Protein crystallization and crystallography are both a science and an art. Every protein crystal is different and unpredictable, even those from the same protein. Crystals that appear to be of “perfect” morphology—whether rhomboid, cubic, trigonal or other morphologies—may not be well-ordered in the crystal lattice, thus never producing high-quality X-ray diffraction data. To find out, you must wade through crystal trials that can last months, or sometimes years. Even then, 99% of your crystal trials typically fail to produce a crystal, or more often, produce a crystal with limited resolution for 3D atomic determination. But when it works, X-ray crystallography is an incredibly powerful tool for resolving the structure of a given protein, offering valuable insight into its function.
In this edition of Bench Tips, we speak with the Stroud lab at the University of California, San Francisco, which is renowned for resolving large membrane-protein structures. Membrane proteins are the target of more than 50% of all modern medicinal drugs. The lab often works on ion channels, transporters or G-coupled protein receptors, approximately 200 to 300 kDa.
“A lot of the bigger, more complex membrane proteins have branched sugar moieties that are notoriously bad for crystallization,” says Joseph D. O’Connell III, a staff research associate in Dr. Robert Stroud’s lab. “You need to remove these by adding a deglycosidase enzyme … [after size-exclusion chromatography], in order to establish identical protein building blocks before crystallization.”
Purify target protein
O’Connell explains that having a homogenous protein sample—in which the protein has identical, molecular building blocks—greatly increases the chances of the molecules coming together to form a well-ordered crystal. (Light-scattering experiments and size-exclusion chromatography (SEC) are usually the best measure of a protein’s homogeneity.) In addition, the protein must be ≥95% pure and meet the following criteria: 1) the protein must form a single Gaussian peak on SEC and 2) the protein must purify as a single band on an SDS-PAGE gel (stained with Coomassie blue). The protein must also be stable—whereby the protein, in concentrated form, stored at 4°C for one week, retains a single Gaussian peak on SEC. Moving from heterologous expression of the target protein to these optimal conditions is a lengthy process, but a requirement before proceeding with crystal trials.
Starting protein concentration
When troubleshooting protein crystallization, the most important variable to optimize is the starting protein concentration. To check the success of a protein-crystallization experiment, crystal drops are examined through stereo light microscopes. A majority of clear drops in the experiment generally signifies that the protein concentration is too low. Alternatively, the appearance of a brown amorphous precipitate in the majority of crystal drops usually indicates the protein concentration is too high. It can also indicate that the protein is unstable, not pure and prone to aggregation.
“If the protein concentration is too high, you may get lots of little crystals, a crystal that is ‘twinned’ or bunched-up crystals, all situations that are not optimal for growing a well-ordered protein crystal,” says O’Connell. “Ideally when we do a protein-crystallization experiment, we look for regular, repeating crystalline precipitate to be present in the drops, which is indicative of a well-behaved protein sample that may produce (or has already produced) a crystal.”
If the crystal drops are sub-optimal, the starting protein concentration must be adjusted to allow the protein to crystallize properly. Some membrane proteins produce crystals at 1 to 2 mg/ml, whereas others require a concentration of 20 to 30 mg/ml (or more) to crystallize. The optimal starting protein concentration is totally protein-to-protein specific and dependent, and it must be determined empirically. Typically, the Stroud lab likes to screen a new protein for crystallization at a minimum concentration of 5 to 10 mg/ml, if possible.
Detergent selection
Optimizing key parameters—like protein concentration, appropriate detergent and optimal detergent concentration—also can facilitate success in crystal trials. A common problem in working with membrane proteins for crystallization is an unstable target protein, where the protein comes out of solution. In such cases, O’Connell recommends adjusting the detergent concentration for solubilization and purification of the target protein, the pH of the buffer used for solubilization and purification, the salt concentration used in the purification, the addition of reducing agents to the protein purification and the addition of osmolytes (i.e., glucose or sucrose) to the protein buffer. One can also add the appropriate substrate or ligand to the protein (if one exists) during purification in the hope of producing a sample that is more homogenous and stable for crystallization trials. If changing these parameters fails to produce a stable protein sample for crystallization, one may switch detergents to guarantee better solubilization and purification of the target protein for crystallization. The concentration of a given detergent required for solubilization and purification of the target protein depends in large part on the critical micelle concentration (CMC) of a particular detergent and the protein system being studied.
Typically, 10x CMC is used to solubilize a membrane protein, thereby ensuring complete extraction of a protein from the cell membrane. For purification of the protein, 1x or 2x CMC is used for a given detergent. The optimal detergent concentration helps keep the protein stable and provides an optimal sample for crystallization. Although the most suitable detergent for a particular protein must be determined empirically, the Stroud lab frequently uses the glucoside (i.e., n-Octyl-β-D-Glucopyranoside (OG), 270 mM for solubilization and 40 mM for purification) and maltoside (i.e., n-Dodecyl-β-D-Maltopyranoside (DDM), 20 mM for solubilization and .05-2 mM for purification) classes of detergents for membrane proteins. Screening for an appropriate detergent should be performed early on. It’s worth noting the detergent used for solubilization does not need to be the same for purification and crystallization.
“You can try to control for all these things, but you just never know what’s going to happen,” says O’Connell. “Even when you think you have nice-looking crystals, you don’t necessarily get good diffraction data. I’ve grown some really ugly-looking membrane-protein crystals over the years that have produced high-resolution diffraction data.”
It should be noted that there can be differences in the diffraction limits and resulting resolution from the same crystal trial. One crystal might give poor data, but another crystal from the same drop may diffract really well. Typically, the lab observes the trials once a day for the first week after setup and then once or twice a week for a month thereafter. (Sometimes the experiment is then repeated at a colder temperature, usually 4°C, when the crystals fail to produce sufficient data.) O’Connell monitors his crystal trials for a minimum of two to three months, as he has seen crystals appear later. But in protein crystallization, he notes, there are never any guarantees—although sometimes a little luck truly does go a long way.
For more detailed information on these techniques, see: Newby, ZE, et al., “A general protocol for the crystallization of membrane proteins for X-ray structural investigation,” Nature Protocols, 4(5):619-37, 2009.
The image at the top of the page is a crystal of a membrane protein, courtesy of the Stroud Lab.