Crystallization of domains involved in self-assembly of the S-layer protein SbsC

Three different truncation constructs of the S-layer protein SbsC containing domains crucial for self-assembly could be crystallized. Native data were collected for the three crystal forms from crystals that diffracted to 3.4, 2.8 and 1.5 Å resolution.


Introduction
Crystalline bacterial cell-surface layers, termed S-layers, represent an almost universal feature of archaeal cell envelopes and have been identified in hundreds of different species of bacteria (Sleytr, 1978;Sleytr & Beveridge, 1999). They can be regarded as the simplest protein membrane developed during evolution. S-layers, in general, are monomolecular isoporous structures composed of a single protein or glycoprotein species with a molecular mass in the range 40-200 kDa (Sleytr, 1978;Sleytr & Beveridge, 1999;Sleytr & Messner, 2009). Electron-microscopic studies revealed that S-layer lattices can exhibit either oblique (p1, p2), square (p4) or hexagonal (p3, p6) symmetry, with a centre-to-centre spacing of the morphological units of approximately 3-35 nm (Pavkov-Keller et al., 2011;. Despite their ubiquitous appearance and their obvious importance in many prokaryotic organisms, the role of S-layers in nature is not completely clear. However, it is now recognized that S-layer lattices can provide the organism with a selective advantage by fulfilling a broad spectrum of functions (Beveridge, 1994;Beveridge & Koval, 1993;Sleytr et al., 1993Sleytr & Sá ra, 1997). Owing to their self-assembly ability, S-layers can be exploited as a patterning element for many nanobiotechnological applications Ilk et al., 2011;Sleytr et al., 2005Sleytr et al., , 2007Sleytr et al., , 2010Sleytr et al., , 2011.
The protein precursor of the oblique lattice-forming S-layer protein SbsC from Geobacillus stearothermophilus ATCC 12980  consists of 1099 amino acids including a 30-amino-acid-long Gram-positive signal peptide (UniProt O68840; Jarosch et al., 2000). Based on the gene sequence, different N-or C-terminally truncated S-layer protein constructs have been recombinantly produced and systematically surveyed for their self-assembly and recrystallization properties (Jarosch et al., 2001). These studies confirmed that the N-terminal part comprising amino acids 31-258 is exclusively responsible for cell-wall binding, whereas the larger, C-terminal part comprises the self-assembly domain responsible for the formation of the crystalline array. This result has been corroborated by surface plasmon resonance (SPR) and isothermal titration calorimetry (ITC) studies showing that the positively charged N-terminal region of SbsC binds specifically to a negatively charged secondary cell-wall polymer (SCWP; Ferner-Ortner et al., 2007;Pavkov et al., 2008).
The property of S-layer proteins to self-assemble into twodimensional crystals makes them very demanding candidates for structural studies. The crystallization and/or X-ray structures of several bacterial and archaeal truncated and soluble S-layer constructs that have been reported to date are as follows: N-terminal and C-terminal parts of SbsC from G. stearothermophilus ATTC 12980 (Kroutil et al., 2009;Pavkov et al., 2003Pavkov et al., , 2008, a truncated derivative of the low-molecular-weight (LMW) S-layer protein from Clostridium difficile (Fagan et al., 2009), the cell-wall-binding domain of the Sap protein from Bacillus anthracis comprised of three S-layer homology (SLH) motifs (Kern et al., 2011) and polypeptide chain constructs from the archaeal surface-layer proteins from Staphylothermus marinus (Stetefeld et al., 2000), Methanosarcina mazei (Jing et al., 2002) and M. acetivorans (Arbing et al., 2012). Recently, the X-ray structure of the S-layer protein SbsB from G. stearothermophilus PV72/p2 lacking the cell-wall-binding domain was reported (Baranova et al., 2012). The protein was crystallized in the presence of a nanobody as a crystallization chaperone preventing the selfassembly of the protein into two-dimensional crystals.
In order to complete the atomic structure of SbsC, we performed crystallization with different soluble truncation forms containing the domains responsible for self-assembly ( Fig. 1).

Cloning, expression and purification
The truncation constructs rSbsC  , rSbsC  , rSbsC  , rSbsC  , rSbsC  and rSbsC  were cloned and expressed according to published protocols (Jarosch et al., 2001;Kroutil et al., 2009). In brief, the gene sequences encoding the truncations were PCR-amplified using the respective oligonucleotide primers (given in Table 1), which introduced the restriction sites NcoI (including an ATG start codon) and XhoI at the 5 0 and 3 0 ends, respectively. For cloning, the resulting PCR products were ligated with plasmid pET28a+ (Novagen) and the recombinant plasmids were electroporated into Escherichia coli TG1 (Stratagene). For expression, the plasmids were established in E. coli One Shot BL21 Star (Invitrogen). Purification of the recombinant proteins was performed by fractionated ammonium sulfate precipitation and sizeexclusion chromatography as described by Pavkov et al. (2003), except that the lyophilization steps between the chromatography runs were omitted. Purified constructs were dialyzed against 50 mM Tris-HCl pH 7.2 and stored at 277 K. The proteins were highly soluble and no degradation was observed within a period of six months.

