Expression of Green Fluorescent Protein in Streptococcus gordonii DL1 and Its Use as a Species-Specific Marker in Coadhesion with Streptococcus oralis 34 in Saliva-Conditioned Biofilms In Vitro

Abstract

Streptococcus gordonii is one of the predominant streptococci in the biofilm ecology of the oral cavity. It interacts with other bacteria through receptor-adhesin complexes formed between cognate molecules on the surfaces of the partner cells. To study the spatial organization of S. gordonii DL1 in oral biofilms, we used green fluorescent protein (GFP) as a species-specific marker to identify S. gordonii in a two-species in vitro oral biofilm flowcell system. To drive expression of gfp, we isolated and characterized an endogenous S. gordonii promoter, PhppA, which is situated upstream of the chromosomal hppA gene encoding an oligopeptide-binding lipoprotein. A chromosomal chloramphenicol acetyltransferase (cat) gene fusion with PhppA was constructed and used to demonstrate that PhppA was highly active throughout the growth of bacteria in batch culture. A promoterless 0.8-kb gfp (′gfp) cassette was PCR amplified from pBJ169 and subcloned to replace the cat cassette downstream of the S. gordonii-derived PhppA in pMH109-HPP, generating pMA1. Subsequently, the PhppA-′gfp cassette was PCR amplified from pMA1 and subcloned into pDL277 and pVA838 to generate the Escherichia coli-S. gordonii shuttle vectors pMA2 and pMA3, respectively. Each vector was transformed into S. gordonii DL1 aerobically to ensure GFP expression. Flow cytometric analyses of aerobically grown transformant cultures were performed over a 24-h period, and results showed that GFP could be successfully expressed in S. gordonii DL1 from PhppA and that S. gordonii DL1 transformed with the PhppA-′gfp fusion plasmid stably maintained the fluorescent phenotype. Fluorescent S. gordonii DL1 transformants were used to elucidate the spatial arrangement of S. gordonii DL1 alone in biofilms or with the coadhesion partner Streptococcus oralis 34 in two-species biofilms in a saliva-conditioned in vitro flowcell system. These results show for the first time that GFP expression in oral streptococci can be used as a species-specific marker in model oral biofilms.

Streptococcus gordonii, a member of the viridans streptococci, is a human oral commensal whose primary colonization site is hard tissue such as enamel (21, 27). S. gordonii may colonize the oral cavity through adhesion to proline-rich proteins in salivary coatings of host surfaces (13, 14, 17) or to bacterial surface sites by specific adhesin-receptor interactions (6, 23, 38, 57). Intrageneric or intergeneric coaggregation between bacterial cells occurs by specific interactions between complementary surface molecules on the partner cells (6, 23). Subsequent growth leads to accumulation of oral bacteria and formation of dental plaque. Maturation of this oral biofilm to include pathogenic late colonizers can lead to the development of periodontal disease (35, 48). Streptococci and actinomyces constitute the majority of gram-positive bacteria in initial plaque, and they are coaggregation partners, a property that likely contributes to their predominance in dental plaque. Fusobacteria are the dominant gram-negative bacteria in samples taken from both healthy and diseased sites, and fusobacteria coaggregate with early colonizers and late colonizers, which are composed of several other gram-negative bacteria (25). Most of the late colonizers do not coaggregate with the early colonizers. Because fusobacteria coaggregate with both early and late colonizers, they have been proposed to be important bridge organisms in the transition from a healthy to a diseased state (22, 28). Elucidation of changes during the transition from the commensal state to disease states requires a detailed understanding of the architecture of these oral biofilms. Our focus is on studying the architecture of very early biofilms composed of streptococci and actinomyces in order to establish a basis for the oral bacterial community structure in its initial stages. The foremost tool for the study of biofilm structure (1, 29, 58) and function (41), particularly at the single-cell level, is confocal scanning laser microscopy (CSLM).

Green fluorescent protein (GFP) has been used as a marker to monitor living bacterial cells in situ (3–5, 42, 50, 52). This stable marker requires no extraneous substrates or cofactors and no sample preparation and is compatible with many fixation techniques, which confers an advantage over other marker systems (2, 40). As an endogenous marker, its ease of detection by CSLM results in no disruption of the microbial community (12). The protein is heat and pH stable (up to 65°C; pH 7 to 11) and resistant to denaturants and proteases. When GFP is continuously synthesized and converted into a mature, stable protein, no dilution of the fluorescence signal occurs during bacterial replication. As a reporter, gfp is broadly applicable for use with many species of bacteria and can be monitored noninvasively at the single-cell level. Thus, GFP is particularly useful in studying autochthonous bacterial populations.

