Dipolar chemical shift correlation spectroscopy for homonuclear carbon distance measurements in proteins in the solid state: application to structure determination and refinement.

Abstract

High-resolution solid-state NMR spectroscopy has become a promising tool for protein structure determination. Here, we describe a new dipolar-chemical shift correlation experiment for the measurement of homonuclear 13C-13C distances in uniformly 13C,15N-labeled proteins and demonstrate its suitability for protein structure determination and refinement. The experiments were carried out on the beta1 immunoglobulin binding domain of protein G (GB1). Both intraresidue and interresidue distances between carbonyl atoms and atoms in the aliphatic side chains were collected using a three-dimensional chemical shift correlation spectroscopy experiment that uses homogeneously broadened rotational resonance recoupling for carbon mixing. A steady-state approximation for the polarization transfer function was employed in data analysis, and a total of 100 intramolecular distances were determined, all in the range 2.5-5.5 A. An additional 41 dipolar contacts were detected, but the corresponding distances could not be accurately quantified. Additional distance and torsional restraints were derived from the proton-driven spin diffusion measurements and from the chemical shift analysis, respectively. Using all these restraints, it was possible to refine the structure of GB1 to a root-mean square deviation of 0.8 A. The approach is of general applicability for peptides and small proteins and can be easily incorporated into a structure determination and refinement protocol.

Full text

Page 1

Dipolar Chemical Shift Correlation Spectroscopy for
Homonuclear Carbon Distance Measurements in Proteins in
the Solid State: Application to Structure Determination and
Refinement
Xiaohu Peng,†,§David Libich,‡,§Rafal Janik,†,§George Harauz,‡,§and
Vladimir Ladizhansky*,†,§
Department of Physics, Department of Molecular and Cellular Biology, and Biophysics
Interdepartmental Group, UniVersity of Guelph, 50 Stone Road East, Guelph,
Ontario, Canada, N1G 2W1
Received September 4, 2007; E-mail: vladimir@physics.uoguelph.ca
Abstract: High-resolution solid-state NMR spectroscopy has become a promising tool for protein structure
determination. Here, we describe a new dipolar-chemical shift correlation experiment for the measurement
of homonuclear13C-13C distances in uniformly13C,15N-labeled proteins and demonstrate its suitability for
protein structure determination and refinement. The experiments were carried out on the ?1 immunoglobulin
binding domain of protein G (GB1). Both intraresidue and interresidue distances between carbonyl atoms
and atoms in the aliphatic side chains were collected using a three-dimensional chemical shift correlation
spectroscopy experiment that uses homogeneously broadened rotational resonance recoupling for carbon
mixing. A steady-state approximation for the polarization transfer function was employed in data analysis,
and a total of 100 intramolecular distances were determined, all in the range 2.5-5.5 Å. An additional 41
dipolar contacts were detected, but the corresponding distances could not be accurately quantified. Additional
distance and torsional restraints were derived from the proton-driven spin diffusion measurements and
from the chemical shift analysis, respectively. Using all these restraints, it was possible to refine the structure
of GB1 to a root-mean square deviation of 0.8 Å. The approach is of general applicability for peptides and
small proteins and can be easily incorporated into a structure determination and refinement protocol.
Introduction
Structural studies of many biological systems, including
membrane proteins and amyloid-forming peptides and proteins,
remain a challenge for X-ray crystallography and solution NMR,
owing to difficulties with crystallization and solubility. Solid-
state NMR (SSNMR) is an alternative approach to structure
determination in these types of systems.1-6Its key advantage
is the ability to obtain structural information from a wide range
of samples, from protein microcrystals, to amyloid fibrils, to
membrane proteins in their native-like lipidic environment.
Recent advances in sample preparation techniques7-9,10-12,13and
in high field instrumentation and pulse sequence design14-18
have led to the creation of new efficient methods for chemical
shift assignments and structural analysis of the proteins in the
solid state. The main source of structural restraints in SSNMR
is either13C-13C distances,5,19,20collected either using proton-
driven spin diffusion (PDSD)21or dipolar assisted recoupling
(DARR),22or interproton distance restraints, obtained through
carbon-detected proton-proton spin diffusion.23-25Both ap-
proaches are capable of establishing the relatively high-
†Department of Physics.
‡Department of Molecular and Cellular Biology.
§Biophysics Interdepartmental Group.
(1) Opella, S. J. Nat. Struct. Biol. 1997, 4, 845-848.
(2) Tycko, R. Curr. Opin. Struc. Biol. 2004, 14, 96-103.
(3) McDermott, A. E. Curr. Opin. Struc. Biol. 2004, 14, 554-561.
(4) Baldus, M. Curr. Opin. Struc. Biol. 2006, 16, 618-623.
(5) Castellani, F.; van Rossum, B.; Diehl, A.; Schubert, M.; Rehbein, K.;
Oschkinat, H. Nature 2002, 420, 98-102.
