Enhanced Biofilm Formation and Loss of Capsule Synthesis: Deletion of a Putative Glycosyltransferase in Porphyromonas gingivalis

Abstract

Periodontitis is a biofilm-mediated disease. Porphyromonas gingivalis is an obligate anaerobe consistently associated with severe manifestations of this disease. As an opportunistic pathogen, the ability to proliferate within and disseminate from subgingival biofilm (plaque) is central to its virulence. Here, we report the isolation of a P. gingivalis transposon insertion mutant altered in biofilm development and the reconstruction and characterization of this mutation in three different wild-type strains. The mutation responsible for the altered biofilm phenotype was in a gene with high sequence similarity (∼61%) to a glycosyltransferase gene. The gene is located in a region of the chromosome that includes up to 16 genes predicted to be involved in the synthesis and transport of capsular polysaccharide. The phenotype of the reconstructed mutation in all three wild-type backgrounds is that of enhanced biofilm formation. In addition, in strain W83, a strain that is encapsulated, the glycosyltransferase mutation resulted in a loss of capsule. Further experiments showed that the W83 mutant strain was more hydrophobic and exhibited increased autoaggregation. Our results indicate that we have identified a gene involved in capsular-polysaccharide synthesis in P. gingivalis and that the production of capsule prevented attachment and the initiation of in vitro biofilm formation on polystyrene microtiter plates.

The clinical importance of dental plaque and its accessibility for in vivo research makes it one of most studied and best-understood biofilm communities. It is now well documented that gram-negative anaerobic bacteria play a significant role in the development of periodontitis, with Porphyromonas gingivalis being implicated as one of the major players in the progression of this chronic disease (10, 19, 30, 61). A number of factors are associated with the virulence of the organism, including a variety of proteases, endotoxins, and collegenases and the production of surface structures, such as fimbriae and capsular polysaccharide (38). This opportunistic pathogen also attaches to and invades human epithelial cells (18, 53, 56, 57), connective tissue, and endothelial cells (15, 17, 55), and invasion has been shown to be mediated by the expression of fimbriae and a variety of surface adhesins. It is evident that the growth of P. gingivalis within the subgingival plaque is central to the disease process; however, although there have been numerous studies of the pathogenicity of the organism and its interactions with other organisms within the biofilm community, little is known about the molecular-genetic basis of biofilm formation in P. gingivalis.

Research on oral biofilms has focused on determining the spatial organization and complex development of plaque by examining the succession of organisms in the growing biofilm. By monitoring the development of the microbial community on the tooth surface after professional cleaning, studies have demonstrated that the early colonizers are primarily gram-positive organisms and the late, or secondary, colonizers are the more fastidious gram-negative anaerobes (47, 48). It is likely that late colonizers take advantage of an altered environment created by their predecessors, e.g., a decrease in redox potential or the availability of substrates from metabolic activity. Early colonizers can also provide new cell-cell attachment sites, and recent in vitro studies using saliva-conditioned flow cells have clearly shown the importance of coaggregation between different genera in the development of this complex community (19, 21). P. gingivalis is recognized as one of the late, or secondary, colonizers during subgingival plaque development. P. gingivalis is a fastidious anaerobe that requires nutritional supplements for growth; therefore, the bacterium likely requires the colonization and metabolic activity of other organisms to become established within the biofilm (78). Coaggregation studies have shown specific interactions between P. gingivalis and a variety of other microbial species, such as Streptococcus species and Treponema denticola (31), thus supporting the model that early colonizers provide attachment sites and possibly a niche for secondary colonizers to become established.

Whether initiated by a single species or a multispecies community, biofilm formation is a complex process (67). Biofilm formation can be viewed as one part of a developmental cycle involving a series of distinct steps. To form a biofilm, the bacteria first attach to a surface and then alter their gene expression in order to adjust to this surface-attached state. If the environment is conducive to growth, the bacteria will then grow on the surface and further develop into a mature matrix-encased biofilm. To complete the cycle the bacteria detach, disseminate, and colonize a new surface. The complexity of this process indicates that many genes are likely required. During the past decade, a series of studies have addressed the regulation and molecular-genetic basis of biofilm development in a variety of disease-causing model organisms, including Pseudomonas sp. (68, 69), Escherichia coli (24, 25, 71, 76), Staphylococcus aureus (34, 35), and Vibrio cholerae (5, 44, 61, 84). These studies have clearly shown that surface appendages, such as pili and flagella, are significant factors in biofilm development and that the availability of nutrients and the regulation of central metabolism can play pivotal roles in biofilm formation (32, 66, 77). In addition, although the exact role of rpoS is still under investigation, this stationary-phase sigma factor has been shown to affect biofilm development in E. coli (1, 11) and in V. cholerae (90). Just recently, the initial attachment phase was found to be a complex process in itself, in which steps of reversible and irreversible attachment have been discovered in two different Pseudomonas species (7, 37). Also, signaling networks (two-component and three-component systems) that simultaneously coordinate the activation of genes required for planktonic growth and the repression of genes involved in biofilm formation were discovered (27, 41, 50).

With the goal of identifying genes involved in biofilm development in P. gingivalis, we screened a transposon insertion library for mutants with altered biofilm phenotypes. In the present report, we focus on a mutation that results in an unusual phenotype, that of enhanced biofilm formation. The disrupted gene was identified, cloned, and sequenced, and the mutation was reconstructed in three different wild-type strains. The gene, encoding a putative glycosyltransferase, is located in a region of the chromosome that was recently identified as the capsular-polysaccharide (K-antigen) biosynthesis locus of P. gingivalis (2, 70). Our results show that a mutation in this gene in a strain that is typically encapsulated (strain W83) results in loss of capsular polysaccharide. The data indicate that lack of capsule production enhances biofilm formation in this organism.

MATERIALS AND METHODS

Strains, media, and chemicals.

