TROSY, transverse relaxation-optimized spectroscopy Konstantin Pervushin Laboratorium für Physikalische Chemie, Swiss Federal Institute of Technology, ETH- Hönggerberg, CH-8093 Zürich, Switzerland. Konstantin Pervushin: [email protected]EMBO Practical Course, Heidelberg September 10-17, 2003 TROSY scope The general problem posed in 3D structure determination of biomolecules by NMR involves the collection of a sufficiently dense set of experimental restraints to define the structure. Experimental restraints can be derived from interproton NOEs [1-3], residual dipole-dipole couplings and CSA interactions [4-9], scalar J couplings [1, 10], J couplings across hydrogen bonds [11, 12] and auto- and cross-correlated relaxation rates [13, 14]. Currently available computer-based 3D structure reconstruction methods require that a large number of experimental restraints are assigned to particular atoms in the protein’s chemical structure at the outset of structure determination [15]. In large proteins this assignment problem is aggravated by fast transverse relaxation of the spins of interest and the complexity of the NMR spectra, both of which increase with increasing molecular size [16-19]. The use of TROSY [20] together with uniform deuteration [21-24] reduces the transverse relaxation rates of 1 H N , 15 N and 13 C aromatic spins. TROSY triple resonance 3-D and 4-D NMR experiments were developed [25-28] enabling successful assignment of the backbone resonances in large biomolecules. 1
30
Embed
TROSY, transverse relaxation-optimized spectroscopycomplexity of the NMR spectra, both of which increase with increasing molecular size [16-19]. The use of TROSY [20] together with
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
EMBO Practical Course, Heidelberg September 10-17, 2003
TROSY scope The general problem posed in 3D structure determination of biomolecules by NMR involves the
collection of a sufficiently dense set of experimental restraints to define the structure. Experimental
restraints can be derived from interproton NOEs [1-3], residual dipole-dipole couplings and CSA
interactions [4-9], scalar J couplings [1, 10], J couplings across hydrogen bonds [11, 12] and auto- and
cross-correlated relaxation rates [13, 14]. Currently available computer-based 3D structure
reconstruction methods require that a large number of experimental restraints are assigned to particular
atoms in the protein’s chemical structure at the outset of structure determination [15]. In large proteins
this assignment problem is aggravated by fast transverse relaxation of the spins of interest and the
complexity of the NMR spectra, both of which increase with increasing molecular size [16-19]. The use
of TROSY [20] together with uniform deuteration [21-24] reduces the transverse relaxation rates of 1HN, 15N and 13Caromatic spins. TROSY triple resonance 3-D and 4-D NMR experiments were developed
[25-28] enabling successful assignment of the backbone resonances in large biomolecules.
Energy level diagram of a two spin 1/2 system IS showing the identification of components of the 2D multiplet with
off-diagonal density matrix elements expressed via single transition basis operators, I13, I24, S12 and S34, connecting
eigenstates j and k of the static Hamiltonian superoperator. The diagram is not drawn to scale. 2D cross-peak from the
[15N,1H]-HSQC spectrum of a backbone amide moiety of u- [15N,2H]-Aldolase, 110 kDa at 20oC measured without 1H and 15N decoupling.
TROSY techniques
1. Coherent evolution: coherence/polarization transfer by unitary rotations and design of NMR
experiments.
2. Stochastic processes: spin relaxation in a multispin system.
3. A combination of both: polarization transfer by relaxation.
5
1. Coherent evolution: coherence/polarization transfer by unitary rotations and design of NMR
experiment.
Set of basis operators for a two spin ½ system.
The vector Basis forms an orthonormal, zero-trace, and by multiplication with i the skew-Hermitian
(see some useful explanations in Appendix) (or antihermitian, A† = -A, where † means adjoint or hermitian
conjugation or A† = Conjugate//Transpose//A operation) basis set for Lie Algebra su(4) special unitary
group for a 4 energy-level problem.
Each individual spin operator can be represented by a 4x4 matrix:
6
This basis set ensures that each of the individual exponentials of the form ebi is unitary.
This would not be the case if, for instance, irreducible spherical tensor or single-element
polarization/step operators were chosen as a basis, in which case bindings between the coefficients in
the exponentials are needed to ensure unitarity of the full propagator [60].)