Crystallization
Initial crystal screening and optimization trials for all constructs were performed with an Oryx8 robot (Douglas Instruments) using commercially available Index (Hampton Research) and Morpheus (Molecular Dimensions Ltd; Gorrec, 2009) screens. All screens and optimization setups were performed in Douglas vapour-batch plates (Douglas Instruments), which were covered with 3 ml of an oil mixture consisting of paraffin (Merck) and silicone oil (Sigma-Aldrich) in a 3:1 ratio. Drops of 1 ml were pipetted for both initial screens and optimizations. For initial screening, a 1:1 ratio of protein and commercial screening solutions was used. Optimization experiments included variation of this ratio as well as variation of the concentrations of the different screening-solution components. Crystallization plates were incubated at 293 K, with the exceptions of those for rSbsC  and rSbsC  , which gave better crystals at 289 K.
Crystals of rSbsC (443-650) obtained using a protein stock solution at 5.5 mg ml À1 in 50 mM Tris-HCl pH 7.2 appeared after five-and-a-half weeks under optimized Index condition No. 40 [0.01 M citric acid pH 3.5, 3.0%(w/v) PEG 3350] with an end protein concentration of 2.25 mg ml À1 in the drop (Fig. 2b).
Several conditions yielded crystals of rSbsC (541-759) with octahedral morphology and with varying size and quality (Fig. 2c). The protein Graphical representation of full-length rSbsC with domains indicated. The SCWP binding domain is marked with dots and domains involved in self-assembly are shown in dark grey. Constructs for which diffraction-quality crystals were obtained are coloured black, whereas constructs that did not yield crystals are coloured light grey. Table 1 Oligonucleotide primer pairs used for PCR amplification of the gene sequences encoding the various rSbsC truncations.
Overhangs are underlined, restriction sites are shown in bold and start and stop codons are shown in italics.

Data collection and processing
Data collection from all crystals was performed at 100 K without additional cryoprotectant, since no ice rings were observed. Data sets from all truncation forms were collected on synchrotron beamlines (EMBL Outstation, Hamburg, Germany and Swiss Light Source, Villigen, Switzerland). Data sets were processed and scaled using the XDS program package (Kabsch, 2010). The unit-cell parameters, assigned space groups and data statistics of the best data set for each truncation form are shown in Table 2. The number of molecules in the asymmetric unit was derived from the self-rotation function calculated with MOLREP (Vagin & Teplyakov, 2010) as well as the approximate solvent content (Table 2).

Results and discussion
The structures of two SbsC C-terminal truncation constructs and the successful crystallization of the N-terminally truncated construct rSbsC  have been reported previously (Pavkov et al., 2008;Kroutil et al., 2009). The structure of rSbsC  , containing the first three N-terminal domains, has been determined to 2.4 Å resolution  Table 2 Data-collection and processing statistics.
Values in parentheses are for the outer resolution shell.
rSbsC  rSbsC  rSbsC  rSbsC (   (PDB entry 2ra1; Pavkov et al., 2008). The structure of a longer construct rSbsC  was at a low resolution and was only partially interpretable. Thus, domain-level information could be derived, but large parts of the structure could only be built as a poly-Ala model. Therefore, two new C-terminal truncation forms of similar length, rSbsC  and rSbsC  , including domains 1-6, were produced. However, these forms yielded no crystals. Owing to the fact that the partial structure of rSbsC  showed that the observed ring-like structure is stabilized by extra residues from domain 7 (Pavkov et al., 2008), a new longer construct rSbsC  was produced, for which crystals could be obtained. At the same time, constructs consisting of domains 4-5, 5-6 and 4-6 were subjected to crystallization. Diffraction-quality crystals were obtained for the first two constructs, rSbsC  and rSbsC  . No crystals were obtained for the construct rSbsC  . It appears that the existence of two flexible linkers between the three domains is detrimental to the formation of an ordered crystal lattice.
A search for structures with significant sequence homology to domains 4, 5 and 6 failed. Therefore, molecular replacement was performed using poly-Ala models of individual domains 4, 5 and 6 from rSbsC   (Pavkov et al., 2008). For all three truncation constructs, molecular replacement was performed with Phaser (McCoy et al., 2007). Manual inspection of the results obtained using Phaser confirmed that rSbsC  contains one molecule in the asymmetric unit, rSbsC  contains two molecules in the asymmetric unit and rSbsC  plus Ca 2+ contains one molecule in the asymmetric unit. No clear solution could be obtained using data from rSbsC (541-759) without Ca 2+ . We believe that the high flexibility of some loop and/or linker regions in the absence of Ca 2+ hampers the formation of stable crystal contacts.
Rebuilding and refinement of the structures is in progress and upon completion will yield the complete structure of the full-length SbsC protein.
This work was funded by projects J2841, P17885-N11 and P19794-B12 from the Austrian Science Fund (FWF). EME was supported by the US Air Force Office of Scientific Research (AFOSR) project FA9550-10-1-0223. TP-K was supported by the Federal Ministry of Economy, Family and Youth (BMWFJ), the Federal Ministry of Traffic, Innovation and Technology (bmvit), the Styrian Business Promotion Agency SFG, the Standortagentur Tirol and ZIT -Technology Agency of the City of Vienna through the COMET Funding Program managed by the Austrian Research Promotion Agency FFG.