Previously, it had been shown that lscA was not expressed in PG4180 while variant BC alleles were functional [3]. When lscA of PG4180 along with its 940-bps upstream sequence was introduced to the lsc-negative E. amylovora mutant Ea7/74-LS6 [15], the resulting transconjugant showed levan formation on agar plates supplemented with 5% sucrose, in contrast to mutant Ea7/74-LS6 (Data not shown). Since lscA was placed in the opposite direction to vector-borne promoters of the plasmid, this result indicated lscA expression from its native promoter. To substantiate this, Western blot analysis with Lsc-specific antiserum and protein extracts of Ea7/74 wild type, its lsc-deficient mutant, and the Ea7/74-LS6 transconjugant carrying lscA was conducted, revealing a clear signal for Lsc in the transconjugant (Figure 1A). Concentrated cell-free supernatants of the Ea7/74 derivatives were spotted on water agar containing 5% sucrose revealing levan formation in the transconjugant thus confirming heterologous lscA expression (Figure 1A).

PG4180 LscA shows 87.5% identity at the amino acid level with variant BC enzymes and shares 75.9% identity to Lsc of Ea7/74. It was speculated that the common ancestor of PG4180 genes lscA, lscB, and lscC might be related to the lsc gene of E. amylovora, and that all three genes present in P. syringae might have diverged from this common ancestor. This hypothesis was supported by a phylogenetic analysis of all available protein sequences of Lsc’s built by the Neighbor-Joining method (Appendix Figure A1). Variant A alleles of all P. syringae strains were found to be clustered closer to the Lsc’s from Enterobacteriaceae including E. amylovora Ea7/74 as opposed to variant BC alleles. Aside of variant A alleles being present in single copy in several P. syringae strains and in E. amylovora, this gene variant was also found in single copy in other enterobacterial species such as Erwinia tasmaniensis and Rahnella aquatilis ATCC33071 (Table 1). Since these bacterial species are commonly found to be associated with host plant species such as soybeans [16,17], P. syringae, E. amylovora, and R. aquatilis might have exchanged genetic information during cohabitation and lscA can be considered as an example of such horizontal gene transfer.

It had been reported that three acidic residues are highly conserved among members of the Glycosyl hydrolase families 32, 43, 62, and 68 at the catalytic active sites termed as block I (Asp/Glu), block II (Asp) and block III (Glu) [12,18,19]. Therefore, conserved regions within the catalytic blocks of variant A and variant BC were compared for Ea7/74 and PG4180 (Figure 1B). The catalytic centers Asp (block II) and Glu (block III) shared a common codon usage for these nucleotide sequences with both organisms (www.kazusa.or.jp/codon/). This further suggested that lscA in PG4180 and lsc of Ea7/74 might share similar regulatory features particularly present in E. amylovora.

In contrast to variant A genes, variant BC lsc alleles are found only in P. syringae pathovars (Table 1, Appendix Figure A1). The coding sequences of all variant BC genes exhibited 94.0% identity at the nucleotide sequence level and 98.1 to 99.8% similarity at the protein sequence level, demonstrating a high degree of conservation among otherwise variable P. syringae pathovars [13,20].

