Identification and Characterization of a Phase-Variable Nonfimbrial Salmonella typhimurium Gene That Alters O-Antigen Production

Abstract

Salmonella typhimurium 798, which was isolated from a pig, is known to phase vary from a nonadhesive to an adhesive phenotype. Cells of the adhesive phenotype adhere to porcine enterocytes, are more readily phagocytized by porcine neutrophils and macrophages, and once phagocytized can survive intracellularly, while cells of the nonadhesive phenotype die rapidly. The effect of phenotypic switching also can be visualized by changes in colony morphologies and the presence of between 10 and 15 proteins in the envelopes of cells in the adhesive phenotype. Mutants previously constructed with cells in the adhesive phenotype and the transposon TnphoA were screened to identify mutants lacking one or more of the unique proteins. One mutation was cloned and sequenced, and the mutation was shown to be in rfaL (O-antigen ligase). Expression of O antigen was shown to be phase variable. The adhesive strain expressed an O antigen that was at least eightfold longer than that for the nonadhesive strain and by virtue of O-antigen production was resistant to porcine complement. The mutant survived intracellularly in phagocytic cells as well as its wild-type parent.

The pathogenesis of enteric disease caused by Salmonella typhimurium requires an intricate and tightly controlled expression of virulence genes that are coordinately regulated. Specific environmental cues most likely act as signals that trigger expression of these virulence genes (38–40). The initial step in pathogenesis is colonization of intestinal villi resulting from the attachment of the organism to the mucosal epithelium (21). Subsequently, S. typhimurium invade these cells. The primary route of invasion is presumed to be through M cells located in Peyer’s patches (8, 9, 19, 23), although S. typhimurium can also invade villous absorptive enterocytes (49). Once S. typhimurium cells enter M cells, apoptosis is induced (41), leading to release of the organism into the lymphoid follicle, where they encounter macrophages. S. typhimurium cells readily enter macrophages and proliferate. To accomplish this, S. typhimurium cells must resist a diverse array of host factors within the macrophage, including oxidative destruction, hydrolytic enzymes, and cationic peptides. The organism also must be able to resist serum complement in blood and starvation conditions. Thus, S. typhimurium has developed several mechanisms resulting in programmed expression of virulence genes that are important for survival within the host and, in particular, within macrophages.

Many virulence genes of S. typhimurium are located on pathogenicity islands, including inv, spa, and hil and genes encoding a type III protein secretion system. A pathogenicity island located on the 90-kb virulence plasmid encodes spv (Salmonella plasmid virulence) and pef (plasmid-encoded fimbriae) (5, 15). At least two global regulatory systems control numerous S. typhimurium virulence genes. The PhoP-PhoQ two-component regulatory system is involved in the activation and repression of more than 40 genes required for host cell invasion and intracellular survival in macrophages (16, 17, 39, 44, 52). In this system, a balanced communication between the environmental sensor PhoQ and the activator PhoP is required. The second system employs the alternative sigma factor, RpoS, which positively regulates genes encoding stress response proteins expressed in stationary phase (47). RpoS regulates expression of plasmid-encoded spv genes that are required for systemic infections.

We have been studying what might represent a third important regulatory mechanism modulating virulence genes, i.e., phase variation. Previously, we showed that S. typhimurium 798, a clinical isolate obtained from a pig with diarrhea, could be grown in two phases with markedly different phenotypes (21). The two phenotypes initially were identified based on adhesiveness to porcine enterocytes and, thus, were termed adhesive and nonadhesive. Subsequent investigations showed that cells of the adhesive phenotype also produced fimbriae, were taken up more efficiently by phagocytes, and survived better intracellularly than cells of the nonadhesive phenotype (21). Cells of the adhesive phenotype also expressed between 10 and 15 envelope-associated proteins that were absent from cells of the nonadhesive phenotype. When cells varied to the nonadhesive phenotype, they coordinately lost expression of all of these traits.