7
However the irreducible spherical tensor T2,0 forms a unitary propagator. It corresponds to the
dipolar hamiltonian, HD = aD Sqrt[6] T2,0.
8
The single-element polarization/step operators:
The multiple quantum operators:
Design of the optimum coherent experiment
9
The conversion of one state to another can be associated with a transfer efficiency. Consider,e.g. [61],
the transfer from a quantum-mechanical state represented by an operator B to a state represented by an
operator A by a unitary transformation U (ignoring dissipative processes)
where Q is a residual operator. The higher the coefficient
the more efficient is the transfer process and thereby the experiment. Determination of amax for a given
coherence or polarization transfer process falls in the field of spin-dynamics bounds. So far, three types
of such bounds have been described for a given transfer B→A.
where ΛX is a vector with the eigenvalues of the Hermitian operator X arranged in descending order
while ΣX is a vector with the singular values of the matrix representation of the operator X also arranged
in descending order. The singular values defined as positive of a matrix M are the nonzero elements of a
diagonal matrix Σ that arise by diagonalization of M by unitary matrices T and V according to M=TΣV†.
Arbitrary matrices can be decomposed in this way and for Hermitian matrices it holds that T = V.
Singular value decomposition is an important element of many numerical matrix algorithms. The basic
idea is to write any matrix m in the form u* mD v, where m is a diagonal matrix, D u and v are row
orthonormal matrices, and u is the Hermitian transpose of u. The function SingularValues[m]
returns a list containing the matrix
*
u, the list of diagonal elements of m , and the matrix D v. The diagonal
elements of m are known as the singular values of the matrix D m. One interpretation of the singular
values is as follows. If you take a unit sphere in n-dimensional space, and multiply each vector in it by
an män matrix m, you will get an ellipsoid in m-dimensional space. The singular values give the lengths
of the principal axes of the ellipsoid. If the matrix m is singular in some way, this will be reflected in the
shape of the ellipsoid. In fact, the ratio of the largest singular value of a matrix to the smallest one gives
a condition number of the matrix, which determines, for example, the accuracy of numerical matrix
inverses. Very small singular values are usually numerically meaningless. SingularValues removes
any singular values that are smaller than a certain tolerance multiplied by the largest singular value. The
10
Hermitian bound applies for Hermitian operators A and B and there is a guarantee that the maximum
efficiency can be achieved by a unitary transformation.
(1) Select the desired polarization transfer pathway.
For the single quantum TROSY in hermitian operators it is:
Sx[2] + 2Sx[2]**Sz[3] → Sx[3] + 2Sx[3]**Sz[2]
In raiselowering (nonhermitian) operators:
S+[2] + 2S+[2]**Sz[3] → S-[3] + 2S-[3]**S-[2]
(note, that only the “minus” operators are directly detectable [62])
(2) Determine amax.
The hermitian bound:
The hermitian bound:
11
Hence, we obtain aherm = aSVD = 1.
(3) Use the energy level diagram to search for appropriate selective rotations.
Double Quantum rotations with the phases x and y.
αα
βα αβ
ββ
αα
βα αβ
ββ
For a selective Double Quantum (DQ) x phase π pulse ( πx
sel) the effective hamiltonian takes the form
with the spin functions ordered |αα>, |βα>, |αβ>, |ββ> in the Hilvert space. It follows that the rotation
in Fig must be of an angle π in order to achieve the maximum possible coherence transfer efficiency.
Explicitly, the rotation is
It corresponds to a π rotation operating selectively on the double quantum transition of the two spins
and can indeed be realized experimentally provided there resolved scalar (J) or dipolar coupling
between them.
12
By constructing the corresponding propagators one can verify achieved hermitian and SVD
bounds:
The symbol i indicates a π/2 phase shift of the detected operator.
For a selective Double Quantum (DQ) y phase π pulse ( πysel) the effective hamiltonian takes the form
with the spin functions ordered |αα>, |βα>, |αβ>, |ββ> in the Hilvert space.
Explicitly, the rotation is
13
Zero Quantum rotations with the phases x and y.
αα
βα αβ
ββ
αα
βα αβ
ββ
For a selective Zero Quantum (ZQ) x phase π pulse ( πx
sel) the effective hamiltonian takes the form
or
representing a planar mixing sequence which earlier has been proposed for the accomplishment of
the coherence order selective anti-phase transfer [63].
