INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

By currently accepted criteria, the gram-positive anaerobic bacterium Peptostreptococcus micros is one of the causative agents of periodontal disease (15, 33). Several potential virulence factors of P. micros have been reported, such as adherence to gingival epithelial cells (7), expression of immunoglobulin G Fc-binding proteins (14), production of hydrogen sulfide from glutathione (2), and production of hyaluronidase (34).

Two types of P. micros have been described, i.e., the smooth (Sm) type and the rough (Rg) type (37). Based on 16S RNA analysis and pyrolysis mass spectrometry, these two types are now considered distinct genotypes (20). Both types can be isolated from subgingival plaque samples from subjects with periodontal disease (38). One of the prominent characteristics of the Rg type is the expression of large fibril-like surface appendages. These auto-aggregating structures often exceed 4 µm in length. Thus far, P. micros is the only gram-positive anaerobic coccus on which such structures have been observed. We can only speculate on their biological function, but they do not seem to be required for adherence to epithelial cells (19) or to other bacteria (18). Remarkably, when grown in broth culture, the Rg type readily converts to a smooth variant (Rg^Sm variant), which lacks the fibril-like structures (20). Since such variants have never been isolated from periodontitis patients (18), the correct assembly of the fibril-like structures probably serves some essential function in vivo.

Characterization of the constituents of these structures may help elucidate their biological function. However, at present their composition is unknown. All our attempts to purify the structures were unsuccessful. In the present study we raised antibodies against the Rg type and absorbed the obtained serum with the Rg^Sm variants to generate a fibril-specific serum. This serum was then used to identify an antigenic constituent of the fibril-like structure of P. micros.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Microorganisms and culture methods. The origin and main characteristics of all strains and plasmids used in this study are listed in Table 1. P. micros strains were routinely cultured on blood agar plates (Oxoid no. 2 agar, supplemented with 5% defibrinated horse blood, hemin [5 mg/ml], and menadione [1 mg/ml]) for 4 days in 80% N₂-10% CO₂-10% H₂ at 37°C. These strains were identified by anaerobic growth, Gram staining, and ATB-32A kits (Analytab Products, Montalieu-Vercieu, France). Rg^Sm variants of all of the Rg strains were obtained after four passages in Schaedler broth (BBL Microbiology Systems, Cockeysville, Md.) and subsequent subculturing on blood agar plates (20). Escherichia coli strains were grown under aerobic conditions at 37°C, except for E. coli BM25.8, which was grown at 31°C on Luria-Bertani (LB) agar containing kanamycin (50 µg/ml) and chloramphenicol (34 µg/ml). E. coli DH5 was routinely cultured on LB agar, E. coli XL-1-Blue was cultured on LB agar with addition of tetracycline (15 µg/ml). When appropriate, ampicillin (50 µg/ml) was added to these media.

Rabbit antisera. Male chinchilla rabbits (2 to 3 kg) were immunized with P. micros HG 1259 (Rg morphotype). A suspension of bacteria in phosphate-buffered saline (PBS) (ca. 10¹⁰ bacteria/ml) was administered intravenously eight times, every other day in increasing amounts (0.25-ml steps). One week after the last injection, a 2.0-ml booster was administered intravenously. The anti-Rg type serum, prepared from blood obtained by cardiac puncture, was inactivated by incubation at 56°C for 30 min and stored at 80°C until use. Aliquots (1 ml) were adsorbed with 200 mg (wet weight) of bacteria of an Sm type (ATCC 33270^T) or an Rg^Sm variant (HG 1259^Sm). For these adsorptions the sera were subjected to four cycles of incubation, each consisting of 1 h at room temperature and 12 h at 4°C, followed by centrifugation at 14,000 × g for 5 min to remove the bacteria.

Electron microscopy (EM). P. micros strains were harvested from blood agar and washed once in distilled water. Grids coated with bacteria were washed once in 0.1 M PBS-0.15 M glycine-1% bovine serum albumin and subsequently incubated for 30 min with the appropriate sera diluted in PBS-glycine-bovine serum albumin. The samples were washed twice in PBS-glycine and incubated with a 1:20 dilution in PBS-glycine of goat anti-rabbit immunoglobulin G-colloidal gold (particle size, 10 nm; Aurodye, Hertogenbosch, The Netherlands) for 30 min at room temperature. After two more washes, the samples were examined with a model EM301 electron microscope (Philips, Eindhoven, The Netherlands).

