Uncovering Roles of Streptococcus gordonii SrtA-Processed Proteins in the Biofilm Lifestyle

S. gordonii interactions with its environment depend on the complement of cell wall proteins. A subset of these cell wall proteins requires processing by the enzyme sortase A (SrtA). The identification of SrtA-processed proteins and their functional characterization will help the community to better understand how S. gordonii engages with its surroundings, including other microbes, integrates into the plaque community, adheres to the tooth surface, and hematogenously disseminates to cause blood-borne infections. This study identified 26 putative SrtA-processed proteins through creation of a markerless deletion mutant library. The library was subject to functional screens that were chosen to better understand key aspects of S. gordonii physiology and pathogenesis.

ABSTRACT

Streptococcus gordonii is a commensal oral organism. Harmless in the oral cavity, S. gordonii is an opportunistic pathogen. S. gordonii adheres to body surfaces using surface adhesive proteins (adhesins), which are critical to subsequent formation of biofilm communities. As in most Gram-positive bacteria, S. gordonii surface proteins containing the C-terminal LPXTG motif cleavage sequence are processed by sortase A (SrtA) to become covalently attached to the cell wall. To characterize the functional diversity and redundancy in the family of SrtA-processed proteins, an S. gordonii DL1 markerless deletion mutant library was constructed of each of the 26 putative SrtA-processed proteins. Each library member was evaluated for growth in rich medium, biofilm formation on plastic, saliva and salivary fractions, cell surface hydrophobicity (CSH), hemagglutination, and integration into an ex vivo plaque biofilm community. Library members were compared to the non-SrtA-processed adhesins AbpA and AbpB. While no major growth differences in rich medium were observed, many S. gordonii LPXTG/A proteins impacted biofilm formation on one or more of the substrates. Several mutants showed significant differences in hemagglutination, hydrophobicity, or fitness in the ex vivo plaque model. From the identification of redundant and unique functions in these in vitro and ex vivo systems, functional stratification among the LPXTG/A proteins is apparent.

IMPORTANCE S. gordonii interactions with its environment depend on the complement of cell wall proteins. A subset of these cell wall proteins requires processing by the enzyme sortase A (SrtA). The identification of SrtA-processed proteins and their functional characterization will help the community to better understand how S. gordonii engages with its surroundings, including other microbes, integrates into the plaque community, adheres to the tooth surface, and hematogenously disseminates to cause blood-borne infections. This study identified 26 putative SrtA-processed proteins through creation of a markerless deletion mutant library. The library was subject to functional screens that were chosen to better understand key aspects of S. gordonii physiology and pathogenesis.

INTRODUCTION

Streptococcus gordonii is a Gram-positive commensal bacterium of the oral cavity and pioneer colonizer during formation of the dental plaque biofilm community (1). Following hematogenous spread, S. gordonii can be an opportunistic pathogen initiating infectious endocarditis (2). Fundamental to colonization within the oral plaque biofilm community and during endocarditis, S. gordonii cells adhere to the salivary pellicle and other proteinaceous films, surfaces, other bacteria, or host cells. Adhesion by S. gordonii cells is initiated by cell wall surface proteins.

Many cell wall proteins are anchored to the cell wall through the transpeptidase/endopeptidase, sortase A (SrtA), which is conserved across Gram-positive organisms; homologs are found in some Gram-negative bacteria and archaea (3). The motif recognized by SrtA on target proteins is a C-terminal pentapeptide, LPXTG, which is followed by a span of hydrophobic residues and a tail of positively charged amino acids (4). The SrtA-processed proteins vary in number, from 18 to 22 in Staphylococcus aureus, with similar numbers in other Gram-positive bacteria; they vary in presumptive function depending on the organism, but many show adhesive properties (5). Certain cell wall proteins can also be processed by accessory sortases (e.g., classes B to F), and their recognized substrate motifs are still being discovered (6, 7). In S. gordonii, sortase B is the only identified accessory sortase, processing the adhesin AbpA (8).

SrtA is required for S. gordonii binding to salivary agglutinin, fibronectin, oral keratinocytes, and saliva-coated glass, for biofilm formation, and for colonization of murine oral mucosa (9–11). In SrtA mutants, substrate proteins are not properly localized to the cell wall, which likely accounts for altered phenotypes. Specifically, S. gordonii SrtA is required for cell wall presentation of SspA, SspB, CshA, CshB, SGO_0707, SGO_0890, SGO_1487, and SGO_1247 (9, 10). SGO_1247 is a 5′-nucleotidase (12), whereas the other proteins are characterized or predicted adhesins (9, 12–16). In S. gordonii, 21 putative LPXTG motif-containing proteins have been identified by searching for orthologs in the PubMed protein database (9), but the family appears to be incomplete.

In this study, we identified 26 putative SrtA-processed proteins by in silico analyses. For each gene encoding the respective proteins, a markerless deletion mutant and library were created. This mutant library was subjected to several screens designed to evaluate properties of cell surface proteins associated with biofilms and adhesion with the goal of uncovering unique and redundant functions within this protein family. Each mutant was characterized by (i) growth in nutrient-rich media, (ii) biofilm formation on plastic, saliva, and salivary components, (iii) hemagglutination, (iv) cell surface hydrophobicity, and (v) integration into an ex vivo plaque biofilm community. These screens begin to characterize functions for the putative SrtA-processed proteins. The library itself serves as a useful tool for the study of these S. gordonii putative cell wall proteins.

