Effects of Lipopolysaccharide Biosynthesis Mutations on K1 Polysaccharide Association with the Escherichia coli Cell Surface

Abstract

The presence of cell-bound K1 capsule and K1 polysaccharide in culture supernatants was determined in a series of in-frame nonpolar core biosynthetic mutants from Escherichia coli KT1094 (K1, R1 core lipopolysaccharide [LPS] type) for which the major core oligosaccharide structures were determined. Cell-bound K1 capsule was absent from mutants devoid of phosphoryl modifications on l-glycero-d-manno-heptose residues (HepI and HepII) of the inner-core LPS and reduced in mutants devoid of phosphoryl modification on HepII or devoid of HepIII. In contrast, in all of the mutants, K1 polysaccharide was found in culture supernatants. These results were confirmed by using a mutant with a deletion spanning from the hldD to waaQ genes of the waa gene cluster to which individual genes were reintroduced. A nuclear magnetic resonance (NMR) analysis of core LPS from HepIII-deficient mutants showed an alteration in the pattern of phosphoryl modifications. A cell extract containing both K1 capsule polysaccharide and LPS obtained from an O-antigen-deficient mutant could be resolved into K1 polysaccharide and core LPS by column chromatography only when EDTA and deoxycholate (DOC) buffer were used. These results suggest that the K1 polysaccharide remains cell associated by ionically interacting with the phosphate-negative charges of the core LPS.

INTRODUCTION

Escherichia coli K1 is a facultative pathogen responsible for severe extraintestinal diseases such as sepsis, meningitis, cystitis, pyelonephritis, cellulitis, pneumonia, and postoperative infections. The production of a polysialic acid K1 capsule, a homopolymer of 5-acetamido-3,5-dideoxy-d-glycero-d-galacto-non-2-ulosonic or N-acetylneuraminic acid (Neu5Ac) residues connected by α2–8 linkage (24) with a phase-variable O-acetylation at position 7 or 9, is the common pathogenic feature of these strains (29). The immunological tolerance of the K1 capsule by molecular mimicry to the host's own polysialic antigen (oncofetal modification of the neural cell adhesion molecule [NCAM]) impedes the development of effective vaccines to prevent diseases caused by the K1 strain.

On the basis of capsule biosynthesis and assembly features, the E. coli capsules have been classified into four groups. The K1 capsule is, together with the K5 capsule, the model system for group 2 E. coli capsules (36). As in other members of group 2, the genes directing the biosynthesis of K1 capsule are organized in three distinct functional regions: region 1, kpsFEDUCS; region 2, neuDBACES, and region 3, kpsMT. Region 2 genes appear to be K-serotype specific and are involved in K1 biosynthesis while regions 1 and 3 appear to be involved in K1 transport and are conserved among members of the E. coli group 2.

Currently, it is thought that the first committed step in the biosynthesis of the K1 polysaccharide (K1-PS) is catalyzed by the epimerase NeuC leading to the formation of N-acetylmannosamine (ManNAc) from UDP-N-acetylglucosamine (UDP-GlcNAc), followed by the synthesis of CMP-N-acetylneuraminic (CMP-Neu5Ac) acid from ManNAc through the action of Neu5Ac synthase NeuB and CMP-Neu5Ac synthetase NeuA. The processive polysialyltransferase NeuS catalyzes the formation of a homopolimer of Neu5Ac using CMP-Neu5Ac as a residue donor. It is also hypothesized that the K1-PS is exported to the outer membrane through a transmembrane protein complex including ATP-binding cassette (ABC)-like transporter proteins KpsM and KpsT, polysaccharide copolymerase protein KpsE, and outer membrane polysaccharide export protein KpsD (reviewed in reference 30). Finally, the group 2 E. coli capsule polysaccharide is thought to be attached at its reducing end to undecaprenyl phosphate (27), a phospholipid, or to a residue of 3-deoxy-d-manno-oct-2-ulosonic acid (Kdo) located between the lipid and the reducing terminal Neu5Ac residue (7, 9), and the lipid domain has been postulated to anchor these capsules to the outer membrane. Nevertheless, only a fraction (20 to 50%) of the isolated capsule polysaccharides has the lipid substitution, opening the possibility that the nonlipidated fraction could be retained at the cell surface via ionic interactions (36).

In E. coli five different core lipopolysaccharide (LPS) structures differing mainly in their outer-core structure have been recognized (K-12, R1, R2, R3, and R4) (14, 22). Studies based on reactions with specific core LPS polyclonal and monoclonal antibodies (4, 12) or on DNA amplification of core-specific diagnostic genes (3) revealed that the R1-type core LPS is the most frequent among E. coli strains, accounting for about 68% of the studied strains (3). In addition, the R1 core type was found to be the most prevalent one in bacteremia (68%) (1) and septicemia (60%) (12) and to constitute 100% of the strains of the group B2 of the E. coli reference (ECOR) collection. The group B2 isolates contain known virulence genes for strains causing extraintestinal infections, like genes involved in group 2 capsule biosynthesis.

In a previous work, we have shown that the Klebsiella pneumoniae K2 capsule, which shares common features with group 1 E. coli capsules, establishes ionic interactions with core LPS negative charges (10) provided by two galacturonic acid residues present in the two known K. pneumoniae core LPS types (23, 33). Thus, we decided to investigate if a similar interaction also occurs in the case of E. coli K1 capsule. In this study, we used a K1-capsulated E. coli strain with core LPS type R1 to construct a series of nonpolar core mutants to analyze the effect of these mutations on K1 antigen interaction or association to the cell surface.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

Bacterial strains (Table 1) were grown in Luria-Bertani (LB) broth and on LB agar at 37°C (19). LB medium was supplemented with ampicillin (100 μg ml⁻¹), chloramphenicol (15 μg ml⁻¹), and kanamycin (50 μg ml⁻¹) when needed to maintain the different plasmids used (Table 1).

Table 1.

