Preliminary crystallographic characterization of PrnB, the second enzyme in the pyrrolnitrin biosynthetic pathway

Crystals of PrnB, the second enzyme in pyrrolnitrin biosynthesis are reported.

The transformation catalyzed by PrnB has no obvious chemical precedent. Although tryptophan-degrading enzymes such as indolamine dioxygenase and tryptophan dioxygenase are well known (Brady, 1975), both result in oxygen-containing compounds. This is not the case for PrnB; there is no additional oxygen in the product and PrnB decarboxylates its substrate. Sequence-comparison methods do not identify any protein with convincing homology to PrnB, further complicating any attempt to rationalize the chemical mechanism. In order to probe the chemistry of the enzyme, we have initiated a crystallographic study.

Purification and site-directed mutagenesis
Genomic DNA was extracted from P. fluorescens (BL915) using the Qiagen Blood and Cell Culture DNA Kit. An overnight culture (10 ml) of P. fluorescens (BL915) was processed on a genomic DNApurification column. Genomic DNA quality was assessed by running DNA gels. The DNA for PrnB was cloned using gene-specific primers and the clones were checked by sequencing. The verified PrnB cDNA was ligated into a variety of Escherichia coli expression vectors, but none gave soluble protein. The failure to express soluble PrnB in E. coli led us to investigate its expression in P. fluorescens (BL915). We employed the same P. fluorescens (BL915) ÁORF1-4 expression strain that had successfully been employed in the study of PrnA (Dong et al., 2004(Dong et al., , 2005Keller et al., 2000). In this strain, all four enzymes in the pyrrolnitrin pathway have been deleted from the chromosome. The prnB ORF was cloned into the pCIB-HIS vector, which was used for PrnA overexpression in P. fluorescens (BL915) ÁORF1-4 (Dong et al., 2004(Dong et al., , 2005Keller et al., 2000). This vector contains an N-terminal His tag. E. coli TOP10 cells were transformed with pCIB-HIS containing the PrnB insert and grown to produce a stock of high-quality plasmid DNA. Some plasmid was transformed into E. coli S17.1 strain. The E. coli strain was mixed with P. fluorescens (BL915) ÁORF1-4 and the plasmid was transferred between the bacteria by conjugation. Plasmid DNA was extracted from tetracycline-resistant clones of P. fluorescens (BL915) ÁORF1-4. Clones carrying the correct plasmids were chosen for protein expression. P. fluorescens (BL915) ÁORF1-4 cells were grown as described previously (Dong et al., 2004).
pCIB-HIS-prnB gave overexpression of soluble His-PrnB, which was confirmed by mass spectrometry. For analysis of in vitro activity, one part cells was resuspended in two parts 50 mM potassium phosphate buffer pH 7.2 and lysed by sonication. The addition of 100 mM d/l-7-chlorotrytophan to the cell lysate containing His-PrnB led to the production of monodechloroaminopyrrolnitrin. The presence of this product was confirmed by HPLC analysis according to a previously published method (van Pee et al., 1980;Hohaus et al., 1997). Control samples (heat-treated or no His-PrnB) gave no such product. For purification of PrnB, cells were harvested by centrifugation and lysed by sonication on ice into 20 mM Tris-HCl pH 7.5, 100 mM NaCl. The soluble fraction was applied onto a 5 ml Ni 2+ Hi-Trap column and eluted with an increasing gradient of imidazole (0-400 mM). Fractions containing His-PrnB were pooled before running on a Superdex S200 gel-filtration column. The protein ran inconsistently between batches, but always showed at least two peaks corresponding to a monomer and a dimer. The presence of 5 mM dithiothreitol (DTT) resulted in predominantly the monomeric species. This protein was concentrated to 16 mg ml À1 , but failed to give useful crystals using Nextal Classics, JSGS+ (Qiagen) and Hampton PEG/Ion crystallization kits.
The sensitivity of the gel-filtration profile to DTT suggested that PrnB was aggregating owing to disulfide-bond formation. As a result, we decided to mutate all three Cys residues to Ser (C21S, C60S and C175S). The P. fluorescens (BL915) PrnB triple mutant (tmPrnB) was created using the QuickChange Multi Site-Directed Mutagenesis Kit. Only one round of mutagenesis was required and the following primers were used (the mutation is marked in bold and the changed triplet is underlined): 5 0 -C21S, ACGCTGCCGTGGCGGCCTCC-

