Assembly of the K40 Antigen in Escherichia coli: Identification of a Novel Enzyme Responsible for Addition of l-Serine Residues to the Glycan Backbone and Its Requirement for K40 Polymerization

Abstract

Escherichia coli O8:K40 coexpresses two distinct lipopolysaccharide (LPS) structures on its surface. The O8 polysaccharide is a mannose homopolymer with a trisaccharide repeat unit and is synthesized by an ABC-2 transport-dependent pathway. The K40_LPS backbone structure is composed of a trisaccharide repeating unit of N-acetylglucosamine (GlcNAc) and glucuronic acid (GlcA) and has an uncommon substitution, an l-serine moiety attached to glucuronic acid. The gene cluster responsible for synthesis of the K40 polysaccharide has previously been cloned and sequenced and was found to contain six open reading frames (ORFs) (P. A. Amor and C. Whitfield, Mol. Microbiol. 26:145–161, 1997). Here, we demonstrate that insertional inactivation of orf1 results in the accumulation of a semirough (SR)-K40_LPS form which retains reactivity with specific polyclonal serum in Western immunoblots. Structural and compositional analysis of the SR-K40_LPS reveals that it comprises a single K40 repeat unit attached to lipid A core. The lack of polymerization of the K40 polysaccharide indicates that orf1 encodes the K40 polymerase (Wzy) and that assembly of the K40 polysaccharide occurs via a Wzy-dependent pathway (in contrast to that of the O8 polysaccharide). Inactivation of orf3 also results in the accumulation of an SR-LPS form which fails to react with specific polyclonal K40 serum in Western immunoblots. Methylation linkage analysis and fast atom bombardment-mass spectrometry of this SR-LPS reveals that the biological repeat unit of the K40 polysaccharide is GlcNAc-GlcA-GlcNAc. Additionally, this structure lacks the l-serine substitution of GlcA. These results show that (i) orf3 encodes the enzyme responsible for the addition of the l-serine residue to the K40 backbone and (ii) substitution of individual K40 repeats with l-serine is essential for their recognition and polymerization into the K40 polysaccharide by Wzy.

Escherichia coli produces two major cell surface polysaccharides that play important roles in virulence. These are the lipopolysaccharide (LPS) O antigens and the capsular polysaccharides (K antigens) (19, 31, 46). The capacity to evade the host complement system can be attributed, in part, to variability in the structures of the O polysaccharide and also to the length of the individual O polysaccharides ligated to the lipid A core component of the LPS molecule (23). The capsular polysaccharides synthesized by E. coli are acidic in nature and often facilitate resistance to phagocytosis (46). Together the O and K polysaccharides allow the cell to evade and/or survive the host immune response during infection.

E. coli K antigens are divided into groups I, II, and III based primarily on (i) their mode of synthesis, (ii) their structural components, and (iii) their linkage to the cell surface. The thermoregulated group II K antigens are the best characterized, both genetically and biochemically, and are encoded by the kps gene cluster near serA (serine biosynthesis) (21, 32). Group III capsules resemble the group II capsules and also map near serA. However, unlike the group II capsules, group III capsules are not thermoregulated (28, 32). Whereas group II K antigens are coexpressed with a large array of O-antigen types when grown at temperatures above 18°C, group I capsules are coexpressed with the homopolymer O8, O9, O9a, and O20 O antigens at all growth temperatures. The chromosomal region encoding the group I biosynthetic gene cluster maps near the his (histidine biosynthesis) and gnd (glucose-6-phosphate-dehydrogenase) loci.

Group I K antigens are found in two distinct forms on the cell surface. The first is a high-molecular-weight (HMW) capsular form which is associated with the cell surface by an unknown mechanism. The second form is termed K_LPS and consists of K oligosaccharides covalently attached to the outer membrane via the lipid A core component of LPS (8, 13, 20, 25). The operational definitions both of a capsule and of an LPS molecule are satisfied by the group I K antigens. The ability of capsules to mask the shorter underlying O polysaccharide molecules in agglutination reactions is achieved by the HMW capsular form of group I K antigens.

Group I K antigens were previously divided into groups IA and IB (22, 46). Group IB K antigens contain amino sugars or amino acids as a component of their repeat structure, whereas group IA capsules do not. Although both group IA and IB K antigens form K_LPS, the K_LPS produced by group IA strains consists primarily of a single repeat unit attached to lipid A core. In contrast, group IB K_LPS is synthesized as longer polysaccharide chains attached to lipid A core which form typical O-antigen-substituted smooth LPS, as observed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western immunoblotting (2, 8, 13, 20, 25). The group IB K antigens are designated as capsules simply because they mask, in agglutination reactions, the neutral O8 or O9 antigen that is coexpressed on the same cell. Notably, there are examples where these strains lack the O8 or O9 antigen, and in these cases the K antigen is reclassified as an O antigen (19).

