Transcriptional Regulation of Aggregatibacter actinomycetemcomitans lsrACDBFG and lsrRK Operons and Their Role in Biofilm Formation

Abstract

Autoinducer-2 (AI-2) is required for biofilm formation and virulence of the oral pathogen Aggregatibacter actinomycetemcomitans, and we previously showed that lsrB codes for a receptor for AI-2. The lsrB gene is expressed as part of the lsrACDBFG operon, which is divergently transcribed from an adjacent lsrRK operon. In Escherichia coli, lsrRK encodes a repressor and AI-2 kinase that function to regulate lsrACDBFG. To determine if lsrRK controls lsrACDBFG expression and influences biofilm growth of A. actinomycetemcomitans, we first defined the promoters for each operon. Transcriptional reporter plasmids containing the 255-bp lsrACDBFG-lsrRK intergenic region (IGR) fused to lacZ showed that essential elements of lsrR promoter reside 89 to 255 bp upstream from the lsrR start codon. Two inverted repeat sequences that represent potential binding sites for LsrR and two sequences resembling the consensus cyclic AMP receptor protein (CRP) binding site were identified in this region. Using electrophoretic mobility shift assay (EMSA), purified LsrR and CRP proteins were shown to bind probes containing these sequences. Surprisingly, the 255-bp IGR did not contain the lsrA promoter. Instead, a fragment encompassing nucleotides +1 to +159 of lsrA together with the 255-bp IGR was required to promote lsrA transcription. This suggests that a region within the lsrA coding sequence influences transcription, or alternatively that the start codon of A. actinomycetemcomitans lsrA has been incorrectly annotated. Transformation of ΔlsrR, ΔlsrK, ΔlsrRK, and Δcrp deletion mutants with lacZ reporters containing the lsrA or lsrR promoter showed that LsrR negatively regulates and CRP positively regulates both lsrACDBFG and lsrRK. However, in contrast to what occurs in E. coli, deletion of lsrK had no effect on the transcriptional activity of the lsrA or lsrR promoters, suggesting that another kinase may be capable of phosphorylating AI-2 in A. actinomycetemcomitans. Finally, biofilm formation of the ΔlsrR, ΔlsrRK, and Δcrp mutants was significantly reduced relative to that of the wild type, indicating that proper regulation of the lsr locus is required for optimal biofilm growth by A. actinomycetemcomitans.

INTRODUCTION

The dental biofilm is a complex and dynamic microbial community that is comprised of up to 700 different species of bacteria (1–6). This biofilm is the prime etiological agent of three common oral diseases in humans: dental caries, gingivitis, and periodontal disease (7–9). The progression of these diseases is associated with major shifts in microbial populations in the oral biofilm, and diseased sites often exhibit increased populations of pathogenic species relative to healthy sites in the oral cavity (7, 8, 10). The stimuli that contribute to these populational shifts have not been well characterized, but the oral cavity is subject to continual environmental flux, including changes in pH, temperature, osmolarity, and nutrient supply. Oral bacteria rapidly detect and respond to these environmental fluctuations, allowing them to successfully coexist and thrive in the oral cavity (8, 11, 12). Both intra- and interspecies communication is known to occur among oral bacteria, and it is likely that these signaling processes enable the organisms to coordinate their behavior and function by regulating gene expression as a community. One mechanism of communication, termed quorum sensing, is a cell density-dependent response (3, 13–15), which, in Gram-negative bacteria, is mediated by the production, release, and detection of soluble signal molecules called autoinducers.

Aggregatibacter actinomycetemcomitans is a Gram-negative organism that is associated with aggressive forms of periodontitis and other systemic infections (16–19). This organism expresses LuxS and secretes autoinducer-2 (AI-2), and AI-2 dependent quorum sensing has been shown to regulate the expression of virulence factors, iron acquisition systems, and biofilm formation (20–23). A. actinomycetemcomitans expresses two periplasmic proteins, LsrB and RbsB, that function as receptors for AI-2 (24, 25), and inactivation of either or both of the genes encoding these proteins has been shown to result in reduced biofilm growth and virulence (21, 23). The importation of AI-2 by LsrB and the Lsr transporter has also been shown to be required for the induction of the QseBC two-component system, which in turn influences biofilm growth and virulence of A. actinomycetemcomitans (21). However, the link between the lsr operons and the expression of qseBC and downstream regulation of gene expression remains to be determined.

In Escherichia coli and Salmonella enterica serovar Typhimurium, LsrB is encoded by an operon consisting of lsrACDBFG (the S. Typhimurium operon also contains an additional gene, designated lsrE), where lsrACD encode the AI-2 transporter, lsrF encodes an aldolase-like protein that cleaves AI-2 (26), and lsrG codes for an isomerase of phospho-AI-2 (27). Upstream and divergently transcribed from the lsrACDBFG operon resides lsrRK, encoding a repressor of lsrACDBFG (lsrR) and an AI-2 kinase (lsrK), which regulate the expression of the lsrACDBFG operon in an AI-2-dependent manner (28, 29). In E. coli and S. Typhimurium, AI-2 is internalized and phosphorylated at high cell density by lsrK, and AI-2-PO₄ binds to LsrR, resulting in derepression of lsrACDBFG. The structure of both operons is conserved in A. actinomycetemcomitans, suggesting that A. actinomycetemcomitans may also import AI-2 and regulate the lsr operon in an AI-2-dependent manner.

