|Home | About | Journals | Submit | Contact Us | Français|
Porphyromonas gingivalis is a late-colonizing bacterium of the subgingival dental plaque biofilm associated with periodontitis. Two P. gingivalis genes, fimR and fimS, are predicted to encode a two-component signal transduction system comprising a response regulator (FimR) and a sensor histidine kinase (FimS). In this study, we show that fimS and fimR, although contiguous on the genome, are not part of an operon. We inactivated fimR and fimS in both the afimbriated strain W50 and the fimbriated strain ATCC 33277 and demonstrated that both mutants formed significantly less biofilm than their respective wild-type strains. Quantitative reverse transcription-real-time PCR showed that expression of fimbriation genes was reduced in both the fimS and fimR mutants of strain ATCC 33277. The mutations had no effect, in either strain, on the P. gingivalis growth rate or on the response to hydrogen peroxide or growth at pH 9, at 41°C, or at low hemin availability. Transcriptome analysis using DNA microarrays revealed that inactivation of fimS resulted in the differential expression of 10% of the P. gingivalis genome (>1.5-fold; P < 0.05). Notably genes encoding seven different transcriptional regulators, including the fimR gene and three extracytoplasmic sigma factor genes, were differentially expressed in the fimS mutant.
Two-component signal transduction systems (TCSTS) are used by bacteria to control the expression of a range of genes in response to a variety of environmental and intracellular stimuli. These systems are found in almost all bacteria and are known to regulate an array of physiological traits, including osmoregulation (4), virulence (8), and quorum sensing (21). The crucial role of TCSTS in governing the signaling and regulatory pathways associated with biofilm development has been well documented in many bacteria, including Escherichia coli, Pseudomonas aeruginosa, and Streptococcus mutans (11, 25, 39). Typically, each TCSTS functions via a phosphorylation cascade and consists of a membrane-bound or cytoplasmic sensor histidine kinase (SHK), which perceives a particular stimulus, and a cytoplasmic response regulator (RR), which allows the cell to respond to the stimulus accordingly via regulation of gene expression (49).
Porphyromonas gingivalis is a Gram-negative anaerobe that has been strongly implicated as a major etiologic agent in the onset and progression of chronic periodontitis (47, 55), a disease of the supporting tissues of the teeth. P. gingivalis is a late colonizer of subgingival dental plaque, a complex and dynamic polymicrobial biofilm (22, 31), and its ability to persist as part of a subgingival plaque biofilm is dependent on its adherence to and colonization of the subgingival niche. P. gingivalis has been shown to adhere to primary plaque-colonizing species, particularly Streptococcus spp. such as Streptococcus gordonii (24, 29, 33, 48). Binding of P. gingivalis to S. gordonii has been shown to result in the formation of a bispecies biofilm with P. gingivalis attached to S. gordonii bound to a salivary pellicle (7). P. gingivalis also adheres to later-colonizing Gram-negative bacteria, including Fusobacterium nucleatum (33, 43, 44) and Treponema denticola (14, 56). Furthermore, P. gingivalis can adhere to host tissues, including gingival epithelial cells (6, 23, 40, 53). Therefore, to colonize and persist within a host P. gingivalis must sense the presence of a variety of surfaces and respond via coordinated gene expression. Given the variation in surfaces to which P. gingivalis may attach, it is likely that numerous cell structures are required to mediate the interactions necessary for specific and stable adherence. Indeed, recent studies have shown that both the major fimbrillin FimA and the minor fimbrillin Mfa1 are required for full P. gingivalis biofilm development by strain ATCC 33277 (26). In addition, the involvement of capsular polysaccharide and lipopolysaccharide O antigen has been implicated in P. gingivalis biofilm formation (10, 35). P. gingivalis W50 is afimbriated (53), so the mechanism of biofilm formation for this strain is unclear. While stable attachment is clearly critical for the establishment of P. gingivalis in the subgingival niche, continued survival at this site requires appropriate bacterial responses to a range of adverse conditions, including oxidative and nutrient stresses as well as variations in temperature and pH. It is likely that many of these responses are regulated by TCSTS.
Bioinformatics analysis of the P. gingivalis W83 genome sequence (36) identified 6 putative TCSTS, one of which, GppX, is a predicted fusion of both SHK and RR proteins (16). A study comparing the transcriptomes of biofilm and planktonic P. gingivalis strain W50 cells identified two genes, pg1431 and pg1432, that were highly upregulated (4.2-fold and 5.2-fold, respectively) (27) during biofilm growth. The pg1431 gene is predicted to encode a 227-amino-acid, 25.5-kDa putative DNA-binding response regulator of the LuxR family, while the pg1432 gene is located upstream of, and in the same orientation as, pg1431 and is predicted to encode a 621-amino-acid, 70.1-kDa putative sensor histidine kinase. Genes homologous to pg1431 and pg1432 have been identified previously in P. gingivalis strain ATCC 33277 and were designated fimR and fimS, respectively (17); therefore, we designate the W50 pg1431 gene as fimR and the W50 pg1432 gene as fimS. It has been assumed that FimR (RR) and FimS (SHK) work in concert as a two-component regulatory system. Inactivation of either fimS or fimR in strain ATCC 33277 resulted in loss of fimbriation (17). Furthermore, FimR has been shown to directly regulate the expression of a limited number of genes, including pg2130, which is associated with FimA fimbriation (37), and mfa1, which encodes the minor fimbrillin Mfa1 (54). The genes associated with the function of FimS have not been experimentally determined.