Synthesis of group IA and IB K antigens appears to occur via a Wzy-dependent pathway similar to that observed for the heteropolymer O antigens (2, 10). In Wzy-dependent pathways, individual O repeat units are assembled on undecaprenol carrier (und) at the cytoplasmic face of the cytoplasmic membrane (CM) and translocated to the periplasmic side of the CM. Wzy then polymerizes the individual units to form a complete polymer which is subsequently ligated to lipid A core and translocated to the surface of the cell (reviewed in references 44 and 45). Wzz (chain length regulator) regulates the extent of polymerization, by Wzy, of individual O-antigen repeats. This results in a serotype-specific chain length modality as observed by SDS-PAGE (44). Differences in K_LPS chain length observed in group IA and IB K antigens reflect the presence or absence of Wzz. In strains expressing a group IB K antigen, Wzz is present and long-chain K_LPS results. However, Wzz is absent in strains expressing a group IA antigen, which leads to an unregulated (nonmodal [45]) pattern of K_LPS carrying only a very short K antigen (2, 8, 13).

We have recently identified, characterized, and sequenced the gene clusters responsible for synthesis of group I K antigens from prototype E. coli strains producing both group IA (10) and group IB K antigens (2). The gene cluster encoding the enzymes required for the biosynthesis of the group IB K40 antigen is typical of those involved in the synthesis of heteropolysaccharide O antigen by Wzy-dependent pathways. The K40 cluster maps near his, at the same chromosomal location as that of the O-antigen (rfb) clusters from E. coli. Analysis of the chromosomal region downstream of the cloned K40 biosynthetic cluster (Fig. 1A) identified a wzz homolog whose gene product was subsequently shown to regulate the modality of the K40_LPS (2).

FIG. 1.

Organization and function of the K40 biosynthetic gene cluster responsible for the expression of the K40 antigen from E. coli 2775 (O8:K40) (2). (A) The chromosomal region from cps (colanic acid biosynthesis) through wzz (K40 chain length regulator) is shown. The six genes of the K40 region are essential for biosynthesis, as is the UDP-glucose dehydrogenase encoded by ugd. This enzyme makes the UDP-GlcA precursor. The plasmid inserts used for the functional analysis are shown below the K40 region. A, AccI; H3, HindIII; HII, HincII; K, KpnI; H, HpaII; N, NcoI; P, PstI; S, SacI; X, XbaI. For clarity, only those restriction sites important for cloning strategies are shown. (B) Structure of the K40 polysaccharide repeat unit (7). The ORF responsible for the addition of each residue is indicated at the respective linkage point. The l-serine residue is amide linked to position 6 of the GlcA residue. The initiating glycosyltransferase (WecA [2]) is located outside the K40 biosynthetic region.

The repeating unit structure of the K40 polysaccharide consists of a trisaccharide backbone carrying an uncommon novel substitution with l-serine amide linked to position 6 of the glucuronic acid (GlcA) residue (Fig. 1B). Substitution of polysaccharides with amino acids is not confined to this serotype of E. coli; others include the K49 and K54 polysaccharides (11). There are also limited examples of amino acid substitutions in the polysaccharides of Proteus penneri, Proteus mirabilis, and Haemophilus influenzae (4, 14, 29, 35, 40–42). Substitution of P. penneri O polysaccharide with l-threonine has been shown to provide the immunodominant epitope and precludes the generation of a strong capsular-polysaccharide-specific antibody (34).

The mechanism(s) by which such amino acid substitutions are added to polysaccharides is unknown. In this study, we identify a gene required for addition of l-serine to the K40 repeat unit and show that this residue is essential for the antigenicity of the K40 antigen and for polymerization of the K40 polysaccharide. The gene encoding the K40_Wzy (K40 polymerase) is identified, and its unusual specificity is defined.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

All bacterial strains and plasmids used in this study are described in Table 1. Bacteria were grown in Luria-Bertani (LB) broth (26) at 37°C. Where appropriate, LB broth was solidified by the addition of 15 g of agar (ICN Biochemicals) liter⁻¹. Antibiotics, when required, were added to final concentrations as indicated: ampicillin, 100 μg ml⁻¹; chloramphenicol, 34 μg ml⁻¹; kanamycin, 30 μg ml⁻¹; and gentamicin, 20 μg ml⁻¹. For expression of genes cloned in the vector pBAD18, l-arabinose was added to a final concentration of 0.02% (wt/vol).