(6) Lange, A.; Giller, K.; Hornig, S.; Martin-Eauclaire, M. F.; Pongs, O.;
Becker, S.; Baldus, M. Nature 2006, 440, 959-962.
(7) Pauli, J.; van Rossum, B.; Forster, H.; de Groot, H. J. M.; Oschkinat, H. J.
Magn. Reson. 2000, 143, 411-416.
(8) Martin, R. W.; Zilm, K. W. J. Magn. Reson. 2003, 165, 162-174.
(9) Lorch, M.; Fahem, S.; Kaiser, C.; Weber, I.; Mason, A. J.; Bowie, J. U.;
Glaubitz, C. ChemBioChem 2005, 6, 1693-1700.
(10) Li, Y.; Berthold, D. A.; Frericks, H. L.; Gennis, R. B.; Rienstra, C. M.
ChemBioChem 2007, 8, 434-442.
(11) Hiller, M.; Krabben, L.; Vinothkumar, K. R.; Castellani, F.; van Rossum,
B. J.; Kuhlbrandt, W.; Oschkinat, H. ChemBioChem 2005, 6, 1679-1684.
(12) Kloepper, K. D.; Woods, W. S.; Winter, K. A.; George, J. M.; Rienstra, C.
M. Protein Expression Purif. 2006, 48, 112-117.
(13) Frericks, H. L.; Zhou, D. H.; Yap, L. L.; Gennis, R. B.; Rienstra, C. M. J.
Biomol. NMR 2006, 36, 55-71.
(14) Bennett, A. E.; Griffin, R. G.; Vega, S. Recoupling of homo- and
heteronuclear dipolar interactions in rotating solids. In Solid State NMR
IV: Methods and Applications of Solid-State NMR; Blumich, B., Ed.;
Springer-Verlag: Berlin, 1994; pp 1-77.
(15) Dusold, S.; Sebald, A. Annu. Rep. NMR Spectrosc. 2000, 41, 185-264.
(16) Tycko, R. Annu. ReV. Phys. Chem. 2001, 52, 575-606.
(17) Baldus, M. Prog. Nucl. Magn. Reson. Spectrosc. 2002, 41, 1-47.
(18) Hughes, C. E.; Baldus, M. Annu. Rep. NMR Spectrosc. 2005, 55, 121-
158.
(19) Castellani, F.; van Rossum, B. J.; Diehl, A.; Rehbein, K.; Oschkinat, H.
Biochemistry 2003, 42, 11476-11483.
(20) Zech, S. G.; Wand, A. J.; McDermott, A. E. J. Am. Chem. Soc. 2005, 127,
8618-8626.
(21) Szeverenyi, N. M.; Sullivan, M. J.; Maciel, G. E. J. Magn. Reson. 1982,
47, 462-475.
(22) Takegoshi, K.; Nakamura, S.; Terao, T. Chem. Phys. Lett 2001, 344, 631-
637.
Published on Web 12/12/2007
10.1021/ja076658v CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 359-369 9 359

Page 2

resolution fold of a protein.5,19,20,26Various structure refinement
methods have also been developed for use in proteins in the
solid state. For example, isotropic chemical shift indexing is
commonly used for the analysis of a secondary structure,27-29
while correlating the relative orientations of anisotropic tensors
providesinformationonlocalandmedium-rangestructures.30-36,37,38
More recently, the introduction of paramagnetic labels with the
purpose of obtaining paramagnetic relaxation enhancements,
originally introduced in solution NMR,39,40was also applied in
solids.41-43
One obvious way to establish precise structural constraints
in solids is to exploit the fact that molecular motions are
generally not sufficiently fast to provide an efficient mechanism
for averaging dipolar and chemical shift anisotropy interactions.
The anisotropic line broadening associated with the presence
of these interactions can be removed through the application of
magic angle spinning (MAS),44,45whereas the structural infor-
mation contained in these interactions can be reintroduced
through the application of radio frequency (rf) pulses.14,15Most
notably, heteronuclear rotational echo double resonance (RE-
DOR)46and frequency-selective homonuclear rotational reso-
nance (R2)47,48have been successfully used to accurately
measure distances between specifically placed nuclear spin
labels.49-52More recently, some of the recoupling methods have
been extended for applications in uniformly13C,15N labeled
solids53-58and applied to determine the high-resolution struc-
tures of a number of peptides.59,60
The main purpose of this paper is to investigate if recoupling
techniques can be used to provide accurate structural constraints
in uniformly
resonance47,48to probe carbonyl side chain distances. In our
experiments, the polarization transfer dynamics is monitored
as a function of spinning frequency, a so-called rotational
resonance width approach.61We have recently introduced a
homogeneously broadened version of the rotational resonance
width experiment, where the broadening of the R2matching
condition is achieved through the reduction of decoupling power
during R2mixing.62The key advantages of this approach are
that (i) the use of reduced decoupling power broadens the R2
resonance width, so that multiple spin pairs can be recoupled
in a single experiment, and (ii) the multispin effects are reduced
in the presence of short zero quantum relaxation, enabling a
two-spin approximation to be employed for data analysis. A
56 residue protein, ?1 immunoglobulin binding domain of
protein G (GB1) was chosen as a model system for our
experiments. It has been extensively studied by various bio-
physical techniques, including solution and solid-state NMR.
