Background

Global transcriptional regulators execute the co-regulation of multiple pathways simultaneously, which allows microbes to adapt quickly to environmental changes. CodY is one such conserved regulator that controls the expression of both metabolic and virulence genes in response to nutrient starvation in Gram positive bacteria with low GC content [1, 2]. CodY functions primarily as a repressor during exponential growth, and its repressive action is inhibited when nutrients become scarce [3]. The crystal structures of Bacillus subtilis CodY fragments show that CodY harbors a GAF domain at the N-terminal region which interacts with branched-chain amino acids (BCAAs, e.g. isoleucine, leucine, and valine) [4]. The binding of the effectors leads to a change in conformation and activation of the C-terminal helix-turn-helix DNA binding domain [4, 5]. Functional analysis indicates that both BCAAs and GTP could activate the DNA binding activity of CodY in B. subtilis [6,7,8,9], whereas BCAAs are the only known effectors for CodY in Lactococcus lactis [10], Streptococcus mutans [11], and Streptococcus pneumoniae [12]. A recent crystal structure study of full-length CodY demonstrates that GTP binds to a site located in the hinge region connecting the N- and C-terminal domains. In comparison to the CodY proteins that are activated by GTP, the CodY proteins of L. lactis and S. pneumoniae exhibit the substitution of three amino acids in the proposed GTP binding site, which may explain the specificity of the cofactors [13]. The CodY binding consensus sequence, 5’-AATTTTCWGAAAATT, has been defined in B. subtilis and L. lactis [14, 15].

Although CodY is activated by BCAAs and mainly acts as a repressor, Listeria monocytogenes CodY remains active in the absence of BCAAs. Under such conditions CodY can activate the expression of prfA, encoding a master virulence regulator [16], and subsequently initiate a multifaceted regulatory network. In Bacillus thuringiensis, CodY upregulates the PapR-PlcR quorum sensing system by optimizing the uptake of PapR, the signaling peptide of the transcriptional regulator PlcR. The surplus PapR activates the DNA binding activity of PlcR, and subsequently upregulates the expression of the PlcR regulon and virulence of B. thuringiensis [17]. The above examples demonstrate the diversity of CodY regulation.

Streptococcus salivarius is the primary ureolytic species in the human mouth. Since urea is the most abundant nitrogen source in the saliva and crevicular fluids [18], the presence of ureolytic S. salivarius is critical for maintaining the pH homeostasis of the oral ecosystem and preventing the development of dental caries [19]. Past studies revealed that the expression of S. salivarius 57.I urease operon was upregulated during growth at acidic pH values and in the presence of excess carbohydrate [20]. Under these conditions, ureolysis is beneficial for the survival of S. salivarius [21]. In vivo and in vitro analyses have confirmed that the repression of the urease operon at pH 7 with limited glucose supply is mainly governed by CodY [22]. It is intriguing that the activity of CodY is modulated by growth pH and carbohydrate availability in S. salivarius, as neither factor is known to affect CodY activation. This observation leads to the hypothesis that growth pH and carbohydrate availability modulate the regulatory activity of CodY by altering the intracellular metabolite profile. This study analyzed the transcriptomes of chemostat-grown S. salivarius 57.I and its CodY-null derivative (strain ΔcodY) to construct the regulatory network of CodY in response to different growth conditions. The impact of CodY regulation on the stress responses and pathogenic capacity of S. salivarius was examined.

Results

The binding consensus sequence of CodY in streptococci

To define the CodY binding consensus in streptococcal species, the CodY regulons of Streptococcus agalactiae, Streptococcus dysgalactiae subsp. equisimilis, Streptococcus gallolyticus UCN34, Streptococcus gordonii CH1, Streptococcus mitis B6, S. mutans UA159, S. pneumoniae TIGR4, Streptococcus pyogenes M1, Streptococcus sanguinis SK36, Streptococcus suis 05ZYH33, Streptococcus thermophilus CNRZ1066, and Streptococcus uberis 0140 J were downloaded from the database and their putative CodY binding consensus sequence was determined. The consensus sequence (Additional file 1: Figure S1A), 5’-AATTTTCAGAAAATT, is close to the known CodY binding consensus sequence (Additional file 1: Figure S1B).