Phase variation is a well-described meta-stable regulatory mechanism employed for the transient expression of many bacterial surface proteins. The most commonly studied phase-variable genes include S. typhimurium flagella (18, 44, 45) and fimbriae such as Escherichia coli type 1 fimbriae (1, 10, 12, 20, 35) and Pap pili (6, 51). Recently, nonfimbrial bacterial surface structures also have been shown to be phase variable, including Haemophilus influenzae lipopolysaccharide (LPS) (42, 53, 54), Coxiella burnetii LPS (29), Neisseria gonorrhoeae Opa proteins (2, 43, 48), and E. coli Ag43 (7). In the present study, we identified an additional phase-variable trait of S. typhimurium that appears to control the production of O antigen. As a result of altering O-antigen synthesis, nonadhesive cells became sensitive to serum complement.

MATERIALS AND METHODS

Bacteria and media.

S. typhimurium ι518 (nonadhesive) and ι519 (adhesive) strains are phenotypic variants of S. typhimurium 798. Strain ι519′ carries a phoN mutation in ι519. A TnphoA mutant bank in S. typhimurium ι519′ was previously constructed and contains 520 mutants (3a). All bacteria were grown overnight in tryptone-phosphate broth (TPB; Becton-Dickinson, Cockeysville, Md.) at 37°C with shaking, except as indicated below. For preparation of DNA, Luria-Bertani (LB) was used (34). Antibiotics were used when appropriate at the following concentrations: ampicillin, 50 μg/ml; kanamycin, 30 μg/ml; and tetracycline, 12 μg/ml. XP plates contained 5-bromo-4-chloro-3-indolyl phosphate (40 μg/ml) in LB agar for detection of alkaline phosphatase activity expressed by TnphoA fusion proteins.

Envelope protein extractions.

Bacterial cultures (150 ml) were incubated overnight in TPB at 37°C with shaking. Cells were harvested by centrifugation at 10,000 × g for 10 min. Pellets were resuspended in 50 ml of phosphate-buffered saline (PBS; pH 7.4) and were homogenized with an Omni Mixer (Omni International, Waterbury, Conn.) at a speed setting of 5 for 30 min on ice. Homogenates were centrifuged at 10,000 × g for 10 min to remove bacterial cells. Supernatants were collected, and proteins were precipitated by the addition of ammonium sulfate to a final concentration of 60% (36.1 g/100 ml). The precipitated proteins were collected by centrifugation at 12,000 × g for 10 min at 4°C, resuspended in 0.5 ml of PBS, and desalted by dialysis. Protein concentrations were determined by Microbicinchoninic acid (Pierce Chemical Co., Rockford, Ill.).

Protein electrophoresis.

Protein samples (20 to 25 μg) were loaded onto sodium dodecyl sulfate (SDS)-12 or 15% polyacrylamide gels with 4% stacking gels (26). Electrophoresis was carried out at constant voltage (200 V). Proteins were fixed and stained with 0.05% Coomassie blue.

Recombinant DNA techniques.

Total genomic DNA was extracted by a protocol employing CTAB (4). The genomic DNA was digested with restriction endonuclease KpnI or SacI (Promega and New England Biolabs) and ligated into the KpnI or SacI site in pGem4z (Promega). Ligated DNA was electroporated into competent E. coli DH5α with an Invitrogen (Carlsbad, Calif.) electroporation device (11). Clones with TnphoA insertions were identified by the presence of kanamycin resistance carried on the transposon, ampicillin resistance carried on the vector, and alkaline phosphatase from the active phoA fusions.

Sequencing of DNA flanking the TnphoA insertions was performed with Sequenase version 2.0 (Amersham). The sequencing primer used was a 20-bp DNA fragment (5′-GACGAGTCCCGCTATAATGA-3′) designed to anneal to the 5′ end of the IS50 element of the TnphoA transposon and read into the flanking DNA. The sequences were analyzed by using BLAST (3).