In this report, we show that the intergenic region (IGR) separating the A. actinomycetemcomitans lsrACDBFG and lsrRK operons contains the promoters that drive the expression for both operons and that each is negatively regulated by LsrR and positively regulated by the consensus cyclic AMP (cAMP)-cyclic AMP receptor protein (CRP) complex. The IGR contains two inverted repeat sequences that are bound by LsrR and two regions that resemble the consensus CRP binding site that interact with the purified cAMP-CRP complex. However, in contrast to what occurs in E. coli, lsrK did not play an essential role in regulating lsrACDBFG in A. actinomycetemcomitans since its deletion did not affect transcription of lsrACDBFG at high cell density relative to that of the wild type (WT). These results suggest that A. actinomycetemcomitans may express another kinase that is capable of phosphorylating AI-2 in the absence of LsrK or alternatively that phosphorylation of AI-2 may not be required for expression of lsrACDBFG. Finally, deletion of lsrR or crp reduced biofilm growth of A. actinomycetemcomitans, suggesting that proper regulation of the lsr locus is required for optimal formation of A. actinomycetemcomitans biofilms.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The bacterial strains and plasmids used in this study are listed in Table S1 in the supplemental material. Luria-Bertani (LB) broth, LB agar (LB broth plus 1.5% agar), brain heart infusion (BHI) broth, and BHI agar (all from Difco) were routinely used for the propagation and plating of bacteria. Aggregatibacter actinomycetemcomitans (afrimbriated, smooth-colony-morphotype strain 652) was grown at 37°C under microaerophilic conditions and was also used for biofilm experiments (see below). We previously showed that biofilm formation by both fimbriated and afimbriated A. actinomycetemcomitans strains is regulated by AI-2 (30). When required, the medium was supplemented with 25 μg/ml kanamycin (Km), 12.5 μg/ml tetracycline (Tc), or 100 μg/ml ampicillin (Amp). LB broth and TYE broth (1% tryptone, 0.5% yeast extract [Difco]) supplemented with 0.2, 0.4, and 0.6% glucose were used to standardize the minimal concentration of glucose required for regulation of the lsr operon.

DNA procedures.

DNA manipulations were carried out as described previously (31). Transformation of E. coli and A. actinomycetemcomitans was done by electroporation (Bio-Rad). Transformants containing plasmids were selected on LB agar plates supplemented with the appropriate antibiotics. Plasmid DNA was isolated using the QIAprep spin miniprep kit (Qiagen). Restriction enzymes were used as recommended by the manufacturer (New England BioLabs). All primers used in this study (Integrated DNA Technology) were flanked with restriction enzyme recognition sites (underlined in the primer sequences) and are shown in Table S2 in the supplemental material. Primer sequences were designed based on the genome information of the A. actinomycetemcomitans D11S-1 strain available from the Pathosystems Resource Integration Center (http://patricbrc.vbi.vt.edu). All constructs were verified by DNA sequencing (University of Louisville Core Sequencing Facilities).

Construction of lsrR, lsrA, and lsrK promoter/lacZ fusion plasmids.

Various fragments of the intergenic region between lsrACDBFG and lsrRK and surrounding coding regions were amplified by PCR as follows using A. actinomycetemcomitans genomic DNA as a template. The typical amplification profile used was 94°C for 2 min for 1 cycle and then 94°C for 30 s, 60°C for 1 min, and 72°C for 2 min for 25 cycles. For the lsrR promoter fusion constructs (pATE23, pATE13, and pATE68), the entire intergenic region was amplified using the primer set lsrR255-f3 and lsrR255-r6 (see Table S2 in the supplemental material). The 255-bp PCR product was then digested with KpnI-BamHI and cloned into KpnI-BamH-digested pJT3 (M. D. Juarez-Rodriguez, A. Torres-Escobar, and D. R. Demuth, submitted for publication) to create pATE23. This same approach was followed to create pATE11 and pATE68 plasmids, using the primer sets lsrR894-f1/lsrR894-r2 and lsrR255-f3/lsrR173-r174, respectively. A similar approach was used to construct the lsrA promoter fusion plasmids pATE21, pATE26, pATE33, pATE34, and pATE71. Primer sets used to amplify the appropriate promoter fragment for these constructs were lsrA255-f4/lsrA551-r43, lsrA255-f4/lsrA255-r7, lsrA551-f42/lsrA551-r43, lsrA551-f42/lsrA1600-r47, and lsrA156-f175/lsrA156-r167, respectively. For the lsrR promoter fusion construction, pATE13, an 88-bp DNA fragment containing the promoter region of the lsrR gene flanked with sticky KpnI-BamHI ends, was obtained by annealing two complementary single-stranded 100- and 92-bp oligonucleotides, lsrR92-f13 and lsrR100-r14. The 88-bp synthetic DNA was cloned into KpnI-BamHI-digested pJT3 to create pATE13. This same approach was followed to create the lsrA promoter fusion construction pATE14, using the primer sets lsrA92-f5 and lsrA100-r16.

To construct the lsrK promoter fusion plasmid pATE35, the intergenic region between lsrR and lsrK was amplified using the primer set lsrRK597-f51 and lsrRK597-r52. The 597-bp product was subsequently digested with KpnI-BamHI and cloned into KpnI-BamHI-digested pJT3 to create pATE35.

Construction of lsrR and crp expression plasmids.

The structural lsrR and crp genes were PCR amplified from A. actinomycetemcomitans genomic DNA using primer sets lsrR-f33/lsrA-r29 and crp-F113/crp-r114, respectively. Each PCR product, 951 bp containing the lsrR gene and 609 bp containing the crp gene, was digested with NcoI-ApaI and cloned individually into NcoI-ApaI-digested pYA3883 (32) expression vector to create pATE28 and pATE51, respectively.

Growth kinetics.

A single colony of A. actinomycetemcomitans harboring each recombinant plasmid was independently inoculated into 10 ml of BHI medium (20) supplemented with 25 μg/ml Km and was grown standing for 24 h at 37°C. The next day, the overnight culture (optical density at 600 nm [OD₆₀₀] of 0.6) was diluted at a 1:30 ratio to inoculate 20 ml of BHI (20 ml in a 50-ml conical centrifuge tube) with 25 μg/ml Km and grown standing at 37°C. For the first 12 h of growth, an aliquot of 1.1 ml (1 ml to read the OD₆₀₀ and 0.1 ml for the β-galactosidase [β-Gal] activity assay) was removed each hour. Additional aliquots were taken from each culture for analysis at the 24-h, 48-h, and 72-h time points. β-Galactosidase activity was also determined as described below.