Bacterial strains, plasmids, and bacteriophages used

Strain, plasmid, or phage	Relevant characteristic(s)	Reference or source
E. coli strains
W3310	Wild-type, core K-12
JC7623	recB21 recC22 sbcB15	27
KT1094	K1:O1:H7, LPS core type R1	J. Blancoa
KT1094 ΔneuA	Nonpolar neuA mutant	This study
KT1094 ΔwaaC	Nonpolar waaC mutant	This study
KT1094 ΔwaaF	Nonpolar waaF mutant	This study
KT1094 ΔwaaQ	Nonpolar waaQ mutant	This study
KT1094 ΔwaaP	Nonpolar waaP mutant	This study
KT1094 ΔwaaY	Nonpolar waaY mutant	This study
KT1094 ΔwaaG	Nonpolar waaG mutant	This study
KT1094 ΔwaaO	Nonpolar waaO mutant	This study
KT1094 ΔwaaL	Nonpolar waaL mutant	This study
KT1094 Δwaa-Km	hldD-waaQ deletion mutant	This study
Plasmids
pKO3	Cmr sacB temperature-sensitive replication suicide vector	17
pKO3ΔneuA	pKO3 with engineered neuA deletion from strain KT1094	This study
pKO3ΔwaaC	pKO3 with engineered waaC deletion from strain KT1094	This study
pKO3ΔwaaF	pKO3 with engineered waaF deletion from strain KT1094	This study
pKO3ΔwaaQ	pKO3 with engineered waaQ deletion from strain KT1094	This study
pKO3ΔwaaP	pKO3 with engineered waaP deletion from strain KT1094	This study
pKO3ΔwaaY	pKO3 with engineered waaY deletion from strain KT1094	This study
pKO3ΔwaaG	pKO3 with engineered waaG deletion from strain KT1094	This study
pKO3ΔwaaO	pKO3 with engineered waaO deletion from strain KT1094	This study
pKO3ΔwaaL	pKO3 with engineered waaL deletion from strain KT1094	This study
pWSB10	Contains the 5′ end of yibB and the gmhD, waaF, waaC, and waaL genes	6
pFFM991	pGEMT with a 1,554-bp insert from pWSB10 the 5′ ends of yibD and gmhD genes	This study
pJSC2	Contains the 5′ end of waaQ, the waaA and coaD genes, and the 3′ end of fgp	5
pWSKA	Contains the 5′ end of waaA and promoter region derived from pJSC2	This study
pGA	containing the 5′ end of yibD and gmhD fused to the 5′ end of waaA and the promoter region	This study
pGKMA	pGA with a kanamycin resistance cassette from pUC4K between the gmhD gene and the 5′ waaA promoter region	This study
pBAD33	Arabinose-inducible expression vector, Cmr	13
pBAD33-waaC	Arabinose-inducible waaC	This study
pBAD33-waaCF	Arabinose-inducible waaCF	This study
pBAD33-waaCFQ	Arabinose-inducible waaCFQ	This study
pBAD33-waaCFQG	Arabinose-inducible waaCFQG	This study
pBAD33-waaCFQGP	Arabinose-inducible waaCFQGP	This study
pGEMT	PCR-generated DNA fragment cloning vector, Ampr	Promega
pGEMT-hldD	pGEMT with hldD from strain KT1094	This study
pGEMT-neuA	pGEMT with neuA from strain KT1094	This study
pGEMT-waaC	pGEMT with waaC from strain KT1094	This study
pGEMT-waaF	pGEMT with waaF from strain KT1094	This study
pGEMT-waaQ	pGEMT with waaQ from strain KT1094	This study
pGEMT-waaG	pGEMT with waaG from strain KT1094	This study
pGEMT-waaP	pGEMT with waaP from strain KT1094	This study
pGEMT-waaY	pGEMT with waaY from strain KT1094	This study
pGEMT-waaO	pGEMT with waaO from strain KT1094	This study
Phages
P1clr100	Temperature sensitive, Cmr	25
K1	K1-specific bacteriophage	20

^aLaboratorio de Referencia de E. coli (Spain).

Mutant construction.

E. coli KT1094, serotype K1 and with an R1-type core LPS, was used to mutate individual genes by creating in vitro in-frame deletions. Each mutated gene was transferred to the chromosome by homologous recombination using the temperature-sensitive suicide plasmid pKO3 containing the counterselectable marker sacB (17). Mutations were made in the gene neuA involved in the K1 capsule polysaccharide (CPS) and in eight core LPS biosynthetic genes, waaC, waaF, waaQ, waaP, waaY, waaG, waaO, and waaL. The plasmids containing the individual engineered in-frame deletions derived from KT1094 (pKO3ΔwaaX_KT1094, where X denotes the letter for each particular gene) were transformed into KT1094 by electroporation. Mutants were selected based on growth on LB agar with chloramphenicol at 42°C to select for plasmid integration and subsequent growth on LB agar containing 5% sucrose to identify second recombinants by loss of the chloramphenicol resistance marker of vector pKO3. The mutations were confirmed by sequencing of the whole constructs in amplified PCR products. The ΔwaaC mutant was constructed by asymmetric PCR amplifications using KT1094 chromosomal DNA and primers waaC-R1-A (5′-CGCGGATCCTCGCTCTCATCA GAACGTC-3′), waaC-R1-B (5′-CCCATCCACTAAACTTAAACAGTGGAGAACATCGCCCATC-3′), waaC-R1-C (5′-TGTTTAAGTTTAGTGGATGGGCAGATGGTTTGTAGGGCTCC-3′) (overlapping regions are underlined), and waaC-R1-D (5′-CGCGGATCCCCGTCTGGCAATGTATGAA-3′). The primers include BamHI sites (indicated by boldface letters). DNA fragments of 606 (waaC-R1-A–waaC-R1-B) and 607 (waaC-R1-C–waaC-R1-D) bp were obtained. DNA fragment waaC-R1-A–waaC-R1-B included nucleotide (nt) 4228414, inside waaF, to nt 4228989, corresponding to the third base of codon 16 of waaC (nucleotide positions are according to the genome of E. coli CFT073, GenBank accession number AE014075.1). DNA fragment waaC-R1-C–waaC-R1-D included the last 33 codons of waaC, nt 4229821 to nt 4230398, inside waaL. DNA fragments waaC-R1-A–waaC-R1-B and waaC-R1-C–waaC-R1-D were annealed at their overlapping region (underlined in primers waaC-R1-B and waaC-R1-C) and amplified by PCR as a single fragment using primers waaC-R1-A and waaC-R1-D. The fusion product was purified, BamHI digested, ligated into BamHI-digested and phosphatase-treated pKO3 vector, electroporated into E. coli KT1094, and plated on chloramphenicol-LB agar at 30°C to obtain plasmid pKO3ΔwaaC_KT1094. Oligonucleotides waaC-R1-E and waaC-R1-F were used to amplify the mutated region by DNA amplification and nucleotide sequence determination leading to amplicons of 2,493 and 1,562 bp for the wild-type and mutant strains, respectively. Similarly, seven primers sets of four oligonucleotides (Table 2) were used to construct the seven additional pKO3ΔwaaX_KT1094 plasmids (Table 1). The same approach was used to generate an unencapsulated derivative of strain KT1094 by generation of an in-frame nonpolar mutation en the neuA gene (Table 1 and 2).

Table 2.