Comparison of the up- and downstream sequences of known variant BC alleles interestingly revealed a common ~1.8-kb highly conserved nucleotide sequence with an average of 87.3% identity (Figure 2, Appendix Figure A2). This conserved region comprises 450 to 452 bp of upstream sequence, 1,296 bp of the lsc ORF and 49-51 bp of downstream sequence, possibly involved in the formation of stem-loop structure for ρ-independent transcriptional termination (Appendix Figure A3). To map the promoter element of variant BC genes, a nested deletion analysis of the lscB upstream sequences ranging from -666 to -50 bp of the respective lsc translational start (TS) was conducted (Figure 3). For this, the lscB- and lscC-deficient mutant PG4180.M6 was complemented with various plasmid-borne deletion constructs and the phenotypes of the transconjugants were analyzed with respect to levan production (Appendix Figure A4). Deletion constructs ending 5’ at position −440-bp upstream of the TS and any larger upstream sequence fully complemented the mutant PG4180.M6 with respect to levan formation. In contrast, the deletion construct ending at position −332-bp did not complement this mutant, thereby identifying the minimal promoter required for lscB expression (Figure 3). The experiment was repeated with respective deletion constructs for lscC giving identical results (Data not shown). Since nucleotides 450–452 bp upstream of variant BC ORFs are highly conserved in all four P. syringae strains analyzed (Figure 2, Appendix Figure A2), it may be speculated that the minimal promoter sequences of the variant BC alleles of the other P. syringae pathovars not experimentally studied are similar to those of strain PG4180. Nucleotide sequences flanking the conserved ~1.8 kb region varied considerably among the strains investigated, with the exception of two plasmid-borne variant BC alleles in strains DC3000 and 1448A, for which 93% of the nucleotides up to −2,800 bp with respect to the translational start sites were identical (Figure 2). This 2,800-bp upstream sequence of plasmid-borne variant BC genes is bordered by a truncated transposase gene resembling that of transposon ISPsy16 [6] (Data not shown).

Aside from the 49–51-bp highly conserved DNA sequences, downstream of the translational stop of variant BC genes, the nucleotide sequences of those alleles further downstream exhibited ~87% identity until +1,530 bp with respect to lsc translational stop codon, except for lscC of B728a and lsc-1 of DC3000, which both had an additional 130-bp 85% similar downstream sequence (Figure 2, Appendix Figure A2). Since strain B728a lacks any native plasmid [7] but possesses conserved sequences surrounding lscC (−478 bp with respect to translational start and +1,420 bp with respect to the translational stop of the lsc gene) to that of DC3000 lsc-1 (Figure 2), it was speculated that lsc-1 of the pv. tomato strain might have been associated with the phylogenetic source of lscC in B728a.

Nucleotide sequence alignments revealed sequences downstream of variant BC alleles with significant similarities to sequences of transposases (www.pseudomonas.com). These could be considered as presumable evolutionary scars of transposable elements involved in the course of variant BC allele distribution in P. syringae (Figure 2). In the genome of strain B728a a sequence showing 92% identity with transposase orfA of ISPsy24 [6] was located 1,450–1,488 bp downstream of the lscC stop codon. Interestingly, the sequence of 1,420 bp downstream of B728a lscC coincided with the 1,290-bp downstream sequences of lscC in strains 1448A and PG4180 as well as of lsc3 in strains DC3000 and T1 (Appendix Figure A2). This finding further supported the hypothesis that lscC of strain B728a might be derived from a similar source to that of lsc-1 from strain DC3000. Previously, studies on E. coli transposon IS911 showed that production of OrfA either in cis or in trans stimulated production of excised circular transposon copies, suitable for intermolecular transposition into a plasmid target [21]. Previously, IS elements had been predicted to be involved in the horizontal transmission of avirulence genes and the coronatine gene cluster in P. syringae [22,23]. In a comparative study, lsc-1 (chromosomal) was located on the variable region (VR) 38 and supposed to be in a hotspot zone [24]. It has been reported that the same location is functionally classified as a lineage-specific region (LSR) i.e., virulence type LSR no. 14, in the DC3000 chromosome. The majority of DC3000-specific genes were suggested to be linked with lateral gene transfer events and responsible for the fitness as well as adaptation to the environment. Such LSRs were also predicted to help in utilizing plant-derived energy sources, e.g., sucrose [6].

Interestingly, the 25-100-bp downstream sequences of lscC in strain B728a, of lscB in strains 1448A and PG4180, and of lsc-1 in strain DC3000, showed an overall 72% sequence identity with the 75-bp downstream sequences of variant A lsc alleles. Forty nucleotides of this 75-bp sequence were found to be conserved in the 3’-end of the nasT gene coding for a response regulator protein, required for expression of nitrite-nitrate reductase genes (Figure 2) suggesting but not proving a phylogenetic link between a putative initial insertion of an ancestral variant A allele to P. syringae, which later diverted into several chromosomal or plasmid-borne variant BC alleles.

