Combining Wet and Dry Lab Techniques to Guide the Crystallization of Large Coiled-coil Containing Proteins

Summary

We describe a framework incorporating straightforward biochemical and computational analysis to guide the characterization and crystallization of large coiled-coil domains. This framework can be adapted for globular proteins or extended to incorporate a variety of high-throughput techniques.

Abstract

Obtaining crystals for structure determination can be a difficult and time consuming proposition for any protein. Coiled-coil proteins and domains are found throughout nature, however, because of their physical properties and tendency to aggregate, they are traditionally viewed as being especially difficult to crystallize. Here, we utilize a variety of quick and simple techniques designed to identify a series of possible domain boundaries for a given coiled-coil protein, and then quickly characterize the behavior of these proteins in solution. With the addition of a strongly fluorescent tag (mRuby2), protein characterization is simple and straightforward. The target protein can be readily visualized under normal lighting and can be quantified with the use of an appropriate imager. The goal is to quickly identify candidates that can be removed from the crystallization pipeline because they are unlikely to succeed, affording more time for the best candidates and fewer funds expended on proteins that do not produce crystals. This process can be iterated to incorporate information gained from initial screening efforts, can be adapted for high-throughput expression and purification procedures, and is augmented by robotic screening for crystallization.

Introduction

Structure determination via X-ray crystallography has made fundamental contributions to every field of modern biology; providing an atomic view of the macromolecules that support life and how they interact with one another in a variety of contexts; allowing us to understand the mechanisms that cause disease and providing opportunities to rationally design drugs to treat disease. Crystallography has long been the dominant experimental technique for determining macromolecular structure, and currently accounts for 89.3% of the structural database (www.rcsb.org). This technique has many advantages, including the potential for very high resolution, the ability to visualize macromolecules with a broad range of sizes, relatively easy data collection, and the opportunity to visualize how the macromolecule interacts with solvent as well as ligands.

Despite numerous technological improvements in recombinant protein expression^1,2, purification³, and molecular biology used to generate these systems⁴, the single biggest obstacle in the crystallographic process remains the ability to grow diffraction quality crystals. This has been especially true for proteins which contain large coiled-coil domains. It has been estimated that as much as 5% of all amino acids are found within coiled-coils^5,6, making this a very common structural feature⁷, yet these proteins are often more difficult to purify and crystallize than globular proteins^8-10. This is further compounded by the fact that coiled-coil domains are often found within the context of a larger protein, therefore correctly predicting the boundaries of these domains is critical to avoid the inclusion of unstructured or flexible sequence that is often detrimental for crystallization.

Here we present a conceptual framework combining web-based computational analyses with experimental data from the bench, to help guide users through the initial stages of the crystallographic process including: how to select protein fragment(s) for structural studies, and how to prepare and characterize protein samples prior to crystallization attempts. We focus our analysis on two proteins containing large coiled-coil domains, Shroom (Shrm) and Rho-kinase (Rock). These proteins were chosen as they both contain coiled-coil domains and are known to form a biologically relevant complex^11-16. Shroom and Rho-kinase (Rock) are predicted to contain ~200 and 680 residues of coiled-coil respectively, many portions of which have been characterized structurally^17-20. The method described here provides a streamlined workflow to quickly identify fragments of coiled-coil containing protein that will be amenable for crystallization, however, the techniques described can easily be adapted for most protein or protein-complexes or modified to incorporate high-throughput approaches as available. Lastly, these methods are generally inexpensive and can be performed by users at nearly all experience levels.

Protocol

NOTE: A diagram of the conceptual framework or workflow is described in Figure 1 for reference. The protocol can be broken down into four stages: computational or sequence based predictions, protein expression and purification, biochemical characterization, and crystallization. The examples shown analyze Shroom SD2 domains and/or Shroom-Rock complexes, but can be utilized with any protein.