Serum sensitivity.

Fresh pig serum was collected and aliquoted. To inactivate complement, serum was heated at 56°C for 60 min (36). S. typhimurium cells were grown overnight, and cells (10⁸/ml) were incubated at 37°C for 30 or 60 min with equal volumes of either PBS, complete serum, or heat-inactivated serum. Samples were serially diluted and plated on LB agar to measure viable cells.

LPS extractions.

LPS samples were extracted in hot phenol buffered with 20 mM morpholinepropanesulfonic acid (MOPS; pH 6.9) following the methods described by Slauch et al. (46). Samples were resuspended in 150 mM NaCl–20 mM MOPS (pH 6.9) and incubated with 100 U of DNase I and 5 U of RNase at 37°C for 1 h, followed by incubation with 3 mg of proteinase K at 65°C for 3 h. The LPS was separated on SDS-12% polyacrylamide gels at constant voltage (200 V) until the bromophenol-blue tracking dye reached the bottom of the gel and was silver stained by the method of Tsai and Frasch (50).

Biochemical assays.

The amount of 2-keto-3-deoxyoctulosonic acid (KDO) was measured by the method of Karkhanis et al. (24), in which absorbance was measured spectrophotometrically at 548 nm. The amount of total carbohydrate in LPS samples was assayed colorimetrically by hydrolyzing in concentrated sulfuric acid containing 5% phenol, and absorbance was measured spectrophotometrically at 480 nm (37).

Phagocyte uptake and killing assays.

Uptake and intracellular survival of all S. typhimurium strains were measured as previously described (21, 27) with leukocytes isolated from freshly drawn porcine blood.

Slide agglutination assay.

Antiserum in 6-week-old rabbits against envelope proteins extracted from adhesive S. typhimurium ι519 was previously prepared. The antiserum was absorbed with cells of the nonadhesive phase-variant S. typhimurium ι518 as previously described (21) to eliminate agglutinating components in the serum common to both phenotypes. The absorbed anti-ι519 serum exclusively agglutinates ι519 cells and not ι518 cells. Throughout this investigation, all strains were periodically checked with this anti-ι519 serum to confirm that the cells were in the proper phase (i.e., ι518 cells do not agglutinate while ι519 cells do agglutinate).

RESULTS

Colony morphologies.

Colony morphology differences between the S. typhimurium adhesive- and nonadhesive-phase variants ι519 and ι518 were initially visualized on blood agar as being different in size and mucoid texture. S. typhimurium ι518 exhibits small, less mucoid, cohesive colonies, while S. typhimurium ι519 colonies are larger and more mucoid in texture (21). Additional colony morphology differences between adhesive and nonadhesive S. typhimurium phase variants were observed when colonies were left at 25°C for 5 days after an overnight incubation at 37°C (Fig. 1). At that time, the colonies became quite large (diameter, 8 to 10 mm). Nonadhesive ι518 had a crinkly or rugose texture (Fig. 1A), while adhesive ι519 maintained a smooth appearance on LB agar plates (Fig. 1B). The outer texture of ι518 colonies also appeared thinner and less mucoid than ι519 colonies.

FIG. 1.

Wild-type nonadhesive ι518 (A) and adhesive ι519 (B) cells growing on LB. ι518 has a rough, crinkly texture after growth on LB for 5 days, while ι519 remains smooth.

Screening of TnphoA mutants.

Previously, we demonstrated that ι519 produces 10 to 15 unique proteins not produced by ι518 (21). An objective of the present study was to identify TnphoA insertions in one of the nonfimbrial, phase-variable genes by comparing envelope extracts prepared from individual mutants from the TnphoA mutant bank with extracts prepared from wild-type cells in the adhesive (ι519 and ι519′) and nonadhesive (ι518) phases. Envelope proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE) and mutants that lost one of the unique proteins of the adhesive-phase cells were sought. With a 12% polyacrylamide gel, one mutant was identified by this procedure and was shown to be lacking proteins with estimated molecular masses of 15 and 17 kDa (Fig. 2) (several smaller proteins also appeared to be missing in this mutant). This mutant was designated mutant #55 (Fig. 2, lane E). Mutant #55 retained its ability to be agglutinated by the absorbed anti-ι519 antiserum, demonstrating that the loss of the two proteins was due not to phase variation but to mutation.

