Production, Crystallization and Structure Determination of C. difficile PPEP-1 via Microseeding and Zinc-SAD

Summary

Proline-proline endopeptidase-1 (PPEP-1) is a secreted metalloprotease and promising drug-target from the human pathogen Clostridium difficile. Here we describe all methods necessary for the production and structure determination of this protein.

Abstract

New therapies are needed to treat Clostridium difficile infections that are a major threat to human health. The C. difficile metalloprotease PPEP-1 is a target for future development of inhibitors to decrease the virulence of the pathogen. To perform biophysical and structural characterization as well as inhibitor screening, large amounts of pure and active protein will be needed. We have developed a protocol for efficient production and purification of PPEP-1 by the use of E. coli as the expression host yielding sufficient amounts and purity of protein for crystallization and structure determination. Additionally, using microseeding, highly intergrown crystals of PPEP-1 can be grown to well-ordered crystals suitable for X-ray diffraction analysis. The methods could also be used to produce other recombinant proteins and to study the structures of other proteins producing intergrown crystals.

Introduction

Clostridium difficile is one of the major causes of nosocomial antibiotic-associated diarrhea infections¹. This Gram-positive anaerobic bacterium is transmitted through its spore form via the fecal-oral route. In the past decade, new ''epidemic'' or ''hypervirulent'' strains (e.g. BI/NAP1/027) caused a drastic increase in new infections and fatality rates in North America and Europe². C. difficile-associated disease (CDAD) is a life threatening colon inflammation with high fatality rates³. The symptoms range from diarrhea⁴ to pseudomembranous colitis⁵ and the often-fatal toxic megacolon⁶.

Treatment of CDAD is difficult as the virulent strains are multidrug-resistant and the recurrence rate is high⁷. At the moment therapy includes the antibiotics metronidazole, fidaxomicin or vancomycin, or in repetitively recurrent cases fecal microbiota transplantation. New therapeutic strategies are urgently needed⁸. Some progress is recorded as the therapeutic monoclonal antibody Bezlotoxumab, targeting C. difficile toxin B⁹, has recently successfully passed phase III clinical trials and was filed for approval with the FDA and EMA. Additionally, new antibiotics are being tested at the moment at different stages of clinical trials¹⁰.

To develop effective treatment new therapeutic targets must be identified. The recently discovered C. difficile protease proline-proline endopeptidase-1 (PPEP-1; CD2830/Zmp1; E.C. 3.4.24.89) is such a promising target, as the lack of PPEP-1 in a knock-out strain decreases virulence of C. difficile in vivo¹¹. PPEP-1 is a secreted metalloprotease^12,13 cleaving two C. difficile adhesins at their C-terminus¹³ thus releasing the adherent bacteria from the human gut epithelium. Therefore, it is involved in maintaining the balance between the sessile and motile phenotype of C. difficile. To develop selective inhibitors against PPEP-1 and to understand how it recognizes its substrates intimate knowledge of its three-dimensional structure is indispensable. We have solved the first crystal structure of PPEP-1 alone and in complex with a substrate peptide¹⁴. PPEP-1 is the first known protease that selectively cleaves peptide bonds between two proline residues¹⁵. It binds the substrate in a double-kinked manner and stabilizes it via an extended aliphatic-aromatic network of residues located in the S-loop that covers the protease active site¹⁴. This substrate-binding mode is unique to PPEP-1 and not found in human proteases to date. This makes it a promising drug target, and off-target effects of inhibitors very unlikely.

To develop and screen selective PPEP-1 inhibitors in the future a large amount of pure and monodisperse PPEP-1 protein is needed. Furthermore, to determine the mode of binding of first inhibitors, co-crystal structures with PPEP-1 will have to be determined. In our hands PPEP-1 constantly produces intergrown crystals. Thus we have developed an optimization procedure to produce single diffraction-quality crystals of PPEP-1. In this protocol we describe in detail the production, purification, crystallization and structure solution of PPEP-1¹⁴. We use intracellular expression in Escherichia coli of a PPEP-1 variant lacking the secretion signal sequence, affinity chromatography and size exclusion chromatography with removal of the purification tag, followed by microseeding¹⁶ into an optimization screen and structure determination via zinc single-wavelength anomalous dispersion (zinc-SAD)¹⁷. This protocol can be adapted for production and structure determination of other proteins (e.g. metalloproteases) and in particular for proteins producing intergrown crystals. On request, plasmid DNA of the construct (pET28a-NHis-rPPEP-1) and diffraction data can be provided for educational purposes.

Protocol

1. Cloning and Construct Design

1. Clone the codon-optimized sequence (for E. coli) of C. difficile PPEP-1 without the signal peptide [amino acids 27-220, named hereafter recombinant PPEP-1 (rPPEP-1)¹¹] into the pET28a vector using NdeI and XhoI restriction sites (Figure 1) with a stop codon at the 3'-end (resulting vector pET28a-NHis-rPPEP-1). This produces a N-terminally 6xHis-tagged protein (NHis-rPPEP-1) with a thrombin cleavage site allowing removing the tag during purification (Figure 1). The plasmid contains a kanamycin resistance cassette for selection. The primers used for cloning are described elsewhere¹⁴.