SDS-PAGE. Whole-cell protein patterns were determined by standard protein polyacrylamide gel electrophoresis (PAGE) (22). Cells harvested from blood agar were washed with PBS, resuspended in 0.5 M Tris-HCl (pH 6.8), and diluted 1:1 in sample buffer (4% sodium dodecyl sulfate [SDS], 2% 2-mercaptoethanol, 20% glycerol, 125 mM Tris-HCl [pH 6.8], bromphenol blue [0.1 mg/ml]). The samples were heated for 10 min at 100°C, and the insoluble debris was removed by centrifugation at 14,000 × g for 10 min. Electrophoresis was performed on a 1.5-mm-thick, 10% homogeneous polyacrylamide gel at 100 V for 2 h, and the proteins were stained with Coomassie brilliant blue R-250 (Bio-Rad Laboratories, Hercules, Calif.).

Immunoblotting. Whole-cell proteins were separated by SDS-PAGE (10% polyacrylamide) and transferred to a nitrocellulose membrane (pore size, 0.45 mm; Schleicher & Schuell, Dassel, Germany) by Western blotting (35). After blotting at 100 V for 1 h, the nitrocellulose sheets were incubated for 1 h at room temperature with blocking buffer TTBS (100 mM Tris-HCl [pH 7.5]-0.9% NaCl supplemented with 0.1% [vol/vol] Tween 20). The blots were incubated overnight with sera (1:100 to 1:500 in TTBS). Subsequently, they were washed four times with TTBS, incubated for 1 h with goat anti-rabbit immunoglobulin G-horseradish peroxidase (1:1,500 in TTBS; Nordic, Tilburg, The Netherlands), and washed again four times in TTBS. Antibody-binding antigens were visualized with 4-chloro-1-naphtol (Bio-Rad) and H₂O₂.

N-terminal amino acid sequencing. SDS-soluble proteins of P. micros HG 1259 were separated on an SDS-10% polyacrylamide gel. The proteins were subsequently transferred to methanol-prewetted polyvinylidene difluoride membrane (Immobilon-P; Millipore Corp., Bedford, United Kingdom), by semidry blotting with 10 mM CAPS (3-cyclo-hexyl-amino-1-propanesulfonic acid) blotting buffer containing methanol (10% [vol/vol]) at pH 11. Afterwards, proteins on the membrane were visualized by Coomassie brilliant blue R-250 (Bio-Rad) staining, and the protein band of interest was excised and subjected to N-terminal sequence analysis by Edman degradation.

DNA isolation. All P. micros strains indicated in Table 1 were harvested from blood agar plates, washed twice in PBS, and resuspended in 25 mM Tris-HCl-10 mM EDTA-50 mM glucose (pH 8.0). The bacteria were disrupted by four cycles of freeze-thawing, followed by incubation for 1 h at 37°C with lysozyme (10 mg/ml) and proteinase K (1 mg/ml) and an incubation with SDS (4%) for 1 h at 56°C. After two phenol-chloroform-isoamyl alcohol extractions and two chloroform-isoamyl alcohol extractions, sodium acetate (0.3 M, pH 7.6) was added before ethanol precipitation. The DNA was air dried and dissolved in 10 mM Tris-1 mM EDTA (pH 8.0) containing RNase (20 mg/ml).

Construction of an expression library. Chromosomal DNA of P. micros HG 1259 was partially digested with Tsp509I (New England Biolabs, Inc., Beverly, Mass.) at 65°C. This partial digest was size fractionated on a sucrose gradient (28). DNA fragments of 1,000 to 3,000 nucleotides (nt) were used for ligation into EcoRI-predigested -TriplEx arms (CLONTECH Laboratories, Inc., Palo Alto, Calif.), followed by  phage packaging reactions using the Gigapack II Plus Packaging Extract (Stratagene Cloning Systems, La Jolla, Calif.), as recommended by the manufacturer. After amplification of the primary library in XL-1-Blue, 10% chloroform was added and the library was stored at 4°C.

