0)/F3(N, t2)/F1(C 0, t3)


Correlations observed : Cbi –C 0

i–NiC1; Cai –C 0

i–NiC1

Panel 8

CCCON–IPAP: The delays are: DZ9 ms, D1Z25 ms, 31Zt1(0), 3Zt(0). The phase cycle is: f1Zx,Kx; f2Z2x, 2(Kx); f3Z4x, 4(Kx); f4Z8x, 8(Kx); fIPAP(IP)Zx; fIPAP(AP)ZKy; frecZ2(x, (Kx), (Kx), x), 2((Kx), x, x, (Kx)). Quadrature detection in

the F1 and F2 dimensions is obtained by incrementing f1 and f2, respectively, in a States-TPPI manner.

Transfer pathway: F1(Cali, t1)/F1(Ca)/F1(C 0)/F3(N, t2)/F1(C 0, t3)

Correlations observed : Calii –C 0

i–NiC1; Cai –C 0

i–NiC1

Panel 9

CANCO–IPAP: The delays are: DZ9 ms, D1Z25 ms, D2Z32 ms. The phase cycle is: f1Zx,Kx; f2Z8x, 8(Kx); f3Z2x,

2(Kx); fIPAP(IP)Z4x, 4(Kx); fIPAP(AP)Z4(Ky), 4y; frecZx, (Kx), (Kx), x, (Kx), x, x, (Kx). Quadrature detection in the F1

and F2 dimensions is obtained by incrementing f1 and f3, respectively, in a States-TPPI manner.

Transfer pathway: F1(Ca, t1)/F3(N, t2)/F1(C 0, t3)

Correlations observed : Cai –Ni–C

0iK1; Ca

iK1–NaiK1–C 0

iK1


Panel 10

CANCO–IPAP ‘selective’: The delays are: DZ9 ms, D1Z25 ms, D2Z32 ms and 3Zt2(0). The phase cycle is: f1Zx,Kx; f2Z8x, 8(Kx); f3Z2x, 2(Kx); fIPAP(IP)Z4x, 4(Kx); fIPAP(AP)Z4(y), 4(Ky); frecZx, (Kx), (Kx), x, (Kx), x, x, (Kx). Quadrature detection in the F1 and F2 dimensions is obtained by incrementing f1 and f3, respectively, in a States-TPPI

manner.

Transfer pathway: F1(Ca, t1)/F3(N, t2)/F1(C 0, t3)

Correlations observed : CaiK1–Ni–C

0iK1

Panel 11

CBCANCO–IPAP: The delays are: DZ9 ms, D0Z8 ms, D1Z25 ms, D2Z32 ms. The phase cycle is: f1Zx,Kx; f2Z8x,

8(Kx); f3Z2x,2(Kx); fIPAP(IP)Z4x,4(Kx); fIPAP(AP)Z4(Ky),4y; frecZx,(Kx),(Kx)x,(Kx),x,x,(Kx). Quadrature detection

in the F1 and F2 dimensions is obtained by incrementing f1 and f3, respectively, in a States-TPPI manner.

Transfer pathway: F1(Ca/b, t1)/F3(N, t2)/F1(C 0, t3)

Correlations observed : Cai –Ni–C

0iK1; Ca

iK1–Ni–C0iK1; Cb

i –Ni–C0iK1; Cb

iK1–Ni–C0iK1

Panel 12

COCA-DIPAP: The delays are: DZ9 ms, D1Z14.4 ms. The phase cycle is: f1Zx,Kx; f2Z4x, 4y; fDIPAP(IP-IP)Z2x, 2(Kx);

fDIPAP(AP-IP)ZfDIPAP(IP-AP)Z2(Ky), 2y; fDIPAP(AP-AP)Z2(Kx), 2x; frecZx, (Kx), (Kx), x, (Kx), x, x, (Kx). Quadrature

detection in the F1 dimension is obtained by incrementing f1 in a States-TPPI manner.

Transfer pathway: F1(C0

, t1)/F1(Ca, t3)

Correlations observed : C 0i–C

ai :


Panel 13

Ca–TOCSY–DIPAP: The delays are: DZ9 ms, D1Z14.4 ms, 3Zt1(0). The phase cycle is: f1Zx,Kx; f2Z4x, 4y; f3Z8x, 8(Kx);

fDIPAP(IP-IP)Z2x, 2(Kx); fDIPAP(IP-AP)Z2(Ky), 2y; fDIPAP(AP-IP)Z2y, 2(Ky); fDIPAP(AP-AP)Z2x, 2(Kx); frecZx, (Kx),

(Kx),x, (Kx),x, x, (Kx), (Kx),x, x, (Kx),x, (Kx), (Kx),x. Quadrature detection in the F1 dimension is obtained by incrementingf1 in a

States-TPPI manner.

