The Modality of Enterobacterial Common Antigen Polysaccharide Chain Lengths Is Regulated by o349 of the wec Gene Cluster of Escherichia coli K-12

Abstract

The assembly of the phosphoglyceride-linked form of enterobacterial common antigen (ECA_PG) occurs by a mechanism that involves modulation of polysaccharide chain length. However, the genetic determinant of this modulation has not been identified. Site-directed mutagenesis of o349 of the Escherichia coli K-12 wec gene cluster revealed that this locus encodes a Wzz protein that specifically modulates the chain length of ECA_PG polysaccharides, and we have designated this locus wzz_ECA. The Wzz_ECA-mediated modulation of ECA_PG polysaccharide chains is the first demonstrated example of Wzz regulation involving a polysaccharide that is not linked to the core-lipid A structure of lipopolysaccharide.

Enterobacterial common antigen (ECA) is a unique cell-surface glycolipid that is present in all gram-negative enteric bacteria (14, 17, 21, 30). The carbohydrate portion of ECA consists of a linear heteropolysaccharide chain comprising the trisaccharide repeat unit →3)-α-d-Fuc4NAc-(1→4)-β-d-ManNAcA-(1→4)-α-d-GlcNAc-(1→, where GlcNAc is N-acetyl-d-glucosamine, ManNAcA is N-acetyl-d-mannosaminuronic acid, and Fuc4NAc is 4-acetamido-4,6-dideoxy-d-galactose (16, 18). The major form of ECA, ECA_PG, consists of polysaccharide chains of various lengths that are covalently linked to phosphoglyceride via phosphodiester linkage involving the GlcNAc residue of the first repeat unit (15, 28). The occurrence of two other minor forms of ECA, ECA_LPS and ECA_CYC, has also been demonstrated. ECA_LPS is characterized by the covalent linkage of ECA polysaccharide chains to the core region of lipopolysaccharide (LPS) (12, 14, 31, 34), whereas ECA_CYC is a water-soluble cyclic form of ECA containing four to six trisaccharide repeat units (9, 14, 33). Although ECA_PG is present in all members of the Enterobacteriaceae, the occurrence of ECA_LPS and ECA_CYC appears to be restricted to certain gram-negative enteric bacteria (12, 14, 31, 34).

The known genetic determinants of ECA are located in the wec (formerly rfe/rff) gene cluster located at min 85 of the Escherichia coli chromosome (14, 22, 30). (The genetic nomenclature used here is in accordance with that proposed by Reeves et al. [27] for genes involved in the synthesis of bacterial polysaccharides.) The nucleotide sequence of this gene cluster includes 12 open reading frames (Fig. 1B). Biochemical and genetic studies have identified the function of 7 of these genes (30); however, the functions of o349, wecD, o416, o359, and o450 have not been unequivocally demonstrated. Many of the early steps involved in ECA synthesis have been firmly established (Fig. 1A). The ECA trisaccharide repeat unit is assembled as an undecaprenylpyrophosphate (Und-PP)-linked intermediate (lipid III) by the same general mechanism established for the assembly of the repeat units of Wzy-dependent O-antigens (for a review, see reference 35). Although details concerning subsequent steps of ECA assembly have not yet been determined, the available information suggests that these steps occur by the same general mechanism established for the assembly of Wzy-dependent O antigens. In this regard, previous studies led to the conclusion that wecF (o716) of the wec gene cluster encoded the Fuc4NAc transferase involved in ECA synthesis (5, 8). However, recent examinations of the nucleotide sequence of this region revealed that it actually contains two putative open reading frames, o359 and o450 (1a). The results of preliminary investigations support the conclusion that o359 encodes WecF, and the considerable homology between the predicted o450 gene product and several Wzy proteins (24) suggests that o450 of the wec gene cluster encodes the Wzy protein involved in ECA polysaccharide chain elongation. Thus, it is likely that lipid III is also synthesized on the cytoplasmic face of the inner membrane and subsequently translocated to the periplasmic face of the inner membrane, where assembly of ECA polysaccharide chains occurs by a Wzy-dependent “block polymerization” mechanism.

FIG. 1.

