|Home | About | Journals | Submit | Contact Us | Français|
Nuclear magnetic resonance (NMR) is a versatile technique providing structural, dynamic and kinetic information on a variety of biological systems. All these applications rely on the ability to assign the signals observed in the NMR spectra to the nuclei of the molecule under investigation. For backbone nuclei of proteins, this is traditionally achieved by recording a series of triple resonance experiments that each correlate an amide proton-nitrogen pair with carbon nuclei belonging to sequentially adjacent residues (e.g. Cα(i) and Cα(i-1)).1 Strips along the carbon dimension are generated and are then compared for matching carbon frequencies. In many challenging systems, such as alpha-helical or unfolded proteins or molecules with undiversified amino-acid sequences (e.g. ...QQEQQ... or SDLA...SDLA...), this task becomes tedious and often unsuccessful due to spectral crowding and resonance degeneracies, in particular for carbon nuclei. In this case, other methods are used to supplement or supplant the conventional suite of experiments. Among these are the hNcocaNH and HncocaNH pair of experiments (a.k.a. (H)N(COCA)NNH2 or HN(C)N 3 and H(NCOCA)NNH4 or HNN3), which allow for a simple and efficient “HN-N backbone walk” assignment procedure that directly correlates HN and N nuclei.5 However, this necessitates the acquisition of two NMR spectra. A recently described alternative method exploits redundancies in the HNCA and HACAN experiments to provide a “stairway” assignment procedure in which chain extension of sequential residues is accomplished by navigating between the 1H-13C planes of the 3D spectrum.6 This hybrid spectrum can be obtained with a single time-shared experiment. Resonance degeneracy is also a limitation for large proteins, which feature an increased number of residues. Large proteins in addition suffer from faster relaxation resulting in line-broadening and signal losses during spin manipulations and evolution periods in the experiments. Thus, while additional experiments such as those described above are needed to overcome ambiguities in assignment, their application is limited by these constraints.
Here, we present a novel NMR experiment that exhibits visually prominent spectral patterns and enables facile sequential backbone assignments in a single experiment. The hNCAnH pulse sequence is a modification of the recently developed TROSY-hNcaNH-TROSY7, which has been used successfully to assign large proteins 8. The major difference between these two experiments is that the chemical shift evolution of nitrogens attached to the detected nuclei (which would provide an HN-HSQC type of correlation) is replaced by a Cα constant time evolution. This modification creates easily recognizable rectangular spectral patterns in the 1H-15N planes and allows for easy and unambiguous sequential assignment of HN, N and Cα nuclei in a single experiment. Several versions of the pulse sequence, appropriate for proteins of different sizes, are discussed in the Supporting Information. For large proteins, the new experiment was designed to overcome relaxation losses by using the double TROSY method7, which not only minimizes transverse relaxation for a pair of chemically bound HN and N nuclei, as in conventional TROSY, but also for nuclei that are linked by a 3D cross peak but are not directly connected by a single scalar coupling. This results in sensitivity which is about half that of the HNCACB experiment. The magnetization flow can be summarized as (see Supporting Information):
Here, H denotes HN and C denotes Cα. A summary of the coherence flow between correlated nuclei is shown in Figure 1.
Offsetting all source protons by one residue (Supporting Information) reveals that, at a given carbon frequency, four cross-peaks arranged in a rectangular pattern are to be expected in each H/N plane: (HNi,Ni,Cαi), (HNi,Ni+1,Cαi), (HNi+1,Ni+1,Cαi) and (HNi+1,Ni,Cαi). Sequential correlation peaks along both H and N dimensions provide unambiguous identification of proton and nitrogen signals of successive residues in a manner reminiscent of NOESY cross-peaks (HN-N backbone walk). Since sequential information also appears in the carbon dimension, a cross-section along C, at (HNi,Ni) enables the identification of the Cα frequency of the previous residue (stairway). The H/N plane at the resulting Cαi-1 frequency can then be analyzed as described above, leading to the assignment of HNi-1 and Ni-1.
The advantages of the method are illustrated with the 33 kDa, 297 residue component B of the E.coli enterobactin synthetase. 9 In addition to its relatively large size, the protein is mainly alpha-helical10 and subject to spectral overlap. Conventional strategies assigned 124 residues out of 239 (picked in HNCA). With the hNCAnH experiment, 151 signals were assigned, two mistakes were corrected and 11 additional connectivities were unambiguously established. Ambiguities in strip comparisons often require multiple hypotheses (that need to be verified or corrected a posteriori) involving knowledge of the sequence, only to establish a connectivity. The hNCAnH identifies sequential residues without a need to know the amino acid sequence.
The main advantage of the new experiment is the relative ease and speed of assignment based on striking visual patterns. This is demonstrated in Figure 2, which shows an example of chain elongation in EntB. First, we choose the (H, N) coordinates of a given residue, e.g. those subsequently assigned to Glu 220. We then look at the H/N plane of the hNCAnH experiment, at the frequency of the corresponding alpha carbon, Cα220. The sequential cross-peaks (with opposite phases) enable the identification of the (H, N) coordinates of the successor, Val 221 (a). A cross-section along C at (HN221, N221) reveals the frequency of Cα221. The corresponding H/N plane is then visualized (b) and displays the correlations between (HN221, N221) and (HN222, N222). As can be judged from the number of peaks in each plane, a number of degeneracies exist along the carbon dimension. In this example, conventional strips comparison would not allow for a determination of successive residues (Figure 2c). The power of the technique presented here relies in its ability to identify which carbons belong to sequentially neighboring residues, as opposed to chance correspondence of their frequencies.
The hNCAnH experiment enables the direct sequential connection of 1H-15N HSQC or TROSY-HSQC cross peaks and delivers the 13Cα assignments in addition. It facilitates the assignment of crowded spectra and enables NMR studies of systems heretofore inaccessible. For large systems, it is a useful supplement to the existing pool of triple-resonance experiments. For any system size, it provides a trivial and visually-simple assignment procedure. For smaller proteins, the hNCAnH can be used to entirely replace conventional backbone assignment experiments, such as HNCA, HNCACB or HNCACO, and save time. The experiment promises spectacular results when non-uniform sampling is used to obtain ultra-high resolution spectra.11 Because long transfer delays are used in the experiment, high resolution can be achieved with little reduction in sensitivity. For very large systems, we have developed an alternative CANCA 13C detected version.12 The perdeuterated protein can then be dissolved in D2O, so that all nuclei are subject to small relaxation rates, which further extends the limit in size of systems that can be assigned. This is however only beneficial when relaxation losses in the hNCAnH surpass the polarization advantages of proton detected experiments. Overall, the hNCAnH experiment provides a new prospective in studies of many systems such as large proteins, a number of alpha-helical proteins and unfolded proteins.
This research was supported by the National Institutes of Health (grants GM47467, AI37581 and RR00995) and a Human Frontier Science Program long-term fellowship from 2007-2009 (A.K.). A.K. is a Director's fellow of the Los Alamos National Laboratory awarded in 2009.
Supporting Information Available: This contains details of sample preparation, a description of the pulse sequence with alternative versions and further discussions. This material is available free of charge via the Internet at http://pubs.acs.org. The pulse sequences can be downloaded at http://gwagner.med.harvard.edu.