Characterization of three novel genetic loci encoding bacteriocins associated with Xanthomonas perforans

Abstract

Bacterial spot is a destructive disease of tomato in Florida that prior to the early 1990s was caused by Xanthomonas euvesicatoria. X. perforans was first identified in Florida in 1991 and by 2006 was the only xanthomonad associated with bacterial spot disease in tomato. The ability of an X. perforans strain to outcompete X. euvesicatoria both in vitro and in vivo was at least in part associated with the production of three bacteriocins designated Bcn-A, Bcn-B, and Bcn-C. The objective of this study was to characterize the genetic determinants of these bacteriocins. Bcn-A activity was confined to one locus consisting of five ORFs of which three (ORFA, ORF2 and ORF4) were required for bacteriocin activity. The fifth ORF is predicted to encode an immunity protein to Bcn-A based on in vitro and in vivo assays. The first ORF encodes Bcn-A, a 1,398 amino acid protein, which bioinformatic analysis predicts to be a member of the RHS family of toxins. Based on results of homology modeling, we hypothesize that the amino terminus of Bcn-A interacts with a protein in the outer membrane of X. euvesicatoria. The carboxy terminus of the protein may interact with an as yet unknown protein(s) and puncture the X. euvesicatoria membrane, thereby delivering the accessory proteins into the target and causing cell death. Bcn-A appears to be activated upon secretion based on cell fractionation assays. The other two loci were each shown to be single ORFs encoding Bcn-B and Bcn-C. Both gene products possess homology toward known proteases. Proteinase activity for both Bcn-B and Bcn-C was confirmed using a milk agar assay. Bcn-B is predicted to be an ArgC-like serine protease, which was confirmed by PMSF inhibition of proteolytic activity, whereas Bcn-C has greater than 50% amino acid sequence identity to two zinc metalloproteases.

Fig 1. Identification of genes associated with Bcn-A and characterization of Bcn-A.

(A) Schematic map of ORFs in pXV12.1 and Bcn-A activity when wild-type clone and deletion mutants of four ORFs were transformed into X. euvesicatoria 91–106 and tested against X. euvesicatoria 91–106. (B) Subcloning of Bcn-A gene cluster to identify immunity gene. Immunity function associated with ORF5. (C) Supernatant from X. perforans was separated based on size exclusion. Activity was associated mostly with size exclusion > 100 kDa. (D) Bcn-A activity is associated with supernatant but not with cell fraction. (E) Alignment of Bcn-A repeating motif region. Number represents position of Y in YD conserved sequence. X- Conserved in WAPA, RHSA, and ligand-binding consensus sequences; X- Only found in RHS consensus sequences; X- Only found in ligand-binding consensus sequences. (F) Lowest energy homology model of BcnA generated by iTasser. The amino terminal plug domain is colored blue, the beta-propeller domain is colored cyan, the YD-barrel is colored green and the carboxy-terminal region is colored red.

Fig 2

(A) Effect of Bcn-A on the viability of X. euvesicatoria 91–106 expressing (x) or not expressing ORF5 (ν) in vitro. A bacterial suspension of X. perforans 91–118 MutBC was added to 100 ml nutrient broth to reach a bacterial concentration of 10⁷ CFU/ml. After 7 h incubation, 10⁶ CFU of challenge strains (X. euvesicatoria 91–106 containing pLAFR119 or pLAFR119ORF5) were added. After challenge strains were inoculated, broth was sampled at various times points and plated on nutrient agar amended with nalidixic acid and streptomycin to select for X. euvesicatoria 91–106. (B). In in vivo phyllosphere antagonism assay populations of X. euvesicatoria strain 91–106 carrying empty vector pLAFR119 (o) or pLAFR119-ORF5 (ο) were monitored at various time points in leaflets at various time points following infiltration of suspensions adjusted to 5 x 10⁷ CFU/mL into Bonny Best tomato leaflets that were infiltrated 18 h earlier with X. perforans 91–118 ΔBcnBC suspension adjusted to 5 x 10⁷ CFU/mL. Note that X. euvesicatoria 91–106 populations not expressing ORF5 declined significantly compared to X. euvesicatoria 91–106 expressing ORF5. Error bars indicate the standard error.