Bacterial strains and plasmids used in this study are shown in Table 1. P. gingivalis wild type and the DEL0106::Erm mutant strain were maintained in a COY anaerobic chamber on Trypticase soy agar plates supplemented with 5% defibrinated sheep blood (North-East Laboratory, Waterville, ME), 1 μg/ml hemin, and 1 μg/ml menadione (BAPHK) or grown in Trypticase soy broth containing hemin and menadione (TSBHK) or Todd-Hewitt broth containing hemin and menadione (THBHK). Escherichia coli strains were grown aerobically on Luria-Bertani plates. Antibiotics were added at the following concentrations: for E. coli, carbenicillin (Cb), 100 μg/ml, and erythromycin (Erm), 200 μg/ml; for P. gingivalis, Erm, 5 μg/ml, and tetracycline (Tc), 1 μg/ml. Unless otherwise stated, all chemicals were obtained from Sigma. Plasmids were constructed in E. coli DH5α and then transformed into P. gingivalis by electroporation (see the protocol below).

TABLE 1.

Strains and plasmids used in this study

Strain (relevant genotype)	Source or reference
P. gingivalis strains
33277 (wild type)	American Type Culture Collection
W83 (wild type)	Christian Mouton, Laval University, Quebec City, Quebec, Canada
381 (wild-type)	Howard Kuramitsu, State University of Buffalo, Buffalo, NY
PG0106::Tn4400 (Tcr)	This study
DEL0106::Erm (Emr) in strains
W83, 33277, and 381	This study
Plasmids
p0106-Tn4400 (Tcr)	This study
p0106TOPO (Emr Cbr)	This study
pHS17	20
pCR 4-TOPO (Kmr Cbr)	Invitrogen, Carlsbad, CA
pT-COW (Cbr Tcr)	22
pT-C106 (Cbr Tcr)	This study

Transposon insertion library.

The P. gingivalis strain 33277 transposon insertion library used in this study was described previously (8). Cloning of the DNA flanking the transposon insertions and determination of the DNA sequence with the L78 primer that reads out from IS4400L of Tn4400 were carried out as described previously (8).

Generation of cassettes for allelic exchange and complementation plasmid.

To confirm that the mutation was linked to the altered biofilm phenotype, the mutation was reconstructed in the parent strain, 33277. In addition, for comparison, the deletion was also constructed in two other wild-type strains (W83 and 381). The cassette for allelic exchange to create these knockout strains was generated by PCR using splicing by overlap extension (36, 39). Three individual fragments were amplified separately with primers designed to overlap one another. All primers used in this study are presented in Table 2. The PCR was carried out in 100-μl volumes using high-fidelity SuperMix reaction mixture (Invitrogen, Carlsbad, CA). The PCR products were purified by the QIAquick PCR purification protocol (QIAGEN). Chromosomal DNA was isolated from P. gingivalis 33277 with the MasterPure DNA purification kit (Epicenter technologies, Madison, WI). The coding region for erythromycin resistance (ErmF-ErmAM) was amplified using plasmid pHS17 as a template and primers ErmF and ErmR (20). The PCR consisted of 30 cycles with a temperature profile of 1 min at 94°C, 30 seconds at 50°C, and 3 min at 72°C for extension. For amplification of the F1-R2 (∼1 kb upstream of PG0106) or the R1-F2 (∼1 kb downstream of PG0106) amplicon, the annealing temperature was dropped to 40°C instead of 50°C, since the R2 and F2 primers have nonhybridizing tag sequences at their 5′ ends. To generate the final three-way product, the PCR consisted of 35 cycles with a temperature profile of 1 min at 94°C, 30 s at 40°C, and 6 min at 72°C for extension. The final PCR product was cloned into the TOPO plasmid by blunt-end ligation, and the sequence of the cassette was determined using the M13 forward and reverse primers.

TABLE 2.

Primers used in this study

Primer name	Sequence (5′-3′)
Mutant construction and verification
ErmF	CCGATAGCTTCCGCTATTGC
ErmR	GAAGCTGTCAGTAGTATACC
Ermchk1	CGTAAATGTTCAACCAAAGCTGTG
Ermchk2	CTCAAGTCTCGATTCAGCAATTGC
F1_106	ATTGCTACGCTCGGTTATCAAACA
R1_106	ATTAGTTCCATGCAAGGTTCCTTT
R2_106	TGTAGATAAATTATTAGGTATACTACTGACAGCTTCATCCGCCCGTCTTCACTACGATCT
F2_106	ACCGATGAGCAAAAAAGCAATAGCGGAAGCGATCGGTTTATATTCGAGTGGTGTTTTATA
F3_106	TTAGAGAAGTCTGCCGATATTGTT
R3_106	TACTCCACTACCGGGCTGAAGAAT
Probes for Northern blotting
F0106	AGAACGGTACATAAAGGGGCTATA
R0106	GCTTTGAACTTTTGCCGATAAGAT
F0108	GCAACAAGATCAATACTTCCCTTG
F0120	CAGGGCTTCGCACGCACAA
R0120	CGCCCAGCCTCTAAACGATTCA
5′ RLM-RACE
5′ RACE adapter	GCUGAUGGCGAUGAAUGAACACUGCGUUUGCUGGCUUUGAUGAAA
Outer adapter primer	GCTGATGGCGATGAATGAACACTG
Inner adapter primer	CGCGGATCCGAACACTGCGTTTGCTGGCTTTGATG
106_5′_outer	GCCCCTTTATGTACCGTTCTTT
106_5′_inner	AATCATCTCTTCTGCCCATCC

Isolation of total RNA.

Total RNA was isolated from cells in early exponential growth (optical density at 550 nm [OD₅₅₀] of 0.2) and late exponential growth (OD₅₅₀ of 0.8) in TSB and from colonies grown on BAPHK agar plates for 24 h. RNA was isolated using the MasterPure RNA purification kit (Epicenter) according to the manufacturer's instructions. Contaminating DNA was removed by digestion with DNase I (Ambion), followed by acid-phenol chloroform extraction. The removal of DNA was confirmed by PCR using a primer set specific for the groEL gene, as described previously (40). The absence of a PCR product was used as an indicator that contaminating DNA was removed. The positive control contained 10 ng of P. gingivalis DNA. The quality of the RNA was determined by agarose gel electrophoresis, and the concentration was determined spectrophotometrically. The RNA was stored at −80°C.