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Genotype or description	Serotype	Source or reference
Strains
DH5α	K-12 f80d lacZΔM15 endA1 recA1 hsdR17(rK− mK−) supE44 thi-1 gyrA96 relA1 Δ(lacZYA-argF)U169 F−	O−:K−	33
2775	Prototroph	O8:K40	B. Jann
CWG291	2775 ugd::aacC1; Gmr	O8:K−	2
CWG294	2775 wzy::aphA-3; Kmr	O8:K40a	This study
CWG295	2775 orf3::aphA-3; Kmr	O8:K−b	This study
Plasmids
pMAK705	pUC19-pMAK700 hybrid suicide vector containing the temperature-sensitive pSC101 replicon; Cmr		16
pYA3265	A nonpolar aphA-3 gene cassette conferring kanamycin resistance; Kmr		A. Honeymoon via E. Vimr (5)
pBCSK+	pUC19-based phagemid cloning vector; Cmr		Stratagene
pBluescript II	pUC19-based phagemid cloning vector; Apr		Stratagene
pBAD18	Expression vector containing the PBAD promoter; Apr		15
pET30a(+)	T7-based expression vector; Kmr		Novagen
pWQ113	pBCSK(+) derivative containing the 9.4-kb HindIII-KpnI fragment carrying orf1.3, galF, the K40 biosynthetic cluster, and the 5′ end of gnd; Cmr		2
pWQ950	pBCSK(+) derivative containing the 1.8-kb HpaII fragment from pWQ113; carries the orf1 gene from the K40 biosynthetic cluster; Cmr		This study
pWQ951	pWQ950 with a SmaI-digested aphA-3 gene cassette inserted into the HincII site within orf1; Cmr Kmr		This study
pWQ952	pMAK705 derivative containing the 3.4-kb XhoI-SacI fragment from pWQ951; carries the orf1::aphA-3 insertion mutation cloned into SalI-BamHI-digested pMAK705; Cmr Kmr		This study
pWQ953	pBCSK(+) derivative containing the 2.6-kb BamHI-AccI fragment from pWQ113; carries the orf3 gene from the K40 biosynthetic cluster; Cmr		This study
pWQ954	pWQ953 with an aphA-3 gene cassette replacing a 300-bp PstI fragment within orf3; Cmr Gmr		This study
pWQ955	3.2-kb XbaI-KpnI fragment from pWQ954 containing the orf3::aphA-3 insertion mutation from the K40 biosynthetic cluster cloned into XbaI-KpnI-digested pBluescript II; Amr Kmr		This study
pWQ956	PMAK705 derivative containing the 3.2-kb XbaI-KpnI fragment from pWQ955; carries the orf3::aphA-3 insertion mutation from the K40 biosynthetic cluster, cloned into XbaI-KpnI-digested pMAK705; Cmr Kmr		This study
pWQ957	pET30(a) derivative containing a 1.6-kb PCR-amplified fragmentc from E. coli 2775; carries orf3 from the K40 biosynthetic cluster; engineered internal restriction sites NcoI and SacI were used for cloning; Kmr		This study
pWQ958	pBAD18 derivative containing the 1.7-kb XbaI-HindIII fragment from pWQ957; Apr		This study

^aThis strain produces an SR-K40_LPS containing a complete single repeat unit attached to lipid A core. 

^bThis strain produces an SR-K40_LPS containing the trisaccharide backbone of the K40 repeat unit but lacks the l-serine substitution on GlcA. 

^cPrimer sequences used for amplification are given in the text. 

PCR amplification and sequencing of chromosomal DNA.

Oligonucleotide primers were synthesized by using a Perkin-Elmer 394 DNA synthesizer. Amplification of chromosomal DNA from E. coli 2775 (O8:K40) was performed by using a Perkin-Elmer GeneAmp PCR system 2400 thermocycler under optimal conditions. Chromosomal DNA was amplified by using PwoI DNA polymerase (Boehringer Mannheim) and was subsequently purified by passage through QIAquick PCR purification columns (Qiagen). Sequencing of PCR-amplified products and cloned PCR products ensured that they were error free. Sequencing was carried out by using an ABI 377 DNA sequencing apparatus (Perkin-Elmer) at the Guelph Molecular Supercentre (University of Guelph, Guelph, Ontario, Canada).

Generation of chromosomal insertion mutations.

Individual genes were inactivated through insertion of a nonpolar aphA-3 (kanamycin resistance) cassette into the target open reading frame (ORF). The mutated gene was delivered to the chromosome via homologous recombination using a previously published procedure (2, 9). Briefly, plasmid pMAK705 contains a temperature-sensitive origin of replication which is functional at 30°C but not at 44°C (16). Selectable replication of the plasmid allows for the isolation of homologous recombination events, where the plasmid has integrated into the host chromosome. Growth at the permissive temperature (30°C) allows for a second homologous recombination event accompanied by resolution of the plasmid. Allelic exchange is detected by screening of colonies for kanamycin resistance and chloramphenicol sensitivity.

