NOESY-Based Strategy for Assignments of Backbone and Side Chain Resonances of Large Proteins without Deuteration (original) (raw)

Daiwen Yang , dbsydw@nus.edu.sg, Department of Biological Sciences, National University of Singapore, 14 Science Drive 4, Singapore 117543. Tel: +65-65161014.

Related Journal & Article Information

Manuscript: A02152C

Journal: Nature Methods

NMR spectroscopy has been widely used to determine high resolution structures of proteins and protein complexes in solution1, 2. However, it is is not suitable for larger proteins (> 30 kDa) due to increased signal decay rates. Deuterium labeling of protein samples overcomes this problem3, but deuteration makes sample preparation much harder and structure determination more difficult4, 5. Here we introduce a strategy that uses non-deuterated samples6. Backbone and side chain resonances of large proteins are assigned from 3D TROSY-HNCA7, 3D MQ-CCH-TOCSY8 and 4D 15N,13C-edited NOESY9 spectra. The three spectra with reasonable quality can be obtained within ~11 days using a uniformly 13C,15N-labeled sample at a protein concentration of ~1 mM. Unlike the conventional strategy mainly relying on through-bond correlation experiments, the strategy used here mainly uses a through-space correlation experiment (4D 15N,13C-edited NOESY) to obtain intra-residue and sequential correlations, which are separated from other inter-residue NOE correlations using the 3D HNCA and CCH-TOCSY experiments. Similar to the conventional strategy, sequential assignment is achieved from matching intra-residue and sequential correlations. Using this strategy, one can obtain nearly complete backbone assignments and >80% side chain assignments for monomeric proteins < 42 kDa and multimeric proteins < 65 kDa.

• 13C and 15N labeled protein sample with ~1 mM monomer and 5-10% D2O; if a cryoprobe is available, lower concentration is needed

• 800 MHz NMR spectrometer; for multimeric proteins with less complexity, a 500 MHz or higher field machine equipped with a cryoprobe can be used as an alternate
• A personal computer (Linux operating system)
• NMRPipe and Sparky softwares; other softwares with similar functions can be used

Recording data • TIMING ~11 days
1. Record 3D TROSY-HNCA, 3D MQ-CCH-TOCSY and 4D 13C, 15N-edited NOESY for resonance assignment. Use half-point delays for the evolution periods of Cα spins (t1) in the HNCA, 13C spins (bonded with the directly detected 1H, t2) in the MQ-CCH-TOCSY, and both 13C and 15N spins (t2 and t3) in the 4D NOESY. In this way, the peaks aliased with odd times have opposite signs to those aliased with even times. Record a 2D 1H-15N TROSY-HSQC with enough spectral width in the 15N dimension to facilitate the identification of peaks aliased in the 15N dimension in 3D HNCA and 4D 13C, 15N-edited NOESY.

Processing data • TIMING ~1 day
2. Process the data with NMRPipe. Use linear prediction to improve the resolution of all indirectly detected dimensions.
Covert the processed spectra into ucsf format for further analysis with Sparky.

Picking peaks and grouping peaks into clusters • TIMING ~1-2 days
3. Use peak picking functions in Sparky to pick HNCA peaks automatically. Group them into clusters according to (HN, N) chemical shifts. Usually one cluster contains two peaks: one from the intraresidue Cα, another from the Cα of the preceding residue. However, some clusters contain only one peak due to degenerate Cα chemical shifts, while some clusters may contain more than two peaks due to the overlap of amide chemical shifts. Identify peaks from NH2 groups according to their paired patterns and unusual “Cα” chemical shifts. These peaks can be excluded during backbone assignment.
Check the number of clusters identified and the expected number and ensure that there is not a large difference.
4. Pick peaks in 4D 13C,15N-edited NOESY manually or automatically. Group NOE peaks into their corresponding clusters according to (HN, N) chemical shifts of individual clusters. Due to lower resolution of the 4D NOESY, some peaks cannot be grouped into only one cluster unambiguously. These peaks will remain ungrouped.
5. Correct chemical shifts of the aliased peaks in 3D TROSY-HNCA and 4D NOESY with Sparky.
6. Pick peaks in CCH-TOCSY.