FIG. 2.

SDS-PAGE of proteins extracted from ι518 (nonadhesive [lane B]), ι519 (adhesive [lane C]), ι519′ (adhesive [lane D]), and mutant #55 (lane E). The arrows indicate the locations of the proteins that are specific to ι519 and are absent in mutant #55. The upper arrow points to the 17-kDa protein, while the lower arrow points to the 15-kDa protein. Molecular weight standards are shown in lane A, and their molecular weights are indicated.

To verify that mutant #55 contained only a single TnphoA insertion, total DNA was obtained, digested with the restriction endonuclease KpnI or SstI, and subjected to Southern hybridization with a 7.7-kb SspI DNA fragment from TnphoA as a probe. Only a single DNA fragment hybridized with this probe (data not shown), which is consistent with there being only one copy of TnphoA inserted in to the chromosome.

Genetic identification of mutants.

Mutant #55 was analyzed to determine the location of the TnphoA insertion. To do this, the TnphoA in mutant #55 and flanking DNA was cloned into the vector pGem-4z as described in Materials and Methods and subjected to DNA sequencing. Based on an analysis of this sequence, the TnphoA element in mutant #55 was shown to have inserted into the gene rfaL (GenBank accession no. M73826). rfaL encodes the enzyme O-antigen ligase, which is essential for the biosynthesis of O antigen. It catalyzes the addition of O-antigen subunits to the growing O-antigen chain (25, 31, 32).

To demonstrate that rfaL expression was controlled by phase variation, phase variants of mutant #55 were sought. The phase variants were identified by the characteristic differences in colony morphologies. Cells from each phase subsequently were plated on XP plates. Colonies containing cells in the adhesive phase were blue, indicating the production of PhoA, and thus RfaL, while colonies containing cells in the nonadhesive phase were white. The rate of shift from the adhesive to the nonadhesive phenotype was approximately 10⁻⁵, which is characteristic of this strain.

Serum sensitivity assay.

Resistance to complement has been associated with the production of a complete O antigen in Salmonella (22). The identification of rfaL as a potentially phase-variable gene led us to determine whether mutant #55 and nonadhesive-phase cells (ι518) were sensitive to complement. As expected, mutant #55 was sensitive to complement (Fig. 3), since it contained a mutation in rfaL and, thus, lacked the ability to produce O antigen. There was also a clear difference in serum sensitivity between the adhesive and nonadhesive phenotypes. Nonadhesive ι518 cells were as sensitive to complement as were mutant #55 cells. Only 2% of the inoculum of mutants #55 and I518 survived after exposure to serum containing active complement for 2 h (Fig. 3). On the other hand, ι519 was completely serum resistant when it was incubated with complete serum for identical lengths of time. Incubation with PBS or heat-inactivated serum had no effect on the viability of any of the strains. When a smaller inoculum of bacteria was used (10²), ι519 increased in concentration while ι518 and mutant #55 were killed when exposed to 50% porcine serum under the same assay conditions (data not shown).

FIG. 3.

Sensitivity of nonadhesive ι518, adhesive ι519, and rfaL mutant #55 to porcine serum. Overnight cultures were incubated with equal volumes of PBS, 50% porcine serum, or heat-inactivated 50% porcine serum.

LPS structures.