A genome-wide comparison of the 450–452-bp upstream sequences of variant BC lsc alleles with the genomic sequences of strains 1448A, DC3000, and B728a, respectively, revealed 452-bp and 70% conserved sequences upstream of two putative family 2 glycosyl hydrolase genes located within the sequences of pro-phage PSPPH01 of strain 1448A and pro-phage GH5 of strain B728A (Table 2). The sequence focused-on herein was also found to be associated with a putative bacteriocin gene in DC3000 although to a lesser extent (Table 2). In Klebsiella sp., a gene cluster, required for production of the bacteriocin klebcin, was previously reported to be associated with a phage sequence suggesting its lateral gene transfer and diversification [25].

Interestingly, the 48-bp 5’-coding sequences of variant BC lsc alleles and the glycosyl hydrolase genes were found to be 80% identical. These 48 nucleotides encode with almost identical N-termini of variant BC lsc gene products as well as the putatively pro-phage-borne glycosyl hydrolases (Figure 4). Furthermore, the conserved upstream sequences included the experimentally determined promoter region of variant BC alleles and a phage-associated com gene encoding a putative translational regulator [26] (Figure 3).

The distribution of the ~500-bp conserved upstream and N-terminus-encoding sequences at various genomic positions in diverse P. syringae strains suggested their mobile nature (Table 2). Mobile DNA sequences possessing potential promoter regions are generally termed mobile promoter elements [27,28]. These DNA elements, often phage-associated, allow for expression of adjacent genes or re-activation of silent genes such as shown for the IS3-mediated activation of argE in E. coli [27], and are therefore termed phage-associated promoter elements (PAPE). Interestingly, the herein observed PAPE of P. syringae seems to be associated with genes encoding for extra-cellular levansucrases and putative glycosyl hydrolases, both of which might play an important role for nutrient acquisition and in planta fitness of the pathogen.

The location of the mapped minimal promoter required for lsc expression appeared to be located ≥332-bp upstream from the translational start of the lscB/C ORF. This stimulated some interest to scan this inter-genic region for additional genes by comparing this sequence with entries of the GenBank nucleotide sequence database using the method BLAST-N. The search for potential additional coding sequences revealed the presence of a 192-bp ORF starting at position -204 and ending -12 bp upstream of the translational start sites of lscB and lscC, respectively. This ORF was homologous to the com gene [29,30] (Figure 3). Its predicted amino acid sequence exhibited a high degree of similarity to Com translational regulators found in pseudomonad bacteriophages, including Mu-like phage of P. entomophila L48 (85.4% similarity), phage B3 of P. aeruginosa (78.3% similarity), and phage DVM 2008 of P. fluorescens Pf-5 (65% similarity). In-silico structural prediction at the ‘SUPERFAMILY database of the structural and functional protein website (supfam.mrc-lmb.cam.ac.uk/SUPERFAMILY/index.html) suggested that the predicted com gene products possess classic C2-H2 zinc-finger domain as reported for Com earlier [31]. Homologs of the putative com gene in PG4180 were also found upstream of the lsc and glycosyl hydrolase genes in 1448A, DC3000 and B728a (Figure 2). Moreover, data derived from the www.pseudomonas.com website suggested that the putative phage-borne cellulase gene (PSPPH0655) in 1448A and its homolog in B728a (Psyr_4600) are surrounded by phage-associated genes, which with caution suggested potential ancestral pro-phage insertions.

A potential impact of com on levan production was investigated by inserting a premature stop codon to com using site-directed mutagenesis of plasmid −666-lscB. The mutated plasmid was introduced to the levan-deficient mutant PG4180.M6 [3] and compared to a transformant of PG4180.M6 carrying a non-mutated plasmid. Both transformants exhibited similar levels of levan formation and Lsc secretion (Data not shown) demonstrating that the putative translational regulator, Com, was not involved in expression of lscB. Its potential role in controlling translation of the glycosyl hydrolase gene needs to be tested in future studies.

