Identification and characterization of three novel EsaI/EsaR quorum-sensing controlled stewartan exopolysaccharide biosynthetic genes in Pantoea stewartii ssp. stewartii

Authors

  • Aurélien Carlier,

    1. Institut für Pflanzenbiologie, Universität Zürich, Zürich, Switzerland.
    Search for more papers by this author
  • Lindsey Burbank,

    1. Department of Microbiology and Plant Pathology, University of California Riverside, Riverside, CA, USA.
    Search for more papers by this author
  • Susanne B. von Bodman

    Corresponding author
    1. Departments of Plant Science and Molecular and Cell Biology, University of Connecticut, Storrs, CT 06269-4163, USA.
      E-mail susanne.vonbodman@uconn.edu; Tel. (+1) 860 486 4408; Fax (+1) 860 486 0534.
    Search for more papers by this author

E-mail susanne.vonbodman@uconn.edu; Tel. (+1) 860 486 4408; Fax (+1) 860 486 0534.

Introduction

Pantoea stewartii ssp. stewartii (P. stewartii), formerly Erwinia stewartii, is a Gram-negative bacterium that causes Stewart's wilt disease in susceptible maize cultivars by colonizing the xylem as cell wall-adherent biofilms (Braun, 1982; Koutsoudis et al., 2006). This mode of growth requires the production of large amounts of stewartan exopolysaccharide (EPS), which impedes the flow of xylem sap, leading to plant wilt. Stewartan EPS is a high-molecular-weight heteropolysaccharide and represents the primary virulence factor in P. stewartii (Bradshaw-Rouse et al., 1981; Jumel et al., 1997). Mutants unable to secrete EPS adhere strongly to surfaces, tend to generate compact, amorphous biofilms and fail to spread beyond the site of infection in the xylem vessels (Koutsoudis et al., 2006).

Stewartan EPS is an anionic polymer composed largely of heptasaccharide repeat units that contain galactose, glucose and glucuronic acid in a 3:3:1 ratio (Fig. 1) (Nimtz et al., 1996a; Yang et al., 1996). Its chemical structure is related to that of amylovoran, a polysaccharide and virulence determinant in Erwinia amylovora (Nimtz et al., 1996b). E. amylovora is also a xylem-dwelling pathogen and causes Fireblight disease in rosaceous plants (Geider, 2000). Sequence homology and partial genetic and biochemical verification indicate that both polymer repeat units are assembled on a polyisoprenoid lipid carrier and translocated via the Wzx membrane-associated polysaccharide specific transport (PST) protein across the inner membrane. A predicted inner membrane-localized Wzy polymerase is thought to facilitate oligomerization of the repeat units, suggesting that the mechanism of stewartan and amylovoran synthesis is related to colanic acid synthesis in Escherichia coli (Coplin et al., 1996; Reeves et al., 1996; Geider, 2000; Whitfield, 2006). All three biosynthetic pathways are under the control of the Rcs phosphorelay signal transduction system (Torres-Cabassa et al., 1987).

Figure 1.

Structure of the stewartan, amylovoran and glucoamylovoran repeating units. The stewartan and amylovoran repeating units consist of Gal, d-galactopyranose residue; Glc, d-glucopyranose; and GlcA, d-glucopyranuronic acid; Pyr, Pyruvyl (Nimtz et al., 1996a,b; and our data). Text in parentheses indicate the anomeric configuration of the residues and their respective linkages. The subscript percentage values indicate the fraction of β(1,6) linked Glc (VI) residues at the branching Gal (I) (90% for stewartan and 70% for amylovoran and gluco-amylovoran). Grey boxes highlight the arbitrary designation of the residues used in Table 2 and in the text following the abbreviation for the polysaccharide; i.e. S-VI refers to stewartan residue VI. Gluco-amylovoran EPS is produced by Eam Ea273 expressing the P. stewartii wce-II gene cluster from plasmid pAUC45. Residue GA-VII is typical of stewartan and a key feature of the heteropolymer. All three exopolymers contain roughly 1000 subunits (our data and Nimtz et al., 1996a,b). The Gal A-V residues of amylovoran can be modified with 2-, 3- or 2,3-linked O-acetyl groups (Ac) (Nimtz et al., 1996a). Whether the GA-V residue of glucoamylovoran is also modified with O-acetyl groups is not known.

In P. stewartii, the biosynthesis of stewartan EPS is cell density-dependent governed by the EsaI/EsaR cell–cell signalling or quorum-sensing (QS) system (von Bodman and Farrand, 1995; von Bodman et al., 1998). The EsaR regulator dimerizes and binds target DNA in the absence of inducing levels of the acyl-homoserine lactone (AHL) signal, and loses DNA binding affinity in its presence (Minogue et al., 2002). EsaR inhibits EPS synthesis by repressing the expression of rcsA at low cell density (Carlier and von Bodman, 2006). The rcsA gene encodes an important regulatory component of the Rcs environmental signal sensing phosphorelay system (Majdalani and Gottesman, 2005). At high cell density, inducing levels of AHL, neutralize EsaR DNA binding at the rcsA promoter allowing expression of RcsA to levels required for formation of the RcsA/RcsB activation complex. This complex is necessary for the stimulated expression of the stewartan biosynthetic cps gene cluster. We have adopted the gene designation wce in place of cps following the suggested nomenclature for various bacterial polysaccharide biosynthetic genes (Reeves et al., 1996).