Figure 1: Schematic representation of construct pET28a-NHis-rPPEP-1 and SDS-PAGE analysis of expression and all purification steps. (A) Vector map of NHis-rPPEP-1 cloned into pET28a vector using NdeI/XhoI created with PlasMapper. (B) Schematic representation of the NHis-rPPEP-1 construct (upper panel) and the final construct after thrombin-cleavage of the 6xHis-tag with the resulting additional GSHM-sequence at the N-terminus (lower panel). SDS-PAGE analysis (C) of the expression in BL21 (DE3) Star at 37 °C for 4 hr and (D) of samples from all purification steps (M: molecular weight marker). Please click here to view a larger version of this figure.

2. Expression and Purification of rPPEP-1

1. Expression of NHis-rPPEP-1
   1. Make up and autoclave LB (lysogeny broth) medium (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, adjust to pH 7.5 with NaOH). Supplement with kanamycin sulfate (50 µg/ml) just before use (LB/Kan medium).
   2. Inoculate a 200 ml overnight culture from freshly transformed E. coli in LB/Kan medium. Grow overnight at 37 °C with shaking at 220 rpm.
   3. On the next morning, check the OD₆₀₀ (optical density at 600 nm wavelength) of the overnight culture. Inoculate two 2.8 L baffled flasks containing 1 L LB/Kan medium each with the overnight culture to an OD₆₀₀ of 0.1. Supplement with three drops of aqueous-silicone emulsion to prevent excessive foam formation. Grow cells at 37 °C shaking at 180 rpm until the OD₆₀₀ reaches 0.6.
   4. Take a pre-induction sample for SDS-PAGE analysis (equivalent of 1 ml from a culture at OD₆₀₀ = 1); add IPTG to 0.5 mM final concentration to induce expression of NHis-rPPEP-1. Continue growing at 37 °C/180 rpm for 4 hr.
   5. Determine the OD₆₀₀ in a 10x dilution and take a harvest sample (equivalent of 1 ml from a culture at OD₆₀₀ = 1).
   6. Collect cells by centrifugation for 20 min at 7,000 x g and 4 °C. To remove residual LB medium resuspend cell pellets from 1 L of culture in 40 ml TBS buffer (Tris-buffered saline: 20 mM Tris-HCl, pH 7.5, 200 mM NaCl) and transfer to a 50 ml centrifuge tube. Collect cells by centrifugation for 10 min at 10,000 x g and 4 °C and store at -80 °C until use. Analyze expression (total lysates and soluble fractions) via SDS-PAGE¹⁸.
2. Purification of untagged rPPEP-1
   1. Take 50 µl samples of each purification step for SDS-PAGE analysis. Resuspend the cell pellet from 1 L of culture in TBS buffer supplemented with 10 µg/ml DNaseI. Use 5 ml of TBS/DNaseI per g of cells.
      1. Lyse the cells by sonication on ice/water using 30% amplitude for 15 min (2 sec pulses with 2 sec pause). Remove debris by centrifugation for 10 min at 10,000 x g and 4 °C and transfer supernatant to an ultracentrifuge tube. Clear lysate in an ultracentrifuge for 30 min at 165,000 x g and 4 °C.
   2. Work at 4-6 °C. Using a peristaltic pump or chromatography system equilibrate 2 ml of nickel-nitrilotriacetic acid (NiNTA) resin in a glass column with TBS buffer supplemented with 10 mM imidazole pH 7.5. Alternatively, use gravity flow.
      1. Adjust the cleared lysate with 1 M imidazole pH 7.5 to a final concentration of 10 mM. Apply the lysate to the column and wash stepwise with TBS buffer supplemented with 10 mM and 30 mM imidazole, respectively, until the UV absorption at 280 nm has reached the baseline.
      2. Elute the protein with TBS buffer plus 250 mM imidazole. Re-equilibrate the column to TBS supplemented with 10 mM imidazole and store overnight.
   3. Determine the protein concentration either at 280 nm by using the extinction coefficient of 25,900 M^-1 cm^-1 or by any other method (e.g. Bradford method¹⁹). Add 2 units of thrombin per mg of protein and dialyze the protein solution overnight at 4 °C against a 50x volume of TBS (50x of the NiNTA elution volume).
      NOTE: Take the correct blank for determination of protein concentration, as imidazole absorbs strongly at 280 nm.
   4. Pass the protein solution over the equilibrated NiNTA resin to remove uncleaved protein. Next, apply the same volume of TBS supplemented with 10 mM imidazole to the column to recover all cleaved protein. To clean the column, elute all remaining protein with 250 mM imidazole. Analyze samples via SDS-PAGE (Figure 1).
   5. Concentrate the protein solution to 4 ml in 10 min intervals at 4,000 x g and 4 °C using a centrifugal ultrafiltration unit. Mix the concentrating protein after each interval to prevent precipitation and aggregation. At this step occasionally some precipitation is observed for rPPEP-1 despite the mixing procedure.
      1. Apply the concentrated protein to a pre-equilibrated size exclusion chromatography column (in TBS buffer) at 4-6 °C. Elute the column with TBS buffer, collect 1 ml fractions and subject 5 µl of every second fraction to SDS-PAGE analysis. rPPEP-1 elutes in a single peak corresponding to a monomer (Figure 2). Occasionally a minor peak at larger molecular weight is observed (fronting peak), that corresponds to a dimer of the protein. The yield should be around 50 mg of pure protein per L of culture. Analyze all samples via SDS-PAGE¹⁸ (Figure 2).