5′ RLM-RACE and Northern blot analysis.

RNA ligase-mediated rapid amplification of cDNA ends (RLM-RACE) was used to determine the transcription start site(s). Total RNA was extracted from a P. gingivalis strain W83 culture grown to early exponential growth (OD₅₅₀ of 0.2). Rapid amplification of 5′ cDNA ends was carried out using a FirstChoice RLM-RACE kit (Ambion) according to the manufacturer's instructions. The nested-PCR conditions for 5′ outer PCR were 10 pmol gene-specific outer primer 106_5′_outer, 1.25 units of Ambion's SuperTaq polymerase, 10 pmol 5′ RACE outer primer (Ambion), 1× SuperTaq PCR buffer (Ambion), 100 μM deoxynucleoside triphosphates, 1 ng/μl first-strand cDNA reaction mixture, and H₂O to 50 μl. The PCR conditions were as follows: 1 cycle of 94°C for 3 min; 35 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s; and 1 cycle of 72°C for 7 min. The 5′ inner PCR was carried out with 10 pmol gene-specific inner primer 106_5′_inner and 10 pmol 5′ RACE inner primer (Ambion) using the products generated with the outer primer set as a template. The same conditions as for the 5′ outer PCR were used. The PCR products were observed on 2% agarose gels and then cloned into pCR4-TOPO for sequencing.

For Northern blot analyses, total RNA was mixed with NorthernMax formaldehyde loading dye (Ambion), heat denatured, and run out on a 1% denaturing agarose gel. After electrophoresis, the RNA was transferred to a positively charged nylon membrane (Hybond N+; Amersham) and then cross-linked with a UV Stratalinker (Stratagene). The probe for detection of PG0106 transcript was a 233-bp PCR amplicon generated from 33277 DNA using the F106 and R106 primer set. The probe for the detection of the PG0108 transcript was a 303-bp fragment amplified from 33277 DNA using the F108 and R1_0106 primer set. The probe for detection of PG0120 transcript was a 283-bp fragment amplified from W83 DNA using the F120 and R120 primer set. The primers are listed in Table 2. The labeling of the probe, hybridization of the blot, and detection of the signal were performed using the AlkPhos Direct labeling and detection kit (Amersham) according to the manufacturer's instructions.

Electroporation of P. gingivalis.

Electroporation of cells was performed as described previously (62) with some slight modifications. To prepare competent cells, 2 ml of an actively growing culture of P. gingivalis was used to inoculate 200 ml of prewarmed and prereduced TSBHK. The culture was then incubated overnight at 37°C to an OD₅₅₀ of between 0.3 and 0.6. The cells were harvested by centrifugation at 2,600 × g for 10 min at 4°C and then washed twice with 200 ml followed by 100 ml of electroporation buffer (10% glycerol, 1 mM MgCl₂ filter sterilized and stored at 4°C). After being washed, the pellet was suspended in 2 ml of electroporation buffer, aliquoted into microcentrifuge tubes (100 μl/tube), and frozen at −80°C for future use. A 100-μl sample of cells to which 5 μg of plasmid DNA was added was placed in a sterile electrode cuvette (0.2-cm gap). The cells were pulsed with a Bio-Rad gene pulser at 2,500 V, and then 1 ml of TSBHK was added and the cells were incubated anaerobically at 37°C for 16 h. The cells were concentrated by centrifugation, and the entire sample was plated on BAPHK containing erythromycin (5 μg/ml) for selection of DEL0106::Erm mutants or tetracycline (0. 5 μg/ml) for selection of strains harboring pT-COW and its derivatives. The plates were incubated anaerobically at 37°C for 4 to 7 days.

Biofilm formation assays.

Biofilm formation was assayed in 96-well microtiter dishes made of polystyrene (Falcon 353936) using a slightly modified previously published protocol (29). In brief, THBHK (200 μl/well) was inoculated from cells patched on BAPHK agar plates containing tetracycline. The microtiter plate was incubated anaerobically at 37°C for 48 h to provide a starter culture for the biofilm formation assay. The THBHK culture (20 μl) was then added to 180 μl of chemically defined medium (CDM) containing 1% tryptone (28, 59) and 0.02% l-cysteine-HCl as a reducing agent. The plates were then incubated anaerobically at 37°C for 48 h to allow time for biofilms to develop. The culture supernatant was removed, and the plate was washed twice by immersion in distilled water. The plates were air dried for 1 h, and then the biofilm was stained with 0.1% safranin for 15 min (100 μl per well) and washed again two times in distilled water. The amount of biofilm was then assessed macroscopically. To quantify the amount of biofilm, 95% ethanol (100 μl per well) was added to solubilize the safranin for 5 min. The ethanol was transferred to a new microtiter dish, and semiquantitative data were obtained by determining the absorbance (492 nm) with a plate reader.

A method to grow and microscopically examine P. gingivalis biofilms on polystyrene using 12-well plates (Falcon 353225) was also developed. To assay biofilm formation in this system, the P. gingivalis cultures were diluted to an OD₅₅₀ of 0.1 in CDM, as described above. Samples (750 μl) were aliquoted into the wells and incubated at 37°C under anaerobic conditions for 48 h. The plate was removed from the incubator, and the wells were washed twice with fresh CDM. CDM (1 ml) containing Syto-9 stain (Molecular Probes, Eugene, OR) was added to the well, and the slide was incubated for 15 min in the dark. The chamber was then washed two times with CDM. The wells were examined by confocal laser scanning microscopy (Leica TCSSP2 with AOBS) using a 40× water immersion objective.

Autoaggregation and hydrophobicity assays.

