793 Research Article Received: 10 March 2010 Revised: 11 May 2010 Accepted: 1 July 2010 Published online in Wiley Online Library: 27 August 2010 (wileyonlinelibrary.com) DOI 10.1002/mrc.2664 Root-mean-square-deviation-based rapid backbone resonance assignments in proteins Ashok K. Rout, Ravi P. Barnwal, Geetika Agarwal and Kandala V. R. Chary ∗ We have shown that the methodology based on the estimation of root-mean-square deviation (RMSD) between two sets of chemical shifts is very useful to rapidly assign the spectral signatures of 1 H N , 13 C α , 13 C β , 13 C ′ , 1 H α and 15 N spins of a given protein in one state from the knowledge of its resonance assignments in a different state, without resorting to routine established procedures (manual and automated). We demonstrate the utility of this methodology to rapidly assign the 3D spectra of a metal-binding protein in its holo-state from the knowledge of its assignments in apo-state, the spectra of a protein in its paramagnetic state from the knowledge of its assignments in diamagnetic state and, finally, the spectra of a mutant protein from the knowledge of the chemical shifts of the corresponding wild-type protein. The underlying assumption of this methodology is that, it is impossible for any two amino acid residues in a given protein to have all the six chemical shifts degenerate and that the protein under consideration does not undergo large conformational changes in going from one conformational state to another. The methodology has been tested using experimental data on three proteins, M-crystallin (8.5 kDa, predominantly β -sheet, for apo- to holo-state), Calbindin (7.5 kDa, predominantly α-helical, for diamagnetic to paramagnetic state and apo to holo) and EhCaBP1 (14.3 kDa, α-helical, the wild-type protein with one of its mutant). In all the cases, the extent of assignment is found to be greater than 85%. Copyright c  2010 John Wiley & Sons, Ltd. Supporting information may be found in the online version of this article. Keywords: NMR; HSQC; M-crystallin; Calbindin (D9K); EhCaBP1; BMRB; RMSD; PCS Introduction Identification of the nuclear spin in the molecule from which a particular signal arises is essential to make use of the information contained in the NMR spectrum. In other words, a 1 : 1 correlation is required between the chemical shifts and the atoms, which contribute to such signals. Such correlation, popularly known as sequence-specific resonance assignment of 1 H, 13 C and 15 N spins is a tedious and time-consuming task. To overcome this problem, a number of automated assignment strategies have been proposed. In particular¸for backbone resonance assignments, the steps involved are often common and normally involve, selecting an appropriate 3D spectral data set, picking peaks from the selected data set, classification of peaks to specific spin systems according to amino acids, and carrying out sequence-specific resonance assignment. The approaches proposed so far differ in the experimental inputs used and the algorithm for the analysis of spectral data. The choice of particular software generally depends on the type and quality of the available spectral data. Most methods utilize information from a suite of standard 3D triple-resonance experiments. These are coupled with algorithms such as simulated annealing, [1,2] Bayesian statistics and artificial intelligence, [3,4] characteristic 13 C α and 13 C β chemical shifts, [5,6] threshold accepting algorithms, [7] connectivity tracing [8] and neural networks. [9,10] Some strategies utilize prediction of chemical shifts for proteins which have high sequence homology to a previously assigned protein. Some of the popularly used programs for automated backbone protein resonance assignments are AUTOASSIGN, [11] GANNA, [12] GARANT, [13,14] IBIS, [15] MAPPER, [16] MARS, [17] PASA [18] and TATAPRO. [19] The precision and accuracy of chemical shifts obtained from the spectral data play an important role in the success of an algorithm for automated assignments. This is primarily because the chemical shifts from different types of spectra and spin systems need to be grouped and matched. Most of the automated assignment protocols consist of an amino acid residue type recognition algorithm and a primary sequence mapping algorithm. In this article, we demonstrate the utility of the approach proposed earlier, [19] which is based on the root-mean-square deviation (RMSD) of chemical shifts, for automated resonance assignment of 1 H α , 1 H N , 13 C α , 13 C β , 13 C ′ and 15 N spins of a given protein in one state from the knowledge of its resonance assignments in a different state. In brief, we demonstrate the utility of this methodology to rapidly assign the 3D spectra of a metal-binding protein in its holo-state from the knowledge of its assignments in apo-state, the spectra of a protein in its paramagnetic state from the knowledge of its assignments in diamagnetic state and, finally, the spectra of a mutant protein from the knowledge of the chemical shifts of the corresponding wild- type protein, without resorting to routine established procedures (manual and automated) mentioned above. Materials and Methods Protein systems The protein systems include Calbindin in its metal-free form (apo-Cb), calcium loaded form ([Ca 2+ ] 2 -Cb) and lanthanide sub- ∗ Correspondence to: Kandala V. R. Chary, Department of Chemical Sciences, Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Mumbai- 400005, India. E-mail: chary@tifr.res.in Department of Chemical Sciences, Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Mumbai-400005, India Magn. Reson. Chem. 2010, 48, 793–797 Copyright c  2010 John Wiley & Sons, Ltd.