Campylobacter fetus sap Inversion Occurs in the Absence of RecA Function

Abstract

Phase variation of Campylobacter fetus surface layer proteins (SLPs) occurs by inversion of a 6.2-kb DNA segment containing the unique sap promoter, permitting expression of a single SLP-encoding gene. Previous work has shown that the C. fetus sap inversion system is RecA dependent. When we challenged a pregnant ewe with a recA mutant of wild-type C. fetus (strain 97-211) that expressed the 97-kDa SLP, 15 of the 16 ovine-passaged isolates expressed the 97-kDa protein. However, one strain (97-209) expressed a 127-kDa SLP, suggesting that chromosomal rearrangement may have occurred to enable SLP switching. Lack of RecA function in strains 97-211 and 97-209 was confirmed by their sensitivity to the DNA-damaging agent methyl methanesulfonate. Southern hybridization and PCR of these strains indicated that the aphA insertion into recA was stably present. However, Southern hybridizations demonstrated that in strain 97-209 inversion had occurred in the sap locus. PCR data confirmed inversion of the 6.2-kb DNA element and indicated that in these recA mutants the sap inversion frequency is reduced by 2 to 3 log₁₀ units compared to that in the wild type. Thus, although the major sap inversion pathway in C. fetus is RecA dependent, alternative lower-frequency, RecA-independent inversion mechanisms exist.

Campylobacter fetus causes infertility and spontaneous abortions in ungulates, as well as intestinal and systemic diseases in both normal and immunocompromised humans (1). C. fetus surface layer proteins (SLPs) form a protective capsule for the organism (41) that is critical for virulence (27). C. fetus cells can express any of eight or nine SLPs that range in mass from 97 to 149 kDa (36). Each SLP is encoded by a promoterless sapA homolog (10), and SLP phase variation occurs by inversion of a 6.2-kb DNA segment, which contains a single outward-facing sapA promoter (8). The sapA (2) and sapA2 (12) homologs, expressing 97- and 127-kDa SLPs, respectively, flank the 6.2-kb invertible region in wild-type strain 23D. The invertible region contains an operon of four genes (sapCDEF), in which SapD, -E, and-F constitute a type I secretion system necessary for the extracellular transport of SLPs, while the function of SapC is unknown (35).

RecA is a highly conserved bacterial protein that facilitates DNA rearrangement via homologous recombination (30). In other bacterial species, including Salmonella species, Bordetella pertussis (24), and Escherichia coli (39), DNA inversion leading to phase variation has been shown to be RecA independent (6, 20, 23, 26, 29, 31). In contrast, in studies using the defined C. fetus mutant strain 23D:AC200 in which a chloramphenicol (cat) resistance cassette without a functional promoter was inserted into sapA, no inversion events could be detected after recA was interrupted (9). Complementation by recA in trans restored inversion, thus suggesting that recA function was required (11), which is consistent with the long (approximately 625-bp) 5′ conserved regions of DNA identity in each of the sap homologs.

However, following recent experimental infection of sheep with a C. fetus recA mutant, one strain (97-209) was recovered in which the molecular mass of the expressed SLP had shifted (19). The aim of the present study was to investigate this observed shift in SLP expression. Two mechanisms were considered: reversion of the recA mutation in 97-209 or occurrence of a DNA rearrangement in the absence of RecA function. We demonstrated that the recA mutation had been maintained in strain 97-209 during in vivo passage but that sap rearrangement had occurred. We found that although RecA function is critical for high-frequency inversion, RecA-independent sap inversion can occur at a lower frequency.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The C. fetus strains used in this study are listed in Table 1. In prior studies (19), a pregnant ewe was challenged with recA mutant 97-211, which expresses a 97-kDa SLP. Although all of the isolates recovered from the animal were expected to express a 97-kDa SLP, as observed in 97-210, one (97-209) expressed a 127-kDa SLP. All C. fetus cells were grown on brucella agar containing polymyxin B sulfate (7,000 U/ml), vancomycin (10 μg/ml), nalidixic acid (15 μg/ml), trimethoprim lactate (10 μg/ml), and, when required, kanamycin (30 μg/ml). Strains were cultured at 37°C under microaerobic conditions. All isolates were kanamycin resistant, reflecting the presence of aphA in recA, and Southern hybridization studies indicated that all shared the same genetic background.

