A Deep-Rough Mutant of Campylobacter jejuni 81-176 Is Noninvasive for Intestinal Epithelial Cells

Abstract

A waaF mutant of Campylobacter jejuni 81-176 showed decreased invasion of INT407 cells in vitro and increased sensitivity to some antibiotics compared to what was seen with the wild-type strain.

The food-borne pathogen Campylobacter jejuni is one of the principal causes of human gastroenteritis worldwide (7, 17). In addition, C. jejuni is associated with the development of a devastating neurological disorder, Guillain-Barré syndrome, perhaps because of molecular mimicry between the outer core of Campylobacter lipooligosaccharide (LOS) and the carbohydrate moiety of human gangliosides (15, 16). C. jejuni strain 81-176 has been shown in two human volunteer feeding studies to cause diarrhea (5) and has been studied extensively for molecular mechanisms of pathogenesis (2, 3, 5, 12-14, 20). C. jejuni 81-176 LOS (Fig. 1) is composed of two covalently linked domains: lipid A, a hydrophobic anchor, and a nonrepeating core oligosaccharide consisting of an inner and an outer core (1, 10). The carbohydrate structure of the outer core, consisting of hexoses and N-acetylglucosamine, primarily mimics GM₂ and GM₃ gangliosides, with minor amounts of GD₂ and GD_1b mimics being detectable (10). Two outer-core mutants have been assayed for their ability to invade human INT407 cells (10). A mutant lacking N-acetylneuraminic acid invaded INT407 cells at the same levels as did the wild-type strain, and interestingly, a mutant lacking the terminal GalNAc showed a modest, but statistically significant, increase in ability to invade INT407 cells (10).

FIG. 1.

Schematic showing the molecular structure of the 81-176 LOS and highlighting the inner-core region. Hep, heptose; PEtn, phosphoethanolamine. The waaF gene encodes heptosyltransferase II, which is involved in the addition of the second heptose as indicated.

The inner-core region of C. jejuni 81-176 LOS is structurally similar to the same region of lipopolysaccharide (LPS) in enteric bacteria such as Escherichia coli and Salmonella enterica serovar Typhimurium (11) (Fig. 1). It consists of a single 3-deoxy-d-manno-2-octulosonic acid residue (Kdo) attached to lipid A and two l-glycero-d-manno-heptose (heptose) residues attached to the Kdo residue. The inner-core region also contains a glucose residue and phosphoethanolamine moiety attached to the first heptose residue. To better understand the relationship between biosynthesis of the LOS and pathogenesis, we initiated a study of the inner-core oligosaccharide of LOS from Campylobacter strain 81-176. The waaF gene was identified in an ordered partial Sau3A-digested genomic library of 81-176 DNA cloned into λ-ZAPII (L. C. Holder and P. Guerry, unpublished data). Four open reading frames (ORFs), organized in a fashion similar to what is seen in other C. jejuni strains (8), were revealed from the sequence analysis of plasmid pLCH3-4 (Fig. 2). ORF2 encoded a protein of 318 amino acids with a predicted molecular mass of 36 kDa that showed significant similarity to several WaaF enzymes that catalyze the transfer of the second heptose to the core oligosaccharide of LPS and LOS. This ORF product was 88% identical to the recently reported WaaF protein from C. jejuni 11828 (19). In addition, sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis followed by silver staining showed that the 81-176 waaF gene is able to restore LPS assembly by complementation of a Salmonella serovar Typhimurium waaF511 mutant strain, SL3789 (22) (data not shown).

FIG. 2.

Schematic representation of the C. jejuni 81-176 genes identified in the insert in plasmid pLCH3-4. The region of DNA that contained the 81-176 waaF gene (ORF2) was cloned into PCR 2.1 TOPO (Invitrogen, Carlsbad, Calif.) to generate pMK10 and into the kanamycin-resistant Campylobacter shuttle plasmid pRY107 (24) to generate pLCH1 for complementation studies in Salmonella serovar Typhimurium and C. jejuni waaF mutants, respectively. The ORF designated waaV encoded a protein that was 91% identical to WaaV (Cj1146c) from the genomic strain of C. jejuni, NCTC 11168. ORF3 encoded a protein that was 100% identical to an unpublished sequence from another strain of C. jejuni (NCTC 11351; accession number AAK08966) but was not present in NCTC 11168. GmhA of strain 81-176 was 97% identical to the corresponding protein from NCTC 11168 (Cj1149c).

A waaF::Cm insertional mutant was generated as previously described (9). One insertion that occurred 146 bp from the translational start of waaF was selected to electroporate 81-176 to Cm^r, and a double crossover was confirmed by PCR (data not shown). The LOS from the waaF::Cm mutant migrated more quickly than that of the wild-type strain (Fig. 3, lane 2 versus 1). The presence of an LOS similar to that of the wild type was restored when the waaF mutant strain was complemented in trans with pLCH1 (Fig. 2), which contains the 81-176 waaF gene (Fig. 3, lane 3). Table 1 lists the strains and plasmids used in this study.

FIG. 3.