Oligonucleotides used for mutant construction and gene complementation

Construct and primer name	5′–3′ Sequence	Product size(s) (bp [primer pair])
KT1094 ΔwaaC
waaC-R1-A	CGCGGATCCTCGCTCTCATCAGAACGTC
waaC-R1-B	CCCATCCACTAAACTTAAACAGTGGAGAACATCGCCCATC	606 (A/B)
waaC-R1-C	TGTTTAAGTTTAGTGGATGGGCAGATGGTTTGTAGGGCTCC
waaC-R1-D	CGCGGATCCCCGTCTGGCAATGTATGAA	607 (C/D)
waaC-R1-E	CTATATCGCGCTGGCCTAT
waaC-R1-F	TCGACCGTGTTGCATTATC	2,493 (wild type; E/F), 1,562 (mutant; E/F)
KT1094 ΔwaaF
waaF-K12-A	CGCGGATCCCGTTCCTGTACGCTTCTTC
waaF-K12-B	CCCATCCACTAAACTTAAACATCATCATCATGTCGCCAACC	691 (A/B)
waaF-K12-C	TGTTTAAGTTTAGTGGATGGGATCGACATTACTCCCCAGC
waaF-K12-D	CGCGGATCCAAATACGGCATATTCGCCAG	648 (C/D)
waaF-K12-E	AGCAACATCGTTAAAGCCC
waaF-K12-F	TGATTCTTCCCATACCCACC	2,837 (wild type; E/F), 1,925 (mutant; E/F)
KT1094 ΔwaaL
waaL-R1-A	CGCGGATCCGGGTCTTTACTTTTCGATTTTT
waaL-R1-B	CCCATCCACTAAACTTAAACAATCCCACAAAGACAAAGTCA	726 (A/B)
waaL-R1-C	TGTTTAAGTTTAGTGGATGGGACTGCTGTGTTTGTCACCAT
waaL-R1-D	CGCGGATCCAGTCTGGGCTATAGCAAACC	708 (C/D)
waaL-R1-E	GCGCTACAAGCAAAGAACTA
waaL-R1-F	GCGCAAGGACAGTTTATATG	2,844 (wild type; E/F), 1,783 (mutant; E/F)
KT1094 ΔwaaQ
waaQ-R1-A2	CGCGGATCCAGGTAGAGAAGGGCGGTGT
waaQ-R1-B	CCCATCCACTAAACTTAAACATTTTGCGTCAGGGTAATTT	588 (A/B)
waaQ-R1-C	TGTTTAAGTTTAGTGGATGGGCGTGACCGAAATGAGATGT
waaQ-R1-D2	CGCGGATCCACGGAAGATTTCACGGCTA	689 (C/D)
waaQ-R1-E	AACGTGCTGAACATCCTTC
waaQ-R1-F	GGAACGATCGACACCTTTA	2,421 (wild type; E/F), 1,595 (mutant; E/F)
KT1094 ΔwaaG
waaG-R1-A	CGCGGATCCTCGTTGGTGAAGCAGACAAC
waaG-R1-B	CCCATCCACTAAACTTAAACAAACCCGAACATGATGACCTC	731 (A/B)
waaG-R1-C	TGTTTAAGTTTAGTGGATGGGCGCGACATTATGCTGATACA
waaG-R1-D	CGCGGATCCCCGGTGCAGGTCAATTACT	675 (C/D)
waaG-R1-E	CGCTTATTAAATGCCCGTGT
waaG-R1-F	CCCTTGTTCCTGCTTGAGAA	2,592 (wild type; E/F), 1,679 (mutant; E/F)
KT1094 ΔwaaO
waaO-R1-A	CGCGGATCCTCATTGCGGATGCCAGTAT
waaO-R1-B	CCCATCCACTAAACTTAAACAGCCATAAGCAACATGGAATG	753 (A/B)
waaO-R1-C	TGTTTAAGTTTAGTGGATGGGGCTCGTTATTGCGCTAAGCA
waaO-R1-D	CGCGGATCCCCCTCGTAAAAGCGTGAGTAA	745
waaO-R1-E	GGACATCATAACGGGTGGTC
waaO-R1-F	CGATATTCAACGCATCCTGA	2,524 (wild type; E/F), 1,729 (mutant; E/F)
KT1094 ΔwaaY
waaY-R1-A	CGCGGATCCACGCGACATACTTCAGGGTA
waaY-R1-B	CCCATCCACTAAACTTAAACAAAATGAGAAGCCGCGATAGC	767 (A/B)
waaY-R1-C	TGTTTAAGTTTAGTGGATGGGCGTCACCTGGGTATTGCCAA
waaY-R1-D	CGCGGATCCATCATTTTTCGCCCATGCTT	739
waaY-R1-E	GCTGGCGTTTCAATTTCTTC
waaY-R1-F	TCTGGCAGGGATAATTCGTC	2,459 (wild type; E/F), 1,928 (mutant; E/F)
KT1094 ΔwaaP
waaP-R1-A	CGCGGATCCCAGCCAATAGCCGTGAAATC
waaP-R1-B	CCCATCCACTAAACTTAAACAACGAAATACCTCACCCTGCA	702 (A/B)
waaP-R1-C	TGTTTAAGTTTAGTGGATGGGGCCACAAAAATCAGGGAAAG
waaP-R1-D	CGCGGATCCAGCATCGCGTTCAGTAACAA	573
waaP-R1-E	CGGGCTGGACGTTTATTATG
waaP-R1-F	GGAAACGACGGCTTTATCAG	2,300 (wild type; E/F), 1,655 (mutant; E/F)
KT1094 ΔneuA
89neuA-A2	CGCGGATCCCCCCTTTTGACGAAGACTC
89neuA-B	CCCATCCACTAAACTTAAACAACTACGGGCTGGAATTATC	817 (A/B)
89neuA-C	TGTTTAAGTTTAGTGGATGGGGAAAACGAAATAGCGGAGAT
89neuA-D2	CGCGGATCCCCAGTGTCAGCGTTACTGC	799
89neuA-E	GCTATTGCACCTAAGGCAG
89neuA-F	AGTTCCCAATTAGCCCACA	3,017 (wild type; E/F), 1,850 (mutant; E/F)
KT1094 Δwaa-Km
1999-1	GAATTCTTTCCATTTATGCTG
1999-2	TCGGTGATATCACCGTTGT	1,554
RFAF	TTTGCATCAGGATAATTCTGC
RFAR	ACCACGTTCAACCAGTGAA	1,511
pGEMT-hldD
HldD1	GAAGGTTACAGTTATGATCATCG
HldD2	GGAGCGTGCGATAGAGACTT	1,028
pBAD33-waaC
C1	CGGGGTACCAGCGCGTACTGGAAGAACTC
C2	CGCTCTAGATTAAATCATGGCAGCTTTTTCA	1,034
pBAD33-waaCF
F1	CGGGGTACCTGAATCGTGACGCATAAGAG
C2	CGCTCTAGATTAAATCATGGCAGCTTTTTCA	2,057
pBAD33-waaCFQ
Q1	CGCTCTAGAAGAGTTACTTGTGGATAAGCCATTTC
Q2	CCCAAGCTTCGGCAACTGTCTGAGCAATA	1,150
pBAD33-waaCFQG
Q1	CGCTCTAGAAGAGTTACTTGTGGATAAGCCATTTC
G1	CCCAAGCTTCCTCACCCTGCAAGGTTTTA	2,264
pBAD33-waaCFQGP
Q1	CGCTCTAGAAGAGTTACTTGTGGATAAGCCATTTC
P1	CCCAAGCTTCGGAAATTACAACGATTTTCG	2,986

Plasmid constructions for mutant complementation studies.

