Structure calculation of biomolecules by NMR relies on unambiguous assignment of all protons in the biomolecule so as to get enough inter-proton distance contraints (from NOEs). Two protons that are in close proximity (less than 5 ang) through space will have a dipolar interaction that is manifested as the NOE (Nuclear Overhauser Effect). More distance constraints from NOEs - better the structure. This is the traditional approach.
As you go to larger macromolecules, there are two main problems - first the number of protons in the molecule increases resulting in spectral overlap - so there is a problem of resolution - being able to resolve two proton frequencies that lie very close to each other or may overlap perfectly. The problem of resolution can be overcome by isotopic labeling (incorporating 15N, 13C, specific labeling of amino acids), going to higher field strengths of NMR magnets for data collection, and greater dimensionality of spectra (3D and 4D data are now routine). The main problem however is that of rotational correlation time which increases with molecular weight. Larger molecules tumble more slowly, thereby increasing the correlation time and decreasing the relaxation rates of nuclei, especially protons. The relaxation time constant T2, is inversely proportional to the apparant linewidth of NMR resonances - larger the linewidth, more the degeneracy in the spectra. In addition, as the T1 and T2 times get shorter, there is less magnetization left at the end of a 3D experiment for detection, since most of the magnetization has already decayed during the experiment.
There are some NOE-independent methods (such as the use of residual dipolar couplings) or recent hybrid methods that rely on chemical shifts to overcome this problem. Perdeuteration also helps to increase the relaxation times. Currently 25-30 kDa is the upperlimit for structure calculation by NMR, although there are a few NMR structures solved for very large proteins or complexes where the spectral dispersion is excellent and they are stable enough that one can raise the temperature to increase the correlation time. These structures also relied heavily on NOE-independent approaches.
The power of NMR is of course is in studying dynamics, and NMR has been used to study motions in some large systems (the 670 kDa proteasome is one example) where it was possible to get sequence specific assignments for a region in the protein.
Hope that helps -
Roopa
******************************************
Roopa Thapar
Research Scientist
Hauptman-Woodward Institute
Assistant Professor, SUNY, Buffalo
700 Ellicott Street
Buffalo, NY 14203
Tel: 716-898-8687 (Office); 716-898-8659 (Lab)
*****************************************************
________________________________________
From: CCP4 bulletin board [[log in to unmask]] On Behalf Of Artem Evdokimov [[log in to unmask]]
Sent: Thursday, June 30, 2011 11:43 PM
To: [log in to unmask]
Subject: Re: [ccp4bb] nmr question
There are about 20 structures solved by NMR with chain lengths above 300 residues. Some of them are solved by combination of restraints and modeling (e.g. 2010 Nature paper describing implementation of Rosetta modeling coupled with backbone-only data, by Raman/Montelione/Baker et al.).
Principal problems from the perspective of a non-specialist are:
a) separation between resonances - complete (multidimensional!) overlaps are ruinous for assignment
b) intensity of each resonance - for larger molecules there is less moles of resonating nuclei
c) relaxation issues in large proteins (the need for expensive and hard to do deuterium labeling is the result of this)
I am sure that true NMR experts would have much more to say on the subject :)
Artem
On Thu, Jun 30, 2011 at 5:20 PM, Lena Griese <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Dear members,
I know that it is not possible to solve a structure by nmr of more than approx. 30 kda. But I have to admit that I dont know why. What exactly is overlapping?
With best regards,
Lena
|