The genetic and structural basis of two distinct terminal side branch residues in stewartan and amylovoran exopolysaccharides and their potential role in host adaptation

Authors

  • Xiaolei Wang,

    1. Plant Science, University of Connecticut, Storrs, Connecticut, USA.
    Search for more papers by this author
  • Fan Yang,

    1. Crop Science, University of Illinois at Urbana-Champaign, Urbana, Illinois, USA.
    Search for more papers by this author
  • Susanne B. von Bodman

    Corresponding author
    1. Plant Science, University of Connecticut, Storrs, Connecticut, USA.
    2. National Science Foundation, Mol. and Cell. Biosciences, 4201 Wilson Blvd., Arlington, VA 22230, USA.
      E-mail svonbodm@nsf.gov; Tel. (+1) 703 292 7139; Fax (+1) 703 292 9061.
    Search for more papers by this author

E-mail svonbodm@nsf.gov; Tel. (+1) 703 292 7139; Fax (+1) 703 292 9061.

Summary

Stewartan and amylovoran exopolysaccharide (EPS) produced by the plant pathogenic bacteria Pantoea stewartii and Erwinia amylovora are virulence factors in the cause of Stewart's vascular wilt and fire blight. The biosynthesis of amylovoran and stewartan is encoded by a set of homologous operons that have been partially characterized, although some annotations are solely on the basis of sequence homology. The major distinguishing features of these two EPS forms are the presence of a terminal pyruvate in amylovoran and glucose in stewartan, even though the gene systems to account for both are conserved and present in each bacterium. This study explores the genetic, structural and functional differences of amylovoran and stewartan, and their potential role in host adaptation. We report that the pyruvyl transferase gene in P. stewartii is non-functional, while the terminal glucosyl transferase is catalytically active. Conversely, in E. amylovora, the homologous glucosyl transferase activity appears to be relatively ineffective, while the pyruvyl transferase function predominates. We also show that the terminally pyruvylated versus glucosylated EPS require specific repeating unit translocases (Wzx). We discuss the evolutionary, functional and biological implications of the terminally pyruvylated and glucosylated polymers and their potential contribution to plant and insect host adaptation.

Introduction

Pantoea stewartii ssp. stewartii (P. stewartii) and Erwinia amylovora (E. amylovora) are Gram-negative, rod-shaped bacteria of the Enterobacteriaceae family, which, respectively, depend on stewartan and amylovoran exopolysaccharide (EPS) synthesis for virulence. P. stewartii is the causal agent of Stewart's wilt disease in maize and other monocots. This bacterium overwinters in certain coleopteran insect hosts, such as the corn flea beetle Chaetocnema pulicaria (Braun, 1982). The association of P. stewartii with maize is well documented (Pataky, 2003; Koutsoudis et al., 2006; Roper, 2011), while the interaction with the overwintering beetle host has only been recently addressed on a cellular and molecular level (Stavrinides et al., 2010). P. stewartii bacteria are introduced into the plant host by the infected beetle vector that emerges from hibernation to feed on the plant tissue thereby creating entry points for the bacteria into the leaf apoplast. Here, the bacteria cause cellular damage and leakage of fluids, which manifest themselves macroscopically as water-soaked lesions (Frederick et al., 2001). The systemic phase of Stewart's vascular wilt occurs when the bacteria colonize the xylem of the host where they form cell wall adherent, EPS-encased biofilms that block the free flow of water leading to the wilting condition (Bradshaw-Rouse et al., 1981; Dolph et al., 1988; Claflin, 2000; Koutsoudis et al., 2006; Roper, 2011).

Erwinia amylovora is also a xylem-dwelling pathogen that causes fire blight disease in rosaceous plants such as apple trees (Geider, 2000). The fire blight disease cycle differs in that the bacteria have only a casual association with insects such as honeybees that carry the infective bacteria from blossom to blossom to establish the primary infection. The bacteria enter the plant host through the blossoms or other natural openings such as stomata, hydathodes, and lenticels. E. amylovora overwinters in the infected canker tissue rather than in an insect host (Geider, 2000).

Amylovoran and stewartan EPS are complex anionic heteropolysaccharides composed of glucose, galactose and glucuronic acid. They are typical of enteric group I polysaccharides, such as colanic acid of Escherichia coli (Whitfield, 2006). Amylovoran and stewartan differ in three major respects (Fig. 1). First, the backbone of amylovoran is composed of three galactose residues, while that of stewartan contains glucose and galactose. Second, stewartan repeating units (RUs) carry a glucose residue that is linked to the branching galactose in about 90% of RUs, while in E. amylovora, the degree of terminal glucosylation at this site ranges from 10% to as much as 70% depending on the growth conditions (Bernhard et al., 1996; Carlier et al., 2009). Third, the side-chain of amylovoran is capped by a ketal pyruvate, while stewartan carries a terminal glucose residue (Yang et al., 1996; Nimtz et al., 1996a,b). Until recently, the synthesis of stewartan and amylovoran EPS were thought to be catalysed exclusively by enzymes encoded in the primary EPS biosynthetic gene clusters, which in P. stewartii is designated wce-I, renamed from cps (Carlier et al., 2009). Carlier and colleagues also described two additional stewartan EPS biosynthetic loci in P. stewartii designated wce-II and wce-III (Carlier et al., 2009). The wce-II locus encodes a second WceG protein, which together with the homologous gene present in wce-I are predicted to initiate RU synthesis by transferring an UDP-galactose onto the undecaprenyl lipid carrier. The wce-III region encodes an essential wceO glucosyl transferase gene and a linked gene, wzx2 that encodes a polysaccharide specific transport (PST) protein, which translocates undecaprenyl-phosphate lipid-linked RUs across the bacterial inner membrane (Whitfield, 2006). Similar gene systems, designated ams-I, ams-II and ams-III exist in E. amylovora (Langlotz et al., 2011 and this study).