The initial observation of colony morphology differences between adhesive and nonadhesive cells, the results indicating substantial differences in susceptibility to serum complement killing, and the identification of rfaL as a potentially phase-variable gene led to the hypothesis that nonadhesive ι518 cells contained a short O antigen and that adhesive ι519 cells contained a long O antigen. To determine if this was true, LPS from S. typhimurium phase variants ι518 and ι519 and mutant #55 was extracted, and the purified LPS preparations were separated by SDS-PAGE (Fig. 4). S. typhimurium ι518 contained an O antigen with a repeat length of only two beyond the core structure. S. typhimurium ι519 contained a typical O antigen with an obvious ladder of repeating subunits (from 1 to 17 subunits). The rough mutant #55 expressed only the LPS core structure.

FIG. 4.

SDS-polyacrylamide gel of LPS extracted from ι518 (lane A), ι519 (lane B), and mutant #55 (lane C). The gel was stained with silver. The two arrows indicate the locations of the O antigen containing one or two subunit repeats.

Biochemical assays were employed to quantify differences in the amount of O antigen between ι518 and ι519. KDO was used as a measure of total LPS core present in a given sample. Total carbohydrate content in the LPS was determined and was used as a measure of the total amount of hexose and pentose sugars present in a given LPS sample which includes the O antigen. The ratio of total carbohydrate to KDO was used to adjust for differences in LPS extraction efficiencies. Adhesive S. typhimurium ι519 had 3.2 times more carbohydrate in LPS extracts compared with nonadhesive S. typhimurium ι518 (Table 1). Collectively, these results demonstrate that ι519 produced a longer and consequently greater amount of O antigen than ι518.

TABLE 1.

Determination of the amounts of carbohydrate in O-antigen preparations

Strain	Amt of total carbohydrate (nM)	Amt of KDO (nM)	Carbohydrate/KDO ratio
ι518	30.8	1.44	21.39
ι519	156.15	2.27	68.79
Mutant #55	78.35	5.35	14.64

Phagocytic uptake and intracellular killing.

The wild-type-phase variants ι518 and ι519 were previously demonstrated to have different capacities for uptake by phagocytes and intracellular survival (21). In the present study, the requirement of rfaL for phagocyte uptake and survival was examined. Mutant #55 was taken up by phagocytes in greater numbers than both ι518 and ι519. This may be due to increased hydrophobic effects resulting from the loss of O antigen. Mutant #55 survived in phagocytes as well as ι519, while ι518 was killed (Fig. 5). Thus, the loss of RfaL did not affect intracellular survival.

FIG. 5.

Intracellular survival of ι518 (■), ι519, (•), and mutant #55 (▴) in murine macrophages. Bacterial cells were mixed with freshly collected and separated porcine leukocytes (1 leukotye/100 bacterial cells), and intracellular survival was measured as described in Materials and Methods.

DISCUSSION

The overall goal of the work described here was to understand the mechanism of long-term colonization of pig intestines by S. typhimurium resulting in persistent, asymptomatic infections. Previously, an S. typhimurium isolate obtained from a pig with diarrhea was shown to cause persistent, asymptomatic infections in pigs (55). This strain was also shown to exist in two phenotypes that were named adhesive and nonadhesive based on their abilities to attach to porcine enterocytes (21). In addition to the ability to attach to enterocytes, adhesive cells were also shown to have 10 to 15 unique proteins that were envelope associated, resistant to intracellular killing by phagocytes, and more readily phagocytized by neutrophils and macrophages. It was also shown that cells could vary between the two phenotypes at a rate consistent with this being phase variation (∼10⁻² to 10⁻⁴/generation). All traits associated with the adhesive phenotype were coregulated (i.e., all shifted in expression as a result of phase variation). Subsequent examination of other S. typhimurium isolates showed that two phase-variable phenotypes were not restricted to just the original isolate but were found with other S. typhimurium isolates isolated from pigs (20a).