TABLE 1.

C. fetus strains used in this study

Strain	recA genotype	Dominant SLP expression (kDa)	Description
23D	+	97	Wild type
23B	+		Spontaneous mutant of 23D
97-209	−	127	recA::aphA, recovered from ewe placenta
97-210	−	97	recA::aphA, recovered from ewe vagina
97-211	−	97	recA::aphA, inoculated into pregnant ewe
97-209s	−	127	MMS survivor
97-211s	−	97	MMS survivor

Immunoblot assay.

Whole-cell preparations of strains 23D, 23B, 97-209, and 97-211 were prepared as described previously (32). Protein concentrations were determined by bicinchoninic acid assay (Pierce, Rockford, Ill.), and 1 μg of protein was assayed by electrophoresis on a sodium dodecyl sulfate–7% polyacrylamide gel. S-layer proteins were detected with polyclonal rabbit serum (1:10,000 dilution) raised against the C. fetus type A 97-kDa SLP, as described previously (28), using goat anti-rabbit immunoglobulin G–alkaline phosphatase (Boehringer Mannheim, Indianapolis, Ind.) (1:2,000 dilution) as the secondary antibody. These antibodies recognize C. fetus SLPs of all molecular masses (28, 38). Strains 97-211s and 97-209s also were assayed by immunoblotting to determine whether any change in SLP phenotype had occurred.

Serum susceptibility assay.

Cells of selected C. fetus strains were harvested from brucella agar plates after 48 h and resuspended in saline (pH 7.0) with 1 mM calcium chloride, as described previously (5). Bacterial suspensions were diluted 10-fold from 10⁻¹ to 10⁻⁷, and the 10⁻⁴ to 10⁻⁷ dilutions were incubated in the presence of 10% normal human serum (NHS) or heat-inactivated NHS at 37°C in an atmosphere with 5% CO₂ for 60 min. Triplicate samples were then inoculated on blood agar plates, and colonies were counted after 72 h as described previously (10).

Sensitivity to MMS.

Bacterial cells were harvested after 48 h of growth and resuspended in 3 ml of brucella broth. The cell suspension then was diluted 10-fold, and 1 ml was incubated either in brucella broth with 0.05% methyl methanesulfonate (MMS) or in broth alone for 60 min 37°C with gentle shaking. Serial 10⁻⁴ to 10⁻⁷ dilutions of the cell suspensions were inoculated in triplicate onto blood agar plates. Cells were sampled both before and after exposure to MMS. Plates were incubated at 37°C for 72 h under microaerobic conditions, and then colonies were counted to determine sensitivity to MMS. Colonies of strains 97-211 and 97-209 that survived MMS incubation were recultured and again exposed to MMS, and survival frequencies were determined as described above.

Southern-hybridization.

C. fetus chromosomal DNA was prepared using the Wizard Genomic DNA Purification Kit (Promega, Madison, Wis.), digested with either NdeI or PstI, electrophoresed on a 0.7% agarose gel, and transferred to a nylon membrane (MSI, Westborough, Mass.)(12,13). The hybridization probes used were PCR products specific for recA (primers B9088 and B9089) (Table 2), the sapA 5′ conserved region (primers B9343 and AN1199), sapF (primers A7308 and 7315), and the gel-purified aphA fragment from pILL600 Smal digestion (25). Probes were labeled using the Renaissance chemiluminescent nonradioactive kit (NEN Research Products, Boston, Mass.).

TABLE 2.