Figure 1.

Structure of the stewartan, pyr-stewartan, amylovoran, depyruvylated amylovoran (Ea100 amylovoran) and gluco-amylovoran repeating units. The stewartan and amylovoran repeating units consist of Gal, d-galactopyranose; Glc, d-glucopyranose; and GlcA, d-glucopyranuronic acid; Pyr, Pyruvyl (Nimtz et al., 1996a,b; Carlier et al., 2009). Text in parentheses indicates 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). Designations I to VII are arbitrary residue designations used in Tables 2 and 3, and in the text following the abbreviation for the polysaccharide; i.e. VI refers to residue VI in stewartan and amylovoran. Pyr-stewartan EPS is produced by Pnss22 (wceO-) expressing the amsJ gene from plasmid pXL2. DC283 (pXL2, amsJ+) produces a mixture of stewartan and pyr-stewartan in a ratio of approximately 1:2. Depyruvylated amylovoran is produced by Ea100. Galactose residues that lack a terminal pyruvate are identified by an asterisk. Overexpression of wceO in Ea101 resulted in the production of gluco-amylovoran that contains the characteristic terminal glucose (VII) of stewartan. All EPS forms contain roughly 1000 subunits (this study and Nimtz et al., 1996a,b).

Although all genes in the two primary EPS biosynthetic gene clusters have been annotated and the mechanism of biosynthesis of the repeating units of amylovoran and stewartan has been proposed (Langlotz et al., 2011), several assigned functions have not been verified experimentally. For example, the activities encoded by the P. stewartii wceJ and wzx1 genes, which are located at the 3′ end of the wce-I gene cluster, have not been further characterized, presumably because mutations in these genes have no obvious effects on stewartan synthesis. The annotations of the corresponding genes in E. amylovora, amsJ and amsL are largely based on sequence homology with proteins of known functions (Langlotz et al., 2011). We provide experimental proof that AmsJ functions as a pyruvyl transferase in E. amylovora and when expressed in the heterologous P. stewartii background. In contrast, we show that WceJ has lost its functionality due to specific mutations in conserved regions of this class of proteins. This deficiency can be restored by site-specific mutagenesis. In addition, we show that Wzx1 and AmsL (here designated AmsL1), are proteins dedicated for the specific translocation of the pyruvylated RUs across the inner membrane, while the second Wzx2 or AmsL2 translocase encoded by the wce-III or ams-III loci, respectively, are dedicated for the translocation of the terminally glucosylated RUs. While the nature of the terminal residues does not influence the pathogenic process when the bacteria are manually inoculated into their respective hosts, we discuss a potential role of the distinguishing terminal residues as an adaptation to different hosts and lifestyles.

Results

Characterization of the E. amylovora AmsJ and P. stewartii WceJ pyruvyl transferase functions

AmsJ and WceJ are annotated as EPS pyruvyl transferases, although their functional roles in the production of stewartan and amylovoran have not been experimentally verified. We were interested in understanding the genetic basis and the potential biological significance of the terminal glucose residue in the side branch of stewartan opposed to the terminal 4,6-ketal pyruvate of amylovoran (Fig. 1). We created mutants of the wceJ and amsJ genes in P. stewartii and E. amylovora respectively (Fig. 2). These genes are predicted to encode pyruvyl transferases specific for the addition of terminal ketal pyruvate groups to the stewartan and amylovoran RUs. As shown in Fig. 3, the P. stewartii mutant strain Pnss24 (wceJ::km) produced mucoid colonies that were indistinguishable from the wild-type DC283 strain when grown on EPS-inducing agar medium, while the corresponding E. amylovora Ea100 (amsJ::km) mutant strain was non-mucoid and exhibited a dry, rough colony morphology. Expression of the amsJ gene from plasmid pXL2 in the Ea100 mutant restored the mucoid phenotype.

Figure 2.

Genetic organization of the stewartan and amylovoran biosynthetic genes. The wce-I (formerly cps) gene cluster of P. stewartii strain DC283 is composed of 12 genes from wceG1/cpsA (ASAPV5b: ACV-0289544) to wzx1 (cpsL) (ACV-0289531) including wceJ (cpsJ) (ACV-0289541). A homologous region derived from genomic DNA from the P. stewartii parent strain SS104 (Coplin et al., 1986) is deposited in GenBank under accession number AF077292. The wce-II and ams-II loci each consist of a single wceG2 (ACV-0286879) and amsG2 gene (GenBank CBJ46616) respectively. These are highly conserved orthologues of wceG1 (cpsA) and amsG1 (CBJ46848), the first genes of the primary wce-I and ams-I genetic loci respectively. The encoded proteins are undecarpenyl-phosphate galactose phosphotransferases, which are predicted to initiate RU synthesis on the lipid carrier (Whitfield, 2006). The wce-III and ams-III loci, each consist of two genes, wceO (ACV-0288191) and wzx2 (ACV-0288192) and amsO (CBJ46720) and amsL2 (CBJ46719) respectively. Previous experimental work suggests that wceG1, wceB (cpsE), wceM (cpsF), wceN (cpsG), wceK (cpsK) and wceO are involved in the addition of sugar residues (Geider, 2000; Carlier et al., 2009; Langlotz et al., 2011). The wzx1 and wzx2 genes encode flippase PST proteins responsible for the translocation of the lipid-linked repeating units across the inner membrane into the bacterial periplasm. The gene product of wceL (cpsD), homologue of amsC, is likely to encode a stewartan-specific Wzy EPS polymerase based on homology and detailed protein structural prediction algorithms (S.B. von Bodman, unpublished information). These enzymes are responsible for the oligomerization of the repeating units (Whitfield, 2006). The predicted products of wza (cpsB), wzb (cpsI) and wzc (cpsC) show high sequence homology to various EPS higher order polymerization and export machineries (Geider, 2000; Whitfield, 2006). The wceJ gene was predicted to encode a pyruvyl transferase, but is, as reported here, non-functional. Genes in ams-I and ams–III are highly homologous to those in wce-I and wce-III. Addition of sugar residues presumably involves amsG, amsD, amsE, amsK and amsO (Langlotz et al., 2011; this study). The gene product of amsJ (CBX81123) is an experimentally verified pyruvyl transferase (this study). The amsA, amsB amsH, amsI, amsL1 and amsL2 genes are predicted to encode proteins with functions similar to their counterparts in wce-I and wce-III. The protein sequences of WceJ and AmsJ share 69% identity. The identity of proteins Wzx1 and AmsL1 is 77%.