To construct the wzy::aphA-3 mutant, a 1.8-kbp HpaII fragment containing orf1 and flanking DNA was subcloned from pWQ113 into the AccI site of pBCSK(+), generating pWQ950. The SmaI-digested aphA-3 gene was ligated into the internal HincII site in orf1, and a derivative with the aphA-3 cassette in the same orientation as orf1 was selected (pWQ951). The mutated orf1::aphA-3 and flanking DNA was removed from pWQ951 as a 3.4-kbp XhoI-SalI fragment and ligated into SalI-BamHI-digested pMAK705. The resulting suicide delivery plasmid, pWQ952, was transferred to E. coli 2775 via electroporation (3), and the double-crossover event was selected to yield E. coli CWG294. The insertion site of the Km^r cassette in the chromosome in E. coli CWG294 was confirmed through (i) Southern hybridization analysis of chromosomal DNA digests and (ii) PCR amplification of the mutated region from the chromosome, with subsequent sequencing of the cassette junctions in the amplified PCR product (data not shown). The nonpolar nature of the orf1::aphA-3 mutation was confirmed through complementation with plasmid pWQ950 to restore the wild-type K40 phenotype.

To generate the orf3::aphA-3 mutant, a 2.6-kbp BamHI-AccI fragment containing orf3 and flanking DNA was ligated into similarly digested pBCSK(+), giving pWQ953. Plasmid pWQ953 was digested with PstI. The PstI-digested aphA-3 gene cassette was used to replace the 300-bp PstI fragment. A derivative with the aphA-3 cassette in the same orientation as orf3 was selected (pWQ954). The mutated orf3::aphA-3 gene and flanking DNA was removed from pWQ954 on a 3.2-kbp XbaI-KpnI fragment and ligated into similarly digested pMAK705, giving pWQ956. The suicide delivery vector pWQ956 was used to transform E. coli 2775, and a double-crossover event was selected, giving E. coli CWG295. The insertion site of the Km^r cassette on the chromosome in E. coli CWG295 was confirmed as described above. The nonpolar nature of the CWG295 orf3::aphA-3 mutation was confirmed by complementation to give the wild-type K40 phenotype by using pWQ958 containing orf3.

To clone orf3, the coding sequence was amplified by PCR and was subsequently cloned into the arabinose-inducible expression vector pBAD18, by using primer-encoded restriction sites (SacI and NcoI) (15). Briefly, PCR amplification of chromosomal DNA from E. coli 2775 using primers PAA130 (5′-TTGAACCATGGCTTTTGTAACAATAAATACAG-3′) and PAA103B (5′-GCATTTCCTTTTTCTGACACAGAGCTC-3′) resulted in a 1.6-kbp product. This product was digested with SacI and NcoI, by using unique sites (underlined) in the primers, and was then ligated into the same restriction sites of pET30(a), generating pWQ957. This cloning resulted in the addition of an in-frame N-terminal His₆ tag to orf3. The 1.7-kbp XbaI-HindIII fragment encoding the His₆-tagged Orf3 was purified and ligated into similarly digested pBAD18, producing pWQ958. This plasmid contains orf3 behind an optimal E. coli ribosome binding site under the control of the inducible arabinose promoter (P_BAD). The nonpolar nature of the orf3::aphA-3 mutation was confirmed through complementation with pWQ958. Restoration of the wild-type phenotype indicated that genes downstream of the Km cassette insertion in orf3 remained functional.

Overexpression and purification of Orf3.

The complete orf3 was cloned into the arabinose-inducible expression vector pBAD18, producing pWQ958. E. coli DH5α containing pWQ958 was used for protein expression. To isolate the Orf3 product, this strain was grown overnight at 37°C in LB broth supplemented with ampicillin. The overnight culture was diluted 1:50 into prewarmed LB broth supplemented with ampicillin and was grown until the culture reached an optical density at 600 nm of 0.2. This culture was induced by the addition of arabinose to a final concentration of 0.02% and was allowed to grow for 3 h at 37°C with shaking. The cells were collected as a pellet and washed twice in phosphate-buffered saline. The cells were suspended in a buffer containing 300 mM NaCl and 10 mM imidazole in 50 mM Tris (pH 8.0) and disrupted by sonication. The cell debris was removed by centrifugation at 10,000 × g for 30 min at 4°C. The His₆-tagged derivative of Orf3 was purified by Ni-nitrilotriacetic acid (NTA) affinity chromatography (Qiagen) according to the manufacturer’s recommendations. This resulted in an Orf3 preparation that was approximately 95% pure. This preparation was further purified by fast protein liquid chromatography (FPLC) on a Superdex 75 size fractionation column eluted in 10 mM ammonium acetate, pH 7.0, with a flow rate of 0.7 ml min⁻¹. The elution profile was monitored at A₂₈₀, and fractions showing increased absorbance were kept for further analysis (data not shown). Peak fractions were lyophilized and resuspended in 50 μl of 10 mM NH₂PO₄, pH 7.0. Individual fractions were routinely monitored for the presence of the Orf3 protein by SDS-PAGE and were visualized by Coomassie brilliant blue staining.