PCR Primers used in this study

Name	Gene or designation	Directiona	Location in gene	Genbank accession no.	Sequence (5′ → 3′)b
B9088	recA	F	561–581	AF020677	GCGTACCAAAAGGAAGAATAG
B9089	recA	R	1410–1390	AF020677	TCTACTGCACCGCTCATTATG
A7308	sapF	F	4752–4772	AF027405	GCTAGTATGTATGAAAATTTA
A7315	sapF	R	5320–5300	AF027405	AAGCTAAGATCCATATTTTCA
B9341	sapA C terminus	F	2014–2033	J05577	AGCTTATTACAGTGAAACTA
B9342	sapA C terminus	R	2775–2758	J05577	GATCTAGCGTACCTGAAA
B9339	sapA2 C terminus	F	2889–2909	S76860	GATGATGCATTAACAATAATA
B9340	sapA2 C terminus	R	3211–3194	S76860	GCAGTGTCTGGAGTAACG
B9337	promF2	F	644–626	S44580	CGATAGTATTTTTGCAAAT
B9338	promR2	R	141–158	S44580	TATGCAATACATCTTCAT
B9343	sapAcon3	F	645–665	S44580	ATAGTAAGGTAAGCAATCCGT
AN1199	sapAcon4	R	590–569	J05577	AGGTAGACGCGTAAGTCGACGTCTCACTCTTCAAAGCATCAATC
B6816	sapFF	F	5907–5927	AF027405	ACTATTAGAAATTTAGAAAGA
A9237	sapA2R	R	3767–3747	S76860	AGCTACTGTGATTGTATTAGC
A9238	sapAR	R	2783–2763	J05577	AAGTTTAAGATCTAGCGTACC
BA3517	sapApromF2	F	596–621	S44580	TATAAAAAATTATGTTATAATTCGCG
C9294	recAF	F	1204–1228	AF020677	CAGCAAAGAAGGAGAGATAATAGAT
C9295	enoR	R	2007–1983	AF020677	CTTTTTTAATGTTTGATATACTTCG

^aF, forward; R, reverse. 

^bRestriction digestion sites are underlined. The sequences are as follows: MluI, ACGCGT; SalI, GTCGAC. 

PCR.

To determine whether aphA remained present in recA (11), chromosomal DNAs from selected strains were amplified with recA primers BA1916 and BA1917, which flank the original aphA insertion site (Table 2). To detect inversion of the C. fetus invertible region, chromosomal DNAs from selected strains were amplified using primer sapApromF2 (forward-facing promoter primer) with either sapAR (reverse-facing sapA-specific primer) or sapA2R (reverse-facing sapA2-specific primer) (Table 2). To determine the frequency of inversion of the invertible region, 10-fold (10⁰ to 10⁻⁷) dilutions of C. fetus chromosomal DNA were amplified with sapAR and either sapApromF2 or sapFF (forward-facing sapF primer). Primers from the recA/eno region downstream of the aphA insertion site were used in a parallel control PCR.

RESULTS

SLP expression of C. fetus cells.

We first sought to confirm that the C. fetus strain (97-209) recovered from the pregnant ewe was indeed expressing a different SLP than the challenge strain, 97-211. Immunoblotting of this strain indicates the expression of a 127-kDa SLP, not the 97-kDa SLP of wild-type strain 23D or of strain 97-211, which was inoculated into the ewe (Fig. 1A). Although for 23D there was a minor 127-kDa band, for strains 97-209 and 97-211, only single SLP bands were present, consistent with their presumed recA phenotype (11). The strains surviving incubation with MMS (97-211s and 97-209s; see below) showed the same SLP expression as the strains from which they were derived. To confirm that the SLPs were present on the cell surface, the C. fetus cells were tested for serum resistance by incubation with NHS or with heat-inactivated NHS to control for nonspecific killing. As expected, the control wild-type strain 23D was serum resistant (<1.0 log₁₀ unit killing) whereas S⁻ strain 23B was highly sensitive (Fig. 1B). Challenge strain 97-211 and the two strains recovered from the ewe all were serum resistant, confirming SLP expression on their cell surface (3).

FIG. 1.