Here, we demonstrate that both FimR and FimS are involved in P. gingivalis biofilm formation, including the regulation of genes associated with fimbriation. Furthermore, DNA microarray analysis of a W50 fimS mutant suggests that this SHK has a broad role in P. gingivalis gene regulation.
The bacterial strains and plasmids used in this study are listed in Table Table1.1. E. coli strain JM109 (Promega, Madison, WI) was grown in Luria-Bertani (LB) broth or on LB agar plates at 37°C under aerobic conditions. Freeze-dried cultures of P. gingivalis strains W50 and ATCC 33277 were obtained from the culture collection of The Melbourne Dental School, The University of Melbourne. P. gingivalis strains were grown and maintained as previously described (46). Growth media were supplemented with 10 μg ml−1 erythromycin (Sigma) or 100 μg ml−1 of ampicillin (Sigma) when appropriate.
P. gingivalis was grown in continuous culture for 30 days, in duplicate, using a Bioflo 110 fermentor with a total volume of 400 ml (New Brunswick Scientific, Edison, NJ), as previously described (9). Planktonic cells were harvested from the fermentor by rapidly pumping them out, and the RNA was harvested using an acidic hot phenol procedure (27). Culture purity was assessed regularly by Gram staining and colony morphology.
Oligonucleotide primers used in this study are listed in Table Table2.2. Genomic DNA from P. gingivalis strains W50 and ATCC 33277 and mutant strains were prepared using the DNeasy blood and tissue kit (Qiagen, Valencia, CA), and plasmid DNA from E. coli was extracted using the Qiagen Miniprep kit (Qiagen). The Pfu DNA polymerase and restriction endonucleases (Promega) and Platinum Taq DNA polymerase High Fidelity (Invitrogen Life Technologies, Carlsbad, CA) were used according to the manufacturer's instructions. Sequencing of DNA was performed by Applied Genetic Diagnostics, The University of Melbourne. Sequence alignments were done with the ClustalW program (http://www.ebi.ac.uk/Tools/clustalw2/) (50). Where a PCR amplicon was sequenced, two separate PCRs were performed to ensure sequence consensus.
The fimR loci from strains W50 and ATCC 33277 were each amplified by PCR using the oligonucleotide primer pair PG1432Seq-For and PG1430Seq-Rev, which annealed to the flanking genes. The W50 fimS locus was amplified using PG1433Seq-For and PG1431Seq-Rev, while the strain ATCC 33277 fimS locus was amplified using the oligonucleotide primer pair PG1431Seq-Rev and fimS_Seq-For. The resulting amplicons were purified using the QIAquick PCR purification kit (Qiagen), and the nucleotide sequences were determined.
To make the fimR mutagenesis cassette, a 353-bp DNA fragment containing the 5′ region of fimR, with flanking AatII and BamHI restriction sites, was generated by PCR using P. gingivalis W50 DNA as template and the oligonucleotide primers PG1431-AatII-For and PG1431-BamHI-Rev. This amplicon was digested with AatII and BamHI and ligated into the AatII and BamHI sites, adjacent to the ermF gene within pAL30 (9), to create pAL31. Similarly, a 264-bp DNA fragment containing the 3′ region of fimR, with flanking KpnI and SpeI restriction sites, was amplified using the oligonucleotides PG1431-KpnI-For2 and PG1431-SpeI-Rev2 and ligated into the KpnI and SpeI restriction sites in pAL31. The resulting plasmid, designated pAL31.1, had the ermF cassette flanked by fimR DNA. Plasmid pAL31.1 was linearized with ScaI and transformed into P. gingivalis strains W50 and ATCC 33277 by electroporation as previously described (12). Transformants were selected after 7 days of anaerobic incubation on horse blood agar plates containing 10 μg ml−1 of erythromycin. Gene disruptions were confirmed by PCR.
A similar strategy for fimS inactivation was followed, in which the 5′ and 3′ regions of fimS were amplified by PCR using the primer pairs PG1432-AatII-For and PG1432-BamHI-Rev (5′ region) and PG1432-KpnI-For2 and PG1432-SpeI-Rev2 (3′ region). These fragments were sequentially ligated to pAL30, to finally produce pAL32.1. This plasmid was introduced into strains W50 and ATCC 33277 by electroporation, and transformants were selected as described above.
Total RNA for reverse transcription-PCR (RT-PCR) was harvested from P. gingivalis W50 chemostat-grown planktonic cells (27) and from exponential-phase ATCC 33277 batch culture cells following the protocol of Dashper et al. (9). The total RNA (1 μg of each) was reverse transcribed to cDNA using the Superscript III first-strand synthesis Supermix kit (Invitrogen) with random hexamer oligonucleotide primers. The cDNA (10 ng) was used as the template for PCR using BIOTAQ Red DNA polymerase (Bioline, Alexandria, Australia) with genomic DNA (10 ng) as a PCR-positive control. Total RNA that had not been subjected to reverse transcription was used as a control to show that the RT-PCR amplicons had not resulted from amplification of contaminating genomic DNA.