RESULTS

Generation of the LPXTG protein markerless deletion mutant library.

We identified putative SrtA-processed proteins in S. gordonii DL1 using three approaches: a search for “LPXTG” in the S. gordonii Challis genome, a BLASTp search for “LPXTG” in S. gordonii Challis, and a LinkDB search for “gram_pos_anchor” domain. These search parameters also led to identification of two proteins, SGO_0208 and SGO_0385, with LPKTA and LPNTA motifs, respectively. In Bacillus, a class D sortase (SrtD) recognizes the LPXTA motif, so we included it in our list, as no SrtD has been identified in S. gordonii (3, 17). From this list, proteins that did not contain the LPXTG/A motif specifically within the C terminus of the protein were excluded from further study. Based on these criteria, 26 LPXTG/A-containing proteins were identified as putative SrtA-processed proteins (Table 1; see Table S1 in the supplemental material) including the 21 proteins identified in the earlier study (9). SrtA may not process all of the listed proteins, whereas others may not have been identified using this strategy. AbpA and AbpB are surface proteins involved in adhesion but are not processed by SrtA (8, 18, 19), and they were included in our studies for comparison (Table 1). Finally, a predicted function was curated for each protein/gene from the National Center for Biotechnology Information Resources (NCBI), including the NCBI annotation and information page for the gene and associated protein, a conserved domain search, and in some cases, a BLASTp search (Table 1).

TABLE 1.

S. gordonii LPXTG/A motif-containing and related proteins identified by in silico analysis

Motif	New locus	Old locus	Protein name or annotation	Reference
LPXTGa
A	SGO_RS08465	SGO_1727	ABC transporter substrate-binding protein/permease	This study
	SGO_RS08090	SGO_1651	Endonuclease	This study
E	SGO_RS06940	SGO_1415	X-prolyl dipeptidyl-aminopeptidase	This study
	SGO_RS06125	SGO_1247	Nt5e; 5′-nucleotidase	This study
	SGO_RS09810	SGO_2005	PadA; adhesin	This study
	SGO_RS04190	SGO_0854	CshA; adhesin	This study
G	SGO_RS03480	SGO_0707	Isopeptide-forming domain-containing fimbrial protein; adhesin	18
K	SGO_RS04375	SGO_0890	CnaB-type domain-containing protein	This study
	SGO_RS00535	SGO_0107	LPXTG cell wall surface protein	This study
	SGO_RS01560	SGO_0317	S8 family (subtilase) serine peptidase	This study
	SGO_RS01035	SGO_0211	SspB; antigen I/II family adhesin	18
	SGO_RS01030	SGO_0210	SspA; antigen I/II family adhesin	18
	SGO_RS01935	SGO_0388	Zinc carboxypeptidase	This study
L	SGO_RS07040	SGO_1435	FtsW/RodA/SpoVE family cell cycle protein	This study
	SGO_RS07295	SGO_1487	Large tandem repeat-containing protein	This study
N	SGO_RS08085	SGO_1650	LPXTG cell wall surface protein	This study
	SGO_RS07290	SGO_1486	BgaA; glycosylhydrolase	This study
	SGO_RS05815	SGO_1182	PavB; adhesin	18
	SGO_RS09805	SGO_2004	G5 domain-containing protein	This study
	SGO_RS02135	SGO_0430	SEC10/PgrA surface exclusion domain-containing protein	This study
	SGO_RS01555	SGO_0316	S8 family (subtilase) serine peptidase	This study
Q	SGO_RS02020	SGO_0405	StrH; β-N-acetylglucosaminidase	This study
R	SGO_RS05650	SGO_1148	CshB; adhesin	This study
	SGO_RS04725	SGO_0966 (SGO_0965)	Hsa; adhesin	This study
LPKTA	SGO_RS01920	SGO_0385	GH32 C-terminal domain-containing protein; exo-beta-d-fructosidase	This study
LPNTA	SGO_RS01020	SGO_0208	EndoD; endo-β-N-acetylglucosaminidase	This study
LPKTS	SGO_RS10300	SGO_2105	AbpA; adhesin	This study
None	SGO_RS00800	SGO_0162	AbpB; dipeptidyl peptidase	This study

^aThe letter in the column denotes the residue in the “X” position within the LPXTG motif.

Since functions of the majority of the S. gordonii proteins have not been well defined, we began characterizing the 26 S. gordonii proteins by creating markerless deletion mutants for each gene using a previously described strategy (20, 21). In the case of sspA and sspB, a double-deletion (ΔsspAB) mutant was used (20). As a negative-control strain, we constructed a ΔsrtA mutant. Finally, we created mutants with mutations of the non-SrtA-processed proteins AbpA and AbpB (ΔabpA and ΔabpB mutants) to be included in our library.

Growth in nutrient-rich media.

