Introduction

Streptococcus mutans is considered a major etiologic agent of dental caries due to three main attributes: (i) a capacity to form biofilms on tooth surfaces (dental plaque), (ii) an ability to convert dietary carbohydrates to lesion-inducing lactic acid, and (iii) an ability to adapt to sudden environmental changes in dental plaque1. To thrive at low-pH values, S. mutans activates the acid tolerance response (ATR), a genetic and physiologic adaptive mechanism that is relatively well-understood1,2. The ATR is accomplished by upregulation of the membrane-associated F-ATPase, induction of pathways that contribute to cytoplasm buffering and changes in membrane fatty acid composition, among other processes1,2.

While the ATR has been studied in some detail, oxygen metabolism and the mechanisms utilized by S. mutans to cope with oxidative stress have received limited attention. The initial notion that the dental plaque (biofilm) environment was virtually anaerobic has been now replaced by evidence that the oral microbial community as a whole has a high capacity to reduce oxygen, resulting in the generation of a variety of toxic reactive oxygen species (ROS) such as H2O2 and superoxide3. For example, members of the mitis group of streptococci (e.g. S. gordonii and S. sanguinis), which cohabit the dental biofilm with S. mutans, are net producers of H2O2. It follows that an inverse correlation between proportions of S. mutans and mitis streptococci has been observed in health and disease, with high numbers of S. mutans associated with caries and high proportions of mitis streptococci associated with oral health4,5. In addition, H2O2 present in oral hygiene and tooth bleaching products may represent another source of peroxide stress for oral bacteria3. Ultimately, via free radical formation as a result of the Fenton reaction in the presence of iron, the presence of high levels of H2O2 can rapidly cause irreversible cellular damage by attacking membrane lipids, triggering mismetallation of enzymes, directly damaging proteins through oxidation of sulfurous amino acids and metal-binding sites, and by disturbing DNA integrity6.

Global transcriptional studies following exposure to H2O2 have been extensively used to obtain new insights into the peroxide stress response mechanisms of bacteria. In Escherichia coli, the peroxide stress response is largely controlled by the OxyR regulator that activates transcription of reactive oxygen species (ROS) scavenging, iron homeostasis and disulfide reduction systems7. In the soil organism Bacillus subtilis, peroxide stress responses are governed by σB and the thiol-sensing OhrR, as well as the peroxide-sensing PerR, which activate (σB and OhrR) or derepress (PerR) transcription of systems involved in ROS scavenging, iron homeostasis, DNA repair and manganese uptake8. In Staphylococcus aureus, microarray analysis has linked DNA repair pathways, iron uptake and storage, and anaerobic metabolism to peroxide stress9. Moreover, Spx is another major transcriptional regulator of Gram-positive bacteria involved in oxidative stress responses, principally by activating transcription of genes involved in thiol homeostasis and detoxification10,11,12,13,14. While the bulk of the work that has contributed to a better understanding of Spx function was conducted with the Gram-positive paradigm B. subtilis, evidence is now accumulating that Spx proteins have similar regulatory functions in many other bacteria and are critical for virulence of several Gram-positive pathogens10,12,14,15,16,17.

Previously, we reported the identification of two Spx proteins in S. mutans that we initially named SpxA and SpxB but have recently renamed SpxA1 and SpxA212,13,18 to avoid confusion with the streptococcal pyruvate oxidase SpxB19. Deletion of the S. mutans spxA1 resulted in increased sensitivity to oxidative stresses, a phenotype that was significantly enhanced in the double ΔspxA1/ΔspxA2 strain12. Transcriptional profiling of the Δspx strains and in vitro transcription assays confirmed that SpxA1 plays a primary role in directly activating transcription of well-known oxidative stress genes such as ahpC (alkyl hydroperoxidase), dpr (iron-binding protein) and sodA (superoxide dismutase)11,12,13,20. SpxA2, however, appears to serve as a backup for SpxA1 in the activation of oxidative stress genes while its primary function may be to control transcription of genes involved in cell envelope homeostasis12.

In this study, we used RNA deep sequencing (RNA-Seq) to identify changes in the transcriptome of S. mutans after a brief exposure to H2O2 and used functional genomics and physiological approaches to characterize pathways newly associated with peroxide stress survival. In its totality, the present study unraveled the scope of the peroxide stress regulon of S. mutans and identified several new Spx-regulated pathways that are important for peroxide survival.