SDS-PAGE analysis of cell surface polysaccharides.

LPS was isolated from exponentially grown cells in liquid culture by using SDS-proteinase K-digested whole-cell lysates according to the method of Hitchcock and Brown (18). The LPS samples were analyzed by SDS-PAGE on commercially prepared 10 to 20% Tricine gels according to the manufacturer’s instructions (Novex, San Diego, Calif.). The PAGE gels were either silver stained (39) or electrophoretically transferred to a Bio-Trace NT membrane (Gelman Sciences) for Western immunoblot analysis as described elsewhere (27, 38). Anti-O8 and anti-K40 polyclonal sera were prepared as described previously (2, 8).

Isolation of the K40 polysaccharides.

The water-soluble fraction of the LPS population was extracted, by the hot-water–phenol method of Westphal and Jann (43), from E. coli CWG294 and CWG295. The lipid-A component of the LPS molecule was released by hydrolysis in 2% acetic acid at 100°C, which cleaves the acid-labile ketosidic linkage between 3-deoxy-d-manno-oct-2-ulosonic acid and lipid A. The water-insoluble lipid-A component of the LPS was removed through centrifugation. The remaining supernatant, containing the K40 polysaccharide and attached core oligosaccharide, was applied to a BioGel P-2 column (1 m by 1 cm) and was eluted with water. The fractions were collected and lyophilized, and those containing the core oligosaccharide plus a single K40 repeat unit were further analyzed.

Structural analysis.

The amide and glycosidic bonds in the fractionated K40_LPS from E. coli CWG294 and E. coli CWG295 were hydrolyzed by incubation in 4 M hydrochloric acid at 100°C for 4 h in vacuo. The acid was removed by distillation in vacuo, and the sample was redissolved in distilled water. The hydrolysate was examined by using a Beckman System Gold amino acid analyzer, by ninhydrin detection. Elution profiles were generated by monitoring the absorbance at 570 nm. Standard mixtures containing all amino acids and N-acetyl-glucosamine (GlcNAc) were used to facilitate identification of all peaks resulting from the K40_LPS samples.

The methylation linkage analyses were carried out by the NaOH-dimethyl sulfoxide-methyl iodide method of Ciucanu and Kerek (6). The permethylated alditol acetate derivatives of the fractionated K40_LPS, isolated from E. coli CWG294 and E. coli CWG295, were fully characterized by gas-liquid chromatography–mass spectrometry in the electron impact mode using a column of DB-17 operated isothermally at 190°C for 60 min. Positive-ion fast atom bombardment-mass spectrometry (FAB-MS) was performed on a fraction of the methylated sample by using a Joel JMS-AX505H mass spectrometer with glycerol-thioglycerol as the matrix and a tip voltage of 3 kV.

¹H nuclear magnetic resonance (NMR) spectra of the core oligosaccharide plus the single K40 repeat unit from E. coli CWG294 and E. coli CWG295 were recorded on a Bruker AMX 500 spectrometer at 300 K with standard Bruker software. Before the experiments were performed, the samples were lyophilized three times with D₂O (99.9%). The internal reference for ¹H NMR was the HOD peak (δ_H 4.786).

RESULTS AND DISCUSSION

Deletion of orf1 results in the loss of K40_LPS polymerization.

The physical map of the K40_LPS biosynthesis cluster and the positions of relevant plasmid inserts are shown in Fig. 1. Synthesis of polysaccharides by Wzy-dependent systems involves formation of individual repeat units on und-phosphate (und-P) at the cytoplasmic face of the CM. Wzx translocates the units across the CM, where Wzy polymerizes them at the reducing end of the growing polysaccharide. Wzy and Wzx homologues are highly hydrophobic proteins that are predicted to contain multiple membrane-spanning domains (44, 45). Based on the hydropathy profiles of the predicted products and on minor sequence similarities, Orf1 and Orf2 were previously tentatively assigned as Wzy (the K40 polymerase) and Wzx (the K40 translocase) respectively (2).

In order to confirm the identification of orf1 as the K40 polymerase gene, a nonpolar chromosomal orf1::aphA-3 mutation was made in E. coli 2775 (O8:K40), producing E. coli CWG294 (orf1::aphA-3). The polysaccharide antigens from whole-cell lysates of E. coli CWG294 were analyzed by SDS-PAGE (Fig. 2). In the wild-type strain (2775), the K40_LPS ladder stains poorly with silver due to its composition and is best visualized through Western immunoblots reacted with specific anti-K40 polyclonal sera (Fig. 2A and B, lanes 1). Silver-stained gels do show some LPS molecules with shorter K40 oligosaccharides, and these react with anti-K40 serum. In E. coli CWG294, the orf1::aphA-3 mutation eliminates all of the higher-molecular-weight K40-immunoreactive material, leaving only a single LPS band that is clearly evident in silver-stained samples (Fig. 2A, lane 3). This fraction contains the K40 antigen (Fig. 2B, lane 3) but does react with anti-O8 antisera (Fig. 2C, lane 3). The migration of this material is consistent with semirough (SR)-LPS, molecules containing a single repeat unit of K40 antigen, and this was confirmed by structural analysis (see below). Complementation of the orf1::aphA-3 mutation in E. coli CWG294 with plasmid pWQ950 restores the complete K40_LPS ladder (Fig. 2A and B, lanes 4), indicating that the K40 defect is not due to polarity effects from the mutation.