1. Use Established Web-based Tools to Generate Computational Predictions of Coiled-coil Domain Boundaries

1. Collect an Evolutionarily Diverse Set of Sequence Homologs.
   1. Go to www.uniprot.org and type in Shroom in the upper search bar.
   2. Select sequences to download by checking the box next to their accession number. Sequences with a star indicate sequences reviewed by Uniprot and are generally more reliable. Take care to ensure that all sequences are complete and accurate.
   3. Save the selected sequences by clicking the Download button.
   4. Align the collection of Shroom sequences using Clustal-Omega²¹ (http://www.ebi.ac.uk/Tools/msa/clustalo/). Click the “Browse” button to select the file with the Shroom sequences and then push Submit.
   5. After the alignment is complete, click the "Download Alignment File" button.
   6. Open the multiple sequence alignment generated in 1.1.5 using Jalview (File|Input Alignment|From file). Within the new alignment window, color by identity (Color|Percent Identity).
2. Calculate predictions of secondary structure, predictions of flexible or disordered sequence, and predicted coiled-coil regions within the protein(s) of interest.
   1. Go to http://gpcr.biocomp.unibo.it/cgi/predictors/cc/pred_cchmm.cgi, to calculate Coiled-Coil predictions. Copy and Paste the sequence into the Sequence Box and click Submit.
      NOTE: This analysis may take a couple of hours to complete. Since Shroom sequences are quite large, the sequences may need to be split into blocks of less than 1,000 amino acids to fit within the size constraint of the webserver. When analyzing Shrm2, it is recommended that the C-terminal 1,000 amino acids are used as this section contains the Shrm SD2 domain which is known to contain coiled-coil regions. When repeating this analysis with other proteins, it is recommended to use the full-length protein sequences when possible. If that is not an option, use the largest fragment possible or dissect the sequence based on known biochemical or functional data.
   2. Go to http://gpcr.biocomp.unibo.it/cgi/predictors/cc/pred_cchmm.cgi, to calculate Coiled-Coil predictions. Copy and Paste the sequence into the Sequence Box and click Submit.
3. Combine the results of steps 1.1 to 1.2.2 into a single comprehensive annotation of the Shroom sequence.
   NOTE: It is highly encouraged to also include any and all relevant information that can be gleaned from the literature or other sources about protein function or purification at this stage.
4. Using this comprehensive annotated sequence, predict domain boundaries (or a series of possible boundaries) for the protein, trying to maximize conservation and predicted structural features while minimizing the amount of disordered or flexible sequence. If known, the resulting protein should retain the functional properties of interest.

2. Express and Purify Proteins with the Domain Boundaries Identified in Section 1

NOTE: The goal of this section is to use a series of quick and easily quantifiable assays to screen hypothetical domain boundaries generated in Section 1.

1. Using standard molecular biology techniques, generate an expression plasmid with the desired coding sequence in frame within the His₁₀-mRuby2-XH2 plasmid (See Figure 3 for the vector map and additional details).
2. Test expression levels for each expression plasmid using BL21(DE3) E. coli or other suitable expression strain in autoinduction media¹ at room temperature for 18-24 hr.
   1. Transform expression plasmids as well as an empty plasmid control into BL21(DE3) following the instructions that came with the cells or using robust transformation protocol, such as the one here (https://www.addgene.org/plasmid-protocols/bacterial-transformation/).
   2. From the freshly transformed plate, pick a single colony and grow a 5 ml starter culture at 37 °C overnight in Lysogeny Broth (LB) media with 34 µg/ml kanamycin.
   3. The following day, pellet the culture and wash the pellet with fresh LB.
   4. Add the starter culture to 50 ml of autoinduction media with 34 µg/ml kanamycin. Allow the culture to grow to saturation at either room temperature or 37 °C. Typically, grow for ~18-24 hr.
   5. Pellet cells and freeze the resulting cell pellets at -80 °C indefinitely prior to purification.
   6. Compare whole cell extracts from each expression strain and the empty vector control by SDS-PAGE to determine the strain that most efficiently produces the desired fusion protein. For His₁₀-mRuby2-Shroom SD2 fusion proteins, run 10% SDS-PAGE gels at 180 V for ~50 min or until the dye front reaches the bottom of the gel²².
   7. Stain gel with Coomassie Blue to visualize total cellular proteins within each strain²³.
3. Perform an initial affinity chromatography step on each strain that successfully expressed mRuby2-fusion protein. Typically, perform purification steps 2.3.2-2.3.7 at room temperature, unless the protein will benefit from altered conditions such as purification at 4 °C.
   1. Thaw cell pellets from step 2.2.5.
   2. Resuspend each cell pellet in 1.5 ml of lysis buffer (10% glycerol, 500 mM NaCl, 40 mM imidazole, 20 mM Tris pH8.0, 1 mM beta-mercaptoethanol). Supplement the lysis buffer with protease inhibitors as appropriate.
   3. Add 30 µl of 10 mg/ml lysozyme and incubate at room temperature for 20 min. Sonicate following the instructions for the instrument being used.
   4. Transfer into a 1.5 ml tube and centrifuge in a table top centrifuge for 30 min at 14,000 x g to pellet insoluble material or unlysed cells. Save both supernatant and pellet for subsequent analysis via SDS-PAGE.
   5. Perform nickel affinity purification, incubating the soluble fraction with 100 µl of Ni-NTA beads. Incubate the beads with soluble lysate for 5-10 min, inverting several times to mix beads.
   6. Centrifuge at 800 x g for 30 sec in a tabletop centrifuge to pellet the beads. Follow this by washing the pellet 3-5 times with lysis buffer to clean off non-specific contaminants.
   7. Elute Ruby fusion protein from the beads with lysis buffer supplemented with 1 M imidazole.
   8. Use SDS-PAGE to compare the behavior of His-mRuby2-Shroom proteins in the "quick and dirty" purification above.

