A New Member of the S-Layer Protein Family: Characterization of the crs Gene from Campylobacter rectus

Abstract

Strains of the periodontal pathogen Campylobacter rectus express a 150- to 166-kDa protein on their cell surface. This protein forms a paracrystalline lattice, called the surface layer (S-layer), on the outer membrane of this gram-negative bacterium. To initiate a genetic analysis of the function of the S-layer in the pathogenesis of C. rectus, we have cloned and characterized its gene. The S-layer gene (crs) from C. rectus 314 encodes a cell surface protein which does not have a cleaved signal peptide at its amino terminus. Although the amino acid sequence deduced from the crs gene has 50% identity with the amino-terminal 30 amino acids of the four S-layer proteins from Campylobacter fetus, the similarity decreases to less than 16% over the rest of the protein. Thus, the crs gene from C. rectus encodes a novel S-layer protein whose precise role in pathogenesis may differ from that of S-layer proteins from other organisms. Southern and Northern blot analyses with probes from different segments of the crs gene indicate that the S-layer gene is a single-copy, monocistronic gene in C. rectus. RNA end mapping and sequence analyses were used to define the crs promoter; there is an exact match to the Escherichia coli −10 promoter consensus sequence but only a weak match to the −35 consensus element. Southern blots of DNA from another strain of C. rectus, ATCC 33238, demonstrated that the crs gene is also present in that strain but that there are numerous restriction fragment length polymorphisms in the second half of the gene. This finding suggests that the carboxy halves of the S-layer proteins from strains 314 and 33238 differ. It remains to be determined whether the diversities in sequence are reflected in functional or antigenic differences important for the pathogenesis of different C. rectus isolates.

Campylobacter rectus (formerly Wolinella recta), a gram-negative, anaerobic bacterium, has been strongly implicated in the etiology of adult periodontitis (31), rapidly progressive periodontitis (11), and periodontitis associated with certain diseases such as AIDS and diabetes (40, 41). However, the pathogenic mechanisms of C. rectus are poorly characterized. One strong candidate for a C. rectus virulence determinant is the paracrystalline cell surface layer (S-layer), which appears to be composed of a single protein (21, 26). Although the first 15 amino acids of the S-layer proteins are identical in several strains of C. rectus, the molecular mass of the S-layer protein varies from 150 to 166 kDa among strains (30). Nitta et al. (30) have also shown, by peptide analysis, that there is internal sequence heterogeneity between the S-layer proteins from strains 314 and ATCC 33238.

Evidence that the C. rectus S-layer is a virulence factor stems from studies of a strain of C. rectus that lost its S-layer during long-term in vitro subculture (7). The S-layer negative cells were more adherent to human gingival fibroblasts than were other strains of C. rectus which had their S-layers (7). In addition, strains which had been passaged 15 to 17 times in vitro formed smaller lesions in a mouse abscess model for soft tissue destruction than did low-passage C. rectus strains (19). These studies have led to the proposal that the C. rectus S-layer helps the organism evade host defense mechanisms. However, the results need to be interpreted cautiously since comparisons were being made between unrelated strains and because the levels of proteins other than the S-layer protein are also different between low-passage and high-passage C. rectus cells (7).

A role for the S-layer in pathogenesis has been shown for other bacteria. For example, the S-layer of the fish pathogen Aeromonas salmonicida protects the bacterial cells against proteolysis and complement lysis and is required for macrophage resistance (14, 15, 29, 36). Similarly, Campylobacter fetus, which causes bovine infertility, has an S-layer that makes the organism resistant to phagocytic uptake and to the bactericidal activity of serum (3, 5, 27). Although the S-layer is involved in evading host defense mechanisms in these two bacteria, the precise roles of the two S-layers are quite distinct. This is not necessarily surprising given that the S-layer proteins from C. fetus and A. salmonicida are not homologous (25).