Results

Overview of the H2O2 stress transcriptome of S. mutans UA159

Previously, we used microarrays to compare the transcriptome of S. mutans UA159 and ∆spx strains, though that study was performed in the absence of any stress12. In addition, we have used quantitative real-time PCR (qRT-PCR) to examine the transcriptional profile of a selected number of Spx-regulated genes in response to H2O2 stress11,13. In these investigations, we found that exposure of mid-log phase cultures to 0.5 mM H2O2 for 5 min was optimal for induction of known oxidative stress genes such as ahpCF, dpr and sodA 11,13. Here, we used the same parameter to uncover the peroxide stress regulon of S. mutans UA159 via RNA-Seq. As compared to cells grown in the absence of stress (control), approximately 7% of the S. mutans UA159 genome showed altered transcription with 100 genes upregulated and 39 genes downregulated after H2O2 stress (Table S1; P ≤ 0.05). The differently expressed genes were grouped into 8 functional categories (Fig. S1) with genes encoding amino acid biosynthesis, DNA metabolism, and hypothetical proteins highly represented in the list of upregulated genes whereas genes coding for hypothetical proteins accounted for more than 50% of the downregulated genes followed by genes involved in carbon metabolism (~20% of the total number of downregulated genes). A subset of the differentially expressed genes was selected and used for quantitative real-time (qRT) PCR analysis for validation of the microarray data and the results were consistent with the expression trends observed in the RNASeq analysis (Tables S3 and S4).

As expected, peroxide stress resulted in the strong and rapid induction of ROS scavenging (ahpCF, tpx, and sodA) and thiol homeostasis (gor, gst, trxA and trxB) genes (Table 1). Recently, we also performed a study focusing on previously uncharacterized genes that were positively regulated by SpxA111. Several of the genes from this previous study were also identified in the present RNA-Seq analysis, including smu143 (putative transcriptional regulator), smu144 (polypeptide deformylase), the sufABCD operon (Fe-S cluster assembly), smu929 (conserved hypothetical protein) and tehB (tellurite resistance protein) (Table 1). While confirming previously reported trends11,12,13,20,21, the RNA-Seq analysis also uncovered new and interesting trends suggesting that peroxide stress triggered important metabolic shifts, often in an SpxA1-dependent manner, as detailed below.

Table 1 Expression changes of selected genes in S. mutans UA159, ΔspxA1, or ΔspxA1ΔspxA2 following exposure to H2O2 stress.
Full size table

Other stress genes affected by peroxide stress

In addition to previously identified Spx-regulated genes, a number of other stress survival genes were induced by the H2O2 treatment (Table 1). Considering that DNA damage is an immediate consequence of an oxidizing environment, it was not surprising to observe that the DNA repair genes, exoA (smu1649), uvrA (smu1851) and mutY (smu1865), were induced by a minimum of 3-fold during peroxide stress. A number of general stress genes were also induced by peroxide stress, including the molecular chaperones clpL, groES-EL and hsp33, as well as the clpE ATPase. With the exception of hsp33, a redox-regulated chaperone that is activated by peroxide stress22, all other stress genes have been previously shown to be important for the acid stress survival of S. mutans 18,23,24,25 albeit their roles in peroxide survival have not been explored. Finally, the relA gene (smu2084, also known as rsh or rel) responsible for the production of the bifunctional (p)ppGpp synthetase/hydrolase was upregulated by 3.4-fold after peroxide stress. The RelA enzyme is responsible for activation of the stringent response, a conserved stress response mechanism to nutrient starvation26 that has been implicated in acid stress survival, biofilm formation and competence development of S. mutans 27,28. Notably, (p)ppGpp was previously shown to accumulate in S. mutans after H2O2 stress29.