Its high-resolution 3D solution63and crystal structures,64as well
as complete13C and15N solid-state NMR resonance assign-
ments, are available.65We demonstrate that a large number of
well-defined13C-13C constraints can in fact be determined from
the resonance width curves and that these constraints can be
used for structure refinement.
13C,15N-labeled proteins. We used rotational
Experimental Section
Sample Preparation. The T2Q mutant of GB1 was heterologously
expressed in E. coli BL21 (DE3) grown in minimal media (1 g/L15-
NH4Cl and 2 g/L13C-glucose), induced with 0.5 mM isopropyl ?-D-
thiogalactoside (IPTG) for 4-5 h. Protein purification was performed
as follows: the cell pellet was disrupted by sonication in 20 mM Tris
at pH 8.0. The supernatant was subjected to anion exchange (HiTrap
Q HP column, Amersham Bioscience, NJ) and size exclusion (HiPrep
16/60 Sephacryl S-100 high-resolution column, Amersham Bioscience,
NJ) chromatography. Peak fractions were pooled and dialyzed three
times against 2 L of distilled water and finally against 1 L of MilliQ
(23) Wilhelm, M.; Feng, H.; Tracht, U.; Spiess, H. W. J. Magn. Reson. 1998,
134, 255-260.
(24) Lange, A.; Luca, S.; Baldus, M. J. Am. Chem. Soc. 2002, 124, 9704-
9705.
(25) Tycko, R.; Ishii, Y. J. Am. Chem. Soc. 2003, 125, 6606-6607.
(26) Lange, A.; Becker, S.; Seidel, K.; Giller, K.; Pongs, O.; Baldus, M. Angew.
Chem., Int. Ed. 2005, 44, 2089-2092.
(27) Wishart, D. S.; Sykes, B. D. J. Biomol. NMR 1994, 4, 171-180.
(28) Cornilescu, G.; Delaglio, F.; Bax, A. J. Biomol. NMR 1999, 13, 289-302.
(29) Luca, S.; Filippov, D. V.; van Boom, J. H.; Oschkinat, H.; de Groot, H. J.
M.; Baldus, M. J. Biomol. NMR 2001, 20, 325-331.
(30) Feng, X.; Lee, Y. K.; Sandstrom, D.; Eden, M.; Maisel, H.; Sebald, A.;
Levitt, M. H. Chem. Phys. Lett. 1996, 257, 314-320.
(31) Hong, M.; Gross, J. D.; Griffin, R. G. J. Phys. Chem. B 1997, 101, 5869-
5874.
(32) Feng, X.; Eden, M.; Brinkmann, A.; Luthman, H.; Eriksson, L.; Graslund,
A.; Antzutkin, O. N.; Levitt, M. H. J. Am. Chem. Soc. 1997, 119, 12006-
12007.
(33) Costa, P. R.; Gross, J. D.; Hong, M.; Griffin, R. G. Chem. Phys. Lett. 1997,
280, 95-103.
(34) Hong, M.; Wi, S. Torsion Angle Determination in Biological Solids by
Solid-State Nuclear Magnetic Resonance. In NMR Spectroscopy of Biologi-
cal Solids; Ramamoorthy, A., Ed.; Taylor & Francis Group, LLC: Boca
Raton, FL, 2006.
(35) Reif, B.; Hohwy, M.; Jaroniec, C. P.; Rienstra, C. M.; Griffin, R. G. J.
Magn. Reson. 2000, 145, 132-141.
(36) Rienstra, C. M.; Hohwy, M.; Mueller, L. J.; Jaroniec, C. P.; Reif, B.; Griffin,
R. G. J. Am. Chem. Soc. 2002, 124, 11908-11922.
(37) Ladizhansky, V.; Jaroniec, C. P.; Diehl, A.; Oschkinat, H.; Griffin, R. G.
J. Am. Chem. Soc. 2003, 125, 6827-6833.
(38) Franks, W. T.; Wylie, B. J.; Stellfox, S. A.; Rienstra, C. M. J. Am. Chem.
Soc. 2006, 128, 3154-3155.
(39) Kosen, P. A. Methods Enzymol. 1989, 177, 86-121.
(40) Battiste, J. L.; Wagner, G. Biochemistry 2000, 39, 5355-5365.
(41) Jovanovic, T.; McDermott, A. E. J. Am. Chem. Soc. 2005, 127, 13816-
13821.