For complementation studies of the nonpolar mutants, neuA and the R1 core type waa genes were amplified from the strain KT1094 chromosomal DNA by using primers waaC-R1-E and waaC-R1-F, which were also used for the testing of the correspondent mutants (Table 2). Amplified DNA fragments were ligated to pGEM-T Easy (Promega) and transformed into E. coli DH5α. For complementation of the strain KT1094 Δwaa-Km(pGEMT-hldD) DNA fragments containing genes waaC and waaCF were amplified from plasmid pWBS10 using oligonucleotides with engineered KpnI and XbaI restriction sites at its ends (Table 2) and ligated to KpnI-XbaI-digested pBAD33 (19) to obtain pBAD33-waaC and pBAD33-waaCF, respectively. DNA fragments containing genes waaQ, waaQG, and waaQGP were amplified from strain KT1094 using oligonucleotides with engineered XbaI and HindIII restriction sites at their ends (Table 2). The amplified fragments were digested with XbaI and HindIII and ligated to XbaI-HindIII-digested pBAD33-waCF to obtain plasmids pBAD33waaCFQ, pBAD33-waaCFQG, and pBAD33-waaCFQGP. These plasmid constructs were tested by nucleotide sequence determination and then electroporated in the corresponding KT1094-derived mutants for complementation assays.

Immunological methods.

Whole cells, culture supernatants, cell surface extracts, or chromatographic fraction extracts were analyzed by enzyme-linked immunosorbent assay (ELISA). ELISAs were performed by dispensing standardized suspensions of each cell or fraction in coating buffer (pH 9.6) into 96-well microtiter plates. The plates were left to stand overnight at 4°C. The wells were blocked with 1% bovine serum albumin in phosphate-buffered saline for 2 h at 37°C. Anti-K1 polyclonal serum (1:500) was added and incubated for 2 h at 37°C. Detection was performed by using peroxidase-labeled sheep anti-rabbit immunoglobulin G (1:1,000) and 2,2′-azino-di-(3-ethylbenzthiazoline sulfonate) as the substrate.

Cell surface extraction and electrophoresis.

E. coli cells grown in LB medium at 37°C were dried and extracted. The phenol-water procedure (35) and the phenol-chloroform-light petroleum ether method (11) were used for the extraction of O-PS-containing and -deficient LPS, respectively. For screening purposes, LPS was obtained after proteinase K digestion of whole cells (8). A cell extract fraction containing both LPS and CPS was obtained using a modified phenol-water method, using phenol and water (1:1) at 37°C. The resulting aqueous phase was dialyzed (molecular-weight cutoff, 3,500) for 4 days. The sample was ultracentrifuged at 100,000 × g at 15°C for 17 h. Successively, the precipitate obtained after ultracentrifugation was dissolved and loaded on a Sephacryl S-200 column, eluted with 50 mM NH₄HCO₃ first and then with a mixture of 0.2 M NaCl, 10 mM Tris, 0.25% DOC, and 10 mM EDTA. Extracts and fractions were separated by SDS-PAGE or SDS-Tricine-PAGE using the method of Laemmli and Favre (16) and visualized by silver and Alcian blue staining as previously described (28, 15, 2).

Isolation of core LPS OS.

The LPS was hydrolyzed in 1% acetic acid (100°C, 90 min), and the precipitate was removed by centrifugation (at 2,500 × g for 1 h). The supernatant was evaporated to dryness, dissolved in water, centrifuged (100,000 × g at 4°C for 4 h), and separated by gel permeation chromatography on a column (2.5 by 70 cm) of Sephadex G-50. The core oligosaccharides (OS) were then separated by high-performance anion-exchange chromatography (HPAEC) on a column (4 by 250 mm; Dionex Corp.) of CarboPack PA100 which was eluted at 1 ml min⁻¹, using a linear gradient program of 15 to 40% 1 M sodium acetate in 0.1 M NaOH over 70 min, and isolated fractions were desalted by gel permeation chromatography on a column (1 by 70 cm) of Sephadex G-10 in 10 mM aqueous NH₄HCO₃.

Compositional analyses.

Monosaccharides were analyzed as acetylated methyl glycosides, which were obtained from the crude LPS (0.5 mg). Methanolysis was performed in 1 M HCl/MeOH (0.5 ml; at 80°C for 20 h), and the sample was extracted twice with hexane. The hexane layer was analyzed by gas chromatography mass spectrometry (GC-MS) to identify fatty acid methyl esters. The methanol layer was dried and acetylated with Ac₂O and pyridine (100°C, 30 min). The samples were analyzed on a Agilent Technologies gas chromatograph 6850A equipped with a 5973N mass selective detector and a Zebron ZB-5 capillary column (Phenomenex; 30 m by 0.25-mm internal diameter, flow rate of 1 ml min⁻¹, and He as the carrier gas). Acetylated methyl glycosides were analyzed accordingly with the following temperature program: 150°C for 3 min, followed by 150°C to 240°C at 3°C min⁻¹.

NMR spectroscopy.

Samples were freeze-dried twice from a 99.9% D₂O solution and dissolved in 99.95% D₂O. Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance II 600-MHz spectrometer (Germany) at 30°C. Internal sodium 3-trimethylsilylpropanoate-2,2,3,3-d₄ (δ_H 0) was used as a reference for calibration. Two-dimensional (2D) NMR spectra were obtained using standard Bruker software, and the Bruker TopSpin, version 2.1, program was used to acquire and process the NMR data.

Mass spectrometry.

Electrospray ionization (ESI) mass spectra were run in the negative ion mode using a micrOTOF II instrument (Bruker Daltonics). Mass spectra were acquired using standard experimental sequences as provided by the manufacturer. Samples (∼50 ng μg l⁻¹) were dissolved in a 1:1 (vol/vol) water-acetonitrile mixture and sprayed at a flow rate of 3 μl min⁻¹. Capillary entrance voltage was set to 4.5 kV, and exit voltage was set to −150 V; drying gas temperature was 180°C.

RESULTS

Effect of R1 core LPS mutants in K1 capsule.

In order to assess the possible interaction between K1 capsule and LPS, we decided to construct a series of nonpolar in-frame mutants in the neuA gene, coding for a CMP-Neu5Ac synthetase involved in the biosynthesis of K1 CPS, and in several genes involved in core LPS biosynthesis. Since most of the K1-encapsulated strains contain an R1-type core LPS, these mutants were constructed in E. coli KT1094 producing K1 capsule and R1-type core LPS, as determined by DNA fragment amplification using R1 core-specific diagnostic oligonucleotides (3). The E. coli KT1094 ΔneuA strain was devoid of K1 CPS, as evidenced by its resistance to K1-specific phage K1 and the lack of reactivity to an anti-K1 polyclonal antibody in ELISAs using whole cells. Analysis of LPS from E. coli KT1094 ΔneuA in SDS-Tricine-PAGE showed no differences in the mobility patterns between the mutant and wild-type strains (Fig. 1A, lanes 1 and 2).

