1.1 Selection of the Labeling Position
Fluorescent adenine and guanine nucleotides have been widely used to report upon binding, protein release and structural changes (6–11). Fluorophores are sensitive probes, easily used at submicromolar concentrations, and can have properties that report rapidly, even on small perturbations in the region of the fluorophore. Thus, ATP and analogs have been modified with fluorophores at several locations in the molecular structure and the range of modifications has been reviewed (12, 13). The choice of attachment site is important, both to get a fluorophore in a position to report but also to modify the parent nucleotide without significant perturbation of the biochemical properties. First, the purine base can be made fluorescent either by modification or by using a fluorescent analog of the natural base, as in the case of formycin triphosphate (FTP) (8). However, such modifications may disrupt the protein–nucleotide interactions, as there is often high selectivity in the base binding site of proteins (13). The phosphate chain can also be modified, as with γ-AMNS-ATP for investigations of Escherichia coli RNA polymerase (14), but these modifications frequently disrupt the cleavage step by preventing correct binding of the phosphates or by blocking access to the γ-phosphate. Ribose modifications are often the most successful, whereby the analog closely mimics the activity of ATP. In many enzymes that bind GTP or ATP, the 2′- and/or 3′-hydroxyl groups of the ribose are partly exposed to the protein surface, while the base and phosphates are well buried. This allows a ribose label to sit at the entrance to the binding site and potentially report on changes in that region, with only a small effect on the binding and catalytic properties.
1.2 Selection of Fluorophore
When fluorescent analogs are to be used for microscopy, the main criteria for choices of fluorophore are high fluorescence intensity (high extinction coefficient and fluorescence quantum yield), excitation and emission maxima best suited to the excitation source, without interference from other components in the system, and stability against photobleaching. With some fluorophores, the extinction coefficient and particularly the quantum yield can change significantly with the chemical environment of the fluorophore. Such changes can be problematic in choosing a suitable fluorophore, but if the intensity changes are monitored, then changes in intensity can be harnessed, as described below for a diethylaminocoumarin. Long-wavelength fluorophores, such as the cyanine dyes (e.g., Cy3), have good properties for fluorescence microscopy and have been used for detecting fluorescence from single molecules. Typically, fluorophores that excite at longer wavelengths are relatively large moieties with multiring structures and may have significantly hydrophobic regions. This may lead to nonspecific binding to proteins and surfaces. There are now many commercially available fluorophore-labeling reagents that give variations on quantum yield, photostability, and wavelengths. A discussion of this variety is outside the scope of this chapter. For additional information and for commercial sources of such labels as well as ATP analogs see: http://www.invitrogen.com, http://www.sigmaaldrich.com, http://www.roche-applied-science.com and http://www.jenabioscience.com. Quantum dots have significant potential for future use, but their large size relative to nucleoside triphosphates is likely to make them difficult to apply generally here.
1.3 Selection of the Linker Between Nucleotide and Fluorophore
As described above, labeling at the ribose hydroxyl groups has been useful because this modification may not lead to large perturbations of the biochemical properties. However, such labeling of the hydroxyl groups leads to the formation of a mixture of 2′- and 3′-isomers, whose biochemical and fluorescent properties may differ when bound (13). This problem can be circumvented by using a parent nucleotide, in which only one hydroxyl is available for modification, for example the commercially available 2′-deoxyATP, or the synthetic 3′-amino-3′-deoxyATP (15). Alternatively, the isomers of some labeled nucleotides interconvert only very slowly and can be successfully separated by chromatography and stored as single isomers (3, 16).
As mentioned in the above section, fluorophores for fluorescence microscopy are relatively large and so may disrupt the biochemical cycle. To ameliorate this problem, several linkers are available to space the fluorophores further away from the catalytic site. Essentially, a zero-length linker is achieved by direct labeling of the amine group on the ribose ring of 3′-amino-3′-deoxyATP. Longer chemical linkers can use different lengths of diamino-n-alkanes, such as 2′(3′)-O-[N-(2-aminoethyl)-carbamoyl]ATP (edaATP (17)) and 2′(3′)-O-[N-(3-aminopropyl)carbamoyl]ATP (pdaATP (15)). In some cases, including Cy3, the commercially available labels include a spacer chain between the fluorophore and the reactive group used for attachment. In these cases, the fluorophore will be positioned further away from the protein and, therefore, should not interfere significantly with the nucleotide association and catalysis. However, moving the fluorophore further away from the protein may reduce any changes to fluorescent properties on binding. All modifications may be deleterious to the enzymic activity, and therefore, it is important to assess the impact of these changes. Methods to assess these effects are described later.
1.4 Other Considerations
The requirements for the synthesis of ATP analogs vary widely. Those described here are relatively simple labeling reactions, performed under aqueous conditions, so potentially high yields can be obtained using conditions and equipment available in most laboratories. The success of the labeling depends both on the chemical reactions per se and on the properties of the fluorophore. For example, very hydrophobic groups may impair the success of a reaction that occurs perfectly well with simpler labels. The purification of the product also may depend on the physical properties of the fluorescent label. Two specific examples are described for the labeling nucleotides at the ribose ring with Cy3 and diethylaminocoumarin (Fig. 1).
The visualization of the binding of fluorescent nucleotides to proteins by light microscope has been limited by technical problems such as the nonspecific binding of the fluorescent nucleotide to the coverslip. This has limited the maximum nucleotide concentration that could be used with analogs such as 2′(3′)-(Cy3-O-[N-(2-aminoethyl)carbamoyl])ATP (Cy3-edaATP, Fig. 1b (3)) to <100 nM. Fluorescent groups may also bind to macromolecules such as proteins, independently of the nucleotide and its binding site, and particularly if used at high concentration. A control, such as displacing the labeled with unlabeled nucleotide, will test if such nonspecific binding occurs.
The fluorescent ATP analog, (3′-(7-diethylaminocoumarin-3-carbonylamino)-3′-deoxyadenosine-5′-triphosphate (deac-aminoATP)) (Fig. 1a) has a low quantum yield when in solution, but this increases dramatically when bound to some proteins. This generates large fluorescence changes, such as the 20-fold increase when bound to myosin Va (18). This enables a distinction between coverslip bound “background” molecules and those bound by proteins (4), which may compensate for the relatively low optimal excitation wavelength and photostability.