Determination of the glucose concentration required for repression of A. actinomycetemcomitans lsr operon expression.

Plasmids pATE23 and pATE33, each containing the entire lsrACDBFG-lsrRK intergenic region, were selected to quantify the expression of both the lsrACDBFG and lsrRK operons in A. actinomycetemcomitans. A single colony of A. actinomycetemcomitans harboring each of these plasmids was independently inoculated into 10 ml of BHI medium containing 25 μg/ml Km and was grown standing for 24 h at 37°C. The next day, the overnight culture was diluted at a 1:30 ratio to inoculate 20 ml of LB (20 ml in a 50-ml conical centrifuge tube) supplemented with 25 μg/ml Km and without or with 0.2, 0.4, or 0.6% glucose in independent tubes and grown standing at 37°C. As described above, an aliquot of each culture was removed hourly between 1 and 12 h of growth and at the 24-h, 48-h, and 72-h time points for determination of culture density and β-galactosidase activity.

β-Galactosidase assays.

β-Gal activity was qualitatively assessed on LB agar plates that were supplemented with 50 μg/ml 5-bromo-4-chloro-3-indolyl-beta-d-galactopyranoside (X-Gal). Quantitative evaluation of β-Gal activity was carried out using permeabilized cells obtained from mid-exponential and early-stationary-phase cultures (OD₆₀₀ 0.3 and 0.5, respectively) incubated with o-nitrophenyl-β-d-galactopyranoside (ONPG) substrate (Sigma) as previously described by Miller (33). Average values (±the standard deviations) for activity units were routinely calculated from three independent assays.

Construction of A. actinomycetemcomitans markerless deletion mutants.

A. actinomycetemcomitans 652 strain deletion mutants (Table S1 in the supplemental material) used in this study were constructed by allelic replacement of the target gene by double homologous recombination using the suicide vector pJT1 (Juarez-Rodriguez et al., submitted). Briefly, the flanking regions of the lsrR were amplified by PCR using A. actinomycetemcomitans chromosomal DNA as a template with the primer sets ATE-86f/ATE-87r and ATE90f/ATE-91r. Each primer was flanked with an appropriate restriction site (see Table S2 in the supplemental material). The 2,814-bp and 2,356-bp products were digested with NotI-XhoI and XhoI-PstI, respectively, and both fragments were cloned adjacent to each other (joined by the XhoI restriction site) into NotI-PstI-digested pJT1 to create pATE48 (see Table S1 in the supplemental material). A similar approach was used to generate suicide vectors pATE49, pATE47, and pATE52 to construct A. actinomycetemcomitans markerless deletion mutants for lsrK, lsrRK (double deletion), and crp, respectively. Primer sets used to amplify the upstream and downstream flanking regions of each target gene were ATE-92f/ATE-93r and ATE-88f /ATE-89r for lsrK and ATE-125f/ATE126r and ATE-127f/ATE-128r for crp. The recombinant suicide plasmids were introduced into A. actinomycetemcomitans by electroporation. Recombinant cells with a single recombinant crossover event were selected on BHI agar supplemented with 50 μg/ml spectinomycin (Sp). Spectinomycin-resistant (Sp^r) colonies were picked and grown in BHI broth standing for 24 h at 37°C in microaerophilic conditions. The next day, the bacterial culture was diluted 1:200 in TYE broth and grown under conditions similar to those described above. This step was repeated for three consecutive days, except that the final cultures on day 3 were grown in the presence of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside). Afterward, to select bacteria with a second event of recombination, and the replacement of the target gene, the bacterial cells were 10-fold diluted, spread onto TYE agar supplemented with 1 mM IPTG and 10% sucrose, and grown at 37°C in microaerophilic conditions. Sucrose-resistant (Suc^r) colonies were replica plated onto TYE agar supplemented with sucrose and onto BHI agar supplemented with spectinomycin. Spectinomycin-sensitive (Sp^s) colonies were selected to perform PCR for the deletion mutation of the target gene. Sp^s colonies that were PCR positive for the appropriate gene deletion were selected for further analysis. Planktonic growth rates of all of the deletion strains were similar to that of the wild type (not shown).

Expression and purification of LsrR- and CRP-hexahistidine fusion proteins.

E. coli LMG194 harboring the pATE28 or pATE51 plasmid was used for the synthesis of the hexahistidine fusion proteins. The expression and detection of the recombinant proteins was performed essentially as described previously by Torres-Escobar et al. (32). Purification was carried out by cobalt-based immobilized metal affinity chromatography under denaturing conditions. Eluted fractions containing the purified LsrR or CRP protein were selected based on sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis (SDS-PAGE). The selected fractions were then pooled and dialyzed in a Slyde-A-Lyzer 10K (for LsrR) or 3K (for CRP) cassette (Pierce) at 4°C against refolding buffer containing 50 mM Tris-HCl (pH 8.5), 4 M urea, 0.5 M l-arginine, 264 mM NaCl, 11 mM KCl, 8 mM MgCl₂, 0.1% Triton X-100, and 50% (vol/vol) glycerol for 24 h. Subsequently, samples were sequentially dialyzed for 6 h each in buffer consisting of 50 mM Tris-HCl (pH 8.5), 100 mM NaCl, 0.1 mM KCl, 8 mM MgCl₂, 2 mM Mg(CH₃COO)₂, 0.1 mM EDTA, 0.1 mM dithiothreitol (DTT), 50% (vol/vol) glycerol, and 2 M, 1 M, or 0.5 M urea. A final dialysis was then carried out for 6 h against the buffer described above without urea. For LsrR, the glycerol concentration in the final dialysis buffer was also reduced to 40% (vol/vol), and for CRP, NaCl was reduced to 75 mM and glycerol to 30% (vol/vol). Aliquots of the purified proteins were stored at −70°C. Protein concentration was determined by the Bradford assay, using bovine serum albumin as a standard.