To begin to determine the role of the C. rectus S-layer in pathogenesis, the gene encoding this putative virulence factor has been cloned, sequenced, and characterized. The C. rectus crs gene from strain 314 encodes an S-layer protein with limited amino acid sequence similarity to S-layer proteins from other bacteria. The S-layer gene is single copy and part of a single-gene operon. Although the gene is present in another strain, 33238, there are many restriction site polymorphisms between the two strains. The polymorphisms are limited to the second half of the crs gene, and the 33238 restriction map is consistent with that reported previously by Miyamoto et al. (28). These results suggest that the precise function of the C. rectus S-layer in pathogenesis may differ from the molecular roles for the S-layers from other organisms.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

C. rectus 314 and ATCC 33238 were originally clinical isolates (7, 35). C. rectus ATCC 33238 (S⁻) is a variant of the ATCC 33238 strain whose S-layer was lost spontaneously during in vitro passage (7). C. rectus cells were grown in mycoplasma-formate-fumarate (MFF) broth (17), supplemented with 5 μg of hemin per ml and 10% horse serum, in a Coy anaerobic growth chamber (5% CO₂, 10% H₂, 85% N₂) at 37°C. The strains were maintained by transfer on MFF agar containing 5 μg of hemin per ml and 10% horse serum twice a week. Plasmid pUC19 was used as a vector for cloning. Recombinant constructs were propagated in Escherichia coli TB-1 in Luria broth medium after transformation by the CaCl₂ procedure or by electroporation (2).

Isolation and labeling of hybridization probes.

DNA fragments used as hybridization probes were isolated from agarose gels by a freeze-thaw-phenol extraction procedure described previously (34). DNA probes were labeled with [α-³²P]dATP by using a nick translation labeling kit or a random primer DNA labeling system from Life Technologies (Gaithersburg, Md.). For oligonucleotide probes, the 5′ end was labeled with T4 polynucleotide kinase and [γ-³²P]ATP (2).

DNA isolations and Southern blots.

Chromosomal DNA was isolated from C. rectus 314 by a detergent-proteinase K lysis procedure that included treatment with cetyltrimethyl ammonium bromide to remove polysaccharides and cell debris (2). Plasmid DNAs from E. coli were prepared by a miniprep method involving alkaline lysis and boiling (2).

Southern blot hybridizations were done as described previously (23). Hybridizations normally were carried out overnight at 65°C for DNA fragment probes and at 42°C for oligonucleotide probes (22). The filters were washed three times in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) at 65°C after DNA fragment probe hybridizations and in 6× SSC at 42°C after oligonucleotide probe hybridizations. For low-stringency experiments with the DNA fragment probe, the hybridization and washes were done at 55°C in 2× SSC.

Cloning the crs gene.

Oligonucleotides CR102 and CR103 (5′-GCICCYTCIGGIACRTCICCRAA-3′ and 5′-GCIYTIACICARACICAAGT-3′, where I = inosine, Y = C or T, and R = A or G) were designed from the previously determined sequence of the amino-terminal 22 amino acids of the purified S-layer protein (30). The primers, 2.5 μg of each, were used in a PCR with 0.5 μg of genomic DNA from C. rectus 314 as a template. After an initial 7-min denaturation step at 93°C, the reaction was amplified for 39 cycles (2 min at 93°C, 1 min at 45°C, and 1 min at 72°C) in a model PTC-100 thermocycler from MJ Research. The 65-bp PCR product generated was cloned into pUC19 and sequenced. A 22-base non-primer-derived sequence from the middle of the PCR fragment was used to design another oligonucleotide, which was used as a hybridization probe to clone a 3.0-kb SacI DNA fragment from a limited genomic library (23) of SacI-digested, size-separated C. rectus 314 DNA in pUC19. DNA sequence analysis of part of the clone, pDK572 (Fig. 1), indicated that it only contained the beginning of the crs gene.

FIG. 1.