Figure 2: Representative size exclusion chromatography and SDS-PAGE analysis of rPPEP-1. Size exclusion chromatogram (A280; absorbance at 280 nm) of purified untagged rPPEP-1 using a (16/600) column in Tris-HCl, pH 7.5, 200 mM NaCl at 6 °C. Based on the elution volume, rPPEP-1 migrates as expected for a 22 kDa protein, suggesting that it is predominantly monomeric. Rarely a minor fronting peak appears that corresponds to a dimer. (inset) SDS-PAGE analysis of the fractions from size exclusion chromatography (M; molecular weight marker). Every second fraction is applied. The faint bands below the main rPPEP-1 band correspond to occasionally occurring minor impurities. Please click here to view a larger version of this figure.

3. Crystallization and Crystal Optimization Using Microseeding

NOTE: rPPEP-1 crystallizes from conditions that constantly produce highly intergrown crystals not suitable for X-ray diffraction analysis (Figure 3). Therefore, an optimization strategy (Figure 4) was developed to obtain high quality crystals (Figure 5).

1. Initial screening of rPPEP-1 using commercial screens
   NOTE: Carry out crystallization trials in the sitting drop format using standard commercially available screens and a crystallization robot.
   1. Concentrate the purified protein to 12 mg/ml using a centrifugal ultrafiltration device in 5 minute intervals at 4,000 x g and 4 °C. Mix the concentrating protein after each interval to prevent precipitation and aggregation. Determine the protein concentration either at 280 nm by using the extinction coefficient of 25,900 M^-1 cm^-1 or by any other method (e.g. Bradford method). Equilibrate the protein to 20 °C. Clear away all particles and dust by centrifugation for 10 min at 16,000 x g and 20 °C.
   2. Either use already pre-filled crystallization plates sealed and stored at 4 °C or fill the reservoir wells of the plates with 70 µl of every crystallization condition. Equilibrate all crystallization plates to 20 °C. Work quickly, as the small volumes quickly dry out. Use a humidity chamber around the dock of the robot, if possible.
      NOTE: Use the following screens as a standard procedure: SaltRx, Index, PEG/Ion, Crystal, Wizard, PACT++, JCSG++.
   3. Set up the screen by pipetting protein and reservoir into subwells 2-4. Drop volume is 300 nl and the ratios (protein:reservoir) are 200:100 (subwell 2), 150:150 (subwell 3) and 100:200 (subwell 4) (in nl). Immediately seal the plate and place in a chamber at 20 °C.
   4. Inspect trays immediately after set-up, and then inspect every day during the first week followed by weekly inspection.
2. Co-crystallization of rPPEP-1 with ligands
   1. For co-crystallization of substrate peptide-rPPEP-1 complexes mix rPPEP-1 at 24 mg/ml in a 1:1 ratio (v/v) with a 7-fold molar excess of peptide solution (Ac-EVNPPVPD-NH₂) lyophilized powder solubilized in TBS buffer), which will give a final concentration of 12 mg/ml r-PPEP-1 protein and a 7-fold molar excess of peptide over PPEP-1. Incubate for 30 min at 20 °C and clear away all particles and dust by centrifugation for 10 min at 16,000 x g and 20 °C. Proceed with crystallization using the microseeding procedure as described for the unbound r-PPEP-1 protein.

Figure 3: Representative crystals from initial screens. Intergrown crystals from rPPEP-1 at 12 mg/ml grown in condition. (A) Crystal screen I/38 (1.4 M sodium citrate tribasic dehydrate, 0.1 M HEPES sodium pH 7.5; 200 nl:100 nl). (B) SaltRx screen/52 (2.4 M ammonium phosphate dibasic, 0.1 M Tris pH 8.5; 100 nl:200 nl) and (C) (200 nl:100 nl). (D) SaltRx screen/96 (60% v/v Tacsimate pH 7.0, 0.1 M BIS-TRIS propane pH 7.0; 200 nl:100 nl). The Scale bar = 0.2 mm. Volume ratios are always protein:reservoir. Please click here to view a larger version of this figure.

Figure 4: Optimization procedure for rPPEP-1 crystallization. Initial crystals from rPPEP-1 at 12 mg/ml of low diffraction quality and with multiple lattices (intergrown) were reproduced in a 24-condition optimization screen. Again, only intergrown crystals were observed in conditions containing 2.55 M ammonium phosphate dibasic. A seed stock was prepared from a single intergrown crystal and diluted 1:1,000 into the same condition (microseeding). A volume of 0.5 µl of the diluted seed stock was added into the remaining clear drops and single crystals grew in almost all conditions. Please click here to view a larger version of this figure.

