Apart from solving solution structures of proteins and protein-ligand complexes, NMR can determine and characterize flexibility and motion ("dynamics") in proteins. These dynamical features can be resolved to the level of individual amino acid residues. They include very fast motions on the picosecond to nanosecond timescales, but also slower motions on the microsecond to millisecond timescale. Whereas fast motions are sometimes correlated to increased temperature factors in crystallographic measurements, slower motions are hardly detectable by crystallography, but are often biologically most relevant (81,86,87). In fact, conformational flexibility on the microsecond to millisecond timescale is essential for the catalytic action of many enzymes, and during enzymatic activity, these very motions have recently been detected (88,89).
Kinases are enzymes, and their ability to perform their catalytic action depends on conformational flexibility. There may even be more importance in kinase dynamics: given the substrate specificity of protein kinases in spite of their abundance, dynamics could be the key to achieving selectivity. It is conceivable that kinases could not attain such a high degree of selectivity if their structure was static, and that only conformational freedom and motions accommodate or reject substrates according to the biological function of the kinase. Analogous considerations may be true for the specificity of kinase inhibitors.
Given the high relevance of protein dynamics for the control of biological processes, it is surprising that protein kinases have not yet been subjected to dynamical investigations by NMR. To the best of our knowledge, there are no literature reports on solution structure or dynamics of protein kinases (status January 2004). Therefore, we have recently initiated efforts to investigate by NMR the solution structure and dynamics of c-Abl kinase, currently one of the most important kinases for pharmaceutical research, and the molecular target for the most innovative therapeutical treatment, Gleevec. The development of resistance in late-stage CML patients, largely because of mutations in Abl kinase, has led to a desire to develop follow-up compounds for the treament of these cases. Knowledge about the solution structure and dynamics of Abl in the presence and absence of inhibitors can be a valuable complement to the X-ray structures (see Subheading 3.1.2.)
Knowledge about protein structure and dynamics can come from NMR, but needs isotopically labeled proteins. For a detailed analysis, double or triple isotopic labeling is necessary, because assignment of the protein resonance is a prerequisite. Resonance assignment is the process of identifying which resonances in the NMR spectrum originate from which amino acid residue in the protein. Typically, protein resonance assignment is achieved from uniformly 15N,13C- (for proteins up to 25 kDa) or 15N,13C,2H-labeled- (for larger proteins) proteins, using the well-known sequential resonance assignment method on the basis of triple resonance NMR experiments (90,91). This poses severe problems for NMR work with kinases: most kinase catalytic domains cannot be readily expressed in E. coli, so that isotopic labeling could be achieved by growing the bacteria in isotopically enriched minimal medium (92). Instead, eukaryotic expression systems have to be chosen, such as Baculovirus-infected insect cells.
The drawback of this expression system is that media for protein isotope labeling are not yet commercially available and are expected to be quite expensive.
In order to facilitate NMR studies of kinase catalytic domains anyway, we have developed a novel protocol for amino-acid type selective isotope labeling in Baculovirus-infected insect cells (24). By this method, all residues of a particular amino acid type (e.g., Phe, Tyr, or Val) are isotopically labeled, whereas all other residues in the expressed protein are unlabeled. This is achieved by establishing culture conditions in which one labeled amino acid is added as 15N- (or 13C-), and all others are unlabeled. The insect cells then incorporate the labeled amino acid into the protein. This method was successfully applied to c-Abl, and four samples of the c-Abl/Gleevec complex were obtained in which Phe, Tyr, Val, or Gly were selectively 15N-labeled (see Fig. 10). Isotope costs for several milligrams of these selectively labeled samples are often only a few hundred dollars. All samples were of high purity with a high label incorporation rate.
Full backbone dynamics data can be extracted from these samples. However, resonance assignment is not straightforward because the sequential assignment procedure does not work in selectively labeled proteins. A possible solution is the preparation of dual-labeled samples in which amino acid Yyy is 15N-labeled, and amino acid Xxx is 13CO-labeled. If Xxx and Yyy are chosen so that an Xxx-Yyy pair occurs at an interesting position, the adjacent 13CO-15N pair can be detected in an HNCO experiment, so that the 15N/1H resonances of Yyy are assigned (93). Alternatively, we developed a strategy for assignment of selectively labeled protein complexes. This strategy is based on the strong relaxation enhancement of protons that are in vicinity to a paramagnetic center ("spin label"). In practice, the ligand (Gleevec in our case) is chemically modified to incorporate a paramagnetic moiety such as TEMPO at a defined position. Measurement of paramagnetic relaxation enhancements are then used to measure distances between protein protons and the paramagnetic center. Because the structure of the complex is known, resonance assignments can be deduced (94).
These novel methods are currently applied in our laboratories to characterize solution structure and dynamics of c-Abl complexes. Detailed results will be presented elsewhere.
Was this article helpful?