A potential problem associated with the use of GFP in anaerobic bacteria is the requirement of oxygen for posttranslational folding of the GFP to generate the fluorophore (8). However it has been reported that in a flowcell with low biomass, a low level of residual oxygen remains (1 to 2 ppm), which is sufficient to allow GFP maturation for facultative anaerobes such as streptococci or oxygen-tolerant bacteria grown in such a flowcell system (15). A second potential problem is that some of the GFP can also precipitate into weakly fluorescent inclusion bodies. GFP mutants with more-intense fluorescence emissions have been identified and have become routinely used (7, 33). For example, a gfp mutant, designated gfp mut3, containing amino acid substitutions S65G and S72A, exhibits 100-fold-higher fluorescence intensity in Escherichia coli than the wild type (7). In addition to the enhanced fluorescence, a shift in the excitation maximum from 395 nm in the wild type to 480 nm in the mutant allows this construct to be more effectively excited by visible light. GFP mut3 also was more soluble, as cells expressing this protein contained fewer inclusion bodies than those expressing wild-type GFP (7).

Several bacterial strains carrying gfp either on a plasmid or chromosomally have been constructed, and GFP has been used as a species-specific marker to label adherent bacteria in biofilms. GFP has been used in enteric dual-species biofilms (11, 47), and GFP-labeled Pseudomonas putida has been used to study bacterial survival in activated sludge (11). GFP was used to analyze the spatial distribution in a biofilm of fluorescent Enterobacter agglomerans and nonfluorescent Klebsiella pneumoniae (46, 47). An endogenous leukotoxin promoter has also been shown to drive expression of the GFP in Actinobacillus actinomycetemcomitans, a human oral pathogen (31). Transcriptional fusions of promoterless gfp can be used to monitor cellular responses to specific environmentally induced signals (55). For instance, gfp has been used to study acid-inducible promoters in Salmonella enterica serovar Typhimurium in macrophage phagosomes (54). Reports of GFP expression in streptococci have been limited to its potential use as a non-antibiotic-based selection marker in Streptococcus thermophilus (49) and its expression from a constitutive Lactococcus lactis P32 promoter in Streptococcus bovis and S. gordonii, although the fluorescence signal has been weak in S. gordonii (44). To our knowledge, there have been no reports of GFP expression from an endogenous promoter in S. gordonii.

In the present study, vectors that expressed GFP mut3 (7) from an active endogenous S. gordonii DL1 promoter, the hppA promoter, were constructed. Insertional inactivation of hppA causes loss of the ability to grow on peptides of 5 to 7 amino acids (19). To investigate the stability of these vectors and their potential use as species-specific markers of S. gordonii DL1 in oral biofilms, we have explored their use in labeling S. gordonii DL1 adherent to a saliva-conditioned in vitro flowcell alone and in conjunction with coadherent Streptococcus oralis 34. A GFP-expressing S. gordonii DL1 variant would replace the need to conduct experiments with fluorescently labeled secondary antibodies against the strain-specific, primary-antibody-coated streptococci. This is particularly important when the primary antibody against S. gordonii DL1 and that against S. oralis 34 are produced in the same animal species.

MATERIALS AND METHODS

Strains and culture conditions.

All bacterial strains used in this study are listed in Table 1. Streptococci are of human origin and were routinely cultured in brain heart infusion (BHI) (Difco Laboratories, Detroit, Mich.) at 37°C under anaerobic conditions with the GasPak system (BBL Microbiology Systems, Cockeysville, Md.). BHI medium supplemented with kanamycin (250 μg per ml), spectinomycin (1,200 μg per ml) or erythromycin (5 μg per ml) was used to select for S. gordonii DL1 cells containing gfp vectors based on E. coli-streptococcal shuttle vectors pMH109 (18), pDL277 (30), and pVA838 (34), respectively. E. coli HB101 was grown on Luria-Bertani (LB; Gibco-BRL) broth or LB agar aerobically at 37°C. LB medium, supplemented with tetracycline (12 μg per ml), spectinomycin (100 μg per ml), and chloramphenicol (25 μg per ml), was used to select for E. coli transformed with various plasmids.

TABLE 1.

List of strains used

Strain	Relevant characteristic(s)	Reference or source
S. gordonii
DL1	Reference strain for streptococcal coaggregation group 1	27
OB416	DL1 containing chromosomally integrated pSFCAT2	This study
PK2585	DL1 containing pMA2	This study
PK2586	DL1 containing pMA3	This study
S. oralis 34	Reference strain for streptococcal coaggregation group 3	27
E. coli HB101	F−mcrB mrr hsdS20(rB− mB−) recA13 supE44 ara-14 galK2 lacY1 proA2 rpsL20(Smr) xyl-5 λ− leu mtl-1	Gibco-BRL

DNA manipulations.

Restriction enzymes and other DNA-modifying enzymes such as alkaline phosphatase were obtained from Promega (Madison, Wis.) or Boehringer Mannheim (Indianapolis, Ind.). Ligations were performed with a vector/insert ratio of 1:3 based on absorbance at 260 nm. Unligated vectors were treated with alkaline phosphatase to prevent vector recircularization. All plasmids were extracted and purified from E. coli by using the QIAGEN (Valencia, Calif.) Miniprep kit. Natural transformation of S. gordonii DL1 was performed as previously described (56) except that an additional 35 min was included during development of cell competence. At least 1,000 transformants were obtained per μg of plasmid DNA. PCR products and DNA extracted from gels were purified using the Bio 101 (Vista, Calif.) Geneclean kit. DNA concentrations were determined using Genequant II (Pharmacia Biotech, Cambridge, United Kingdom). DNA sequences were determined using the Perkin-Elmer Applied Biosystems 377XL Automated DNA sequencer.