An overview of CodY-regulated genes and the role of CodY in global regulation

CodY regulates the expression of S. salivarius urease operon at both pH 7 and 5.5, with a limited (20 mM) or an excessive (100 mM) amount of glucose [22], suggesting that different levels of CodY activity would be detected in the growth conditions in this study. To test this hypothesis, the codY gene of S. salivarius 57.I was insertionally inactivated to generate strain ΔcodY. Genes that are differentially expressed between wild-type 57.I and strain ΔcodY under each growth condition were identified by using edgeR [23] and the regulation pattern of these genes was examined by using hierarchical clustering analysis (Fig. 1a). The expression patterns of wild-type 57.I and strain ΔcodY under different growth conditions were located in two clusters, demonstrating the global impact of CodY regulation in response to growth pH values and glucose concentrations (Fig. 1a). Furthermore, the expression patterns of wild-type 57.I grown in 20 mM and 100 mM glucose were separated, suggesting that the amount of glucose in the culture medium modulates CodY activity more effectively than pH. Finally, the expression pattern of strain ΔcodY under 20 mM glucose at pH 7 was most distant from that of the others, suggesting that CodY is most active at pH 7 with limited glucose supply.

Fig. 1
figure 1

Hierarchical clustering of differentially transcribed genes in chemostat-grown S. salivarius 57.I and ΔcodY. The expression patterns of differentially expressed genes under all growth conditions between the two strains (a), and the genes directly (b) and indirectly regulated by CodY (c) are shown. Bacterial cells were cultivated in TY containing 20 mM or 100 mM glucose, and the cultures were maintained at pH 7 or 5.5. The mode of CodY regulation is indicated on the right hand side of the figure

Full size image

As both activation and repression were observed (Fig. 1a), whether CodY affects the transcription profile of S. salivarius by both direct and indirect regulation was analyzed. Genes whose expression was affected significantly by codY deletion under at least one condition were identified. The direct targets were defined based on the presence of a CodY binding consensus in the 5′ flanking region of the gene or operon. A total of 153 genes was found; 83 and 70 genes are defined as direct and indirect targets of CodY, respectively. Hierarchical clustering analysis also indicated that the expression profile of strain ΔcodY under 20 mM glucose at pH 7 was most distant from the profiles of the other three conditions, in both direct (Fig. 1b) and indirect targets (Fig. 1c). While all direct targets were upregulated in strain ΔcodY, both upregulation and downregulation were observed in the indirect targets. The RPKM (Reads Per Kb per Million mapped reads) values of direct and indirect targets are listed in Additional file 2: Table S1 and S2, respectively. The targets are discussed below.

Genes regulated directly by CodY

In agreement with observations in other bacteria, genes encoding proteins catalyzing the uptake and degradation of peptides, the uptake and biosynthesis of amino acids and urease were repressed directly by CodY in S. salivarius 57.I (Table 1). Although repression by CodY was generally more active under glucose limitation compared to that under excessive glucose, the effect of pH on the regulation of oppA (Ssal_00190), pepC (Ssal_00280) and pepP (Ssal_01946) was unclear. On the other hand, in most of the genes involved in the biosynthesis or uptake of amino acids, the expression level between wild-type 57.I and strain ΔcodY was significantly different under at least three conditions, indicating that the differential expression of these genes in response to growth pH values and carbohydrate concentrations optimizes the fitness of S. salivarius under different growth conditions. Finally, all the genes of the urease operon were derepressed at pH 5.5 and upregulation was further enhanced under excessive glucose in S. salivarius 57.I (Additional file 2: Table S1). Additionally, the differences in the expression level between wild-type 57.I and strain ΔcodY under glucose limitation were generally greater than that under excessive glucose, and the effect of codY deletion at pH 5.5 was less evident under excessive glucose, supporting the hypothesis that CodY is least active at pH 5.5 under excessive glucose.