Figure 3.

Mucoid phenotype of P. stewartii and E. amylovora strains after 24h growth. (A) DC283, (B) Pnss24 (wceJ::Km), (C) Pnss25 (wzx1::Km), (D) Pnss26 (pXL2, amsJ+), (E) Ea273, (F) Ea100 (amsJ::Km), (G) Ea100 (pXL2, amsJ+), (H) Ea101 (amsL1::Km) and (I) Ea101 (pXL3, amsL1+). The colonies marked A, B and C are typical mucoid P. stewartii colonies, while the strain designated D is typical of a non-mucoid colony. The strains designated E, G and I are representative of mucoid E. amylovora, while those designated F and H represent non-mucoid strains.

Extraction of EPS produced by the P. stewartii wild-type and wceJ mutant strains after growth in liquid inducing medium showed that both the mutant and wild-type strains produced comparable amounts of stewartan EPS (Table 1). In contrast, the E. amylovora amsJ mutant strain Ea100 produced about 60% less amylovoran than the wild-type strain Ea273. The fact that this mutant still produced significant levels of EPS allowed us to characterize the polymer as described below. Complementation of Ea100 with amsJ expressed from plasmid pXL2 restored wild-type levels of amylovoran production. These studies confirmed that amsJ plays a critical role in amylovoran biosynthesis, while wceJ appears to have little or no effect on stewartan production.

Table 1.  EPS production and virulence of P. stewartii and E. amylovora strains.
StrainStewartan productiona± 95% CIAmylovoran productionb± 95% CIVirulence on corncVirulence on apple treesd
  • a. 

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

  • b. 

    Total quantity of free amylovoran in culture supernatant, in fg/cfu ± 95% CI.

  • c. 

    Symptoms on corn seedlings 10 days after inoculation (0 = no symptoms; 1 = few scattered lesions; 2 = scattered water soaking symptoms; 3 = numerous lesions and slight wilting; 4 = moderately severe wilt; 5 = death; nd, not determined). The EPS quantification assays and virulence assays were performed in triplicates and repeated once.

  • d. 

    Length of necrotic tissue on apple trees 6 days after inoculation, in cm ± 95% CI (nd, not determined). The EPS quantification assays and virulence assays were performed in triplicates and repeated once.

P. stewartii    
 DC2833.4 ± 0.4 4 
 Pnss24 (wceJ::Km)2.8 ± 0.3 4 
 Pnss25 (wzx1::Km)3.0 ± 1.0 4 
 DC283 (pXL2,amsJ+)4.0 ± 0.6 4 
 Pnss25 (pXL2,amsJ+)1.0 ± 0.2 nd 
 Pnss22 (wceO::frt)<0.01 0 
 Pnss22 (pXL2,amsJ+)2.1 ± 0.7 4 
 Pnss26(wceO-/wzx1::Km) (pXL2)<0.01 nd 
 Pnss27(wceO-/wzx2:Gm) (pXL2)1.5 ± 0.4 nd 
E. amylovora    
 Ea273 613 ± 7 7.6 ± 1.3
 Ea100 (amsJ::Km) 237 ± 2 0.2 ± 0.1
 Ea100 (pXL2,amsJ+) 530 ± 7 11.1 ± 1.3
 Ea100 (pXL1,wceJ+) 253 ± 5 nd
 Ea100 (pXL4,amsO+) 537 ± 12 nd
 Ea100 (pXL1-r,wceJ-r) 567 ± 25 nd
 Ea100 (pXL2-28c, amsJ-28c) 237 ± 4 nd
 Ea101 (amsL1::Km) <1 0
 Ea101 (pXL3,amsL1+) 555 ± 7 12.43 ± 1.42
 Ea101 (pAUC42,wceO+) 573 ± 15 nd
 Ea102 (amsO::Km) 540 ± 40 nd
 Ea273 (pAUC44-Cm,wceO+wzx2+) 507 ± 4 7.6 ± 1.4

Structural analyses of amylovoran and stewartan

The AmsJ and WceJ proteins share 69% identity. Glycosyl composition and linkage analyses by gas chromatography-mass spectrometry (GC-MS) of the respective EPS forms by us, Nimtz et al. (1996a,b) and Busson et al. (2001), allowed the structural assignments of the native and derived EPS RUs schematically displayed in Fig. 1. The linkage pattern of native amylovoran and stewartan was identical to that reported by Nimtz et al. (1996a,b). Specifically, amylovoran produced by the wild-type strain Ea273 (Table 2) is distinguished by the presence of a 4, 6-substituted galactosyl residue (2,3-Di-O-methyl galactitol) due to the 4,6-linked terminal ketal pyruvate. This substitution is absent in the EPS produced by the Ea100 amsJ mutant strain, which instead yielded a predominantly unsubstituted galactosyl V residue (2,3,4,6-Tetra-O-methyl galactitol) and a fractional increase in both the side branch terminal glucose residue VII (2,3,4,6-Tetra-O-methyl glucitol) and 6-linked galactose V (2,3,4-Tri-O-methyl galactitol). In contrast, the wceJ mutant strain Pnss24 yielded a polymer that was indistinguishable from the wild-type strain DC283 (Table 3).