For a selective Zero Quantum (ZQ) y phase π pulse ( πy
sel) the effective hamiltonian takes the form
14
or
(4) Find out whether the experiment can be realized in a nonselective manner.
We start with the notion that operators Sx[2]**Sx[3] and Sy[2]**Sy[3] commute as well as the
Sx[2]**Sy[3] and Sy[2]**Sx[3] operators.
and
In general,
The propagator
may straightforwardly be converted into a practical pulse sequence using relations of the type
15
transforming the transverse bilinear rotations into mixtures of linear and longitudinal bilinear rotations
which can be realized by rf irradiation and evolution under heteronuclear JIS coupling, respectively. The
latter is typically accomplished by free precession under the unperturbed Hamiltonian
which additionally shows dependence on the isotropic chemical shifts for the two spin species.
Undesired influence from the chemical shifts may by standard means be eliminated by p-pulse
refocusing, i.e.,
resulting in
The practical pulse sequence propagator is (t = 1/(4 JIS)
where the final z rotation is irrelevant in the practical implementation [64].
t2t1
y
15N
1H-y
-y
y
y
The ZQ propagator
represents an original pulse sequence proposed by Pervushin et al. [20] and subsequently called the
SingleTrasition-to-Single Transition Polarization Transfer element (ST2-PT) [47].
16
t2t1
y
15N
1Hy
y
A practical implementation of the TROSY experiment based on the ST2-PT element.
Relevant polarization transfer pasways can be represented by the diagram
When both pathways indicated are retained, two diagonally shifted signals represent two out of the four 15N–1H multiplet components in the resulting [15N, 1H]-correlation spectrum. The undesired polarization
17
transfer pathway,12 → 13, is suppressed either by 2-step cycling of the phases ψ1 and ψ5 or by
application of PFGs during t1 and at time point d. The remaining anti-echo polarization transfer
pathway, 34 → 24, connects a single transition of spin S with a single transition of spin I, and in
alternate scans with inversion of the rf-phases ψ2 and ψ4, the corresponding echo transfer, 12 → 13, is
recorded. In the experiments of Fig. 2 we used both the 1H and 15N steady-state magnetizations. The
product operator analysis accounts for this by the following density matrix at time b in the experimental
scheme
The constant factors u and v reflect the relative magnitudes of the steady-state 1H and 15N
magnetizations, respectively, which are determined by the gyromagnetic ratios, the spin–lattice
relaxation rates and the delay between individual data recordings. Since the S34 operators are transferred
to observable magnetization, both the 1H and 15N steady-state magnetizations add up to the signal
obtained with the pulse sequence, which is proportional to (u-v)/2.
Zero quantum TROSY experiment [65]
(1) The desired polarization transfer pathways are
2S+[2]**S-[3] → S-[3] + 2S-[3]**S-[2] (ZQ)
2S+[2]**S+[3] → S-[3] + 2S-[3]**S-[2] (DQ)
(2) The SVD bound is:
amax
SVD = 1.
(3) The energy level diagram to search for appropriate selective rotations.
18
αα
βα αβ
ββ
αα
βα αβ
ββ
For a selective αα−αβ x phase π pulse ( πx
sel) the effective hamiltonian takes the form
with the spin functions ordered |αα>, |βα>, |αβ>, |ββ> in the Hilvert space. The effective
hamiltonian is
(4) Implementation of the effective hamiltonian.
For the transfer process in question we first eliminate the ‘‘selective element’’ represented by the
polarization operators out of the rotations
Since involved operators commute, one obtains
The first z-rotation is immaterial, resulting in the ZQ-TROSY experimental scheme:
19
t2t1
y
15N
1Hy -y
A practical implementation of the TROSY experiment based on the ZQ element.
Relevant polarization transfer pasways can be represented by the diagram
ZQ-TROSY is effectively used in the context of 3D 15N-resolved NOESY in order to suppress
strong diagonal peaks.