Table 1 Genes directly regulated by CodY in S. salivarius 57.I
Full size table

The expression of genes encoding proteins involved in regulation, metal ions transportation, carbon metabolism, virulence, and other cellular functions was also regulated by CodY (Table 1). Specifically, the expression of Ssal_00908, encoding a putative Mn2+/Fe2+ transporter, and of hmuU-fhuC-Ssal_01464, an operon encoding a putative ferric ATP-binding cassette transporter was repressed by CodY under all conditions, suggesting that CodY modulates metal homeostasis in S. salivarius 57.I. CodY also repressed the expression of gdhA (Ssal_01768), encoding glutamate dehydrogenase (GDH), and the nrgA (Ssal_01668)-glnB (Ssal_01667) operon, encoding an ammonium transporter and a PII-like nitrogen regulator, respectively. The repression of these genes was maximized under glucose limitation, indicating that CodY regulates nitrogen utilization in response to carbon availability. Furthermore, the expression of pgmB (Ssal_01810), encoding β-phosphoglucomutase for converting β-D-glucose-1-phosphate to glucose 6-phosphate, was up regulated at pH 7 under excessive glucose in strain ΔcodY. Finally, the identification of lysM (Ssal_01428, encoding a peptidoglycan binding protein) and the cidAB operon (Ssal_00555 and Ssal_00556, encoding a putative murein hydrolase exporter and a regulator) suggests that CodY regulates cell wall biosynthesis.

Indirect targets of CodY

Forty-eight indirect targets were downregulated in strain ΔcodY under at least 3 growth conditions, while 22 genes were upregulated (Table 2). Among the upregulated genes, Ssal_00630, Ssal_00982 and Ssal_01876 are annotated in GenBank as transcriptional regulators. Particularly, the protein product of Ssal_01876, CovR, is a dominant global regulator controlling virulence expression in streptococci [24,25,26]. Sequence analysis revealed that 33 of the indirect targets harbor a CovR binding consensus sequence (5′-ATTARA) [27] at the 5′ flanking region, and both upregulation (11 genes) and downregulation (22 genes) were observed in strain ΔcodY. Specifically, a putative cell wall-binding hydrolase gene (Ssal_00236) was downregulated in ΔcodY, suggesting that CodY indirectly modulates cell wall structure.

Table 2 Genes indirectly regulated by CodY in S. salivarius 57.I
Full size table

Inactivation of CodY also led to the down-expression of a gene cluster (fabH-acpP-fabKDGF-accB-fabZ-Ssal_01799-accCDA) encoding proteins involved in membrane fatty acid biosynthesis. Although the effects of growth pH values and glucose concentrations on the regulation of these genes are unclear, the differences between wild-type 57.I and ΔcodY in cells grown at pH 7 were generally greater than in those grown at pH 5.5 under the same concentration of glucose, suggesting that CodY was essential for optimal membrane biosynthesis at pH 7.

Inactivation of CodY promoted clumping, prolonged the doubling time, and reduced the biofilm formation in S. salivarius 57.I

To analyze the impact of CodY regulation, a codY complemented strain, CΔcodY, was generated as detailed in the methods, in which a single copy of codY was established on the chromosome of strain ΔcodY. During the preparation of S. salivarius cultures for growth studies, strain ΔcodY formed aggregates in overnight cultures, while wild-type 57.I did not (Fig. 2a). The colony size of strain ΔcodY was smaller than that of the wild-type 57.I (Fig. 2b). To reduce clumping during the course of the growth kinetic studies, the inocula of all strains were gently sonicated prior to inoculation and the cultures were shaken before each reading. The estimated generation time of wild-type 57.I in BHI was approximately 22 min, whereas strain ΔcodY exhibited a generation time of approximately 44 min (Fig. 3). A wild-type doubling time was observed in the complemented strain (CΔcodY), although the lag phase in strain CΔcodY was longer than that of wild-type 57.I under the experimental condition for unknown reasons.