Table 2.  Carbohydrate linkage analysis of EPS preparations of E.amylovora strains.
Peracylated derivative ofaDerivative of residuebPosition of substitutionEa273; Ea100(pXL2); Ea100(pXL1-r); Ea102Ea100; Ea100(pXL1); Ea100(pXL2-28c)Ea100(pXL4); Ea101(pAUC42); Ea273(pAUC44-Cm)
  • a. 

    The most significant signals for distinction of wild-type and Ea100 amylovoran are marked by an asterix (*).

  • b. 

    See Fig. 1 for corresponding sugar residue in the respective EPS RU structures.

  • The −, +, ++, +++ characters indicate the absence or approximate relative abundance of peracylcated derivatives of the mutant and wild-type EPS forms.

Galactitol     
 *2,3,4,6-Tetra-O- methyl-EA-V +
 2,4,6-Tri-O-methyl-A-III, EA-III GA-III3++++++
 2,3,4-Tri-O-methyl-A-II,EA-II, GA-II, GA-V6++++++++
 2,6-Di-O-methyl-A-I, EA-I, GA-I3;4+++
 *2,3-Di-O-methyl-A-V4;6+
 2-Mono-O-methyl-A-I, EA-I, GA-I3;4;6++++++
Glucitol     
 2,3,4,6-Tetra-O- methyl-A-VI,EA-VI, GA-VI, GA-VII ++++++
Table 3.  Compositional and carbohydrate linkage analysis of EPS preparations of P. stewartii strains.
Peracylated derivative ofaDerivative of residuebPosition of substitutionDC283; Pnss25(pXL2)DC283(pXL2)Pnss22(pXL2)DC283(pXL1-r)
  • a. 

    The most significant signals for distinction of wild-type and pyruvylated stewartan (pyr-stewartan) are marked *.

  • b. 

    See Fig. 1 for corresponding sugar residue in EPS structure.

  • c. 

    Proportional amount of the total alditol acetate derivatives (AA) after acid hydrolysis of EPS.

  • The −, +, ++, +++ characters indicate the absence or approximate relative abundance of peracylcated derivatives of the mutant and wild-type EPS forms.

Galactitol percentage of total AAc  51576156
 2,4,6-Tri-O-methyl-S-III, PS-III3++++++++
 *2,3,4-Tri-O-methyl-S-V6++++
 2,6-Di-O-methyl-S-I, PS-I3;4++++
 *2,3-Di-O-methyl-PS-V4;6++++
 2-Mono-O-methyl-S-I, PS-I3;4;6++++++++
Glucitol percentage of total AAc  49433944
 *2,3,4,6-Tetra-O-methyl-S-VI, PS-VI, S-VII ++++++++
 2,3,4-Tri-O-methyl-S-II, PS-II6++++++++

Because carbohydrate methylation analysis provides only indirect proof for the presence of the ketal pyruvate, we sought to generate direct verification by high-resolution 1H-NMR spectroscopy. The spectra displayed in Fig. 4 show that native amylovoran has a chemical shift signature at 1.46 ppm, which is indicative of the presence of a ketal pyruvate group (Busson et al., 2001). This signal is notably absent in the EPS produced by the amsJ mutant Ea100 and reappears when the mutant is complemented with a functional copy of amsJ. These results corroborate the GC-MS methylation results and provide experimental validation that AmsJ is the enzyme responsible for the addition of the 4,6 ketal pyruvate to the amylovoran RUs. In contrast, both 1D and 2D-NMR data obtained from native stewartan and those produced by the wceJ mutant strain Pnss24 are identical, all lacking the pyruvate signals, further supporting the prediction that wceJ encodes a non-functional entity. We know that the wceJ gene directs the translation of a relatively stable protein based on the analysis of a His-tagged version, His-WceJ, which when expressed in the Ea100 (amsJ-) mutant, allows the immunodetection of a protein with the predicted molecular weight (Fig. S2). However, this protein product was not able to restore amylovoran production of Ea100 mutant to the wild-type level (Tables 1 and 2).

Figure 4.

1H-NMR spectra of amylovoran EPS. Amylovoran isolated from wild-type E. amylovora strain Ea273 yields a characteristic pyruvate signal at 1.46 p.p.m. (arrow) (A). This specific signal is lacking in the amsJ mutant strains Ea100 (B), and reappears in the strain complemented with a functional amsJ gene carried on plasmid pXL2 (C). Note: although there is some degree of spectral variations due to the sample preparation and solubility, the spectra clearly show the presence and absence of the signature ketal pyruvate signal.