3. Crystal optimization using microseeding
   NOTE: Highly intergrown crystals of rPPEP-1 appear after two days in a condition containing 2.4 M ammonium phosphate dibasic, 0.1 M Tris-HCl, pH 8.5 (SaltRx screen, condition E4, all three subwells) (Figure 3). An optimization procedure using a grid screen around the initial condition combined with microseeding was applied (Figure 4).
   1. Prepare a grid screen (Figure 4) comprising 24 conditions with 1.8 - 2.55 M ammonium phosphate dibasic (in steps of 0.15 M) and 0.1 M Tris-HCl pH 7.5-9.0 (in steps of 0.5 pH-units) from appropriate stock solutions (4 M ammonium phosphate and 1 M Tris buffers).
      NOTE: Use the Make Tray applet (http://hamptonresearch.com/make_tray.aspx) to calculate the volumes and the pipetting scheme to obtain 2 ml of every condition that allows to perform 10 optimization screens. The 4 M ammonium phosphate stock solution is difficult to prepare. Heat the solution during stirring to completely dissolve the powder in water.
   2. Pipette 200 µl of every grid screen solution into the wells of a 24-well plate and equilibrate at 20 °C.
   3. Manually set up the crystallization plate. The drop volume is 3 µl and the ratios (protein:reservoir) are 2:1, 1.5:1.5 and 1:2 (in µl). Here, use a positive displacement pipette to avoid air bubble formation. Immediately seal the plate and place in a chamber at 20 °C. Avoid air bubble formation.
      NOTE: After one to four days highly intergrown crystals appear in the four conditions containing 2.55 M ammonium phosphate dibasic and 0.1 M Tris-HCl pH 7.5 - 9.0 (Figure 4). No crystals are formed in the remaining 20 conditions with ammonium phosphate concentrations below 2.55 M. The microseeding procedure is used to obtain single crystals of rPPEP-1 in these conditions.
   4. Prepare a microseed stock by harvesting a single intergrown crystal from one of the two conditions with 2.55 M ammonium phosphate dibasic and 0.1 M Tris-HCl pH 8.0 or 8.5. The crystals might be attached to the plastic surface. Careful deformation of the surrounding plastic with an acupuncture needle helps to detach the crystals.
      1. Transfer 50 µl of the respective mother liquor into a 1.5 ml tube containing a small highly polished glass bead (beads-for-seeds). Using a mounted nylon loop transfer the crystal into 1 µl of the mother liquor placed onto a glass cover slide.
      2. Transfer the liquid containing the crystal into the tube and vortex at high speed for 30 sec. Make a 1:1,000 dilution of the seed stock into a new 1.5 ml tube containing the same freshly prepared condition and vortex thoroughly for 5 seconds.
         NOTE: Seeds stocks can be stored at -80 °C for later use.
   5. Remove the seal of the plate covering the 20 conditions with clear drops and pipette 0.5 µl of the seed stock (1:1,000 dilution) into the wells. Seal the plate and place in a chamber at 20 °C. Single crystals of high diffraction quality appear in 2-7 days (Figure 5).

Figure 5: Representative crystals from optimization screen. Single crystals from rPPEP-1 at 12 mg/ml seeded with 1:1,000-dilution seed stock grown in the following conditions: (A) 2.1 M ammonium phosphate dibasic, 0.1 M Tris pH 7.5; 1.5 µl:1.5 µl; (B) 2.1 M ammonium phosphate dibasic, 0.1 M Tris pH 7.5; 2 µl:1 µl; (C) 2.25 M ammonium phosphate dibasic, 0.1 M Tris pH 8; 2 µl:1 µl; (D) 2.1 M ammonium phosphate dibasic, 0.1 M Tris pH 8; 1 µl:2 µl. (E) Mounted crystal in 0.1-0.2 µm nylon loop, grown in 2.1 M ammonium phosphate dibasic, 0.1 M Tris pH 8 (2 µl:1 µl) and cryo-protected in 2.1 M ammonium phosphate dibasic, 0.1 M Tris pH 8, 20% glycerol. The Scale bar = 0.2 mm in (A-D). Volume ration are always protein:reservoir. Please click here to view a larger version of this figure.

4. Crystal Mounting and Data Collection

NOTE: To obtain the best quality of diffraction data crystals should be mounted at the peak of their quality and size. Crystals can be stored in liquid nitrogen until they are subjected to X-ray diffraction analysis at 100 K. Therefore, the condition from which they originate must be adjusted to cryo-conditions. rPPEP-1 crystals can be cryo-protected by addition of either 20% glycerol or 30% sucrose (replacement of water in the condition by the cryo-protectant).

1. Crystal mounting
   NOTE: All crystal manipulation steps should be performed under the stereomicroscope.
   1. Choose the optimal size of nylon loop for the maximal length of the chosen crystals. The typical longest axis of rPPEP-1 crystals is about 100-200 µm (Figure 5). Prepare a cover slide and the appropriate cryo-condition (e.g. 2.1 M ammonium phosphate dibasic, 0.1 M Tris, pH 8.0, 20% glycerol).
      1. Fill the foam dewars with liquid nitrogen, load the vial clamp with a vial and pre-cool it in the liquid nitrogen-filled 800 ml foam dewar. Place a cryo sleeve and a cryo cane holder marked with a suitable identifier in liquid nitrogen-filled 2 L foam dewar. Load the magnetic wand with a mounted nylon loop.
         NOTE: Wear protective clothing (eyeshield/glasses, gloves) when working with liquid nitrogen. Warm objects plunged into liquid nitrogen may produce spills.
   2. Cut open the sealing tape with a sharp scalpel and remove it with the forceps. Pipette 1 µl of the cryo-condition onto the cover slide (or alternatively in an empty well on the same plate) and remove the crystal from the drop by fishing it with the mounted nylon loop (Figure 5). Attached crystals can be easily detached from ground by deforming the surrounding plastic with an acupuncture needle.
      1. Quickly transfer the crystal to the drop of cryo-condition and let it equilibrate for 1 second. Fish the crystal out as quickly as possible and plunge-freeze in liquid nitrogen.
      2. When the liquid nitrogen around the mounted loop stops boiling, place the loop in the vial. Place the vial on the cryo cane holder and when loaded with 6 vials place a cryo sleeve around the holder. Store the crystals in a tank filled with liquid nitrogen until use.
2. Data collection
   NOTE: Data collection can be carried out at the home diffractometer, if available, or at a synchrotron beamline. For rPPEP-1 data were collected at the beamline X06DA of the Swiss Light Source, Paul-Scherrer-Institute, Villigen, Switzerland using a hybrid photon counting detector. The original data and all files used in structure determination can be provided upon request.
   1. Set up the wavelength of the beam to 1.282 Å (9,667 keV), which is the characteristic x-ray absorption edge energy (peak) of the element zinc. rPPEP-1 is a metalloprotease that contains a single zinc per molecule in the active site.
   2. Collect data at 100 K in the inverse-beam mode in 10° wedges for a total of 270° in each direction. The exposure time is 0.1 sec with 0.1° rotation per image. Set the transmission to 14% (0.14).
   3. To collect a native high-resolution dataset from a second crystal originating from the same crystallization condition set up the wavelength of the beam to 1.00 Å (12,398 keV). Collect data at 100 K. The exposure time is 0.1 sec with 0.1° rotation per image. Set the transmission to 70% (0.7).