Considering that SrtA is regarded as a “housekeeping” sortase and potentially responsible for processing proteins essential to general cell physiology and metabolism (7), S. gordonii ΔsrtA or other mutants could have defects under normal growth conditions. To evaluate bacterial growth in nutrient-rich, complex medium, each mutant was inoculated into brain heart infusion (BHI) broth and incubated aerobically (Fig. 1). Stationary phase was typically reached by 12 h. No mutants displayed any dramatic growth changes compared to wild-type S. gordonii; the ΔsrtA, ΔSGO_0388, and ΔSGO_1415 mutants, however, showed slightly lower growth rates than the wild type. In contrast, the ΔSGO_1727 and ΔSGO_0405 mutants showed growth rates noticeably higher than those of the wild type.

FIG 1.

Growth of S. gordonii wild-type (WT) and mutants in rich medium. Strains were incubated aerobically in BHI for 12 h, and optical density at 600 nm (OD₆₀₀) was measured every 30 min. All strains were grown at the same time, but data were split into four graphs for visualization. Data are the means from 3 separate experiments.

Biofilm formation on plastic, saliva, and salivary components.

Several S. gordonii proteins in our list are known to be involved in adhesion or biofilm formation, specifically, AbpA, CshA, CshB, Hsa, PadA, SspA, SspB, SGO_0707, and SGO_1182 (13, 14, 16, 20, 22–25). To determine whether any of the other proteins are involved in biofilm formation, we tested each S. gordonii mutant on several surfaces. First, we evaluated each strain for biofilm formation after 24 h on tissue culture-treated (uncoated) or saliva-coated polystyrene (Fig. 2A). Many mutants displayed changes in biofilm biomass compared to wild-type S. gordonii on saliva-coated and/or uncoated plates.

FIG 2.

Biofilm formation by S. gordonii wild-type and mutants on several substrates. Bacteria were incubated in microplates coated with saliva or medium only (uncoated) (A) or MUC5B or LDP salivary fractions (B) for 24 h. Data are the percentage of the wild-type S. gordonii result from ≥3 separate experiments (means and standard errors of the means [SEM]). *, P ≤ 0.01 as determined by Student’s t test comparing each mutant to the wild type, with the wild-type value set at 100.

One of the dominant mucins in saliva is MUC5B, a component of the salivary pellicle that coats the teeth, serving as a substrate for bacterial adhesion and a nutrient source for the consortia of colonizing bacteria (26, 27). Using ultracentrifugation, saliva was partitioned into a MUC5B-enriched fraction and a second fraction enriched for the remaining low-density proteins (LDP), including MUC7 and gp340, two other glycoproteins involved in oral bacterial colonization (28). Following 24-h incubation on MUC5B and LDP, members of our mutant library were evaluated for biofilm formation and potential tropisms for binding to salivary components (Fig. 2B). Several mutants showed changes in biofilm biomass compared to wild-type S. gordonii on MUC5B-coated and/or LDP-coated plates.

Biofilm formation was compared in the four different growth conditions. Some mutants behaved similarly under all conditions. Others showed substrate-dependent biofilm formation. Overall, the extent of biofilm formation by mutants and wild-type was more similar on MUC5B and LDP than on whole saliva or polystyrene. Using a cutoff at least 20% greater than or less than the values for wild-type S. gordonii, notable mutant phenotypes became apparent. Not surprisingly, the ΔsrtA mutant had severe defects in biofilm formation on all substrates. Confirming previous observations, the ΔSGO_1182 and Δhsa mutants both formed significantly less dense biofilms on saliva than wild-type S. gordonii (20, 29). The Δhsa mutant also had significant defects on both MUC5B and LDP, whereas the ΔSGO_1182 mutant displayed a defect only on whole saliva. In addition to the ΔSGO_1182 and Δhsa strains, significantly less dense biofilms (≤80% of the wild type) were formed in at least one of the four conditions by ΔcshA, ΔsspAB, ΔSGO_0385, ΔSGO_0405, ΔSGO_0707, ΔSGO_0890, ΔSGO_1415, ΔSGO_1435, ΔSGO_1487, and ΔSGO_1650 strains. In contrast, ΔabpB, ΔpadA, ΔSGO_0107, ΔSGO_0316, ΔSGO_0388, ΔSGO_0430, ΔSGO_0707, ΔSGO_1415, ΔSGO_1435, ΔSGO_1650, ΔSGO_1727, and ΔSGO_2004 strains formed significantly denser biofilms (≥120% of the wild type) in at least one of the four conditions.

Hemagglutination.

Hsa is an adhesin that binds erythrocytes and promotes hemagglutination (25). Considering that other LPXTG/A proteins have adhesive properties, we screened all of the mutants for effects on hemagglutination (Fig. 3A). We considered the 100% effective dose (ED₁₀₀) the minimal dilution of bacterial cells that promotes agglutination of erythrocytes in the well. The average ED₁₀₀ values for each deletion mutant strain were compared to those of the wild type (Fig. 3B). As expected, Δhsa and ΔsrtA mutants had strong deficits in promoting hemagglutination (higher ED₁₀₀). Similarly, ΔcshA and ΔSGO_1435 mutants resulted in moderately less hemagglutination (higher ED₁₀₀).