(42) Pintacuda, G.; Giraud, N.; Pierattelli, R.; Bockmann, A.; Bertini, I.; Emsley,
L. Angew. Chem., Int. Ed. 2007, 46, 1079-1082.
(43) Nadaud, P. S.; Helmus, J. J.; Hofer, N.; Jaroniec, C. P. J. Am. Chem. Soc.
2007, 129, 7502-7503.
(44) Andrew, E. R.; Bradbury, A.; Eades, R. G. Nature 1958, 182, 1659.
(45) Lowe, I. J. Phys. ReV. Lett. 1959, 2, 285.
(46) Gullion, T.; Schaefer, J. J. Magn. Reson. 1989, 81, 196-200.
(47) Raleigh, D. P.; Levitt, M. H.; Griffin, R. G. Chem. Phys. Lett. 1988, 146,
71-76.
(48) Colombo, M. G.; Meier, B. H.; Ernst, R. R. Chem. Phys. Lett. 1988, 146,
189-196.
(49) McDowell, L. M.; Lee, M. S.; McKay, R. A.; Anderson, K. S.; Schaefer,
J. Biochemistry 1996, 35, 3328-3334.
(50) Mcdermott, A. E.; Creuzet, F.; Griffin, R. G.; Zawadzke, L. E.; Ye, Q. Z.;
Walsh, C. T. Biochemistry 1990, 29, 5767-5775.
(51) Creuzet, F.; Mcdermott, A.; Gebhard, R.; Vanderhoef, K.; Spijkerassink,
M. B.; Herzfeld, J.; Lugtenburg, J.; Levitt, M. H.; Griffin, R. G. Science
1991, 251, 783-786.
(52) Lansbury, P. T.; Costa, P. R.; Griffiths, J. M.; Simon, E. J.; Auger, M.;
Halverson, K. J.; Kocisko, D. A.; Hendsch, Z. S.; Ashburn, T. T.; Spencer,
R. G. S.; Tidor, B.; Griffin, R. G. Nat. Struct. Biol. 1995, 2, 990-998.
(53) Jaroniec, C. P.; Tounge, B. A.; Herzfeld, J.; Griffin, R. G. J. Am. Chem.
Soc. 2001, 123, 3507-3519.
(54) Jaroniec, C. P.; Filip, C.; Griffin, R. G. J. Am. Chem. Soc. 2002, 124,
10728-10742.
(55) Verhoeven, A.; Williamson, P. T. F.; Zimmermann, H.; Ernst, M.; Meier,
B. H. J. Magn. Reson. 2004, 168, 314-326.
(56) Williamson, P. T. F.; Verhoeven, A.; Ernst, M.; Meier, B. H. J. Am. Chem.
Soc. 2003, 125, 2718-2722.
(57) Ramachandran, R.; Ladizhansky, V.; Bajaj, V. S.; Griffin, R. G. J. Am.
Chem. Soc. 2003, 125, 15623-15629.
(58) Ladizhansky, V.; Griffin, R. G. J. Am. Chem. Soc. 2004, 126, 948-958.
(59) Rienstra, C. M.; Tucker-Kellogg, L.; Jaroniec, C. P.; Hohwy, M.; Reif, B.;
McMahon, M. T.; Tidor, B.; Lozano-Perez, T.; Griffin, R. G. Proc. Natl.
Acad. Sci. U.S.A. 2002, 99, 10260-10265.
(60) Jaroniec, C. P.; MacPhee, C. E.; Bajaj, V. S.; McMahon, M. T.; Dobson,
C. M.; Griffin, R. G. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 711-716.
(61) Costa, P. R.; Sun, B. Q.; Griffin, R. G. J. Magn. Reson. 2003, 164, 92-
103.
(62) Janik, R.; Peng, X.; Ladizhansky, V. J. Magn. Reson. 2007, 188, 129-
140.
(63) Gronenborn, A. M.; Filpula, D. R.; Essig, N. Z.; Achari, A.; Whitlow, M.;
Wingfield, P. T.; Clore, G. M. Science 1991, 253, 657-661.
(64) Gallagher, T.; Alexander, P.; Bryan, P.; Gilliland, G. L. Biochemistry 1994,
33, 4721-4729.
(65) Franks, W. T.; Zhou, D. H.; Wylie, B. J.; Money, B. G.; Graesser, D. T.;
Frericks, H. L.; Sahota, G.; Rienstra, C. M. J. Am. Chem. Soc. 2005, 127,
12291-12305.
A R T I C L E S
Peng et al.
360 J. AM. CHEM. SOC.9VOL. 130, NO. 1, 2008

Page 3

water. Pure protein was concentrated with Centriplus 3 kDa MWCO
filters and stored at -20 °C before use. Precipitation of GB1 for NMR
studies was performed using a previously published protocol, without
modifications.65Approximately 15 mg of GB1 precipitate were packed
into a 3.2 mm NMR rotor. The sample was centered in the rotor between
two cylindrical Vespel inserts. The HBR2W distance measurements and
cross-peak assignment experiments were performed on a single
U-13C,15N labeled sample. For the identification of intermolecular
contacts, some of the experiments were repeated with a uniformly
13C,15N labeled sample mixed in a 1:1 molar ratio with nonlabeled
protein. Approximately 15 mg of the diluted protein sample were packed
into the rotor.