EMSA.

The DNA fragments used for nonradioactive electrophoretic mobility shift assay (EMSA) were obtained by PCR using sets of primers described in Table S2 in the supplemental material. A biotin 3′-end-labeling kit (Thermo Scientific) was used for labeling of DNA fragments according to the manufacturer's instructions. Binding reactions were performed with a total of 50 fmol of each probe mixed with various amounts of purified LsrR or CRP (1, 2, and 4 μM) in 20 μl of binding buffer: for LsrR, 10 mM Tris-HCl (pH 7.5), 50 mM KCl, 1 mM dithiothreitol, and 50 ng of poly(dI-dC) used as nonspecific protein blocking; for CRP, 20 μl of the same binding buffer but also 100 μM cAMP. Reaction mixtures were incubated for 20 min at room temperature. Afterward, 5 μl of gel loading buffer (0.25× Tris-borate-EDTA [TBE], 60%; glycerol, 40%; bromphenol, 0.2% [wt/vol]) was added, and mixtures were electrophoresed in a 6% native polyacrylamide gel in 0.5× TBE buffer (45 mM Tris-borate, 1 mM EDTA, pH 8.0) (NuPAGE 4 to 12% Bis-Tris; Invitrogen) and immunoblotted. DNA bands were detected using the LightShift chemiluminescent EMSA kit according to the manufacturer's instructions.

Biofilm formation and analysis.

A. actinomycetemcomitans biofilms were grown on a saliva-coated cover glass in an FC81 polycarbonate flow chamber (Biosurface Technologies Corp, Bozeman, MT) (chamber dimensions are 50.5 mm by 12.7 mm by 2.54 mm) at a flow rate of 10 ml per hour at 25°C, essentially as described by Shao et al. (23) with two exceptions (see below). Briefly, whole saliva was collected according to a guideline approved by the Institutional Review Board of the University of Louisville and diluted 1:2 with MilliQ water, centrifuged at 13,000 × g for 1 min to remove insoluble material, filter sterilized (pore size, 0.22 μm), and incubated on the cover glass (60 mm by 24 mm) for 30 min at 37°C. Although clarification and filtration may remove some salivary constituents that may contribute to biofilm formation, A. actinomycetemcomitans adhered well to the treated slides and formed robust biofilms under the conditions used. The saliva-coated cover glass was then fixed in the flow chamber and washed with phosphate-buffered saline (PBS; 136.8 mM NaCl, 2.68 mM KCl, 10.14 mM Na₂HPO₄, 1.76 mM KH₂PO₄) for 30 min at a flow rate of 10 ml per hour using a peristaltic pump (Manostat Sarah cassette; Fisher Scientific, Pittsburgh, PA). A frozen glycerol stock of A. actinomycetemcomitans 652 WT, mutants, or complemented mutant strains were streaked onto BHI agar plates (when required it was supplemented with 25 μg/ml Km) and incubated at 37°C. A single colony from each strain from the plates was inoculated in BHI (10 ml in a 15-ml polypropylene tube) and was grown overnight, standing at 37°C. Each culture was used to inoculate BHI (45 ml in a 50-ml polypropylene tube) at a dilution of 1:20. The cultures were grown standing at 37°C to an OD₆₀₀ of 0.4. Each culture (45 ml) was inoculated for 3 h into the polycarbonate flow chamber at a flow rate of 10 ml per hour. After unbound cells were removed, bound cells were fed with BHI medium for 62 h at a flow rate of 10 ml per hour. The resulting biofilm was stained with 0.2 mg/ml fluorescein isothiocyanate (FITC; Sigma-Aldrich) for 1 h in the dark and then washed with PBS for 2 h. Biofilms were visualized using an Olympus Fluoview FV500 confocal scanning laser microscope (Olympus, Pittsburgh, PA) at a magnification of ×600, using an argon laser. Confocal images were captured from 20 randomly chosen frames from each flow chamber, and z-plane scans from 0 to 100 μm at 1-μm intervals were performed above the glass surface for each frame. The images were analyzed using Volocity image analysis software (PerkinElmer Inc.) to generate simulated three-dimensional images. Biofilm depth (average and maximal), biofilm biomass (average and total), and total surface of biofilm were also determined using the Volocity software package. The values reported are means of data from the different frames obtained. Biofilm assays were repeated independently three times with each strain. Differences among biofilms were determined by one-way analysis of variance (ANOVA) followed by Tukey's multiple-comparison test. Differences with P values of <0.05 were considered significant. Data were analyzed with GraphPad Prism version 5 software.

RESULTS

Determination of the minimal regulatory region of the A. actinomycetemcomitans lsrACBFG and lsrRK operons.

The divergent lsrACDBFG and lsrRK operons are conserved in all six serotypes of A. actinomycetemcomitans and are structurally organized similarly to the corresponding operons in E. coli and Salmonella (28, 29). To map the promoter regions that drive the expression of each operon, several DNA fragments encompassing a portion of or the entire 255-bp intergenic region (IGR) were amplified by PCR and cloned into the low-copy-number promoterless lacZ plasmid pJT3 (see Fig. 1). The recombinant plasmids and a control vector without a promoter were introduced individually into A. actinomycetemcomitans 652 by electroporation, and β-galactosidase (β-Gal) activity was determined. As shown in Fig. 1, plasmids pATE13 and pATE68, containing nucleotides −1 to −88 and −82 to −255, respectively, of the IGR transcribed lacZ, suggesting that multiple promoters may drive lsrR transcription. However, β-Gal activity expressed by these constructs was 5- to 6-fold lower than plasmid pATE23 containing the entire 255-bp IGR. Furthermore, pATE13 and pATE68 exhibited minimal increase in activity in early stationary phase, whereas pATE23 was induced by approximately 2-fold in early stationary phase. Finally, plasmid pATE11 that contains a translational fusion encompassing the 255-bp IGR and 408 bp of the lsrR structural gene produced levels of β-Gal activity similar to that of pATE23 in both mid-exponential and early stationary phases of growth (Fig. 1). β-Gal activity was not detected in the A. actinomycetemcomitans 652 or A. actinomycetemcomitans 652/pJT3 control strains. These results indicate that the entire 255-bp IGR is required for expression of lsrR and that sequences important for the optimal transcription of lsrRK are located between −89 and −255 bp.

