Soaking of crystals for macromolecular crystallography in a capillary

A method for soaking macromolecular crystals in ligand solutions is described in which a crystal is placed in a capillary. The length of the capillary is used to form a gradient of the ligand concentration. This set up reduces the crystal susceptibility to cracking. An often-used approach in crystal studies of protein complexes is to grow native protein crystals and then soak the crystals in an artificial mother liquor containing desired ligands in addition to the precipitant, buffer etc. This approach works because protein crystals contain large channels between the protein molecules which are filled with solvent. The ligand molecules are usually small enough to diffuse into the interior of protein crystals through these channels and to bind to the protein molecules. There are some caveats, however. The process of the exchange of mother liquor sometimes leads to crystal cracking and/or significant loss of the X-ray scattering power of the crystals. This could in some cases be related to crystal crushing by osmotic pressure, especially when high-ionic-strength precipitant is exchanged for polyglycols (Ray, Puvathingal, Bolin, Minor, Liu & Muchmore, 1991). More often, however, crystal cracking is due to changes in the unit-cell dimensions that result from conformational adjustments induced by ligand binding to the protein molecules and subsequent changes in intermolecular contacts. If we consider the process of crystal soaking, a ligand molecule entering a solvent channel in the crystal may either bind to one of the binding sites on the protein molecules forming the channel or progress inwards through the maze of the channels. A competition between these two processes will lead to a gradient of ligands bound to protein molecules. If there is a unit-cell change connected with ligand binding, this gradient will lead to different unit-cell dimensions in the outer parts of the crystal and in the interior and in consequence will cause crystal strain and cracking and/or increased mosaicity. In many cases, ligand binding is a reversible process and one way to avoid crystal cracking is to use initially a very low concentration of the ligand, wait until equilibrium is reached, and gradually increase the ligand concentration. In some applications up to 40 exchanges of mother liquor were used (Ray et al. 1991). It should be mentioned, however, that if the ligand binding is virtually irreversible then even very low concentrations of the ligand may lead to a large gradient of the protein-ligand complex in the crystals. We report here a very simple procedure that allows one to eliminate laborious multiple exchanges of mother liquor. This procedure is essentially a modification of the crystalmounting technique used in many laboratories. A crystal is dropped into a capillary filled with the mother liquor. The crystal then sinks to the bottom of the capillary where it is supported by the lower meniscus, as shown in Fig. 1. For the crystal-soaking experiments, mother liquor from the capillary head is then removed with a syringe and replaced with a solution also containing the ligand. The capillary is then sealed with vasoline and left for a period of 3-4 d. During this time the ligand molecules diffuse through the capillary and their concentration in the crystal gradually increases. We have carried out tests with 1 × 10 -3 M permanganate ions diffusing through 50% saturated ammonium sulfate solution in a 1 mm capillary with a head-tocrystal distance of about 45 mm. The characteristic coloring of permanganate ions was noticeable at the crystal, at the bottom of the capillary, after about two days. After another two days, the concentration of permanganate ions within the capillary had equilibrated and its gradient was not apparent. Obviously, for larger ligands or more viscous solutions, the diffusion is slower and the soaking time should be adjusted. Using this approach, we were successful in