Fig 3. Identification of Bcn-B and characterization by homology modeling.

(A) Bcn-B is identified as an endoprotease Arg-C following subcloning of pXV6.0 and mutation by placing a stop codon (TAA) at the N terminus of the putative endoprotease. (B) A superposition of the lowest energy models of Bcn-B is shown. In (i), the models generated by the Phyre2 (blue) and RaptorX (tan) servers are shown. The Phyre2 server did not generate a model structure for amino acids 1 to 213 of Bcn-B, only the carboxy-terminal catalytic domain. In (ii), the model of the two domains predicted by RaptorX to exist at the amino terminus of Bcn-B are shown superposed with the x-ray crystallographic structures of CUB domains from Homo sapiens neurophilin-2 (light blue, PDB ID 6GH8) and H. sapiens TSG-6 (light purple, PDB ID 2WNO). (C) Superposition of homology models of the Bcn-B catalytic domain. In (i), the models generated by RaptorX and Phyre2 are colored tan and light blue, respectively. A superposition of the serine protease domain with the putative catalytic residues shown as sticks is found in (ii). In (iii), the lowest energy RaptorX homology model of Bcn-B is shown in C with the catalytic amino acids (Asp337, His289 and Ser414) labeled and (iv), superposed with trypsin from Fusarium oxysporium, the closest homologue identified by a structural search of the PDB (PDB ID: 1XVM).

Fig 4. Identification of Bcn-C and characterization.

(A) Bcn-C was identified as a metalloprotease following subcloning of pXV5.1 in pLAFR119 followed by expression in X. euvesicatoria strain 91–106 and using deferred antagonism assay (B)Superpositions of Bcn-C homology models. In (i), the lowest energy homology model of Bcn-C generated by the Phyre2 and RaptorX alogrithms are shown in tan and green, respectively. The putative zinc coordinating residue side chains in each model are drawn as sticks. In (ii), the Phyre2 homology model of Bcn-C (tan) is superposed onto the x-ray crystallographic structure of a zinc metalloprotease from Grifola fondosa (blue, PDB ID: 1GE7). The active site zinc ion in the Grifola fondosa structure is shown as a sphere with its coordinating water molecules (red).

Fig 5. Bcn-A inhibitory activity and Bcn-B and Bcn-C protease activity are associated with the general secretory pathway.

(A) 91–106 wild-type and 91–106ΔxpsD containing pXV12.1 (BcnA+) were compared for Bcn-A activity using the deferred antagonism assay. (B) 91–106 wild-type and 91–106ΔxpsD containing pXV5.8 (+Bcn-B) and pXV5.1 (+Bcn-C) were assayed for protease activity using a milk agar assay after 48 hr. Note that XpsD mutants were negative for activity.

Fig 6. Protease activity of bacteriocins.

(A) Protease activity of 91–118 wild-type, and 91–106 conjugated with empty vector or vector expressing Bcn-A, Bcn-B or Bcn-C were shown by a diffusion assay after 24 hr. Zones of clearing around the bacteria due to the degradation of the substrate were observed. Note that the wild-type strain and 91–106 expressing Bcn-B and Bcn-C but not Bcn-A produced clear zones indicating protease activity. (B) Protease activity of 91–118 wild type and mutants were detected by Protease fluorescent Detection kit (Sigma, Missouri, USA). Fluorescence measurements were read by cytofluor II (PerSeptive Biosystems USA). Mutations in Bcn-B and Bcn-C reduced protease activity slightly, whereas the double mutant of both genes resulted in a large drop in protease activity. (C) Inhibition assay with PMSF. Protease activity was determined for ME90 expressing Bcn-B and Bcn-C after treatment with the serine protease inhibitor, PMSF, using the Protease fluorescent Detection kit. Red is Bcn-C and dark lines are Bcn-B. Note that PMSF inhibits Bcn-B at all concentrations but that Bcn-C in not inhibited by PMSF.