Construction and characterization of the PhppA-′cat fusion.

For amplification of the S. gordonii DL1 PhppA region, oligonucleotides were synthesized based upon the nucleotide sequence of hppA and the intergenic region upstream, including a portion of the 3′ region of hppG, which is adjacent and in the upstream direction from hppA (GenBank accession no. L41358). To facilitate PCR product cloning into pMH109, which contains a promoterless cat gene (′cat) from a gram-positive bacterium (18), restriction sites (underlined in the primer sequences below) were incorporated into the primers. The primer pair comprised HJ25 (nucleotides 1335 to 1357; hppA locus), 5′-GCTTCTAGATAAACCAGTAACCAG-3′, and HJ26 (complementary to nucleotides 1718 to 1746), 5′-TATGAGCTCCATTTCAATTTCAAATTTAG-3′. By using the Expand System (Boehringer), a PCR product of the predicted size (411 bp) was amplified (30 cycles; primer annealing temperature, 50°C) from phage lambda gt11-cp2 DNA that carried a 1.85-kb EcoRI fragment of S. gordonii DL1 genomic DNA (20) containing the hppG-hppA intergenic region (19). The product was digested with XbaI and SacI and was cloned into similarly digested pMH109. The ligation mixture was transformed into E. coli HB101 with selection for tetracycline resistance (Tc^r) and chloramphenicol resistance (Cm^r). The recombinant pMH109 was then digested with XbaI and BamHI, and the excised 1.3-kb PhppA-′cat fragment was cloned into similarly digested pSF143 (51) to generate plasmid pSFCAT2. The pSFCAT2 plasmid was introduced onto the S. gordonii chromosome by transformation and selection for Tc^r to generate the recombinant strain OB416. This insertion also conferred on S. gordonii OB416 resistance to 1 to 2 μg of chloramphenicol/ml. To confirm that the plasmid had integrated with insertion-duplication of the target sequence at the predicted hppA chromosomal region, a nylon blot of HindIII-digested S. gordonii genomic DNA was probed with the labeled PCR product and, separately, with the BamHI fragment from pMH109 containing the cat gene.

To further ensure that the correct fusion constructs on the streptococcal chromosome had been generated, the oligonucleotide primer CATPCR (5′-AACACTAATATCAATTTCTGTGG-3′; complementary to nucleotides 1348 to 1370 of the staphylococcal cat gene [GenBank accession no. J01754]) was used for PCR with primer HJ25, to amplify the promoter-cat fusion DNA region from OB416 genomic DNA. The PCR product was sequenced directly using a cat-defined nested primer, CATSEQ (5′-CTAAAAGTCGTTTGTTGG-3′; complementary to nucleotides 1325 to 1342).

Primer extension analysis.

Purified RNA (60 μg) was prepared from S. gordonii OB416 by a modification of the method of Lunsford (32) as described previously (39). The detailed procedure for primer extension has been published previously (39).

Preparation of cell extracts and CAT assay.

Cell extracts were prepared by spheroplasting with mutanolysin as described previously (39). Protein concentrations were determined by using a Bio-Rad protein assay kit with bovine gamma globulin as the standard. Chloramphenicol acetyltransferase (CAT) enzyme activity was determined by the spectrophotometric method of Shaw (45), utilizing a UV-240 recording spectrophotometer (Shimadzu Corp., Kyoto, Japan) with a temperature-controlled cuvette chamber. The reaction rate was determined from the linear portion of the graph, corrected for background change in absorbance at 412 nm, and divided by 0.0136 to yield CAT activity expressed as nanomoles of chloramphenicol acetylated per minute at 37°C.

Construction of gfp plasmid vectors.

The plasmids used are listed in Table 2. E. coli-streptococcal shuttle vectors pMH109, pDL277, and pVA838 formed the basis, respectively, of pMA1, pMA2, and pMA3, the three gfp vectors constructed in this study. A promoterless gfp (′gfp) cassette from pBJ169 was PCR amplified (DNA Thermal Cycler; Perkin-Elmer Cetus) and subcloned into pMH109-HPP, replacing the ′cat downstream of PhppA to construct pMA1. The ′gfp cassette used encoded GFP mut3, which exhibits 100-fold-greater intensity than the wild-type protein when expressed in E. coli (7). To facilitate PCR product cloning into pMH109-HPP, restriction sites were incorporated into the primers designed to amplify (30 cycles; primer annealing temperature, 55°C) the ′gfp cassette from pBJ169. Primer pairs (with restriction sites underlined) were 5′-GTACGAGCTCGGAGGCATATCAAATGGGTAAAGGAGAAGAACTTT-3′ (Pr5F1, incorporating a SacI site) and 5′-GCACGTATAACGATCGCGGATCCTTGTATAGTTCATCCATGCCAT-3′ (Pr5R, incorporating a BamHI site). The PhppA-′gfp cassette was then PCR amplified (30 cycles; primer annealing temperature, 55°C) from pMA1 using primer pairs 5′-CGCGGATCCTAAACCAGTAACGAAGAAAGAGTATCA-3′ (Pr6F, incorporating a BamHI site) and 5′-CGCGGATCCTTGTATAGTTCATCCATGCCATGTGTA-3′ (Pr6R, incorporating a BamHI site). The amplified cassette was subcloned into the BamHI sites of pDL277 and pVA838 to generate pMA2 and pMA3, respectively. Plasmids pMA2 and pMA3 were used to transform S. gordonii DL1 to obtain the transformants S. gordonii PK2585 and PK2586, respectively (Table 1).