The primary gene cluster for stewartan EPS synthesis, now termed wce-I, is structurally and functionally related to the ams gene cluster of E. amylovora (Coplin et al., 1996; Geider, 2000). For example, it is possible to complement specific wce-I mutants with corresponding genes of the ams gene cluster and vice versa (Bernhard et al., 1996).

In our attempt to more fully understand the regulated functions involved in stewartan EPS synthesis, we realized that additional functions, not contained within the primary wce-I gene system, must exist elsewhere in the P. stewartii genome. First, it seemed unlikely that biosynthesis of the hexasaccharidic subunits of amylovoran and the heptasaccharidic subunits of stewartan involve the same number of glycosyl-transferases (GTs). Second, ams mutants complemented with P. stewartii wce-I genes produced EPS lacking some characteristic features of stewartan, suggesting that the wce-I gene system alone is insufficient for complete stewartan synthesis (Bernhard et al., 1996).

In this study, we identified and characterized the previously unrecognized bicistronic wce-II and monocistronic wce-III loci, thereby extending the inventory of stewartan EPS biosynthetic genes in P. stewartii. Each of these gene systems encodes a putative GT with specific functional roles in the priming or completion of the stewartan repeat units. Interestingly, the wce-II and wce-III gene systems are also controlled by EsaI/EsaR QS and the Rcs signal transduction system. These findings underscore the significance of the co-ordinated regulation of stewartan synthesis in response to cell density and environmental cues.

Results

Identification of two novel stewartan EPS biosynthetic gene systems

Discrepancies in the number of genes involved in stewartan synthesis encoded by the wce-I gene cluster (formerly cps) (Fig. 2) and the number of proposed glycosidic linkages in the backbone of stewartan (Fig. 1) prompted us to scan the P. stewartii genome for putative additional GT genes with a potential role in EPS synthesis. We were interested in P. stewartii gene sequences annotated in the ASAP annotation database (Glasner et al., 2003) as GT with Gene Ontology numbers GO:0008194 (UDP-glycosyltransferase activity), GO:0016757 (transferase activity, transferring glycosyl groups) and GO:0016758 (transferase activity, transferring hexosyl groups). Two potential loci emerged that were not evidently linked to lipopolysaccharide, peptidoglycan or enterobacterial common antigen synthesis. One of these loci, designated wce-II, comprised two open reading frames, which we named wceO and wzx2. The predicted product of wceO (ASAP: ACV-0283221) is a 340-amino-acid protein with a predicted molecular weight of 39 kDa. It is a putative GT of the CAZy (carbohydrate active enzymes) GT2 family (Campbell et al., 1997; Coutinho et al., 2003). The protein sequence is 21% identical and 31% similar to the inverting GT WbbE of Salmonella enterica serovar Borreze (GenBank: AAC98401.1), which is an experimentally confirmed N-acetylmannosamine transferase enzyme (Keenleyside et al., 2001). WceO is 24% identical and 35% similar to the nucleotide-diphospho-sugar transferase SpsA (GenBank: CAB15817.1) that is involved in Bacillus subtilis spore coat formation. The crystal structure of this enzyme has been published (Charnock and Davies, 1999). Alignment and domain structural analyses of WceO, WbbE and SpsA identify key signatures found in this class of non-processive GT2 proteins (Keenleyside et al., 2001). The greatest degree of conservation is in the N-terminal Domain A that features invariant residues throughout; but in particular two highly conserved aspartate (D) residues (motifs 1 and 2 shown in Fig. S1) that are predicted to play a role in activated donor substrate (Mn2+-UDP-galactose) binding and catalysis (Charnock and Davies, 1999; Keenleyside et al., 2001; Fulton et al., 2008). The C-terminal acceptor-binding Domain B exhibits lower overall conservation, but features an ED(H) motif that aligns with the conserved WbbE ED(Y) and SpsA TDD motifs (Fig. S1). Recent structural analysis of the MAP2569c GT2 of Mycobacterium shows that the catalytic Domain B accommodates the acceptor substrate hydrogen bonded to conserved threonine and arginine residues within this region. This structure also shows the spatial proximity of the donor and acceptor substrates (Fulton et al., 2008).

Figure 2.

Genetic organization of the stewartan biosynthetic genes. The wce-I gene cluster comprises 12 genes from wceG to wzx and is linked to the galFE locus. The wce-II locus encodes the wceO and wzx2 genes. The third locus involved in stewartan biosynthesis is the monogenic wce-III locus. Putative GT genes required for stewartan subunit biosynthesis are depicted in light grey. Putative genes specifying stewartan polymerization and export functions are shown in black. Previous experimental work and sequence analysis presented here suggest that synthesis of stewartan subunits begins with the addition of a galactose residue to a lipid carrier by the products of the wceG1 (Geider, 2000) and wceG2 (this study) genes. Subsequent addition of sugar residues presumably involves genes wceB, wceM, wceN, wceK and wceO, resulting in a heptameric repeat unit (galactose : glucose : glucuronic acid 3:3:1) (Nimtz, 1996a; Geider, 2000). Flippase or PTS functions encoded by wzx1 and wzx2 are predicted to transfer the lipid-linked repeat units to the periplasm. The gene product of wceL was originally described as a GT, but secondary structure analysis indicates that this gene is likely to encode a stewartan-specific Wzy EPS polymerase (S.B. von Bodman, unpub. information) (Whitfield, 2006). Finally, the predicted products of wza, wzb and wzc show high sequence homology to various EPS export machineries (Geider, 2000; Whitfield, 2006). The precise function of the wceF and wceJ genes in stewartan EPS synthesis remains to be clarified.