Restoration of WceJ functionality by PCR-based random mutagenesis

Based on the above experiments, wceJ appeared to encode a non-functional, but relatively, stable protein. Comparison of the deduced amino acid sequences of WceJ [ASAP annotation database: ACV-0289541 (Glasner et al., 2003)] and AmsJ (GenBank: CBX81123.1) revealed a few key differences in regions of predicted protein structural conservation (Fig. S1). Exposure of the native wceJ gene to PCR-based random mutagenesis, and transfer of the mutated gene pool into the amsJ mutant strain, recovered one Ea100 transformant that exhibited a mucoid phenotype. The EPS produced by this strain gave the same methylation profile as the native pyruvylated amylovoran (Table 2). The established DNA sequence of the mutated version of wceJ, designated wceJ-R (R for reversion), revealed several nucleotide substitutions that account for five amino acid changes as summarized in Fig. 5 and Table 4. The most dramatic change of the predicted translated WceJ-R protein relates to a proline to serine substitution at position 10, which, in fact, corresponds to a threonine residue in the functional AmsJ protein at this position (Fig. 5, Fig. S1). Secondary structural analysis using the PSIPRED protein structure prediction algorithm (http://bioinf.cs.ucl.ac.uk/psipred/) showed that this proline replacement appears to disrupt and shorten a prominent helical secondary structure that is preserved in the AmsJ protein (data not shown). Based on this information, we created a corresponding site-directed mutation in the native amsJ gene carried on plasmid pXL2 to create an adenine to cytosine substitution at nucleotide position 28, thereby creating a codon that specifies a proline instead of a threonine residue at position 10. As shown in Table 2, this single nucleotide substitution inactivated the AmsJ pyruvyl transferase activity as inferred from the linkage profile of the EPS produced by Ea100(pXL2-28c), which was identical to that obtained from the EPS produced by the Ea100 parent strain. The expression of the wceJ-R gene, carried on plasmid pXL1-r in the P. stewartii wild-type DC283 strain, led to the production of a hybrid stewartan with about 60% of the RUs carrying a ketal pyruvate (Table 3). Together, these data show that AmsJ in E. amylovora is a functional pyruvyl transferase that is responsible for the addition of a terminal ketal pyruvate to amylovoran, while WceJ in P. stewartii is inactive, but can be restored by site-specific mutagenesis.

Figure 5.

Protein sequence alignment of WceJ, WceJ-R and AmsJ. Only regions where substitutions occur are shown. Conserved residues are highlighted in grey. Stars (*) indicate substitutions in WceJ-R and numbers indicate the positions of amino acid residues in WceJ.

Table 4.  DNA and amino acid substitutions in WceJ-R and AmsJ-28C.
 BaseBase substitutionPosition in codonAmino acidAmino acid substitution
WceJ-R28C→T110Pro→Ser
215G→A272Gly→Asp
244C→T182Silent
253A→G185Asn→Asp
1066A→G1356Met→Val
1117A→G1373Ser→Gly
AmsJ-28c28A→C110Thr→Pro

To ensure that the mutation giving rise to the proline 10 residue was characteristics of P. stewartii isolates and not merely a function of prolonged propagation of strain DC283 in the laboratory, we compared the wceJ DNA sequence of strain DC283 with that of the parent strain SS104 (Coplin et al., 1986) and three additional, independent natural isolates DC424, DC441, DC442 (Coplin et al., 2002). First, we found that the full-length DNA sequence of the wceJ gene (GenBank database: AF077292.2) of the minimally propagated parent strain SS104 and the laboratory strain DC283 (ASAP annotation database: ACV-0289541) are a 100% conserved both carrying a CCC codon specifying a proline 10 residue in the translated protein sequence (Fig. S1). Most importantly however, the three other independent natural P. stewartii isolates have identical N-terminal protein coding sequences that all feature a proline at amino acid position 10 instead of a threonine residue that is present in AmsJ (Fig. S1).

The effect of AmsJ on stewartan synthesis

Bernhard et al. (1996) showed that expression of the ams-I gene cluster in P. stewartii led to the production of partially pyruvylated stewartan. To confirm that this activity related directly to the amsJ gene, we transferred plasmid pXL2 (amsJ+) into the wild-type P. stewartii background. This strain produced essentially wild-type levels of EPS (Table 1), which, based on composition analysis, showed a relative increased ratio of total galactitol to glucitol residues. Moreover, linkage analysis of this polymer established that the shift from glucitol to galactitol derived primarily from a decrease in terminal glucose residues (2,3,4,6,Tetra-O-methyl glucitol) that are β-(1,6)-linked to the sub-terminal galactose residue V (2,3,4,Tri-O-methyl galactitol). These structural changes were accompanied further with a proportional increase in the amount of ketal pyruvate that is 4,6-linked to galactose V. Composition analysis showed that the amount of terminally pyruvylated RUs was roughly 65% and terminally glucosylated RUs about 35% in the wild-type strain that carries a functional WceO terminal glucosyl transferase. In contrast, expression of AmsJ in the Pnss22 (wceO-) mutant genetic background yielded a fully pyruvylated stewartan polymer (pyr-stewartan) (Fig. 1 and Table 3), although at reduced overall EPS levels (Table 1).

The role of the ams-III locus in amylovoran production

Erwinia amylovora harbours a gene system, designated ams-III that corresponds to the wce-III locus in P. stewartii (Fig. 2). The two gene systems have conserved promoter regions and are regulated by the RcsA/RcsB regulatory complex (Carlier et al., 2009; X. Wang, unpubl. data). Schematically shown in Fig. 2, the ams-III locus encodes a predicted terminal glucosyl transferase, AmsO (CBJ46720.1) together with a second Wzx-like protein, AmsL2 (CBJ46719.1). An insertion mutation in the amsO gene had no effect on amylovoran synthesis both in terms of quantity (Table 1) and structure (Table 2). In contrast, overexpression of amsO from plasmid pXL4 in the Ea100 (amsJ-) genetic background produced wild-type levels of EPS that, by structural analysis, proved to be 100% terminally glucosylated as indicated by the lack of pyruvylated, unsubstituted galactosyl V residues and increased amounts 6-linked galactosyl V (2,3,4-Tri-O-methyl galactitol) and terminal glucosyl VII residues (2,3,4,6-Tetra-O-methyl glucitol). These results show that AmsO is only fully functional when overexpressed and in the absence AmsJ in E. amylovora. These results also explained the observation that the residual EPS produced by the amsJ mutant yielded a fractional increase in both the terminal side branch glucosyl VII (2,3,4,6-Tetra-O-methyl glucitol) and the 6-linked galactose V (2,3,4-Tri-O-methyl galactitol) residues.