Isolation of fib clones. The -TriplEx phagemids containing DNA fragments encoding the FibA protein were selected by immunoscreening, essentially as described by the manufacturer. The immunoscreening of the resulting filters with Rg^Sm-adsorbed anti-Rg serum was performed as described in the immunoblotting section. Phages of positive plaques were eluted from isolated agar plugs, and this phage eluate was replated and rescreened to obtain single clones, which were subsequently converted to plasmids in E. coli BM25.8, according to the manufacturer's recommendations. The inserts of the converted plasmids were sequenced by using the 5'-Texas Red-labeled T7 (TAATACGACTCACTATAGGG) and 5'TriplEx (TTTTCTCGGGAAGCGCGCCAT) primers (Isogen Bioscience BV, Maarssen, The Netherlands). Subclones were obtained by cloning PCR fragments by using pGEM-T Vector System I (Promega Corp., Madison, Wis.) in E. coli DH5F'; these subclones were sequenced with T7 and Sp6 (CGATTTAGGTGACACTATAG) primers. Sequence reactions were performed with a Thermo Sequenase premixed cycle sequencing kit (catalog no. RPN2444; Vistra Systems, Amersham Pharmacia Biotech, Roosendaal, The Netherlands) on a Vistra DNA Sequencer 725 (Vistra Systems, Amersham). The sequencing data were analyzed with Lasergene software (DNAstar Inc., Madison, Wis.). The obtained DNA sequences were subsequently compared with sequences in the NCBI database (National Center for Biotechnology Information, Los Alamos, N.Mex.) by using the BLAST software.

fibA analysis in P. micros types. Genomic DNA that was isolated from P. micros Sm and Rg strains (Table 1) was used for PCR amplification with fibA-specific primers. Primers used for this analysis, i.e., Fib-FN (TAATCGTTGGAGAGGCTAAGG) and Fib-RN (TGCCTATCTTTTTCGAATTC), were designed to amplify the N-terminal part of the fibA gene since no sequences homologous to this part were found in other bacteria by a BLAST homology search. The fibA PCR product of HG 1259 was subsequently labeled with DIG-HighPrime (Boehringer Mannheim). This probe was used for dot blotting on genomic DNA of Sm and Rg genotypes of P. micros, according to the manufacturer's instructions.

Nucleotide sequence accession number. The DNA sequence of fibA, orf2, and orf3 of P. micros HG 1259 (Rg type) has been deposited in the GenBank database under accession no. AF097909.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of an antigen of the fibril-like structures. A previous study provided strong indications that the fibril-like appendages of the Rg type of P. micros are proteinaceous structures (19). To isolate proteins that constitute these structures an anti-Rg type serum was adsorbed with Sm-type and Rg^Sm-type bacteria. Serum adsorbed with the Sm type (ATCC 33270) strongly recognized the fibril-like structures as well as some other surface structures on the Rg-type cells as determined by immunogold EM (Fig. 1A). Adsorption with the Rg^Sm variant of HG 1259 increased the specificity for the fibril-like structures, although there still was substantial labeling at the cell-to-cell junctions (Fig. 1B). On Rg^Sm-type bacteria, gold particles were observed at the intersections of adjacent bacterial cells and on structures resembling fibril-like structures that were not attached to cells (Fig. 1C). No gold particles bound to Sm-type bacteria when the Sm type-adsorbed or Rg^Sm type-adsorbed anti-Rg type serum was used (data not shown). These findings indicate that the adsorption of the anti-Rg type serum with the Rg^Sm type had effectively removed all common antibodies, leaving only antibodies recognizing Rg type-specific structures.

View larger version (172K): [in this window] [in a new window]   FIG. 1.   Micrographs of immunogold-labeled P. micros HG 1259 (Rg type). Bacterial cells were incubated with anti-Rg serum adsorbed with ATCC 33270T (Sm type) (A) or adsorbed with HG 1259Sm (RgSm type) (B). The latter serum was also incubated with HG 1259Sm (RgSm variant) (C). Bars, 1 µm.

View larger version (37K): [in this window] [in a new window]   FIG. 2.   (A) Immunoblots incubated with anti-Rg rabbit serum adsorbed with Sm type cells (lanes 1 to 3) or adsorbed with RgSm type cells (lanes 4 to 6). The latter serum was adsorbed with a sonicate of E. coli BM25.8 containing pTX-Fib2 (lanes 7 to 9). SDS-soluble whole-cell components of HG 1259 (Rg type) (lanes 1, 4, and 7), HG 1259Sm (RgSm type) (lanes 2, 5, and 8) and HG 1256 (Sm type) (lanes 3, 6, and 9) were separated on an SDS-10% polyacrylamide gel and transferred to nitrocellulose. (B) Immunoblot with anti-Rg serum adsorbed with RgSm-type cells. Shown are SDS-soluble whole-cell components of P. micros HG 1259 (lane 1), of E. coli BM25.8 containing pTX-Fib1 (lane 2) or pTX-Fib2 (lane 3), and of E. coli BM25.8 (lane 4). A 45-kDa protein encoded on pTX-Fib2 (indicated by a star) was recognized by the fibril-specific serum.