The wzx2 gene (ASAP: ACV-0283220) encodes a putative PST protein (flippase) involved in polysaccharide repeat translocation across the bacterial inner membrane. The predicted Wzx2 protein shares relatively weak amino acid sequence homology with RbfX of E. coli K12 (GenBank: NP416541) (26% sequence identity and 46% similarity) and only 18% identity and 28% similarity with the predicted amino acid sequence of the P. stewartii wce-I-encoded wzx1 gene. However, like most PST proteins, the wce-I encoded Wzx1 and the predicted Wzx2 product possess 10 transmembrane domains, six of which are located within a loosely conserved Pfam Polysacch_synt domain (Marolda et al., 2004) (http://pfam.sanger.ac.uk/family?acc=PF01943) (Fig. S2). The third stewartan biosynthetic locus, called wce-III, contains one gene, which we designated wceG2 (ASAP: ACV-0283377). The predicted WceG2 protein is 59% identical to WceG1, which is encoded by the first gene of the wce-I operon. WceG2 also shares 62% amino sequence identity with the well-characterized WbaP undecaprenyl-phosphate UDP-galactose phosphotransferase of S. enterica (GenBank: AAC27321 and CAA40130). This protein transfers galactose phosphate residue to the undecaprenyl phosphate lipid carrier in the first committed step of EPS and O-antigen synthesis (http://pfam.sanger.ac.uk/family?acc=PF02397) (Liu et al., 1993; Saldías et al., 2008). Significantly, the predicted protein products of WceG1, WceG2 and WbaP display a highly conserved overall membrane topology with five predicted transmembrane domains as detailed by Saldías et al. (2008).

The wce-II- and wce-III-encoded functions are required for stewartan EPS synthesis

To verify a role for wce-II and wce-III in stewartan polysaccharide synthesis, we created non-polar deletion mutants and evaluated their impact on stewartan production using a quantitative stewartan-specific I-ELISA immunodetection assay. The wce-II mutant strains, Pnss22 (wceO::GmR) and Pnss23 (wzx2::GmR.) were grown separately in stewartan-inducing medium to mid-exponential phase. As summarized in Table 1, strains Pnss22 (wceO::GmR) and Pnss23 (wzx2::GmR) were completely abolished for stewartan EPS synthesis demonstrating the absolute requirement of WceO and Wzx2 in stewartan production. Mutant complementation with the respective genes expressed in trans from plasmids pAUC46 and pAUC47 restored stewartan production to wild type levels (Table 1).

Table 1.  Stewartan production and virulence of P. stewartii strains.
StrainStewartan productiona ± 95% CIVirulenceb
  • a. 

    Total quantity of stewartan, free and cell-bound, in pg/cfu ± 95% confidence interval (CI).

  • b. 

    Symptoms on corn seedlings 10 days after inoculation (0, no symptoms; 5, death; nd, not determined.

DC2834.8 ± 0.84
DC283 (pAUC45)4.6 ± 1.0nd
Pnss21 (wceG2)3.6 ± 0.64
Pnss21 (pAUC45)5.0 ± 1.1nd
Pnss22 (wceO)< 0.010
Pnss23 (wzx2)< 0.010
Pnss22 (pAUC46)4.3 ± 0.7nd
Pnss23 (pAUC47)4.9 ± 0.9nd

In contrast, the wce-III mutant strain Pnss21 (wceG2::GmR) produced approximately 25% less stewartan compared with the wild-type strain DC283, indicating that WceG2 contributes to stewartan synthesis, but is not essential. Redundancy of the wceG1 and wceG2 undecaprenyl-phosphate UDP-galactose phosphotransferase genes was further demonstrated by expressing wceG1 (wce-I) from a lac promoter on plasmid pAUC45 in the wceG2 (wce-III) deletion mutant Pnss21, which restored wild-type levels of authentic stewartan EPS production (Table 1). These data indicate that wceG1 and wceG2 have identical functions both of which contribute to maximal stewartan EPS synthesis if present in multiple copies, while single-copy-mutant strains exhibit reduced EPS.