FIG 3.

Hemagglutination in the presence of S. gordonii wild-type (WT) and mutants. (A) Hemagglutination assay plate with selected mutants. (B) Hemagglutination induced by each S. gordonii mutant compared to wild-type following 24-h incubation with erythrocytes. The highest dilution of bacteria that resulted in hemagglutination was considered the 100% effective dose (ED₁₀₀). Data are consolidated from ≥4 independent experiments and reported as means and SEM. *, P ≤ 0.001 as determined by Student’s t test comparing each mutant to the wild type.

Cell surface hydrophobicity.

Bacterial cell surface hydrophobicity (CSH) is often correlated with bacterial attachment or biofilm formation on solid abiotic surfaces (30). Therefore, we screened the mutant library for changes in CSH (Fig. 4). The ΔsrtA mutant showed 20% lower CSH than the wild type. The loss of individual cell wall proteins was expected to show more modest effects. Indeed, ΔabpB, ΔcshB, ΔpadA, ΔSGO_0107, ΔSGO_0317, ΔSGO_0388, ΔSGO_0430, and ΔSGO_1651 mutants showed small but statistically significant decreases in CSH compared to the wild type. Interestingly, ΔabpA and ΔSGO_0385 strains showed lower CSH than the ΔsrtA strain. There were no mutants with hydrophobicity that was significantly greater than that of the wild type.

FIG 4.

Cell surface hydrophobicity of S. gordonii wild-type and mutants. CSH of S. gordonii mutants as measured by hexadecane binding. Data compare each mutant as a percentage of wild-type S. gordonii CSH from ≥3 independent experiments and are reported as means and SEM. *, P ≤ 0.01 as determined by Student’s t test.

Considering that the loss of a hydrophobic or hydrophilic protein may alter CSH in a given mutant, we assessed the predicted grand average of hydropathicity (GRAVY) for proteins whose mutants displayed a significant decrease (31). Each protein’s GRAVY value failed to correlate with measured cell surface hydrophobicity for the respective mutant (data not shown).

Integration into an ex vivo plaque biofilm community.

S. gordonii is commonly found in dental plaque and other complex oral biofilm communities. As a measure of fitness contributed by putative SrtA-processed proteins, each S. gordonii mutant was tested for integration into an ex vivo plaque biofilm community under anaerobic conditions (Fig. 5). Most mutants integrated into the plaque community at least as readily as the wild type. Only ΔsrtA, ΔabpA, ΔabpB, Δhsa, and ΔSGO_1182 strains were less able to integrate into the ex vivo plaque community than the wild-type strain.

FIG 5.

Integration into an ex vivo plaque community by S. gordonii wild-type and mutants. Integration of each S. gordonii mutant into ex vivo plaque biofilm community as determined by qPCR amplification of the JHMD1 cassette as described in Materials and Methods. Data are presented as fold change in the mutant compared to the wild type from ≥2 independent experiments and reported as means and SEM. *, P ≤ 0.001 as determined by one-way analysis of variance (ANOVA), with Dunnett’s multiple-comparison test comparing each mutant to wild-type S. gordonii.

DISCUSSION

We created a library of markerless deletion mutants of all of the 26 putative SrtA-processed proteins in S. gordonii identified by the presence of the C-terminal LPXTG/A motif. Although there are predicted functions published in key databases, experimental confirmation for many of the proteins has not been reported. Therefore, we subjected our mutant library to several screens that begin to characterize their singular and parallel roles in S. gordonii biology. Based on analysis of the data from the screens, most proteins in the family show potential overlapping or redundant functions, whereas other functions may be unique to particular proteins (Table 2).

TABLE 2.

Summary of assay data for each mutant

Mutation	Resulta for:
	Growth on BHI	Biofilm formation				Hemagglutination	CSH	Plaque integration
		Plastic	Saliva	MUC5B	LDP
ΔsrtA	↓	↓	↓	↓	↓	↓	↓	↓
ΔabpAs	–	–	–	–	–	–	↓	↓
ΔabpB	–	–	↑	↑	–	–	–	↓
ΔcshA	–	↓	–	↓	–	↓	↓	–
ΔcshB	–	–	–	–	–	–	–	↑
Δhsa	–	↓	↓	↓	↓	↓	–
ΔpadA	–	–	–	↑	–	–	↓	↑
ΔsspAB	–	↓	–	–	↓	–	–	↑
ΔSGO_0107	–	–	↑	–	–	–	↓	↑
ΔSGO_0208	–	–	–	–	–	–	–	↑
ΔSGO_0316	–	–	–	–	–	–	–	↑
ΔSGO_0317	–	–	–	–	–	–	–	↑
ΔSGO_0385	–	–	↓	–	–	–	↓	↑
ΔSGO_0388	↓	–	–	–	↑	–	–	–
ΔSGO_0405	↑	–	↓	–	–	–	–	↑
ΔSGO_0430	–	↑	↑	↑	–	–	↓	↑
ΔSGO_0707	–	↓	–	↑	–	–	–	↑
ΔSGO_0890	–	–	↓	–	–	–	–	–
ΔSGO_1182	–	–	↓	–	–	–	↓	↓
ΔSGO_1247	–	–	–	–	–	–	–	↑
ΔSGO_1415	–	–	↓	↑	–	–	–	↑
ΔSGO_1435	–	–	↓	↑	–	↓	–	–
ΔSGO_1486	–	–	–	–	–	–	–	↑
ΔSGO_1487	–	–	↓	–	–	–	–	↑
ΔSGO_1650	–	–	↑	–	↓	–	–	–
ΔSGO_1651	–	–	–	–	–	–	–	–
ΔSGO_1727	↑	–	↑	↑	↑	–	–	↑
ΔSGO_2004	–	–	↑	–	–	–	–	↑