Fig 1.

Schematic diagrams of the 255-bp Aggregatibacter actinomycetemcomitans lsrA-lsrR intergenic region as annotated in the genome database of A. actinomycetemcomitans D11S-1 (not to scale) and the reporter constructs that were analyzed. Portions of the intergenic sequence (IGR) are represented by thin lines, and lsrA or lsrR coding sequences are shown by the open boxes. The lacZ gene is shown with the thick black arrow. β-Galactosidase activity for each construct is expressed as Miller units and was measured in A. actinomycetemcomitans strains transformed individually with each plasmid and grown in BHI medium as described in Materials and Methods. Measurements were made at the mid-exponential and early stationary phases of growth (OD₆₀₀ of 0.3 and 0.5, respectively), since these are the stages where AI-2 levels are highest (44). The values are means of results from three independent experiments with (±) standard deviations.

For the lsrA-lacZ transcriptional fusions, no β-Gal activity was detected in strains transformed with pATE14 or pATE26, containing 88 bp or 255 bp upstream from the putative lsrA gene start codon, respectively (Fig. 1). Bioinformatics analysis of the putative lsrA coding sequence identified a second potential in-frame start codon at position +160 that was preceded by a putative ribosome binding site. To determine if the +1 to +159 lsrA fragment promoted transcription of the lsr operon, pATE71 was tested. As shown in Fig. 1, pATE71 directed low but detectable levels of β-Gal activity. However, pATE21, which contains the intact IGR fused to the +1 to +159 lsrA fragment, produced significantly higher β-Gal activity than pATE71, and similar results were obtained for pATE33 and pATE34. Thus, the lsrA promoter (lsrA-P) may require sequences within the putative lsrA coding region, or alternatively, the start codon of lsrA may be ATG at +160.

Further bioinformatic analysis of the lsrA-lsrR IGR identified two inverted repeats (see Fig. 2) that resemble the operator sequences of E. coli LsrR binding sites (34, 35). The first is located between nucleotides −123 to −85 (referred to as O₂-O₁), and the second resides upstream at nucleotides −226 to −197 (O₄-O₃), relative to the lsrR start codon. In addition, two putative CRP binding sites were identified, one residing between O₂-O₁ at nucleotide −113 to −98 and the second further upstream at nucleotides −173 to −158 (referred to as CRP₁ and CRP₂, respectively). CRP₂ (5′-TGAGA-N₆-TCACA-3′) closely resembles the consensus recognition sequence of E. coli CRP (5′ TGTGA-N₆-TCACA-3′), including G at position 4 and C at position 13 (36). CRP₁ exhibits lower similarity to the consensus CRP binding site but appears to contribute to CRP interaction with the promoter (see below and Fig. 4). These results suggest that both LsrR and catabolite repression may regulate the expression of the A. actinomycetemcomitans lsr operon.

Fig 2.

Double-stranded DNA sequence of the lsrA-lsrR IGR. The numbers on the right-hand side represent nucleotides upstream from the start codon of lsrR to the internal ATG codon of lsrA. Inverted repeat sequences labeled O₁-O₂ and O₃-O4 that represent potential LsrR binding sites are shown in bold text. Regions that resemble the consensus CRP binding site are underlined and labeled CRP₁ and CRP₂. The currently annotated lsrA start codon, the in-frame ATG codon in lsrA, and the lsrR start codon are indicated with arrows. Potential Shine-Dalgarno sequences are labeled SD.

Fig 4.

Binding of purified LsrR and cAMP-CRP complex to the lsrA and lsrR intergenic region. (A) Schematic representation of the lsrA-lsrR IGR showing the putative binding regions for LsrR (black boxes) and CRP (gray boxes). The PCR fragments encompassing the lsrA-lsrR coding region and IGR that were used in EMSA reactions are numbered from I to XII, and the position of each fragment is indicated to the right. Numbering is relative to the start codon of lsrR. PCR probes were incubated with 4 μM purified LsrR (B) or purified CRP (C) for 20 min at room temperature, and DNA-protein complexes (indicated by asterisks) were resolved in 6% polyacrylamide gels. A 200-bp DNA fragment from the psaA gene of Yersinia pseudotuberculosis was used as the negative control.

LsrR and CRP interact with the lsrA-lsrR intergenic region of A. actinomycetemcomitans.

In order to determine if the LsrR and CRP proteins interact with lsrA-P and/or lsrR-P, LsrR- and CRP-hexahistidine fusion proteins were purified (see Fig. 3) as described in Materials and Methods, and EMSA reactions were performed with a family of DNA fragments that spanned the 414-bp region that was characterized above (see Fig. 4A). Preliminary experiments showed that the optimal protein and probe concentrations required to form a DNA-protein complex was 4 μM and 50 fmol, respectively (data not shown). As shown in Fig. 4B, LsrR bound to probes that contained the putative binding sites identified in Fig. 2 (i.e., V, VI, VII, VIII, and IX), whereas probes comprising other regions of the IGR or coding sequences of lsrA or lsrR did not form DNA-LsrR complexes. However, probe X that contains only O₁ at its 5′ terminal end did not form a detectable complex with LsrR, suggesting that a complete inverted repeat sequence may be required for LsrR binding. CRP bound only to probes VI, VII, and VIII (see Fig. 4C), with the maximal shift occurring with probes VII and VIII that encompass both putative CRP binding sites. This suggests that CRP₁ may contribute to CRP binding even though it diverges in sequence from the consensus CRP binding site. Consistent with this, probe VII displayed two shifted bands, which might arise if CRP binds poorly to CRP₁ when it resides at the 3′ terminus of a probe. CRP binding to the other probes was not detected (not shown), nor did LsrR or CRP interact with a 200-bp DNA fragment from the Yersinia pseudotuberculosis psaA gene (probe XII), which was used as a negative control.