The wce-II operon encodes a stewartan 1,6-glucosyltransferase activity

Direct biochemical characterization of bacterial GT enzymatic activities is difficult due to stringent substrate specificity and limited availability of the lipid-linked intermediates. For this reason, we used a combined genetic and biochemical approach to define the function of WceO by exploiting the structural and functional similarities of the P. stewartii stewartan and E. amylovora amylovoran biosynthetic pathways. We knew from previous studies that the expression of the ams gene cluster in P. stewartii wce-I structural mutants leads to the production of a polysaccharide with a characteristic amylovoran backbone, but lacking the distinctive pyruvate and acetate adducts of amylovoran (Bernhard et al., 1996). Most significantly, the heterologous polysaccharide featured a terminal glucose residue typical of stewartan in place of the pyruvyl group on the side-chain galactose residue (A-V) of amylovoran (Fig. 1) (Bernhard et al., 1996). This distinctive feature indicated the presence of a dedicated GT for the addition of terminal glucose residues to the polymer repeat units that must be located outside the primary (wce-I) EPS biosynthetic locus. The genes located within the wce-II gene cluster were the most likely candidates to fulfil this functional role. To verify this prediction, we expressed wceO and wzx2 carried on plasmid pAUC44-Cm in E. amylovora Ea273, which normally synthesizes genuine amylovoran EPS (Fig. 1) (Nimtz et al., 1996b). The neutral monosaccharide composition of the two forms of EPS isolated from culture supernatants of E. amylovora Ea273 with and without pAUC44-Cm were analysed by trifluoroacetic acid (TFA) hydrolysis and alditol-acetate derivative detection by gas chromatography-electron impact mass spectrometry (GC-MS). This analysis showed that the EPS produced by E. amylovora Ea273 (pAUC44-Cm) contains substantially more glucose with a glucose : galactose molar ratio of 30:70 instead of the 15:85 ratio of native amylovoran (Table 2). This neutral sugar content corresponds to 1.7 glucose and 4 galactose residues per polymer repeat unit of the gluco-amylovoran heteropolymer versus 0.7 glucose and 4 galactose residues per repeat unit of native amylovoran. As reported by Nimtz et al. (1996b), amylovoran contains a β(1,6)-linked glucose residue at the branching galactose (A-I) in about 70% of the repeat units (Fig. 1). We infer that the additional glucose residue in gluco-amylovoran is linked to the terminal side-chain galactose (GA-V) as in stewartan (S-V). Confirmation of this prediction comes from the methylation analysis of the different purified EPS samples as summarized in Table 2. Specifically, gluco-amylovoran shows a sharp decline of 4,6-linked galactose characteristic of the 4,6 pyruvate-ketal substitution of the amylovoran side-chain galactose (A-V; Fig. 1) and a concomitant increase in 6-substituted galactose residues similar to stewartan. Interestingly, the ratio of 3,4,6-linked (A-I, GA-I) to 3,4-linked galactose residues is unchanged in gluco-amylovoran, indicating that the branching galactose residue in the gluco-amylovoran backbone remains glucose-substituted in 70% of the subunits similar to native amylovoran. These data suggest that the wce-II operon encodes a β(1,6) glucosyltransferase that is responsible for the addition of terminal glucose residues to both stewartan and gluco-amylovoran repeat units.

Table 2.  Compositional and carbohydrate linkage analysis of preparations of stewartan, amylovoran and gluco-amylovoran.
Peracylated derivative ofaDerivative of residuebPosition of substitutionStewartanb (S)Amylovoranb (A)Gluco-amylovoran (GA)b
  • a. 

    The most significant signals for distinction of wild-type and WceO-modified amylovoran are marked*.

  • b. 

    See Fig. 1 for corresponding sugar in EPS structure.

  • c. 

    Proportion of the total alditol acetate derivatives (AA) after hydrolysis of the polymers.

Galactitol percentage of Totalc AA  518570
 2,4,6-Tri-O-methyl-S-III, A-III, GA-III3++++++
 *2,3,4-Tri-O-methyl-S-V, A-II, GA-II, GA-V6+++++++
 2,6-Di-O-methyl-S-I, A-I, GA-I3; 4+++++
 *2,3-Di-O-methyl-A-V4; 6+
 2-Mono-O-methyl-S-I, A-I, GA-I3; 4; 6++++
Glucitol percentage of Totalc AA  491530
 *2,3,4,6-Tetra-O methyl-S-VI, S-VII, A-VI, GA-VI, GA-VII++++++
 2,3,4-Tri-O-methylS-II6++

The functions encoded by the wce-II gene system are required for virulence on corn seedlings

Stewartan EPS is the primary factor of virulence in the P. stewartii-induced wilt of corn (Bradshaw-Rouse et al., 1981; Dolph et al., 1988; Coplin and Majerczak, 1990). We therefore tested whether mutations in wce-II and wce-III impair or block the infection process based on corn seedling virulence assays. As summarized in Table 1, seedlings inoculated with strains Pnss22 (wceO::GmR) and Pnss23 (wzx2::GmR) were largely asymptomatic as expected. In contrast, the diminished EPS production exhibited by strain Pnss21 (wceG2::GmR) had no noticeable effect on the degree and rate of symptom development compared with wild-type strain DC283. The latter observation implies that maximal EPS production may not be required for the colonization of the maize host, at least under laboratory infection conditions.

The EsaI/EsaR QS system controls the expression of all three stewartan biosynthetic loci

We previously established that the wce-I operon, which encodes the primary stewartan biosynthetic functions, is regulated by QS via control of rcsA. The transition between stewartan EPS ‘on’ and ‘off’ appears to be biphasic in that prior to the threshold cell density or AHL concentration the bacteria produce little, if any, detectable amounts of EPS (von Bodman et al., 1998). The identification of two additional stewartan EPS biosynthetic loci, wce-II and wce-III, begged the question whether these gene systems were also co-ordinately controlled by the EsaI/EsaR and Rcs regulatory systems. As summarized in Table 3, comparative real-time PCR analyses using total RNA isolated from cultures of P. stewartii strain ESN51 (esaI) and PSS11 (esaI, rcsA) grown under AHL inducing and non-inducing conditions showed that all genes tested, namely wceG, wza, wceB, wzx (wce-I); wceO and wzx2 (wce-II) and wceG2 (wce-III) were significantly upregulated in response to inducing AHL concentrations in the signal synthase-deficient mutant P. stewartii strain ESN51 (esaI). However, AHL responsiveness of all these genes was abolished in the esaI, rcsA double mutant strain PSS11 (Table 3; Minogue et al., 2005). Taken together, these data show that stewartan EPS production requires the co-ordinated, cell density/RcsA-dependent upregulation of the apparent tri-partite stewartan biosynthetic gene system.