Fig 1.

Polyacrylamide gels showing the migration of LPS from E. coli KT1094 and derived nonpolar waa mutants. Shown are SDS-Tricine-PAGE analyses of LPS from E. coli mutants containing (A) and lacking (B) O antigen. (A) LPS samples from KT1094 (lane 1), KT1094 ΔneuA (lane 2), KT1094 ΔwaaY (lane 3), KT1094 ΔwaaQ (lane 4), KT1094 ΔwaaP (lane 5), and KT1094 ΔwaaL (lane 6). (B) LPS samples from KT1094 (lane 1), KT1094 ΔwaaL (lane 2), KT1094 ΔwaaO (lane 3), KT1094 ΔwaaG (lane 4), KT1094 ΔwaaF (lane 5), and KT1094 ΔwaaC (lane 6).

LPS samples obtained from strain KT1094-derived nonpolar in-frame waa mutants were analyzed by SDS-Tricine-PAGE which showed the expected mobility shift due to core LPS truncations (Fig. 1). Among the studied mutants, only LPS from strains KT1094 ΔwaaQ and KT1094 ΔwaaY (Fig. 1A, lanes 3 and 4) retained the ability to produce full O antigen, while only a faint amount of O antigen was apparent for KT1094 ΔwaaP (Fig. 1A, lane 5).

To determine the effect of each core LPS mutation on K1 capsule, these mutants were assayed for sensitivity to phage K1 using as controls E. coli KT1094 (K1 capsule positive [K1⁺]) and KT1094 ΔneuA (K1 capsule negative [K1⁻]). The results obtained (Table 3) showed that no differences were detected between wild-type KT1094 and KT1094 ΔwaaL (O antigen ligase mutant), indicating that the absence of O antigen did not affect the presence of K1 capsule on the cell surface. The efficiency of plating (EOP) for phage K1 also remains unchanged in E. coli KT1094 ΔwaaO [UDP-glucose:(glucosyl) LPS α1,3-glucosyltransferase mutant), suggesting that the presence of a core lacking four hexose (Hex) residues from the outer core [KdoHep(PPEtN)Hep(P)(Hep)Glc, where P is phosphate and PPEtN is pyrophosphorylethanolamine] does not affect the expression and localization on the cell surface of the K1 capsule. In contrast, inner-core LPS mutants KT1094 ΔwaaC (heptosyltransferase I mutant), KT1094 ΔwaaF (heptosyltransferase II mutant), and KT1094 ΔwaaP (unphosphorylated l-glycero-d-manno-heptose I [HepI] mutant) became completely resistant to phage K1 as well as outer-core mutant KT1094 ΔwaaG [UDP-glucose:(heptosyl) LPS α1,3-glucosyltransferase mutant] (Table 3). Inner-core mutants KT1094 ΔwaaY (unphosphorylated HepII mutant) and KT1094 ΔwaaQ (heptosyltransferase III mutant) were still sensitive to phage K1 but showed a clear reduction in EOP values compared to wild-type strain KT1094 levels (Table 3). To confirm the results observed for this series of core LPS mutants, another approach was used to measure K1 capsule in these strains based on an ELISA with anti-K1-specific polyclonal antibodies yielding results equivalent to those found using phage K1 (Table 3). Genetic complementation of mutants showing increased resistance to phage K1 restored wild-type levels of phage sensitivity and K1 capsule productions as determined by ELISA (data not shown). These results suggest that changes in the inner-core LPS structure could affect the production and/or localization of the K1 capsule antigen on the cell surface. To differentiate between these two possibilities, the amount of the K1-PS was determined in the culture supernatants of the mutant strains by ELISA (data not shown). The K1-PS was found in the supernatant of all of the waa mutants, indicating that mutation in the inner-core LPS do not affect the production of K1-PS. Furthermore, a 7-fold increase in K1-PS was found in the culture supernatant of mutants KT1094 ΔwaaF, KT1094 ΔwaaP, and KT1094 ΔwaaG and a 2-fold increase in that of mutants KT1094 ΔwaaQ and KT1094 ΔwaaY compared to that of KT1094 ΔwaaL and KT1094 ΔwaaO.

Table 3.

Determination of K1 capsule antigen in E. coli strain KT1094 and waa gene mutants

E. coli strain	Core OS structure	EOPa	OD405b
KT1094	KdoHep(PPEtN)Hep(P)(Hep)GlcGlc(Glc)GalGal	1	1.8 ± 0.15
KT1094 ΔneuA	KdoHep(PPEtN)Hep(P)(Hep)GlcGlc(Glc)GalGal	<0.001	<0.1
KT1094 ΔwaaL	KdoHep(PPEtN)Hep(P)(Hep)GlcGlc(Glc)GalGal	1	1.7 ± 0.12
KT1094 ΔwaaO	KdoHep(PPEtN)Hep(P)(Hep)Glc	0.99	1.6 ± 0.09
KT1094 ΔwaaG	KdoHepHep	<0.1	<0.1
KT1094 ΔwaaQ	KdoHep(PPEtN)Hep(P)GlcGlc(Glc)GalGal	0.50	0.4 ± 0.03
KT1094 ΔwaaY	KdoHep(PPEtN)Hep(Hep)GlcGlc(Glc)GalGal	0.25	0.3 ± 0.03
KT1094 ΔwaaP	KdoHepHepGlc	<0.001	<0.1
KT1094 ΔwaaF	KdoHep	<0.001	<0.1
KT1094 ΔwaaC	Kdo	<0.001	<0.1

^aK1 capsule was measured by determination of the multiplicity of infection (EOP) to K1 capsule-specific phage K1.

^bK1 antigen detection by ELISA with K1-specific polyclonal antibody. OD₄₀₅, optical density at 405 nm.

Core LPS structure of KT1094-derived mutants.

In order to reveal a possible correlation between the core LPS structure and the expression of K1 capsule, we studied by ESI mass spectrometry and/or NMR spectroscopy the chemical structure of the LPS for each of the core LPS nonpolar mutants constructed in this work; to facilitate this study nonpolar waaQ and waaY mutants were constructed in KT1094 ΔwaaL to obtain double mutants KT1094 ΔwaaL ΔwaaQ and KT1094 ΔwaaL ΔwaaY. The core OS structure for mutants retaining the full sensitivity to phage K1 was, as expected, based on the known waa gene functions. Two-dimensional NMR spectroscopic studies of the OS fraction from E. coli KT1094 ΔwaaL confirmed the presence of the full R1 core structure reported earlier (32). The mass spectrum of the core OS from this strain (Fig. 2A) showed [M-H]⁻ ion peaks at m/z 1,703.48, 1,783.45, 1,826.49, and 1,906.46 corresponding to Hex₅Hep₃KdoP₁, Hex₅Hep₃KdoP₂, Hex₅Hep₃KdoPPEtN, and Hex₅Hep₃KdoP₁PPEtN oligosaccharides, respectively. Each of these peaks was accompanied by a peak differing in 18 u for the corresponding compounds with the Kdo residue in an anhydro form. These data are in agreement with the reported R1 core structure as well (Table 3). A similar analysis showed that the core OS structure from E. coli KT1094 ΔwaaO corresponds to the whole inner core plus one outer-core hexose residue (data not shown) (Table 3).