Immunoscreening of a -TriplEx expression library and isolation of fib clones. To isolate the gene encoding the 42-kDa constituent of the fibril-like structures, a -TriplEx expression library was constructed from partially Tsp509I-digested DNA of strain HG 1259. The phagemids of the two seropositive plaques that resulted from immunoscreening with Rg^Sm type-adsorbed anti-Rg type serum were subsequently converted to plasmids by transduction into E. coli BM25.8. The two resulting plasmids were designated pTX-Fib1 and pTX-Fib2. Restriction enzyme analysis revealed that pTX-Fib1 contained an insert of approximately 1,400 nt, whereas pTX-Fib2 contained an insert of approximately 2,600 nt.

Sequence of pTX-Fib1 and pTX-Fib2. The combined sequence of pTX-Fib1 and pTX-Fib2 was 3,248 nt long; the 3' end of the insert of pTX-Fib1 overlapped the 5' end of clone pTX-Fib2 for a length of 738 nt (Fig. 3). A sequence analysis of this combined DNA fragment revealed the presence of one complete open reading frame (ORF), designated fibA, spanning a total of 1,182 nt, encoding 393 amino acids (aa). The initiation codon of this ORF at nt 662 was preceded by a putative Shine-Dalgarno sequence, AGGA (25), at 6 nt from the start codon, and three putative 35 and 10 transcription initiation sequences homologous to the consensus promoter sequences of gram-positive bacteria. A hairpin-loop sequence that resembled a putative rho-independent transcription terminator was present directly downstream of fibA (nt 1,861 to 1,886).

View larger version (19K): [in this window] [in a new window]   FIG. 3.   Schematic representation of the sequence of the inserts of clones pTX-Fib1 and pTX-Fib2. The ORFs fibA, orf2, and orf3 deduced from the DNA sequence are indicated; the arrows indicate the orientations of these ORFs. These clones do not contain the complete sequences of orf2 and orf3.

FibA, the fibril-like subunit protein. To confirm that fibA encoded the antigen that was recognized by the fibril-specific serum, the 42-kDa protein was isolated from a nitrocellulose blot and subjected to N-terminal amino acid sequencing. This analysis revealed a 15-residue sequence: Ser-Ile-Asn-Arg-Gly-Glu-Ala-Lys-Glu-Lys-Tyr-Asp-Val-Ile-Pro. This amino acid sequence was completely present in the deduced peptide sequence of fibA, starting at aa 39. The predicted size of the 355-aa mature FibA was 41,368 Da, which agreed with the size estimated by SDS-PAGE. Analysis of the deduced FibA amino acid sequence revealed that the most abundant amino acids were lysine (11.4%), asparagine (8.6%), and valine (8.4%). The hydrophobicity profile of the FibA peptide sequence (21) indicated that except for the first 30 residues, which comprised a highly hydrophobic region, the protein was mainly hydrophilic. Conformational analysis using the algorithms of Garnier et al. (12) and Chou and Fasman (3) indicated no large -helical domains in any part of the FibA protein. The Chou-Fasman method predicted primarily turns.

Deduced amino acid sequence of orf2 and orf3. The second ORF on pTX-Fib2 exhibited high homology with a large part of fibA both in nucleotide sequence and in deduced amino acid sequence. Kyte and Doolittle (21) analysis revealed that this peptide is also primarily hydrophilic except for a small N-terminal domain. The Orf2 protein had a similar C-terminal six-repeat domain (aa 243 to 361), which was 87.5% identical to the FibA repeat domain. Furthermore, the domain preceding the repeat domain, i.e., aa 111 to 242, also showed high similarity to the FibA domain preceding the repeat domain. The N-terminal domain of Orf2 (aa 31 to 173) exhibited homology with myosin proteins of various eukaryotic organisms (5). Similarities of 45 to 52% with the myosin proteins, as indicated by a BLAST homology search, were based on a putative -helix domain, which was indicated by analyses according to the methods of both Garnier et al. (12) and Chou and Fasman (3). The signal sequence of this protein contained features similar to those in the FibA leader peptide; the methionine precedes two charged lysines and a hydrophobic core. Putative protease cleavage sites were identified after alanine residues at positions 23 and 29. The deduced amino acid sequence of Orf3 was also highly homologous to the C termini of FibA and Orf2 (Table 2).