The ECA biosynthetic pathway and genes of the wec gene cluster. (A) Current status of our knowledge concerning the enzymatic reactions involved in the biosynthesis and assembly of ECA. The genetic determinants of the various enzymes are indicated next to the reaction catalyzed by each of the respective enzymes. Abbreviations are as follows: Und-P, undecaprenylmonophosphate; Und-PP, undecaprenylpyrophosphate; Glc, glucose; GlcNAc, N-acetyl-d-glucosamine; ManNAc, N-acetyl-d-mannosamine; ManNAcA, N-acetyl-d-mannosaminuronic acid; Glc-1-P, glucose 1-phosphate; FucNH₂, 4-amino-4,6-dideoxy-d-glucose; Fuc4NAc, 4-acetamido-4,6-dideoxy-d-glucose; CoASH, coenzyme A; α-KG, α-ketoglutaric acid; PPi, inorganic pyrophosphate; ECA_PG, phosphoglyceride-linked ECA polysaccharide chains. The structures of lipids I, II, and III are defined in the inset. (B) The wec gene cluster. Arrowheads designate the direction of transcription. The 14-kb scale below the gene cluster is provided for reference.

The chain length of Wzy-dependent O antigens (35, 36), as well as some lipid A-core-linked capsular polysaccharides (10, 11), is regulated by Wzz (Rol, Cld). A number of Wzz homologues have also been identified in a variety of bacteria that are involved in the assembly of bacterial cell surface polysaccharides that are not linked to core-lipid A (36), but none of these homologues has been demonstrated to function in the regulation of polysaccharide chain length. Wzz-mediated regulation results in a modal or multimodal distribution of polysaccharide chain lengths when preparations are analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). In contrast, a nonmodal random distribution of chain lengths is observed in wzz mutants; however, the mechanism of Wzz-mediated modulation of polysaccharide chain length has not been determined. The degree of polymerization of ECA polysaccharide chains is also modulated, and a single modal distribution of chain lengths has been observed following SDS-PAGE analyses of ECA_PG preparations (1, 2, 23). Although the genetic determinant of ECA polysaccharide chain length regulation has not been identified, significant homology between the putative product of o349 of the wec gene cluster and various Wzz homologues has been noted (3, 4, 6, 25). The hydropathy profiles of all of these proteins—including that of o349—are strikingly similar, and they all possess a large hydrophilic domain that is flanked by apparent transmembrane domains at both the amino and carboxy termini (25). We demonstrate here that o349 of the wec gene cluster of E. coli K-12 does indeed encode a Wzz that functions to modulate ECA polysaccharide chain lengths. In addition, the Wzz_ECA-mediated modulation of ECA_PG polysaccharide chains is the first demonstrated example of Wzz regulation involving a polysaccharide that is not linked to the core-lipid A region of lipopolysaccharide.

Wzz_ECA is encoded by o349 of the wec gene cluster.

The assembly of ECA polysaccharide chains in E. coli AB1133 (23) results in a population of polysaccharide chains whose lengths range from approximately 1 to 14 repeat units with an apparent modal value of 5 to 7 (Fig. 2A, lane 1). A chromosomal o349::mini-TnlacZ insertion mutant was constructed by allelic exchange in order to evaluate the role of this gene in the regulation of ECA polysaccharide chain length. Briefly, the transposon TnlacZ (Km^r) was used to generate TnlacZ insertions in the coding region of o349 by the general strategy of Pradel et al. (26). Plasmid pCA53 (Tc^r Amp^r), a pBR322-based construct containing wecA and o349 on a 2.54-kb ClaI insertion fragment (Fig. 3) (22), was introduced by transformation into E. coli CC170 (19), which carries a chromosomal copy of TnlacZ. Cells in which transposition of TnlacZ to the plasmid had occurred were selected by subsequent growth of the transformants on Luria-Bertani (LB) agar plates containing tetracycline and ampicillin as well as high levels of kanamycin (300 μg/ml). A pool of plasmid DNA prepared from these cells was used to transform strain CC118 (Lac⁻) (19), and kanamycin-resistant transformants were selected on LB plates containing kanamycin as well as 5-bromo-4-chloro-3-indoyl-β-d-galactopyranoside (X-Gal) to differentiate between cells containing potentially in-frame fusions (blue colonies) from those containing out-of-frame insertions (white colonies). Plasmids were isolated from individual transformants, and those containing TnlacZ insertions in o349 were identified by restriction mapping. One of these plasmids, pJK2885, contained a TnlacZ insertion in an orientation opposite that of the direction of transcription of the o349 gene. The insertion in pJK2885 was stabilized by removal of the transposase by digestion with XhoI followed by religation to yield plasmid pJK2887 (o349::mini-TnlacZ, Km^r).