Restriction endonuclease maps of the S-layer gene region from C. rectus 314 and ATCC 33238. Key restriction endonuclease sites are marked, but all restriction endonuclease sites are not shown. Restriction endonuclease sites that differ between the two strains are underlined. The extent of the coding region of the crs gene, determined by sequencing, is delimited by vertical lines. The direction of transcription of the S-layer RNA is indicated by the large arrow, as is the position of the RNA start site. The positions of the probes used in Southern and Northern blot analyses are indicated by the lines labeled I, II, and III. The size of the DNA fragment used in the S1 nuclease analysis of the RNA start site is shown by the small arrow labeled S1 probe. B, BstEII; Bs, BsaI; C, ClaI; H, HindIII; P, PstI; S, SacI; Sc, ScaI.

To clone more of the coding region of the crs gene, a 170-bp PstI/EcoRI fragment from clone pDK572 was used as a hybridization probe of a ClaI-digested, size-separated limited genomic DNA library. Clone pDK580 (Fig. 1) contained the first half of the crs gene on a 1.9-kb ClaI fragment in the AccI site of pUC19. Finally, the second half of the gene was cloned from a SacI-limited genomic DNA library, on a 7.3-kb SacI fragment, using a 150-bp PstI fragment from the most crs promoter-distal region of the insert in pDK580 as probe. This clone was designated pDK594 (Fig. 1).

DNA sequencing and computer analysis.

The nucleotide sequence of the S-layer gene was determined by the dideoxy-chain termination method. Portions of some clones were sequenced by using a Sequenase sequencing kit (United States Biochemical), but the majority of the DNA was sequenced in the Center for Advanced Technology at the University of Texas Health Science Center at San Antonio, using an Applied Biosystems model 373A sequencing system. All sequences were determined independently from both strands. The deduced amino acid sequence of the C. rectus S-layer protein was compared to the nonredundant protein database of the National Center for Biotechnology Information, using the BLAST searching algorithm (1). The Multalin 4.0 software (Cherwell Scientific) was used to find the best alignment between the C. rectus and C. fetus S-layer proteins.

RNA isolation and analyses.

C. rectus RNA was prepared by a sodium dodecyl sulfate lysis-CsCl cushion procedure previously used for isolation of RNA from Actinobacillus actinomycetemcomitans (24). RNAs were isolated from 50-ml samples of C. rectus cells in the mid-logarithmic phase of growth (optical density of 0.2 to 0.4 at 660 nm). Northern blots were performed as previously described (12).

S1 nuclease analyses were done as described previously (8, 24), with minor modifications. The DNA probe was end labeled with T4 polynucleotide kinase and [γ-³²P]ATP, and about 20 ng of probe was mixed with 5 to 45 μg of RNA in 15 μl of hybridization buffer [0.83 M NaCl, 1.7 M piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES; pH 6.5)]. After being boiled for 3 min, the samples were incubated at 67°C for 60 min, and 300 μl of iced S1 nuclease buffer and 983 U of S1 nuclease were added. The samples were incubated at 37°C for 30 min and then extracted with phenol and precipitated with ethanol. The pellets were dissolved in 3 μl of TE buffer (10 mM Tris [pH 7.4], 1 mM EDTA), 2 μl of sequencing sample buffer was added, and the samples were electrophoresed on 6% polyacrylamide-urea gels after being heated at 75°C for 2 min.

The protocol of Kolodrubetz et al. (24), with minor changes, was used for primer extension assays. About 0.7 pmol of end-labeled oligonucleotide primer was mixed with 1 to 45 μg of RNA in 30 μl of reverse transcriptase buffer (50 mM Tris [pH 8.3], 75 mM KCl, 3 mM MgCl₂). The mixture was heated to 85°C for 3 min and then allowed to anneal at 37°C for 90 min. One microliter of Moloney murine leukemia virus reverse transcriptase (200 U/μl; Life Technologies) and 2 μl of 0.1 M dithiothreitol were added to the samples along with each deoxynucleoside triphosphate to 1.25 mM. The total volume of the reaction was brought to 40 μl by the addition of 2 μl of 5× reverse transcriptase buffer and 3 μl of diethyl pyrocarbonate-treated water. After 1 h of incubation at 37°C, each sample was precipitated with ethanol and electrophoresed as described above.