NMR Experiments. All MAS NMR experiments were performed
on a Bruker Avance II spectrometer, operating at 600.130 MHz proton
frequency, using a Bruker triple resonance
BioSolids magic angle spinning probe (Bruker USA, Billerica, MA).
Two pulse sequences employed in this work are presented in Figure 1.
Both start with1H/15N cross polarization of 2 ms duration, implemented
with a radio frequency (rf) field ramped around 55 kHz on the proton
channel and with a constant rf field of ∼43-45 kHz on the nitrogen
channel.15N/13C′ cross polarization was performed with an rf carbon
field ramped around 42-43 kHz (5% ramp), with a constant rf field
of ∼31 kHz on the nitrogen channel and with a CW decoupling proton
field of 100 kHz. The
experiments. A proton field of 65 kHz (optimized experimentally) was
applied for the two-pulse phase modulation decoupling (TPPM) (7.6
µs TPPM pulse, 15° overall phase shift) during nitrogen t1and carbon
t2indirect chemical shift evolutions. Higher TPPM decoupling of 71
kHz (7µs TPPM pulse, 12° phase shift) was applied during direct13C
acquisition. The π/2 pulse widths were 2.7 µs, 5 µs, and 6 µs on the
1H,13C, and15N channels, respectively. The selective rotor-synchronized
Gaussian π pulse applied to carbonyl carbons was around 160-180
µs in length, depending on the spinning frequency. The temperature of
the cooling gas was kept at 260 K during the course of all experiments.
Homogeneously Broadened Rotational Resonance Width Experi-
ments. A pulse sequence for distance measurements is shown in Figure
1a. In this experiment, the1H/15N cross polarization creates transverse
15N coherences, which are subsequently frequency-labeled during t1
1H-13C-15N 3.2 mm
15N/13C′ CP mixing time was 6 ms in all
isotropic chemical shift evolution. The13C π pulse in the middle of
the t1 evolution refocuses one-bond15N-13C J-couplings. Following
t1evolution, a band-selective version of15N/13C′ SPECIFIC CP66creates
13C′ single quantum coherences, which are frequency labeled during t2
isotropic chemical shift evolution. Here, a combination of a hard π
pulse and of a soft13C′-selective Gaussian π pulse refocuses the one-
bond homonuclear13C′-13CR J-couplings,67whereas the hard π pulse
on the15N channel removes the residual line broadening associated
with15N-13C′ J-interactions.
Structurally constraining carbonyl side chain dipolar interactions are
recoupled during13C-13C mixing, following the t2period. We used a
homogeneously broadened version of R2(HBR2). The broadening is
accomplished through the reduction of the decoupling power during
R2mixing.62The experiments were conducted as a function of spinning
frequency, which constitutes the fourth pseudo-dimension of the
experiment. It was shown previously that such a homogeneously
broadened rotational resonance width (HBR2W) implementation has
certain advantages over a conventional time-dependent magnetization
exchange experiment: (i) the data interpretation does not require prior
knowledge or assumptions about zero-quantum relaxation times,57,61,62
reducing systematic errors associated with this uncertainty, and (ii) many
spin pairs can be recoupled in a single experiment, while recoupling
of strong one-bond13C′-13CR interactions can still be avoided.62A
constant HBR2mixing time of 40 ms was used in all experiments. The
CW proton decoupling power during mixing was optimized experi-
mentally to be 47 kHz, the lowest value at which no recoupling of one
bond13C′-13CR dipolar interactions was observed. Fifteen 3D spectra
were recorded at spinning frequencies of 11.3, 11.4, 11.5, 11.6, 11.65,
11.7, 11.75, 11.8, 11.85, 11.9, 11.975, 12.05, 12.15, 12.25, 12.35 kHz.
At a proton field of 600 MHz, these spinning frequencies recouple
carbonyls and aliphatic side chain carbons through the n ) 2 R2
matching condition.
The schematic representation of the evolution of the relevant part
of the density matrix, starting with the nitrogen magnetization following
1H/15N CP, can be written as follows:
In the expression above, Nix, C′i-1x, and Cjx
corresponding to single quantum coherences of nitrogen, carbonyl, and
aliphatic side chain (X ) γ, δ, etc.) spins, and ΩNi, ΩC ′i-1, and ΩCjXare
their respective isotropic chemical shifts. This pathway is collected
according to a TPPI phase sensitive scheme in t1 and t2 dimensions.