20
21
2. Stochastic processes: spin relaxation in a multispin system
Transverse relaxation in coupled spin systems
We consider a system of two scalar coupled spins ½, I and S, with a scalar coupling constant JIS,
which is located in a protein molecule. T2 relaxation of this spin system is dominated by the DD
coupling of I and S and by CSA of each individual spin, since the stereochemistry of the polypeptide
chain restricts additional interactions of I and S to weak scalar and DD couplings with a small number
of remote protons, Ik. The relaxation rates of the individual multiplet components of spins I and S in a
single quantum spectrum may then be widely different due to the effect of interference between IS
dipolar coupling and anisotropic chemical shift on transverse relaxation [66-68]. Shimizy [69]
suggested to use the DD/CSA interference effect for determination of absolute sign of the coupling
constants by measuring the saturation or the width of each component of a NMR spectrum that gives
multiplet lines through spin-spin couplings. Detailed descriptions of the interference effects on
transverse relaxation in coupled spin systems are available [67, 70-74].
The simplest functional form of the transverse relaxation equations can be obtained by using
single-transition and zero- and double-quantum basis operators [2, 65, 67, 75]. In the slow-tumbling
limit for an isolated IS spin system in the absence of rf pulses and rf field inhomogeneities only terms in
J(0) need to be retained resulting in an uncoupled system of differential equations with the diagonal
form of the first-order relaxation matrix.
where J(ω) represents the spectral density functions at the frequencies indicated:
22
ωS and ωI are the Larmor frequencies of the spins S and I,
h is the Plank constant divided by 2π, rIS the distance between S and I, B0 the polarizing magnetic field,
and ∆σS and ∆σI are the differences between the axial and the perpendicular principal components of
the axially symmetric chemical shift tensors of spins S and I, respectively. Ckl = 0.5(3cos2Θkl-1) and Θkl
the angle between the unique tensor axes of the interactions k and l. The single-transition basis operators
To quantitatively evaluate the transverse relaxation optimization effect and contributions from
other mechanisms of relaxation we consider a specific example where I and S are identified as the 1HN
and 15N spins in a 15N–1H moiety. The internuclear distances and CSA tensors in various model
compounds including amide moieties in short peptides were extensively studied by solid state NMR, so
that we use the data obtained there: the internuclear distance r(15N−1HN) = 1.04 Å for the peptide
backbone amide moieties [76]; ∆σ(1HN) = σzz – (σxx + σyy)/2 = 15 ppm, is axially symmetric with the
angle between the zz axis and the 15N−1HN bond = 10o [77, 78]; ∆σ(15N) = σzz – (σxx + σyy)/2 = -155
ppm, is axially symmetric with the angle between zz axis and 15N− 1HN bond = 15o [79].
(protonated; 600 MHz). For the 1HN dimension, the upheld 1HN-{15N} component on average relaxes
slower than the downfield 1HN-{15N} component by a factor of 1.8 (perdeuterated; 800 MHz) and 1.6
(perdeuterated; 600 MHz). The reported data correspond well with the rates expected on the basis of the
theoretical calculations [26].
24
Appendix
25
26
REFERENCES
27
1. Wuthrich, K., NMR of Proteins and Nucleic Acids. 1986, New York: Wiley. 2. Ernst, R.R., G. Bodenhausen, and A. Wokaun, The Principles of Nuclear Magnetic Resonance in
One and Two Dimensions. 1987, Oxford: Clarendon. 3. Clore, G.M. and A.M. Gronenborn, Science, 1991. 252(5011): p. 1390-9. 4. Ernst, M. and R.R. Ernst, Journal of Magnetic Resonance Series A, 1994. 110(2): p. 202-213. 5. Tolman, J.R., et al., Proceedings of the National Academy of Sciences of the United States of
America, 1995. 92(20): p. 9279-9283. 6. Tjandra, N. and A. Bax, Science, 1997. 278(5340): p. 1111-1114. 7. Tjandra, N., et al., Nat. Struct. Biol., 1997. 4(9): p. 732-8. 8. Choy, W.Y., et al., Journal of Biomolecular Nmr, 2001. 21(1): p. 31-40. 9. Wu, Z.R., N. Tjandra, and A. Bax, Journal of the American Chemical Society, 2001. 123(15): p.
3617-3618. 10. Bax, A., Current Opinion in Structural Biology, 1994. 4(5): p. 738-744. 11. Dingley, A.J. and S. Grzesiek, Journal of the American Chemical Society, 1998. 120(33): p.