Nucleotide sequence accession number.

The nucleotide sequence of the crs gene has been deposited in GenBank with accession no. AF010143.

RESULTS AND DISCUSSION

The crs gene from C. rectus encodes a novel member of the S-layer protein family.

The sequence of the amino-terminal 22 amino acids of the purified S-layer protein (30) was used to design two degenerate oligonucleotides for use in PCR with genomic DNA from C. rectus 314 as the template. A 65-bp PCR product encoding the amino terminus of the S-layer protein was generated, as expected. Because the oligonucleotides used to make the PCR product were degenerate, a 22-base oligonucleotide whose sequence was derived from the middle of the PCR fragment was used as a hybridization probe to clone a 3.0-kb SacI DNA fragment from a limited genomic library. Partial DNA sequencing of this clone, pDK572 (Fig. 1), indicated that it contained part of the crs gene, since near one end of the fragment, there was a partial open reading frame whose amino-terminal sequence matched the previously determined amino acid sequence of the protein. In addition to proving that the correct gene had been cloned, this result indicates that the C. rectus S-layer protein is placed on the outside of the cell without the cleavage of a signal peptide. So far, the S-layer proteins from Caulobacter crescentus and Campylobacter spp. are the only ones which are synthesized without a cleavable signal peptide (6).

The rest of the S-layer gene was then cloned, by hybridization, as two overlapping DNA fragments: a 1.9-kb ClaI fragment containing the first half of the gene (pDK580), and a 7.3-kb SacI fragment containing the second half of the gene as well as over 4.5 kb of 3′ flanking sequence (pDK594) (Fig. 1). The nucleotide sequence of the crs gene was determined independently from both DNA strands. Analysis of the DNA sequence revealed a 4.1-kb open reading frame encoding a protein of 1,361 amino acids with a calculated size of 144 kDa (Fig. 1 and 2). When the deduced amino acid sequence was used to search the protein databases with the BLAST program, the four sequenced S-layer proteins (SapA, SapA1, SapA2, and SapB) from C. fetus (4, 9, 10, 38) gave the most significant matches. However, the matches were limited to several short (<30-amino-acid long) regions. The program Multalin was used to find the best alignment of the C. rectus S-layer protein with each C. fetus protein; SapA2 gave the alignment with the most identities, 15.8% (Fig. 2). There were two small regions with more striking identity. The amino termini of the two proteins were identical in 15 of 28 amino acids, and there was a stretch, centered at position 495, in which 9 of 11 amino acids were identical between the two proteins. This latter segment of identity was not present in the other three Sap proteins from C. fetus, and the SapB protein of C. fetus matched the amino terminus of the C. rectus Crs protein at only 10 of 28 amino acids. Overall, since the S-layer protein from C. rectus shows less than 16% identity with any other S-layer protein, we conclude that the C. rectus crs gene encodes a novel S-layer protein.

FIG. 2.

Deduced amino acid sequence of the C. rectus S-layer protein from strain 314 aligned with the S-layer protein A2 from C. fetus (9). The residues that are identical between the two sequences are marked by dots. The two sequences were aligned by using the Multalin 4.0 alignment program (Cherwell Scientific). The alignment shown is the one in which gaps were minimized.

Finally, it is interesting that the first 16 amino acids of the C. rectus S-layer protein are identical to those of a 104-kDa cytotoxin secreted by C. rectus 33238, the only strain in which it was looked for (18). The amino-terminal amino acid sequence for this cytotoxin was derived from protein sequencing, and so the extent of its identity with the S-layer protein is unknown. Having the cloned crs gene will facilitate future genetic analyses to determine whether the cytotoxin and S-layer protein are made from the same gene.

crs from C. rectus is a single-copy monocistronic gene.