Metabolic re-routing as a consequence of peroxide stress

In addition to genes involved in oxidative and general stresses, our RNA-Seq analysis identified a number of genes involved in energy-generation (mainly pyruvate metabolism and fate) as upregulated after H2O2 treatment (Table 1). For example, transcription of genes of the adhABCD operon (smu127- smu130), predicted to encode the acetoin dehydrogenase (AoDH) complex, increased between 4.1 to 7.3 fold. Immediately downstream and apparently co-transcribed with the adh gene cluster is the lplA gene (smu131) encoding a lipoate ligase that may serve as a scavenger of lipoic acid from the environment. Of note, lipoic acid is an enzymatic co-factor of dehydrogenases, including the AoDH complex. Following the same trend of the adh genes, transcription of lplA was increased by 6.9-fold after the stress stimulus. Another operon that relates to acetoin metabolism (smu1451-smu1452, aldB and alsS) was upregulated by 3.6- (aldB) and 2.9-fold (alsS). The alsS gene encodes for an α-acetolactate synthase (ALS) that serves as a pyruvate sink consuming two molecules of pyruvate to yield one of acetoin. The acetoin may then be utilized by AoDH to produce acetaldehyde. In addition to acetoin/acetaldehyde metabolism, the 3.6-fold increased expression of pyruvate formate lyase (pflA, smu1692) was another indicator of a metabolic shift toward mixed fermentation, as this enzyme is involved in formation of formate from pyruvate. Collectively, these transcriptional changes suggest that peroxide stress triggers important changes in the fermentation profile of S. mutans.

SpxA1 is the major transcriptional activator of the peroxide stress response

Above, we discussed the results of the transcriptional changes in the S. mutans UA159 (parent strain) transcriptome after H2O2 exposure. Here, we compare the transcriptional signatures of UA159 and spx mutant strains (∆spxA1, ∆spxA2, ∆spxA1∆spxA2) during peroxide stress (Table S2). A total of 230 genes were differentially expressed in ∆spxA1 when compared to UA159, with 75 genes upregulated and 155 genes downregulated. Remarkably, the transcriptome of ∆spxA2 after incubation with peroxide was nearly identical to that of the parent strain following stress, with only 14 genes upregulated and 5 genes downregulated. Despite the small number of differentially expressed genes in the ∆spxA2 strain, the ∆spxA1∆spxA2 double mutant strain showed an even greater number of differently expressed genes than the ∆spxA1 strain, with 170 upregulated genes and 264 downregulated. Among the 230 genes differently expressed in the ∆spxA1 strain when compared to UA159, 65% of these genes (n = 151) were also found in the comparison between the UA159 and ∆spxA1/∆spxA2 strains (Fig. S2). In many cases, when genes were differentially expressed in both ∆spxA1 and ∆spxA1∆spxA2, the difference in expression values as compared to UA159 were more extreme in the ∆spxA1∆spxA2 strain. This trend supports our earlier observations that SpxA2 serves a supporting or backup role for SpxA111,12,13.

One difference that was noticed when comparing expression trends of the ∆spx mutants to the parent strain was the expression of the peroxide-inducible genes. In most cases, the archetypal oxidative stress genes were among the most strongly repressed genes in the ∆spxA1 and ∆spxA1∆spxA2 strains (Table 1). This was particularly noticeable for ROS scavenger genes such as ahpCF, sodA and tpx, which were downregulated by a minimum of 15-fold and as much as 120-fold in the ∆spxA1 and ∆spxA1∆spxA2 strains. This trend was also observed for genes involved in pyruvate metabolism (adhABCD, aldB, and alsS), DNA repair (smx, uvrA and mutY), histidine biosynthesis and general stress (clpE, clpL and relA) (Table 1). Further, transcription of two additional DNA repair genes (smn and recA) and of the iron-binding protein dpr was also significantly lower in the ∆spxA1 and ∆spxA1∆spxA2 strains when compared to the parent strain (Table 1). The operons encoding two major iron transporters, feoABC and ftsABCD, were strongly induced in ∆spxA1 mutants. While the association of Spx as a repressor of the FeoABC system was previously reported11, the strong upregulation of the genes of the ftsABCD operon in the ∆spxA1∆spxA2 strain (15 to 21-fold higher than the parent strain) confirms that SpxA1 regulates iron trafficking. Finally, transcription of the ferric uptake regulator furR (smu593), a well-characterized transcriptional repressor of iron transporters30, was downregulated in ∆spxA1 and ∆spxA1∆spxA2 suggesting that Spx may control iron uptake via the FurR regulator.