Autoaggregation assays of strains were performed in triplicate as described previously (79). Briefly, the strains were grown to early stationary phase, pelleted, and washed two times with phosphate-buffered saline (PBS). The cells were then suspended in PBS to an OD₅₅₀ of 0.8 and incubated at 37°C. The absorbances of the suspensions were monitored over time. A steady decrease in absorbance indicated autoaggregation and sedimentation. The hydrophobicity assay was performed in triplicate as previously described using hexadecane as the hydrophobic solvent (6, 26).

Microscopic examination.

To examine the presence of capsular polysaccharide, the cells were negatively stained with India ink (Fisher Scientific). The preparations were made by mixing a bacterial colony with 5 μl of India ink on a slide, which was then coverslipped and examined with a phase-contrast microscope at ×1,000 magnification. In addition, for more detailed analysis, cells were prepared for electron microscopy as previously described (23). Briefly, the strains were grown in TSBHK to early logarithmic phase and pelleted by centrifugation at 7,000 × g for 5 minutes. The cell pellets were then washed with PBS. The cell suspensions were fixed at room temperature for 2 h with 3.6% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.2), followed by secondary fixation at room temperature in 2% osmium tetroxide in 0.1 M phosphate buffer (pH 7.2) for 1 h. Ruthenium red (0.075%) and lysine (55 mM) were added to the glutaraldehyde fixative in order to preserve polysaccharide-containing material. The grids were viewed with a JEOL JEM-1200 EX electron microscope.

RESULTS

Isolation of the PG0106 mutant.

A collection of more than 3,000 P. gingivalis ATCC 33277 random-transposon mutants that was generated previously (8) was screened in polystyrene microtiter dishes for the ability to form a biofilm. Single-transposon mutants were grown in wells containing chemically defined medium supplemented with tryptone, and biofilm formation was assessed as described in Materials and Methods. Candidate mutants were tested twice in the same assay to confirm the results. Five mutants altered in biofilm formation were identified. One mutant was significantly enhanced in its ability to form a biofilm compared to the parental strain, and its reconstruction and characterization are the focus of this report.

To determine the site of insertion, we compared the DNA sequence flanking the transposon insertion site to the sequenced genome of P. gingivalis strain W83 (64). The sequence obtained was 100% identical to that of the putative glycosyltransferase gene PG0106 (Comprehensive Microbial Resource [CMR] database, the Institute for Genomic Research [TIGR] [http://cmr.tigr.org/CMR/Genomes.shtml]). The protein is predicted to be 378 amino acids in length, and the site of insertion was 234 base pairs from the 5′ end of the open reading frame (ORF). The PG0106 ORF had closest homology to an undecaprenyl-phosphate alpha-N-acetylglucosaminyltransferase from Bacteroides thetaiotaomicron VPI-5482 (39.5% identity; 66% similarity). In strain W83, the gene is at the beginning of a cluster of 16 genes (Fig. 1), and many of these are predicted to be organized as an operon (CMR database, TIGR). In addition, our sequence data and data from a previously published microarray study (9) indicate that the sequence of the PG0106 gene is conserved in all three wild-type strains used in this study.

FIG. 1.

Schematic representation of the capsule locus. The numbers correspond to TIGR CMR identifications. The region is predicted to contain 17 ORFs, with PG0106 the first in the series.

Reconstruction of the PG0106 mutation.

To confirm that the transposon insertion in PG0106 was responsible for the biofilm phenotype, we reconstructed the mutation in the parental strain, ATCC 33277, as well as two other wild-type backgrounds, W83 and 381. Mutant strains were constructed using pCR-4 TOPO as a suicide plasmid carrying a construct in which the entire PG0106 open reading frame was replaced with an erythromycin resistance gene cassette oriented in the opposite direction, and over 1 kb of flanking DNA was provided as homology for recombination. Integration of the construct by double crossover was verified by PCR using F1_106 with R1_106, Ermchk1 with F1_106, and Ermchk2 with R1_106 primer sets (data not shown). DEL0106::Erm is used to designate a strain with this deletion-insertion mutation.

Biofilm formation by PG0106 mutants.

The growth rates for the three DEL0106::Erm mutants were determined and found to be identical to those of the wild-type strains (data not shown), indicating that the altered biofilm phenotype could not be attributed to a change in growth rate. A 96-well biofilm assay was used to quantify the amounts of biofilm produced by the mutant and parent strains. Although the amounts of biofilm produced by the three parent strains varied greatly, all three of the corresponding mutant strains were enhanced in the ability to form a biofilm (Fig. 2). To assess biofilm formation microscopically, the strains were grown in 12-well polystyrene assay plates and the structures of the biofilms were examined by confocal laser scanning microscopy; top-down images of the biofilms are shown in Fig. 3. Overall, the three DEL0106::Erm mutants showed enhanced ability to form biofilms compared to their respective parent strains, and most notable were the biofilm phenotypes of W83 and the corresponding deletion mutant. The W83 parent strain was unable to initiate biofilm formation in our system; however, the DEL0106::Erm mutant formed a biofilm (Fig. 3), although it did not develop into a mature biofilm consisting of macrocolonies, such as those produced by strain 381 (Fig. 3). Strain 381 formed the most dense biofilms of all three wild-type strains, and as shown in Fig. 3, the mutant was able to form an even more complex biofilm consisting of larger macrocolonies. Although strain ATCC 33277 was the parental strain in which the transposon insertion mutant was first isolated, this strain forms a thin biofilm with very little topology (no macrocolony formation), and the corresponding mutant demonstrated the least enhancement in biofilm formation. However, a more even distribution of growth on the well surface (black subsurface is visible in the wild-type biofilm) was observed in the mutant strain, indicating that the cell-surface interaction is more stable, which could account for the enhanced-biofilm phenotype detected during screening.

FIG. 2.

Quantification of biofilm formation by the three parent strains and the corresponding DEL0106::Erm mutants. The assay was performed in 96-well microtiter plates, as described in Materials and Methods. The error bars represent standard deviations.

FIG. 3.

Microscopic examination of biofilms formed by the three wild-type strains and the corresponding DEL0106::Erm mutants. The confocal laser scanning microscopy fluorescent images are top-down views of 2-day-old biofilms grown under static conditions in a 24-well microtiter dish, as described in Materials and Methods.