Table 3.  Quantification of QS-regulation of stewartan biosynthetic genes by real-time RT-PCR.
Gene system
Genea
wce-I wce-II wce-III
wceG2
wceG1 wza wceB wzx1 wceO wzx2
  • a. 

    Primer set used in real-time RT-PCR experiments (Table S1, e.g. wceG = primers are listed as wceGRT5 and wceGRT3).

  • b. 

    Fold change expression in P. stewartii ESN51 cultured in the presence over absence of AHL. Fold change was calculated using galE as an internal standard. Error values represent the 95% confidence interval (CI).

  • c. 

    Fold change expression in P. stewartii PSS11 cultured in the presence over absence of AHL.

Fold changeb ± 95% CI6.63 ± 0.504.46 ± 0.704.86 ± 0.554.44 ± 1.105.03 ± 0.852.64 ± 0.258.43 ± 0.85
Fold change in absence of rcsAc ± 95% CI1.21 ± 0.350.83 ± 0.500.98 ± 0.451.05 ± 0.551.31 ± 0.601.05 ± 0.501.47 ± 0.55

Discussion

Stewartan EPS is an essential virulence factor of P. stewartii (Bradshaw-Rouse et al., 1981; Coplin et al., 1986). Our present study establishes a critical role of the two previously unidentified wce-II and wce-III gene systems in the biosynthesis of stewartan EPS. Although genetically unlinked, all three loci are governed by the EsaI/EsaR QS system via control of the essential RcsA cotranscription factor, which together with RcsB forms the transcriptional activation complex required for wce expression (Torres-Cabassa et al., 1987; Minogue et al., 2005). These observations are consistent with our predictions that premature stewartan synthesis disrupts the early microbial developmental processes including the ability to adhere to surfaces and form normal biofilms both in vitro and in the xylem of the plant host (Koutsoudis et al., 2006).

The wce-II gene system is essential for stewartan synthesis as demonstrated by a highly sensitive immunodetection assay. Specifically, mutants of wceO and/or wzx2 abolish both cell-bound and cell-free stewartan synthesis completely. These strains are also avirulent. Expression of wce-II in the E. amylovora strain Ea273 results in the production of a heteromeric gluco-amylovoran EPS polymer (GA) that exhibits key structural similarities to both native stewartan and amylovoran EPS. Specifically, linkage analysis shows that GA features a terminal Glc (GA-VII) residue 1,6-linked to the GA-V Gal residue in the side-chain of all repeat units of the polymer, with no evidence of the characteristic pyruvate substitution of native amylovoran. Gluco-amylovoran therefore assumes a distinctive stewartan characteristic. We predict that the GA-VII Glc residue is in the same β anomeric configuration as stewartan because GTs of the Cazy GT2 family tend to follow an inverting catalytic mechanism, resulting in the addition of a glycosyl residue from UDP-glucose to the growing polymer in the β configuration (Carbohydrate Active Enzymes Database, http://www.cazy.org/) (Coutinho et al., 2003).

Interestingly, the heteromeric gluco-amylovoran synthesized by the E. amylovora strain Ea273 expressing the P. stewartii wce-II (wceO, wzx2) locus retains the partial 70% β(1,6)-linked glucose substitution (GA-VI) at the branching galactose (GA-I) typical of native amylovoran. This observation suggests that the stewartan WceO glucosyl-transferase does not contribute to the transfer of additional glucose residues to the branching galactose (GA-I) in the E. amylovora background. The simplest explanation for this observation is that P. stewartii WceO may lack specificity for the branching galactose residue of amylovoran, which is (β1–3) linked to a preceding galactose residue, while stewartan is (β1–3) linked to a glucose residue (Fig. 1). In this connection, it is important to point out that the E. amylovora genome sequence carries a homologous wceO/wzx2 gene system, whose functionality remains undefined, but may be specific for amylovoran repeat unit synthesis.