3. Characterize Protein Sample to Identify Those with Advantageous Properties

1. Use a spectrophotometer set at 280 nm to measure the concentration of the Ruby fusion protein, and then assess the homogeneity of the sample by loading 1-5 µg of fusion protein onto a native PAGE gel²⁴. Run the 10% Native PAGE gel at 4 °C for 140 min at 175 V.
   1. Image the native PAGE using an imager equipped to visualize fluorescence, observing where the mRuby-tagged fusion migrates in this assay.
      NOTE: Be careful to observe fusion protein that is stuck in the well as this protein is likely aggregated. If a fluorescence imager is not available, one can often visualize and image concentrated samples of Ruby tagged protein under a black light or with a UV light box.
2. Perform limited proteolysis to identify stably folded domains. Incubate 95 µl of Ruby-Shrm SD2 fusion at ~1 mg/ml with 5 µl of 0.025% Subtilisin A.
   1. Sample the reaction at time points of 0, 0.5, 2, 5, 15, 60, and 120 min, removing 10 µl from the reaction for each time point and visualizing the progress of the reaction via SDS-PAGE using a 15% acrylamide gel.
   2. Stain with Coomassie Blue as in step 2.2.7 and evaluate whether a protease resistant species can be identified.
3. Integrate data from the purification efficiency, behavior in native PAGE, and limited proteolysis on the partially purified mRuby2-Shroom fusion proteins into a comprehensive assessment of the overall behavior of these proteins in solution.
   1. (Optional) If a suitable functional assay is available, check for activity at this point.

4. Producing High Quality Crystals of the Coiled-coil SD2 Domain from Shroom

NOTE: All steps within section 4.1 are performed at room temperature unless the protein would benefit from purification at a different temperature, usually 4 °C.

