Journal of Biomolecular NMR, 16: 235–244, 2000. KLUWER/ESCOM © 2000 Kluwer Academic Publishers. Printed in the Netherlands. 235 Improved photo-CIDNP methods for studying protein structure and folding Kiminori Maeda a,∗ , Charles E. Lyon a , Jakob J. Lopez a , Masa Cemazar a , Christopher M. Dobson b & P.J. Hore a,∗∗ Oxford Centre for Molecular Sciences, a Physical and Theoretical Chemistry Laboratory, and b New Chemistry Laboratory, Oxford University, Oxford OX1 3QZ, U.K. Received 21 October 1999; Accepted 21 December 1999 Abstract Two new techniques offering considerable improvements in the quality of 1 H photo-CIDNP spectra of proteins are demonstrated. Both focus on the problem of progressive photo-degradation of the flavin dye used to generate polarization in exposed tryptophan, tyrosine and histidine side-chains. One approach uses rapid addition and removal of protein/flavin solution between light flashes to mix the NMR sample and introduce fresh dye into the laser-irradiated region. The other involves chemical oxidation of photo-reduced flavin by the addition of hydrogen peroxide. In both cases a larger number of scans can be accumulated before the flavin is exhausted than would otherwise be possible. The techniques are demonstrated by 600 MHz CIDNP-NOESY spectroscopy of bovine holo-α-lactalbumin, and by real-time CIDNP observation of the refolding of bovine apo-α-lactalbumin following rapid dilution from a high concentration of chemical denaturant. Introduction Chemically induced dynamic nuclear polarization (CIDNP), a phenomenon traditionally used to inves- tigate the mechanisms of free radical reactions (Muus et al., 1977; Salikhov et al., 1984), has been exten- sively exploited as a surface probe of protein structure (Kaptein, 1978, 1982; Kaptein et al., 1978; Hore and Broadhurst, 1993). In the presence of a suitable photosensitiser – usually a flavin – laser irradiation generates non-equilibrium nuclear polarization in the side-chains of exposed aromatic amino acid residues. Although the resulting sensitivity enhancements and spectral simplifications are often welcome, the real attraction of the method is that only the histidine, tryptophan and tyrosine side-chains that are physically accessible to the photoexcited flavin are polarizable (Kaptein, 1978, 1982; Kaptein et al., 1978; Hore and Broadhurst, 1993). A recent example of the use of ∗ Permanent address: Department of Chemistry, Graduate School of Science, Tohoku University, Sendai 980-8578, Japan. ∗∗ To whom correspondence should be addressed. E-mail: hore@physchem.ox.ac.uk CIDNP in this way is a stopped-flow experiment in which differential changes in the exposure of tyro- sine and tryptophan residues in hen lysozyme were observed as the denatured protein folded to its native state (Hore et al., 1997). (The investigation of pro- tein folding using real-time NMR has been reviewed by van Nuland et al. (1998) and Dobson and Hore (1998).) Several extensions of the original one-dimensional (1D) ‘laser flash–radiofrequency pulse–acquire’ 1 H CIDNP experiment have been devised. Kaptein and co-workers have developed COSY and NOESY ver- sions in which each repetition of the pulse sequence is preceded by a light flash (Scheek et al., 1984, 1985). More recently, Lyon et al. (1999) have measured 2D 15 N- 1 H heteronuclear CIDNP correlation spectra to reveal the accessibility of tryptophan side-chains in a denatured protein. A serious problem encountered in these experiments, which require prolonged laser irradiation, is the progressive decay of the polarization generated by successive light flashes. The photochem- ical reactions used to produce CIDNP in proteins are cyclic, so that polarization is observed in the intact