The second gene of the wce-II locus, wzx2, is a wzx PST gene homologue, also referred to as a ‘flippase’. PST proteins facilitate the transport of undecaprenyl-phosphate lipid-linked glycans across the cytoplasmic membrane (Liu et al., 1996; Feldman et al., 1999). Sequences of PST proteins are highly divergent, but seem to group together loosely according to specificity for O-antigen or EPS transport (Paulsen et al., 1997). Regardless of the primary structural divergence, these proteins have highly related helical membrane-spanning domain characteristics as shown in the Fig. S2. It is generally assumed that PST proteins involved in O-antigen biosynthetic systems display relaxed specificity requirements for the carbohydrate composition and structure of the O-antigen polysaccharide (Feldman et al., 1999; Marolda et al., 2004). Marolda et al. (2004) suggested that the specificity of the PST proteins does not extend beyond the first sugar residue linked to the undecaprenyl phosphate lipid carrier. Our data, however, suggest a high degree of specificity based on the observation that the wzx2 null mutant (strain Pnss23) is unable to produce measurable amounts of stewartan EPS even in the presence of a functional wzx1 gene encoded by the wce-I operon. In contrast, a wzx1 mutant strain exhibits a somewhat reduced mucoid phenotype (D.L. Coplin, unpub. obs.), which suggests a primary, but not exclusive role for Wzx2-mediated translocation of lipid-linked stewartan repeat units. It is possible that Wzx2 is specific for complete heptameric repeating units translocation, while Wzx1 may have greater or exclusive specificity for the hexameric repeating units that lack a terminal glucose at the branching galactose. These plausible scenarios could explain the observed reduction in overall polymer synthesis by the wzx1 mutant strains. Work in progress addresses this hypothesis as part of a larger question to explore a potential biological role for the heteromeric stewartan polymer composed of 90% heptasaccharidic and 10% hexasaccharidic repeating units.

The WceG1/WceG2 redundant functions have a high degree of homology and transmembrane helix topology to WbaP in S. enterica (Saldías et al., 2008). The enzymes are therefore likely to catalyse the transfer of galactosyl-phosphate to the undecaprenyl-phosphate lipid carrier from UDP-galactose as a first step in the synthesis of stewartan repeat units (Liu et al., 1993; Reuber and Walker, 1993; Geider, 2000). The synthesis of both the hepta- and hexasaccharidic repeat units initiate with a lipid-linked galactose, which suggests that WceG1 and WceG2 are functionally equivalent. This functional redundancy explains the leaky phenotype of a transposon-insertion mutation in the wceG1 gene (Dolph et al., 1988; Reeves et al., 1996). The high sequence homology between the predicted wceG1 and wceG2 genes and comparable GC content strongly points to a gene duplication event.

What are the potential benefits from such a redundancy? We showed that wceG1 expressed in the Pnss21 wceG2 mutant background restored this strain to full EPS synthesis (Table 1). Thus, gene duplication might enable high-level synthesis of galactosyl-undecaprenyl-phosphate at the onset of EPS synthesis without calling for excessive upregulation of the wce-I operon, which could have deleterious consequences. However, suboptimal EPS synthesis does not apparently diminish the colonization and/or pathogenic potential of the wceG2 mutant (Pnss21) bacteria (Table 1). However, here it is important to emphasize that the laboratory virulence assays are intrinsically artificial, and most certainly do not reflect natural infections mediated by the corn flea beetle vector of Stewart's wilt. Perhaps reduced stewartan production engenders a subtle fitness cost under natural infection conditions, but which are incidental during manual infections with large numbers of cells.

Experimental procedures

Bacterial strains and plasmids

Strains and plasmids are summarized in Table 4. The E. coli strains used as cloning hosts include DH5α (Life Technologies), Top10 (Invitrogen), DH10B (Invitrogen) and S17-1 (Simon et al., 1982) for conjugal transfer of RK2-based plasmid constructs into P. stewartii and E. amylovora strains. E. coli strains were grown at 37°C on nutrient agar (NA) plates or Luria–Bertani broth (LB) in presence of appropriate antibiotics, where applicable. The P. stewartii strains were grown at 28°C in LB in presence of 30 μg ml−1 of nalidixic acid on NA plates, AB minimal medium supplemented with 0.2% glucose (Clark and Maaloe, 1967) or LB. E. amylovora strains were grown at 28°C in LB or in AB minimal medium supplemented with 0.2% (w/v) glucose and 0.1% (w/v) yeast extract (Difco). DNA extractions and manipulations were performed by standard methods as previously described (von Bodman et al., 1995; 1998). DNA amplification was performed using Ex Taq Polymerase (Takara Bio) and synthetic oligonucleotides ordered to specification from Qiagen Operon (Table S1).

Table 4.  Strains and plasmids.
Strain/plasmid designationRelevant genotypeReference or source
  1. Ap R , ampicillin; CmR, chloramphenicol; KmR, kanamycin; NalR, nalidixic acid; resistance.

Strains  
 DH10B E. coliΔlacX74 araΔ139Δ(ara-leu)Invitrogen
 DB3.1 E. coli FgyrA462 rpsL20(SmR) λInvitrogen
 DB3.1 λpirλpir lysogen of DB3.1 House et al. (2004)
 DC283 P. stewartii wild type, NalR Dolph et al. (1988)
 Pnss21 P. stewartii wceG2::Gm, NalR, GmRThis study
 Pnss22 P. stewartii wceO::Gm, NalR, GmRThis study
 Pnss23 P. stewartii wzx2::Gm, NalR, GmRThis study
 Ea273 E. amylovora wild type Momol et al. (1997)
 Gal8 P. stewartiiΔ(wceG-galE), NalR Bradshaw-Rouse et al. (1981)
 S17-1RP4 Mob+ Simon et al. (1982)
Plasmids  
 pDONR221Cloning vector, KmRInvitrogen
 pBBR1MCSBroad host range vector, CmR Kovach et al. (1994)
 spBBR1MCS-4Broad host range vector, ApR Kovach et al. (1995)
 pBBR1MCS-4GWpBBR1MCS-4 modified as a Gateway destination vectorThis study
 pPS856source of gentamicin-resistance cassette, GmR Choi et al. (2005)
 pKNG101suicide vector, SmR Kaniga et al. (1991)
 pES2144 cps gene cluster cloned in pVK100 Coplin et al. (1986)
 pAUC40suicide vector, Gateway attR-CmR cassette cloned in pKNG101, SmR, CmRThis study
 pAUC41 wceG2::GmR cloned in pAUC40This study
 pAUC42 wceO::GmR cloned in pAUC40This study
 pAUC43 wzx2::GmR cloned in pAUC40This study
 pAUC44-Ap wceO-wzx2 cloned in pBBR1MCS-4GW, ApRThis study
 pAUC44-Cm wceO-wzx2 from pAUC44-Ap cloned BamHI/SstI in pBBR1MCS-4GW, CmRThis study
 pAUC45 wceG, wza, wzb and wzc from pES2144 cloned in pBBR1MCS-4, ApRThis study
 pAUC46 wceO cloned in pBBR1MCS-4GWThis study
 pAUC47 wzx2 cloned in pBBR1MCS-4GWThis study