5. Structure Determination via Zinc-SAD

NOTE: In order to determine the structure of rPPEP-1 via zinc-SAD some basic crystallographic knowledge is needed as well as the software packages XDS²⁰, Phenix²¹ and the program Coot²². For visualization of structures the program PyMOL²³ or Chimera²⁴ is needed. Data collected at the wavelength corresponding to the peak at the absorption edge of the element zinc can be used for single-wavelength anomalous dispersion (SAD)²⁵ to obtain phase information that can be extended for all protein atoms.

1. Data processing
   1. Process the two peak datasets (normal and inverse) using the software XDS (alternatively iMosflm or HKL3000) in space group P2₁2₁2₁ (space group 19) separating the Friedel's mates (anomalous data). The unit cell parameters should be around a, b, c (Å) = 43.17, 71.68, 117.70 and α=β=γ (°) = 90. This gives two HKL-files (reflection files).
   2. Inspect the file CORRECT.LP. Use data up to the resolution in which the CC_1/2 is at least 50%. Scale together both datasets/reflection files (HKL-files) using XSCALE. Inspect the file XSCALE.LP. Check how far the anomalous signal extends (SigAno) and note the resolution with an anomalous correlation (Anomal Corr) of about 30%, which is 2 Å in the case of the data used here collected to 1.67 Å. This is the resolution cut-off for the anomalous signal used in Phenix AutoSol.
   3. Convert the (scaled) HKL-file into a CCP4-format reflection file (named, for example, peak_anom.mtz) using XDSCONV creating an R_free subset of 5% and keeping the anomalous data (FRIEDEL'S_LAW=FALSE). Check the mtz-file for consistency with the program mtzdmp inspecting the unit cell parameters, the space group and the existence of the R_free subset (label FreeRflag) and the anomalous data (labels DANO/SIGDANO). Prepare also an additional mtz-file with XDSCONV without extracting the anomalous data (FRIEDEL'S_LAW=TRUE; named, for example, peak_native.mtz) for refinement at a later stage.
2. Substructure solution (phase determination)
   1. Run Phenix AutoSol using the reflection file peak_anom.mtz. Select SAD/MAD peak as data type and choose 2 zinc sites (as there are two molecules per asymmetric unit). Choose either the more exact experimental values for the f'/f'' parameters (determined in a fluorescence scan at the beamline) or the theotetical values f' = -8.245 and f'' = 3.887. Also load the FASTA file containing the amino acid sequence of the crystallized protein.
   2. Set the resolution limit to the resolution with an anomalous correlation (Anomal Corr) of about 30% (determined in 5.1.2), in this case 2 Å and select the "autobuild model" option. Using the phases of the two zinc sites found by Phenix HySS (part of the Phenix AutoSol pipeline) the phases for the whole protein could be deduced and the model built (by Phenix RESOLVE) into the electron density. The best model is called "overall_best.pdb".
3. Model building, refinement and validation
   1. Select the option "autobuild model" to build most of the rPPEP-1 model automatically. Inspect the electron density at 1.0 σ contour level using the program Coot (Figure 6). It should be connective and surrounding the atoms of the model. Ideally also some water molecules should be built into the model (at a resolution better than 2.5 Å). Bulk water (space between the molecules) should contain no density.
   2. Check if the whole model is complete (all amino acids built into the electron density). If not, build them manually using the tools provided by Coot. Refine the structure by running iterative rounds of Phenix Refine with 5 refinement rounds each using the overall_best.pdb model file, the peak_native.mtz reflection file and the FASTA sequence file; and manual model building in Coot.
   3. Validate the quality of the structural model with the respective tools in Coot.

Figure 6: Experimental electron density map and model of rPPEP-1 after the Phenix Autosol run. Electron density in blue at a contour level of 1.0 σ displayed in the program Coot. In this initial map the electron density is nicely resolved and the model build into the electron density. The zoom in shows the residues His142 and Glu189, as well as a water molecule. Please click here to view a larger version of this figure.

6. Structure Determination to High Resolution via Molecular Replacement

NOTE: In order to obtain high-resolution structural information about rPPEP-1 a native dataset is collected. Then a molecular replacement procedure using the software Phaser^26,27 (within the Phenix software package) is employed using the structure determined via zinc-SAD as a model. This procedure can be also used later when solving structures of rPPEP-1 complexed with small molecules.

