Hyperdimensional NMR Spectroscopy with Nonlinear Sampling Victor A. Jaravine, ² Anastasia V. Zhuravleva, ², ⊥ Perttu Permi, ‡ Ilgis Ibraghimov, § and Vladislav Yu. Orekhov* ,², | Swedish NMR Centre, Go ¨teborg UniVersity, Box 465, 40530 Go ¨teborg, Sweden, NMR Laboratory, Structural Biology and Biophysics Program, Institute of Biotechnology, UniVersity of Helsinki, P.O. Box 65, Helsinki FIN-00014, Finland, Saarbru ¨cken UniVersity, Mathematical Department, Saarbru ¨cken D-66041, Germany, and Department of Biological Chemistry and Molecular Pharmacology, HarVard Medical School, 240 Longwood AVenue, Boston, Massachusetts 02115 Received September 20, 2007; E-mail: orov@nmr.gu.se Abstract: An approach is described for joint interleaved recording, real-time processing, and analysis of NMR data sets. The method employs multidimensional decomposition to find common information in a set of conventional triple-resonance spectra recorded in the nonlinear sampling mode, and builds a model of hyperdimensional (HD) spectrum. While preserving sensitivity per unit of measurement time and allowing for maximal spectral resolution, the approach reduces data collection time on average by 2 orders of magnitude compared to the conventional method. The 7-10 dimensional HD spectrum, which is represented as a set of deconvoluted 1D vectors, is easy to handle and amenable for automated analysis. The method is exemplified by automated assignment for two protein systems of low and high spectral complexity: ubiquitin (globular, 8 kDa) and  cyt (naturally disordered, 13 kDa). The collection and backbone assignment of the data sets are achieved in real time after approximately 1 and 10 h, respectively. The approach removes the most critical time bottlenecks in data acquisition and analysis. Thus, it can significantly increase the value of NMR spectroscopy in structural biology, for example, in high-throughput structural genomics applications. Introduction Over the past two decades, NMR spectroscopy has evolved as one of the prime techniques for protein structure determi- nation at the atomic level and for characterizing proteins, protein-ligand complexes, or nucleic acids. X-ray crystal- lography and NMR are two biophysical methods for determining protein structures 1 that have proven to be the most useful in structural genomics, which aims to ascribe a three-dimensional protein structure to each gene product of the human and other genomes. Because these methods rely on distinctly different physical principles and experimental procedures, crystallography and NMR are highly complementary for high-throughput (HTP) structure determination; both are important to ultimate project success. 2-4 When used to determine structure in the pipeline with highly automated and parallelized target selection and protein expression, contemporary NMR often represents a major time bottleneck. 4 Weeks of data collection using an expensive NMR spectrometer are required for every protein target. Measurements are followed by data analysis, which is at least as lengthy and is usually performed manually. The NMR community 5,6 has devoted significant attention to the need to save spectrometer time and to automate the analysis steps. With modern sensitive NMR spectrometer hardware, the duration of a multidimensional experiment is determined by the time needed for one measurement and the number of measure- ments. Both factors are targeted in ongoing efforts to speed up the experiments. Recording individual data points can be accelerated by reducing the delay between consecutive measure- ments 7-12 or by parallel acquisition as in single scan NMR. 13 ² Go ¨teborg University. ‡ University of Helsinki. § Saarbru ¨cken University. | Harvard Medical School. ⊥ Current address: Department of Biochemistry & Molecular Biology 1232P, University of Massachusetts, Amherst, Massachussetts 01003. (1) Brunger, A. T. Nat. Struct. Biol. 1997, 4 (Suppl), 862-865. (2) Yee, A. A.; Savchenko, A.; Ignachenko, A.; Lukin, J.; Xu, X. H.; Skarina, T.; Evdokimova, E.; Liu, C. S.; Semesi, A.; Guido, V.; Edwards, A. M.; Arrowsmith, C. H. J. Am. Chem. Soc. 2005, 127 (47), 16512-16517. (3) Snyder, D. A., et al. J. Am. Chem. Soc. 2005, 127 (47), 16505-16511. (4) Yee, A.; Gutmanas, A.; Arrowsmith, C. H. Curr. Opin. Struct. Biol. 2006, 16 (5), 611-617. (5) Wuthrich, K. Nobel Lecture, Stockholm, 2002. (6) Moseley, H. N. B.; Montelione, G. T. Curr. Opin. Struct. Biol. 1999, 9 (5), 635-642. (7) Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principles of NMR in one and two dimensions; Clarendon: Oxford, 1987. (8) Ross, A.; Salzmann, M.; Senn, H. J. Biomol. NMR 1997, 10 (4), 389- 396. (9) Pervushin, K.; Vogeli, B.; Eletsky, A. J. Am. Chem. Soc. 2002, 124 (43), 12898-12902. (10) Atreya, H. S.; Szyperski, T. Proc. Natl. Acad. Sci. U.S.A. 2004, 101 (26), 9642-9647. (11) Schanda, P.; Brutscher, B. J. Am. Chem. Soc. 2005, 127 (22), 8014-8015. (12) Schanda, P.; Van Melckebeke, H.; Brutscher, B. J. Am. Chem. Soc. 2006, 128 (28), 9042-9043. (13) Frydman, L.; Scherf, T.; Lupulescu, A. Proc. Natl. Acad. Sci. U.S.A. 2002, 99 (25), 15858-15862. Published on Web 03/01/2008 10.1021/ja077282o CCC: $40.75 © 2008 American Chemical Society J. AM. CHEM. SOC. 2008, 130, 3927-3936 9 3927