Sequence analysis

The P. stewartii genome sequence data were generated by the Baylor College of Medicine Human Genome Sequencing Center website at http://www.hgsc.bcm.tmc.edu. Preliminary genome annotation is available trough the ASAP database at http://asap.ahabs.wisc.edu/asap/home.php (accessed February 2008). Pairwise sequence alignments were performed using the EMBOSS Needle algorithm (http://www.ebi.ac.uk/Tools/emboss/align/). The EMBL-EBI ClustalW2 program (http://www.ebi.ac.uk/Tools/clustalw2/) was used to perform multiple-sequence alignments. Membrane topology prediction employed the TMHMM algorithm (http://www.cbs.dtu.dk/services/TMHMM/, accessed August 2009) and conserved domain predictions used the Pfam algorithm (http://pfam.sanger.ac.uk/, accessed August 2009).

Deletion mutagenesis and allelic replacement

The destination vector pAUC40 was created by inserting the Invitrogen Gateway attR-CmR cassette from the commercial plasmid conversion kit into the backbone of BamHI/SmaI pairwise-digested plasmid pKNG101 (Kaniga et al., 1991). Plasmid pAUC40 was maintained in E. coli strain DB3.1 λpir (House et al., 2004). Mutant strains Pnss21, Pnss22 and Pnss23 were created using a modified version of the Gateway method described by Choi and Schweizer (2005). Specifically, the flanking regions of the wceG2, wceO and wzx2 genes were amplified by PCR using oligonucleotide primers listed in Table S1. A gentamicin-resistance cassette derived from plasmid pPS856 (Choi et al., 2005) was inserted between the flanking regions of the gene of interest using a PCR overlap technique with primer sets listed in Table S1. The resulting PCR products containing the Gm-resistance cassette flanked by gene-specific DNA was cloned into the Gateway Entry vector pDONR221 using the BP clonase II kit (Invitrogen). The constructs were genetically transferred into the suicide vector pAUC40 using the LR clonase kit II (Invitrogen), resulting in plasmids pAUC41, pAUC42 and pAUC43 respectively. The plasmids were introduced into E. coli strain S17-1 and conjugally transferred into the wild-type strain Pnss DC283. Allelic replacement events were selected based on dual resistance to gentamicin and sacBR-mediated sucrose sensitivity. Allelic replacement events were verified by PCR. For complementation analysis, DNA regions containing wceO, wzx or both were amplified by PCR using primers listed in Table S1. The PCR products were introduced into the vector pBBR1MCS-4GW using Invitrogen Gateway technology to give rise to plasmids pAUC47, pAUC46 and pAUC44-Ap. The plasmids were introduced into E. coli S17-1 and transferred into P. stewartii.

RNA extraction and real-time RT-PCR

RNA was isolated from P. stewartii strain ESN51 cultures grown to mid-exponential phase in AB medium supplemented with or without 10 μM AHL as previously described (Carlier et al., 2006). First strand cDNA was generated using random hexamers and reagents supplied by the Superscript III kit (Invitrogen). Real-time RT-PCR using a LightCycler 2.0 instrument (Roche) and the LightCycler SYBR green master plus kit allowed detection and relative quantification of cDNAs corresponding to mRNAs expressed from target genes under different conditions. Primers for target genes were designed using the Roche Probe Design Software and are listed in Table S1. Results were analysed using the LightCycler software version 4.0. Cp values reflecting the amount of target cDNA in a sample quantified by extrapolation from a standard curve. The galE gene served as an internal standard for each cDNA sample. This gene is not regulated by QS or the Rcs regulatory system (Torres-Cabassa et al., 1987; Minogue et al., 2005). The ratio of target gene mRNA to galE mRNA was established for all samples tested and used to compare the relative expression of the target genes under QS inducing and non-inducing conditions.