A phenotype enhancement screen reveals hypersensitivity of double-mutant strains

To further investigate the role of the SpxA1-regulated genes identified in the RNA-Seq analysis, we used a markerless strategy to delete the adhD, alsS, hisC and lplA genes, as this is the first time that these genes have been associated with SpxA1 (Fig. 1A). Using this approach, the ∆adhD, ∆hisC and ∆lplA strains were readily obtained. We also created a markerless deletion in gdhA, encoding a glutamate dehydrogenase responsible for the deamination of glutamate to KG. After several unsuccessful attempts to obtain a markerless alsS mutant, we used a non-polar kanamycin cassette to isolate the ∆alsS strain. All mutant strains were viable and, with exception of ∆lplA that showed a slightly longer lag phase, grew as well as the parent strain in the absence of stress (data not shown).

Figure 1
figure 1

Phenotypic characterization of single and double (paired with ∆spxA1) mutant strains lacking gdhA, adhD, lplA, hisC and alsS. (A) Schematic representation of gdhA, adhABCD-lplA, his and aldB-alsS operons and flanking regions. Dashed lines indicate region of gene deletions. (B) H2O2 disc diffusion assay showing the diameters (in mm) of the zone of growth inhibition around discs saturated with 0.5% H2O2. (C) Growth inhibition of S. mutans UA159 and its derivatives by the peroxigenic S. sanguinis SK150 strain. (D) The S. sanguinis-S. mutans competition assay was repeated with catalase overlaid onto the S. sanguinis spot to inactivate H2O2. (*) compared to UA159, (+) - compared to ∆spxA1; p < 0.005 using Student’s t-test.

Full size image

We then tested the ability of the mutants to tolerate peroxide stress. In a H2O2 disc diffusion assay, the ∆adhD and ∆lplA mutants were significantly more sensitive than the parent strain with the ∆lplA strain displaying a zone of inhibition comparable to the highly sensitive ∆spxA1 strain (Fig. 1B). However, in a qualitative competition assay against a peroxigenic S. sanguinis strain, the differences between parent and all single mutant strains were subtle and unlikely to be biologically relevant (Fig. 1C). Nevertheless, it is not entirely surprising that single gene inactivation does not result in strong peroxide-sensitive phenotypes. While the physiological changes associated with those mutations may be part of the oxidative stress response of S. mutans, functional redundancy and genetic buffering may account for the lack of stronger phenotypes31. We and others have shown that genomic buffering could be overcome by introducing mutations of Spx-regulated genes into the ∆spxA1 background strain11,32. We utilized this same approach here by introducing a marked spxA1 deletion into the newly generated single mutants, and then compared the peroxide sensitivity of the double mutants to that of the ∆spxA1 single mutant. This time, with exception of ∆alsS∆spxA1, all other double mutants (∆adhD∆spxA1, ∆hisC∆spxA1, ∆lplA∆spxA1 and ∆gdhA∆spxA1) showed marked increases in peroxide sensitivity when compared to the single ∆spxA1 mutant strain in both disc diffusion and S. sanguinis antagonism assays (Fig. 1B,C).

The histidine biosynthesis operon is activated during peroxide stress

It was also interesting to note that transcription of eight of the thirteen genes comprising the histidine biosynthesis (his) operon were upregulated (~3 to 7-fold) during peroxide stress (Table 1). In Pseudomonas fluorescens, histidine is earmarked for α-ketoglutarate (KG) production during oxidative stress33. A TCA cycle intermediate, KG is a potent antioxidant as it spontaneously reacts with H2O2 to generate succinate and CO2 34,35. More specifically, some bacterial species are able to convert histidine to glutamate and ammonia in a multistep enzymatic reaction catalyzed by the genes of the hut (histidine utilization) operon36. Glutamate dehydrogenase is thereby provided with substrate to catalyze the deamination of glutamate to KG. To assess the possibility that production of KG via histidine-to-glutamate deamination could be part of the S. mutans oxidative stress response, we tested the effects of exogenously added histidine, glutamate or KG on growth of S. mutans in the presence of 0.4 mM H2O2. This concentration of H2O2 induced a prolonged lag phase and dramatically slowed growth of the UA159 parent strain (Fig. 2A) and completely abolished growth of the ∆spxA1 strain (Fig. 2B). The addition of 0.2 mM KG to the growth media partially restored growth of both strains exposed to H2O2, whereas 2 mM KG completely reversed the growth defect of both strains (Fig. 2A and B). On the other hand, addition of histidine (Fig. 2C) or glutamate (Fig. 2D) (up to 5 mM) failed to rescue growth of the two strains.