For quantitative real-time RT-PCR (qRT-PCR) analysis, total RNA was harvested from at least three separate cultures of each P. gingivalis strain, ATCC 33277, ECR221, and ECR223, during exponential-phase and stationary-phase growth in brain heart infusion (BHI) medium batch culture. RNA was harvested following the protocol of Dashper et al. (9) with cDNA synthesis and qRT-PCRs using 100 ng of input RNA and the Superscript III Platinum 2-step qRT-PCR SYBR green kit reagents and protocol (Invitrogen, Van Allen Way, CA). The qRT-PCRs were carried out using a Rotor-Gene 3000 instrument (Qiagen, Sydney, Australia). Melting curve analysis was performed in the temperature range from 50 to 99°C in 0.2°C increments. A no-RT control was used in each run. The mRNA abundance was determined by comparing the values obtained from the qRT-PCR to a standard curve generated using ATCC 33277 genome DNA. The expression of the housekeeping gene galE (27) was used for normalization between samples.
Total bacterial RNA, harvested from duplicate W50 and ECR222 chemostat-grown planktonic cultures, was reverse transcribed to cDNA and then labeled with Cy3 and Cy5 for use in DNA microarray hybridizations with P. gingivalis oligonucleotide arrays as described previously (27). Four DNA microarrays (kindly provided by the Pathogens Functional Genomics Resource Centre; http://pfgrc.jcvi.org) were used for each biological replicate comparison, with a dye-swap design, making a total of eight slides used in the analysis. Scanned images of the hybridized arrays were analyzed using Imagene 6.0 software (Biodiscovery, Los Angeles, CA) with local background correction. Intensity-dependent Lowess normalization was applied using GeneSight 4.1 (Biodiscovery), as described previously (27). Differentially expressed genes were identified at 95% confidence intervals with a fold change threshold value of 1.5. All DNA microarray work in this study was in compliance with MIAME guidelines.
Strains were initially grown overnight in an anaerobe chamber, and an aliquot of cells (1 × 109 CFU ml−1) was added to (i) fresh BHI medium for growth kinetics and temperature stress assays, (ii) BHI medium supplemented with 1 mM H2O2 (H2O2 stress), (iii) 0.1 μg ml−1 instead of 5 μg ml−1 of hemin (hemin limitation stress), or (iv) BHI medium adjusted to pH 5.0 or pH 9.0 (pH stress). The cells (260 μl) were transferred to triplicate wells of a microtiter plate (Falcon 353072; Becton Dickinson, North Rye, NSW, Australia), which was then sealed and inserted into a microtiter plate reader (Labsystems iEMS reader MF; Labsystems, Helsinki, Finland) and incubated at 37°C. The growth of each strain was monitored by measurement of the optical density at 620 nm (OD620). Growth curves were constructed using the mean and standard deviation of three separate assays. Wells containing BHI medium only were used as blank controls.
Static biofilm formation was assayed using the protocol of O'Toole and Kolter (38) with slight modification. Briefly, an overnight culture was diluted with fresh BHI medium to obtain 5 × 107 CFU ml−1. The cells were aliquoted into the wells of a 96-well microtiter plate (260 μl per well) and incubated anaerobically at 37°C for 24 h. The supernatant of the culture was aspirated, and then the well was washed twice with phosphate-buffered saline (150 mM NaCl, 3 mM KCl, 10 mM Na2HPO4, and 1.5 mM KH2PO4, pH 7.4). The biofilms were stained by incubation of each well with 100 μl of 0.1% crystal violet (CV) for 5 min. The plate was then washed twice with distilled water and destained with 95% ethanol (200 μl per well) for 5 min. The solubilized CV was transferred to a new microtiter plate, and the OD540 was measured. Biofilm formation was qualitatively determined to be proportional to the absorbance of the CV.
Static biofilms were generated in a 16-well chambered coverglass system (Grace Biolabs) as described previously (5) and stained using the BacLight bacterial viability assay kit (Invitrogen) according to the manufacturer's instructions. The biofilms were examined using a Zeiss LSM 510 Meta confocal laser scanning microscope (CLSM) with a C-Apochromat 63×/1.2 numerical aperture, water immersion objective lens with correction collar. SYTO 9 fluorescence (green, live cells) was detected by excitation at 488 nm, and emission was collected with a 500- to 550-nm bandpass filter. Propidium iodide (PI) fluorescence (red, dead cells) was detected by excitation at 488 nm, and emission was collected with a 560-nm long-pass filter. All images were obtained over an area 142.9 × 142.9 μm in the x-y plane (parallel to the surface). z-stack images were obtained by taking serial optical slices in this plane over a range of distances at a resolution of 1,024 by 1,024 pixels. Three independent static biofilm experiments were performed for each strain, and at least 9 image stacks were acquired for each experiment. These image stacks were quantitatively analyzed using COMSTAT software (18) (The Math Works, Inc., Natick, MA) to determine the biomass, mean thickness, and roughness measurements of the biofilms formed by all strains.
