Title
Crystallization of protein-nucleic acid complexes
Overview
A prerequisite for the determination of the three-dimensional structure of a protein-nucleic acid complex by x-ray crystallography is a well-ordered crystal of the complex in question. This article describes methods for crystallizing these complexes, including protein sample, choice of oligonucleotide, preparation of complexes, and crystallization conditions.
Introduction
X-ray crystallography has provided a wealth of information on the three-dimensional structures of proteins and nucleic acids, and on the atomic details of how these two types of biological macromolecules interact. With the rapid improvement in data collection and computational methods that have occurred over the last decade, the principal obstacle that lies in the path to determining an atomic structure is the crystallization of the molecule or complex of interest. Typically, a trial and error approach is used to identify conditions under which single crystals will grow from a concentrated solution of macromolecules.
The approach to crystallizing macromolecules remains rather empirical. In the case of protein-DNA complexes, there are many more considerations than for the case of the crystallization of proteins alone, making the task seem daunting. In addition to considerations of protein choice and purity, one must consider the composition and length of the DNA and the way in which the complex is formed. Fortunately, the accumulated experience of the many investigators who have studied protein-DNA complexes makes it possible to come up with basic guidelines for anyone embarking upon the crystallization of a new protein-DNA complex. While this article focuses on complexes between proteins and DNA, many of the considerations outlined here will be the same for crystallizing complexes between protein and RNA.
Material
Protein sample preparation
The first consideration in crystallizing a protein-DNA complex is the protein. While it is always most desireable to crystallize the intact protein or proteins bound to DNA, it is frequently the case that disordered or flexible regions of the protein will interfere with the formation of a well-ordered crystal lattice. A common procedure for identifying the minimal folded protein fragment to use is to carry out limited proteolysis experiments. In these experiments, serial dilutions of proteases such as subtilisin or trypsin are incubated with the protein in the presence and absence of DNA. Reactions are stopped by the appropriate protease inhibitor and analyzed by SDS-polyacrylamide gel electrophoresis. The major products of the protease digestion can be identified by excising bands from the gel and subjecting them to analysis by N-terminal sequencing and mass spectroscopy. Specific cleavage products must also be analyzed for biological activity such as DNA binding or cooperative interactions with partner proteins. If these experiments identify a smaller fragment that retains the activities of interest, PCR cloning methods are used to construct an expression vector to express this fragment.
For the purpose of crystallization, it is desireable to have a highly pure, concentrated preparation of protein. Ideally, the protein should be ? 99% homogeneous and at a concentration of at least 5 mg/ml, and preferably 10 - 20 mg/ml. The protein solution should be buffered with 25 - 50 mM of a non-phosphate buffer in the pH range of 6.0 - 8.0 and should contain the minimum of salt required to maintain protein solubility. In addition, the solution should contain 1 mM EDTA and 0.001% sodium azide to inhibit bacterial growth. Reducing agents such as dithithreitol (DTT) should be included if the protein contains any cysteines. If significantly higher levels of salt or buffer are needed to maintain protein solubility, attempts to reduce the ionic strength can be carried out after the protein-DNA complex has been formed.
Procedure
The DNA: Sequence choice and sample preparation
Like the choice of which protein fragment to use, the choice of the DNA fragment plays a significant role in the success of a crystallization experiment. The experience of numerous investigators has shown that the precise length and composition of the oligonucleotide is the most critical variable and must be experimentally determined for each new protein of interest. Since many protein-DNA complexes crystallize under a rather narrow range of solution conditions, an efficient approach is to focus one? efforts on trying an exhaustive variety of DNA fragments under a relatively limited set of crystallization conditions. While, in theory, one should also try to be exhaustive in exploring potential crystallization conditions, experience has shown the effort is better spent on trying more DNA sequences.
In the choice of the DNA fragment to use in complex formation, a critical variable is the length of the DNA. DNA has a strong preference to stack end-to-end in a crystal, so the precise length of the DNA fragment and the nature of the stacking interactions between fragments are critical determinants of the unit cell size and crystalline order. The lower limit on the length of the DNA fragments will be determined by the size of the minimal binding site necessary for tight complex formation between the protein and DNA. Since protein contacts with the DNA backbone generally extend beyond the minimal DNA sequence recognized by the protein, it is important to use data on DNA binding affinity and not rely solely on studies that identify the base pairs most critical for DNA sequence recognition. Beginning with the minimal double-stranded sequence needed for binding, one can then experiment with sequences of increasing length. Lengths of DNA that have been particularly successful in crystallization trials (11, 15, 16, 20, 21 and 26 base pairs) correspond to multiples of approximate integral or half integral turns of the DNA. Initial crystallization trials should be carried out with a series of DNA fragments of these lengths containing the protein binding site near the center of a DNA fragment. If these oligonucleotides are unsuccessful, additional bases should be added or taken away until the whole spectrum of lengths is examined. One should also consider modifying the binding site sequence, altering the position of the binding site within the oligonucleotide, or adding binding sites to the oligonucleotide.