Fig. 2
figure 2

The phenotypic characteristics of S. salivarius 57.I and its derivatives. a Broth cultures of strains 57.I, ΔcodY and CΔcodY in BHI. b The colonies on BHI agar

Full size image
Fig. 3
figure 3

Growth kinetics of S. salivarius 57.I and its derivatives in BHI. The results are representative of those from three independent experiments

Full size image

In both the static biofilm cultures (Fig. 4a) and the flow-cell system (Fig. 4b), cells of strain ΔcodY failed to adhere to the plastic surface effectively. Presumably, self-aggregation impeded colonization, leading to minimal biofilm formation. Both wild-type 57.I and strain CΔcodY formed thick biofilms containing both live and dead cells (Fig. 4b).

Fig. 4
figure 4

The biofilm formation by S. salivarius 57.I and its derivatives. a The mass of the static biofilms of S. salivarius 57.I, ΔcodY and CΔcodY. Values are expressed in terms of the means and standard deviations of three independent experiments; quadruplicate samples of each strain were used in each experiment. Significant differences were analyzed by the Student’s t test. **, p < 0.01. b Images of the biofilms in the flow cell system. The biofilms were grown in BMG for 10 h. Live (green fluorescence) and dead (red fluorescence) cells were stained with SYTO9 and PI, respectively, and visualized by CLSM

Full size image

Inactivation of CodY reduced the glycolytic rate, acid tolerance and the capacity against oxidative stress in S. salivarius 57.I

Regulation by CodY is likely to play a critical role in energy conservation, which could then influence acid tolerance via the activity of F-ATPase [28]. Since bacterial glycolysis is normally acid-limited, and the terminal pH value reflects the sensitivity of cells to acidification, the acid production from glucose and acid tolerance of all strains were evaluated to test this hypothesis. The acid production of strain ΔcodY was lower than that of wild-type 57.I and strain CΔcodY, and the final pH of strain ΔcodY was 0.5 unit higher than that of wild-type 57.I (pH 4.78 vs. pH 4.26) and strain CΔcodY (Fig. 5), suggesting that the acid tolerance of strain ΔcodY was lower than that of wild-type 57.I. Additionally, the expression levels of glk (Ssal_01383), pfk (Ssal_01270) pyk (Ssal_01268), and pgk (Ssal_00232), encoding glucose kinase, 6-phosphofructokinase, pyruvate kinase and phosphoglycerate kinase in the Embden-Meyerhof-Parnas pathway, in wild-type 57.I were higher than those in strain ΔcodY under at least three growth conditions (Additional file 2: Table S3), indicating that CodY positively modulates the glycolytic activity via an unknown mechanism. Finally, the growth of S. salivarius at pH 5.5 was compromised significantly by codY deletion (Fig. 6).

Fig. 5
figure 5

The acid production by S. salivarius and its derivatives. The cell suspensions at 109 CFU ml− 1 were adjusted to pH 7.2 and glycolysis was initiated by the addition of 55.5 mM glucose. The results are representative of those from three independent experiments

Full size image
Fig. 6
figure 6

The effect of pH on the growth of S. salivarius strains. 10 μl of serially diluted exponential growth phase cultures of strains 57.I, ΔcodY, and CΔcodY were spotted on BHI (pH 7.2) or BHI-HCl (pH 5.5) agar, and were grown for 24 h. The images are representative of three independent experiments

Full size image

CodY also participates in the regulation of the oxidative stress response. A comparable level of growth was observed in wild-type 57.I and strain CΔcodY in medium containing up to 3 mM H2O2, while reduction in growth was detected with strain ΔcodY (Fig. 7), indicating that inactivation of CodY sensitized S. salivarius to oxidative stress.

Fig. 7
figure 7

The effect of H2O2 on the growth of S. salivarius strains. 10 μl of serially diluted exponential growth phase cultures of strains 57.I, ΔcodY, and CΔcodY were spotted on BHI agar containing various concentrations of H2O2 and were grown for 24 h. The images are representative of three independent experiments

Full size image

CodY regulation was essential for the toxicity generated from S. salivarius infection in the G. mellonella larva model

To analyze whether codY deletion would affect S. salivarius against host innate immune clearance, the toxicity of S. salivarius was evaluated in the G. mellonella larva model (Fig. 8). Seven out of 12 larvae infected with S. salivarius 57.I died 2 days post infection, and greater than 107 CFUs were recovered from one dead larva. A similar death rate was observed with the group of larva infected with strain CΔcodY. Less than 102 CFUs was recovered from one live larva infected with strain ΔcodY 7 days post infection. Thus, CodY regulation was crucial for the survival of S. salivarius and the death of the larvae.