In C. fetus, there are multiple S-layer genes which are involved in the S-layer antigenic variation found in that organism (38). To determine whether the crs gene is present in multiple copies in C. rectus, DNA from strain 314 was analyzed by Southern blotting using three nonoverlapping probes (Fig. 1): probe I, a 1.9-kb ClaI fragment which contains the first half of the crs gene; probe II, a 0.7-kb PstI/AflII fragment from the center of the gene; and probe III, a 1.5-kb AflII fragment encompassing the last third of the gene. At normal stringency, the only significant hybridizations with each of the probes were to the DNA fragments of the sizes expected from the sequence of the crs region (Fig. 3A). These data suggest that crs is a single-copy gene in the chromosome of C. rectus, the same conclusion that Miyamoto et al. (28) reached in their Southern blot analysis of the crs gene from another strain, 33238. To extend this conclusion, the Southern blot analyses were repeated under lower-stringency conditions using two DNA fragments which together contain the entire crs gene as hybridization probes. The hybridization and washes were done at a temperature and salt concentration which we have shown previously to generate a strong hybridization signal with as little as 75% identity over 200 bp between the probe and the target DNAs (22). A comparison of the hybridization signals found at high stringency (Fig. 3A) and low stringency (Fig. 3B) shows that there are no new signals found under low stringency. This observation confirms the conclusion that the S-layer gene is a single-copy gene in C. rectus. These results are also consistent with the lack of any reports of antigenic variation of the S-layer within strain 314. Although the antigenicity of the C. rectus S-layer has been reported to differ between strains (20, 30), no antigenic differences are found when a given strain is passaged multiple times in vitro (7). This differs from what is found with C. fetus, where long-term passage does result in a shift to antigenically different S-layer proteins (39).

FIG. 3.

Hybridization of S-layer gene probes to genomic DNA from C. rectus 314. DNA (8 μg) from strain 314 was digested with the indicated restriction endonucleases (C, ClaI; E, EcoRI; H, HindIII; P, PstI), electrophoresed on a 0.75% agarose gel, and analyzed by Southern blot hybridization. (A) Each blot was hybridized with one of ³²P-labeled probes I, II, and III, which are from different segments of the crs gene (Fig. 1). The filters were hybridized and washed under normal-stringency conditions (65°C). (B) The blot was hybridized with two ³²P-labeled probes at the same time; the probes were a 1.94-kb BstEII/ClaI DNA fragment containing the first half of the crs gene and a 2.54-kb SacI/BsaI DNA fragment containing the second half of the gene. The hybridization and washings were done at a lower stringency (55°C). The faint hybridization bands are due to incomplete digestions with some restriction endonucleases. The positions of the molecular size standards are indicated.

Bacterial genes are often organized in operons containing other genes that are involved in the same metabolic or regulatory pathway. To determine whether the crs gene is cotranscribed with other genes, RNA from C. rectus was analyzed by Northern blotting. When RNA from strain 314 was hybridized with a crs probe, a 4.3-kb hybridization signal was seen (Fig. 4A). No larger band was found even after a long exposure (data not shown). Since the transcript is only 200 bases longer than the crs coding region, we conclude that the crs gene in C. rectus is monocistronic. This is similar to what has been found for the S-layer genes in other bacteria where it has been examined (25).

FIG. 4.

Expression of S-layer RNA in various strains of C. rectus. RNAs isolated from C. rectus 314, ATCC 33238 (S⁺), and ATCC 33238 (S⁻) and from A. actinomycetemcomitans Y4 (37) were used in Northern blot analyses. The two S⁻ lanes are RNAs prepared from the same strain at different times. (A) The blot was first hybridized with ³²P-labeled probe I, corresponding to a segment from the first half of the crs gene (Fig. 1). (B) The blot in panel A was stripped and then hybridized to a ³²P-labeled DNA fragment from the glyA gene of A. actinomycetemcomitans.