Biofilm parameters obtained using confocal microscopy and transcript levels measured using qRT-PCR were statistically analyzed using a one-way analysis of variance (ANOVA) with Scheffé's post hoc multiple comparison (32). For all statistical tests, α was set at 0.05. Levene's test was performed to investigate homogeneity of variance. All statistical analyses were performed using Statistical Package for the Social Sciences, version 16.0.
The binding of fluorescein isothiocyanate (FITC)-labeled P. gingivalis to KB cells was carried out as described by Pathirana et al. (40). Bound bacteria were detected using an FC500 flow cytometer (Beckman Coulter, Gladesville, NSW, Australia). KB cells were identified as a homogeneous population of large granular cells, and the bacteria adhering to these cells were identified by fluorescence detected through a 525-nm band-pass filter and compared with unlabeled controls.
All new microarray data have been deposited in the ArrayExpress databases under accession no. E-TABM-546.
Bioinformatic analysis of the predicted W83 strain FimR (36) using the Simple Modular Architecture Research Tool (SMART; http://smart.embl-heidelberg.de) (45) identified it as typical response regulator with a putative receiver domain between amino acid residues 2 and 117 and a helix-turn-helix Lux regulon-type DNA binding domain between amino acid residues 161 and 218 (Fig. (Fig.1).1). SMART analysis of the W83 strain FimS identified an N-terminal sequence typical of a long leader peptide (residues 1 to 40); however, this overlapped a second, shorter, predicted transmembrane domain (residues 21 to 40). With an alternative methionine start codon at residue 19, this second transmembrane domain may also serve as the leader peptide. This suggests that FimS may be secreted across the inner membrane. Another transmembrane domain was predicted between residues 375 and 397 followed by a histidine kinase A phosphoacceptor domain (residues 412 to 479) and a histidine kinase-like ATPase domain (residues 524 to 619). Therefore, the predicted structure of FimS indicates that the sensor domain is localized to the periplasm and linked to the cytoplasmic kinase domain via transmembrane residues 375 to 397 (Fig. (Fig.1).1). The predicted sensor domain has two tetratricopeptide repeats (TPR; residues 143 to 176 and 183 to 216) and a coiled-coil region (residues 338 to 365), motifs that are involved in protein-protein interactions and protein stabilization respectively.
We determined the nucleotide sequence of the fimR and fimS loci from the P. gingivalis W50 (afimbriated) and ATCC 33277 (fimbriated) strains. The W50 fimS locus was amplified by PCR using the oligonucleotide primers PG1433Seq-For and PG131Seq-Rev. However, these oligonucleotide primers did not amplify a product from ATCC 33277. This suggested that the sequences upstream of the W50 fimS and the ATCC 33277 fimS are divergent. To confirm this, we designed a forward primer oligonucleotide (fimS_Seq-For) based on the published ATCC 33277 fimS locus sequence (17), repeated the PCR, and sequenced the amplicon. The result confirmed the data of Hayashi et al. (17) and showed that the nucleotide sequences upstream of the fimS genes of the two strains are significantly different (Fig. (Fig.2A).2A). Indeed, the nucleotide sequence alignment indicates that these genes may use different start codons and the expressed proteins may have different N-terminal sequences, although both can have possible leader peptides consistent with the PG1432 SMART residue 21-to-40 transmembrane domain prediction. Importantly, the sequences of the promoters and putative regulatory elements upstream of these fimS genes must differ, which has implications for the transcriptional regulation of these loci in the different strains.
In addition to these differences, a single base transition was detected within the region encoding the histidine kinase domains of FimS, changing the coded amino acid from isoleucine to lysine in the W50 FimS (Fig. (Fig.2B).2B). This change is not located within the predicted phosphoacceptor or the histidine kinase-like ATPase domains and is unlikely to affect the histidine kinase function. More significant, however, was the presence of an additional adenosine base in the W50 fimS at 1,843 bp from the predicted PG1432 ATG start codon that was not present in the ATCC 33277 fimS (Fig. (Fig.2B).2B). This altered reading frame would result in a shorter FimS product in strain W50. The sequenced W50 fimR and fimS genes are identical to those of genes pg1431 and pg1432 of strain W83.
Typically, genes encoding the response regulator and the sensor histidine kinase of two-component signal transduction systems are contiguous and transcribed together in an operon. It has been presumed but not verified that fimS and fimR, which are separated by only 12 nucleotides in the ATCC 33277 genome and 65 nucleotides in the W50 genome (Fig. (Fig.2B)2B) are cotranscribed. To explore this possibility, the transcription of fimR and fimS in each strain was analyzed by RT-PCR. Using oligonucleotides specific for fimR (LuxR-For and LuxR-Rev) or fimS (SenHis-For and SenHis-Rev) gave RT-PCR products of 113 bp and 131 bp, respectively, as expected (Fig. (Fig.3).3). However, PCR with a primer pair designed to span the intergenic region between fimR and fimS (SenHis-For and PG1431R2) yielded no amplicons (Fig. (Fig.3,3, lanes 14 and 17), whereas PCR using the same oligonucleotides with genomic DNA as the template gave an amplicon of 954 bp, as expected (Fig. (Fig.3,3, lanes 16 and 19). Taken together, these results show that fimR and fimS are not cotranscribed. Furthermore, these data indicate that there must be a promoter specific for fimR expression between fimS and fimR. We identified a sequence motif, TAGGTTTG, that is similar to the highly conserved −7 sequence motif, TAnnTTTG, found as part of the consensus P. gingivalis and other Bacteriodetes promoter sequences (2, 19, 30), and there is also an alternative fimR start codon 18 bp downstream of this sequence (Fig. (Fig.2B2B).