Identifying and classifying spin-systems • TIMING ~5-10 days
7. Each cluster is characterized by a pair of (HN, N) chemical shifts. Typically, one cluster contains two HNCA peaks and a number of HC-NH NOE peaks. First, initialize a spin-system from one HNCA peak in a given cluster. Second, among the NOE peaks in this cluster, identify the peak that matches the HNCA peak in 13C chemical shift within a certain tolerance (e.g., ~0.3 ppm). If only one peak is identified, this peak is considered as a sequential or intra-residue CαHα-NH NOE10. Note that there are two α protons for glycine residues and thus two peaks with identical Cα shifts may be observed. Third, display the TOCSY slice defined by the (H, C) chemical shifts of each NOE peak in this cluster and check if the CH spin-pair of the NOE peak correlates with the Cα spin of this initial spin-system. If yes, the NOE peak is regarded as a sequential or intraresidue side chain NOE in this spin-system. The 13C spin of the CH spin-pair should also correlate with CαHα of this spin-system via TOCSY. In some cases, this correlation cannot be detected because it is too weak or close to the intense water signal. For spin-systems of long side chains, a few more CH spin-pairs may be added into the spin-systems according to their TOCSY correlations with side chain CH spin-pairs that are already identified.
8. In cases where only one HNCA peak is picked for one cluster due to degenerate Cα chemical shifts and two NOE peaks match this HNCA peak, build two spin-systems using the two NOE peaks. The NOE peaks belonging to these two spin-systems can be separated because side chain 13C spins correlate with two non-degenerate CαHα spin-pairs in CCH-TOCSY, provided that 13C-CαHα correlations are detectable.
9. For each spin-system, check whether the CH spin-pairs correlate with each other in CCH-TOCSY and whether the correlations can be explained by one of the 20 amino acids in terms of number of peaks and chemical shifts.
10. Classify a spin-system by residue type according to its Cα chemical shift and NOE peaks. Typically it is possible to unambiguously recognize most glycine, alanine, threonine, valine, isoleucine, and leucine, which account for ~42% amino acid residues in a protein. Glycine can be identified according to its small Cα chemical shift, and sometimes two Hα peaks. Alanine can be identified based on its strong intraresidue CβHβ-NH NOE peak, strong Cα-CβHβ TOCSY peak, and characteristic Cα and Cβ chemical shifts. For threonine, valine, isoleucine and leucine, classification can be easily done based on their characteristic side chain chemical shifts. In particular, the methyl groups of these residues always give rise to well resolved TOCSY correlations with other 13C spins. Some serine, lysine, arginine, and proline can be distinguished, provided that the CβHβ spin-pair for serine, CβHβ, CγHγ and CδHδ spin-pairs for lysine, arginine and proline can be identified. Classify other residues into three groups: (Asp, Asn, Phe, Tyr), (Gln, Glu, Met) and (Cys, Trp, His) according to both 1H and 13C chemical shifts.
If a spin-system contains too few peaks to be classified, leave it without classification. Usually classification of about ~35-50% spin-systems is sufficient for obtaining backbone assignment.

Matching spin-systems to form fragments and Mapping fragments • TIMING ~1-3 days
11. Link clusters to build fragments through matching spin-systems. Two spin-systems match each other when they have i). identical Cα chemical shifts, ii). the largest number of matched NOE peaks, and iii). no inconsistent (C, H) chemical shifts.
12. Compare spin-system types of a connectivity fragment with the protein sequence. If only one segment in the sequence matches the fragment in residue type at every amino acid position, map this fragment onto the segment of the sequence. Note that it is unknown which end of the fragment is the N-terminus since intraresidue and sequential spin-systems are not distinguished. The intraresidue HNCA peak is usually stronger than the sequential one. If a mapping is inconsistent with this, it may contain errors and need to check more carefully.
13. Use common NOEs shared by two clusters to resolve ambiguities in sequential connectivities for unassigned spin-systems6. If two spin-systems have a true sequential relationship, no mismatch of intra-residue and sequential correlations should exist. On the basis of this fact, some ambiguities can be removed too.
14. Repeat steps 11-13 and assign the unassigned fragments and clusters in an iterative way.
15. Note that some initial errors may be introduced during the processes of identifying, classifying and matching spin-systems and mapping fragments to the sequence. Most errors can be recognized during the process of assignment and will not lead to wrong final assignments. Those lead to wrong initial assignments should be recognized when some other clusters can’t be assigned6.

Side chain assignments • TIMING ~2-7 days
16. Most side chain chemical shifts are ready to be extracted according to the identified spin-systems and backbone assignments. Assign some more side chain resonances by comparing two 13C,15N-edited NOESY planes defined by two adjacent amides, and by exploring the MQ-CCH-TOCSY more extensively10.
17. Assign NH2 chemical shifts according to i) Cβ-NH2 and Cγ-NH2 correlations in the HNCA spectrum for asparagine and glutamine respectively, ii) HβCβ-NH2 and HγCγ-NH2 /HβCβ-NH2 correlations in the 4D 13C,15N-edited NOESY for asparagine and glutamine respectively.
18. Assign CδHδ groups of aromatic residues according to their NOE correlations with intraresidue and sequential amides11. Other aromatic resonances can be assigned later with a 4D 13C,13C-edited NOESY ( ~7 days for data collection) or a 3D 15N/13C-simutaneous NOESY ( ~3 days for data collection).