Fig 2.

ESI mass spectra of acid-released core LPS OS. Shown are spectra from E. coli KT1094 ΔwaaL (A), E. coli KT1094 ΔwaaL ΔwaaY (B), and E. coli KT1094 ΔwaaL ΔwaaQ (C) mutants. Shown are the regions of [M-H]⁻ ion peaks.

While KT1094 ΔwaaL and KT1094 ΔwaaO mutants showed phosphoryl modifications in their core LPSs, mutants that are resistant to phage K1 and thus devoid of K1 capsule showed no phosphorylation of the core. Data from E. coli KT1094 ΔwaaC, E. coli KT1094 ΔwaaF, E. coli KT1094 ΔwaaP, and E. coli KT1094 ΔwaaG revealed major core OS structures of Kdo, HepKdo, HexHep₂Kdo, and Hep₂Kdo, respectively (data not shown) (Table 3).

Mutants with a decreased amount of K1 capsule showed a reduced level of phosphorylation. The mass spectrum of the core OS fraction from E. coli KT1094 ΔwaaL ΔwaaY (Fig. 2B) showed [M-H]⁻ ion peaks found in the OS fraction of E. coli KT1094 ΔwaaL at m/z 1,685.47 and 1,808.47 for Hex₅Hep₃KdoP₁ and Hex₅Hep₃KdoPPEtN, with Kdo in an anhydro form, respectively, but was devoid of ion peaks for Hex₅Hep₃KdoP₂ and Hex₅Hep₃KdoP₁PPEtN oligosaccharides. In addition, a [M-H]⁻ ion peak at m/z 1,605.50 corresponding to a nonphosphorylated Hex₅Hep₃Kdo oligosaccharide was found in the OS fraction from E. coli KT1094 ΔwaaL ΔwaaY while this peak was absent from the mass spectrum of the E. coli KT1094 ΔwaaL core OS.

The mass spectrum of the OS fraction from E. coli KT1094 ΔwaaL ΔwaaQ (Fig. 2C) showed [M-H]⁻ ion peaks at m/z 1,413.44, 1,493.39, 1,573.36, 1,616.39, and 1,696.36 for Hex₅Hep₂Kdo, Hex₅Hep₂KdoP₁, Hex₅Hep₂KdoP₂, Hex₅Hep₂KdoPPEtN, and Hex₅Hep₂KdoP₁PPEtN oligosaccharides, respectively. Therefore, in agreement with a heptosyltransferase III function for the WaaQ protein (38), the core OS from E. coli KT1094 ΔwaaL ΔwaaQ lacks HepIII. A closer inspection of this mass spectrum together with NMR analysis revealed a significant change in the relative abundance of phosphoryl modifications on HepI and HepII compared to those of E. coli KT1094 ΔwaaL. The data presented in Table 4 show 8- and 14.5-fold reductions in the percentages of HepI(P)HepII(P) and HepI(PPEtN)HepII(P) species in E. coli KT1094 ΔwaaL ΔwaaQ compared with E. coli KT1094 ΔwaaL. These distinctions could be responsible for the different behaviors of these two mutants in capsule K1 detection by phage K1 sensitivity or ELISA.

Table 4.

Percentage of phosphorylated core OS variants from E. coli KT1094 ΔwaaL and KT1094 ΔwaaL ΔwaaQ mutants

Strain and mass (Da) of core OS	Substituent on HepI	Substituent on HepII	Content	% of totala
E. coli KT1094 ΔwaaL
1,704	P		76	28.5
1,784	P	P	103	38.6
1,827	PPEtN		26	9.7
1,907	PPEtN	P	62	23.2
E. coli KT1094 ΔwaaL ΔwaaQ
1,512	P		83	67.5
1,592	P	P	6	4.9
1,635	PPEtN		32	26.0
1,715	PPEtN	P	2	1.6

^aThe percentage was calculated using the intensities of the [M-H]⁻ ion peaks for the components in the mass spectrum of the isolated core OS.

Core LPS K1 interaction.

A possible interpretation for the above results could be the existence of a direct ionic interaction between the core LPS region and the K1 capsule antigen. In order to study this possibility, the KT1094 ΔwaaL mutant cells were extracted by phenol-water at 37°C in order to reduce the extent of depolymerization of the capsular polysaccharide (CPS) (21).

The sample was analyzed by SDS-PAGE and stained with Alcian blue as well as silver nitrate (data shown for silver nitrate), which revealed that both the LPS and the CPS were present. Moreover, the GC-MS analysis of the acetylated methyl glycosides showed the presence of Gal, Glc, Hep, GlcN, and Kdo belonging to the core LPS region, together with the neuraminic acid corresponding to K1 CPS.

A partial separation between LPS and CPS was achieved by ultracentrifugation. Both the supernatant and the precipitate were analyzed by SDS-PAGE and GC-MS sugar analysis, revealing that only CPS was present in the supernatant, while both LPS and CPS were present in the precipitate. Repeating the process did not improve the separation.

The precipitate obtained after ultracentrifugation was split in two aliquots, the first of which was loaded on a Sephacryl S-200 column and eluted by 50 mM NH₄HCO₃. The chromatogram showed the presence of two peaks (data not shown). SDS-PAGE analysis revealed that both peaks contained LPS together with CPS. The presence of two peaks was probably due to the different sizes of the LPS molecular aggregates. The second aliquot was chromatographed on the same column, this time eluting with a buffer of increased ionic strength (0.1 M NaCl and 10 mM Tris), which contained the metal ion chelator EDTA and the dissociating detergent DOC. In this case, the chromatogram showed the presence of three peaks (Fig. 3A), named A, B, and C. These three peaks were obtained only when EDTA was present in the eluting buffer (Fig. 3B). The chemical composition of these three peaks determined by GC-MS of the acetylated methyl glycosides revealed the presence of LPS residues in peaks B and C and K1 CPS residue in peaks A and B. Furthermore, ELISAs with anti-K1 polyclonal antibody showed the presence of K1 antigen in peaks A (A₆₆₀ of 1.8) and B (A₆₆₀ of 0.5) but not in C (A₆₆₀ of <0.1). The material from these peaks was subjected to SDS-Tricine-PAGE and SDS-PAGE analysis (Fig. 3C and D, respectively) and visualized by silver staining. The gel clearly showed that fraction A contained only high-molecular-mass bands (Fig. 3C, lane 3, and D, lane 1) corresponding to K1 CPS, according to its chemical composition and reactivity with anti-K1 polyclonal antibody. Fraction C contained an LPS banding material that did not react with anti-K1 antibody (Fig. 3C, lane 4, and D, lane 3). Fraction B contained LPS plus low-molecular-mass K1 CPS bands (Fig. 3C, lane 5, and D, lane 2).