To explore the significance of the FimR and FimS in P. gingivalis biofilm formation, fimR and fimS of strains ATCC 33277 and W50 were each disrupted by insertion of an ermF cassette. The W50 fimR mutant was designated ECR220, the ATCC 33277 fimR mutant was designated ECR221, the W50 fimS mutant was designated ECR222, and the ATCC 33277 fimS mutant was designated ECR223.
The ability of each of the fimR and fimS mutants to form biofilms was initially examined using a rapid 24-h, 96-well microtiter plate static biofilm assay. ATCC 33277 formed measurable CV-stained biofilms in this assay; however, P. gingivalis strain W50 did not, even though this strain is known to produce biofilms in continuous culture (1, 27). In an attempt to induce W50 to form biofilms in this system, the microtiter wells were coated with filter-sterilized human saliva (5 to 10 μg of protein) or fibronectin from human plasma (5 to 10 μg), but these strategies were unsuccessful. The wells were also coated with 5 × 107 formalin-killed S. gordonii or F. nucleatum cells prior to seeding with P. gingivalis and overnight incubation. However, incubation with P. gingivalis W50 caused the S. gordonii and F. nucleatum cells to release from the wells. As we were unable to find conditions amenable to P. gingivalis W50 biofilm formation in this system, we confined the rapid biofilm studies to ATCC 33277 and the ATCC 33277 mutants ECR221 and ECR223.
The fimR and fimS mutants ECR221 and ECR223 both formed significantly less biofilm (P < 0.001, t test) than the ATCC 33277 parent strain (Fig. (Fig.4),4), with ECR221 and ECR223 having, respectively, 81% and 60% less CV-staining biofilm biomass than ATCC 33277. Furthermore, the biofilm biomass formed by ECR221 was significantly smaller than that of the fimS mutant ECR223 (P < 0.001). To investigate the possibility that the reduced biofilm formation by ECR221 and ECR223 could be attributed to reduced growth rate, the growth kinetics of the mutants and the wild-type strains were compared. The mean generation times for the mutants (ECR223, 5.0 ± 0.4 h; ECR221, 4.4 ± 0.2 h) were not significantly different from that of the wild type (ATCC 33277, 4.6 ± 0.5 h). Furthermore, at 24 h, the time point at which biofilm formation was assessed, ECR221 and ECR223 achieved optical densities (0.98 ± 0.03 and 0.90 ± 0.01, respectively) that were comparable to that of the ATCC 33277 wild type (0.96 ± 0.02). Therefore, the reduced biofilm formation by strains ECR221 and ECR223 could not be attributed to reduced growth rates.
Confocal laser scanning microscopy was used to analyze the architecture of the biofilms formed by ATCC 33277, ECR221, and ECR223. Strains were grown in a 16-well culture chamber coverglass system and stained using BacLight live-dead stain. Both ECR221 and ECR223 formed biofilms that were visibly more sparse than the biofilm formed by the parent ATCC 33277, with the ECR221 biofilm being composed of smaller, more dispersed aggregates in comparison to ECR223 (Fig. (Fig.5).5). Quantitative analysis using COMSTAT (Table (Table3)3) showed that the calculated biomasses of ECR221 and ECR223 biofilms were significantly reduced (P < 0.01) to 7% and 25%, respectively, of that produced by the wild-type strain ATCC 33277. The mean thickness of the biofilm produced by each mutant (0.3 ± 0.12 μm for ECR221 and 1.33 ± 0.30 μm for ECR223) was also significantly reduced relative to ATCC 33277, which was 4.52 ± 1.31 μm thick (P < 0.01). However, both mutants displayed similar maximum thickness to that of the wild type (Table (Table3),3), suggesting that microcolony tower formation was not limited, but rather there was a less effective establishment of intercolony contacts in these biofilms. This supposition was supported by the roughness coefficients (R) of 1.8 and 1.2 for ECR221 and ECR223 biofilms, respectively, indicating the formation of rough and heterogenous biofilms that are often associated with many pillars and towers of cells separated by areas devoid of cells (18). The roughness coefficients of the biofilms produced by each of the mutants were significantly different (P < 0.01), indicating that the mutants produced biofilms with dissimilar morphologies. In contrast, the wild-type biofilm had a very low R (0.08), indicating that this strain forms a homogenous and uniform biofilm (Table (Table33).
It has been shown that production of the fimbrillins Mfa1 and FimA is essential for biofilm formation by P. gingivalis ATCC 33277 (26). Furthermore, it has been shown that FimR binds directly to the mfa1 promoter region (54) and to the promoter of the fimbriation-associated gene fimX that is upstream of fimA (37) (annotated as pg2130 in the W83 genome  and pgn_0178 in the recently released ATCC 33277 genome sequence ). We used qRT-PCR to quantify mfa1, pgn_0178, and fimA transcripts in the fimS and fimR mutants and ATCC 33277, normalizing the transcript levels relative to that of the control gene galE (27).