Cell surface characteristics of P. gingivalis strains.

The screening for biofilm mutants was carried out using hydrophobic polystyrene plates, and we reasoned that the increased ability of mutants to form biofilms could be related to increased hydrophobicity. We used a hexadecane-partitioning assay to qualitatively assess the hydrophobicities of parent and mutant strains. As shown in Fig. 4, the W83 parent was hydrophilic; however, the W83 DEL0106::Erm mutant partitioned almost completely into the n-hexadecane solvent (a shift from ∼5% to 90%), indicating that the mutation resulted in a cell with a surface that was much more hydrophobic. This assay was previously used to assess the hydrophobicity of P. gingivalis strains. It was shown that strains ATCC 33277 and 381 are hydrophobic and do not make significant amounts of capsular polysaccharide (16, 82), and indeed, we found that these strains partitioned into hexadecane by approximately 90%. Increased partitioning was not observed in the respective PG0106 mutants.

FIG. 4.

Hydrophobic properties of the three wild-type strains and the corresponding DEL0106::Erm mutants. The graph shows the percentages of cells partitioning into the hydrophobic solvent hexadecane. The experiment was performed as described in Materials and Methods. The error bars represent standard deviations.

Cell surface characteristics were also assessed using an autoaggregation assay. Wild-type strain W83 showed very little autoaggregation during 5 h of incubation; however, the DEL0106::Erm mutant exhibited increased autoaggregation and sedimentation within 2 h (Fig. 5A). The absorbance of the W83 mutant culture dropped steadily during the 5 hours of assay. Also shown in Fig. 5D is a picture of the cultures at 5 h in which cell aggregates can be seen sedimented to the bottom of the culture tube. Strain 381 autoaggregated at a very high rate, but autoaggregation was slower with cells in the early logarithmic growth phase; therefore, cells in that growth phase were used in the assay with this strain. The DEL0106::Erm mutant of strain 381 autoaggregated at a slightly higher rate than its parental strain, as shown in Fig. 5C. Surprisingly, unlike the other two wild-type strains, strain ATCC 33277 demonstrated more autoaggregation than its DEL0106::Erm mutant. This is shown in Fig. 5B. Thus, the DEL0106::Erm mutant of 33277 appears to have less cell-cell interaction.

FIG. 5.

Autoaggregation of the three wild-type strains and the corresponding DEL0106::Erm mutants. A decrease in absorbance over time is the result of aggregation and precipitation. The error bars represent standard deviations. (D) Image of cultures of W83 and the DEL0106::Erm mutant at the final time point (5 h) showing precipitated cell aggregates.

Microscopic examination of capsule and complementation analysis.

According to the annotation of strain W83, the putative glycosyltransferase encoded by PG0106 was clustered with several genes predicted to be involved in capsular-polysaccharide synthesis. Therefore, we examined the W83 parent and DEL0106::Erm mutant strain for the presence of capsule by bright-field microscopy using India ink to negatively stain the cells. As shown in Fig. 6A, the encapsulated wild-type strain W83 demonstrated a clear zone around the cell, indicative of capsular polysaccharide. However, the DEL0106::Erm mutant did not show a zone of clearing, indicating that the strain does not produce a capsule (Fig. 6B). As noted above, it has been shown that strains ATCC 33277 and 381 are hydrophobic and do not make significant amounts of capsular polysaccharide (16, 82). India ink staining of these two strains is shown in Fig. 6E and F and clearly demonstrates the lack of capsule production by our laboratory stocks.

FIG. 6.

Microscopic examination for the production of capsule. Shown are phase-contrast images of cells negatively stained with India ink. (A) The clear zone around the cells indicates the presence of capsule. W83 (wild type) is encapsulated. (B) The corresponding DEL0106::Erm mutant does not produce capsule. (C) Capsule expression in the deletion mutant was rescued by complementation with plasmid pT-C106 containing PG0106 and its putative promoter region. Panel D shows the mutant containing the pT-COW vector only (vector control), and panel E and panel F show the other two wild-type strains, 381 and 33277, respectively.

We also performed complementation analysis in wild-type strain W83, because it is encapsulated and thus more relevant for further investigations and because this parent strain does not form a biofilm (Fig. 2). For complementation, a PCR amplicon of W83 DNA was created using the F3_106 and R1_106 primer set. The ORF of the PG0106 gene plus 1,010 bp of upstream DNA containing a putative promoter region and 259 bp downstream, which included 11 bp of the 5′ end of the next putative ORF (PG0108), was cloned into vector pT-COW. The plasmid (pT-C106) was introduced into the W83 PG0106 mutant, and the phenotype was determined in this complemented strain. The inability to form a biofilm was not restored in the complemented strain (data not shown); however, a partial rescue of capsule expression was observed. The addition of the wild-type gene in trans rescued capsule production when cells were grown on agar plates for 48 h (Fig. 6C); however, in contrast to the wild-type W83 strain, no capsule was detected in cells in liquid cultures (data not shown). Although a complete rescue of the phenotype was not observed, these data do confirm that the mutation in this putative glycosyltransferase gene is linked to the loss of capsule production in strain W83. Failure to restore the phenotype to match the parent strain is likely due to the fact that the mutation not only affected PG0106 expression, it also affected the expression of the downstream genes that are predicted to be involved in capsule synthesis and transport. The complex transcription of this locus is further addressed in Discussion below.

To further investigate the effect of the mutation on capsule production, we examined the W83 parent and the DEL0106::Erm mutant strain by electron microscopy after treatment with ruthenium red, a nonspecific stain for polysaccharide. As shown in Fig. 7, the mutant exhibited more cell-cell interaction than the parent and lacked the amorphous capsular polysaccharide clearly found to coat the wild-type cells. These results indicated that the putative glycosyltransferase is indeed involved in the production of capsule. Furthermore, since the mutant strain demonstrated more cell-cell and cell-surface interactions, it is likely that the lack of capsule reveals surface structures, such as surface proteins or lipopolysaccharide (LPS), that are otherwise masked by the capsule. Thus, the presence or absence of capsular polysaccharide can potentially alter the overall surface characteristics of cells.