FIG. 2.

Silver-stained SDS-PAGE gel and corresponding Western immunoblots of various E. coli strains expressing the K40 antigen. (A) Silver-stained SDS-PAGE gel; (B) Western immunoblot reacted with polyclonal anti-K40 serum; (C) Western immunoblot reacted with polyclonal anti-O8 serum. Strains are identified above each lane. A chromosomal wzy::aphA-3 insertion in E. coli CWG294 eliminates polymerization of K40_LPS, and only a single K40 repeat unit (SR-K40_LPS) is ligated to lipid A core. E. coli CWG295 contains a chromosomal orf3::aphA-3 insertion, which results in the loss of polymerized K40 polysaccharide. This strain also accumulates SR-K40_LPS, but, unlike the LPS from E. coli CWG294, it does not react with polyclonal anti-K40 serum. The location of the SR-K40_LPS is indicated by the arrow. The mutations did not influence the LPS-linked O8 antigen (C). E. coli CWG291 contains a chromosomal ugd::aacC1 mutation as a K40-negative control. Ugd is responsible for synthesis of the UDP-GlcA precursor required for the K40 antigen.

The accumulation of SR-K40_LPS is the phenotype expected for a wzy mutation, as the cell would still have all enzymes required for assembly of the repeat unit, ligation to lipid A core, and translocation to the cell surface. However, wzy mutants lack the polymerase activity required to form an extended polysaccharide chain. Together, the results of these experiments show that the K40 polymerase (Wzy_K40) is encoded by orf1. Assembly of polysaccharides via the Wzy-dependent pathway also requires a second highly hydrophobic protein (Wzx), with multiple membrane-spanning domains, for translocation of individual repeat units across the CM (24). The protein predicted to be encoded by orf2 meets these criteria. With the identification of orf1 as the K40 polymerase gene, orf2 presumably encodes Wzx, the K40 translocase protein (Wzx_K40).

Mutations and manipulations of wzy have no influence on O8-substituted LPS, as evident in both the silver-stained gels (Fig. 2A) and western immunoblots (Fig. 2C). The O8_LPS is synthesized via a fundamentally different, Wzy-independent mechanism and requires an ATP-binding-cassette (ABC) transporter for its export across the plasma membrane (12, 37, 44).

Inactivation of orf3 eliminates polymerization and alters synthesis of the K40_LPS.

The repeat unit structure of the K40 polysaccharide contains the uncommon side branch substitution of GlcA with l-serine (Fig. 1B). Similar amino acid substitutions of LPS and capsular polysaccharides from Proteus have been shown to provide the primary epitope recognized by polyclonal serum in immunological studies (29). However, to date the enzymes responsible for such substitutions have not been identified.

With the analysis above, all genes required for the synthesis and assembly of the K40 polysaccharide backbone have been identified (2) (Fig. 1). Orfs 4, 5, and 6 were previously defined as serotype-specific glycoslytransferases, required for the assembly of the K40 backbone (2). The only remaining gene in the sequenced K40 cluster with no assigned function is orf3. Sequence database searches identified no homologies to orf3 at either the nucleotide or the amino acid level. To investigate the function of the orf3 gene products, E. coli CWG295 (orf3::aphA-3) was constructed. This mutation has no effect on the O8-substituted LPS (Fig. 2A and C, lanes 5). Silver-stained gels show the accumulation of a band with a migration similar to that of the SR-K40_LPS (similar to the orf1 mutant), but this fraction was distinguished by its lack of immunoreactivity with K40-specific antisera (Fig. 2B, lane 5). The effect of the orf3 mutation is therefore not explained by a simple wzy-like polymerization defect.

One interpretation of these results is that the orf3 mutant makes K40_LPS with an intact carbohydrate backbone, but the structure is missing one component. The l-serine residues provided a likely candidate. Consistent with such a minor structural difference, slight differences in migration of the SR-LPS fraction were evident, with the molecules in the orf1 mutant showing a slightly slower migration than those from the orf3 mutant (data not shown). Structural analysis confirmed these predictions (see below).