Fig 3.

LPS and K1 CPS separation on a Sephacryl S-200 column. (A) Cell extract from E. coli KT1094 ΔwaaL containing both LPS (O⁻) and K1 CPS was subjected to Sephacryl S-200 chromatography using 0.2 M NaCl, 10 mM Tris, 10 mM EDTA, and 0.25% DOC as eluant and with the same buffer but without EDTA (B). SDS-Tricine-PAGE (C) and SDS-PAGE (D) analysis of the obtained fractions. Shown are LPS preparations from E. coli KT1094 (C, lane 1) and E. coli KT1094 ΔwaaL (C, lane 2) and samples from peaks A (C, lane 3; D, lane 1), B (C, lane 5; D, lane 2), and C (C, lane 4; D, lane 3). R.I., relative intensity.

Reconstruction of core LPS.

To further prove the above correlation, we used a complementary approach to individual gene mutant construction. In this approach a mutant strain was constructed (Fig. 4) with a deletion from codon 169 of the 3′ end of hldD (ADP-l-glycero-d-manno-heptose epimerase) to upstream of waaQ gene. Thus, the DNA located between codon 169 of hldD and waaA (bifunctional Kdo transferase) was deleted and replaced by a kanamycin resistance cassette. To construct this mutant, plasmid pWSB10 (6) (containing the 5′ end of the gene yibB and the hldD, waaF, waaC, and waaL genes) and primers 1999-1 and 1999-2 (engineered to include an EcoRV restriction site) (Table 2, underlined A in the sequence 1999-2) were used to amplify a 1,554-bp DNA fragment containing the 5′ ends of the yibD and hldD genes. Plasmid pFFM991 was obtained by ligation of the amplified fragment to pGEM-T. Plasmid pJSC2 (5) (containing the 5′ ends of the waaQ and the waaA and coaD genes and the 3′ end of fpg) and primers RFAF and RFAR (Table 2) were used to amplify a 1,511-bp DNA fragment. This fragment was isolated and digested with EcoRV and KpnI to obtain a 1,253-bp fragment containing the promoter region and the 5′ end of waaA and was ligated to EcoRV-KpnI-digested pWSK29 (34) to obtain plasmid pWSKA. A 1,250-bp DNA fragment obtained by EcoRI-EcoRV digestion of pFFM991 was ligated to EcoRI-EcoRV-digested pWSKA to obtain plasmid pGA containing the 5′ ends of yibD, hldD, and waaA as well as the waaA promoter. The kanamycin resistance gene from pUC4K was excised by HincII digestion and ligated to EcoRV-digested pGA to obtain plasmid pGKMA (Fig. 4). The engineered deletion was introduced into strain JC7623 (37), because of its ability to recombine with linear incoming DNA fragments, by homologous recombination (18) by electroporation of a 4,050-bp fragment obtained and purified from EcoRI-KpnI-digested pGKMA. Once the mutant was constructed and checked by nucleotide sequence determination, the mutation (Δwaa Km) was transduced using the phage P1Cmts (25) to the E. coli strains W3310 and KT1094 containing K-12- and R1-type core LPSs, respectively.

Fig 4.

Diagram of the construction of E. coli KT1094 Δwaa-Km. Shown are steps leading to the construction of plasmid pGKMA containing a deletion extending from codon 169 of hldD to upstream of waaQ. Plasmid pGKMA was transformed into E. coli JC7623 to obtain by homologous recombination JC7623 Δwaa-Km. This mutation was transduced to E. coli KT1094 to obtain KT1094 Δwaa-Km. EI, EcoRI EI; EV, EcoRV; K, KpnI. EV*, EcoRV restriction site engineered on primer 1999-2. Numbers 1, 2, 3, and 4 denote primers 1999-1, 1999-2, RFAF, and RFAR, respectively.

As expected, both mutant strains produced a core LPS containing only lipid A and Kdo residues (data not shown), and strain KT1094 Δwaa-Km was resistant to K1-specific phage K1 and devoid of K1 capsule in ELISAs. On the other hand, colonies of this mutant grown on LB agar plates at 37°C showed a mucous phenotype, suggesting that K1-PS was being produced but not retained at the cell surface. In agreement with this hypothesis, material reacting with anti-K1 polyclonal antibody was found in ELISAs using KT1094 Δwaa-Km cells lysed with a French press, indicating that K1-PS was still synthesized in this mutant. The mutant KT1094 Δwaa-Km was transformed with plasmid pGEMT-hldD, and the transformant was used to reintroduce, step by step, the waa gene(s). Introduction of waaC (pBAD33-waaC) into KT1094 Δwaa-Km(pGEMT-hldD) leads to a decrease in LPS mobility in SDS-Tricine-PAGE in agreement with the presence of inner-core residue HepI (KdoHep) (Fig. 5, lane 4). Similarly, introduction of waaF and waaC (pBAD33-waaCF) produces a further reduction in LPS mobility in agreement with the presence of inner-core HepI and HepII (Hep₂Kdo) (Fig. 5, lane 5). In contrast, introduction of the three heptosyltransferases encoded by the genes waaC, waaF, and waaQ (pBAD33-waaCFQ) leads to the production of LPS with the same mobility as that of KT1094 Δwaa-Km(pGEMT-hldD/pBAD33-waaCF) (Fig. 5, lane 6), indicating that the HepIII residue cannot be added to the core OS until other main chain residues are transferred to the inner-core Hep disaccharide. Introduction of waaCFQG (pBAD33-waaCFQG) (Fig. 5, lane 7) and waaCFQGP (pBAD33-waaCFQGP) (Fig. 5, lane 8) further reduces the mobility of the corresponding LPS samples in SDS-Tricine-PAGE, in agreement with their core LPS structures, HexHep₂Kdo and HexHep₂KdoP, respectively. As shown in Table 5, core OS resulting from the introduction into KT1094 Δwaa-Km(pGEMT-hldD) of pBAD33-waaC (KdoHep), pBAD33-waaCF (Hep₂Kdo), pBAD33-waaCFQ (Hep₂Kdo), and pBAD33-waaCFQG (HexHep₂Kdo) did not become phage K1 sensitive. Furthermore, all these constructs showed a mucous colonial phenotype in LB agar plates at 37°C. Only when the five waa genes waaC, waaF, waaQ, waaG, and waaP (pBAD33-waaCFQGP) were simultaneously introduced into the KT1094 Δwaa-Km(pGEMT-hldD) strain, a reduction in the colonial mucous phenotype was found, and it was possible to detect K1 capsule with phage K1 and with polyclonal K1-specific antibody on whole cells. These results establish a role for the phosphate substitution at the HepI residue of the inner-core LPS in the interaction between core LPS and the K1 CPS.