^aSymbols for BHI growth and plaque integration: ↓, less than wild type (WT); –, same as/similar to WT; ↑, greater than WT. Symbols for biofilm formation: ↓, ≤80% of WT; –, same as/similar to WT; ↑, ≥120% WT. Symbols for CSH: ↓, ≤90% WT; −, same as/similar to WT; ↑, ≥110% WT. Symbols for hemagglutination: ↓, statistically significant decrease from WT; –, no statistically significant difference from WT; ↑, statistically significant increase from WT.

In rich medium, the wild type and all of the mutants grow at largely similar rates (Fig. 1), suggesting that none of the proteins are required for proliferation in nutrient-rich medium. ΔSGO_0388 and ΔSGO_1415 strains showed a slight growth defect, whereas ΔSGO_1727 and ΔSGO_0405 strains showed a higher rate of growth, and these changes in growth could have impacted the biofilm formation by these mutants (Fig. 2). We cannot exclude the possibility that some family members contribute to processes impacting growth under certain conditions not tested in this study.

Certain mutants presented striking defects in one or more of the additional functional screens, as summarized in Table 2. For biofilm formation on plastic, for example, five mutants had a >20% decrease (loss of function) relative to wild-type and one mutant had >20% gain of function. On saliva, nine mutants showed a >20% decrease in biofilm formation, whereas six mutants had a >20% increase. Biofilm formation on saliva appeared to involve more genes encoding LPXTG/A-motif proteins than those on uncoated plastic. The many proteins that contribute to interactions with saliva appear to reflect functional redundancy with this complex substrate. Except for Hsa, no one protein appeared to be dominant. Similarly, on plastic alone, many proteins appeared to contribute, although the specificity of interaction is likely to differ from that on the saliva substrate.

On the other hand, some mutants showing biofilm defects on saliva were unaffected on MUC5B and LDP. These salivary-fraction substrates do not appear to recapture the complement of S. gordonii binding sites found in whole saliva. There appears to be functional redundancy in interactions with either of these salivary fractions. However, specific biofilm phenotypes were not generally reflected by other behaviors, such as growth in BHI, CSH, hemagglutination, or fitness in the ex vivo plaque community. Hence, the LPXTG/A proteins do play greater roles in some functional screens than others, suggesting that there are distributed functions within the family.

CshA is a fibrillar adhesin on S. gordonii known for its catch-clamp mechanism of attachment to fibronectin (32–34). An insertional mutant of cshA decreased CSH and interactions with Actinomyces naeslundii and Candida albicans (16, 35). CshB is a structurally similar fibrillar protein that binds fibronectin. When insertionally mutated, CshB, however, proved to have less impact on S. gordonii hydrophobicity and was not required to bind A. naeslundii and C. albicans (36). Coexpression of CshA and CshB, however, was required for S. gordonii colonization of the murine oral cavity (16, 35). Consistent with these studies, both ΔcshA and ΔcshB mutants showed slightly lower CSH than the wild-type strain, with a greater decrease in the ΔcshA mutant (Fig. 4). Additionally, we found that the ΔcshA mutant (but not the ΔcshB mutant) showed a significant decrease in hemagglutination (Fig. 3) and in biofilm formation on plastic and MUC5B (Fig. 2). In the remaining assays, both mutants displayed similar phenotypes, reflecting their primary sequence homology.

Hsa is a sialic acid-binding adhesin that recognizes different cell types and surfaces (37–39). S. gordonii Hsa also mediates coaggregation with Veillonella species, which express a corresponding S. gordonii-binding hemagglutinin (40). An hsa mutant showed defective biofilm formation on sialic acid-rich saliva compared to the wild type (29). Generally confirming earlier studies, we showed that the Δhsa mutant was defective in biofilm formation on all surfaces (Fig. 2) and unable to agglutinate erythrocytes (Fig. 3). We now show that the Δhsa strain was compromised in fitness in the ex vivo plaque community (Fig. 5). These Hsa-mediated adhesive features, including hemagglutination, appear to partially explain virulence in a rat model of S. gordonii-induced infective endocarditis (38, 41).