The altered SR-LPS phenotype produced by E. coli CWG295 shows that the mutation causes an effect on both the structure and the polymerization of the K40 antigen. Complementation experiments were carried out to clearly demonstrate that the effects of the orf3 mutation on LPS polymerization and structure were due to a single mutation and that the nucleotide sequence for orf3 correctly predicted a single ORF in this region. The orf3::aphA-3 mutation in CWG295 was complemented with plasmid pWQ958, to restore a wild-type profile of the immunoreactive K40_LPS ladder (Fig. 2A and B, lanes 6). Plasmid pWQ958 contains only the orf3 region cloned in pBAD18, under the control of the arabinose-inducible P_BAD promoter. The orf3 coding region is cloned so that the product contains an N-terminal His₆ tag. The orf3 gene product was barely detectable in whole-cell lysates, regardless of the attempts to optimize the induction conditions. The His₆-tagged Orf3 protein was therefore purified from a cell-free lysate of DH5α(pWQ958) by using the QIAexpress Detection System (Qiagen) and an FPLC step. The resulting purified fraction was homogeneous in Coomassie blue-stained SDS-PAGE gels and showed a single polypeptide with a molecular mass of 62 kDa (Fig. 3). This is the size predicted from the sequence data. Together these results confirm the expression of the His₆-tagged Orf3 protein and the fact that the complementation of the orf3::aphA-3 mutation in E. coli CWG295 was due to its expression.

FIG. 3.

SDS-PAGE analysis monitoring the purification of the orf3 gene product. The protein preparations were separated by SDS-PAGE and then stained with Coomassie brilliant blue. Lane 1, protein molecular weight standards. Lane 2, cell-free lysate of E. coli DH5α(pWQ958). Plasmid pWQ958 carries the orf3 gene, which is under the control of the arabinose-inducible P_BAD promoter. The cell-free lysate was passed through a Ni-NTA column (Qiagen), and the eluant containing the partially purified His₆⁺-tagged Orf3 protein is shown in lane 3. The partially purified Orf3 protein from lane 3 was further purified by FPLC (lane 4) (arrow). The purified Orf3 protein was approximately 99% pure when observed by these methods and was shown to have the expected molecular weight of 62 kDa.

Serine content in the SR-LPS molecules in E. coli CWG294 and CWG295.

The structures of the SR-LPS molecules from E. coli CWG294 and CWG295 were compared by compositional analysis. Purified SR-LPS samples were hydrolyzed in HCl, releasing the individual residues of the polysaccharide structure. These components were then separated by anion-exchange high-performance liquid chromatography, and any free amino groups were reacted with ninhydrin. These analyses would be expected to identify the GlcNAc residues of the carbohydrate backbones in the K40 repeat units (7), as well as l-serine residues (as determined by reference standards). GlcNAc is present in the SR-LPS samples from both E. coli CWG294 and CWG295 (Fig. 4). As expected, a second peak, containing an amino group, is present in the E. coli CWG294 (wzy::aphA-3) SR-K40_LPS sample. This peak corresponds to serine and is absent in the SR-LPS fraction of E. coli CWG295. The presence of serine in E. coli CWG294 SR-K40_LPS was confirmed by ¹H NMR. The CH₂OD group of serine has previously been reported to give a signal at δ 3.88 (4, 7). This signal was used as a diagnostic for the presence or absence of l-serine in the LPS samples. A signal at δ 3.9 was evident in the wild-type strain (2775 [7]) and the orf1 mutant (CWG294) but not in the orf3 mutant, E. coli CWG295 (data not shown). Together the compositional data prove that serine is lacking in the SR-LPS produced in E. coli CWG295 (orf3::aphA-3) and establish a requirement for Orf3 in the addition of l-serine to the K40 backbone.

FIG. 4.

Compositional analysis of the LPS isolated from E. coli CWG294 (orf1::aphA-3) (A) and E. coli CWG295 (orf3::aphA-3) (B). Both of these mutants produce a truncated LPS containing a single repeat unit attached to lipid A core (SR-K40_LPS). The polysaccharides were hydrolyzed and separated by Superdex 75 size fractionation chromatography. In both structures a peak corresponding to GlcNAc is present at approximately 31 min. Panel A also shows a peak at approximately 9.6 min, which corresponds to the l-serine residue attached to GlcA, prior to hydrolysis. The SR-K40_LPS in panel B contains no l-serine, indicating that the orf3 mutation eliminates the addition of l-serine to the SR-K40_LPS.

Determination of the carbohydrate backbone sequence of the SR-LPS molecules in E. coli CWG295.