The experiment is generally performed with all the pulses before the spin-lock, as well as the spin-lock itself, to cover the

aliphatic region. In this case, the experiment yields all the correlations of Ca with the remaining aliphatic carbon atoms within

the side chain of the aminoacid. However, the experiment can also be performed with all the pulses before the spin-lock, as well as

the spin-lock itself, that are selective over the Ca region and with a much longer duration of the spin-lock. In this case the

experiment yields Ca–Ca sequential correlations.

Transfer pathway: F1(Cali, t1)/F1(Ca, t3)


i

Panel 14

CAN–DIPAP: The delays are: DZ13.6 ms, dZ3.6 ms, xZ2.25 ms, 3Zt1(0). The phase cycle is: f1Zx,Kx; f2Z2x, 2(Kx);

fDIPAP(IP-IP)Z4x, 4(Kx); fDIPAP(AP-IP)Z4(Ky), 4y; fDIPAP(IP-AP)Z4y, 4(Ky); fDIPAP(AP-AP)Z4x, 4(Kx); frecZx,

(Kx), x, (Kx), (Kx), x, (Kx), x. Quadrature detection in the F1 dimension is obtained by incrementing f1 in a States-TPPI

manner.


Transfer pathway: F1(Ca)/F3(N, t1)/F1(Ca, t2)

Correlations observed : Cai –Ni; Ca

iK1–Ni

Panel 15

CACO–NOESY–IPAP: The delays are: DZ9 ms, 3Zt1(0), tMZmixing time. The phase cycle is: f1Zx,Kx; f2Z4x, 4(Kx);

fIPAP(IP)Z2x, 2(Kx); fIPAP(AP)Z2(Ky), 2y; frecZx, (Kx), (Kx), x, (Kx), x, x, (Kx). Quadrature detection in the F1 dimension

is obtained by incrementing f1 in a States-TPPI manner.

Transfer pathway: F1(Ca, t1)/F1(C0

, t2)

Correlations observed : Cai –C 0

i ; and other correlations arising from spin � diffusion

Panel 16

Ca–NOESY–DIPAP: The delays are: DZ9 ms, D1Z14.4 ms, 3Zt1(0), tMZmixing time. The phase cycle is: f1Zx,Kx; f2Z4x, 4(Kx); fDIPAP(IP-IP)Z2x, 2(Kx); fDIPAP(IP-AP)Z2(Ky), 2y; fDIPAP(AP-IP)Z2y, 2(Ky); fDIPAP(AP-AP)Z2x, 2(Kx);

frecZx, (Kx), (Kx), x, (Kx), x, x, (Kx). Quadrature detection in the F1 dimension is obtained by incrementing f1 in a States-TPPI

manner.

Transfer pathway: F1(Cb/ali, t1)/F1(Ca, t2).


i and other correlations arising from spin � diffusion


References

[1] K. Pervushin, R. Riek, G. Wider, K. Wuthrich, Proc. Natl Acad. Sci. USA

94 (1997) 12366.

[2] K. Pervushin, R. Riek, G. Wider, K. Wuthrich, J. Am. Chem. Soc. 120

(1998) 6394.

[3] V. Tugarinov, P.M. Hwang, J.E. Ollerenshaw, L.E. Kay, J. Am. Chem.

Soc. 125 (2003) 10420.

[4] E. Miclet, D.C. Williams Jr., G.M. Clore, D.L. Bryce, J. Boisbouvier,

A. Bax, J. Am. Chem. Soc. 126 (2004) 10560.

[5] D.M. LeMaster, F.M. Richards, Biochemistry 27 (1988) 142.

[6] M.A. Markus, K.T. Dayie, P. Matsudaira, G. Wagner, J. Magn. Reson.

Ser. B 105 (1994) 192.

[7] M. Sattler, S.W. Fesik, Structure 4 (1996) 1245.

[8] R.A. Venters, R. Thompson, J. Cavanagh, J. Mol. Struct. 602-603 (2002)

275.

[9] C.H. Arrowsmith, Y.S. Wu, Progr. NMR Spectrosc. 32 (1998) 277.

[10] L.Y. Lian, D.A. Middleton, Progr. NMR Spectrosc. 39 (2001) 171.

[11] V. Tugarinov, W.Y. Choy, V.Y. Orekhov, L.E. Kay, Proc. Natl Acad.

Sci. USA 102 (2005) 622.