There are generally two sets of peaks observed in the Fourier
transformed 3D spectra. The pseudo-diagonal peaks of relative intensity
(1-R) result from the singly underlined component of the density matrix
in eq 1 and have ΩNi/ΩC ′i-1/ΩC ′i-1shifts in the F1/F2/F3 dimensions,
respectively. These peaks determine the origin of the R2magnetization
transfer. The cross-peaks of relative intensity R, doubly underlined in
eq 1, occur at shifts of ΩNi/ΩC ′i-1/ΩCjXand correspond to the aliphatic
carbons
intensities of these peaks, R, depend on the strengths of the dipolar
interaction between
frequency, and on the residual coupling to the proton bath.
Cross-Peak Assignment Experiments. Within each interacting pair
of spins13C′[i-1]-13CX[j], the carbonyl spin can be identified in the
3D correlation spectra according to the known
correlation pattern.65In contrast, the assignments of the aliphatic cross-
Xdenote spin operators
13CX[j], through-space coupled to
13C′[i-1]. The relative
13C′[i-1] and
13CX[j] spins, on the spinning
15N[i]-13C′[i-1]
(66) Baldus, M.; Petkova, A. T.; Herzfeld, J.; Griffin, R. G. Mol. Phys. 1998,
95, 1197-1207.
(67) Straus, S. K.; Bremi, T.; Ernst, R. R. Chem. Phys. Lett. 1996, 262, 709-
715.
Figure 1. Experimental pulse sequences. The thin and thick dark bars
represent π/2 and π pulses, respectively. In (a), a three-dimensional chemical
shift correlation experiment used for HBR2W distance measurements is
shown. A selective rotor-synchronized Gaussian π pulse is applied to the
carbonyl region. The following phase cycling was used: φ1, φ2, φ3) (x, x,
y, y, -x, -x, -y, -y), φ4) (y, y, -x, -x, -y, -y, x, x), φ5) (-y, -y, x,
x, y, y, -x, -x), φrec ) (x, -x, y, -y, -x, x, -y, y). In (b), a
three-dimensional chemical shift correlation experiment for cross-peak
assignments is shown. The phase cycling was as follows: φ1) (x, x, y, y,
-x, -x, -y, -y), φreceiver) (x, -x, y, -y, -x, x, -y, y).
Homonuclear13C−13C Measurements in Proteins
A R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 130, NO. 1, 2008 361

Page 4

peaks are generally ambiguous, because they are only based on the
chemical shifts of the13CX[j] spins. The cross-peak assignments were
assisted with a series of three-dimensional chemical shift correlation
experiments shown in Figure 1b. The polarization transfer pathway
here is the same as the one explained in eq 1 for the HBR2experiment,
except that the carbonyl chemical shift evolution is replaced by the
chemical shift evolution of the aliphatic spins and by an additional
13C-13C mixing step. The resulting15N[i]-13CX[j]-13CY[j] (X ) γ, δ;
Y ) R, ?, etc.) correlation shares15N[i] and13CX[j] shifts with the HBR2
15N[i]-13C′[i-1]-13CX[j] correlation and also allows identification of
13CX[j] spins according to the unique intraresidue
correlation pattern. A total of four experiments for cross-peak assign-
ments were recorded at spinning frequencies of 11.5, 11.65, 11.9, and
12.05 kHz. The assignment strategies are further discussed in the Results
and Discussion section.
For the identification of intermolecular contacts, three additional
HBR2experiments were recorded at spinning frequencies of 11.3, 11.65,
and 12.25 kHz in a diluted sample. Spectra at 11.3 and 11.65 kHz
were recorded with 16 scans, and the experiment at 12.25 kHz was
recorded with 8 scans.
Proton-Driven Spin Diffusion Experiments. To get additional
constraints for structure calculation and refinement, HBR2W measure-
ments were supplemented with a set of distance restraints derived from
2D and 3D spin diffusion spectra. A 3D NCOCX spectrum was
recorded using the pulse sequence of Figure 1a, except that no
decoupling was applied during COCX mixing. The 3D NCACX spin
diffusion experiment was recorded using a previously published pulse
sequence.19In addition, a 2D
spectrum was taken. All spin diffusion experiments were recorded with
a 500 ms mixing time. The13C-13C distance restraints derived from
these spin diffusion spectra were used for structure calculation as
discussed in the following section.
Data Analysis. Experimental data were processed with NMRPipe.68
Processing parameters are provided in the respective figure captions.
Chemical shift was indirectly referenced to DSS using the chemical
shift of the downfield13C resonance in adamantane (40.48 ppm).69The
general positions of the cross-peaks in our spectra were found to be
very similar to the published data, with most peaks found to be within
0.2 ppm from previously published chemical shifts,65although some
15N shifts were found to deviate by as much as 0.5 ppm. For reference
purposes, the assignment table derived from a series of 2D (13C-13C,
NCA, NCO) and 3D (NCACX, NCOCX) correlation experiments and
used in the analysis of the HBR2experiments is given in the Supporting
Information. Processed data were analyzed using Sparky software,
version 3.1 (T. D. Goddard and D. G. Kneller, University of California,
San Francisco, CA). Peak volumes were extracted with automated fitting
to three-dimensional Gaussians.