Figure 2
figure 2

Addition of α-ketoglutarate (KG) abolishes S. mutans growth defect caused by H2O2 stress. Strains UA159 (A,C) or ∆spxA1 (B,D) were grown in BHI broth in the absence or presence of 0.4 mM H2O2 with catalase (positive control), KG, histidine or glutamate added to the growth media using the concentration indicated in the figure.

Full size image

Despite this initial evidence that KG protects S. mutans from peroxide stress, we cannot rule out that the spontaneous reaction between KG and H2O2 is occurring outside the cell since a similar level of protection was observed with catalase that is not expected to enter the cellular compartment (Fig. 2A and B). Thus, we also took a reductionist approach and used a disc inhibition assay to test whether histidine or glutamate deprivation would affect the peroxide tolerance of S. mutans. In this case, sensitivity to H2O2 was significantly increased when either histidine or glutamate were omitted from the growth media but not when glycine (randomly chosen as a control) was omitted (Fig. 3A). Of note, the omission of histidine, glutamate or glycine from the growth media did not affect growth rates of S. mutans (data not shown). As a further measure to address the concern that KG might be detoxifying only extracellular H2O2, a KG pre-loading experiment was performed. S. mutans cells were inoculated in the presence or absence of 2 mM KG. Upon reaching early log phase, the cultures were washed to remove any extracellular KG, then resuspended in medium containing H2O2. The H2O2-exposed cultures demonstrated a considerable defect in growth yield as compared to controls that had not been exposed to the stress (Fig. 3B). However, the cultures that had been provided with KG prior to the stress showed a significant recovery of final growth yield.

Figure 3
figure 3

Physiological evidence that histidine, glutamate, and KG are associated with oxidative stress survival. (A) Depletion of histidine or glutamate increased sensitivity of S. mutans UA159 to H2O2. Sensitivity of S. mutans to 0.5% H2O2 delivered on paper discs was determined on agar plates composed of the chemically defined medium FMC, altered to omit histidine, glutamate, or glycine as indicated. The diameters of the zone of growth inhibition around the H2O2 discs were measured in mm. (*) indicates statistical significance as compared to FMC complete (white bar). (B) Pre-loading with KG offered protection from later exposure to H2O2. Growth of S. mutans ∆spxA1 was initiated in the absence or presence of 2 mM KG. Upon reaching early log phase (OD600 = 0.3, red horizontal line), cultures were washed twice in phosphate buffered saline to remove extracellular KG. The cells were then resuspended in BHI in the absence or presence of 0.4 mM H2O2. (*) indicates statistical significance when comparing strains exposed to H2O2 in the presence or absence of KG (blue and red lines). (*) p < 0.01 using Student’s t-test.

Full size image

The fermentative profile of peroxide-treated cells supports the altered transcriptome

As a preliminary step toward assessment of the impact of peroxide stress and SpxA1 on the fermentation profile of S. mutans, enzymatic assays were performed to measure metabolic end products stemming from the metabolism of pyruvate. Abundance of metabolites was measured from the culture supernatants of cells exposed to 0.5 mM H2O2 for 60 minutes, as compared to unstressed controls grown to the same optical density (OD600~0.6). Exposure to H2O2 resulted in minimal impact on lactic acid production by either the wild-type UA159 or ∆spxA1 strains, albeit the ∆spxA1 strain produced significantly more lactic acid (~25% more) than UA159 cultures (Fig. 4A). Interestingly, peroxide stress resulted in reduced ethanol production of approximately 10-fold in the parent strain as compared to unstressed UA159 cells, (Fig. 4B). On the other hand, the ∆spxA1 strain failed to produce large quantities of ethanol regardless of the growth condition. Similar results were observed for formate that showed a 4-fold reduction after H2O2 stress in the parent strain but not in the ∆spxA1 strain (Fig. 4C). In agreement with the H2O2-induced activation of the ALS genes (alsS and aldB) that convert pyruvate into the non-acidic end product acetoin, acetoin pools rose approximately 16-fold after peroxide treatment in the parent strain but not in the ∆spxA1 strain (Fig. 4D). Because the adhABCD operon is predicted to convert acetoin into acetaldehyde and is also induced by peroxide stress, we also attempted to quantify acetaldehyde using a biochemical approach. However, we were unable to detect acetaldehyde above background levels under any growth condition, possibly due to its highly volatile nature.