TABLE 2.

List of plasmids used

Plasmid	Relevant characteristic(s)a	Reference or source
pMH109	Tcr Kmr ′cat cassette; Streptococcus-E. coli shuttle vector	18
pBJ169	Apr; ′gfp cassette	B. Julien and D. Kaiser
pDL277	Spr; Streptococcus-E. coli shuttle vector	30
pVA838	Emr Cmr; Streptococcus-E. coli shuttle vector	34
pMH109-HPP	Tcr Kmr Cmr; pMH109 containing PhppA-′cat cassette; Streptococcus-E. coli shuttle vector	This study
pSF143	Tcr; Streptococcus integration vector	51
pSFCAT2	Tcr Cmr; pSF143 containing PhppA-′cat cassette	This study
pMA1	pMH109-HPP containing an 0.8-kb SacI-BamHI ′gfp cassette in place of the ′cat cassette	This study
pMA2	pDL277 containing a 1.2-kb BamHI-BamHI PhppA-′gfp cassette	This study
pMA3	pVA838 containing a 1.2-kb BamHI-BamHI PhppA-′gfp cassette	This study

^aResistance phenotypes: Ap, ampicillin; Sp, Spectinomycin; Em, Erythromycin; Km, kanamycin; Cm, chloramphenicol; Tc, tetracycline. 

Flow cytometric analysis.

The time course experiments for GFP fluorescence maintenance in S. gordonii DL1(pMA2) and S. gordonii DL1(pMA3) were performed by flow cytometric analysis using a Becton Dickinson FACScan (San Jose, Calif.). Data analysis was performed using LYSIS II software (Becton Dickinson). Debris and dead cells were excluded from the analysis by using forward and side scatter light gating, with 10,000 events acquired for analysis. Cells for fluorescence-activated cell sorter (FACS) analysis and optical density measurements (600 nm) were obtained by diluting 1:10 an overnight culture grown anaerobically in BHI (without antibiotics for S. gordonii DL1 and with appropriate antibiotics for other strains) into 20 ml of fresh, warm aerobic BHI with appropriate antibiotics or no antibiotics (S. gordonii DL1) in a 250-ml culture flask. A sample was taken at time zero, and the remaining diluted culture was placed on a shaker for vigorous aeration at 37°C. The time zero sample was immediately processed by pelleting the cells (3,000 × g, 5 min, 4°C), washing once with phosphate-buffered saline (PBS), and resuspending in PBS. Flow cytometric readings were taken in triplicate at 0, 2, 4, 6, 8, 10, and 24 h and processed as described above.

Preparation and inoculation of in vitro flowcell chamber.

An overnight culture of S. gordonii DL1(pMA2) was grown anaerobically in a medium consisting of tryptone, yeast extract, Tween 80, and glucose buffered to pH 7.5 with K₂HPO₄ (36) in the presence of spectinomycin (CAMG/Sp). The culture was diluted 1:20 in warm, anaerobic CAMG/Sp and then incubated anaerobically for 4 h. Cells were then pelleted, washed in diluted saliva (pooled, filter-sterilized, human saliva diluted 1:4 in sterile water [26]), pelleted again, and finally suspended in diluted saliva to a density of 8.3 × 10⁸ cells/ml. Glass flowcells for microscopy (250-μl capacity; laminar flow) (26) were coated statically for 15 min with diluted saliva. The chamber was inoculated with a 400-μl bacterial suspension of S. gordonii DL1(pMA2) and incubated statically for 15 min prior to initiating the flow of diluted saliva (flow rate, 0.2 ml/min). Saliva flow was continued for 15 min, after which monospecies biofilms were observed by CSLM. For dual-species biofilms, S. oralis 34 cells were prepared as described above and added after the 15-min saliva flow. A second 15-min static incubation was followed by a 15-min saliva wash as described above. The biofilm was observed by CSLM. To visualize S. oralis 34, dual-species biofilms were labeled with rabbit anti-S. oralis 34 antibody absorbed against wild-type S. gordonii DL1 to remove cross-reacting antibodies, followed by a Cy5-conjugated goat anti-rabbit antibody (Jackson ImmunoResearch Laboratories, Inc., Westgrove, Pa.).

CSLM analysis of flowcell biofilm.

Microscopic observations and image acquisition were performed on a CSLM (TCS 4D system; Leica Lasertechnik, Heidelberg, Germany) with detectors and filters set for fluorescein (GFP) and far-red (Cy5) detection. Extended focus images were generated by CSLM software, and images were processed for display using Adobe Photoshop 5.5 (Adobe, Mountain View, Calif.).