FIG. 7.

Transmission electron micrographs of the wild-type strain W83 (left) and the corresponding DEL0106::Erm mutant (right). The cells were stained with ruthenium red to determine the presence of capsular polysaccharide, as described in Materials and Methods. The arrows point to capsular polysaccharide on the cell surface.

Northern blot analysis of the capsule locus.

A Northern blot was prepared with RNA from cells in early exponential growth as described in Materials and Methods. The blot was hybridized with a 233-bp PCR amplicon amplified from within the PG0106 ORF. As shown in Fig. 8A, a transcript greater than 9 kb was detected in mRNA prepared from strain W83. This transcript is considerably larger than the 1.2 kb predicted for the PG0106 ORF, indicating that PG0106 is part of a polycistronic message in this strain. As also shown in Fig. 8A, hybridization did not occur in the W83 DEL0106::Erm mutant, thus confirming deletion of the PG0106 ORF and probe specificity. Interestingly, no transcript was detected in the wild-type strains 381 (Fig. 8A) and 33277 (data not shown), indicating that either this gene was not expressed or it is an unstable transcript or expressed at levels below detection.

FIG. 8.

Northern blot analysis of PG0106 and PG0108 expression. Total RNA was isolated from cells during early exponential growth and late exponential growth (EXP) and from P. gingivalis cells grown on TSBHK agar plates (plate) as described in Materials and Methods. (A) Expression of PG0106 during early exponential growth in strains W83 and 381 and the corresponding deletion mutants. (B) Expression of PG0106 in strain W83 during late exponential growth compared to expression in all three wild-type strains grown on agar plates. (C) Same blot as shown in part B hybridized with a probe for PG0108. (D) Expression of PG0120 during early exponential growth in strains W83 and 381 and the corresponding deletion mutants. The arrows in all four panels indicate the areas where transcripts were detected with the probes.

To evaluate expression of the PG0106 transcript in cells growing in colonies, RNA was prepared from cells grown on agar plates. In addition, RNA was prepared from cells in liquid to late exponential growth for comparison. A Northern blot of the three wild-type strains was prepared and hybridized with the PG0106 probe (Fig. 8B). Unlike the data in Fig. 8A, multiple transcripts were detected. The predominant transcripts in strain W83 during late exponential growth were between 4 and 9 kb in size, although the large transcript (>9 kb) was also detected. Also, an additional transcript (approximately 1 kb) hybridized to the PG0106 probe. On the other hand, although a similar hybridization pattern was detected in RNA prepared from cells grown on agar plates (lane 2), the >9-kb transcript was not detected, indicating that the transcriptional profiles are different in liquid-grown cells versus colony-/agar plate-grown cells. Strain 381 did not demonstrate strong hybridization to the probe, although some faint bands were visible by eye and a smear was detected below 2.0 kb in strain 33277. These data again indicate that there are very low levels of PG0106 transcription in strains 381 and 33277 or that the transcript is not stable.

To further evaluate transcription from this locus, the blot was stripped of the PG0106 probe and hybridized with a probe to PG0108. PG0108 is predicted to be the next ORF downstream of PG0106 and to be transcribed with PG0106; however, a noncoding intergenic region of 249-bp is located between the stop codon for PG0106 and the start codon for PG0108, which may encode a promoter region (Fig. 1). As shown in Fig. 8C, the PG0108 probe had a hybridization pattern similar to that of the probe for PG0106, except that the band that hybridized below 1 kb with the PG0106 probe was absent. We propose that this small transcript is likely the PG0106 gene, to which the PG0108 probe would not hybridize. Also, the hybridization pattern with the PG0108 probe also indicates that the transcriptional profiles are different in liquid-grown cells versus colony-/agar plate-grown cells. In addition, the blot shown in Fig. 8A, prepared with RNA from cells in early exponential growth, was stripped of the PG0106 probe and hybridized with the probe for PG0108. A similar hybridization pattern was detected, indicating that these two genes are parts of the same large transcript and that the deletion mutants do not express PG0108. These data indicate that the mutation in PG0106 has a polar effect on PG0108 expression, at least during early exponential growth (data not shown).

To further analyze operon size, another Northern blot was prepared with RNA from cells in early exponential growth, and this blot was hybridized with a probe for PG0120. This ORF is predicted to be at the 3′ end (14,974 base pairs downstream of PG0106) of the putative operon. The probe was also found to hybridize to multiple transcripts. One predominant transcript that was greater than 9 kb hybridized, and two other transcripts (4 kb and 6 kb) were also detected, albeit at very low levels (Fig. 8D, lane 1). Interestingly, although the mutation (lane 2) eliminated expression of the predominant transcript (>9 kb), the two smaller transcripts were still detected in the W83 mutant, indicating that although transcription of the large polycistronic message (PG0106 through PG0120) was eliminated by the mutation in PG0106, the mutation did not eliminate the expression of the other two PG0120 transcripts.

Identification of PG0106 transcriptional start sites.

RLM-RACE was carried out with total RNA extracted from a P. gingivalis strain W83 culture grown to early exponential growth and was used to determine transcription start points. Rapid amplification of 5′ cDNA ends was carried out using a FirstChoice RLM-RACE kit (Ambion) according to the manufacturer's instructions. The adapter and primer sequences used are shown in Table 2. In brief, a 45-base RNA adapter was ligated to the RNA population using T4 RNA ligase. This RNA population was then used as a template for a random-primed reverse transcription reaction. The cDNA product of this reaction was then used as a template for PCR using a PG0106-specific outer primer and a reverse primer to the adapter. Specifically, a reverse primer was designed that was located 191 bp from the start of PG0106 translation (106_5′ outer primer) (Fig. 9A). The product was confirmed using the inner gene-specific primer, and then the amplicon was purified, cloned into pTOPO, and sequenced. The junction of the adapter with P. gingivalis DNA is the 5′ end of the transcript. As shown in Fig. 9, four transcriptional start points were identified in the putative promoter region of PG0106. These sites were located 31, 48, 54, and 64 base pairs upstream from the first nucleotide of the translation initiation codon.