Over-expression of the glycosyl hydrolase gene associated with the herein discovered PAPE in PG4180 and the subsequent protein purification yielded no detectable cellulose-degrading activity [32,33] (Data not shown), suggesting that the encoded enzyme is not a cellulase. A growth-phase dependent transcriptional analysis of the glycosyl hydrolase gene of PG4180 was conducted using qRT-PCR to compare if its expression resembles that of lsc as published earlier [4]. Cells of PG4180 wild type were grown in minimal medium containing glutamate at 18 °C. Results indicated that lsc and the glycosyl hydrolase gene associated with the PAPE showed maximal expression at the early exponential growth stage (Figure 5). A very similar pattern of expression had previously been observed for the lscBC genes [4]. Thus, the result indicated that the PAPE might be the site of transcriptional regulation where common, regulator(s), yet-to-be-identified, might bind and lead to gene expression.

Due to its sequence similarity, its size, and the heterologous expression of the variant A lsc gene of PG4180 observed herein, it is tempting to speculate that variant A alleles were initially obtained by horizontal gene transfer from enterobacterial species such as E. amylovora. This assumption is fueled by the fact that P. syringae is the only known organism, in which multiple copies of lsc exist. Consequently, variant A might represent the most ancestral lsc variant in P. syringae. However, due to its potentially inactive promoter sequence, this variant remained cryptic in P. syringae. In an unknown sequence of events initially involving a gene duplication of lscA, a pro-phage insertion bearing an active promoter, a potential translational regulator, and the pro-phage-borne N-terminal sequence might have inserted upstream of one of the two variant A gene copies yielding an ancestral variant BC lsc copy. Subsequently, transposon-mediated transposition events might have led to a spreading of variant BC copies, now functional, in various P. syringae genomes. The latter assumption is indirectly supported by the fact that P. syringae pv. tomato PT23 and pv. glycinea race 4 contain multiple copies of IS1240 with the tnpA gene coding for a transposase [20,22,34].

Since there are several plasmid-borne variant BC alleles, it is tempting to speculate furthermore that conjugative transfer of lsc-bearing plasmids might have led to an accelerated distribution of variant BC alleles among P. syringae pathovars. Interestingly, the plasmid-borne lsc-3 of DC3000 shows an upstream sequence very similar to that of plasmid-borne variant BC alleles of strains PG4180 and 1448A. However, its downstream sequence compares better to the downstream sequences of chromosomal orthologs in strains PG4180 and 1448A, respectively, (Figure 2 and Appendix A2) thus complicating the clarity of further definition of the phylogenetic pathway. However, the presence of two copies of variant BC, of which one is located on a plasmid, might indicate the importance of Lsc for the evolutionary fitness of the leaf pathogen, P. syringae.

Bacterial strains and plasmids used in this study are listed in Table 3. Escherichia coli and Erwinia amylovora strains were maintained and grown on Luria-Bertani (LB) medium at 37 °C and 28 °C, respectively [14,35]. P. syringae cultures were grown in a modified Hoitink Sinden (HS) medium [36] supplemented with glutamate as sole carbon and nitrogen source at 18 °C. Bacterial growth in liquid LB media was continuously monitored by measuring the optical density at 600 nm (OD₆₀₀) and harvested for protein sampling at an OD₆₀₀ of 1.0. Antibiotics were added to the media at the following concentrations (µg/mL), respectively: ampicillin, 50; kanamycin, 25; tetracycline, 25. Cellulolytic activity was assessed according to Kasana et al., 2008 [32].