RESULTS

Construction of PhppA-′cat fusion and PhppA activity.

A segment of S. gordonii DL1 chromosomal DNA (411 bp) containing PhppA was obtained following PCR amplification and cloned 5′ to a promoterless cat gene (′cat) present in plasmid pMH109 to yield pMH109-HPP (Table 2). To generate a chromosomal promoter fusion, the PhppA-′cat fragment (1.3 kb) was excised from pMH109-HPP and cloned into pSF143, which does not replicate in S. gordonii (51), to generate pSFCAT2 (Table 2). Transformation was used to obtain integration of pSFCAT2 into the S. gordonii DL1 chromosome and insertion duplication of the PhppA promoter expressing cat; transformants, S. gordonii OB416 (PhppA-′cat), were selected for tetracycline resistance. Southern blot analysis using the labeled PCR product and, separately, the BamHI fragment from pMH109 containing cat confirmed insertion of pSFCAT2 at the hpp locus (data not shown).

To check that the predicted promoter fusion had been obtained, PCR was used to amplify the corresponding chromosomal region from S. gordonii OB416 (PhppA-′cat). The promoter-fusion construct sequence obtained is shown in Fig. 1 and is aligned with the respective wild-type sequence. The ribosome binding site (RBS) sequence for hppA (GGAGA) was replaced during construction of the PhppA-′cat fusion with the cat RBS sequence (GGAGG). The RBS-ATG spacing was increased by 1 nucleotide, from 7 nucleotides found in PhppA to 8 nucleotides. Primer extension was used to map the transcriptional start site of hppA by utilizing a cat-defined oligonucleotide (see Materials and Methods). The PhppA-′cat mRNA initiated at nucleotide position −51 relative to the start codon of cat (data not shown), equivalent to −37 nucleotides from the ATG start codon of hppA (Fig. 1). This defined a potential extended −10 sequence within PhppA (underlined in Fig. 1) and a potential −35 sequence for RNA polymerase binding.

FIG. 1.

Comparisons of sequence containing the PhppA region with corresponding sequences generated by the PhppA-′cat or PhppA-′gfp fusion. HJ26 is the complementary-strand oligonucleotide primer used for amplification of the PhppA sequence. Lowercase letters within this primer sequence indicate changes made from the chromosomal sequence to introduce a restriction enzyme recognition site to facilitate cloning of the PCR product. Potential −10 (underlined) and −35 (overlined) promoter sequences are indicated, and RBS sequences are boldfaced. The experimentally defined transcriptional start site for PhppA-′cat is indicated. An HJ26 primer synthesis error (T instead of C) was revealed by sequencing of the PhppA-′cat PCR product and is lowercased.

The activity of PhppA was determined during growth of S. gordonii OB416 (PhppA-′cat) by measuring CAT activity in cell extracts. Half-maximal specific activity of CAT was already present in cells as they entered the exponential phase of growth (Fig. 2). CAT specific activity continued to increase during exponential growth and maintained maximal levels through stationary phase. The PhppA-′cat promoter region contains strong transcriptional terminator sequences upstream (1,463 to 1,501 bp; accession no. A41358), suggesting that transcriptional read-through of cat from upstream was unlikely. Because the CAT activities of extracts of S. gordonii OB416 (PhppA-′cat) were up to twofold higher than the CAT activities observed previously with another endogenous promoter-cat fusion, PcshA-′cat (39), we chose PhppA as the candidate representing a strong, endogenous promoter to drive GFP expression in S. gordonii DL1.

FIG. 2.

Specific CAT activity (squares) in cells of S. gordonii OB416 (PhppA-′cat) at points during growth (circles) in batch culture. Results presented are from one representative experiment of four. Error bars, standard deviations of the means of triplicate samples.

Construction of gfp-expressing Streptococcus-E. coli shuttle vectors.

The hppA promoter was then fused to ′gfp to generate plasmid pMA1 (Fig. 3). Transformation of E. coli with this plasmid yielded fluorescent transformant colonies. However, attempts to transform S. gordonii DL1 with pMA1 by selection with 250 μg of kanamycin per ml of agar were unsuccessful. Lower concentrations of kanamycin were tested, and they yielded either no transformants (200 μg per ml) or confluent growth (150 μg per ml). It is possible that the origin of replication in pMA1, which is the same as the Staphylococcus and Bacillus origin of replication in pMH109 (18), does not function in S. gordonii DL1. Therefore, the PhppA-′gfp cassette in pMA1 was PCR amplified and subcloned into pDL277 to generate pMA2 and into pVA838 to generate pMA3 (Fig. 3). Transformation of E. coli with these shuttle vectors yielded fluorescent transformants in each case. The insert within pMA2 was sequenced to determine the sequence of the PCR-amplified promoter region of the PhppA-′gfp insert. The pMA2 sequence containing the hppA promoter region and including the gfp translational start codon is shown in Fig. 1. The promoter region is identical to that of PhppA-′cat, indicating that no change in sequence occurred during PCR amplification.

FIG. 3.