Repeating unit translocases

The amsL1 (formerly amsL) gene located in the ams-I amylovoran biosynthetic operon has been annotated as a Wzx-type ‘flippase’ thought to be responsible for translocating the amylovoran RUs across the inner membrane during amylovoran biosynthesis. Consistent with this prediction is that the AmsL1 protein features characteristic eleven transmembrane segments (TMS) (Table S2), and a loosely conserved domain typical of all polysaccharide specific transport (PST) proteins (Polysacc_synt domain, pfam01943). In this study, we created an amsL1 mutant strain, designated Ea101. When grown on EPS-inducing minimal agar medium, the strain was unable to produce detectable amounts of amylovoran. Moreover, this strain had considerable difficulty growing on EPS inducing medium (Fig. 3H). Expressing amsL1 from plasmid pXL3 restored wild-type growth and amylovoran production. These data confirm that amsL1 is essential for normal amylovoran synthesis. Interestingly, overexpression of wceO in the Ea101 (amsL1-) background also restored EPS production to wild-type levels (Table 1), suggesting another Wzx-like protein, most likely AmsL2, is responsible for translocating gluco-amylovoran specifically.

Pantoea stewartii also has two Wzx-like ‘flippase’ proteins associated with stewartan EPS biosynthesis (Fig. 2). The wzx1 gene (ASAP database: ACV-0289531) is part of the wce-I locus; the wzx2 gene (ASAP database: ACV-0288192) is genetically linked to wceO (ASAP database: ACV-0288191), and the two genes comprise the wce-III stewartan biosynthetic region. As reported by Carlier et al. (2009), wzx2 mutation eliminates stewartan synthesis completely. In this study, we show that a mutation in the wzx1 gene (strain Pnss25) had no effect on stewartan synthesis (Fig. 3C, Table 1). However, overexpression of amsJ from plasmid pXL2 in the Pnss25 (wzx1-) background reduced EPS synthesis to 25% of the wild-type strain (Table 1) presumably because multiple copies of AmsJ were generating terminally pyruvylated RUs, which could not be translocated for lack of Wzx1. Methylation analysis of the residual EPS produced by Pnss25 (pXL2, amsJ+, wzx1-) showed this polymer to be 100% terminally glucosylated. In contrast, expression of pXL2 (amsJ+) in the wceO-/wzx1- double mutant strain Pnss26 completely abolished EPS synthesis, while introduction of pXL2 (amsJ+) into the wceO-/wzx2- double mutant, wxz1+ background of strain Pnss27 yielded significant levels of pyruvylated EPS (Table 1). These data together show that Wzx1 is functional and dedicated for the translocation of pyr-stewartan RUs, while the translocation of glucosylated RUs requires Wzx2.

AmsJ and AmsL1 are required for E. amylovora virulence on apple trees, while WceJ and Wzx1 are not required for P. stewartii virulence in corn

Pantoea stewartii and E. amylovora require stewartan and amylovoran for full virulence in corn and on apple trees respectively. Manual inoculations of corn seedlings with the wceJ and wzx1 mutant strains, Pnss24 and Pnss25, respectively, induced Stewart's wilt symptoms similar to those inoculated with the wild-type strain (Table 1). This is not surprising because these mutant strains produce wild-type amounts of genuine stewartan EPS. In contrast, apple tree seedlings inoculated with the amsJ mutant Ea100 or amsL1 mutant Ea101 were basically asymptomatic, while the respective complemented strains regained virulence (Table 1). These results confirm that amsJ and amsL1 are essential for the development of fire blight on apple trees. Interestingly, the P. stewartii strains DC283 (pXL2, amsJ+) and Pnss22 (pXL2, amsJ+), which produce partially and fully pyruvylated stewartan, respectively, were also fully virulent in maize inoculation assays. Similarly, E. amylovora strain Ea273 (pAUC44-Cm, wceO+/wzx2+) producing the heterologous gluco-amylovoran was fully virulent on apple. These data suggested that the chemical nature of the terminal residue of the respective RUs has no effect on virulence when the host plants are infected by artificial, manual inoculation.

Discussion

In this study, we focused on a set of genes present in P. stewartii and E. amylovora that dictate the nature of the terminal side branch residue of the respective RUs, which in stewartan is a β-(1,6)-linked terminal glucose, and in amylovoran a 4,6-ketal pyruvate. We confirmed the pyruvylation activity of AmsJ by carbohydrate linkage analysis, as well as 1H-NMR (Table 2 and Fig. 4). Parallel studies in P. stewartii established that the wceJ gene, although specifying a protein product (Fig. S2), has no role in stewartan production (Fig. 3 and Table 1). Comparative genetics, protein structure/function predictions, coupled with random, site-specific mutagenesis allowed us to pinpoint a structurally significant proline residue in WceJ that renders the protein non-functional. Interestingly, the same proline 10 residue is present in four independent natural isolates of P. stewartii, which strongly supports our hypothesis that this proline substitution and loss of WceJ functionality is the likely result of a divergent evolutionary trajectory in which one capping enzyme system (WceO/Wzx2) has taken on a more dominant role over the competing mechanism (WceJ/Wzx1). Because the basal and induced transcriptional levels of all six gene systems are comparable (Carlier et al., 2009; X. Wang, unpubl. data), the observed functional divergence appears to occur at the protein, not at the transcriptional level.