In comparison to wild-type cells, the expression of mfa1, fimA, and pgn_0178 in each of the mutants was decreased, but to different extents for each gene. The level of fimA and pgn_0178 expression was significantly reduced, by at least 28-fold, in the fimS and fimR mutants (P < 0.05), from levels up to 35-fold greater than galE expression in the wild type to levels below galE expression in the mutants (Fig. (Fig.6).6). Notably, in the wild-type strain ATCC 33277 the level of FimA-encoding transcript was 9-fold higher than that of the pgn_0178 transcript, indicating that transcription of more fimA mRNA is initiated from the fimA proximal promoter than from the promoter proximal to pgn_0178. In the fimS and fimR mutants, although the abundances of the pgn_0178 and fimA mRNAs were both very low, this differential transcript abundance between fimA and pgn_0178 continued, with the ratio being 19-fold in the fimR mutant and 13-fold in the fimS mutant. However, there was no statistically significant difference between the level of either fimA or pgn_0178 expression in each of the mutants.
The decrease in expression of the minor fimbrillin-encoding gene mfa1 in the fimS and fimR mutants was less profound, although statistically significant. Interestingly, the expression of mfa1 was found to be influenced by growth phase. In ATCC 33277, the level of mfa1 expression in exponential-phase cells (OD650 of 0.5 to 0.6) was significantly higher (1.9-fold; P < 0.01) than that measured in stationary-phase cells (OD650 of 1.4). Similarly, in the fimR mutant there was a 1.6-fold difference in mfa1 expression (P < 0.05) between exponential- and stationary-phase cells (Fig. (Fig.6).6). In contrast, in the fimS mutant there was no significant change in the level of mfa1 transcript between the exponential and stationary phases. In the exponential growth phase, the fimS mutant expressed decreased levels of mfa1 relative to wild-type cells (2.7-fold; P < 0.001), a difference that decreased to 1.5-fold in the stationary phase. In the fimR mutant, the decrease in mfa1 expression was not statistically significant. There was no significant difference between the levels of mfa1 expression in any of the strains during the stationary phase of growth.
Overall, the data clearly show that FimR and FimS each have a function in the control of expression of genes important for P. gingivalis fimbriation and that each affects the expression of these genes to differing extents. Reduction in fimbriation due to decreased fimbriation-associated gene expression may in part explain the reduced capacity of the ATCC 33277 FimR and FimS mutants to form biofilms in vitro.
The P. gingivalis W50 wild-type strain and fimR and fimS mutants were grown to mid-exponential phase, harvested, labeled with FITC, and incubated with KB cell monolayers. P. gingivalis strain ATCC 33277 was not used for the KB cell binding assays due to the propensity of the strain to self-aggregate (40). The percentage of KB cells with bound P. gingivalis was determined using flow cytometry. There was no difference in the numbers of KB cells with bound P. gingivalis W50 wild-type or fimR/fimS cells, and mean fluorescence intensities were also equivalent (data not shown), indicating that equal numbers of mutant and wild-type P. gingivalis cells were bound per KB cell (data not shown). Therefore, this suggests that the binding of P. gingivalis W50 to KB cells was not dependent on expression of fimR or fimS.
The involvement of FimR and FimS in the P. gingivalis response to various physiological stresses such as H2O2, elevated temperature, hemin limitation, and altered pH was investigated. The fimR and fimS mutants, in the presence of 1 mM H2O2, showed no difference in growth kinetics from their respective wild-type strains (ATCC 3327 and W50), nor was any difference observed under temperature stress (42°C), hemin limitation (0.1 mg ml−1), or growth at altered pH (data not shown). Thus, under the test conditions used, neither FimR nor FimS had any significant role in the responses of P. gingivalis W50 or ATCC 33277 to H2O2, hemin limitation, elevated temperature, or altered pH.
To gain further insight into the role of FimS in P. gingivalis gene expression, we used DNA microarrays to compare the transcriptome of the wild-type strain, W50, with that of the fimS mutant ECR222. As the W50 fimS mutant would not grow as a biofilm, we used wild-type and fimS mutant (ECR222) cells grown in planktonic culture for the microarray analysis.
Relative to strain W50, the disruption of fimS resulted in the altered expression (>1.5-fold up- or downregulated; P < 0.05) of 199 genes in ECR222 cells, of which 110 genes showed increased expression and 89 genes exhibited decreased expression relative to the parent strain (see Tables S1 and S2 in the supplemental material). This represents 10% of the P. gingivalis genome. The level of fimR expression was reduced by 2.3-fold in the fimS mutant. The signal intensity of the hybridized fimS probe was equal to the array background signal intensity, confirming that there was no expression of fimS in ECR222. These data indicate that fimR expression is influenced by, but not completely codependent on, fimS transcription and support the RT-PCR analysis that these genes are not cotranscribed.