Comparison of LOS from C. jejuni 81-176 with that from the waaF mutant. Proteinase K-digested whole-cell preparations were electrophoresed on 16% Tricine gels and silver stained. Lane 1, 81-176; lane 2, waaF mutant; lane 3, waaF mutant complemented in trans (pLCH1).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant property or characteristic(s)	Reference or source
C. jejuni strains
81-176	Penner serotype 23, 36	5, 14
81-176waaF::cat	Nonpolar insertion in waaF, Cmr	This study
Serovar Typhimurium     strains
SL3770	waa+	SGSCa
SL3789	waaF511	SGSC
E. coli strain
DH5α	F− φ80d lacZ ΔM15	Invitrogen
Plasmids
pCR2.1	Cloning vector, Apr	Invitrogen
pMK10	pCR2.1 with a 1.5-kb PCR-derived insert containing C. jejuni 81-176 waaF, Apr	This study
pRY107	E. coli-C. jejuni shuttle vector, Kmr	24
pLCH1	pRY107 with a 1.5-kb XbaI-BamHI fragment containing waaF from pMK10, Cmr Kmr	This study

^aSGSC, Salmonella Genetic Stock Centre.

A previous study has shown that the permethylated core oligosaccharide of the parental strain 81-176 possesses pseudomolecular ions with m/z values of 1,943, 2,188, 2,548, and 2,752 [M + H]⁺ corresponding to cores exhibiting mimicry of GM₃, GM₂, GD₂, and GD_1b gangliosides, respectively (10). In contrast, the mass spectrum of the truncated LOS species in the waaF mutant possessed a pseudomolecular ion with an m/z value of 926 [M + H]⁺ and related daughter ions that may be interpreted as those of an LOS species that contains glucose, heptose, phosphoethanolamine, and Kdo (Fig. 4). This result supports the loss of the second heptose residue from the LOS. Furthermore, glycosylation of a complete LOS requires the transfer of the sugar catalyzed by the heptosyltransferase II, WaaF.

FIG. 4.

Analysis of the positive-ion fast atom bombardment mass spectrometry of the permethylated core oligosaccharide from LOS of 81-176 waaF. Numbers refer to m/z values for ions. The asterisk indicates that derivatization resulted in partial degradation of the Kdo with a loss of 46 mass units.

Since there was no difference between the wild-type and waaF mutant strains in growth or motility (data not shown), the ability of the C. jejuni 81-176 deep-rough mutant to invade INT407 human intestinal epithelial cells was examined as previously described (2, 3, 18, 25). The data shown in Fig. 5 are expressed as the percentage of the input inoculum internalized during the 2-h gentamicin kill assay. The level of wild-type 81-176 invasion (1.238% ± 0.634%) was 14-fold higher than that of the waaF mutant (0.08% ± 0.074%). When the waaF mutant was complemented in trans with pLCH1, the level of invasion was restored to wild-type levels (2.0% ± 1.41%).

FIG. 5.

Comparison of levels of invasion of INT407 cells. Invasion levels are shown as percentages of the input inoculum internalized during a 2-h invasion assay. Data shown are the mean results from two invasion assays for the complemented mutant and from four invasion assays for the wild type and the waaF mutant.

Deep-rough mutants of enteric bacteria are hypersensitive to hydrophobic antibiotics and detergents due to major alterations in the outer membrane (11, 21). As shown in Table 2, the C. jejuni waaF mutant showed hypersensitivity to the antibiotics fusidic acid and novobiocin and the detergent SDS compared to what was seen with the wild type. However, neither the wild-type nor the waaF mutant strain was sensitive to bacitracin, and there was no difference in the sensitivities to gentamicin, which was used in the invasion assays. Wild-type levels of resistance to fusidic acid, novobiocin, and SDS were restored when the waaF mutant was complemented in trans.

TABLE 2.

Sensitivity to antibiotics and SDS

Strain	Zone diam (mm) with agenta:
	Novobiocin	Bacitracin	Fusidic acid	SDS
81-176	11	7	10	14
waaF mutant	16	7	15	18
waaF(pMK10)	13.5	7	10	11

^aCompounds were impregnated onto 7-mm-diameter disks. A zone size of 7 mm is equal to the diameter of the disk and corresponds to complete resistance.

Although a waaF mutant (19) of another C. jejuni strain with a core structure distinct from that of 81-176 (10) has been described, this is the first report of both the structure of a waaF core and complementation in trans and the first report to examine the effect of deep-rough mutations on virulence. waaF mutants of a number of gram-negative bacteria have been studied (4, 6, 22), and Haemophilus ducreyi and Neisseria gonorrhoeae waaF mutants have been shown to be less virulent (4, 23). This reduction in invasion by the C. jejuni waaF mutant may reflect the need for a complete LOS for proper membrane function. As with other deep-rough mutants, mutation of waaF led to hypersensitivity to various antibiotics. We are in the process of determining whether loss of WaaF activity may result in additional changes in the composition or structure of the outer membrane.

Nucleotide sequence accession number.

The nucleotide sequence of the region of the chromosome of C. jejuni 81-176 presented in Fig. 2 has been deposited in GenBank under accession number AY423899.