Why do it? X-ray crystallography of protein-ligand complexes is frequently used during the drug discovery process to determine the molecular details of binding between the chemical entity of interest (ligand) and the target protein. This information can be very useful to guide the modification of atoms and groups in the potential drug to make it more potent, and to improve its chemical or biological properties. The molecule can be introduced into the protein crystals in two ways; either by crystallising the pre-formed protein-ligand complex or by soaking an apo crystal in a ligand solution after the crystallisation has taken place. There is no universal answer as which is the method of choice, it is very protein dependent. The solubility and potency of the ligand has a big effect on the experimental design.
A soakable system can permit a higher throughput if there are many ligands of interest and is less protein-hungry. However, it requires the ability to reproducibly grow apo crystals, which are singular and transferable into a stable soaking solution. In practical terms, this translates into having not too many volatiles or near saturated components in crystallisation solutions and having crystals which are physically robust (definitely a relative term as protein crystals are delicate at the best of times 😰). That is, they are reasonably chunky and three dimensional – wafer-thin plates or needles can add to the handling challenge.
The crystals will usually need to be able to tolerate at least a few % DMSO, to allow the addition of ligand in suitable excess over the protein concentration. DMSO stocks of around 100mM compounds would be a typical starting point. Proteins going into crystallisation would typically be 10-20mg/ml (a ballpark of 0.2-0.5mM). For potent compounds, (where the Kd is considerably less than the protein concentration) the ligand is close to full occupancy at an equivalent molarity to the protein. For compounds with higher Kds, at least a 10-fold excess of compound over protein is recommended. Ligands are not always sufficiently soluble to achieve this; extra cunning and wizardry is then required.
Soaking times can vary from minutes to several days. After transfer to the soaking solution (in simple cases, the crystallisation reservoir with added compound) it’s a case of watching the crystal down the microscope. Some die quickly, some just fade away, some live happily ever after (or a few days at least!). With a successful system, hundreds of structures containing ligands can be obtained from sub milligram amounts of protein upwards leading to structure-guided discovery of potent and selective inhibitors e.g. Erk2 Ward et. al., 2017
Sometimes, however, an alternative approach is needed. Many proteins are only stable in solution when complexed with co-factors or nucleotides. Without a stabilising ligand present they may not fold correctly, often aggregating and hence can be difficult to purify to homogeneity. This limits the range of proteins which can be crystallised as an apo, potentially soakable, form. Sometimes an apo protein may be stable, and crystallises well, but if there is a conformational change upon binding to the ligand, then the soaking method is unlikely to work. Attempting to soak active compounds into such proteins can vary and crystals are often observed as gradually becoming more jelly-like or quite dramatically shattering.
In contrast to soaking, co-crystallisation experiments involve forming the protein-ligand complex prior to setting up the crystallisation screens. Incubation times required vary nearly as much as soaking times – from around 30 minutes to days, with a few hours being a good starting point. Similar guidelines for amounts of ligand required apply as to those for soaking. This can often be the method of choice with low solubility ligands where complex formation can take place at low protein concentration, allowing the ligand to be present in suitable excess of protein. After the complex has formed, it can often be concentrated to put into crystallisation. Sometimes just lumping the precipitate and all into the crystallisation drop can work for a slow crystallisation. The rationale here is that having compound present as a solid will ensure the amount in solution will be maintained by more compound dissolving as some moves across into the protein-ligand complexes.
Less soluble ligands can be encouraged into solution sometimes by using other solvents e.g. PEG400 or alcohols, particularly diols, instead of, or in combination, with DMSO (this can be applicable to soaking experiments as well). Occupancy of the ligand can sometimes be increased by soaking a co-crystal in some additional ligand solution.
A great example of stabilising a protein with ligand during purification is the kinase, JAK2. Purification of the apo form resulted in protein which was very prone to aggregation. Purifying the protein in the presence of staurosporine (a potent, non-selective inhibitor of protein kinases) produced a stable complex which could be purified and put into crystallisation trials. These crystals could be soaked with sufficiently potent ligands of interest but the limitation was that only ligands with a greater affinity for the protein than staurosporine could be successfully soaked in (Ioannidis et al, 2011)
Despite our best efforts, however, independent proteins can make their own decisions! In some cases, proteins will retain their natural, tightly bound ligands during the purification process. Many can pick up a bonus ligand from somewhere in the purification procedure, e.g. buffers, cryoprotectants, even protease inhibitors. There are numerous examples of this in the PDB. And finally,….don’t wash that ligand out! In both systems it is important to remember to include an appropriate concentration of the ligand in any cryo-protecting solutions used to freeze the crystals. Weaker binding ligands could easily be diluted and soaked out of the protein if the local concentration was reduced.
Have a look at our protein crystallography service page to find out more about how we can help you with your protein crystallography challenges.
For further background information have a look at Guidelines for the successful generation of protein–ligand complex crystals and Hampton Research: Tip and tricks archive