The other interesting theme emerging from this study is that each of the terminal transferases comes equipped with a Wzx-type RU translocase function that appears to be specific for the transfer of either the terminally glucosylated or pyruvylated RUs. This is in contrast to reports suggesting that Wzx-like PST proteins involved in O-antigen biosynthetic systems have relaxed specificity requirements toward the structure and composition of polysaccharides (Marolda et al., 2004), and that the Wzx-type proteins display specificity only for the first sugar linked to the undecaprenyl phosphate lipid carrier (Feldman et al., 1999). This first residue is galactose in both stewartan and amylovoran RUs, which would suggest that the two translocases should be cross-functional. A recent report indicates that substrate specificity of the Wzx PST proteins is related to the number of charged residues within the TMS of these proteins (Islam et al., 2010). As shown in Table S2, the TMS of Wzx1 comprise a higher percentage of positively charged residues than those of Wzx2, which is consistent with the observed substrate selectivity for the pyruvylated versus glucosylated RUs.

Taken together, the observed functional divergence between the enzyme systems that specify two distinct terminally capped EPS RUs is consistent with the respective biosynthetic output, namely pyruvylated amylovoran in E. amylovora and terminally glucosylated stewartan in P. stewartii. Thus, it would appear that the two organisms derived from a common progenitor, but have diverged to adapt to different, complex biotrophic lifestyles in which P. stewartii benefits from the synthesis of a polar, more readily hydrated EPS as a function of terminally glucosylated RUs, while E. amylovora may have gained an advantage by producing a more acidic, terminally pyruvylated EPS. However, we showed that virulence of either pathogen is not influenced by the nature of the EPS produced when the host plants are infected by manual inoculation (Table 1). This is consistent with the finding by Bernhard et al. (1996) showing that the P. stewartii strain producing a hybrid stewartan was virulent when assayed by manual inoculation. However, here it is important to consider that these inoculation practices deal with bacteria that are most likely physiologically distinct from the bacteria that are carried by the insect vectors to the host. Manual inoculation also sidesteps critical early events of the natural infection process, which commence the moment the bacteria are deposited onto or into the plant tissue and before establishing the systemic, biofilm-type xylem colonization (Koutsoudis et al., 2006). The Stewart's wilt and fire blight disease processes differ in how the bacteria are carried to the plant host, and how and where these organisms overwinter. P. stewartii maintains a relatively persistent association with a beetle insect vector that also serves as an overwintering host for the bacterium, while E. amylovora overwinters in the canker tissue of the infected plant host. The different physicochemical properties of a terminally pyruvylated versus glucosylated form of EPS (Rinaudo, 2004) may be important for some aspects of the different infection schemes. Such modifications play a critical role in the host range determination of other plant-associated bacteria, most notably, the rhizobia plant symbionts (Philip-Hollingsworth et al., 1989; Pellock et al., 2000; Fraysse et al., 2003). For example, the terminal glucose-substituted EPS may be less viscous and more readily hydrated (Fidanza et al., 1989), thereby providing an advantage for the pathogen after deposition into the beetle feeding wounds, where water may be limiting. Furthermore, P. stewartii might benefit from a terminal glucose residue for protection from phage attacks while residing in the beetle gut particularly considering that certain phage tailspike-associated polysaccharide lyases have specificity for terminally pyruvylated polymers (Sutherland, 1976; Ahlgren, 1991; Labrie et al., 2010). In case of E. amylovora, the terminal pyruvyl group of amylovoran contributes to the overall anionic properties that may render the amylovoran EPS more stable and ‘sticky’ particularly in the presence of divalent cations (Qin et al., 2007). Another intriguing possible role for the pyruvylated amylovoran is the potential to suppress Ca2+ induced innate immunity as a function of sequestering apoplastic calcium ion (Aslam et al., 2010).

In summary, our work demonstrates an interesting comparative evolutionary dynamic in two related bacterial plant pathogens that resulted in the production of distinct chemical forms of EPS. Unquestionably, microbial polysaccharides constitute anything but inert barriers between the microbe and its environment, and their distinctive physicochemical properties may well hold important clues about their biology in nature and adaptability to different hosts and environments.

Experimental procedures

Bacterial strains and plasmids

Strains and plasmids are summarized in Table S1. The E. coli strains used as cloning hosts include DH5α (Life Technologies) 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 Luria–Bertani broth (LB) in presence of appropriate antibiotics (Carlier et al., 2009). The P. stewartii strains were grown at 28°C in LB in presence of 30 mg ml−1 of nalidixic acid or AB minimal medium supplemented with 0.2% glucose (Clark and Maaloe, 1967). 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 and Farrand, 1995; von Bodman et al., 1998). DNA amplification was performed using Ex Taq Polymerase (Takara Bio) and synthetic oligonucleotides ordered to specification from Eurofins Scientific, Demoines IA (Table S2). Plasmids were introduced into electrocompetent cells by electroporation (Hanson and Phillips, 1981).

Sequence analysis

The P. stewartii DC283 genome sequence data (version 5b) and preliminary genome annotation are available through the ASAP database at https://asap.ahabs.wisc.edu/asap/home.php?returnVars=YTowOnt9&formSubmitReturn=1 (accessed September 2011). The sequence data of E. amylovora Ea273 (ATCC 49946) were also accessed through the ASAP database (Sebaihia et al., 2010). 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. The HMMTOP algorithm (http://www.enzim.hu/hmmtop/) and TMHMM algorithm (http://www.cbs.dtu.dk/services/TMHMM/, accessed August 2011) were employed to predict the membrane topology and conserved domains used the Pfam algorithm (http://pfam.sanger.ac.uk/, accessed August 2011). Secondary structural analysis of WceJ and AmsJ relied on the PSIPRED protein structure prediction algorithm (http://bioinf.cs.ucl.ac.uk/psipred/).