Construction of the streptococcal gfp vectors pMA2 and pMA3. Promoterless gfp incorporating SacI and BamHI sites was PCR amplified from pBJ169 (a) and subcloned into pMH109-HPP (b) downstream of PhppA to generate pMA1 (c). PhppA-′gfp incorporating BamHI sites was PCR amplified from pMA1 and subcloned into the BamHI site of pDL277 (d) to generate pMA2 (e) and into the BamHI site of pVA838 (f) to generate pMA3 (g).

Epifluorescence analysis.

Identification of GFP fluorescence by epifluorescence microscopy (with an HBO 50-W mercury vapor lamp) revealed only a small percentage (approximately 1%) of visibly fluorescent cells in randomly selected microscope fields of view for S. gordonii DL1(pMA2) and no fluorescence in S. gordonii DL1(pMA3). Fluorescence was detectable in E. coli carrying pMA2 to a level of approximately 60 to 70% fluorescent cells in any given viewing field. Further analysis, described below, indicated the low sensitivity of this technique for GFP detection. Flow cytometric analysis and CSLM were used to obtain more sensitive and quantitative fluorescence detection in S. gordonii DL1(pMA2) and S. gordonii DL1(pMA3).

Growth and flow cytometric analysis of GFP-expressing S. gordonii during aerobic incubation.

The stability of GFP expressed from pMA2 and pMA3 was determined by performing flow cytometric readings on cells harvested from aerobically grown cultures. Cells were incubated in Erlenmeyer flasks containing a volume of medium less than 10% of the flask capacity and were aerated by shaking to enhance posttranslational folding of the fluorophore and acquisition of fluorescence. Cell suspensions of E. coli HB101(pMA2) representing the positive control, untransformed wild-type S. gordonii DL1 representing the negative control, S. gordonii DL1(pMA2), and S. gordonii DL1(pMA3) were subjected to flow cytometric analysis after 0, 2, 4, 6, 8, 10, and 24 h of aerobic growth. The optical densities of S. gordonii DL1(pMA2) and DL1(pMA3) at the same time points indicated that S. gordonii DL1(pMA2) grew more slowly than S. gordonii DL1(pMA3) and that S. gordonii DL1(pMA3) grew at about the same rate as wild-type S. gordonii DL1 (Fig. 4A). All streptococci reached about the same optical density after 8 h.

FIG. 4.

Optical densities (A) and median fluorescence intensities (B) of untransformed wild-type S. gordonii DL1 (open triangles), E. coli HB101(pMA2) (open circles), S. gordonii DL1(pMA2) (solid circles), and S. gordonii DL1(pMA3) (open squares) at 2, 4, 6, 8, 10, and 24 h of incubation under aerobic conditions in shaking flasks at 37°C. Data in panel B are averages of triplicate readings from a representative assay.

The greatest increase in fluorescence relative to that of wild-type S. gordonii DL1 was observed in the first 2 h for S. gordonii DL1(pMA2); fluorescence was maintained and gradually increased for 24 h (Fig. 4B). Fluorescence was not detectable with S. gordonii DL1(pMA3). The relative differences in fluorescence intensity between the nonfluorescent wild-type S. gordonii DL1 and the gfp-carrying strains are revealed by comparing the overlay of the S. gordonii DL1 cell profile (Fig. 5, upper left panel) to the fluorescence shifts observed with the gfp-carrying strains at two selected time points, 0 and 24 h (Fig. 5, left and right panels, respectively). At 0 h, fewer than 1.0% of the untransformed S. gordonii DL1 cells in the population have a median fluorescence intensity of 34 U (Fig. 5, upper left panel). At 24 h, both S. gordonii DL1 (Fig. 5, upper right panel) and a second negative control, S. gordonii DL1 transformed with the pDL277 vector containing the ′gfp insert (data not shown), exhibited fewer than 1.0% of the cells in the population with a median fluorescence intensity of 62 U, indicating the requirement for PhppA to drive GFP expression.

FIG. 5.

Comparison of flow cytometric analysis of S. gordonii DL1 and S. gordonii DL1(pMA2) at 0 h (left panels) and 24 h (right panels). Cells were analyzed immediately after a 1:10 dilution of an overnight culture of transformed cells grown anaerobically (0 h) or after 24 h of aerobic growth in a shaker flask (24 h). Open jagged peaks on the left sides of all six panels represent a negative control of the overnight culture of untransformed wild-type DL1. Shaded peaks, fluorescent shifts of gated cell populations. Net % pos, percentage of cells exhibiting a median (med) fluorescence intensity of positive cells only.

The positive control, E. coli HB101(pMA2), showed 98% of the gated cell population exhibiting a median fluorescence intensity of 625 U (Fig. 5, lower left panel) at 0 h. After 24 h, E. coli HB101(pMA2) showed 98% of the gated cell population exhibiting a median fluorescence intensity of 599 U (Fig. 5, lower right panel). At 0 h, 9.1% of the gated population of S. gordonii DL1(pMA2) showed a median fluorescence intensity of 32 U (Fig. 5, middle left panel). After 24 h, S. gordonii DL1(pMA2) exhibited a median fluorescence intensity of 179 U, with 75% of the cells in this gated population (Fig. 5, middle right panel). In contrast, after 24 h, only 2.2% of the gated cell population of S. gordonii DL1(pMA3) had a median fluorescence intensity of 63 U (data not shown), equal to that of untransformed S. gordonii DL1 or to that of DL1 transformed with pDL277 containing a ′gfp cassette. Thus, it appears that S. gordonii DL1(pMA2) is distinct from S. gordonii DL1(pMA3) in its ability to stably express GFP. S. gordonii DL1(pMA2) was chosen for studies in multispecies biofilms.