8293-8297. 12. Pervushin, K., et al., Proceedings of the National Academy of Sciences of the United States of
America, 1998. 95(24): p. 14147-14151. 13. Tjandra, N., et al., Nat Struct Biol, 1997. 4(6): p. 443-9. 14. Reif, B., M. Hennig, and C. Griesinger, Science, 1997. 276(5316): p. 1230-1233. 15. Clore, G.M. and C.D. Schwieters, Current Opinion in Structural Biology, 2002. 12(2): p. 146-
153. 16. Wagner, G., Journal of Biomolecular Nmr, 1993. 3(4): p. 375-385. 17. Kay, L.E. and K.H. Gardner, Current Opinion in Structural Biology, 1997. 7(5): p. 722-731. 18. Clore, G.M. and A.M. Gronenborn, Nat Struct Biol, 1997. 4 Suppl: p. 849-853. 19. Clore, G.M. and A.M. Gronenborn, Curr Opin Chem Biol, 1998. 2(5): p. 564-70. 20. Pervushin, K., et al., Proceedings of the National Academy of Sciences of the United States of
America, 1997. 94(23): p. 12366-12371. 21. LeMaster, D.M., Methods in Enzymology, 1989. 177: p. 23-43. 22. LeMaster, D.M., Annual Review of Biophysics and Biophysical Chemistry, 1990. 19: p. 243-
266. 23. LeMaster, D.M., Quarterly Reviews of Biophysics, 1990. 23(2): p. 133-174. 24. Venters, R.A., et al., Journal of Molecular Biology, 1996. 264(5): p. 1101-1116. 25. Salzmann, M., et al., Journal of the American Chemical Society, 1999. 121(4): p. 844-848. 26. Salzmann, M., et al., Proceedings of the National Academy of Sciences of the United States of
America, 1998. 95(23): p. 13585-13590. 27. Yang, D.W. and L.E. Kay, Journal of the American Chemical Society, 1999. 121(11): p. 2571-
2575. 28. Konrat, R., D.W. Yang, and L.E. Kay, Journal of Biomolecular Nmr, 1999. 15(4): p. 309-313. 29. Tugarinov, V., et al., Journal of the American Chemical Society, 2002. 124(34): p. 10025-
10035. 30. Salzmann, M., et al., J. Am. Chem. Soc., 2000. 122(31): p. 7543-7548. 31. Eletsky, A., A. Kienhofer, and K. Pervushin, Journal of Biomolecular Nmr, 2001. 20(2): p. 177-
180. 32. Eletsky, A., et al., Journal of Biomolecular Nmr, 2002. 24(1): p. 31-39. 33. Hu, K.F., A. Eletsky, and K. Pervushin, Journal of Biomolecular Nmr, 2003. 26(1): p. 69-77. 34. Fernandez, C., et al., Febs Letters, 2001. 504(3): p. 173-178. 35. Arora, A., et al., Nature Structural Biology, 2001. 8(4): p. 334-338. 36. Pervushin, K., D. Braun, and K. Wuthrich, J. Biomol. NMR, 2000. 17: p. 195-202. 37. Fernandez, C., et al., Proceedings of the National Academy of Sciences of the United States of
America, 2002. 99(21): p. 13533-13537. 38. Hwang, P.M., et al., Proc. Natl. Acad. Sci. USA., 2002. 99(21): p. 13560-13565.
28
39. Sanders, C.R., et al., Biophysical Journal, 1998. 74(2): p. A130-A130. 40. Sanders, C.R. and K. Oxenoid, Biophysical Journal, 2002. 82(1): p. 1721. 41. Vinogradova, O., et al., Biophysical Journal, 1997. 72(6): p. 2688-2701. 42. Schubert, M., et al., Chembiochem, 2002. 3(10): p. 1019-1023. 43. Pellecchia, M., et al., Nature Structural Biology, 1999. 6(4): p. 336-339. 44. Takahashi, H., et al., Nature Structural Biology, 2000. 7(3): p. 220-223. 45. Frickel, E.M., et al., Proceedings of the National Academy of Sciences of the United States of
America, 2002. 99(4): p. 1954-1959. 46. Cavanagh, J., et al., Protein NMR Spectroscopy: Principles and Practice. 1996, New York:
Academic Press. 47. Pervushin, K., G. Wider, and K. Wuthrich, Journal of Biomolecular Nmr, 1998. 12(2): p. 345-
348. 48. Pervushin, K., et al., Journal of Biomolecular Nmr, 2000. 16(1): p. 39-46. 49. Dingley, A.J., et al., Journal of the American Chemical Society, 1999. 121(25): p. 6019-6027. 50. Majumdar, A., et al., Journal of Biomolecular Nmr, 1999. 15(3): p. 207-211. 51. Majumdar, A., A. Kettani, and E. Skripkin, Journal of Biomolecular Nmr, 1999. 14(1): p. 67-70. 52. Wohnert, J., et al., Nucleic Acids Research, 1999. 27(15): p. 3104-3110. 53. Scheurer, C. and R. Bruschweiler, Journal of the American Chemical Society, 1999. 121(37): p.