Fig. 8
figure 8

The toxicity of S. salivarius 57.I, ΔcodY and CΔcodY in the G. mellonella larva model. Heat inactivated S. salivarius 57.I suspension and PBS were used as negative controls. 12 larvae were used for each bacterial strain. Kaplan-Meier survival analysis showing significant reduction of toxicity in S. salivarius ΔcodY (p < 0.001). The results are representative of those from three independent experiments

Full size image

Discussion

Microbes in the oral cavity frequently encounter fluctuations in carbohydrate availability and pH. This study was designed to gain insight into the impact of these two environmental factors on CodY regulation. Using the chemostat culture system, we found that CodY is most active under glucose limitation. Reduced CodY repression under excessive glucose, when the level of nitrogen containing nutrients is likely to be limiting, could enhance the overall nutrient utilization and energy generation by promoting ammonium uptake and amino acid biosynthesis. Specifically, S. salivarius is able to utilize ammonium as a sole nitrogen source [21], therefore the expression of the nrgA-glnB operon under excessive glucose allows for the utilization of an alternative nitrogen source. In S. thermophilus [29], GDH mediates the conversion of carbon nutrients into amino acids to promote lactose utilization. Conceivably, the reduced repression of gdhA by CodY in S. salivarius under excessive glucose might promote the utilization of carbon nutrients by the upregulation of glutamate biosynthesis. On the other hand, variations in the effects of pH between CodY-regulated genes may result from additional pH-dependent regulation by other regulator(s), as seen in the regulation of the urease operon [22, 30]. Taken together, the current data suggest that the profile of intracellular metabolites, rather than the growth pH plays major roles in modulating CodY activity in S. salivarius.

In S. pyogenes, CovR is a master regulator of the CovRS two-component system, which predominantly acts as a transcriptional repressor [25]. Among the identified targets that potentially belong to the CovR regulon, some are activated by CodY in S. salivarius, suggesting that CodY and CovR regulate common genes in opposite directions. Similar regulation has been reported in S. pyogenes, in which CodY modulates CovRS-regulated transcriptomes in response to nutritional conditions [31]. The opposite regulation in response to specific growth conditions may fine tune the transcriptomes to enhance the fitness and virulence of S. salivarius. Thus, regulation by CovR in the absence of CodY modulation contributes, at least in part, to some of the phenotypes of ΔcodY.

The clumping of strain ΔcodY cells in liquid culture, which has not been reported in other CodY-null bacterial strains, is likely to result from CovR regulation in the absence of CodY modulation. It is established that defects in enzymes catalyzing cell wall metabolism could alter cell morphology, the chain length of streptococci [32], or the cluster size of staphylococci [33]. Among CodY-regulated genes that potentially participate in cell wall metabolism in S. salivarius 57.I, the expression of Ssal_00236, a potential member of the CovR regulon, is most affected by CodY deletion. Ssal_00236 is highly expressed in S. salivarius 57.I, the expression level of which is approximately 3–8 fold higher than that of dnaA under the same growth conditions (data not shown). Inactivation of CodY almost abolished the expression of Ssal_00236. Although the function of the protein encoded by Ssal_00236 and its homologs has not been characterized experimentally in streptococci, this protein is predicted to harbor a LysM motif and a CHAP domain in a manner similar to the Cse protein of S. thermophilus [34]. Cse interacts with the cell wall via the LysM motif and catalyzes cell wall hydrolysis via the endopeptidase activity of the CHAP domain. Inactivation of this locus results in abnormal daughter cell segregation and long chains of cells. The indirct regulation of Ssal_00236 by CodY might be required for the development of proper chain morphology. While the function of the CidAB system in cell wall hydrolysis has been demonstrated in both S. mutans [35] and S. aureus [36], in both cases the hydrolytic activity of the CidAB system is counteracted by the activity of the LrgAB system in response to the growth conditions, and neither system is regulated by CodY. Thus, the contribution of the up-expression of the cidAB operon and lysM (Ssal_01428) in the clumping of ΔcodY cells is unclear.