Figure 4
figure 4

Metabolic profiles of stressed S. mutans UA159 and ∆spxA1. Cultures grown to early log phase (OD600 = 0.35) were exposed to 0.5 mM H2O2 for 60 minutes, while unstressed controls were incubated for the same period of time. Culture supernatants were harvested by centrifugation and used to determine the concentrations of lactic acid (A), ethanol (B), formic acid (C), and acetoin (D). (*) indicates statistical significance as compared to unstressed UA159. (∆) indicates statistical significance as compared to the equivalent UA159 culture. (*) or (∆) p < 0.02 using Student’s t-test.

Full size image

Discussion

In this report, we showed that in addition to the archetypal oxidative stress genes such as those involved in ROS scavenging and thiol homeostasis, Spx mediates transcription of genes involved in DNA repair, pyruvate metabolism and amino acid biosynthesis. To our knowledge, the study presented here is the first to uncover the peroxide transcriptome of a streptococcal species. Nonetheless, two microarray studies whereby S. mutans cells were exposed to oxidative stress by aeration37 or excess (8.4%) oxygen20 support our results as both studies reported increased expression of several genes identified in our study (adhABCD, ahpCF, gor, lplA, sodA, tpx, among others). It was also interesting to note the remarkable overlap between the peroxide transcriptome of the present study and the transcriptome signature of S. mutans during acid stress38,39. For example, progressive environmental acidification to pH 5.5 via glycolysis led to increased expression of the histidine operon, adhABCD, aldB and alsS, as well as several antioxidant (ahpCF, sodA, tpx), thiol homeostasis (trxB), DNA repair (mutY and smn) and general stress (clpE, clpL and rel) genes38. Similarly, a study evaluating the global transcriptional profile of S. mutans acid-shocked to pH 5 for two hours also described the induction of several peroxide-induced genes including adhD, ahpCF, dpr and mutY 39.

Previously, we showed that SpxA1 plays a major role in iron homeostasis by serving as a transcriptional activator of genes encoding for iron-binding protein (dpr), tellurite resistance (tehB), Fe-S cluster assembly (sufABCD) and peptide deformylase11 as well as a repressor of the ferrous iron transporter system (feoABC). If one takes into account that iron is a catalyst of the Fenton reaction, the involvement of Spx as a negative regulator of iron transport is logical. In fact, we have previously shown that the ∆spxA1 strain is significantly more sensitive to the iron-dependent antibiotic streptonigrin11. Here, we expand the relationship between Spx and iron trafficking by confirming that transcription of feoABC was greater in the ∆spxA1 and ∆spxA1∆spxA2 strains (4 to 6-fold) but also by showing that the ferrichrome permease operon (ftsABCD) was strongly induced (15 to 21-fold) in the double mutant ∆spxA1∆spxA2 strain. While iron is an essential micronutrient, the increased expression of iron transporters in ∆spxA1 strains is likely to exacerbate the ROS stress imposed upon these strains that already have an impaired ability to activate antioxidant defenses and maintain iron homeostasis.

To begin to determine the significance of some of the genes identified in the RNASeq analysis in oxidative stress, we isolated and characterized isogenic deletion mutants lacking the alsS (∆alsS), adhD (∆adhD), lplA (∆lplA), hisC (∆hisC) and gdhA (∆gdhA) genes. While the ∆adhD and ∆lplA strains showed modest increases in sensitivity in H2O2 disc diffusion assays, our interpretation is that these single gene inactivations have little impact on the ability of S. mutans to cope with peroxide stress. It is not uncommon to observe a complete lack of detectable phenotypes from single gene deletions due to functional redundancy and, in this particular case, genetic interactions within the Spx regulon40. For example, when a single gene of the SpxA1 regulon is inactivated, a detectable phenotype may be masked by the robust response of other Spx-regulated genes. This “genetic buffering” phenomenon is evident in systematic gene deletion libraries whereby gene functions are rarely assigned based upon single gene deletions40. To overcome this limitation, we have used a “phenotype enhancement” approach32 whereby we paired each new mutation of an Spx-regulated gene with the spxA1 deletion mutant creating a panel of double mutant strains11. By decreasing overall expression of SpxA1-regulated genes in the ΔspxA1 strain background, we were able to unequivocally demonstrate that loss of adhD, gdhA, hisC and lplA markedly increased the peroxide stress sensitivity of the ΔspxA1 strain.