To establish that the orf3 mutation influenced only the l-serine residue, the structure of the carbohydrate backbone in the SR-LPS from E. coli CWG295 was determined by using the published K40 structure as a guide (7). The terminal sugar of the repeat (at the nonreducing end) and the order of the repeat unit were established in FAB-MS experiments (Fig. 5), in conjunction with methylation linkage analysis. The methylation data (not shown) confirmed that the carbohydrate backbone structure was the same as that already published (7). The SR-LPS fraction in this bacterium contains an R1 core oligosaccharide, for which detailed methylation data are available (17, 47). The FAB-MS spectrum of the methylated SR LPS from E. coli CWG295 identified the following primary glycosyl oxonium ions: m/z 260 (GlcNAc)⁺, m/z 478 (GlcNAc-GlcA)⁺, m/z 723 (GlcNAc-GlcA-GlcNAc)⁺, and m/z 927 (GlcNAc-GlcA-GlcNAc-Hex)⁺ (Fig. 5). The secondary ions at m/z 228, m/z 478, and m/z 691 are derived from the primary ions at m/z 260, m/z 478, and m/z 723, respectively, via the β-elimination of methanol. These data identify the biological repeat unit structure of the K40 polysaccharide backbone as -6)-α-d-GlcNAc-(1→4)-β-d-GlcA-(1→4)-α-d-GlcNAc-(1-. The m/z 927 ion reflects the attachment of the modified K40 repeat unit to the β-glucosyl side branch in the R1 core oligosaccharide (17).

FIG. 5.

FAB-MS was used to determine the sequence of sugars in the repeating unit of the K40 polysaccharide from E. coli CWG294. The peaks at m/z 260.1, m/z 478.2 and m/z 723.3 correspond to the primary oxonium ions as shown. (Note that all hydroxyl and amine groups were methylated prior to the FAB-MS analysis.) The peaks at m/z 228.1, m/z 446.2, and m/z 691.3 are secondary ions arising from the β-elimination of methanol (subtracting 32 from each of the primary ion peaks for the loss of CH₃OH fragmented structures of GlcNAc, GlcNAc-GlcA and GlcNAc-GlcA-GlcNAc, respectively). The additional peak at m/z 927.4 corresponds to the α-d-GlcNAc-(1→4)-β-d-GlcA-(1→4)-α-d-GlcNAc-(1→3)-β-d-Glc⁺ oxonium ion. The Glc residue is the attachment point of the K40 repeat unit to the R1 core oligosaccharide. Therefore the repeat unit of the K40 polysaccharide backbone is -6)-α-d-GlcNAc-(1→4)-β-d-GlcA-(1→4)-α-d-GlcNAc-(1-.

Conclusions.

The data presented here indicate that the product of orf3 is the serine transferase. The loss of antigenicity in the SR-LPS of E. coli CWG295 indicates that the l-serine moiety provides the immunodominant serospecific epitope recognized by polyclonal serum.

The addition of the l-serine residue is clearly required for polymerization of the K40 polysaccharide. This result is quite surprising given that l-serine is a substitution on the middle sugar of the biological repeat, as established here by the FAB-MS-derived structure of the SR-LPS (Fig. 1B). Initiation of the K40 repeat unit synthesis occurs by the addition of GlcNAc-1-P (from UDP-GlcNAc) to the carrier lipid und-P, producing und-P-P-GlcNAc. The enzyme responsible for this addition is encoded by wecA (2), which is the initiator for the synthesis of other O antigens (1, 2, 30) and enterobacterial common antigen (30). The und-P-P-GlcNAc intermediate then forms the substrate for the addition of the two remaining sugars, GlcA and GlcNAc. The chemical structure of the K40 antigen previously determined by Dengler et al. (7) does not predict the sequence of glycosyltransferase reactions, but this “biological” repeat structure is now defined by the data presented here. These results indicate that the l-serine is added prior to the polymerization of the K40 repeat unit and before it is ligated to lipid A core. Furthermore, the terminal GlcNAc in the repeat unit can be transferred to the und-P-P-GlcNAc-GlcA by Orf5-Orf6 without the prior addition of l-serine. The most likely interpretation is that addition of the l-serine residue to the K40 repeat unit occurs in the cytoplasm. The nature of the precursor for this reaction is currently under investigation.

It is well established that Wzy enzymes are specific for a given repeat unit structure (44), and it is further believed that Wzy recognizes the terminal residue of the repeat unit structure. In support of this, polymerization of the Salmonella typhimurium O5 antigen is not influenced by modifications such as O-acetylation of the repeat unit (36). These results are in contrast to the unique specificity of the Wzy enzyme for the K40 system. In this system, the Wzy activity is highly dependent on the penultimate GlcA–l-serine component of the repeat unit. One key element in this specificity may be the influence of the l-serine in altering the distribution of the negative charge contributed by the carboxyl group on GlcA.

Our results clearly show that lack of K40 polymerization in the Wzy-dependent systems can occur from various mutations, and therefore caution is required in interpreting SDS-PAGE gels showing SR-LPS. The mutations shown here are indistinguishable by this technique, and more rigorous structural analyses are required to confirm such mutations in related systems.