FIG. 2.

Effect of mutations in o349 of the wec gene cluster and waaL on the modality of ECA polysaccharide chain length. Exponentially growing cultures were incubated with [³H]GlcNAc for 30 min. The incorporation of radiolabeled GlcNAc into ECA was subsequently analyzed by SDS-PAGE and fluorography. (A) Lane 1, strain AB1133 (wild type); lane 2, strain CS2942 (o349::mini-TnlacZ). (B) Strain PR4100 (o349::mini-TnlacZ + pRL123 [wecB⁺ wecC⁺ rmlB⁺]). (C) Lane 1, strain PR4129 (o349::mini-TnlacZ + pRL139 [o349⁺]); lane 2, strain PR4130 (o349::mini-TnlacZ + pRL123 [wecB⁺ wecC⁺ rmlB⁺] + pRL139 [o349⁺]). (D) Lane 1, strain AB1133 (wild type); lane 2, strain CS2334 (waaL::TnphoA).

FIG. 3.

Physical map of the 5′ half of the wec gene cluster. The site and orientation of the mini-TnlacZ insertion in o349 are indicated. The cross-hatched rectangle for rmlA denotes a partial open reading frame. The chromosomal regions contained in plasmids pCA53, pRL123, and pRL139 are indicated by the respective open rectangles.

Introduction of the o349::mini-TnlacZ insertion into the E. coli K-12 chromosome by allelic exchange was accomplished by transformation of strain VJS803 (recBC sbcB) (32) with plasmid pJK2887. Kanamycin-resistant transformants were selected and subsequently screened for sensitivity to ampicillin. The insertion in one of these transformants, strain CS2898 (Km^r Amp^s), was then transduced into strain PR21548 (wecA::Tn10) (23) by using bacteriophage P1. Kanamycin-resistant transductants were selected, and subsequent screening for sensitivity to tetracycline revealed that all of the kanamycin-resistant transductants were sensitive to tetracycline. The recA56 allele of JC10240 (recA56 srl-300::Tn10) (7) was introduced into one of the transductants, CS2900, by conjugation in order to prevent recombination between genes harbored on plasmids and their chromosomal counterparts during subsequent complementation studies. Kanamycin- and tetracycline-resistant transconjugants were selected and screened for sensitivity to UV light. One such UV-sensitive transconjugant, CS2942, was used in all subsequent studies.

The precise site of the mini-TnlacZ insertion in the chromosome of strain CS2942 was determined by analysis of the nucleotide sequence of the PCR product obtained by using the ′lacZ forward primer (5′ GTC ACG ACG TTG TA 3′) and the reverse primer (5′ GCG CTC ACC CAG CAG 3′). The forward primer was complementary to the coding sequence of ′lacZ at the left end of TnlacZ (19), and the reverse primer was complementary to the flanking nucleotide sequence immediately downstream of the o349 stop codon (23). Nucleotide sequencing of the PCR product revealed that the mini-TnlacZ was inserted into o349 between nucleotides 11,400 and 11,401 of the published wec gene cluster sequence (GenBank accession no. AE000454).

Wzz_LPS function is not required for the synthesis of Wzy-dependent O antigens. In contrast, the null mutation in o349 resulted in the apparent abrogation of ECA synthesis, as indicated by the lack of incorporation of N-acetyl-d-[1-³H]glucosamine into ECA polysaccharide chains in strain CS2942 (o349::mini-TnlacZ) (Fig. 2A, lane 2). In vivo incorporation of [³H]GlcNAc (6.6 Ci/mmol) into ECA and analysis of radioactive ECA by SDS-PAGE and fluorography were carried out as previously described (29). The fast-moving broad bands in each lane that migrate immediately ahead of ECA and at the bottom of the gel were not identified; however, they are believed to be radiolabeled lipid A and phospholipid, respectively. Subsequent experiments revealed that the lack of ECA synthesis by strain CS2942 was due to a polar effect of the insertion in o349 on required downstream genes. The wecB and wecC genes are located immediately downstream of o349 (Fig. 1B) (20), and these genes are required for the synthesis of UDP-ManNAcA, the donor of ManNAcA residues for ECA synthesis. In addition, o349 and the wecBC genes appear to be located in the same transcriptional unit (5). Thus, ECA synthesis was restored in strain PR4100, which was constructed by the introduction of plasmid pRL123 containing the wecBC genes and the rmlB gene into strain CS2942 (Fig. 2B). It should be noted that the rmlB_ECA and rmlA_ECA genes are located immediately downstream of the wecC gene, and these genes are required for synthesis of TDP-Fuc4NAc, the donor of Fuc4NAc residues for ECA synthesis (20). Although the rmlBA_ECA genes appear to be located in the same transcriptional unit as o349 and the wecBC genes (5), plasmid pRL123 contained only the rmlB_ECA gene (Fig. 3). Thus, it seems likely that synthesis of ECA in strain PR4100 was also dependent on the expression of the chromosomal rmlA allele contained in the wbb (formerly rfb) gene cluster of E. coli K-12 (27).