FIG. 9.

The promoter region of PG0106. (A) Determination of the transcriptional start sites by RLM-RACE. Outer PCR (lane 1) was performed using total cDNA as a template with the RACE outer adapter primer and the gene-specific outer primer (106_5′ outer primer) set. Subsequently, inner PCR (lanes 2 and 3) without template (lane 2) or with template (lane 3) from the previous outer reaction were performed using the RACE inner adapter and the inner gene-specific (106_5′ inner primer) primer set. M, molecular size ladder. (B) The coding region starting with GTG is shown, with the upstream promoter region in italics. The annealing sites of the gene-specific inner and outer primers used in RLM-RACE are designated by arrows, and the identified transcriptional start sites are boldface and shaded.

DISCUSSION

A key element of bacterial survival is the ability to create a protective niche where bacteria can propagate successfully. Bacteria that inhabit the oral cavity persist by adhering to a variety of surfaces, such as the saliva-coated tooth, epithelial cells, and one another. Without secure attachment, bacteria are washed away; hence, attachment and the development of biofilm communities are fundamental to persistence in this environment. Physical and/or chemical properties that operate at interfaces can profoundly influence attachment. Bacteria can actively overcome these forces by excreting products that alter the properties of an interface, and they can adapt to the surroundings by altering their own surface properties. Biofilm development involves the attachment of cells to and growth on a surface, followed by detachment and dissemination to a new site to start the cycle again (67). It is likely that much of biofilm-specific physiology is devoted to dynamic changes that occur at the cell surface throughout this cycle. Here, we discuss how capsule expression affects biofilm formation in P. gingivalis.

To our knowledge, there is only one report of a genetic locus affecting biofilm development in this anaerobic bacterium. That study showed that defects in polyphosphate kinase activity result in a strain that is attenuated in biofilm formation, indicating that the synthesis of this storage polymer [short-chain poly(P)] is involved in biofilm development (10). A number of genes have been discovered that affect cell-cell and cell-surface interactions, and by virtue of these properties, they have the potential to affect biofilm development. In a recent study, it was found that a mutation in gtfA, which is a putative glycosyltransferase gene (PG0750), inhibits autoaggregation, attachment to epithelial cells, and the maturation of fimbriae on the surface (63). Research on the pathogenicity of P. gingivalis has clearly demonstrated a role for long peritrichous fimbriae in attachment to human epithelial cells, in coaggregation, and in autoaggregation (3, 30, 54, 80). In addition, gingipains, which are surface-associated proteases, have been found to be involved in the adherence of P. gingivalis to other bacteria, collagen, immobilized extracellular matrix proteins, and human epithelial cells (79). All of these interactions are likely to be essential to the initiation and development of biofilm in vivo.

Interestingly, the major fimbriae project from the cell far beyond the capsule structure, yet strains of P. gingivalis that are encapsulated are typically deficient in the ability to adhere to bacterial and epithelial cells. Little is known about the interrelationship between the expressions of these two surface structures that are potentially antagonistic with regard to their effects on attachment. It has been shown that the encapsulated W83 strain does not typically produce fimbriae (82), although the genes required for the synthesis of this surface structure are present in the chromosome of the strain (64). Possibly, these two virulence factors are not expressed at the same time. The in vivo signal that triggers fimbrial expression is not known; however, it has been shown that expression of fimA, which encodes the fimbrillin protein subunit, is affected by environmental and nutritional parameters, since expression was repressed under elevated temperatures or hemin limitation (4, 58, 88). A two-component system (FimS-FimR) that regulates the expression of fimA was recently discovered (33, 81, 83), and it has been shown that the response regulator FimR controls expression at the level of transcription (65). Further studies of the function of PG0106 and the regulation of the expression of capsule are required to determine if the expressions of fimbriae and capsule are coordinated.

In order to delimit the promoter region of PG0106, we performed 5′ RACE to identify transcriptional start sites. Our analyses discovered four transcriptional start sites, indicating that there is potential for complex regulation, and given the importance of capsule in cell structure and surface properties, as well as the energy cost of its synthesis, it is likely that expression is regulated. The complexity of transcription was also reflected in our Northern blot analyses. Our studies determined that multiple transcripts are encoded in the capsule locus, and although PG0106 is present in multiple polycistronic transcripts, it is also transcribed separately. Furthermore, although expression of the transcript including the entire operon is eliminated by the mutation in PG0106, the mutation does not appear to eliminate the expression of transcripts including downstream genes under all growth conditions. We propose that this accounts for the ability of the wild-type gene to partially complement the phenotype in trans.

Although it is well documented that excreted exopolysaccharide is an integral part of biofilm development (13), a role for cell-associated capsular polysaccharide in biofilm formation is only beginning to be investigated. Like that in many encapsulated pathogenic bacteria, capsule expression in P. gingivalis is linked to pathogenicity. Interestingly, strains that produce a capsule have been shown to cause a spreading type of infection in mice, with recovery of the organism from blood, spleen, and kidneys (16, 52), while infections with strains that do not synthesize capsule (strains 381 and 33277) typically attach and result in localized abscesses (43, 52), suggesting that encapsulation is linked to dissemination and not attachment. Our results show that capsule expression in P. gingivalis blocks biofilm formation and hence could enhance dissemination. It was recently reported that capsule expression inhibits biofilm formation in Vibrio vulnificus, an opportunistic pathogen that attaches to a variety of surfaces in marine estuaries and causes severe infections associated with raw-oyster consumption (42). It has also been determined that capsule expression in V. vulnificus is regulated by growth phase (87) and by phase variation (86), indicating that both environmental and genetic parameters impact V. vulnificus biofilm development. Another encapsulated organism whose biofilm formation phenotype has been investigated is Neisseria meningitidis. This bacterium is the causative agent of bacterial meningitis, and its ability to colonize the human nasopharyngeal mucosal surface is a key aspect of its pathogenicity. Non-biofilm-forming strains of N. meningitidis were encapsulated, and a capsule-transport mutant (ctrA) gained the ability to form a biofilm (89), similar to the phenotype we observed here with P. gingivalis. Also, it was shown that the expression of capsule, as well as pili, is repressed in N. meningitidis upon contact with host epithelial cells, and this down-regulation facilitated attachment (14). Hence, a correlation of the expression of capsule with planktonic growth and the inability to form a biofilm is being discovered in a variety of bacteria.