Fig 5.

Polyacrylamide gels showing the migration of LPS from E. coli KT109 Δwaa-Km and reintroduced waa genes. Shown are SDS-Tricine-PAGE analyses of LPS from E. coli KT1094 ΔwaaL (lane 1), KT1094 Δwaa-Km (lane 2), KT1094 Δwaa-Km(pGEMT-hldD) (lane 3), KT1094 Δwaa-Km(pGEMT-hldD/pBAD33-waaC) (lane 4), KT1094 Δwaa-Km(pGEMT-hldD/pBAD33-waaCF) (lane 5), KT1094 Δwaa-Km(pGEMT-hldD/pBAD33-waaCFQ) (lane 6), KT1094 Δwaa-Km(pGEMT-hldD/pBAD33-waaCFQG) (lane 7), and KT1094 Δwaa-Km(pGEMT-hldD/pBAD33-waaCFQGP) (lane 8). The strains with pBAD33 plasmid derivatives were grown under inducing conditions. LPS samples were extracted and analyzed according to Darveau and Hancock (8).

Table 5.

Determination of K1 capsule antigen in E. coli mutant strain KT1094 Δwaa-Km(pGEMT-hldD) containing different waa genes

Added gene(s)	EOPa	OD405b
None	<0.001	<0.1
waaC	<0.001	<0.1
waaC, waaF	<0.001	<0.1
waaC, waaF, waaQ	<0.001	<0.1
waaC, waaF, waaQ, waaG	<0.001	<0.1
waaC, waaF, waaQ, waaG, waaP	0.87	1.72 ± 0.11

^aK1 capsule was measured by multiplicity of infection (EOP) to K1 capsule-specific phage K1.

^bK1 antigen detection by ELISA with K1-specific polyclonal antibody. OD₄₀₅, optical density at 405 nm.

DISCUSSION

The analysis of the core OS structures of the individual waa gene nonpolar mutants generated by an in-frame deletion method revealed that mutations affecting inner-core residues decrease the ability to detect the K1 antigen. Furthermore, mutations precluding the phosphorylation of both HepI and HepII, such as KT1094 with mutations ΔwaaC, ΔwaaF, ΔwaaP, and ΔwaaG, result in the lack of K1 antigen as determined by phage K1 EOP and ELISA. In addition, the core OS structures obtained from the KT1094 ΔwaaG mutant clearly indicate that phosphorylation of HepI and HepII residues depends on the transfer of the first Glc residue of the outer core, in agreement with previous results obtained with an E. coli waaG insertion mutant (39). The same conclusion can be drawn from the analysis based on the step-by-step reintroduction of individual waa genes into a strain with a deletion spanning from hldD to waaQ (KT1094 Δwaa-Km).

Among the different waa mutants leading to a decrease in K1 antigen, the ones with a longer core OS correspond to KT1094 ΔwaaQ and KT1094 ΔwaaY. The WaaY protein has been proposed to be responsible for the transfer of phosphate to HepII (38), and accordingly the core LPS from KT1094 ΔwaaY is a full-length R1 core type but without phosphoryl modifications at the HepII residue (Fig. 2C), suggesting that this substitution is important for K1 CPS binding to the cell surface. In the KT1094 ΔwaaQ mutant, the core OS contains the same residues as wild-type KT1094 or KT1094 ΔwaaL with only one Hep (HepIII) residue missing. Initially, it is not obvious how the absence of HepIII could lead to a decrease in K1 antigen, and only after a careful analysis of the phosphorylation pattern in this particular mutant can a hypothesis be made. Comparison of the degrees of phosphorylation of heptose residues in the core OS structures from KT1094 ΔwaaL and KT1094 ΔwaaQ (Table 4) revealed a strong reduction in the percentage (from 38.6 to 4.9%) of OS being simultaneously phosphorylated at both HepI and HepII residues, similar to what has been previously found with the waaQ insertion mutant (38). Another major difference between these OS was the decrease from 23.2 to 1.6% in the OS fraction containing PPEtN and phosphate at HepI and HepII, respectively. Thus, the transfer of HepIII by WaaQ does not appear to be a prerequisite to phosphorylation of HepII by WaaY in contrast to what has been found in a waaQ insertion mutant (38). But HepII phosphorylation in the absence of HepIII proceeds inefficiently; i.e., transfer of HepIII creates a better substrate for WaaY. These changes in the phosphorylation pattern of the KT1094 ΔwaaQ strain would explain the reduction of cell-bound K1-PS in this mutant.

In addition, mutants KT1094 ΔwaaY and KT1094 ΔwaaQ, as has been previously shown (38), did not appear to show pleiotropic effects since they showed essentially the wild-type sensibility to hydrophobic compounds, such as deoxycholate, SDS, and polymyxin B, and outer membrane protein pattern in SDS-PAGE. These results suggest that there could be an ionic interaction between the K1 capsule and negative charges contributed by phosphate groups at HepI and HepII.

In agreement with this hypothesis, it was possible to recover by a modified water-phenol extraction method a cell surface fraction containing both K1-PS and LPS. This fraction could only be resolved into individual high-molecular-mass K1-PS and LPS peaks, as judged by SDS-PAGE, chemical composition analysis and ELISA, and gel permeation chromatography using a high-ionic-strength eluting buffer containing EDTA and DOC. The requirement for an ion chelator suggests that divalent cations bridge the negative charges between phosphate core LPS and K1 residues.

A previous study has suggested that a waaR mutation precluding the addition of a GlcIII residue in an E. coli K-12 strain avoids K5- and K1-PS binding to the cell surface (26). In contrast, our results indicate that a waaO mutant precluding GlcII incorporation to core LPS does not affect K1-PS binding to the cell surface. These two mutants were constructed using different approaches, and it should be noticed that waaR is just upstream of waaY and that possible polar effects cannot be ruled out. Unfortunately in the previous work the actual core LPS structure of the waaR mutant was not determined. Furthermore, the E. coli waaR mutant used by Taylor et al. (26) was constructed in a K-12 strain, where the genes for K1 production were introduced by Hfr conjugation from a K1 strain (31), and not in a natural K1-PS producer strain.

Thus, although it has been suggested that the K1 CPS and other group 2 E. coli capsules are anchored to the cell surface via a CPS lipid domain (36), our results strongly suggest that a major fraction of the K1 CPS is retained on the cell surface via ionic interaction with the negative phosphate charges of the core LPS. This fact is similar to what has been described in K. pneumoniae, where core LPS galacturonic negative charges interact with K2 polysaccharide (10).