Restoration of ECA synthesis in strain PR4100 resulted in a random nonmodal distribution of ECA polysaccharide chain lengths (Fig. 2B). This distribution pattern is characteristic of the patterns observed for O-antigen synthesis in wzz mutants (36). In addition, the phenotype of PR4100 was not due to overexpression of the wecBC and rmlB genes in pRL123 since introduction of this plasmid into the wild-type parental strain, AB1133, did not result in altered ECA chain length regulation (data not shown). Wild-type modal chain length distribution was rescued in strain PR4130 (wzz_ECA::mini-TnlacZ) that contained plasmid pRL123 (wecBC rmlB) as well as plasmid pRL139 (o349) (Fig. 2C, lane 2). These results clearly support the conclusion that o349 of the wec gene cluster functions as a regulator of ECA polysaccharide chain length; accordingly, we have designated this open reading frame wzz_ECA.

A modal distribution of ECA polysaccharide chains was also observed in strain PR4129, which was constructed by the introduction of only plasmid pRL139 (wzz_ECA) into strain CS2942 (Fig. 2C, lane 1). However, the level of [³H]GlcNAc incorporation into ECA chains was significantly reduced in strain PR4129 in comparison to that observed in strain PR4130, which contained plasmid pRL139 (wzz_ECA) as well as plasmid pRL123 (wecBC rmlB). These observations suggest that wzz_ECA::mini-TnlacZ insertion does not completely abolish expression of genes located immediately downstream. Indeed, the apparent absence of radiolabeled chains in strain CS2942 most likely reflects a markedly reduced level of ECA synthesis that was not detected by SDS-PAGE due to the dispersal of the low level of incorporated radioactivity into a randomly distributed population of ECA chains (Fig. 2A, lane 2). However, complementation of the defect in wzz_ECA facilitated the detection of the radiolabeled chains by effectively concentrating the radioactivity into a smaller modal population of polymers. It is also important to note that overexpression of wzz_ECA in the wild-type parental strain had no effect on either the modality of ECA chains or the amount of incorporation of [³H]GlcNAc into ECA (data not shown). With the exception of the data shown for strains CS2942 (Fig. 2A, lane 2) and PR4129 (Fig. 2C, lane 1), the apparent differences in the amount of radiolabel incorporated into ECA chains primarily reflect variations in the amount of radiolabeled sample analyzed by SDS-PAGE and the length of time the gels were exposed to X-ray film during fluorography.

Wzz_ECA regulates the chain length of ECA_PG polysaccharides.

ECA_LPS occurs only in gram-negative enteric bacteria that are able to synthesize a complete R1, R4, or K-12 LPS core structure but that are unable to synthesize O side chains (14, 31, 34). In contrast, all members of the Enterobacteriaceae synthesize ECA_PG, and ECA_PG is the major form of ECA in those organisms capable of synthesizing both ECA_PG and ECA_LPS. Thus, it has been estimated that less than 5% of E. coli K-12 core-lipid A structures are substituted with ECA polysaccharide chains (14). Furthermore, a modal distribution of ECA polysaccharide chains has been observed in extracts obtained from organisms, such as Salmonella enterica serovar Typhimurium, that do not synthesize ECA_LPS (1).