Fig 3.

SDS-PAGE analysis of proteins expressed in the cytoplasm of the E. coli LMG109 strain. (A) Lane 1, molecular-weight marker (MWM); lane 2, hexa-histidine LsrR purified by cobalt-based immobilized metal affinity chromatography. (B) Lane 1, molecular-weight marker; lane 2, hexa-histidine CRP purified by cobalt-based immobilized metal affinity chromatography.

A. actinomycetemcomitans LsrR regulates the expression of both lsrACDBFG and lsrRK.

In E. coli and Salmonella, LsrR functions as a repressor of lsrRK and lsrACDBFG. At high cell density, LsrK phosphorylates AI-2, and AI-2-PO₄ binds to LsrR, inducing a conformational change resulting in derepression (35). To determine if A. actinomycetemcomitans LsrR functions similarly, the reporter plasmids pATE23 and pATE33 (see Fig. 1) were introduced into A. actinomycetemcomitans strains harboring a chromosomal lsrR deletion (ΔlsrR), a chromosomal lsrK deletion (ΔlsrK), or a chromosomal lsrRK double-deletion (ΔlsrRK), and β-Gal activity was measured in cultures at late exponential and stationary phases of growth. In the absence of LsrR, the transcriptional activity of both lsrR-P and lsrA-P was increased relative to that of the wild-type strain, and similar results were obtained when the reporters were introduced into an lsrRK-deficient background (Table 1). However, the transcriptional activities of lsrR-P and lsrA-P did not differ significantly in the wild-type and lsrK-deficient backgrounds, suggesting either that in contrast to what occurs in E. coli, another kinase is capable of phosphorylating AI-2 in A. actinomycetemcomitans in the absence of LsrK or that LsrK is not essential for regulating the lsrACDBFG and lsrRK operons in A. actinomycetemcomitans.

Table 1.

Deletion of lsrR increases expression of the lsr locus

A. actinomycetemcomitans 652 strain	β-galactosidase activity (Miller units)
	Late exponential phase	Stationary phase
With pATE23 (lsrR-P)
Wild type	787.4 ± 38.7	924.4 ± 199.6
ΔlsrR mutant	1,163.1 ± 20.1	1,550.3 ± 23.4
ΔlsrRK mutant	1,146.0 ± 12.3	1,438.0 ± 112.2
ΔlsrK mutant	853.5 ± 12.3	991.6 ± 158.9
With pATE33 (lsrA-P)
Wild type	49.1 ± 2.7	64.7 ± 5.6
ΔlsrR mutant	63.4 ± 10.8	124.1 ± 20.9
ΔlsrRK mutant	73.0 ± 2.7	129.9 ± 12.2
ΔlsrKv mutant	22.1 ± 1.9	77.4 ± 10.5

Glucose influences lsrACDBFG and lsrRK expression through CRP.

To determine if glucose influences the activity of lsrR-P and lsrA-P, A. actinomycetemcomitans strains containing pATE23 and pATE33 (see Fig. 1) were grown in LB broth with and without 0.2% glucose, and β-Gal activity was measured after 24 h. As shown in Fig. 5A, β-Gal activity expressed by A. actinomycetemcomitans/pATE23 was reduced ∼40% in the presence of glucose. A significantly greater effect was observed with lsrA-P, as β-Gal activity of A. actinomycetemcomitans/pATE33 was ∼10-fold lower in the presence of 0.2% glucose than in the control (Fig. 5B). Similar results were obtained when the strains were cultured in TYE medium supplemented with glucose (not shown). To determine if CRP upregulates lsrACDBFG and lsrRK expression, A. actinomycetemcomitans 652 and an isogenic strain containing a genomic crp deletion (Δcrp) were transformed either with pATE23 or pATE33 and grown in the absence and presence of glucose. As shown in Fig. 5A, β-Gal activity expressed by the Δcrp/pATE23 was similar to that of the wild type grown in the presence of glucose. However, lsrR-P is transcriptionally active even in a CRP-deficient background or in the presence of glucose, suggesting that the lsrRK operon remains expressed during catabolic repression, albeit at lower levels. As shown in Fig. 5B, β-Gal activity was barely detected in Δcrp/pATE33, suggesting that CRP is essential for optimal expression of the lsrACDBFG operon. Together, these results indicate that CRP positively regulates both lsrR-P and lsrA-P, but it influences expression from lsrA-P to a significantly greater extent.

Fig 5.

β-Galactosidase activity of wild-type A. actinomycetemcomitans 652 (Aa) and an isogenic Δcrp mutant harboring either pATE23 (A) or pATE33 (B), which contain P_{lsrR −1 to −255}-lacZ, or P_{lsrA −414 to +182}-lacZ promoter fusions, respectively. Bacterial cells were grown statically in BHI medium overnight at 37°C and were then diluted (1:30) in fresh LB medium alone or LB supplemented with 0.2% glucose. After incubation for 24 h, β-Galactosidase activity was measured. The values are means of results from three independent experiments with (±) standard deviations.

lsrK and crp are required for optimal biofilm formation by A. actinomycetemcomitans.