To identify potential promoter sequences for the crs gene, the position of the 5′ start site of the crs mRNA was determined. In S1 nuclease protection assays with a 543-bp end-labeled HindIII/PstI fragment encompassing the region from +112 to −431 (Fig. 1), two RNA-dependent bands were protected from S1 nuclease (Fig. 5A). The major protected band is approximately 250 bp in size, which indicates that the major RNA start site is around position −140 relative to the crs coding region (Fig. 5C). The minor protected band, with a size of 195 bp, suggests that there might be a second RNA start site at position −86. To confirm these results and to map the crs RNA start site(s) more accurately, primer extension was performed with a primer that is 73 bp downstream of the minor band. One RNA-dependent band was found (Fig. 5B); its size indicated that the crs RNA initiated at position −140, consistent with the position found for the major RNA start site by S1 nuclease analysis. There was no primer extension product at the position expected for the minor band found in the S1 analysis. The minor band could be an S1 artifact or could represent the 3′ end of an upstream RNA since the HindIII end of the S1 probe was also labeled. We conclude that the crs RNA initiates at one position 140 bp before the crs coding region.

FIG. 5.

Determination of the 5′ end of the S-layer RNA. (A) S1 nuclease protection experiment using an end-labeled 543-bp PstI/HindIII fragment (probe) encompassing the promoter region of the crs gene (Fig. 1). In reaction c, the probe was hybridized to RNA from strain 314 and then subjected to S1 nuclease treatment as described in Materials and Methods. The lane marked stds contains a radiolabeled 123-bp ladder DNA (Life Technologies). (B) Primer extension reaction products obtained with the 5′-end-labeled primer CR120, which is the reverse complement of the sequence at the start of the crs coding region (C), and increasing amounts of RNA (1, 3, 15, and 45 μg) from C. rectus 314. The samples (lanes a to d) were electrophoresed on a high-resolution sequencing gel alongside the products of a dideoxy sequencing reaction (seq. rxn.) of 314 DNA with primer CR120. (C) Sequence of the region upstream of the S-layer coding sequence. The position of the codon for the first amino acid is designated +1. The dot marks the S-layer RNA start site. The putative −10 and −35 promoter elements are indicated by lines. The position of oligonucleotide CR120, which is the reverse complement of the sequence shown, is marked by the arrow.

The DNA upstream of the crs RNA initiation site was examined for potential promoter elements. A hexamer sequence that is identical to the E. coli −10 consensus promoter element was found 10 bp in front of the crs RNA start site (Fig. 5C). When the region 16 to 20 bp further upstream was examined, the best match to the E. coli −35 consensus sequence was identical only at three of six bases (Fig. 5C). Interestingly, the 20 bp immediately upstream of this putative −35 element are very A/T rich. This resembles the UP element that has been identified at the same position in the promoters for some strongly expressed genes in E. coli (32, 33) and other organisms (16). Proof that this sequence has a similar function in the C. rectus crs promoter awaits further experimentation.

As has also been found with S-layer genes in other bacteria (25), the leader sequence for the C. rectus crs RNA is rather long, 140 bp. Interestingly, even though the promoter regions of the crs gene and the sapA S-layer gene from C. fetus show no significant similarities, the first 27 bases of the C. rectus S-layer RNA are almost identical (26 of 27 bases) to the 5′ end of the sapA S-layer RNA from C. fetus (38). Contained in this conserved region is a 12-base palindrome 5′-ACTATCGATAGT-3′ whose function is unknown, although such sequences in the leader segments of RNA may play a role in RNA stability (13).

The second half of the S-layer gene from strain ATCC 33238 shows restriction fragment length polymorphisms.