In the current study, using a collection of TnphoA fusion mutants, a new phase-variable, nonfimbrial trait that was expressed only by adhesive-phase cells was identified (ι519). This mutant was identified on the basis of lacking proteins specific to the adhesive phenotype by SDS-PAGE to screen the mutant collection. The mutant was shown to contain a TnphoA fusion in rfaL, the gene that encodes O-antigen ligase. O-antigen ligase is an enzyme required for the ligation of O-antigen subunits to the growing lipopolysaccharide-based O antigen (25, 31, 32).

The mutation in rfaL was identified based on the loss of proteins with molecular masses of 17 and 15 kDa. RfaL has a molecular size of 45.9. Because other similarly sized proteins comigrate with RfaL on SDS-PAGE gels, we were not able to determine if RfaL was also missing. It is assumed that the smaller proteins that were missing somehow were relevant to the function of RfaL or the rfa operon. Data from other experiments were consistent with the mutation affecting the rfa operon. First, the mutant did not produce O antigen, and, second, the colonies appeared rough when grown on semisolid media. Consistent with rfaL expression being phase variable was the discovery that ι518, the nonadhesive phase, produced a very short O antigen (one to two subunits in length) while ι519 produced a long complete O antigen. O antigens are known to confer resistance to serum complement by inhibiting the activation of complement or sterically hindering the formation of an effective membrane attack complex. This occurs by activating complement at a distance from the bacterial membrane (22, 28). Both mutant #55 and ι518 were shown to be sensitive to porcine complement, while ι519 was not. When the mutation in rfaL was tested in porcine neutrophils and macrophages, it was shown that it did not alter the cell’s ability to survive intracellularly. The mutant did retain its parental heritage by retaining the phagocyte uptake and intracellular survival characteristics of ι519 and retained its agglutinability with the absorbed anti-ι519 antiserum.

Our results show that O-antigen production in S. typhimurium is phase variable. Whether the phase-variable production of O antigen is related to variable expression of rfaL or some other gene in the rfa operon still needs to be determined. To our knowledge, this is the first report of phase-variable expression of O antigen in S. typhimurium. There is precedence for phase-variable expression of LPS in other bacterial genera, including Vibrio (14), Neisseria (2, 43, 48), Haemophilus (42, 53, 54), and Chlamydia (30).

There is little doubt that phase variation of S. typhimurium 798 between the two phenotypes alters its virulence capabilities. The adhesive phenotype expresses more virulence properties than the nonadhesive phenotype. In addition to the phase-variable traits previously associated with the adhesive phenotype, we identified a new phase-variable trait, i.e., O-antigen synthesis. Due to the modulation of rfa, adhesive cells are resistant to serum complement while nonadhesive cells are not. Thus, virulence properties conferred by phase switching give the adhesive cells several survival advantages as they encounter host defenses in the gut, the host’s humoral immune response, and intracellular killing within macrophages and neutrophils. Phenotypic switching by S. typhimurium appears to control a large regulon involving more than one operon, including surface fimbriae (21) and LPS expression.

Environmental cues probably are important parts of this regulatory process. Altering the expression of outer membrane structures at an optimal time under precise regulation would contribute to the adhesive and immunity-evading capabilities of the bacteria. This switching ability should be advantageous within different physiological environments of the host body and is an efficient way for S. typhimurium to attain optimal virulence properties throughout the course of infection. Alternatively, the expression of unnecessary surface structures at specific sites of residence within the host body may be a waste of energy or resources for the invasive pathogen.

The regulation of virulence factors by phase variation may provide a mechanism for how S. typhimurium 798 can cause persistent, asymptomatic infections. Nonadhesive cells do not express sufficient virulence attributes to survive well in animals. Cells that do invade would be rapidly cleared by macrophages and those that do not attach or invade would be washed out. Cells expressing the entire repertoire of S. typhimurium virulence factors, some of which are regulated by phase variation, should be virulent. However, by modulating the fraction of cells in either phase, a population sufficient to maintain an in vivo population yet not enough to cause disease can be achieved. Thus, phase variation could be responsible for this phenotypic modulation.