[12] R.K. Harris, Nuclear Magnetic Resonance Spectroscopy, Longman,

London, 1986.

[13] J.C. Lindon, E. Holmes, J.K. Nicholson, Prog. NMR Spectrosc. 45 (2005)

109.

[14] H. Kovacs, D. Moskau, M. Spraul, Prog. NMR Spectrosc. 46 (2005)

131.

[15] E.R. Zuiderweg, Biochemistry 41 (2002) 1.

[16] S. Grzesiek, J. Anglister, H. Ren, A. Bax, J. Am. Chem. Soc. 115 (1993)

4369.

[17] J. Cavanagh, W.J. Fairbrother, A.G. Palmer III, N.J. Skelton, Protein

NMR Spectroscopy. Principles and Practice, Academic Press, San Diego,

1996.

[18] K.T. Dayie, G. Wagner, J. Am. Chem. Soc. 119 (1997) 7797.

[19] M. Pellecchia, Y. Pang, L. Wang, A.V. Kurochkin, A. Kumar,

E.R.P. Zuiderweg, J. Am. Chem. Soc. 121 (1999) 9165.

[20] L. Zeng, M.W.F. Fischer, E.R.P. Zuiderweg, J. Biomol. NMR 7 (1996)

157.

[21] W. Bermel, I. Bertini, L. Duma, L. Emsley, I.C. Felli, R. Pierattelli,

P. Vasos, Angew. Chem. Int. Ed. 44 (2005) 3089.

[22] M. Salzmann, K. Pervushin, G. Wider, H. Senn, K. Wuthrich, Proc. Natl

Acad. Sci. USA 95 (1998) 13585.

[23] R.R. Ernst, G. Bodenhausen, A. Wokaun, Principles of Nuclear Magnetic

Resonance in One and Two Dimensions, Oxford University Press,

London, 1987.

[24] A. Bax, G.M. Clore, P.C. Driscoll, A.M. Gronenborn, M. Ikura,

L.E. Kay, J. Magn. Reson. 87 (1990) 620.

[25] B.-H. Oh, W.M. Westler, P. Darba, J.L. Markley, Science 240 (1988)

908.

[26] I. Bertini, L. Duma, I.C. Felli, M. Fey, C. Luchinat, R. Pierattelli,

P.R. Vasos, Angew. Chem. Int. Ed. 43 (2004) 2257.

[27] W. Bermel, I. Bertini, I.C. Felli, R. Pierattelli, P.R. Vasos, J. Magn.

Reson. 172 (2005) 324.

[28] N. Shimba, H. Kovacs, A.S. Stern, A.M. Nomura, I. Shimada, J.C. Hoch,

C.S. Craik, V. Dotsch, J. Biomol. NMR 30 (2005) 175.

[29] W. Bermel, I. Bertini, I.C. Felli, R. Kummerle, R. Pierattelli, J. Am.

Chem. Soc. 125 (2003) 16423.

[30] E. Babini, I. Bertini, F. Capozzi, I.C. Felli, M. Lelli, C. Luchinat, J. Am.

Chem. Soc. 126 (2004) 10496.

[31] B. Vogeli, H. Kovacs, K. Pervushin, J. Biomol. NMR 31 (2005) 1.

[32] R. Tycko, A. Pines, J. Guckenheimer, J. Chem. Phys. 83 (1985)

2775.

[33] N. Shimba, A.S. Stern, C.S. Craik, J.C. Hoch, V. Dotsch, J. Am. Chem.

Soc. 125 (2003) 2382.

[34] K. Pervushin, A. Eletsky, J. Biomol. NMR 25 (2003) 147.

[35] F. Bloch, A. Siegert, Phys. Rev. 57 (1940) 522.

[36] J. Weigelt, A. Hammarstroem, W. Bermel, G. Otting, J. Magn. Reson.

Ser. B 110 (1996) 219.

[37] M. Ottiger, F. Delaglio, A. Bax, J. Magn. Reson. 131 (1998) 373.

[38] P. Andersson, J. Weigelt, G. Otting, J. Biomol. NMR 12 (1998) 435.

[39] L. Duma, S. Hediger, A. Lesage, L. Emsley, J. Magn. Reson. 164 (2003)

187.

[40] M.D. Sørensen, A. Meissner, O.W. Sørensen, J. Biomol. NMR 10 (1997)

181.

[41] M. Ottiger, F. Delaglio, J.L. Marquardt, N. Tjandra, A. Bax, J. Magn.

Reson. 134 (1998) 365.

[42] J.H. Prestegard, K.L. Mayer, H. Valafar, G.C. Benison, Methods

Enzymol. 394 (2005) 175.