It is likely that CodY participates in acid tolerance of S. salivarius 57.I by regulating the metabolic status, urease production, and possibly, membrane integrity. Regulation of metabolic pathways by CodY could augment ATP generation for acid tolerance, as seen in non-ureolytic S. mutans [11, 37,38,39]. Additionally, the regulation of the urease operon by CodY not only contributes to pH homeostasis but also provides a surplus of nitrogen-containing nutrients, which could again, promote ATP generation for pumping intracellular H+ via the activity of F-ATPase at acidic growth pH values. Finally, studies in S. mutans have demonstrated that the upregulation of unsaturated membrane fatty acid biosynthesis enhances acid tolerance [40, 41], indicating that membrane integrity is essential for optimal acid resistance. Thus, the down regulation of the fabH-acpP-fabKDGF-accB-fabZ-Ssal_01799-accCDA gene cluster in strain ΔcodY could potentially lead to reduced membrane integrity and acid tolerance in S. salivarius.

CodY participates in the oxidative stress response in S. thermophilus [42] and S. pneumoniae [43] by activating the expression of genes encoding glutathione synthetase (gshF) and thiol peroxidase (tpxD), respectively. However, neither gshA (Ssal_00663, an ortholog of gshF) nor tpx (Ssal_01091) is regulated by CodY in S. salivarius 57.I. Instead, CodY regulates directly two iron uptake systems (the products of Ssal_00908 and hmuU-fhuC-Ssal_01464). Intracellular iron homeostasis is critical for controlling the generation of reactive oxygen species resulting from the Fenton reaction. In S. pyogenes, both upregulation of an Fe(II) exporter (PmtA) [44] or downregulation of iron uptake [45] enhances the oxidative stress resistance. Thus, it is possible that the elevated iron uptake in strain ΔcodY could lead to the perturbation of iron homeostasis, leading to reduced resistance against H2O2.

A putative fibronectin-binding protein (FBP) (encoded by Ssal_00672) was more actively expressed in strain ΔcodY than in wild-type 57.I under all growth conditions. Although homologs of FBP in Gram positive pathogens are known to participate in adherence, invasion and immune evasion [46, 47], the toxicity of strain ΔcodY in the G. mellonella infection model was significantly lower than that of wild-type 57.I. Thus, either Ssal_00672 is not required for the survival against innate immune responses, or the upregulation of Ssal_00672 enhances recognition by host immune cells, and subsequently enhances clearance.

Conclusions

In conclusion, this study demonstrated that metabolic adaptation in response to nutrient availability and growth pH is tightly linked to stress responses and virulence expression in S. salivarius (Fig. 9). The regulation of metabolism by CodY allows for the maximal utilization of available nutrients and ATP production. The counteractive regulation of the CovR regulon fine tunes the expression of the regulon in response to the nutrient availability and growth pH.

Fig. 9
figure 9

The regulation circuit of CodY in S. salivarius. The model suggests that the activity of CodY is modulated by growth pH and glucose concentrations via the amount of intracellular BCAA produced under each growth condition. Positive regulation is indicated by a plus sign and repression is indicated by a minus sign

Full size image

Methods

Construction of a codY-deficient and a codY-complemented derivative of S. salivarius 57.I