Stewartan purification and antibody production

Stewartan EPS was purified from 1 l cultures of P. stewartii strain DC283 grown in AB medium to stationary phase. EPS was precipitated from cultures supernatants with 3 vols of ethanol and collected by centrifugation at 5000 g for 20 min in a Beckman J2HS centrifuge. Remaining cells and debris were removed by centrifugation at 17 000 g for 2 h. Precipitated material was solubilized in 20 ml PBS and treated with 1 mg Proteinase K (Invitrogen) at 65°C for 1 h. A total of 2 mg of carbohydrates as estimated by the phenol-sulphuric acid method (Masuko et al., 2005) was separated by size-exclusion chromatography on a 100 cm × 1.5 cm Sephacryl S500 H column (GE Biosciences). Void volume carbohydrate fractions were pooled, dialysed extensively against water and lyophilized. Preparations of stewartan EPS were sent to Covance Custom Immunology Services (Denver, PA) for polyclonal antibody production in New Zealand White rabbits. Stewartan specific antibodies were enriched by absorbing stewartan antisera to the stewartan-deficient mutant P. stewartii strain Gal8 to remove non-specific antibodies.

Quantification of stewartan by I-ELISA

Stewartan was quantified in solution using an Inhibition Enzyme-Linked Immunosorbent Assay (Inzana and Champion, 2007). Biotin-LC-hydrazide (Pierce Chemical, Rockford, IL) was coupled directly to the carboxyl groups of glucuronic acid of purified stewartan with stoichiometric amounts of ethyl-dimethylaminopropyl-carbodiimide HCl (EDC, Pierce Pierce Chemical, Rockford, IL) according to manufacturer's instructions. The resulting biotin-EPS was dialysed against deionized water. Biotin-EPS was bound to streptavidin-coated microplates (Thermo-Fisher Scientific, Vantaa, Finland) by incubating biotin-EPS (10 μg ml−1) dissolved in phosphate-buffered saline (PBS) containing 20 mM MgCl2 for 1 h at 37°C. Non-specific binding sites were blocked with Blocking Buffer (BB, PBS 1% casein, 0.05% Tween 20, pH 7.4) for 1 h at 37°C. Purified EPS extracts or crude culture samples were incubated with a 1:1000 dilution of rabbit anti-stewartan serum in BB overnight at 4°C. One hundred microliters of preincubated, inhibited serum samples was added to a streptavidin-coated microplate in triplicates and incubated at 37°C for 1 h. The plates were washed with PBS 0.05% Tween 20. Addition of a ×5000 dilution of goat anti-rabbit immunoglobulin G conjugated to horseradish peroxidase in BB was added to the wells, incubated and washed following standard protocol. Stewartan concentrations were quantified based on peroxidase-mediated immunohistochemistry (TMB substrate kit, Pierce) and absorbance readings at 490 nm in a FLUOstar OPTIMA microplate reader (BMG laboratories). Values were fitted on a standard curve prepared with known amounts of purified stewartan to estimate the concentration of stewartan in a sample.

Carbohydrate composition and linkage analysis

Stewartan EPS was isolated from P. stewartii and E. amylovora culture supernatants by precipitation with 1% cetylpyridinium chloride (CPC) (ACROS organics) followed by resuspension in PBS. We found that adding this CPC precipitation step to the stewartan isolation procedure greatly improved recovery of stewartan and facilitated elimination of smaller neutral contaminating glucans from preparations. Excess CPC and other impurities were removed by precipitation with 3 vols of ethanol. Proteins were extracted with hot phenol and the remaining carbohydrates were precipitated with ethanol, dialysed extensively against deionized water and lyophilized. For compositional analysis, EPS samples were first hydrolysed in 2 M TFA at 121°C for 2 h. Compositional analysis was performed by GC-MS separation of alditol-acetate derivatives (Stenutz et al., 2004) on an Agilent 6890 series gas chromatograph fitted with a Supelco SP2330 capillary column (30 m × 0.25 mm) and an Agilent 5973 Network Mass Selective Detector. Elution times and mass spectra of sample compounds were compared with that of authentic standards (Sigma). For linkage analysis dry EPS was accurately weighed and solubilized in DMSO for 24 h at 28°C under constant shaking. Partially methylated alditols acetates were prepared by the method of Anumula and Taylor using sodium hydroxide in DMSO followed by permethylation using methyl iodide (Anumula and Taylor, 1992). Permethylated EPS were hydrolysed in 2 M TFA, reduced with NaBH4 and acetylated. Structural information was inferred from the analysis of the partially methylated alditols acetatess by GC-MS.

Virulence assay on sweet corn seedlings

Sweet corn seedlings (Zea Mays cv. Jubilee, Rogers seeds) were grown in a mixture of 45% peat, processed pine bark, perlite and vermiculite (Fafard 3B mix) in a controlled environment chamber at 28°C, 70% relative humidity, 16 h light and 8 h dark cycle, 355 μE m−2 s−1 light intensity. Plants were inoculated 5 days after germination with 5 μl of bacterial suspension in PBS containing roughly 1 × 104 cells. Stems were wounded twice at right angles 1 cm above the cotyledon. Six plants were inoculated for each strain tested. Symptom severity was rated 10 days post inoculation on the following scale (von Bodman et al., 1998): 0 = no symptoms; 1 = few scattered lesions; 2 = scattered water soaking symptoms; 3 = numerous lesions and slight wilting; 4 = moderately severe wilt; 5 = death.

Acknowledgements

We thank David Coplin for providing mutant strains and insightful discussions; Bruce Link and Wolf-Dieter Reiter for their assistance the EPS biochemical and structural characterization. This research was supported by the National Science Foundation (Grant MCB-0619104), the Agricultural Experiment Station (Grant CONS00775) and the University of Connecticut Research Foundation.

Ancillary