In agreement with the qRT-PCR findings for the ATCC 33277 fimS mutant ECR223, there was reduced expression of the fimX (pgn_0178) homologue pg2130 (13-fold) and fimA orthologue (pg2132) (6.5-fold). It is interesting to note that the fimA orthologue is expressed in W50 and the protein monomer detected by proteomic analysis (9), but the fimbriae are not assembled. Also downregulated were other fimbriation-associated genes (pg2131 and 2133-pg2136) (Table (Table4).4). In the genome sequence of strain W83 (36), mfa1 is interrupted by insertion element ISPg4. We used PCR and DNA sequencing and showed that mfa1 is also interrupted by ISPg4 in strain W50 (data not shown). The transcript that would be the truncated mfa1 mRNA (pg0176) was downregulated in the array 3-fold, in agreement with the reduced mfa1 expression measured by qRT-PCR in the ATCC 33277 fimS mutant. The disruption of fimS also resulted in the upregulation of most of the genes of a 6-gene cluster, PG0508 to PG0513, that also has reduced expression in P. gingivalis mature biofilms (27), suggesting that FimS negatively affects the expression of these genes in both biofilm and planktonic cells through an as-yet-unidentified repressor. Also potentially derepressed was PG0718, encoding a conserved hypothetical protein (16-fold upregulation), and PG0862, encoding a putative restriction endonuclease (12-fold increase in expression).
When the differentially regulated genes were grouped into TIGR (www.tigr.org) role categories (Fig. (Fig.7),7), more than half of the differentially expressed genes observed in ECR222 (55 genes upregulated and 48 genes downregulated) encoded hypothetical proteins, conserved hypothetical proteins, or proteins with similarity to proteins with uncharacterized functions. Of the other functional categories, the expression of genes encoding proteins predicted to be involved in the binding and transport of substrates was most affected in the fimS mutant (23 genes were up- and downregulated, respectively, in ECR222) (Table (Table4).4). Genes involved in amino acid biosynthesis and fatty acid and phospholipid metabolism were apparently unaffected by the disruption of fimS.
During the establishment of a biofilm and during biofilm growth, cells must remain responsive to environmental changes and be able to maintain an appropriate balance between the planktonic and biofilm growth phases, a process that requires the interplay of various cell factors. As a late colonizer of subgingival plaque, P. gingivalis persistence is dependent on its ability to stably attach to a variety of surfaces, including other bacteria and the subgingival plaque extracellular matrix, as well as possibly epithelial cells of the subgingival crevice. Given that each of these surfaces will have distinct properties, it is likely that P. gingivalis must use specific extracellular protein(s) to facilitate these diverse interactions. Thus, P. gingivalis would need to regulate the production of an array of extracellular proteins to adapt to the various surfaces it encounters. This may be reflected in the activation of fimS, fimR, and genes associated with the biogenesis of a cell envelope component such as the FimA and Mfa1-type fimbriae when P. gingivalis is grown as biofilm (27).
Two-component signal transduction systems have been identified and characterized in various bacterial species (49). Roles for some of these TCSTS in biofilm development, as well as other phenotypic traits, have been demonstrated in both Gram-negative and Gram-positive bacteria, and the regulatory systems often display remarkable complexity. For example, biofilm formation by E. coli involves production of curli (11). Curli production is influenced by the CpxA-CpxR sensor kinase-response regulator system (11) but also by CsgD, EnvZ/OmpR, Rcs, and H-NS (20). Furthermore, the regulator CsgD influences cell aggregation and cellulose production and affects the expression of at least 24 genes, including other regulators (3).
Species that have a restricted ecological niche, such as intracellular parasites, have few sensor proteins encoded within their genomes, whereas in contrast, species such as Mesorhizobium loti, P. aeruginosa, and Vibrio cholerae, which occupy more variable habitats have more than 100 sensor proteins (13). The detection of only 6 putative sensor histidine kinases in the P. gingivalis W83 genome suggests that it may be programmed to sense few environmental factors. The independent transcription of fimS and fimR may provide added flexibility to the response regulation cascades in P. gingivalis with FimS and/or FimR able to interact with alternative sensor histidine kinases and response regulators in a manner that would not be possible if they were transcriptionally linked. The concept that FimS interacts with response regulators other than FimR is supported by the microarray data that showed change in the expression of a large number of genes following the disruption of fimS in ECR222 (10% of the genome). In comparison, transcriptomic analysis of a fimR mutant revealed a relatively small regulon with only 7 genes identified as differentially expressed (37). Our microarray data revealed that the expression of four predicted transcriptional regulators (PG0173, PG0543, PG1044, and PG2125) was downregulated in ECR222, while two other regulators (PG0928 and PG1181) were upregulated. In addition, three putative RNA polymerase extracytoplasmic sigma-70 factors (PG0214, PG0985, and PG1827) were upregulated in ECR222. These data strongly suggest that FimS is part of a complex cascade of regulatory effectors. Cross-talk and cross-regulation phenomena, albeit a new concept with P. gingivalis, have been documented in other species, including E. coli and Pseudomonas spp. (42, 52). Given the limited number of TCSTSs in the P. gingivalis genome, the broad specificity of FimS may increase the adaptability of the response of the organism to environmental change.