Phenotypic characterization of C. fetus recA strains. (A) Identification of SLPs in whole-cell preparations of C. fetus strains by immunobloting with polyclonal rabbit serum against C. fetus SLPs. Lanes: a, 23D (wild type); b, 97-211; c, 97-209; d, 97-211s (strain surviving incubation with MMS); e, 97-209s (strain surviving MMS); f, 23B (spontaneous S−strain). A 97-kDa SLP is present in lanes a, b, and d. A major 127-kDa SLP is present in lanes c and e. Immunoblotting indicates that strain 97-209 has changed to expression of a 127-kDa SLP, and only a single SLP predominated for each strain. Immunoblotting of strains 97-211s and 97-209s indicates that SLP expression was not affected by exposure to MMS. (B) Susceptibility of C. fetus strains 23D (S⁺), 23B (S⁻), 97-211, 97-210, and 97-209 to NHS. Strains 97-211, 97-210, and 97-209 were highly resistant to serum killing, consistent with the expression of SLPs on their cell surfaces. (C) Sensitivity of C. fetus to MMS, using strains 23D (recA⁺), 97-211 (recA), 97-211s (97-211 survivor of MMS incubation), and 97-209s (97-209 survivor of initial MMS incubation). Results represent log₁₀ killing after incubation of C. fetus cells at 37°C for 60 min in the presence of 0.05% MMS. For each strain, the mean (± standard deviation) log₁₀ kill is shown for triplicate determinations. Strains 97-211 and 97-209 are highly sensitive to MMS, indicating a lack of RecA function. For each of the assays of strains 97-211 and 97-209, several colonies were observed after incubation with MMS. These survivors (97-211s and 97-209s) showed the same phenotype (SLP expression and susceptibility to MMS) as their parental strains.

Susceptibility of cells to MMS.

Previous studies have shown that wild-type C. fetus cells survive treatment with the mutagenizing agent MMS, whereas recA strains are highly sensitive (11). Our studies confirmed that recA strain 97-211 was substantially more sensitive to MMS than was wild-type strain 23D (Fig. 1C). Strain 97-209, which changed to expression of a 127-kDa SLP (Fig. 1A) also was highly sensitive to MMS, consistent with a RecA phenotype. For both strains 97-211 and 97-209, several colonies survived a 60-min incubation with MMS. A representative colony from each (97-211s and 97-209s, respectively) was picked and reincubated with MMS to determine whether there had been selection for MMS resistance or whether their presence merely reflected the limits of the assay system. These MMS survivors were as susceptible to MMS as were their parental strains (Fig. 1C), indicating no selection for MMS resistance. These results indicate that during the ovine passage, there had been no change from the original RecA phenotype (11).

recA genotype of C. fetus strains.

To assess the recA genotype of the ewe isolates, both Southern hybridizations and PCR analyses were used. Southern hybridization using both recA and aphA probes indicated that aphA was stably integrated within recA in both the strain used to challenge the ewes (97-211) and the strains recovered (97-209 and 97-210), regardless of their SLP phenotype (Fig. 2). PCR using recA primers B9088 and B9089 (Table 2) showed a 2.0-kb product in wild-type strain 23D and 3.4-kb products in strains 97-210, 97-209, and 97-211, confirming the presence of aphA within recA (data not shown). In total, both the phenotypic and genotypic data indicate that in strain 97-209, the change in SLP phenotype occurred in a recA background.

FIG. 2.

Analysis of the recA mutation in C. fetus strains by Southern hybridization of PstI-digested chromosomal DNA. The probes used are specific for recA (left panel) and aphA (right panel), and the strains examined are 23D (wild type), 97-211, 97-210, and 97-209. The sizes (in kilobases) of hybridizing fragments are indicated at left. The results show that the aphA cassette is stably present within recA in strains 97-211, 97-210, and 97-209.

Southern hybridization analyses provide evidence for sap inversion in recA strain 97-209.