NOE assignment and structure determination • TIMING ~2-6 weeks
19. With a relatively complete chemical shift list, assign other NOE peaks in the 4D 13C,15N-edited NOESY and 13C,13C-edited NOESY. Follow the conventional procedure for NOE assignment and structure calculation. The chemical shifts of a few more side chain atoms may be assigned during this process.

Some amide groups have degenerate (HN, N) chemical shifts, which lead to some clusters containing more than one residues. These clusters can be identified since they contain more than two HNCA peaks or more than one HNCO peak. They are treated as normal clusters during spin-system identification. However, when linking clusters through matching spin-systems, we place these clusters at the ends of fragments. In this way, each spin-system in such a cluster is linked and mapped independently. When the number of degenerate amides is so large that many fragments are too short to be mapped onto the sequence, backbone assignment may be difficult to be obtained.
For very large proteins, the CCH-TOCSY may give weak signals for many correlations, especially for those involved in Hα and Hβ spins, and the 4D 13C,15N edited-NOESY may also lack many intraresidue and sequential NOE peaks. This leads to spin-systems with very few NOE peaks, which are difficult to be classified and matched unambiguously. In such cases, one may rely on Cα chemical shifts and common NOE peaks between two clusters, although there will be more ambiguity in spin-system matching and mapping.

For monomeric proteins smaller than ~42 kDa or multimeric proteins smaller than ~65 kDa, the 3D and 4D spectra used in our method can be obtained with reasonable quality on an 800 MHz NMR spectrometer equipped with a normal probe using a sample of 13C,15N-labeled protein at a monomer concentration of ~1 mM. The number of the clusters obtained from the spectra should be very close to the expected one. Most spin-systems generated from the spectra should contain at least one HC-NH NOE peak. ~60% spin-systems can be classified by residue types. ~50% spin-systems can be unambiguously matched to their corresponding partners. Nearly complete backbone assignment can be obtained, provided that all amides give rise to 1H-15N HSQC correlations. On the basis of the backbone assignment, >80% side chain resonances can be assigned from the 3D MQ-CCH-TOCSY and 4D 13C,15N-edited NOESY spectra. With these assignments, NOE assignments should be possible and a high resolution structure can be determined.

1. Wuthrich, K. NMR of Proteins and Nucleic Acids, (Wiley, New York, 1986).
2. Cavanagh, J., Palmer, A. G., Fairbrother, W. & Skelton, N. Protein NMR Spectroscopy: Principles and Practice (Academic, San Diego, 1996).
3. LeMaster, D.M. & Richards, F.M. NMR sequential assignment of Escherichia coli thioredoxin utilizing random fractional deuteriation. Biochemistry 27, 142-150 (1988).
4. Tugarinov, V., Kanelis, V. & Kay, L. E. Isotope labeling strategies for the study of high-molecular-weight proteins by solution NMR spectroscopy. Nature Protocols 1, 749-754 (2006).
5. Kainosho, K. et al. Optimal isotope labelling for NMR protein structure determinations. Nature 440, 52-57 (2006).
6. Xu, Y.Q., Zheng, Y., Fan, J.S. & Yang, D.W. A novel strategy for solution structure determination of large proteins without deuteration. Nature Methods (in press).
7. Yang, D.W. & Kay, L.E. TROSY triple-resonance four-dimensional NMR spectroscopy of a 46 ns tumbling protein. J. Am. Chem. Soc. 121, 2571-2575 (1999).
8. Yang, D.W., Zheng, Y., Liu, D. & Wyss, D.F. Sequence-specific assignments of methyl groups in high-molecular weight proteins. J. Am. Chem. Soc. 126, 3710-3711 (2004).
9. Kay, L.E., Clore, G.M., Bax, A. & Gronenborn, A. M. Four-dimensional heteronuclear triple-resonance NMR spectroscopy of Interleukin-1β in solution. Science, 249, 411-414 (1990).
10. Xu, Y.Q., Lin, Z., Ho, C. & Yang, D.W. A general strategy for the assignment of aliphatic side-chain resonances of uniformly 13C,15N-labeled large proteins. J. Am. Chem. Soc. 127, 11920-11921 (2005).
11. Lin, Z., Xu, Y.Q., Yang, S. & Yang, D.W. Sequence-specific assignment of aromatic resonances of uniformly 13C,15N-labeled proteins by using 13C- and 15N-edited NOESY spectra. Angew. Chem. Int. Ed. 45, 1960-1963 (2006).

NMR, assignment, large protein, NOESY, protein structure