Primers used in this study are listed in Additional file 2: Table S5. Plasmid pSky2 [22], in which the region encoding the 25th to 78th amino acids (aa) of CodY was replaced by an erythromycin (Em) resistance gene (erm) [48], was used to insertionally inactivate codY in S. salivarius 57.I. The genotype of the Em resistant isolates was verified by colony PCR using primers Ssal_00403_S and Ssal_00405_AS. The correct strain was designated ΔcodY. An Ωkan-tagged codY gene was then inserted into the intergenic region between Ssal_00777 and Ssal_00779 of strain ΔcodY using ligation mutagenesis [49] to generate a codY complemented strain. A 1.1-kbp DNA fragment containing the codY gene, its 5′ flanking region of 250 bp and its 3′ flanking region of 90 bp was generated from S. salivarius 57.I with primers 57.I_CodY_800_XhoI_S and 57.I_CodY_1900_SphI_AS. A DNA fragment containing the Ωkan cassette was generated from plasmid pVT924 [50] with primers kan_BamHI_S and kan_XhoI_AS. Two DNA fragments for integration onto the chromosome were generated from S. salivarius 57.I using primer pairs 57.I_lacZ_1241_S/57.I_lacZ_4970_BamHI_AS and 57.I_lacZ_4970_SphI_S/57.I_ssal_00779_5950_AS, respectively. All four PCR products were digested, mixed in a ligation reaction and then used to transform S. salivarius ΔcodY by natural transformation [51]. The genotype of the kanamycin resistant isolates was verified by colony PCR. The correct strain was designated CΔcodY.

Bacterial strains and growth conditions

Bacterial strains used in this study are listed in Additional file 2: Table S4. S. salivarius 57.I [52] and ΔcodY were grown in a chemostat culture system (Sartorius) in 3% tryptone-0.5% yeast extract medium (TY) supplemented with 20 mM or 100 mM glucose at a dilution rate of 0.3 h− 1, corresponding to a generation time of 2.3 h. Cultures were maintained at pH 7 or pH 5.5 by the addition of 2 N KOH. Cells were grown for a minimum of 10 generations to achieve the steady state at each condition.

RNA-seq library construction, sequencing and gene expression analysis

Total cellular RNA was isolated from streptococcal strains as described [53] and further purified using the RNeasy kit (Qiagen). Ribosomal RNA was removed from the RNA preparation using the RiboMinus transcriptome isolation kits (Invitrogen). The RNA-seq library was constructed using SOLiD™ small RNA expression kit (ABI). All procedures were performed per manufacturer’s instruction. The sequence of the sample was determined on an ABI SOLiD 4.0 sequencer. Reads with a quality value greater than 8 were mapped to the genome sequence of S. salivarius 57.I (accession number CP002888.1) by using the SOLiD system analysis pipeline tool (Corona Lite) with mismatches up to 5 bases. RPKM were used to calculate the expression level of each gene. The RPKM values were log10 transformed and the correlation between biological and technical replicates under the same growth condition was examined to verify the quality of the analysis. The significant difference in the expression level of each gene between wild-type 57.I and strain ΔcodY under one specific growth condition was analyzed using a Bioconductor software package edgeR with a p-value cut off at 0.05 [23].

Identification and characterization of the targets of CodY

Information on the CodY regulons of various streptococcal species was retrieved from RegPrecise version 3.0 database [54]. A position weight matrix (PWM) of the 5′ flanking regions of the CodY regulon was constructed with the MEME software at the default parameter settings. The PWM model was used to search for potential CodY binding motif in the 400-bp region 5′ to the translation start site of all genes in S. salivarius 57.I by using MAST with a p value less than 10− 4 [55]. The transcriptomes of wild-type 57.I and strain ΔcodY under each growth condition were used to seek targets of CodY by using the following criteria. Firstly, a significant difference in the expression level was observed between strains in at least one growth condition, and the regulation pattern (repression or activation) was identical in all conditions. Secondly, the regulation by CodY was consistent in genes of one operon. Thirdly, genes that are differentially expressed between strains and are with a CodY binding consensus were considered direct targets; genes without a CodY binding consensus were considered indirect targets.