1. Using the 50 ml growths as a rough estimate of expression and purification potential, perform large scale expressions of mRuby2-Shroom SD2 with the goal of achieving 5-20 mg of completely purified sample. Typically 2 L of culture provides adequate starting material. After growth pellet cells by centrifugation at 8,000 x g for 10 min.
   1. Resuspend the pellet from the 2 L growth in lysis buffer as before, using ~2 ml of lysis per gram of frozen cell pellet if using lysozyme for lysis. Alternatively, resuspend at ~8 ml/g of pellet if using a homogenizer or French press. Pellet cellular debris via centrifugation at 30,000 x g for 30 min.
   2. Batch bind the soluble fraction from 4.1.1 using 10 ml of Ni-NTA resin. Pour into appropriate gravity column and allow unbound proteins to drain off.
   3. Wash resin 3-5 times with 40 ml of lysis buffer, followed by a wash with lysis buffer supplemented to 1 M NaCl.
   4. Wash resin with 40 ml of lysis buffer supplemented with 80 mM imidazole.
   5. Elute the Ruby-Shroom fusion protein with 40 ml of lysis buffer supplemented to 1 M imidazole. Manually fractionate the eluted protein into 4-10 ml fractions.
   6. Confirm fractions containing Ruby-Shroom fusion protein using 10% SDS-PAGE.
   7. Dialyze appropriate fractions (typically fractions 2-4) into 8% glycerol, 250 mM NaCl, 15 mM imidazole, 20 mM Tris pH8.0, 1 mM β-ME, using 6-8 kDa MWCO dialysis tubing. Add 1 mg of TEV protease for every 25-50 mg of fusion protein. Dialyze overnight at room temperature with slow stirring.
   8. Repeat nickel affinity purification steps 4.1.2. to 4.1.6. The Shroom SD2 domain should now remain in the unbound fraction while the His₁₀-mRuby2 will remain bound to the resin and will be found in the elution fractions. Confirm this using 12% SDS-PAGE staining with Coomassie Blue.
   9. Dialyze fractions containing Shroom SD2 domain into 8% glycerol, 100 mM NaCl, 20 mM Tris pH 8.0, 5 mM beta-mercaptoethanol buffer overnight.
   10. Perform anion exchange chromatography²⁵ using an FPLC, and eluting with a NaCl gradient from 0.1-1.0 M NaCl. Analyze fractions using 12% SDS-PAGE.
   11. Using a spin concentrator (MWCO of 10 kDa or greater depending on the protein being studied), concentrate peak fractions to ~ 0 mg/ml, and then perform size exclusion chromatography²⁶, analyzing peak fractions via 12% SDS-PAGE.
   12. Pool peak fractions and dialyze into 20 mM Tris pH 8.0, 50 mM NaCl, 2% glycerol, 1 mM β-ME, and concentrate to 10 mg/ml prior to crystallization.
       Optional: During this step, remove 5 µl samples at concentrations of 1, 2, 5, and 10 mg/ml during the course of concentrating the sample. Run a native PAGE loading 2-5 µl of each sample to look at the behavior of the sample during this step.
2. Screen a small array of 12 conditions to identify an optimal protein concentration for crystallization trials. The composition of these conditions is described in Figure 7.
   1. Using 24-well sitting drop crystallization trays, perform crystallization trials using the vapor diffusion method, pipetting 500 µl of each of the 12 conditions into separate wells followed by drops that initially contain 1 µl of well solution and 1 µl of protein sample on the coverslips. Please see²⁷ for additional information on this technique.
   2. Quickly seal the tray with clear tape and move the tray to a suitable temperature controlled environment that is vibration free. The temperature used can vary, but 4 °C, 16 °C, and 20 °C are quite common.
   3. Examine the drops in the trays over the next 3 days using a microscope at up to 100X magnification. At the end of the 3-day period, score each drop as containing no precipitation (clear), light precipitation, heavy precipitation (brown), or crystals. A suitable protein concentration should contain no more than 6 heavy precipitations.
   4. Using the protein concentration identified in 4.2, screen a broad range of commercially available crystallization screens. The use of a liquid handling or crystallization robot greatly speeds up this process while also minimizing error in the drops. It requires far less protein as well, allowing the user to screen more conditions with a single sample.
   5. Improve initial conditions identified from broad screens by systematically varying each of the variables within the crystallization drop using 24-well screening trays as before.

Results

A diagram depicting the workflow utilized in this system is shown in Figure 1 and includes three main stages. Computational analysis of the sequence is utilized to develop hypotheses about the domain boundaries of the coiled-coil protein of interest. An example of an annotated analysis of the Shrm2 SD2 domain is shown in Figure 2. In this diagram, the goal was to identify possible domain boundaries for a conserved domain at the C-terminus of the cytoskele...

Discussion

The protocol described here is designed to help the user identify domain boundaries within large coiled-coil proteins to facilitate their crystallization. The protocol relies on a holistic incorporation of a variety of data from computational predictions and other sources to generate a series of potential domain boundaries. These are followed by a set of biochemical experiments which are quick and inexpensive, and are used to further refine these initial hypotheses. Using this approach, the user could quickly eliminate p...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
BL21(DE3) Rosetta	Emd Millipore	70954-3
BL21(DE3) Star	ThermoFisher Scientific	C601003
BL21(DE3) Codon Plus	Agilent Technologies	230245
Lysozyme	Spectrum Chemical Mfg Corp	L3008-5GM
Ni-NTA resin	Life Technologies	25216
SubtilisinA	Spectrum Chemical Mfg Corp	S1211-10ML
24 well Cryschem Plate	Hampton research	HR3-160
INTELLI-PLATE  96:	Art Robbins Instruments	102-0001-03
PEG 3350	Hampton research	HR2-591
PEG 8000	Hampton research	HR2-515
PEG 400	Hampton research	HR2-603
PEG 4000	Hampton research	HR2-605
pcDNA3.1-Clover-mRuby2	Addgene	49089
Overnight Express Autoinduction System 1	Emd Millipore	71300
Lysogeny Broth powder	ThermoFisher Scientific	12795027