Although all three wild-type strains show enhanced biofilm formation when this putative glycosyltransferase is deleted, its role in strains 33277 and 381 (strains that do not typically produce a capsule) is more difficult to ascertain. Moreover, since strain ATCC 33277 demonstrated more autoaggregation than its DEL0106::Erm mutant while the opposite effect was observed in the 381 wild-type strain and its corresponding DEL0106::Erm mutant, PG0106 may have different activities in these two wild-type strains. According to TIGR annotation, the gene is predicted to be involved in the biosynthesis and degradation of surface polysaccharides and LPS, and therefore, the deletion could result in a strain that is defective in LPS synthesis. A defect in LPS synthesis (expression of a truncated LPS structure) has been shown to enhance biofilm formation in Salmonella enterica (60) and in Neisseria meningitidis (89). The altered biofilm phenotype could also be the result of inactivation of genes downstream of PG0106. Clearly, further studies of the roles of this locus in these two wild-type strains are required.

Recently, the sequence surrounding PG0106 in wild-type strain 381 was submitted to NCBI (accession number AJ969093) (2). The region showed similarity to the W83 locus, including conservation of PG0106, PG0108, and PG0120; however, there are significant differences, including replacement of PG0109 and PG0110 with three distinct open reading frames. Also, PG0112, PG0113, and PG0114 are missing in strain 381. In addition, a study using gene microarrays to compare the genome of strain W83 with that of strain 33277 indicated that genes in this region are either absent or divergent in strain 33277 (9), and these data are in agreement with the sequence data described above for strain 381. Our sequence data showed that the PG0106 ORF is conserved in all three strains, and this result is supported by microarray and PCR analysis studies (9). Interestingly, the submitted 381 sequence predicted that PG0106 was interrupted by a change from a “G” at base number 637 to a “T” that generated an internal termination codon; however, although our sequence data (not shown) from both strains 381 and ATCC 33277 were identical to the NCBI 381 sequence in all other respects, we found base number 637 to be a “G” and not a “T,” and therefore, an internal stop codon in PG0106 was not found. The resulting ORF demonstrated 98% amino acid identity to PG0106 from strain W83. Our Northern blot analyses also clearly indicate that transcription from this region is distinctly different in the three wild-type strains. Further studies are required to determine the functions of this putative glycosyltransferase in the different P. gingivalis strains.

In this study, it was apparent that a mutation in PG0106 in strain W83 disrupted expression of an operon and abolished capsule production. This caused a dramatic change from a hydrophilic to a hydrophobic bacterial surface and an increase in autoaggregation of P. gingivalis cells as a consequence. Similar hydrophobicity data were reported in a previous study that compared the rapid adhesion of strain ATCC 33277 to hexadecane (75% of cells in 4 min) to that of strain W50, a close relative of W83, where only 25% of input cells adhered in the same time period (73). In those experiments, the initial coadhesion of Actinomyces viscosus to P. gingivalis on the surfaces of hexadecane droplets appeared to be preferential and specific. As noted above, it is well documented that interspecies aggregation plays a large role in the development of dental plaque and other mixed-species biofilms (72), and because of the specificity of these interactions, they are considered to be biochemically driven, e.g., mediated by specific lectins and their cognate receptors/ligands. However, autoaggregation driven by nonspecific hydrophobic mechanisms may also contribute to multi- and monospecies biofilm formation through the formation of aggregation foci/nuclei. We propose that the change in surface properties of W83 DEL0106::Erm resulted from a lack of capsule and hence exposure of short hydrophobic surface structures. Previously, it was proposed that autoaggregation could be attributed to the long peritrichous fimbriae on the cell surface (51), but this does not seem to be the cause of the effect observed here, since strain W83 does not produce fimbriae. However, a variety of surface proteins are potentially exposed when the capsular polysaccharide is not present. Unlike cell-cell interactions mediated by polymeric structures, such as fimbriae, that reach far out from the bacterial cell surface, aggregation through surface proteins reflects intimate contact. A well-studied example of protein-mediated intimate contact is self-recognition of the surface adhesin antigen 43 (Ag43) autotransporter from E. coli. Ag43 protrudes approximately 10 nm from the cell surface, and its activity, as well as that of another potent adhesin protein (AIDA-I), can be blocked by the presence of a capsule (74). Ag43 is encoded by the flu gene, which was found to be highly expressed in E. coli biofilms (75). Moreover, Ag43 has been shown to greatly enhance biofilm formation (12) and to facilitate the development of multispecies biofilms (45). It has even been proposed that the primary function of this protein in E. coli is to promote cell-cell interaction for biofilm development (46).

Our results indicate that P. gingivalis has the potential to modulate interactions with the surface and with other cells by differential capsule expression, and it is possible that the expressions of capsular polysaccharide, surface adhesins, and fimbriae are regulated in concordance. Recent studies of Bacteroides fragilis (a close relative of P. gingivalis) have shown that capsular-polysaccharide expression is phase variation controlled through reversible inversions of promoters (49, 85). Studies of P. gingivalis have shown that elevated temperatures, which can occur in inflamed periodontal tissue, can reduce the expression of fimbriae (4). Since phase variation events can be random or modulated by environmental cues, it will be interesting to study the promoter region of PG0106 and to determine the mechanism(s) controlling expression and if elevated temperatures or other environmental parameters affect the expression of this locus required for capsular-polysaccharide production.