Our previous work showed that biofilm growth by A. actinomycetemcomitans is dependent on AI-2 and that inactivation of lsrB significantly reduced biofilm biomass and average biofilm depth (21). To determine if disruption of the regulatory mechanisms that control lsrACDBFG and lsrRK expression influence biofilm formation, A. actinomycetemcomitans 652 and the isogenic ΔlsrR, ΔlsrK, ΔlsrRK, and Δcrp mutants were cultured in flow cells as described in Materials and Methods, and the resulting biofilms were visualized using confocal laser scanning microscopy and analyzed using Volocity software. Representative simulated three-dimensional images of the biofilms formed by each strain are shown in Fig. 6, and measurements of biofilm biomass, depth, and surface coverage are shown in Table 2. Deletion of lsrR or crp resulted in reduced biofilm formation, and the biofilms formed by these strains exhibited significantly lower biomass, average biofilm depth, and surface coverage than the wild-type strain. Deleting the entire lsrRK operon resulted only in a slight further reduction of each parameter (see Table 2). Interestingly, deleting lsrK alone produced a different phenotype than the lsrR mutants in that the ΔlsrK strain formed larger spreading microcolonies that were reflected in the greater biomass-per-microcolony value. However, the lsrK mutant did not exhibit significant differences in biofilm depth, total biomass, or surface coverage relative to the wild-type strain. Complementation of the ΔlsrRK and Δcrp mutant strains with the low-copy-number plasmids pATE52 and pATE60 containing the promoter and structural lsrRK and crp genes, respectively, resulted in increased biofilm depth, total biomass, or surface coverage to values similar to those of the wild-type strain (see Fig. 6 and Table 2). Together, these results suggest that proper regulation of the lsr locus by LsrR and CRP, but not LsrK, is required for optimal biofilm formation by A. actinomycetemcomitans.

Fig 6.

Three-dimensional renditions of biofilms formed by wild-type A. actinomycetemcomitans 652 and isogenic deletion mutants lacking lsrR (ΔlsrR), lsrK (ΔlsrK), lsrR and lsrK (ΔlsrRK), and crp (Δcrp). Mutations were complemented by a plasmid-borne copy of lsrRK or crp in plasmids pATE52 and pATE60, respectively. Biofilms were cultured in open flow cells as described in Materials and Methods and analyzed using confocal laser scanning microscopy. Image stacks were assembled and analyzed with Volocity image analysis software.

Table 2.

Measurements of biofilm biomass, depth, and surface coverage^a

Strain	Avg biofilm depth (μm)	Avg biomass (μm3)/ microcolony	Total biomass (μm3 × 103)	Total surface coverage (μm2 × 103)
652	10.4 ± 5.1	1,462.0 ± 876.8	665.2 ± 340.8	708.4 ± 260.5
ΔlsrR mutant	7.0 ± 1.3*	593.1 ± 398.3**	410.1 ± 179.7*	438.8 ± 104.9***
ΔlsrK mutant	7.7 ± 2.4	3,747 ± 1,712***	727.3 ± 158.5	568.6 ± 510.8
ΔlsrRK mutant	5.3 ± 0.3***	386.9 ± 79.5***	265.5 ± 280.8***	359.2 ± 397.0***
ΔlsrRK/pATE52 mutant	8.7 ± 1.2	1,689.0 ± 107.2	677.0 ± 140.4	849.5 ± 82.3
Δcrp mutant	6.5 ± 1.2*	353.7 ± 111.3***	336.2 ± 921.1**	458.2 ± 110.4**
Δcrp/pATE60 mutant	10.4 ± 0.1	1,845.0 ± 936.4	477.7 ± 136.5	581.4 ± 85.1

^a***, P < 0.001; **, P < 0.01; *, P < 0.05, in comparison to the wild-type strain.

DISCUSSION

Shao et al. (23, 25) previously demonstrated that A. actinomycetemcomitans biofilm growth was dependent on AI-2 and identified two periplasmic proteins, LsrB and RbsB, that function as receptors for the quorum-sensing signal. Inactivation of lsrB significantly reduced biofilm growth, and the mutant strain also exhibited reduced virulence in vivo (21). Shao et al. (23, 25) also showed that lsrB was present in an operon comprised of lsrACDBFG and that the genetic organization of the A. actinomycetemcomitans lsr locus was conserved with the divergent lsrACDBFG and lsrRK operons in E. coli and Salmonella. Genome sequences are now available for all six A. actinomycetemcomitans serotypes (37), and the lsrACDBFG and lsrRK operons are highly conserved in each.

In E. coli and Salmonella, lsrCADBFG is regulated by lsrRK, where lsrK codes for an AI-2 kinase and lsrR encodes a repressor of the lsr operon. At high cell density, lsrK phosphorylates AI-2 and AI-2-PO₄ binds to LsrR, resulting in derepression of lsrACDBFG (28, 29, 38). To determine if the regulatory mechanisms that control lsrACDBFG and lsrRK expression in A. actinomycetemcomitans are similar to enteric organisms and influence biofilm growth, the lsrA and lsrR promoters (lsrA-P and lsrR-P) were identified. Our results suggest that transcription of the lsr locus is complex and may be driven by several promoters. The lsrACDBFG and lsrRK operons are separated by a 255-bp IGR, and this IGR was shown to contain lsrR-P, and important regulatory sequences that are essential for optimal activity of lsrR-P were localized between nucleotides −89 and −255. However, the 255-bp IGR did not possess lsrA-P. Instead, part of this promoter was also localized to nucleotides +1 and +159 of the putative lsrA coding region. One possible explanation is that the in-frame ATG at nucleotide +160 to +162 in lsrA is the functional start codon for lsrA, and the IGR is 414 bp in length in A. actinomycetemcomitans. If so, lsrA codes for a protein of 453 amino acids rather than the 506-amino-acid polypeptide that is currently annotated in the A. actinomycetemcomitans genome sequences (LsrA of E. coli is 511 amino acids in length). Alternatively, it is possible that translation initiates at the currently annotated start codon but that sequences within the lsrA coding region contribute to lsrA-P activity. More detailed analyses of the +1 to +159 region are currently being carried out to address these possibilities. Finally, in addition to lsrR-P, the lsrRK operon appears to possess an internal promoter (lsrK-P) that drives expression of lsrK, since a fragment comprised of a portion of the lsrR coding sequence and the lsrR-lsrK IGR directed lacZ expression. However, β-Gal expression from this promoter did not increase as cells entered stationary phase and was approximately 20- to 25-fold lower than the growth-phase-inducible lsrR promoter (Fig. 1).