[43] I. Bertini, I.C. Felli, R. Kummerle, C. Luchinat, R. Pierattelli, J. Biomol.

NMR 30 (2004) 245.

[44] N.C. Nielsen, H. Thøgersen, O.W. Sørensen, J. Am. Chem. Soc. 117

(1995) 11365.

[45] A. Meissner, J.O. Duus, O.W. Sørensen, J. Magn. Reson. 128 (1997)

92.

[46] L. Duma, S. Hediger, B. Brutscher, A. Bockmann, L. Emsley, J. Am.

Chem. Soc. 125 (2003) 11816.

[47] K. Hu, A. Eletsky, K. Pervushin, J. Biomol. NMR 26 (2003) 69.

[48] D. Lee, B. Vogeli, K. Pervushin, J. Biomol. NMR 31 (2005) 273.

[49] Z. Serber, C. Richter, V. Dotsch, ChemBioChem. 2 (2001) 247.

[50] E. Babini, I.C. Felli, M. Lelli, C. Luchinat, R. Pierattelli, J. Biomol.

NMR, 33 (2005) 137.

[51] M. Schubert, D. Labudde, D. Leitner, H. Oschkinat, P. Schmieder,

J. Biomol. NMR 31 (2005) 115.


[52] B. Vogeli, H. Kovacs, K. Pervushin, J. Am. Chem. Soc. 126 (2004)

2414.

[53] Z. Serber, C. Richter, D. Moskau, J.-M. Boehlen, T. Gerfin, D. Marek,

M. Haeberli, L. Baselgia, F. Laukien, A.S. Stern, J.C. Hoch, V. Dotsch,

J. Am. Chem. Soc. 122 (2000) 3554.

[54] M. Sattler, J. Schleucher, C. Griesinger, Progr. NMR Spectrosc. 34

(1999) 93.

[55] L.E. Kay, M. Ikura, R. Tschudin, A. Bax, J. Magn. Reson. 89 (1990) 496.

[56] S. Grzesiek, A. Bax, J. Magn. Reson. 99 (1992) 201.

[57] S. Grzesiek, A. Bax, J. Am. Chem. Soc. 114 (1992) 6291.

[58] W. Bermel, I. Bertini, I.C. Felli, R. Kummerle, R. Pierattelli, J. Magn.

Reson., in press.

[59] J. Yao, H.J. Dyson, P.E. Wright, FEBS Lett. 419 (1997) 285.

[60] S. Balayssac, B. Jimenez, M. Piccioli, unpublished results.

[61] M.W.F. Fischer, L. Zeng, E.R.P. Zuiderweg, J. Am. Chem. Soc. 118

(1996) 12457.

[62] I. Bertini, I.C. Felli, R. Kummerle, D. Moskau, R. Pierattelli, J. Am.

Chem. Soc. 126 (2004) 464.

[63] W. Bermel, I. Bertini, I.C. Felli, R. Kummerle, R. Pierattelli,

unpublished results.

[64] W.M. Westler, M. Kainosho, H. Nagao, M. Tomonaga, J.L. Markley,

J. Am. Chem. Soc. 110 (1988) 4093.

[65] W.M. Westler, B.J. Stockman, J.L. Markley, J. Am. Chem. Soc. 110

(1988) 6256.

[66] K. Hu, B. Vogeli, K. Pervushin, J. Magn. Reson. 174 (2005) 200.

[67] A. Eletsky, O. Moreira, H. Kovacs, K. Pervushin, J. Biomol. NMR 26

(2003) 167.

[68] I. Bertini, C. Luchinat, G. Parigi, Solution NMR of Paramagnetic

Molecules, Elsevier, Amsterdam, 2001.

[69] E. Fermi, Z. Phys. 60 (1930) 320.

[70] H.M. McConnell, J. Chem. Phys. 25 (1956) 709.

[71] I. Solomon, Phys. Rev. 99 (1955) 559.

[72] N. Bloembergen, E.M. Purcell, P.V. Pound, Phys. Rev. 73 (1948)

679.

[73] D.R. Eaton, R.D. Fischer, C.J. Hawkins, W.D. Horrocks Jr., J.P. Jesson,

R.W. Kreilick, R.J. Kurland, G.N. La Mar, C.H. Langford,

W.D. Phillips, L.H. Pignolet, M.F. Rettig, T.J. Swift, NMR of

Paramagnetic Molecules, Academic Press, New York, 1973.

[74] I. Bertini, C. Luchinat, M. Piccioli, Methods Enzymol. 339 (2001) 314.

[75] T.E. Machonkin, W.M. Westler, J.L. Markley, Inorg. Chem. 44 (2005)

779.