The HBR2W curves (i.e., cross-peak intensities as functions of
spinning frequencies) were fitted using a steady-state approximate
solution for a two-spin system under short zero-quantum relaxation
conditions:62
S(∆) )〈
13CX[j]-13CY[j]
13C-13C spin diffusion correlation
Here tmixand RZQare the HBR2constant mixing time (40 ms) and zero
quantum relaxation rate, RZQ ) 1/T2
2νRdepends on the spinning frequency νRand measures the deviation
from the exact n ) 2 rotational resonance condition. δI, δS are the
chemical shifts of spins I (carbonyl) and S (aliphatic spin), respectively.
The effective dipolar interaction ωD
ZQ, respectively, ∆ ) δI - δS -
(2)was calculated as70
where bIS ) (-µ0/4π)(γ2p/r3) is the dipolar coupling constant for a
homonuclear spin pair, r is the internuclear distance, and ? is the Euler
angle defining the orientation of the internuclear vector with respect
to the rotor-fixed frame. The effects of chemical shift anisotropy (CSA)
were neglected in the fitting but were taken into account through the
introduction of the systematic error as explained in the Results and
Discussion section.
We have shown previously that the steady-state approximation
accurately describes R2polarization transfer in situations when the zero
quantum relaxation rate is large compared to the effective interaction
strength.62This assumption holds well under low decoupling conditions
used in this work.
Equation 2 assumes that the aliphatic cross-peaks are normalized
per initial intensity of the corresponding carbonyl peak. To avoid the
additional time-consuming procedure of collecting reference spectra
at zero mixing time, the aliphatic cross-peak intensities were normalized
with respect to the sum of all peaks occurring at ΩNi/ΩC ′i-1shifts in the
F1and F2indirect dimensions, including spinning sidebands:
Here, SNiC ′i-1CjXis the volume intensity of the cross-peak of interest,
SNiC ′i-1Ckis the intensity of the peak occurring at shifts ΩNi/ΩC ′i-1/ΩCk,
and the summation is extended over all cross-peaks at ΩNi/ΩC ′i-1/ΩCk
shifts along F3. This normalization procedure assumes that the
longitudinal T1relaxation times of carbon spins are long on the time
scale of HBR2mixing of 40 ms, and thus the total polarization is
conserved during R2.
Statistical analysis of random errors was performed using in-house
Monte Carlo simulations, implemented in FORTRAN 77. Briefly, the
analytical expression 2 was used to generate a two-dimensional grid
of R2W curves simulated as a function of relaxation rate RZQand an
internuclear distance for each spin pair, using experimental mismatch
values ∆. A typical size of the grid was 2500 curves, for distances in
the range 2.5-7.5 Å incremented in a step of 0.1 Å and for relaxation
times from 0.1 to 5 ms incremented in a step of 0.1 ms. Such a grid
can be generated in less than 1 min, benefiting from the use of the
analytical solution of eq 2 for the signal intensity. Once the analytical
solution grid was generated, the best fit to the experimental R2W curve
was found, and the root-mean-square deviation (rmsd) was computed.
The rmsd was used to generate random noise according to a Gaussian
distribution with zero mean and a standard deviation equal to the rmsd.
The simulated noise was added to each point of the best fit, and the
resulting curves were refit using the same solution grid. This procedure
was repeated 10 000 times. A histogram showing the frequencies of
occurrence of each distance in this procedure was used to extract a
range of internuclear distances at a 95% confidence level.
Results and Discussion
15N[i]-13CO[i-1] Spectral Resolution in GB1. The indirect
F1-F2 projection of the15N-13CO spectrum recorded at a
spinning frequency of 11.65 kHz is shown in Figure 2. The
cross-peaks in the spectrum identify the origin of the polarization
transfer process, and the resolution provides the basis for the
site-specific analysis of the 3D spectra. Generally, the resolution
in the indirect carbonyl chemical shift dimension observed in
this work is higher than the one observed directly, owing to
the homonuclear JCOCRdecoupling implemented in the pulse
sequence.
(68) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax, A. J.
Biomol. NMR 1995, 6, 277-293.
(69) Morcombe, C. R.; Zilm, K. W. J. Magn. Reson. 2003, 162, 479-486.
(70) Levitt, M. H.; Raleigh, D. P.; Creuzet, F.; Griffin, R. G. J. Chem. Phys.
1990, 92, 6347-6364.
1
2(1 - exp{-4|ωD
(2)|2RZQ
∆2+ RZQ
2
tmix})〉powder
(2)
ωD
(2)) bIS1
4sin2?
(3)
SNiC ′i-1CjX
norm
)
SNiC ′i-1CjX
∑
k
SNiC ′i-1Ck
(4)
A R T I C L E S
Peng et al.