1. To obtain a crystal structure with higher resolution (in this case up to 1.4 Å) process the native dataset using the software XDS (alternatively iMosflm or HKL3000) in space group P2₁2₁2₁ (space group 19). The unit cell parameters should be around a, b, c (Å) = 43.17, 71.77, 117.80 and α=β=γ (°) = 90. This gives one HKL-files (reflection file).
2. Inspect the file CORRECT.LP. Use data up to the resolution in which the CC_1/2 is at least 50%. Convert the HKL-file into a CCP4-format reflection file (named, for example, native.mtz) using XDSCONV creating an R_free subset of 5%. Check the mtz-file for consistency with the program mtzdmp inspecting the unit cell parameters, the space group and the existence of the R_free subset (label FreeRflag).
3. Prepare the PDB-file containing the model from overall_best.pdb determined earlier and remove all water molecules and all ligands (i.e. the zinc atom). Also load the FASTA file containing the amino acid sequence of the crystallized protein. Run Phaser in Phenix using the reflection file native.mtz. Search for two molecules per asymmetric unit.
4. After successful structure solution (TFZ-score larger than 8; here 10.2) inspect the model (named native_phaser.1.pdb) and the electron density map in Coot. Build and refine the structure by running iterative rounds of Phenix Refine with 5 refinement rounds each using the native_phaser.1.pdb model file, the native.mtz reflection file and the FASTA sequence file; and manual model building in Coot.
5. Validate the quality of the structural model with the respective tools in Coot.

Representative Results

rPPEP-1 is overexpressed in several E. coli strains, with the highest yield in E. coli BL21 (DE3) Star (Figure 1C). After the first NiNTA affinity chromatography step the 6xHis-tag can be successfully cleaved off from most of the protein and in the second NiNTA step undigested protein can be completely separated from thrombin-digested protein (Figure 1D). On a S200 16/600 column untagged rPPEP-1 migrates as monomer with occasional fronting most probably corresponding to the dimeric species (Figure 2A). The protein is pure (Figure 2B) and the yield is about 50 mg protein from 1 L of E. coli culture. rPPEP-1 crystallizes in various conditions (Figure 3), often containing phosphate. The crystals are highly intergrown in all conditions, so crystal optimization had to be performed. Figure 4 depicts the scheme of the optimization procedure performed for crystals originating from condition commercial screen SaltRx E4 (52) (Figure 3B-C). Again intergrown crystals appeared in wells A6-D6 (Figure 4). A microseed stock was prepared from crystals grown in condition C6 (2.55 M ammonium phosphate dibasic, 0.1 M Tris pH 8.5) of the optimization screen and diluted 1:1,000 in mother liquor. Two to seven days after seeding with the diluted seed stock into the other remaining 20 conditions with clear drops large single-lattice crystals of high diffraction quality were observed (Figure 5) in most drops. After mounting of the crystals into nylon loops (Figure 5E) diffraction data to 1.67 Å resolution were collected for structure determination via zinc-SAD (Table 1), with an anomalous signal extending to 2 Å. Additionally, native diffraction data to 1.4 Å resolution were collected for determination of a high-resolution structure (Table 1). The crystal structure of rPPEP-1 could be solved (Figure 6) and the model of the native structure refined to R/R_free factors of 0.156/0.182 at 1.4 Å resolution (PDB ID: 5A0P)¹⁴ (for further refinement statistics see Table 1).

Structure/dataset	peak Zn-SAD	native
Data collection
Space group	P212121	P212121
a, b, c (Å)	43.17, 71.68, 117.70	43.17, 71.77, 117.80
α, β, γ (°)	90, 90, 90	90, 90, 90
Wavelength (Å)	1.28254	1
Resolution (Å)	45.5-1.67 (1.72-1.67)	45.5-1.40 (1.49-1.40)
No. of observations	779717 (34025)	310074 (45851)
No. of unique reflections	80960 (5597)	71874 (11172)
Multiplicity	9.6 (6.1)	4.3 (4.1)
Completeness (%)	99.2 (93.0)	98.2 (95.6)
Rmerge (%)	6.8 (56.5)	5.2 (52.6)
<I/σ(I)>	21.86 (2.81)	15.66 (2.48)
CC(1/2) (%)	100 (85.1)	99.9 (85.5)
Refinement statistics
Rwork/Rfreed (%)		15.60/18.17
No. of non-H protein atoms		3109
No. of water molecules		532
No. of ions/heavy atoms		2 Zn
No. of other molecules		-
No. of TLS groups/chain		9
Root-mean-square deviations
Bond lengths (Å)		0.006
Bond angles (°)		1.022
Average B factor (Å2)
All protein atoms		16.12
Water		29.74
Other atoms		10.12
Ramachandran plote (%)
Most favored		98.72
Additionally allowed		1.28
Disallowed		0
PDB entry		5a0p

Table 1: Data collection and refinement statistics.

Discussion

X-ray crystallography is still the fastest and most accurate method to determine three-dimensional near-atomic resolution structures of proteins²⁸. However, it requires the growth of well-ordered single crystals. These are often difficult to get and the crystalline state is artificial. However, a comparison of protein structures determined by X-ray crystallography with those determined by other methods, especially NMR, shows generally a very good agreement. In the case of PPEP-1, an NMR structure published recently²⁹ shows excellent agreement with our crystal structure¹⁴, including the mobility of the S-loop.

This protocol describes the production and purification of N-terminally His-tagged rPPEP-1 protein for structural studies and the crystallization and structure determination of untagged rPPEP-1. Single crystals were difficult to grow in this case and required a special microseeding procedure. In the following section we will discuss the results for PPEP-1 and indicate how the protocol could be adapted for the production and crystallization of any other protein.