Another interesting finding from the RNA-Seq analysis was that genes involved in pyruvate metabolism (adhABCD-lplA, aldB, alsS, and pflA) followed the same trend as the oxidative stress genes, e.g. induced by H2O2 stress in an Spx-dependent manner. The lplA gene encodes a lipoate ligase, the function of which may be to scavenge lipoic acid for use as an antioxidant and also to lipoylate enzymes that require this co-factor41,42. For example, lipoylation is known to be essential for the activity of dehydrogenases and the genetic proximity of lplA to the adh operon (AoDH) suggests that lplA may serve to provide this co-factor to AoDH. The protein products of the aldB-alsS operon are responsible for the conversion of pyruvate to acetoin, while genes of the adhABCD operon (acetoin dehydrogenase, AoDH) are predicted to catalyze the conversion of acetoin to acetaldehyde but may also work in reverse, augmenting acetoin production. To obtain the first glimpses into the metabolic profile of S. mutans under peroxide stress, we compared the production of lactate, ethanol, formate and acetoin in cells subjected to H2O2 stress. When compared to cultures grown in the absence of stress, the production of ethanol and formate was drastically reduced (10- and 4-fold, respectively) in the wild-type strain UA159 after peroxide stress whereas acetoin levels increased by 16-fold after stress. In addition, there was a small increase in lactate production after stress but the difference observed was not statistically significant. While the pflA gene, coding for the pyruvate formate lyase, is induced after H2O2 stress, the reduction in formate production after H2O2 stress is not unexpected given that the PflA enzyme is highly sensitive to oxidation43. The increased production in acetoin during peroxide stress is in line with the increased transcription of the ALS genes (alsS and aldB) that convert pyruvate into the non-acidic end product acetoin. Interestingly, acetoin pools have been shown to increase during aerobic growth in S. mutans 44. Moreover, acetoin can spontaneously react with oxygen and H2O2 in the presence of Fe3+  45, even though it is not known if this reaction can occur in vivo. While these results indicate a linkage between acetoin production and oxidative stress that deserves further investigation, it should be noted that the very small amounts of acetoin produced (~0.01% of the total lactate produced under the same growth condition) may have little, if any, impact on cell physiology. Finally, ethanol, formate and acetoin pools remained largely unaltered in the ∆spxA1 strain after H2O2 indicating that SpxA1 regulation is critical for the metabolic alterations observed during stress. Collectively, these results validate the transcriptome data and open new doors for more detailed metabolic studies.

In addition to the genes involved in pyruvate metabolism, transcription of eight of the thirteen genes comprising the histidine biosynthesis operon was induced by H2O2, also in an Spx-dependent manner. There are several, not mutually exclusive, explanations for the apparent increase in histidine biosynthesis during peroxide stress. First, histidine is particularly prone to metal-catalyzed oxidation that results in the formation of 2-oxo-histidine – a biological marker for assessing protein oxidation during stress46. Thus, it is possible that S. mutans increases histidine biosynthesis during oxidative stress to simply restore cellular histidine pools. Second, histidine has been proposed to function as a scavenger of hydroxyl radical and singlet oxygen, but not of H2O2 and superoxide anion47. In vertebrates, histidine-containing dipeptides such as carnosine protect neural cells by acting as antioxidants48. Finally, histidine can be enzymatically converted to glutamate, which in turn can be deaminated to form KG, a potent antioxidant that spontaneously reacts with H2O2 to generate succinate and CO2 33,34. While S. mutans has an incomplete TCA cycle and cannot rely on this pathway for generation of KG, it can utilize the citrate metabolism pathway to generate KG from isocitrate via the aconitase enzyme (CitB). However, CitB is a cysteine-rich enzyme that becomes inactive during oxidative stress49, an indication that glutamate deamination may be the only source of KG for S. mutans during oxidative stress. Because histidine was shown to be earmarked for KG production in P. fluorescens during oxidative stress33,35, we tested the protective effects of KG, as well as histidine and glutamate, on growth and survival of S. mutans during oxidative stress. Our results clearly demonstrate that KG is a physiologically relevant antioxidant capable of attenuating the oxidative stress sensitivity of S. mutans. However, a homologous system to the histidine utilization (hut) operon responsible for the conversion of histidine to glutamate in some bacteria, was not identified in S. mutans 36. While S. mutans may use an alternative pathway to convert histidine into glutamate, the possible association of the histidine-to-glutamate-to-KG pathway in S. mutans oxidative stress response remains to be confirmed. Our future efforts will include direct quantifications of intracellular pools of amino acids and organic acids as well as the detection of 2-oxo-histidine levels.