Another important sequence variable is the composition of the DNA ends. Blunt-ended oligonucleotides have yielded high-quality crystals in a number of cases, although single or double complimentary overhanging bases at either the 3?or 5?end are frequently included to promote end-to-end stacking of oligonucleotides in the crystal. While a number of different types of overhangs have been used successfully, the most common one is a single 5?overhanging A on one strand and a 5?overhanging T on the other. Another approach is to engineer DNA ends that promote triplex interactions through Hoogsteen base-pairing of the overhanging C with a G?C base pair in the opposing fragment. Double-stranded oligonucleotides containing the sequence CpC at the 5?end of each strand and a G at each 3?end can potentially form these interactions.
DNA for crystallization trials can be produced by 1 mM synthesis of each strand on commercially available DNA synthesizers. Purity of greater than 99% can be achieved by two rounds of purification by reverse-phase HPLC. In the first step, the synthetic single-stranded oligonucleotide with the 5?trityl group is loaded onto a reverse-phase Varian PureDNA column in the presence of 0.1 M triethylamine acetate (TEAA) pH 7 and eluted with a gradient of increasing acetonitrile in 0.1 M TEAA. The pooled peak fractions are dialyzed into 0.01 M triethylamine bicarbonate (TEAB) pH 7 and then lyophilized. The DNA is then resuspended in 0.1 M TEAA and reloaded onto the column. The trityl group is removed by flowing 0.5% triflouroacetate (TFA) over the column for 10 minutes, after which the TFA is washed away with 0.1 M TEAA and the column developed with an acetonitrole gradient in 0.1 M TEAA. The peak fractions are pooled, dialyzed into 0.01 M TEAB, then lyophilized until needed. Complementary strands are annealed by resuspending each strand in 0.5 ml of 0.01 M TEAB and combining stoichiometric amounts, heating the solution for 5 minutes in a 70oC water bath, and then allowing the DNA to come slowly to room temperature. The concentration of the annealed DNA is determined spectroscopically and the DNA is aliquoted in fixed amounts, lyophilized, and stored at -80oC until needed.
Complex preparation
The protein-DNA complex is prepared by adding the concentrated protein solution directly to the lyophilized DNA. The goal is to have a slight excess of DNA (typically 10 - 20%) over the amount needed for a 1:1 (or 2:1 or 3:1, as the case may be) ratio of protein to DNA. For more complex systems involving two different proteins, it would be advisable to look closely at the
ratio of the proteins, examining the complex at a series of different protein-protein ratios by gel shift to maximize complex formation.If there is more than one DNA binding site and binding of the proteins is not cooperative, it may not be advisable to have much excess DNA. In this case, the presence of extra DNA may lead to a solution containing a mixture of 1:1 and 2:1 protein-DNA complexes, which may not be conducive to crystal growth.
Once a complex of the appropriate stoichiometry is formed, it is dialyzed into the most minimal buffer conditions under which the complex remains soluble. Since these conditions might be quite different for the complex than they are for either the protein or the DNA alone, these conditions must be determined for each complex, although not necessarily for each new DNA sequence. The minimal buffer would ideally be distilled water, although that is usually not possible. A more achievable minimal buffer might be 10 mM buffer, 1 mM EDTA (if you?e concerned about nucleases), 0.001 % sodium azide, and 1 mM DTT if necessary. The underlying idea is that certain components might interfere with crystallization, so it is best to eliminate as much as possible from the solution. Moreover, it is easier to manipulate the pH in crystallization trials if there is little or no buffer present in the complex solution.
The precise buffer conditions that will allow the complex to remain soluble will depend somewhat on the concentration of the protein and DNA. Sometimes, the solubility of the complex is significantly higher than the protein alone, making it possible to further concentrate the complex at this stage. This is particularly useful if it was not possible to concentrate the protein to more than a few mg/ml. In that case, a second concentration step following formation of the protein-DNA complex would be advisable. The entire complex can then be further purified chromatographically, but this has been determined to be unnecessary in the majority of cases. After the final preparation of each complex, a small sample should be analyzed on a non-denaturing polyacrylamide gel to ascertain whether a complex with the desired proportion of excess DNA is present.
Crystallization conditions
A large proportion of protein-DNA complexes whose structures have been determined were crystallized under a relatively limited set of conditions. These include polyethylene glycol (PEG) as the primary precipitant, 20 - 100 mM buffer between pH 5 - 8, and frequently some amount of salt (either NaCl of KCl) and/or divalent cation (usually CaCl2 or MgCl2). In addition, small amounts of various additives are sometimes required to produce high quality crystals. These additives include polyamines such as spermine or cobaltic hexamine chloride, non-volatile organics such as glycerol or 2-methyl-2, 4-pentanediol (MPD), and ions such as zinc.