fibA in Rg and Sm genotypes. To survey the presence of fibA in other Rg strains, in Sm strains, and in an Rg^Sm variant, PCR analysis was performed with primers Fib-FN and Fib-RN. These primers were designed to specifically amplify the N-terminal region of the fibA gene, since the DNA encoding for the C-terminal region of fibA is apparently more generally used in genes encoding surface proteins in P. micros. This PCR analysis revealed the presence of this region of the fibA gene in all five Rg isolates and one Rg^Sm strain but also in all six Sm-type strains (data not shown). These data were confirmed by dot blot analysis on chromosomal DNA of the indicated Rg strains, Rg^Sm variant, and Sm strains of P. micros (Table 1), with the digoxigenin-labeled PCR product of HG 1259 (Rg type) as probe (data not shown).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Long protruding fibril-like structures are a prominent morphological characteristic of the Rg type of P. micros. Morphologically similar structures are present on other oral bacteria, such as Porphyromonas gingivalis, Actinobacillus actinomycetemcomitans, Prevotella spp., Actinomyces spp., and streptococci. The fimbrial structures on these bacteria were shown to be key components in cell-to-surface and cell-to-cell adherence and therefore are assumed to be important in the pathogenesis of some oral and nonoral diseases (16). The fibril-like structures of P. micros seem to have a different function: compared to the Rg^Sm variant and the Sm type, which both lack these structures, the Rg type adhered slightly less to epithelial cells (18). Additionally, no differences in coaggregation were observed among these three morphotypes of P. micros; they all displayed similar levels of aggregation with Fusobacterium nucleatum strains and nonencapsulated P. gingivalis strains (19).

In the present study a component of the fibril-like structures, designated FibA, was identified. Adsorption of an anti-Rg type serum with the Rg^Sm variant, which lacks the fibril-like surface appendages (20), resulted in a serum that had an enhanced specificity for the fibril-like structures. This antiserum specifically recognized a 42-kDa protein of the Rg type. Two clones of an expression library of the Rg type, i.e., pTX-Fib1 and pTX-Fib2, were selected by immunoscreening with this serum. The gene, designated fibA, was present on both clones and encoded a putative 45,564-Da protein. IPTG-induced expression of fibA in E. coli resulted in the translation of a protein of approximately 45-kDa that was recognized by the fibril-specific antibodies. Furthermore, the N-terminal amino acid sequence of the 42-kDa antigen was present at aa 39 to 53 of the deduced FibA protein. Adsorption of the fibril-specific antiserum with a sonicate of the E. coli clone expressing FibA precursor protein removed the antibodies recognizing the 42-kDa FibA on a Western blot, and in immunogold EM no gold particles attached to the fibril-like structures. These results indicated that FibA is indeed an antigenic component of the fibril-like structures. However, it does not exclude the presence of other constituents in the fibril-like structures.

Remarkably, the Rg type-specific serum also recognized structures at cell-to-cell interfaces of the Rg bacteria. The most likely explanation for this is that the fibril-like structures originate from every single cell in a bacterial cluster and there aggregate to form the fibril-like structures. Hence the individual components of these larger structures like the 42-kDa FibA protein will also appear at cell-to-cell junctions. Although adsorption to generate an Rg type-specific serum was done with whole cells of the Rg^Sm variant, antibodies from this serum still recognized a 42-kDa antigen in whole-cell preparations of the Rg^Sm variant. Furthermore, in immunogold EM samples of the Rg^Sm variant, substantial immunogold labeling was observed at cell-to-cell junctions and on sporadically observed structures that were not attached to bacterial cells, resembling the fibril-like structures (Fig. 1B). These observations indicate that the lack of fibril-like structures on the Rg^Sm strain is not the result of a difference in the level of expression of FibA per se. Conformational changes in FibA, but more likely changes in the expression of other components of the fibril-like structure, may impair the attachment of these structures to the cell wall. Further research should focus on revealing the composition of the fibril-like structures and identification of the differences between the Rg and Rg^Sm type, that are responsible for the loss of the fibril-like structures.