8661-8662. 54. Wagner, G., Current Opinion in Structural Biology, 1993. 3(5): p. 748-754. 55. Boisbouvier, J., et al., Journal of Biomolecular Nmr, 1999. 14(3): p. 241-252. 56. Loria, J.P., M. Rance, and A.G. Palmer, Journal of Biomolecular Nmr, 1999. 15(2): p. 151-155. 57. Riek, R., et al., Proceedings of the National Academy of Sciences of the United States of
America, 1999. 96(9): p. 4918-4923. 58. Riek, R., et al., Journal of the American Chemical Society, 2002. 124(41): p. 12144-12153. 59. Fiaux, J., et al., Nature, 2002. 418(6894): p. 207-211. 60. Untidt, T.S. and N.C. Nielsen, Journal of Chemical Physics, 2000. 113(19): p. 8464-8471. 61. Nielsen, N.C., H. Thogersen, and O.W. Sorensen, Journal of Chemical Physics, 1996. 105(10):
p. 3962-3968. 62. Levitt, M.H., Journal of Magnetic Resonance, 1997. 126(2): p. 164-182. 63. Sattler, M., et al., Journal of Biomolecular Nmr, 1995. 6(1): p. 11-22. 64. Yang, D.W. and L.E. Kay, Journal of Biomolecular Nmr, 1999. 13(1): p. 3-10. 65. Pervushin, K., et al., Proceedings of the National Academy of Sciences of the United States of
America, 1999. 96(17): p. 9607-9612. 66. Gueron, M., J.L. Leroy, and J. Griffey, J. Am. Chem. Soc., 1983. 105(25): p. 7262-7266. 67. Goldman, M., Journal Magnetic Resonance, 1984. 60: p. 437-452. 68. Farrar, T.C. and T.C. Stringfellow, eds. Encyclopedia of NMR: Relaxation of Transverse
Magnetization for Coupled Spins. Encyclopedia of NMR, ed. D.M. Grant and R.K. Harris. Vol. 6. 1996, Wiley: New York. 4101-4107.
69. Shimizy, H., Journal of Chemical Physics, 1964. 40(11): p. 3357-3364. 70. Blicharski, J.S., Physics Letters, 1967. 24(11): p. 608-610. 71. Blicharski, J.S., Acta Physica Polonica, 1972. 41: p. 223-236. 72. Vold, R.R. and R.L. Vold, Journal of Chemical Physics, 1975. 64(1): p. 320-332. 73. Vold, R.R. and R.L. Vold, Progress in NMR Spectroscopy, 1978. 12: p. 79-133. 74. Mayne, C.L. and S.A. Smith, eds. Encyclopedia of NMR: Relaxation Processes in Coupled-Spin
Systems. Encyclopedia of NMR, ed. D.M. Grant and R.K. Harris. Vol. 6. 1996, wiley: New York. 4053-4071.
75. Sorensen, O.W., et al., Progress in Nuclear Magnetic Resonance Spectroscopy, 1983. 16: p. 163-192.
76. Roberts, J.E., et al., Journal of the American Chemical Society, 1987. 109(14): p. 4163-4169. 77. Gerald, R., et al., Journal of the American Chemical Society, 1993. 115(2): p. 777-782.
29
78. Ramamoorthy, A., C.H. Wu, and S.J. Opella, Journal of the American Chemical Society, 1997. 119(43): p. 10479-10486.
79. Wu, C.H., et al., Journal of the American Chemical Society, 1995. 117(22): p. 6148-6149. 80. Kontaxis, G., G.M. Clore, and A. Bax, Journal of Magnetic Resonance, 2000. 143(1): p. 184-