Crystallization trials should employ the hanging drop vapor diffusion method, which allows rapid screening of many conditions and oligonucleotides. In this technique, a small amount of the protein-DNA complex is pipetted onto a siliconized cover slip. An equal amount of crystallization solution is added to the drop and the cover slip is inverted over a well containing the same crystallization solution. Equilibration of conditions between the drop and the well solution occurs by diffusion of vapor within the sealed well. Since the precipitation point for complexes composed of various oligonucleotides is frequently similar, a simple screen can be used to quickly test the conditions required for crystallization. Due to the flexibility of the number of drops that can be placed on a single coverslip, up to four complexes with different oligonucleotides can be tested simultaneously with the same crystallization solution.
It is frequently the case that once the correct oligonucleotide is found, crystallization occurs under numerous conditions. Therefore, a limited crystallization screen can be used to rapidly test the various oligonucleotides used in complex formation. First, a 24-well tray should be set up to determine the precipitation point of the various complexes. Various concentrations of a specific PEG can be screened as a function of four different pH values. For example, the concentration of PEG4000 can be varied from 5 - 30% (w/v) in intervals of 5%, in the presence of 100 mM buffer: citrate pH 5, MES pH 6, HEPES pH 7, and Tris-HCl pH 8 (Figure 1). Once the precipitation point is found for the specific precipitant and pH, a second tray can be set up to address the effect of different additives. In this tray, the range of precipitant concentrations should be chosen so that the maximum concentration of precipitant is just below that required for complex precipitation. A good starting point is to test 2 mM spermine, 5 mM cobaltic hexamine chloride, 100 mM NaCl, 10 mM MgCl2 and 100 mM NaCl + 10 mM MgCl2. Another tray, in which CaCl2 is used in the presence and absence of other salts, should also be set up. Precipitation and crystal formation from these solutions should then be used to direct the choice of conditions for further crystallization trials. While conditions to be explored are limited only by one? imagination, it is advisable to start by exploring several different molecular weights of PEG, varying the type and concentration of salt, and experimenting with the presence of different divalent cations. Once some sign of crystal formation has been observed, crystal quality can sometimes be dramatically improved by adding small amounts of some type of salt, organic solvent, or other compound. A little creativity and adventurousness in exploring the chemical shelf can pay off in the form of large, single crystals.
Once crystals have been obtained, it is important to verify that they indeed contain the protein-DNA complex of interest. Crystal content can be analyzed by washing a crystal in well solution, then dissolving the crystal in dilute buffer and subjecting the solution to electrophoresis in a non-denaturing polyacrylamide gel as described above. Standard SDS-PAGE of dissolved crystals can also be used to verify that the protein in the crystal has not become degraded in the crystal drop. If degradation appears to have occurred, further analysis of the sample by mass spectrometry may be necessary.
Summary of what has worked for others
It is not possible in the scope of this article to review the crystallization conditions for all protein-DNA complexes whose stuctures have been determined. We summarize here some of the findings from a survey of 72 crystal structures determined between 1988 and 1999. Of these DNA-protein complexes, three-quarters were crystallized with PEG as the primary precipitant, with about 10% utilizing MPD and the remainder a variety of salts and organic solvents. Complexes were crystallized over a broad pH range, extending from pH 4 to 9, with 6.4 being the average pH. Over half of the complexes were crystallized in the presence of on average 0.2 M salt, with the concentration of NaCl or KCl ranging from 0.02 to 0.6 M. About 65% of the co-crystals were formed in the presence of divalent cations. They are, in decreasing order of popularity, Mg2+, Ca2+, Zn2+, Ba2+, and Cd2+. The average divalent cation concentration was 0.05 M, with a range of 2 mM - 0.2 M. Another commonly occuring additive was a polyamine: either 0.1 - 3 mM spermine or 2 - 50 mM cobaltic hexamine chloride. Organic solvents such as glycerol, MPD, or ethylene glycol, which are most commonly used as primary precipitating agents, have also been used in small amounts (less than 15% w/v) as additives that improve crystal quality.
Figure 1# (
http://muddy.med.jhmi.edu/wolberger/Fig1.jpg) Sample crystallization tray showing conditions in each well for an initial screen for precipitation points of the protein-DNA complexes in question. A skilled practitioner can easily fit four drops in a single cover slip, thereby facilitating the rapid screening of many different oligonucleotides.
Troubleshooting
Reference
Adrian H. Batchelor*, Derek E. Piper #, and Cynthia Wolberger #
* Department of Infection and Immunity, Walter and Eliza Hall Institute, Royal Melbourne Hospital, Victoria 3050 Australia
# Howard Hughes Medical Institute and the Johns Hopkins University School of Medicine, 725 N. Wolfe St., Baltimore, MD 21205 USA