PCR analysis and dot blot analysis of the presence of the fibA gene revealed that the N-terminal region of fibA is present in Rg- and Sm-type P. micros strains. Although sequences similar to those of fibA are present in the Sm type, FibA-specific antibodies recognized no antigen in whole-cell preparations of this type. This might be caused by changes in the epitope of the translated protein or by an obstruction of transcription or translation of the Sm-type fibA gene. Similar phenomena have been described for other human pathogens; for example, H. pylori isolates that contain the pathogenicity island-associated cagA gene but lack CagA expression as a result of insertion upstream of the gene have been previously described (43).

The precursor of the FibA protein contained a putative 38-residue leader peptide. This leader peptide exhibited structural similarities to signal peptides of secreted proteins of other organisms, such as Bacillus spp. (31) and Mycoplasma spp. (4). The potential peptidase cleavage site at position 24 does not directly precede the N terminus of the mature FibA. This implicates the presence of a short propeptide, like the ones described for several secreted enzymes of Bacillus spp. (31). The presence of a secretion signal within the FibA precursor protein confirms that this protein is indeed an exported protein, which is in line with its presence in the fibril-like structure of P. micros. A structurally similar signal peptide was present in the deduced protein sequence of orf2, indicating that the product of this gene might also be an exported protein.

In gram-positive microorganisms, a hexameric LPXTGX motif followed by a highly hydrophobic membrane-spanning domain in the C-terminal region of proteins is probably important for anchoring of surface proteins in the cell membrane (10, 24, 29). These membrane-anchoring regions could not be identified in FibA or the deduced peptides of orf2. Both proteins contained a C-terminal repeat domain comprising a relatively large number of aromatic amino acids, which was homologous to the C termini of a family of surface proteins of streptococci and surface proteins and toxins of Clostridium spp. (39, 42, 44). This family of proteins includes surface-associated proteins PspA (44), SpsA (17), CbpA (27), and autolysin (11) from Streptococcus pneumoniae; glucosyltransferases from Streptococcus mutans (30, 36) and from Streptococcus downei (9, 13); and toxins A and B from Clostridium difficile (1, 6). These C-terminal repeats were shown to be involved in anchoring surface proteins to carbohydrate structures (39); more specifically, in S. pneumoniae PspA these repeats represent an anchoring mechanism via choline-mediated interaction with lipoteichoic acid (46). At present no information is available on the presence of choline in the cell wall of P. micros. Molecular characterization of PspA of S. pneumoniae revealed that a minimum of five 20-residue repeats is required for anchoring (45). In FibA and the deduced peptide of orf2, six repeats were present. The proximal two of these repeats exhibited only minor similarity to the four C-terminal repeats. These differences may result in a less-rigid anchoring to the cell surface carbohydrates, thus explaining why FibA is not only directly attached to the cell surface but is also part of the protruding fibril-like structures. From the fact that these C-terminal repeats were present in FibA and in the deduced peptides of orf2 and orf3, we hypothesize that this surface-anchoring mechanism is generally used in P. micros surface proteins.

Since the proteins encoded by fibA and orf2 are highly homologous, they both might be involved in the assembly of the fibril-like structures. Like the surface protein PspA from S. pneumoniae (44), the N-terminal region of the deduced protein of orf2 contained heptad repeats, which are implicative of an -helical coiled-coil region. This region is homologous to fibrous proteins like myosin proteins of eukaryotic organisms (5). Therefore, this protein may account for the fibril-like appearance of the surface structures.

The rapid loss of attached fibrils in vitro, resulting in the Rg^Sm variants, and the absence of these variants from subgingival plaque samples (18) suggest that the fibril-like structures are essential features for the in vivo survival of the Rg type of the periodontal pathogen P. micros. Although the present study did not elucidate the biological function of the fibril-like structures of the Rg type of P. micros, it has identified an antigenic proteinaceus component (FibA) of the fibrillar structure. Sequence analysis of the gene encoding FibA did not reveal the biological function or its potential role in the assembly of the fibril-like structures. To determine the biological function of FibA, deletion mutants are required; however, at present no molecular tools for producing these mutants are available. At present experiments are focused on the development of molecular tools for P. micros.