CSLM of S. gordonii DL1(pMA2) biofilm formation.

GFP-labeled cells bound to the saliva-coated surface after 15 min of static conditions and 15 min of salivary flow. A representative biofilm of S. gordonii DL1(pMA2) in the in vitro flowcell chamber is shown in Fig. 6. The adhesion pattern of S. gordonii DL1(pMA2) is the same as that seen with biofilms formed by wild-type S. gordonii DL1 under these conditions (26), indicating that S. gordonii DL1(pMA2) is equivalent to the wild type in its initial adhesion characteristics and could replace the wild type in in vitro model systems. Differential interference contrast (DIC) microscopy (Fig. 6A) and CSLM (Fig. 6B) of the same field of view showed differences in fluorescence intensity between single cells. The DIC image was overlaid onto the CSLM image and shows the varied expression of GFP intensity (Fig. 6C). The three cells indicated in Fig. 6 are in the same focal plane of this monolayer biofilm, as observed by the DIC image. Therefore, the fluorescence intensities in panel B (Fig. 6) are varied not because of variance in the cell depth in the biofilm but as a function of each cell's GFP content. The presence of large numbers of cells expressing pMA2-encoded GFP with intensities sufficiently detectable by CSLM confirms the earlier flow cytometric finding that GFP can be stably maintained in large populations of S. gordonii DL1(pMA2).

FIG. 6.

CSLM analysis of a GFP-labeled S. gordonii DL1(pMA2) biofilm after 30 min of incubation under conditions of salivary flow. A single field of view is shown by DIC, which shows all the cells in the field (A); by extended focus CSLM fluorescence detection of GFP, which shows the fluorescing cells (B); and by superimposing the DIC image from panel A at 60% opacity on panel B (C). The opacity of DIC was decreased to 60% so that the GFP-fluorescing cells from panel B can be seen beneath the slightly transparent cell ultrastructure visible in panel A. Although arrows 1, 2, and 3 point to cells of similar sizes and shapes (A), their fluorescence intensities vary (B and C) from bright fluorescence (arrow 1) to intermediate fluorescence (arrow 2) and dim fluorescence (arrow 3).

CSLM of coadhesion of S. gordonii DL1(pMA2) and S. oralis 34.

S. gordonii DL1(pMA2) and S. oralis 34 were inoculated sequentially into a saliva-conditioned flowcell and allowed to incubate with salivary flow for 15 min before analysis by CSLM (Fig. 7). Because S. oralis 34 is detected only by antibody coating of its surface and subsequent treatment with dye-conjugated secondary antibody, these cells appear larger than the S. gordonii DL1(pMA2) cells, which are detected by intracellular GFP fluorescence (Fig. 7). Examples of intrageneric coadhesion of S. gordonii DL1(pMA2) and S. oralis 34 are shown (Fig. 7). The cells of the two species are in close proximity both in the xy plane shown in Fig. 7 and in the z plane of this monolayer biofilm (data not shown). Note that there is no cross-reaction of the antibody with the GFP-expressing cells, indicating the species-specific nature of the two probes, GFP and specific antiserum, for identifying the spatial organization of these two streptococci in biofilms.

FIG. 7.

Extended-focus CSLM of coadhesion of GFP-labeled S. gordonii DL1(pMA2) and Cy5-labeled S. oralis 34 in a saliva-conditioned flowcell. GFP-expressing cells are green and are imaged directly. S. oralis 34 (So34) cells are first coated with rabbit anti-So34 (preabsorbed with wild-type S. gordonii DL1 cells) antiserum and then coated with secondary Cy5-conjugated goat anti-rabbit immunoglobulin G (red), which makes the cells appear larger than uncoated cells. Arrows indicate examples of intrageneric coadhesion of S. gordonii DL1(pMA2) and S. oralis 34.

DISCUSSION

This is the first report of the expression of GFP in S. gordonii using an endogenous promoter to drive gfp transcription. Transcriptional mapping of PhppA indicated that the RBS-proximal −10 sequence was likely to be active, and a potential −35 signal (TTTACA) was positioned accordingly, 18 nucleotides 5′ to the −10 sequence. The PhppA sequences thus strongly resembled sigma-70 consensus promoter sequences of E. coli (16). However, it is of interest that PhppA contained an extended −10 sequence (consensus, TNTGNTATAAT, where N is any nucleotide) (43) and that this was identical to the extended −10 sequence in the amiA promoter in Streptococcus pneumoniae (43). We speculate that the presence of a full −10 extension may be associated with the constitutive exponential-growth-phase expression of PhppA and the relatively higher strength of PhppA compared with PcshA of S. gordonii (39). The higher activity of PhppA is also likely to be significant in obtaining sufficient expression of GFP for detection by FACS and CSLM in our present study.