Variations on construct design and expression

rPPEP-1 is expressed as an N-terminally His-tagged variant (Figure 1A), as for crystallization it is preferable to remove the His-tag by thrombin digest due to its possible impact on crystallization success and protein structure³⁰. For other proteins it may be advisable to additionally test a C-terminally His-tagged version (e.g. cloned using the same restriction sites, the vector pET22b and a reverse primer without stop codon at the 3' end) or to leave the N-terminal His-tag uncleaved, as in some cases a tag may help during crystallization. For rPPEP-1 the yield and stability of a C-terminally His-tagged construct was inferior. Additionally, an initial testing for best soluble protein expression should be included in the following E. coli strains: BL21 (DE3), BL21 (DE3) pLysS or Lemo21 (DE3), BL21 (DE3) codon plus RIPL or Rosetta 2 (DE3), BL21 (DE3) Star and C41 (DE3) at three different temperatures / incubation times (3-4 hr at 37 °C, 5 hr at 30 °C and overnight at 20 °C). Before cloning, check for the existence of an internal thrombin cleavage site by using the ExPASy PeptideCutter tool³¹ to prevent cleavage of the protein of interest. Alternatively, there are vectors available providing a cleavage site for HRV3C (i.e. EMBL's pETM-14³²) or TEV protease (i.e. EMBL's pETM-11).

Protein purification

For rPPEP-1 constructs cell lysis is performed via sonication on ice/water. For more sensitive proteins that tend to aggregate or precipitate a more "gentle" cell lysis method involving a cell disruptor might be used. Check for large amounts of insoluble protein after the first centrifugation step, which are not detected when using lysis method such as chemical cell lysis. For the purification of rPPEP-1 usually 2 ml of the NiNTA resin were used. Read the manufacturer's instructions on how large the protein binding capacity of the chosen resin is and adjust accordingly. In the first purifications of rPPEP-1, where only 1 ml of resin was used, a lot of protein was not bound to the column and was found in the flow-through (Figure 1D). On the other hand using too much resin might lower the purity of the purified protein. Adjust the imidazole concentration to the binding affinity of the used His-tagged construct to the NiNTA matrix. A stepwise washing procedure is preferred here (i.e. 10 mM, 30 mM, 50 mM, 70 mM imidazole), in which the A280 is monitored and is allowed to reach the baseline during each step. In that way no protein is lost, even if it would elute during a high imidazole wash step (e.g. at 50 or 70 mM). In the case of NHis-rPPEP-1 some protein already elutes at 30 mM imidazole, but it is unclear what kind of protein species it represents. Some proteins precipitate/aggregate when imidazole is dialyzed or diluted out of the protein solution. In such cases reduce the concentration of imidazole in the elution buffer to 150 mM and the time of exposure to high imidazole concentrations, e.g. decrease the concentration of imidazole by elution into a beaker containing a 5-10-fold volume of buffer without imidazole compared to the planned elution volume. For NHis-rPPEP-1 incubation with 2 units thrombin per mg of protein overnight at 4 °C are sufficient to cleave the His-tag to almost 100%. Adjust the amount of thrombin for the protein of interest by using 1-10 units per mg protein at 4 °C to 20 °C. When concentrating the protein using an centrifugal ultrafiltration unit pause in intervals of 5-10 min and mix the protein to homogenize the concentration gradient building up in the concentrator. Otherwise highly concentrated protein could aggregate/precipitate.

Crystallization

rPPEP-1 constantly produces highly intergrown crystals (single crystals growing on top of each other, thus having multiple crystal lattices) that are not suitable for X-ray diffraction analysis. The explanation might be that too many nucleation events are happening in the nucleation zone of the phase diagram (Figure 7).

Figure 7: Phase diagram of a protein crystallization experiment. Crystals can only form, when the protein is supersaturated. Nucleation takes place in the nucleation zone and crystal growth in the metastable zone. When a protein is undersaturated, the drop will remain clear.

Lowering of protein or precipitant concentrations does not improve the situation as then no nucleation of rPPEP-1 is observed anymore and the drops stay clear. Microseeding was the method of choice, as tiny nuclei are brought directly into the metastable zone, in which rPPEP-1 crystals eventually grow (Figure 7). By designing the optimization screen in a way that the original condition (2.4 M ammonium phosphate dibasic, 0.1 M Tris pH 8.5) is located at the position C5 (Figure 4) (rather the high end of pH and precipitant concentration) most of the new conditions correspond to a condition with lower supersaturation with a high propensity to represent a part of the metastable zone¹⁶ (Figure 7). Thus, the nuclei that are brought in during seeding can potentially grow to large single crystals. To optimize that procedure (amount and size of newly formed crystals) the original seed stock can be diluted 10^-2 to 10^-5. Here several dilutions of the seed stocks could be tested to determine the best seed dilution for production of large single crystals. Alternatively streak seeding using an animal whisker or tail hair (rabbit, cat, chinchilla or horse) could be employed³³.

The procedure can be used to co-crystallize product-peptides and substrate-peptides of rPPEP-1. Crystals in space group P2₁ diffracting up to 1.25 Å were obtained from seeds stocks diluted 1:250 in the optimization procedure. Crystals grow in the two space groups P2₁2₁2₁ (unbound protein) and P2₁ (complex structures), while originating from very similar conditions within the ammonium phosphate screen of the optimization procedure. Due to the crystal packing found in both r-PPEP-1 crystal forms, principally all small molecules and peptides addressing the unprimed side of the active site alone could be soaked into the crystals. This side of the molecule is accessible from the solvent present in the crystal. However molecules addressing both sub-sites or the primed side only need to be co-crystallized with rPPEP-1, as S-loop opening would be required to accommodate them in the active site. Additionally, rPPEP-1 is contacted by the neighboring molecules at the exit from the substrate-binding site near the S3'site (promotion of crystal contacts in this area) — a fact that also limits the length of substrate peptides to 3 residues at the primed side in these two crystal forms.