Small scale isolation of plasmid DNA, restriction enzyme digests, agarose gel electrophoresis, purification of DNA fragments from agarose gels, electroporation, ligation of DNA fragments and other routine molecular methods were performed using standard protocols [35]. Nucleotide sequencing was carried out commercially (Eurofins MWG Operon, Ebersberg, Germany). The 3.1-kb Pst1 fragment containing PG4180 lscA was obtained from pSKL3 [3] and re-cloned in pBBR1MCS-3 [37] in the opposite direction to the vector-borne lac promoter yielding pBBR3(lscA). This construct was subsequently electroporated into competent cells of Ea7/74-LS6 yielding the transformant LS6(lscA), which was then grown on LB agar plate containing Km^r and Tc^r. Later, the transconjugant was streaked on 5% sucrose-containing LB agar media.

Nested deletion analysis of the upstream region of lscB in plasmid pRB7.2 [3] was conducted using the Erase-a-Base^® kit (Promega, Madison, USA). For analysis of the lscC upstream region, PCR was used to generate products covering the same region as in the deletion constructs of lscB (Table 3). PCR products of the lscC upstream region were cloned in vector pBBR1MCS-3 (Table 4). All constructs were introduced to E. coli DH5α via electroporation and then transferred by tri-parental mating [38] with helper plasmid pRK2013 [39] to the lscB lscC mutant PG4180.M6 [3]. Transconjugants were streaked on 5% sucrose-containing MG agar medium for assessment of levan production.

Extra-cellular fractions obtained from Ea7/74, Ea7/74-LS6, LS6 (lscA), PG4180, and lsc deletion constructs-harboring PG4180.M6 transformants and the use of polyclonal antibodies were carried out as described previously [4]. For immunological detection of Lsc enzyme, Western blot experiments were performed with total extra-cellular fractions using polyclonal antibodies raised against purified Lsc of P. syringae pv. phaseolicola as described earlier [3]. Water agar plates with 1.5% agar and 5% sucrose were used for the qualitative visualization of extra-cellular Lsc. Lsc activity was quantified by measuring the amount of glucose liberated during incubation with sucrose using the Gluco-quant Glucose/HK assay kit (Roche Diagnostics, Mannheim, Germany) at an absorbance of 340 nm. One unit of Lsc activity corresponded to the amount of enzyme which liberates 1 μmol glucose per minute from sucrose. The experiments were repeated three-fold and mean values were expressed as the quantity of glucose release.

Bacterial cells were grown in HS + glutamate medium at 18 °C. When cultures reached distinct OD₆₀₀ values, total RNA was isolated by acid phenol/chloroform extraction as described previously [40]. The yield and the purity of RNA were determined by measuring absorption at 260 and 280 nm. Total mRNA samples were treated with TURBO DNA-free (Applied Biosystems, Darmstadt, Germany) to remove remaining traces of genomic DNA as described by the manufacturer’s recommendation.

SYBR green-based qRT-PCR was performed with 1 ng RNA template and 200 nM primers (cel_RT_fwd and cel_RT_rev) (Table 4) using the QuantiTect SYBR Green one-step RT-PCR Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The thermocycler program comprised an initial step of 95 °C for 15 min followed by 40 cycles of 95 °C for 30 s, 55 °C for 30 s, 72 °C for 30 s. Reactions were performed in technical duplicates and biological triplicates with a Mastercycler^® ep realplex2 real-time PCR system (Eppendorf, Hamburg, Germany) as described by the manufacturer using their universal program. Reactions with no addition of reverse transcriptase served as negative controls and proved lack of DNA contamination. Specificity of amplification was assessed by analyzing the melting curve of the amplification product. Due to very high sequence identity between lscB and lscC, it was not possible to design primers discriminating between these two mRNAs, thereby the expression profile of lsc is always referred to as a combination of both genes.

Vector NTI Advance 10.1.1 (Invitrogen Corporation, USA) was used for the nucleotide and amino acid sequence alignments and for dendrogram generation. BLAST-N and BLAST-P programs were used for online sequence analyses and for identifying transposase-like sequences and mobile promoter elements [41]. The website www.pseudomonas.com was consulted for the determination of P. syringae gene orthologs and paralogs.