Nitta et al. (30) used peptide analysis to show that the S-layer proteins from strains 314 and ATCC 33238 differed. Although the amino-terminal sequences and the sequences of one peptide were identical between the two strains, another peptide from 33238 showed no homology to any peptide from strain 314. This finding suggests that the sequences of the crs genes from 314 and ATCC 33238 should show some polymorphisms. To test this possibility, we looked for restriction site differences between the crs genes of the two strains by Southern blot analysis. The hybridization patterns of DNA from strains 314 and ATCC 33238 were compared after digestion with one of several different restriction endonucleases and subsequent hybridization with probes from three regions of the crs gene. With all three probes, there were several restriction fragment length polymorphisms (compare Fig. 3 and 6). The restriction map generated from the Southern blot data for strain 33238 is consistent with that of Miyamoto et al. (28), and comparison to the map for strain 314 suggests that the sequences between strains 314 and 33238 are homologous for the first half of the crs gene but diverge in the second half (Fig. 1). This result is also supported by the previous peptide analysis of the S-layer proteins from the two strains (30). The peptides whose sequences were identical between 314 and 33238 are identical to deduced sequences from the beginning of the gene, whereas the peptide from strain 33238 that is different has homology with the deduced sequence from the second half of the crs gene. It is yet to be determined whether the differences in the carboxy portion of the S-layer proteins from the two C. rectus strains impart a difference in function or simply reflect the antigenic diversity of proteins with the same function.

FIG. 6.

Hybridization of S-layer gene probes to genomic DNA from various C. rectus strains. DNA (8 μg) from ATCC 33238 (S⁺) and ATCC 33238 (S⁻) was digested with the indicated restriction endonucleases (C, ClaI; E, EcoRI; H, HindIII; P, PstI), electrophoresed on a 0.75% agarose gel, and analyzed by Southern blot hybridization. Each blot was hybridized with one of the ³²P-labeled probes I, II, or III, which are from different segments of the crs gene (Fig. 1). The filters were hybridized and washed under normal-stringency conditions. The positions of the molecular size standards are indicated.

A previously identified S-layer-negative isolate of C. rectus has the crs gene but not the full-length crs RNA.

ATCC 33238 (S⁻) is a strain whose S-layer was lost spontaneously during long term in vitro passage of strain 33238. The reason for the loss of the S-layer is unknown, as is the possible occurrence of other mutations in this strain. To begin to determine the molecular basis for the loss of the S-layer in this strain, Southern and Northern blot analyses were performed. On Southern blots, the hybridization patterns were identical for DNA from the CRS⁺ and CRS⁻ strains of ATCC 33238 (Fig. 6). This finding indicates that the crs gene is present in the spontaneous CRS⁻ isolate and that there are no insertions or deletions larger than 100 bp in the crs gene from the spontaneous CRS⁻ isolate. Miyamoto et al. (28) reached the same conclusion in their analysis of strain 33384 but did not delve further into the reason for the lack of an S-layer in the spontaneous mutant.

Although the crs gene is present in the spontaneous CRS⁻ isolate, the intact 4.4-kb crs RNA was not found by Northern blot analysis (Fig. 4A). Instead, a broad hybridization signal ranging from 1 to 2 kb in size was found. This signal was seen only when probe I, which recognizes the first half of the crs RNA, was used in the hybridization (data for probes II and III are not shown). All three probes hybridized to a 4.3-kb RNA in the parental strain, CRS⁺ 33238, indicating that the hybridization to a smaller RNA in the CRS⁻ 33238 isolate is not due to strain differences with strain 314. Finally, the smaller crs RNA in the CRS⁻ isolate is not due to general RNA degradation in that sample since another hybridization probe, the glyA gene from A. actinomycetemcomitans, gave a similar signal on Northern blots with RNA from all three strains (Fig. 4B). These results suggest that the initiation of crs RNA synthesis may be normal in the spontaneous CRS⁻ isolate of 33238 but that the RNA either terminates early in the gene or is rapidly degraded from the 3′ end.

The availability of the cloned S-layer gene from C. rectus will allow molecular genetic approaches to be used in the analysis of the function of the S-layer protein in pathogenesis. In particular, the tools are being developed to create isogenic crs and crs⁺ strains of C. rectus in order to elucidate directly the importance of the S-layer as a virulence factor in in vivo and in vitro studies.