362 J. AM. CHEM. SOC.9VOL. 130, NO. 1, 2008

Page 5

The typical experimentally observed line widths were ∼0.4-
0.5 ppm for the13C indirect dimension and ∼0.5-0.6 ppm for
the
projection spectra.
Of the expected 55 backbone15N[i]-13CO[i-1] peaks, a total
of 45 are well resolved, and their corresponding volume
intensities can be reliably extracted from the 3D spectra. The
remaining 10 backbone cross-peaks show some degree of
spectral overlap. In particular, the Q2N-M1C′ and T19N-T18C′
peaks at about 126/171.2 ppm overlap completely, and their
contributions to the formation of the aliphatic cross-peaks cannot
be disentangled. There is a similar situation for the heavily
overlapping group of Y3N-Q2C′, K4N-Y3C′, A23N-D22C′
peaks at around 123.6/174.8 ppm and for the V21N-A20C′ and
E27N-A26C′ peaks at about 116.8/177.3 ppm. Partially overlap-
ping peaks T51N-K50C′ and T53N-F52C′ at around 112.2/155.6
ppm could not be integrated accurately, but their peak volumes
could be estimated from the 3D spectra. The D22N-V21C′ at
116/175 ppm cross-peak is very weak, probably because of the
high mobility of V21.65
In addition to15N[i]-13CO[i-1] backbone correlations, two
side chain cross-peaks corresponding to the nitrogen-carboxy-
late correlations in asparagines are observed. A total of five
side chain15N-13C correlations of this type are expected, but
only two,15Νδ-13Cγ of N35 and N37 at 114.3/175.8 ppm, and
115.1/176.6 ppm, respectively, can be clearly seen in the 2D
indirect projection of Figure 2. Both of these residues are located
inside the R-helical secondary structure, and their side chains
are relatively immobile. In contrast, the side chain correlations
in Q32, Q2, and N8 are of much lower intensity in the full 3D
15N dimension, as estimated from the 2D
15N-13CO
spectrum, probably due to their increased mobility. These peaks
are missing in this 2D projection.
HBR2Spectra. At 600 MHz, most carbonyl-aliphatic side
chain spin pairs have their n ) 2 R2conditions in the range of
11.3-12.35 kHz. Matching this condition within the HBR2
recoupling bandwidth results in the formation of the cross-peaks.
Figure 3 shows representative 2D planes (F1-F3,15N indirect
and13C direct dimensions) of the full 3D HBR2experiments
measured at different spinning rates. The left panel shows a
2D plane of the15N[i]-13CO[i-1]-13CO[i-1] part of the
spectrum taken at 11.65 kHz spinning frequency, while three
panels on the right show aliphatic peaks detected at 11.5 kHz,
11.6 kHz, and 11.65 kHz, respectively. Cross-peaks aligned
horizontally represent interacting spin pairs. For example, there
is a strong interaction between N37C′ and V39γ1 carbons,
whereas V39γ1 also interacts strongly with G38C′. Most of the
peaks observed in the spectra in Figure 3 correspond to either
intraresidue (e.g., T17C′-T17Cγ2 or T44C′-T44Cγ2, with exact
n ) 2 R2conditions at 11.483 kHz, and 11.536 kHz, respec-
tively) or sequential (e.g., E15C′-T16Cγ2, G38C′-V39γ1 with
exact n ) 2 R2conditions at 11.618 kHz and 11.468 kHz,
respectively) connectivities. These contacts mostly constrain
local side chain conformations. There are also some medium
and long range interactions (e.g., N37C′-V39γ1 and E15C′-
T44Cγ2, with exact matching conditions at 11.476 kHz and
11.544 kHz, respectively), which report on the secondary
structure and global fold of the protein. In the latter case, the
E15C′-T44Cγ2 cross-peak corresponds to an intermolecular
interaction, as discussed in the following section.
Figure 2. 2D15N[i]-13C′[i-1] indirect F1-F2projection of the 3D HBR2spectrum, recorded at a spinning rate of 11.65 kHz. 84 real data points with 200
µs dwell time were taken in t1, and 52 real data points with 125 µs dwell time were taken in t2. The spectrum was recorded with 12 scans and with a
recycling delay of 2.3 s. Forward prediction of 35 points and 52 points was applied to the raw data in the t1and t2dimensions, respectively. Data were
processed with exponential line broadening of 5.0 Hz in the t1dimension, Lorentzian-to-Gaussian apodization (5.0 Hz of Gaussian line broadening and 10
Hz of Lorentzian line narrowing) in the t2indirect dimension, and with the π/2.5-shifted squared sinusoidal function in the t3direct13C dimension. All peaks
are labeled according to their carbonyl/carboxylate shifts.
Homonuclear13C−13C Measurements in Proteins
A R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 130, NO. 1, 2008 363