Taken together, these observations indicate that Wzz_ECA is capable of modulating the chain length of ECA_PG polysaccharides. These observations also suggest that the data presented in this study primarily reflect the regulation of ECA_PG. In order to demonstrate this, the pattern of incorporation of [³H]GlcNAc into ECA polysaccharide chains was examined in strain CS2334 (waaL::TnphoA). The transfer of ECA chains from lipid-carrier to core-lipid A receptor requires the same O-translocase (O-ligase) required for translocation of O antigen from lipid-carrier to core-lipid A (14, 30, 31). This enzyme is encoded by the waaL gene, and mutations in this gene that preclude synthesis of a functional WaaL are unable to synthesize ECA_LPS (31). Radiolabeled GlcNAc was efficiently incorporated into ECA polysaccharide chains in strain CS2334 (Fig. 2D, lane 2), and these chains were distributed in the same modal pattern as were the ECA polysaccharide chains synthesized by wild-type cells (Fig. 2D, lane 1). These data support the conclusion that Wzz_ECA does indeed modulate the chain length of ECA_PG polysaccharide chains. Furthermore, the data also demonstrate that WaaL is not required for the synthesis of ECA_PG. In this regard, it is not known whether the chain length of ECA_LPS-linked polysaccharide chains is modulated by Wzz_ECA in a manner similar to that observed for ECA_PG polysaccharide chains, and experimental approaches to this problem have thus far been hampered by the inability to isolate mutants that are specifically defective in the synthesis of ECA_PG. Similarly, water-soluble ECA_CYC has been reported to contain four to six trisaccharide repeat units; however, it is not known whether the number of trisaccharide repeat units in this cyclic polymer is regulated in some manner.

The available data support the conclusion that Wzz O-antigen (Wzz_OAg) proteins are not O-antigen or species specific; however, modal values appear to be Wzz_OAg-dependent (3, 13). In contrast, the activity of Wzz_ECA appears to be restricted to the modulation of ECA polysaccharide chain length since the presence of a wild-type wzz_ECA gene in mutants that possess a null mutation in Wzz_OAg does not complement the defect in O-antigen chain length regulation.

The biochemical mechanism involved in the modulation of polysaccharide chain length is not known; however, two models have been proposed for the assembly of Wzy-dependent O antigens (3, 25). Both of these models suggest that the modulation of O-antigen chain length involves a complex interaction between Wzy_OAg, Wzz_OAg, WaaL, and lipid-carrier-linked O antigen; however, they differ with regard to the relative importance of interactions between Wzy_OAg and either Wzz_OAg or WaaL. The results of experiments presented in this study demonstrate that the Wzz_ECA-mediated modulation of ECA_PG polysaccharide chain length is not dependent on WaaL. The lack of a dependence on WaaL for the Wzz_ECA-mediated modulation of ECA_PG polysaccharide chain length may indicate that this regulation is primarily dependent on an interaction between Wzz_ECA and Wzy_ECA. Alternatively, modulation of ECA_PG polysaccharide chain length may involve specific interactions between Wzz_ECA, Wzy_ECA, and an as yet unrecognized protein that is the functional counterpart of WaaL. In this regard, the mechanism involved in the synthesis of the linkage between ECA polysaccharide chains and the phospholipid aglycone of ECA_PG has not yet been determined. It seems likely that this reaction involves either the transfer of polysaccharide or polysaccharide-1-phosphate from polysaccharide-pyrophosphorylundecaprenol to phosphatidic acid or diacylglycerol, respectively. In either case, it is possible that the enzyme that catalyzes this reaction functions in agreement with the proposed role of WaaL in the regulation of O-antigen chain length as suggested by Morona et al. (25).

ADDENDUM IN PROOF

Analysis of the hydropathy profile of Wzz_ECA, using the SOAP program of PC/GENE, revealed a large hydrophilic region flanked by two transmembrane regions, TM1 and TM2, located at the amino and carboxy terminus, respectively. This profile is characteristic of Wzz homologs found in other gram-negative bacteria (25). Wzz_ECA also contains the proline-rich sequence R₃₀₆X₄PX₂PX₄SPRX₂X₇IXGX₃GAG₃₃₉ in the carboxy-terminal region, which is in excellent agreement with the consensus sequence RX₄PX₂PX₄SPKX₂X₇IXGXMXGAG found in other Wzz homologs (A. Becker and A. Pühler, J. Bacteriol. 180:395–399, 1998). Residues predicted to be part of the membrane-spanning helix TM2 are italicized.