SGO_1182 is a homolog of the Streptococcus pneumoniae adhesin PavB, which contributes to nasopharyngeal colonization and respiratory infections (20, 42), which we previously reported to be an adhesin required for biofilm formation (20). Here, we confirmed the contribution to biofilm formation on saliva (Fig. 2) and now show that SGO_1182 contributes to integration into the oral plaque community (Fig. 5). We plan to determine whether the defect in plaque integration caused by ΔSGO_1182 reflects an inability to bind to a salivary substrate or to other microbes.

Not all of the LPXTG/A family proteins are predicted to be adhesins. SGO_1415 is a predicted X-prolyl-dipeptidyl aminopeptidase. In bacteria, including streptococci, X-prolyl-dipeptidyl aminopeptidases are involved in metabolism of polypeptides. The ΔSGO_1415 mutant had a very strong defect in biofilm formation on saliva and a mild defect on plastic and LDP. The phenotype is rather remarkable, since the ΔSGO_1415 mutant forms biofilm better than the wild type on MUC5B (Fig. 2). When present, X-prolyl-dipeptidyl aminopeptidase may digest a binding motif on MUC5B or on its own cell surface, while processing salivary substrates or another cell wall protein(s) for optimum S. gordonii biofilm formation. In contrast to the activity in monospecies S. gordonii biofilms, proteolytic digestion by X-prolyl-dipeptidyl aminopeptidases may remove or degrade binding sites for the wild-type strain in the plaque community, since the ΔSGO_1415 strain integrates better than the wild-type strain (Fig. 5).

The ΔSGO_0385 mutant had a significant defect in monospecies biofilm formation on saliva but no defects in the other biofilm conditions (Fig. 2) or in integration into the plaque community (Fig. 5). Also, although it showed a significant decrease in CSH (Fig. 4), no other phenotypic characteristics were affected. SGO_0385 is annotated as an exo-beta-d-fructosidase, clustering on the chromosome with other glycosyl hydrolases. One can speculate whether this enzyme contributes to metabolism of oral foodstuffs or to processing of a structural entity on S. gordonii or in the oral environment, although fructosides are not likely present among mammalian glycoproteins.

The ΔSGO_1435 mutant also had a significant defect in biofilm formation on saliva (Fig. 2) and was one of the few mutants impacting hemagglutination (Fig. 3). Functionally, SGO_1435 has conserved domains with the FtsW/RodA/SpoVE family of proteins, which function in cell division, elongation, or spore formation. Whereas growth was unaffected in rich medium (Fig. 1), the ΔSGO_1435 mutant could impact cell division under certain conditions, such as in the presence of saliva or following attachment to the salivary pellicle. The observation that ΔSGO_1435 resulted in less hemagglutination than the wild type suggests that SGO_1435 binds to erythrocytes or potentially impacts the expression of Hsa or its interaction with erythrocyte sialic acid.

For functional comparisons of unrelated surface proteins, the non-LPXTG/A family members AbpA and AbpB were included. Amylase-binding protein (AbpA) is an adhesin that binds salivary amylase (8, 18) and an abpA mutant is significantly impaired in biofilm formation after 20 min in a flow system (24). We observed no difference from the wild type under all biofilm conditions tested for the ΔabpA mutant (Fig. 2). Therefore, AbpA may function during initial tethering to the substrate in the presence of shear flow. Since by 24 h, our static biofilms successfully establish in the absence of AbpA, other adhesins are likely to be expressed to compensate for the loss of this early tethering adhesin. Interestingly, the ΔabpA mutant integrated more poorly into the plaque community than the wild type (Fig. 5), suggesting a previously unrecognized role during development of a multispecies biofilm. Despite the similar nomenclature, amylase binding-protein B (AbpB) is not homologous to AbpA and is a dipeptidyl peptidase with demonstrated proteolytic activity (19, 43). Nevertheless, like AbpA, AbpB is not required for monospecies biofilm formation (Fig. 2), yet it is required for full integration of S. gordonii into the plaque community (Fig. 5). This is consistent with a previous study demonstrating that AbpB is required for colonizing rat teeth (44).

Most mutants remarkably showed fitness in the ex vivo plaque biofilm community that is equal to if not better than that of wild-type S. gordonii (Fig. 5). In the competitive community environment, cells appear to compensate for the variability in monospecies biofilm formation on the various substrates; notably, no single LPXTG/A family protein was responsible for biofilm formation (Fig. 2), hemagglutination (Fig. 3), and CSH (Fig. 4). We hypothesize that LPXTG/A family proteins are functionally redundant to favor successful competition and inclusion in the complex oral environment. Moreover, as summarized in Table 2, some proteins show slight or no defects in all assays compared to the wild type. In some cases, the lack of phenotype may merely reflect screening assay conditions. Phenotypes may be uncovered following simple modification of protocol. For example, in biofilm formation, we evaluated biofilm biomass after 24 h. Some mutants may only show a defect at earlier time points, but the cells may be able to compensate for the loss. Indeed, S. gordonii can compensate for the loss of some adhesins to correct a biofilm defect, as we reported earlier (10, 20, 45).