Crystallization and X-ray data collection
For crystallization experiments, tmPrnB was concentrated to 16 mg ml À1 (measured by the Bradford assay) in 20 mM Tris-HCl pH 7.2. Saturated solutions of each enantiomer of tryptophan were used; saturated solutions were made by agitating excess solid material with the protein solution. The precise concentration of the tryptophan is therefore unknown. Crystallization trials were set up at the Scottish Structural Proteomic Facility with a nanodrop crystallization robot (Cartesian HoneyBee) on sitting-drop plates (Hampton Research CrystalQuick 96) using Nextal Classics, JCSG+ (Qiagen) and Hampton PEG/Ion crystallization kits. Each drop contained 200 nl protein solution plus 200 nl precipitant in vapour equilibrium with 100 ml precipitant. We failed to obtain crystals of protein unless tmPrnB was first incubated with tryptophan. Crystals were obtained from only one condition (during the four weeks of observation), which was the same for both d-and l-tryptophan. This crystallization condition was optimized independently for both incubations by increasing drop size and systematically varying the pH and precipitant concentration. The best results (judged by crystal appearance) were obtained using sitting-drop plates (Hampton Research Crys-talClear Strips 96-well plates) with 1.5 ml protein solution (16 mg ml À1 ) plus 1.5 ml precipitant [0.22 M magnesium sulfate, 16%(w/v) PEG 3350] irrespective of whether protein was incubated with d-or l-tryptophan. Crystals tended to grow (but not always) as cluster of plates/rods, both habits being visible within the same well. We have been able to grow crystals of protein incubated with l-7-chlorotryptophan (using the same method of saturating the solution) under the same conditions. Irrespective of the incubating ligand, the crystals typically measure 0.2 Â 0.2 Â <0.05 mm. A plate was detached using a loop from a crystal cluster obtained from protein pre-incubated with l-tryptophan. The plate was cryoprotected with 20% (R,R)-2,3-butanediol (Fig. 2). Data were collected at 100 K on Station 9.6 at the Synchrotron Radiation Source, UK (SRS) using an ADSC Quantum 4 CCD. A single plate-shaped crystal grown in the presence of d-tryptophan was cryoprotected in the same way and a data set was collected at 100 K on beamline ID29 using an ADSC Quantum 4 CCD. A single rod-shaped crystal grown in the presence of l-7-chlorotryptophan was cryoprotected in the same way and a data set was collected on Station 20 at the SRS using an ADSC Quantum 4 CCD. All three data sets were collected as a series of 0.5 oscillations. All images were indexed in MOSFLM (Leslie, 1992) and merged with SCALA (Evans, 1997) as implemented in the CCP4 package (Collaborative Computational Project, Number 4, 1994). Full statistics are given in Table 1. Interestingly, the unit cells of the l-tryptophan-incubated crystals appear to be related to the other two (with a and b common), but the data are distinct. The lower symmetry space group for the crystals grown in the presence of d-tryptophan and 7-chlorotryptophan appears to be correct; we failed to index either in higher symmetry. Matthews solvent-content calculations suggest that all crystals contain a monomer in the asymmetric unit, but the l-tryptophan-grown crystals contain significantly less solvent.

Results and discussion
We have obtained good-quality crystals of PrnB from P. fluorescens (BL915) grown in the presence of substrate and two substrate mimics. The lack of homology of PrnB to proteins of known structure has ruled out molecular replacement as a possible method of structure solution. Therefore, we will have to use experimental phasing methods relying on the incorporation of anomalous scatters. We have shown the P. fluorescens (BL915) system can be used to incorporate selenium in much the same way as in E. coli (Dong et al., 2004). As the crystals give high-quality data, we expect this to be straightforward. In both crystals, we have obtained the result that there is a monomer in the asymmetric unit. The mutation of Cys to Ser to prevent aggregation has been used in other systems (Pesce et al., 2003;Vallone et al., 2004) and was essential here. Once again, the P. fluorescens (BL915) expression system has provided a relatively simple method of rescuing apparently 'insoluble' protein from E. coli expression systems. The yield of the pure PrnB is around 5 mg from 1 l of culture. The system therefore possibly merits wider use in the structural biology community. We expect that a structural analysis of  A crystal of PrnB grown in the presence of l-tryptophan and mounted for data collection.