Crystal mounting and cryo-protection

Mounting of protein crystals is a method that requires some skill in manipulation under the stereomicroscope and thus needs some practice. Choose the optimal length of nylon loop (or other loop of choice, e.g. litholoop) to prevent the excess of solvent around the crystal that contributes to background scattering and thus a lower signal to noise ratio of the diffraction data. The optimal loop size/length also makes fishing of the crystals easier as the crystal does not slip through the loop. For rPPEP-1 crystals, which are about 100-200 µm in the longest dimension, nylon loops of 0.1-0.2 mm size were chosen. The cryo-protectant may also contribute to worsening of data quality, as the crystal may encounter an osmotic shock when transferred to the cryo-condition. This may hamper or even destroy the inner order of the crystal. Carefully select the type and concentration of the cryo-protectant. rPPEP-1 crystals were cryo-protected with either 20% glycerol or 30% sucrose. If crystallizing a substrate-peptide complex or a complex of rPPEP-1 with another ligand, sucrose should be chosen as cryo-protectant as glycerol binds to the primed site of the substrate-binding site of rPPEP-1¹⁴.

Structure determination

The AutoSol solution obtained a BAYES-CC of 49.2 ± 18.4, a FOM (figure of merit) of 0.41, a skew of 0.17 and a correlation RMS of 0.85. The model map correlation is 0.86 and the R/R_free factors are 0.21/0.24. All these parameters indicate a successful structure solution.

Materials

Name	Company	Catalog Number	Comments
Genes / Vectors / cell strains
pET28a vector	Merck-Millipore	69864	Thrombin cleavable N-terminal His-tag
E. coli strain BL21 (DE3) Star	ThermoFisher Scientific	C601003	RNase H deficient
Codon-optimized gene (for E. coli) of PPEP-1 (CD630_28300)	Geneart (Thermo Fisher Scientific)	custom	amino acids 27-220
Name	Company	Catalog Number	Comments
Chemicals
Yeast extract	any
Tryptone	any
Antifoam B	Sigma-Aldrich	A5757	aqueous-silicone emulsion
Agar	any
Kanamycin	any
IPTG	AppliChem	A1008
Tris-HCl	AppliChem	A1087	Buffer grade
NaCl	any		Buffer grade
DNaseI	AppliChem	A3778
Imidazole	AppliChem	A1073	Buffer grade
Thrombin	Sigma-Aldrich	T4648
Ammonium phosphate dibasic	Sigma-Aldrich	215996
Glycerol 100%	any		purest grade
Sucrose	Sigma-Aldrich	84097
Liquid nitrogen	any		for storage and cryocooling of crystals
Name	Company	Catalog Number	Comments
Equipment (general)
Shaking incubator	any		providing temperatures of 20 °C - 37 °C
Glassware	any		baffled Erlenmeyer flasks (50 ml - 2.8 L)
Centrifuge for large culture volumes	any		centrifuge for processing volumes up to 12 L
Sonicator Vibra-Cell VCX500	Sonics	SO-VCX500	or any other sonicator / cell disruptor
Ultracentrifuge	any		centrifuge providing speeds up to 150,000 x g
NiNTA Superflow resin	Qiagen
Empty Glass Econo-Column	Bio-Rad	7371007	or any other empty glass or plastic column
Size exclusion chromatography column HiLoad Superdex 200 16/600	GE Healthcare	28989335
Chromatography system Äkta Purifier	GE Healthcare	28406264	or any other chromatography system
Dialysis tubing Spectra/Por 3	Spectrum Labs	132724
Dialysis tubing closures	Spectrum Labs	132738
Ultrafiltration units (concentrators) 10,000 NWCO	any
UV-Vis spectrophotometer	any
Name	Company	Catalog Number	Comments
Equipment (crystallography)
Low volume pipette 0.1-10 µl	any
Positive displacement pipette Microman M10	Gilson	F148501
Crystallization robot	any
96-well crystallization plates TTP IQ with three protein wells	TTP	4150-05810	or any other 96-well crystallization plate
24-well CombiClover Junior Plate	Jena Bioscience	EB-CJR
Crystal Clear Sealing Tape	Hampton Research	HR3-511
Siliconized Glass Cover Slides	Hampton Research	HR3-225
Commercial crystallization screens: SaltRx, Index, PEG/Ion, Crystal	Hampton Research	diverse
Commercial crystallization screens: Wizard, PACT++, JCSG++	Jena Bioscience	diverse
JBS Beads-for-Seeds	Jena Bioscience	CO-501
CrystalCap SPINE HT (nylon loops)	Hampton Research	diverse	loop sizes 0.025 mm - 0.5 mm
CrystalCap Vial	Hampton Research	HR4-904
Cryogenic Foam Dewar 800 ml	Hampton Research	HR4-673
Cryogenic Foam Dewar 2 L	Hampton Research	HR4-675
Vial Clamp, Straight	Hampton Research	HR4-670
CrystalWand Magnetic, Straight	Hampton Research	HR4-729
CryoCane 6 Vial Holder	Hampton Research	HR4-711
CryoSleeve	Hampton Research	HR4-708
CryoCane Color Coder - White	Hampton Research	HR4-713
Scalpel	any
Straight microforcep	any		for manipulation of sealing tape. etc.
Acupuncture needle	any		e.g. from a pharmacy
Stereo microscope	any		for inspection of crystallization plates and crystal mounting, magnification up to 160X