Bioinformatics analysis suggested that the IGR contained two inverted repeat sequences that represent potential LsrR binding sites and two additional sites that resemble the consensus CRP binding site, and indeed purified hexa-His LsrR and CRP fusion proteins bound to probes containing these sequences using EMSA. Furthermore, deletion of lsrR or lsrRK increased expression of lsrR-P and lsrA-P 0.75- and 2.0-fold, respectively, confirming that LsrR functions to repress expression of the lsr locus. In contrast, deletion of crp decreased expression of lsrA-P by ∼200-fold and lsrR-P by only 0.6-fold, suggesting that CRP activates expression transcription of the lsr locus, albeit much more dramatically for the lsrACDBFG operon. This finding confirms the results of a microarray expression analysis of A. actinomycetemcomitans JP2 and an isogenic Δcrp mutant (39) which identified a potential CRP-binding site upstream of an open reading frame designated Aa02226, which corresponds to lsrA. Together, this suggests that regulation of the A. actinomycetemcomitans lsr locus is similar to that of E. coli and Salmonella. Consistent with this, glucose also decreased expression from both lsrA-P and lsrR-P, as in E. coli (40), but again more dramatically from lsrA-P.

Surprisingly, deletion of lsrK had little to no effect on lsrA-P or lsrR-P activity at high cell density, suggesting that the role of lsrK in A. actinomycetemcomitans may differ from that in E. coli. In E. coli and Salmonella, LsrK phosphorylates AI-2 at high cell density, which is a necessary step to derepress lsrACDBFG (38, 41), and inactivation of lsrK results in a significant reduction in expression. This clearly did not occur in our experiments and might be explained by several possibilities, such as (i) a second kinase exists in A. actinomycetemcomitans that is capable of phosphorylating AI-2 and compensating for the loss of lsrK in the mutated strain, (ii) a signal other than AI-2-PO₄ is able to induce derepression via LsrR, or (iii) other regulatory mechanism(s) not yet identified may activate (or derepress) the lsrACDFG operon in A. actinomycetemcomitans.

Although regulation of the A. actinomycetemcomitans lsr locus by LsrR and CRP is similar to that in E. coli, the structure and architecture of the lsrA and lsrR promoters differ from their E. coli counterparts. First, the inverted repeat sequences representing the putative LsrR operator sites are A-T rich in A. actinomycetemcomitans, and the lengths of sequence between operator elements differ between the two sites (e.g., AAAAAA-N₂₇-TTTTTT and GTATTTT-N₁₆-AAAATAC), whereas the E. coli operators (34) have higher G-C content and are spaced by 21 nucleotides (TGAACA-N₂₁-TGTTCA and TGAACA-N₂₁-TGTTCA). In addition, the E. coli LsrR binding sites reside 16 bp and 33 bp upstream from lsrR and lsrA, respectively, and overlap the putative −10 and/or −35 sequences of these promoters (34, 35). Thus, in E. coli, LsrR may repress transcription at least in part by blocking access of RNA polymerase to the basal promoter elements. In contrast, the binding sites in A. actinomycetemcomitans reside 84 bp and 188 bp upstream from lsrR and lsrA, respectively, and do not overlap the putative −10 and −35 sequences of lsrR since pATE13 lacks the LsrR binding sites but still promotes transcription of lacZ. This suggests that the mechanism of LsrR repression may differ in A. actinomycetemcomitans and E. coli, and additional work will be necessary to define the mechanism of LsrR repression of the lsr locus in A. actinomycetemcomitans.

The importation of AI-2 by LsrB and the Lsr transporter has been shown to be required for the induction of the QseBC two-component system, which in turn influences biofilm growth and virulence of A. actinomycetemcomitans (21). The results from this study also indicate that the proper regulation of the lsr locus is important for optimal biofilm formation since the ΔlsrR and ΔlsrRK strains exhibits a biofilm phenotype, with reduced biomass, biofilm depth, and surface coverage, that was very similar to that observed in a ΔqseC strain (21). However, the link between the lsr operons and the expression of qseBC and downstream regulation of gene expression remains to be determined. LsrR has been shown to regulate genes other than lsrACDBFG and lsrRK in E. coli (42) and S. Typhimurium (43), but the transcriptional activity of the qseBC promoter was unchanged in the ΔlsrR and ΔlsrRK strains (D. R. Demuth, unpublished), indicating that LsrR does not directly regulate qseBC expression. Deletion of lsrK had little significant effect on biofilm depth, biomass, or surface coverage, which is consistent with our results that lsrK did not play an essential role in regulating the lsr locus in A. actinomycetemcomitans. However, biofilm architecture was altered in the lsrK mutant, which produced large spreading microcolonies rather than the more punctate microcolonies exhibited by the wild type or lsrR mutant. In E. coli, lsrK regulates a distinct set of genes relative to lsrR (42), and it is possible that this also occurs in A. actinomycetemcomitans and that the dysregulation of one or more of these genes accounts for the unique biofilm phenotype exhibited by the lsrK deletion mutant. We are currently comparing transcriptional profiles of the wild type with both the ΔlsrR and ΔlsrK strains to define the LsrR and LsrK regulons in A. actinomycetemcomitans in order to identify regulatory genes that may couple lsr and qseBC expression or influence biofilm architecture in the lsrK-deficient background.