We next sought to determine the mechanism for the change in SLP phenotype in recA strain 97-209. To accomplish this, we performed a series of Southern hybridizations using strain 97-209 and relevant controls to determine whether the changed phenotype could be explained by sap locus inversion. In Southern hybridizations of PstI-digested chromosomal DNA, if inversion of the 6.2-kb invertible region had occurred such that sapA2 now was downstream of the unique sap promoter and able to express the 127-kDa SLP, with use of a probe to the sapA 5′ conserved region, the loss of 4.8- and 5.7-kb hybridizing fragments and the gain of 6.9- and 3.5-kb fragments would be expected. The C. fetus strains expressing a 97-kDa SLP (23D, 97-210, and 97-211) showed the 5.7- and 4.8-kb bands, as expected, but these were absent in 97-209 and a new band at 3.5 kb was seen (data not shown). When an NdeI digestion of chromosomal DNA was done and the same sapA probe was used, a shift from a 1.2- to a 1.5-kb hybridizing band was expected in strain 97-209 but not in the other strains, and this was clearly observed (data not shown). Since use of the sapA probe is associated with multiple bands, hybridization of PstI-digested chromosomal DNA also was done using a probe to sapF, which is present in a single copy within the invertible DNA fragment. Rearrangement permitting expression of the sapA2 product would be expected to produce a shift from a 4.8- to a 3.5-kb sapF-hybridizing fragment (Fig. 3A), and this was observed for strain 97-209 but not for the control strains (Fig. 3B). Therefore, these studies indicated that the change in SLP expression in strain 97-209 was due to inversion of the 6.2-kb invertible region, which placed sapA2 downstream of the unique sap promoter.

FIG. 3.

Analysis of the sap invertible region in C. fetus strains by Southern hybridization of PstI-digested DNA. (A) Postulated restriction maps of strains 23D (97-kDa SLP) and 97-209 (127-kDa SLP). Arrows indicate direction of transcription, and diagonal bars indicate 5′ conserved region of sap homologs. PstI sites (P) are shown, and brackets indicate expected products when using a sapF probe (black bar). (B) Southern hybridization of PstI-digested chromosomal DNAs from C. fetus strains 23D (97 kDa), 97-209 (127 kDa), 97-210 (97 kDa), and 97-211 (97 kDa) using the sapF probe.

PCR confirmation of inversion of the sap invertible region and estimation of its frequency.

Since inversion of the 6.2-kb invertible region is believed to occur spontaneously in wild-type strains, we sought to test this phenomenon in strain 23D. Using primer sapApromF2, which is a forward-facing sapF primer in the region of the unique sap promoter, and reverse primers specific to either sapA or sapA2, we were able to obtain PCR products, indicating that routine culture of 23D resulted in a mixture of cells with the invertible region in either orientation (data not shown). Using these same primers for the recA strains 97-211 and 97-209 also showed that both products were present, confirming that populations of cells with the 6.2-kb invertible region in either orientation were present in both strains.

To estimate the frequency of inversion, these PCRs were repeated using serial dilutions of template DNA. A control PCR for a chromosomal locus (recA/eno), which does not rearrange, also was used (Fig. 4). For each strain studied, we could detect the recA/eno product in 10⁻⁶ dilutions of the chromosomal DNA template. For wild-type strain 23D, the frequency of inversion was approximately 10⁻¹, comparing the results of the two competing sapApromF2-based PCRs (Fig. 4). In contrast, for recA strains 97-211 and 97-209, the frequencies were about 10⁻³ and 10⁻⁴ respectively, with the number of positive dilutions reflecting the strain's phenotype (expression of a 97-kDa protein favors the reaction with the sapAR primer, whereas expression of the 127-kDa protein favors the reaction with the sapA2R primer) (Fig. 4). These results indicate that mutation of recA substantially diminished but did not eliminate the sap invertible-region inversion.

FIG. 4.

Quantitative PCR of C. fetus DNA inversion. Reciprocal 10-fold dilutions (10⁰ to 10⁻⁷) of chromosomal DNAs from strains 23D (wild type, 97-kDa SLP), 97-211 (97-kDa SLP, recA), and 97-209 (127-kDa SLP, recA) were amplified with sapApromF2 (forward-facing sap promoter primer) and either sapAR (reverse-facing sapA-specific primer) or sapA2R (reverse-facing sapA2-specific primer). The same samples also were amplified using primers recAF and enoR to control for DNA template dilutions. The lanes indicate dilutions from 10⁰ to 10⁻⁷. For strain 23D, PCR products were observed at about 1 dilution further for primers sapApromF2 and sapAR, consistent with the dominant expression of the 97-kDa sapA product. For strain 97-211, primers sapApromF and sapAR yielded products about 3 dilutions further than did primers sapApromF2 and sapA2R, also consistent with its expression of a 97-kDa SLP. For strain 97-209, there was about a 4-log-unit difference but with more dilutions for sapApromF2 and sapA2R, consistent with its expression of the 127-kDa sapA2 product.