Growth kinetics

The growth kinetics of S. salivarius strains was analyzed using the Bioscreen C Microbiology Reader (Oy Growth Curve AB Ltd.) as described [56]. Overnight cultures were diluted at 1:20 in brain heart infusion (BHI) and grown to an optical density at 600 nm (OD600) of 0.8. The culture was sonicated at 10% amplitude twice, 5 s each, diluted at 1:1000 in BHI and then 300 μl of the diluted suspension was inoculated in a microtiter plate. The OD600 value was monitored at 10-min intervals for 10 h. The plate was shaken for 10 s before each reading.

The static and the flow-cell biofilm systems

The static biofilm formation of S. salivarius strains was examined by the described method [57] with modifications [56]. Overnight cultures in biofilm medium (BM) [58] containing 20 mM glucose (BMG) were diluted at 1:20 in BMG and grown to an OD600 of 0.8. The culture was sonicated as described above, followed by dilution at 1:100 in BMG. The diluted suspension was then inoculated in the wells of a 96-well microtiter plate. The plate was incubated at 37 °C in 10% CO2 for 24 h. The biofilm was stained with crystal violet and the absorbance at 562 (OD562) nm was determined with a microplate reader (SoftMax®Pro, Molecular Devices).

The biofilm grown in a flow-cell system was examined as described [56]. Overnight cultures in BMG were diluted at 1:20 and grown to an OD600 of 0.8. The cultures were diluted at 1:800 in BMG and 300 μl of the diluted cell suspension was injected into the flow cell chamber. The chamber was kept at 37 °C in an upside down position without medium flow for 4 h. The medium flow rate was set up at 5 ml h− 1 channel− 1 and the biofilm was grown for 10 h. The biofilm was stained with SYTO 9/propidium iodide (PI) reagent (Invitrogen) and observed by using confocal laser scanning microscope (CLSM).

The pH drop assay

The glycolytic activity of S. salivarius 57.I and its derivatives was evaluated by the acid drop analysis [59]. Cultures at an OD600 of 0.6 in BHI were harvested, washed once with 50 mM KCl-1 mM MgCl2, and concentrated in the same solution with an OD600 of 1.0. The cell suspension was equilibrated at pH 7.2 with 0.01 N KOH at room temperature, and the pH drop was initiated with the addition of 55.5 mM glucose. The pH value was recorded every 30 s for 1 h.

Assays for acid resistance and oxidative stress tolerance

The growth of S. salivarius 57.I and it derivatives at pH 5.5 was evaluated using the described method [11] with modifications. Overnight cultures were diluted at 1:20 in BHI and grown to an OD600 of 0.8. S. salivarius ΔcodY was subject to sonication three times prior to reading. The cultures were serially diluted and 10 μl of each dilution was spotted on BHI agar (pH 7.2) or BHI agar that has been adjusted to pH 5.5 by the addition of 2 N HCl (BHI-HCl). The plates were incubated at 37 °C in 10% CO2 for 24 h.

A similar approach was use to evaluate oxidative stress tolerance. Cultures were prepared as described above, serially diluted and spotted onto BHI agar plates supplemented with 0.5, 1, 2 or 3 mM H2O2. Plates were incubated at 37 °C in 10% CO2 for 24 h.

Galleria mellonella larva model

The toxicity of S. salivarius strains in the G. mellonella larva model was determined using the described method [60] with modifications. Bacterial cells were grown to an OD600 of 0.8 in BHI, harvested, washed twice with PBS, and suspended in PBS at 1.0 × 109 CFU ml− 1. 10 μl of the suspension was injected into the proleg of each larva with a 26 s ga needle (Hamilton). PBS and heat inactivated (75 °C for 30 min) bacterial suspensions with the same CFU were used as negative controls. For each bacterial strain, 12 larvae were infected. The infected larvae were kept at 37 °C without feeding and recorded daily until the death of all larvae. The significant difference between infection groups was analyzed by using Kaplan Meier curve with log-rank test.

To determine the CFU in a larva, the larva was surface-cleaned with 70% ethanol, placed in a 2-ml microfuge tube containing 1 ml PBS and 12 zirconia/silica beads (2.3 mm diameter). The larva was homogenized in a Mini-Beadbeater (Biospec Products) for a total of 40 s at room temperature. The bacteria count in the homogenate was determined by serial dilutions and plating.