The use of a strong, active promoter contributed to the strength of gfp expression in S. gordonii DL1(pMA2). However, other factors also determine GFP expression, because the same cassette in the pVA838 transformant, S. gordonii DL1(pMA3), exhibited no fluorescence. The endogenous DL1 PhppA drives CAT expression in E. coli HB101 and S. gordonii DL1 under normal culture conditions. The plasmid copy number may also be important in the expression of plasmid inserts. The plasmid vector of choice for subcloning the gfp cassette, pDL277, is a low- to medium-copy-number Streptococcus-E. coli shuttle vector (10). Use of a high-copy-number plasmid may impose unnecessary physiological stress on a bacterium by diverting vital cellular functions into maintaining the high replication requirements of such a plasmid. The impact of the strength of the promoter expressing gfp versus the gene dosage of the plasmids carrying it must be taken into consideration, particularly in gram-positive hosts.

Transcriptional termination sequences flank the multiple cloning site of pDL277 (30) and thus are expected to flank the PhppA-′gfp insert in pMA2, but such sequences are absent in pVA838, the vector used to generate pMA3. These termination sequences may help stabilize inserts with active promoters such as PhppA, which if uncontrolled could interfere with plasmid copy number or cause plasmid deletion. These sequences may contribute to the success of gfp expression in pMA2, since both pMA2 and pMA3 are derived from low- to medium-copy-number plasmids, suggesting that copy number alone is an insufficient determinant for successful expression of vector inserts. Other constitutive streptococcal promoters capable of expressing gfp to higher levels would be desirable for use in a species-specific marker.

Many broad-host-range gfp-carrying plasmids have been used to label certain species of bacteria. One possible limitation of applying gfp-based vectors in bacterial strains used in environmental studies is the concern about plasmid stability under natural or simulated environmental conditions, such as the in vitro flowcell system used to simulate the in vivo conditions of the oral cavity. One way to circumvent this is to use bacterial strains chromosomally marked with a single copy of the gfp gene to maximize genetic stability as well as reduce the risk of transfer of the genetic marker to other microorganisms in the same environment. However, the problem with chromosomally based gfp expression is the low sensitivity of detection in cells containing a single copy of the gfp marker. Transposon-based suicide delivery vectors with gfp have been constructed to remedy this problem by generating strains with multiple transposon insertions (37, 50).

The use of a mutant GFP with a higher fluorescence yield enhanced GFP detection (7, 54). Low temperatures (49) and aerobic conditions (8) have been cited as crucial in rendering posttranslational formation of the chromophore more efficient, thereby enhancing fluorescence. We found that incubating plates containing S. gordonii DL1(pMA2) at 4°C did not enhance fluorescence. Aerobic incubation of S. gordonii DL1(pMA2) and S. gordonii DL1(pMA3) was conducted to augment GFP conformational changes, although at the expense of suboptimal growth conditions for facultative anaerobes such as the streptococci. One possible reason for the reduced levels of fluorescence detectable in streptococci is geometric quenching of the fluorescence signal by the thick peptidoglycan cell wall of gram-positive bacteria. However, GFP is known to be expressed well even under chromosomal expression in less-studied gram-positive bacteria such as Arthrobacter sp. strain A-6 (53).

Although GFP in our construct is cytoplasmic, refinement of the expression system for GFP beyond the methods used here could be considered. For example, targeting GFP to a defined area of the cell may remove GFP from the reducing environment of the cytoplasm. Debarbieux and Beckwith (9) have shown that the reductive enzyme thioredoxin 1 can change its function from that of a reductant to that of an oxidant when it is artificially exported to the E. coli periplasm. Exporting GFP to a potentially less reducing cell surface environment may prove conducive to the folding of the GFP fluorophore into the active fluorescent form. Fusion of GFP to surface lipoproteins such as ScaA (24) or HppA (19, 39), which are anchored in the cytoplasmic membrane but exposed to the cell exterior in gram-positive bacteria, would test this hypothesis.

In summary, GFP is expressed in S. gordonii DL1 by using a plasmid vector construct, but expression varies with the plasmid construct. The stable expression of GFP over time in a planktonic phase also suggests that GFP may be used as a stable marker for S. gordonii thriving in a saliva-conditioned biofilm. This knowledge has been applicable in employing GFP as a specific marker for S. gordonii alone or in conjunction with specifically tagged antibody-coated S. oralis 34 to identify the locations of these bacteria in an in vitro flowcell system. Application of GFP-expressing streptococci in studies of mixed-species biofilms in the oral cavity will elucidate the spatial arrangement of early colonizing members of the plaque community. Improvement in the level of gfp expression in S. gordonii and other oral streptococci will make it possible to use gfp as a reporter to study the regulation of contact-inducible or -repressible genes involved in the various intrageneric and intergeneric physical associations that occur within oral biofilms. Reporters of this kind are central to understanding the metabolic as well as contact-induced communication required for the transition from a commensal flora to a pathogenic flora in oral bacterial communities.