Expression of fimS also had a negative influence on the expression of genes with very diverse cellular functions. Thus, we propose a role for FimS as the major physiological “switch” as part of P. gingivalis biofilm development, with a major effect being reduction in fimbriation. The importance of FimS in P. gingivalis biofilm formation coupled with the fact that fimS is highly upregulated during P. gingivalis mature biofilm growth (27) suggests that fimS may be activated during both of these distinct phases of P. gingivalis biofilm development.
The disruption of either fimS or fimR of ATCC 33277 resulted in significant impairment (but not abolition) of biofilm formation by P. gingivalis in 96-well microtiter plate and glass culture chamber systems. Furthermore, a W50 fimS mutant also formed only sparse, poorly attached biofilm in a fermentor vessel. CLSM analysis of the biofilms formed by the ATCC 33277 fimR and fimS mutants showed that they produced biofilms that were distinct from the wild-type strain and also from each other, with rough and heterogeneous biofilms suggesting an apparent reduction of microcolony contacts. In vitro analysis of ATCC 33277 fimA and mfa1 mutants has shown FimA to be involved in adhesion of P. gingivalis to saliva-coated glass surfaces (26) and cultured KB epithelial cells (51), while mfa1 expression mediates cell aggregation (26). Neither the fimA nor mfa1 mutants formed a confluent biofilm (26). In view of this, reduced expression of fimA and mfa1 in the ATCC 33277 fimS and fimR mutants, with resulting lowered FimA and Mfa1 production, would explain the biofilm architecture we observed using CLSM and would also explain the lowered biofilm production by these mutants. However, this is not sufficient to explain the impairment of biofilm formation by the afimbriated strain W50, indicating that there are factors other than fimbriation that are involved in the initiation of P. gingivalis biofilm formation. As the distinct physiological steps of P. gingivalis biofilm development have yet to be fully elucidated, it is probable that in concert with fimbriae, the products of the extensive number of hypothetical genes shown to have altered expression in ECR222 have a role in P. gingivalis biofilm formation.
P. gingivalis W50 has previously been shown to bind to KB cells (40) and fibroblasts (41), with the RgpA-Kgp cysteine proteinase-adhesin complexes produced by P. gingivalis having some role in this adherence. Using isogenic mutants, it has been shown that production of the Kgp proteinase had the most significant role in this effect (40, 41). We could measure no difference in the adherence of strain W50 and the ECR222 fimS mutant to KB cells, and the microarray data indicated that there was no change in the expression of kgp or rgpA, which encode the components of the RgpA-Kgp complexes. Measurement of Arg- and Lys-specific whole-cell proteinase activity also showed no difference between the strains (data not shown). Together, these data indicate that strain W50 KB epithelial cell adhesion is mediated by factors other than those associated with FimS function and that FimS does not function in the regulation of the expression of rgpA or kgp. Interestingly, other than a slight upregulation of hagA (2.1-fold), none of the genes that encode the numerous surface-associated CTD family proteins to which RgpA and Kgp belong (46), many of which are adhesins, were differentially expressed in ECR222, indicating that FimS does not function to regulate expression of genes encoding these surface proteins.
The qRT-PCR data showed that, during exponential growth, mfa1 expression was decreased more in the fimS mutant than in the fimR mutant, suggesting that there may be at least one other activator of mfa1 expression which is dependent on phosphorylation by FimS. Possible candidate transcription factors revealed by the microarray data are PG0543, MntR (PG1044), PG2125, and PG0173, as all of the genes encoding these proteins were downregulated in ECR222 (Table (Table44).
The presence of an N-terminal sequence typical of a signal peptide followed by two transmembrane helices indicates that the sensor domain of FimS with the TPR is likely to be localized to the periplasm of the cell (Fig. (Fig.1).1). TPR domains are found to be widely distributed from prokaryotes to eukaryotes and are known to act as molecular scaffolds in mediating specific protein-protein interaction (15), while coiled-coil structures are known to facilitate and stabilize protein-protein interactions (28). Given this and the result that FimS was highly induced in a P. gingivalis W50 biofilm (27), we propose that FimS may sense a signal that is important in the development of P. gingivalis biofilm. We observed no altered ability of the FimS mutants to respond to a range of stresses that P. gingivalis would experience during biofilm growth in the subgingival niche, including H2O2 (which may result from neutrophil attack), temperature changes (as occurs during inflammation), hemin limitation, and altered pH.
To our knowledge, this is the first study to directly show the important involvement of both FimS and FimR in P. gingivalis biofilm development. We have demonstrated that fimR and fimS are not part of an operon and also that fimS and fimR mutants form altered biofilm phenotypes compared to both wild-type strains and each other. In contrast to the situation observed in an fimR mutant, where altered regulation of only a limited number of genes was observed, the disruption of fimS resulted in the altered expression of a large number of genes encoding products with very diverse cellular functions. We hypothesize that FimS may be important in monitoring the specific environmental signal that is required for or associated with P. gingivalis fimbriation and growth as a biofilm and forms part of a complex regulatory network regulating biofilm formation and development.
This work was funded by National Health and Medical Research grant no. 509326 and Australian Dental Research Foundation grant 44/2006.
Published ahead of print on 8 January 2010.