In the oral environment, S. gordonii can compete for environmental niches by variously regulating adhesins typically to compensate for the lack of expression (10, 20, 45). The mechanism involves recognition and negative signaling in response SrtA-cleaved C-terminal peptides (C-peps). Following SrtA processing of LPXTG/A family proteins, C-peps remain in the cell membrane to interact with an intramembrane two-component system SGO_1180–SGO_1181 (20). In the absence of C-pep, adhesin gene expression and protein presentation are modified to effect binding to surface substrates using alternative adhesins.

Use of alternative adhesins and binding substrates by S. gordonii can also occur when substrate sialic acids are reduced or absent after hydrolysis by microbial neuraminidases (46). Therefore, the availability of a particular substrate on a surface or another microbe may result in the utilization of different adhesins or cell wall enzymes. For example, SGO_0208 (EndoD), SGO_0405 (StrH), and SGO_1486 (BgaA) are LPXTG/A-containing glycosyl hydrolases that promote S. gordonii growth in saliva (47). A triple mutant displayed the greatest growth defect in saliva and a significant defect in biofilm formation when grown in flow cells perfused with saliva (47). In our screens, each of the single mutants had variable performance in biofilm formation, with the ΔSGO_0405 strain showing the greatest defect, although it was modest (Fig. 2). Hence, these compensatory mechanisms, including functional redundancy, may explain why mutants can colonize the model plaque community better than the wild type.

In conclusion, we have created a markerless deletion mutant library of all S. gordonii LPXTG/A-motif containing proteins and AbpA and AbpB. This library can serve as a tool to aid our understanding of specific and redundant functions of the LPXTG/A family proteins, which are common to Gram-positive bacteria. These studies could lead to a deeper understanding of how S. gordonii colonizes within the complex oral community as a pioneer commensal and how S. gordonii can adapt to colonize distal sites of infection.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

S. gordonii DL1 was grown in BHI broth or on BHI agar plates (Difco, Sparks, MD) at 37°C in 5% CO₂ unless otherwise stated. For experiments requiring anaerobic culture condition, S. gordonii was cultured anaerobically at 37°C in an anaerobic mixed-gas chamber (Coy Laboratory Products).

Mutant construction.

We created markerless deletion mutants for each gene using a previously described strategy (20). In brief, a S. gordonii-specific selection cassette, JHMD1, was used in which the S. gordonii ldh promoter drives expression of ermAM for positive selection, and a mutated S. gordonii pheS gene was used for negative selection. To create constructs for allelic exchange, the JHMD1 cassette and regions of chromosomal homology of 500 bp immediately upstream and downstream of the gene to be deleted were PCR amplified from pJHMD1 and DL1 genomic DNA, respectively. The upstream and downstream regions and JHMD1 were linked using splice overlap extension PCR (SOE-PCR). The PCR product was transformed into wild-type S. gordonii DL1, and cells were plated on Todd-Hewitt broth (THB) supplemented with 5 μg/ml erythromycin and grown anaerobically. JHMD1 integrants were selected by resistance to erythromycin and sensitivity to 0.4% (wt/vol) dl-4-chlorophenylalanine (p-Cl-Phe). These integrants were stored in 10% glycerol at −80°C and used in the plaque integration assays. A second SOE-PCR linking the same regions of chromosomal homology was then transformed into the JHMD1-containing strain, plated on THB supplemented with p-Cl-Phe, and grown anaerobically. Resistance to p-Cl-Phe and sensitivity to erythromycin identified potential markerless mutants. The mutated regions were confirmed by multiple PCRs.

Growth curves in BHI.

Wild-type S. gordonii and each mutant were inoculated 1:100 (vol/vol) into BHI in tissue culture-treated, 96-well plates (Costar), and each plate was sealed with a Breathe-Easy (Sigma-Aldrich) membrane. The plates were incubated aerobically for 16 h at 37°C in a BioTek Synergy II plate reader. Every 30 min, the plates were shaken for 10 s and the optical density at 600 nm (OD₆₀₀) of each well was recorded.

Saliva collection and preparation for coating plates.

Stimulated whole saliva was collected and pooled from at least three healthy volunteers using protocols that were reviewed and approved by the Institutional Review Boards of the University of Minnesota and Malmö University. Saliva was collected by expectoration into tubes on ice followed by centrifugation for 10 min at 4°C and 1,000 × g (Beckman SX4750 rotor) to remove large debris and bacterial aggregates. The supernatants were collected, and the planktonic bacteria were pelleted by centrifugation at 15,000 × g for 20 min at 4°C.

Biofilm formation.

Following overnight culture in BHI, each strain was adjusted to an OD₆₀₀ of 1.0, inoculated at 1:100 (vol/vol) in BHI into round-bottom, tissue culture-treated 96-well plates (Costar), and incubated at 37°C and 5% CO₂ for 24 h. For some experiments, prior to inoculation, plates were coated with saliva or salivary fractions as described below. Biofilm formation was measured by crystal violet (CV) staining of adherent cells. To each well was added 50 μl of 0.1% crystal violet solution and plates incubated at room temperature for 15 min. Plates were gently washed 4 times with 200 μl PBS. Plates were air dried for 1 min and then 200 μl of acidified ethanol (4% 1 N HCl–96% ethanol [EtOH]) was added to each well. After 2 to 3 min, samples were mixed by pipetting up and down, and 125 μl of CV-ethanol mix was transferred to a flat-bottomed, polystyrene 96-well plate, and the absorbance at 570 nm was measured using a ChroMate 4300 microplate plate reader (Awareness Technology).