Deletion mutagenesis and allelic replacement

Mutant strains Pnss24, Pnss25, Pnss26 and Pnss27 were created as previously described (Carlier et al., 2009). Mutant strains Ea100, Ea101, Ea102 were created using the red cloning strategy (Wang et al., 2009). In all cases, plasmid pkD46 encoding the Red recombinase of λ phage are thermosensitive for replication and was cured at 37°C.

Construction of plasmids pXL1, pXL2, pXL3, pXL4

The wceJ, amsJ, amsL1 and amsO coding sequences were amplified by PCR using primers specifying appropriate restriction sites (Table S2). The PCR products were digested with the restriction enzymes and ligated into the vector pBBR1MCS-4 (Kovach et al., 1995), giving rise to pXL2, pXL3 and pXL4 respectively. Plasmid pXL1 specifying a C-terminal his-tagged WceJ protein was generated by PCR amplification of wceJ with primers containing histidine tandem repeats, enzyme pairwise EcoRI/XbaI digestion and ligation into the vector plasmid pBBR1MCS-4. All these genes are inserted downstream of a lac promoter.

Random mutagenesis of WceJ

Random mutagenesis was carried out using Diversify PCR random mutagenesis kit (Clontech) as described by manufacturer. Two oligonucleotides wceJhisfor and wceJhisrev (Table S2) were used to amplify the full-length wceJ coding region. The resulting PCR product was ligated into pBBR1MCS-4 and the ligation mixture was transformed into E. coli DH5α. The mutant library was isolated from E. coli DH5α and electroporated into Ea100 (amsJ-) competent cells for screening of mucoid colonies.

PCR-based site-directed mutagenesis of AmsJ

Site-directed mutagenesis was carried out using Quikchange II XL site-directed mutagenesis kit (Stratagene) as described by manufacturer. Two complementary oligonucleotides amsJ28c and amsJ28c-anti (Table S2) were used to amplify the mutated amsJ. The non-mutated template was digested with DpnI restriction enzyme and DpnI-treated DNA was transformed into XL10-Gold ultracompetent cells.

Purification and quantification of stewartan and amylovoran

The purification of stewartan and amylovoran was performed as previously described (Carlier et al., 2009). Briefly, wild-type and mutant EPS forms were isolated from cells grown for 24 hours at 28°C in AB minimal liquid media supplemented with 0.2% glucose and 0.1% yeast extract. The bacterial cells were removed by centrifugation at 5000 g and the supernatant was precipitated with cetylpyridinium chloride (1% final concentration). The precipitate was resuspended in phosphate-buffered saline (PBS) from which EPS was recovered by standard ethanol precipitation. The crude EPS was treated with Benzonase (EMD) and proteinase K (Invitrogen) to remove nucleic acids and proteins. The EPS samples were further purified on a 100 × 5 cm column packed with Sephacryl S500-HR (GE Healthcare) in 0.1 M NaCl and then dialysed extensively against deionized water (12 kDa cut-off) and lyophilized. Stewartan was quantified using inhibition enzyme-linked immunosorbent assay (Carlier et al., 2009). A turbidity assay was employed for the quantification of amylovoran (Bellemann et al., 1994).

Carbohydrate composition and linkage analysis

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 equipped 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, 200 µg of dry EPS was methylated by addition of a 0.2 ml volume of a NaOH-DMSO and 0.1 ml volume of methyl iodide (Anumula and Taylor, 1992). The sample was vortexed repeatedly and methylated carbohydrates were extracted with chloroform and dried under a nitrogen stream. The methylated sample was then hydrolysed by trifluoroacetic acid (TFA), reduced with NaBD4 and acetylated by acetic anhydride. The resulting partially methylated alditol acetates (PMAAs) were dissolved in chloroform and injected into the GC-MS instrument.

Structural analysis of EPS by nuclear magnetic resonance (NMR)

Purified polymerized amylovoran samples were dissolved in D2O and placed in a 5 mm Shigemi NMR tube for proton NMR experiment. Wild-type and wceJ mutant stewartan oligomers were obtained after Dpo-specific enzymatic hydrolysis (Kim and Geider, 2000) and dissolved in D2O. 1-D proton and 2-D NMR data were collected using a Varian Inova 500 or 600 MHz spectrometer at 25°C. Chemical shifts were measured relative to internal acetone.

Virulence assay on sweet corn seedlings and apple trees

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 mE m−2 s−1 light intensity. Plants were inoculated 5 days after germination with 50 µl of bacterial suspension in PBS containing roughly 1 × 103 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.

Virulence assays on apple trees were performed using young annual shoots of ‘Gala’ apple, approximately 20 cm in length, by pricking the tip with a needle, and transferring 2 µl of bacterial suspension (OD600 = 0.1) onto the wounded tissue. For each bacterial strain, six to ten shoots were inoculated. Plants were kept in a greenhouse at 25°C and 16 h light photoperiod, and evaluated for disease development after 6 days following inoculation by measuring length of the necrotic tissue.

Acknowledgements

We thank Dr Martha Morton for her service and help in conducting and interpreting the NMR experiments. We also thank Professor Peczuh for his assistance in analysing the NMR results. We particularly thank Dr David Coplin and Doris Majerczak at the Ohio State University for providing unpublished DNA sequence information of the wceJ genes present in natural isolates of P. stewartii. We thank Dr Youfu Zhao for his laboratory's assistance in performing the virulence assays on apple trees and also for critical reading of this manuscript. The effort by co-author Yang was supported by a grant to Y.Z. from the USDA NIFA Grant No. 2010-65110-20497. The majority of the work presented was supported by the National Science Foundation through grant nos. 0619104 and 1053869, the USDA CREES CONS00775, and the University of Connecticut Research Foundation. In addition, this material is based on work supported by the National Science Foundation while S.vB. was working at the Foundation. Any opinion, finding, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the National Science Foundation.

Ancillary