[76] I. Bertini, C. Luchinat, G. Parigi, R. Pierattelli, ChemBioChem. 6 (2005)

1536.

[77] U. Kolczak, J. Salgado, G. Siegal, M. Saraste, G.W. Canters,

Biospectroscopy 5 (1999) S19–S32.

[78] I. Bertini, Y.-M. Lee, C. Luchinat, M. Piccioli, L. Poggi, ChemBio-

Chem. 2 (2001) 550.

[79] T.E. Machonkin, W.M. Westler, J.L. Markley, J. Am. Chem. Soc. 124

(2002) 3204.

[80] M. Kostic, S.S. Pochapsky, T.C. Pochapsky, J. Am. Chem. Soc. 124

(2002) 9054.

[81] F. Arnesano, L. Banci, I. Bertini, I.C. Felli, C. Luchinat,

A.R. Thompsett, J. Am. Chem. Soc. 125 (2003) 7200.

[82] T.E. Machonkin, W.M. Westler, J.L. Markley, J. Am. Chem. Soc. 126

(2004) 5413.

[83] I. Bertini, B. Jimenez, M. Piccioli, J. Magn. Reson. 174 (2005) 125.

[84] S. Balayssac, B. Jimenez, M. Piccioli, J. Biomol. NMR, in press.

[85] I. Bertini, B. Jimenez, M. Piccioli, L. Poggi, J. Am. Chem. Soc. 127

(2005) 12216.

[86] D.S. Wishart, B.D. Sykes, J. Biomol. NMR 4 (1994) 171.

[87] L. Ma, A.M.M. Jørgensen, G.O. Sorensen, J. Ulstrup, J.J. Led, J. Am.

Chem. Soc. 122 (2000) 9473.

[88] S.J. Wilkens, B. Xia, B.F. Volkman, F. Weinhold, J.L. Markley,

W.M. Westler, J. Phys. Chem. B 102 (1998) 8300.

[89] V. Gaponenko, A. Dvoretsky, C. Walsby, B.M. Hoffman, P.R. Rosevear,

Biochemistry 39 (2000) 15217.

[90] L.W. Donaldson, N.R. Skrynnikov, W.-Y. Choy, D.R. Muhandiram,

B. Sarkar, J.D. Forman-Kay, L.E. Kay, J. Am. Chem. Soc. 123 (2001)

9843.

[91] V. Gaponenko, A.S. Altieri, J. Li, R.A. Byrd, J. Biomol. NMR 24 (2002)

143.

[92] J. Wohnert, K.J. Franz, M. Nitz, B. Imperiali, H. Schwalbe, J. Am.

Chem. Soc. 125 (2003) 13338.

[93] V. Gaponenko, S.P. Sarma, A.S. Altieri, D.A. Horita, J. Li, R.A. Byrd,

J. Biomol. NMR 28 (2004) 205.

[94] T. Ikegami, L. Verdier, P. Sakhaii, S. Grimme, P. Pescatore, K. Saxena,

K.M. Fiebig, C. Griesinger, J. Biomol. NMR 29 (2004) 339.

[95] M. Prudencio, J. Rohovec, J.A. Peters, E. Tocheva, M.J. Boulanger,

M.E. Murphy, H.J. Hupkes, W. Koster, A. Impagliazzo, M. Ubbink,

Chem. Eur. J. 5 (2004) 3252.

[96] M.R. Jensen, C. Lauritzen, S.W. Dahl, J. Pedersen, J.J. Led, J. Biomol.

NMR 29 (2004) 175.

[97] G. Pintacuda, A. Moshref, A. Leonchiks, A. Sharipo, G. Otting,

J. Biomol. NMR 29 (2004) 351.

[98] Y. Hiyama, C.H. Niu, J.V. Silverston, A. Bavoso, D.A. Torchia, J. Am.

Chem. Soc. 110 (1988) 2378.

[99] M. Tessari, H. Vis, R. Boelens, R. Kaptein, G.W. Vuister, J. Am. Chem.

Soc. 119 (1997) 8985.

[100] L. Emsley, G. Bodenhausen, J. Magn. Reson. 97 (1992) 135.

[101] A.J. Shaka, J. Keeler, R. Freeman, J. Magn. Reson. 53 (1983) 313.

[102] A.J. Shaka, P.B. Barker, R. Freeman, J. Magn. Reson. 64 (1985) 547.

[103] J.M. Boeheen, G. Bodenhausen, J. Magn. Reson. Ser A 102 (1993)

293.

13C-detected protonless NMR spectroscopy of proteins in solution

Documents