For biofilm experiments using saliva, sterilized saliva prepared as described above was dispersed 100 μl per well and incubated with rocking for 1 h at room temperature. Unbound saliva was then removed, and bound saliva was sterilized by exposure to UV irradiation for 15 min (Spectroline UV Crosslinker FB-UVXL-1000; Spectronic, Westbury, NY) prior to biofilm inoculation.

For biofilm experiments using MUC5B and LDP, fractions were provided by Claes Wickström at Malmö University, Malmö, Sweden, after purification from human saliva as previously reported (28). The fractions were used to coat microtiter wells for biofilm assays at concentrations approximating their native proportions in human saliva (i.e., 100 μl per well of 0.2 mg/ml MUC5B or 0.35 mg/ml LDP after dilution in 1× phosphate-buffered saline [PBS]). To coat, plates were incubated with rocking for 1 h at room temperature, and the salivary solution was gently aspirated.

Hemagglutination.

For preparation of erythrocytes, defibrinated sheep’s blood (Hemostat Laboratories) was centrifuged at 150 × g for 10 min (Beckman SX4750 rotor), supernatant was aspirated, and cells were washed three times with sterile 1× PBS using repeated centrifugation at 150 × g for 5 min. After the final wash, sterile 1× PBS was added to erythrocytes (10% [vol/vol]) and then diluted to 0.5% (vol/vol) for use in experiments. Following overnight culture in BHI, each S. gordonii strain was adjusted to an OD₆₀₀ of 1.0 and serially diluted 2-fold in PBS in a tissue culture-treated 96-well round-bottom plate (Costar); 50 μl of the erythrocyte suspension was added and the mixture was then incubated for 24 h at 4°C. The highest dilution of bacteria for which hemagglutination seen was considered the 100% effective dose (ED₁₀₀).

Cell surface hydrophobicity.

CSH was determined using hexadecane binding as previously described, with modifications (48). An overnight culture of wild-type S. gordonii and each mutant strain was adjusted to an OD₆₀₀ of 1.0, 1 ml was placed into a glass tube, and 100 μl of hexadecane (Sigma-Aldrich, St. Louis, MO) was added. The mixture was vigorously vortexed for 2 min and allowed to stand for 10 min at room temperature, when phase separation was complete. The OD₆₀₀ of the lower aqueous phase was recorded. Percent CSH was calculated using the following formula: [1 − (OD₆₀₀ after vortex mixing/OD₆₀₀ before vortex mixing)] × 100. The predicted grand average of hydropathicity for each protein was determined using the program ProtParam (31).

Ex vivo plaque biofilm integration.

The ex vivo plaque biofilm integration assay was performed as previously described (20). Four healthy volunteers provided supragingival plaque from the maxillary molars 24 h after brushing their teeth. The plaque was washed with prereduced sterile PBS and combined for overnight growth in SHI medium (49) at 37°C anaerobically. The microbial community was frozen at −80°C in 10% glycerol and served as the stock for biofilm integration experiments.

For biofilm integration, the ex vivo plaque community was grown anaerobically overnight in SHI medium and diluted to an OD₆₀₀ of 0.1 in SHI medium. For the S. gordonii strains, we used the markerless deletion mutant intermediates, which contain JHMD1, and a wild-type S. gordonii in which we integrated the JHMD1 cassette at the attB site (20). S. gordonii strains containing the JHMD1 cassette were grown anaerobically overnight in BHI and inoculated at an OD₆₀₀ of 0.0001. Biofilms were grown anaerobically for 24 h using 12-well polystyrene culture plates that had been coated with saliva as described above for biofilm formation (Corning). The planktonic cells were aspirated, and the surface-attached biofilm was washed once with 500 μl of prereduced sterile 1× PBS. Biofilms that remained attached to the wells were collected and centrifuged at 21,000 × g for 2 min, PBS was aspirated, and 50 μl of fresh 1× PBS was added to each pellet. Pellets were stored at −20°C until use. The presence of wild-type or mutant S. gordonii in the community was determined by qPCR amplification of the JHMD1 cassette with the primers pheSermAMFor (5′-TGG ATT TGG GTT CGG CTT AG-3′) and pheSermAMRev (5′-GTA ATC ACT CCT GGG ATC CTC TA-3′) and normalized to total 16S rRNA gene DNA, which was amplified with primer pair (BAC1 and BAC2) (49). Reactions were performed with the Mx3000 real-time PCR system (Stratagene) in 20-μl reaction mixtures (1× QuantaBio Perfecta Sybr green FastMix, with ROX reference dye, 1 μl of thawed bacterial pellets, and 100 nM each forward and reverse primer). The PCR conditions were (i) 95°C for 10 min and (ii) 40 cycles of 95°C for 15 s and 55°C for 30 s. The ΔΔC_T method was used to calculate relative abundance of each mutant strain compared to the wild type.