DISCUSSION

In this study, we confirmed that C. fetus strain 97-209, recovered from an experimentally infected ewe, expressed a 127-kDa SLP rather than the 97-kDa SLP expressed in the challenge strain (19). That strain 97-209 was serum resistant (4) is indicative of the surface localization of this protein (18). These results demonstrate that the shift in SLP expression that had occurred in vivo was not to produce a strain lacking SLP encapsulation (15, 16) but rather to an S⁺ variant expressing an alternative SLP.

Studies of susceptibility to MMS demonstrated that the variant strain 97-209 maintained the RecA phenotype, and both Southern hybridizations and PCR analyses showed that the original recA genotype of the challenge strain (97-211) remained in 97-209. Further, the few survivors of incubation with MMS showed a fully susceptible phenotype, which indicates that rather than selecting for MMS resistance, they represent the chance survivors at the limit of assay efficacy, rather than reversion to wild-type RecA function.

Given that RecA previously has been demonstrated to be necessary for SLP switching, how then did C. fetus change its phenotype from expression of a 97-kDa SLP to expression of a 127-kDa SLP? Again, both Southern hybridization and PCR analyses are consistent in indicating that DNA inversion, involving the 6.2-kb invertible region had occurred in strain 97-209. DNA inversion involving the sap invertible region is known to occur spontaneously, and in C. fetus mutant strain 23D:ACA2K101 (sapA::cat sapA2::aphA) it was estimated to occur at a frequency of 10⁻⁴ to 10⁻³ (9). This rate of inversion recently has been confirmed in both PCR analyses and studies of (phenotypic) shifts in antibiotic resistance for that strain (Z.-C. Tu et al., unpublished data). For wild-type strain 23D, our PCR dilution results indicate that the frequency of inversion is about 10⁻¹ (Fig. 4), which is approximately 100- to 1,000-fold greater than the previously published result (11). Both results have been confirmed in multiple experiments. One explanation for this 2- to 3-log₁₀-unit difference in inversion rate between wild-type strain 23D and mutant strain 23D:ACA2K101 is that in the latter strain, antibiotic resistance cassettes (cat and aphA in sapA and sapA2, respectively) were inserted in the sapA 5′ conserved regions. Their presence at that location might interfere with the inversion process, especially since it substantially disrupts the DNA homology that would be required for RecA-requiring recombination events. For E. coli, the frequency of homologous recombination depends in part on the length of the homologous sequence (7, 18, 22, 33, 34, 40). Thus, our present studies of strain 23D (Fig. 4) provide analysis of spontaneous in vitro inversion frequency in a strain with wild-type rather than mutated sapA homologs.

Our results further indicate that sap inversion continues to occur in the recA strains, but at a frequency also about 2 to 3 log₁₀ units lower than for the wild-type strain (Fig. 4). Thus, while RecA function is critical for high-frequency (10⁻¹ to 10⁻²) inversion, there is a residual recA-independent inversion mechanism that operates at lower frequency. In other bacteria, recA-independent (site-specific) inversion is well recognized (14, 21). That at least two independent pathways can lead to inversion of the sap invertible region is an indication of the functional significance of SLP antigenic variation for C. fetus. These results suggest that the ability to change SLP expression is a critical in vivo function for C. fetus (36). Thus, the recovery of strain 97-209 during the course of experimental ovine infection (19) could reflect strong in vivo selection for the spontaneous variant.

The residual recA-independent inversion mechanism could reflect general (homologous) recombination using other pathways, analogous to the recBCD and recEF pathways of E. coli (14, 21, 37), for example. Alternatively, the inversion could reflect a site-specific mechanism involving a unique enzymatic activity. Studies to examine these possibilities are under way.