Based on the evidences provided in this study, we propose to expand the Spx-regulated pathways that are important for oxidative stress survival to include activation of DNA repair enzymes, histidine biosynthesis and alterations in pyruvate metabolism (Fig. 5). While it remains to be further dissected, our RNA-Seq analysis also indicates the existence of a crosstalk between the Spx regulators and classic stress survival systems, which may include activation of the stringent response and of stress chaperones. While other transcriptional regulators such as PerR and SloR participate in the oxidative stress responses of S. mutans 50,51, our collective results indicate that SpxA1, and to some extent SpxA2, function as the master regulators responsible for launching a rapid, multi-strategy defense towards oxidative insults in S. mutans. Because the Spx regulation is conserved among Gram-positive bacteria, the findings presented here are likely to have broader implications.

Figure 5
figure 5

Antioxidant pathways of S. mutans regulated by Spx. Solid arrows indicate traits that have been validated through transcriptional and/or mutational analyses in the present (gray) or previous (black)11,12,13 studies. Dashed arrows denote pathways identified through transcriptome analysis that require further validation. (*) SpxA1 acts as a repressor of the feoABC and ftsABCD operons.

Full size image

Methods

Bacterial strains and growth conditions for RNA-seq analysis

The bacterial strains used in this study are listed in Table 2. S. mutans UA159 and its derivatives were routinely grown in brain heart infusion (BHI) at 37 °C in a 5% CO2 atmosphere or, in the case of oxidative stress sensitive Δspx strains, under anaerobic conditions (BBL Gaspack system, BD, Franklin Lakes, NJ). For RNA-Seq analysis, duplicate cultures were grown in BHI to an OD600 of 0.4, at which point control samples were harvested by centrifugation, while experimental samples were exposed to 0.5 mM H2O2 for 5 min before cell pellets were harvested by centrifugation and stored at −80 °C until use.

Table 2 Bacterial strains and plasmids used in this study.
Full size table

Construction of mutant strains

S. mutans strains bearing unmarked deletions in the gdhA (smu913), lplA (smu131), hisC (smu1273), or adhD (smu130) genes were created using natural genetic transformation techniques as described elsewhere52. Briefly, approximately 3-kb stretches of the DNA flanking the gene of interest at both the 5′ (primers 1 and 3) and 3′ (primers 4 and 5) ends were amplified using the primers listed in Table S5. For each gene, primers 3 and 4 encoded complementary sequences facilitating the annealing of the 5′ PCR product to the 3′ PCR product in a ligase-free PCR reaction. The desired overlap product was then amplified using nested primers (primers 2 and 6). The gene of interest was thereby absent in this final PCR amplicon, which was used for transformation of a highly competent population of S. mutans UA159. Briefly, S. mutans was grown overnight in peptide-free chemically defined medium (CDM)53, the cells were collected by centrifugation, washed twice and resuspended in PBS, and used to inoculate 500 μl peptide-free CDM broth at a dilution of 1:20. The cultures were grown to an OD600 of 0.1 when 1 μM of the competence pheromone ComX-inducing peptide (XIP) (GenScript, Piscataway, NJ) and 0.4 μg of the PCR amplicon were added to the culture. The cultures were incubated for an additional 3 hr before plating on BHI agar. After 48 hr incubation, colonies were screened by PCR using the primers listed in Table S3 to ensure that a double recombination event resulted in deletion of the target gene. The selected clone was confirmed by DNA sequencing of the PCR product. The alsS (smu1492) mutant strain was created by replacing the coding region of the gene with a non-polar kanamycin resistance cassette using a PCR ligation mutagenesis approach54. Briefly, PCR fragments flanking alsS were ligated to the kanamycin cassette and this ligation mix used to transform S. mutans UA159. Double mutants were obtained by amplifying the mutated spxA1 region of the previously constructed ΔspxA1 strain (spectinomycin-resistant, SpcR); this PCR product was then used to transform the newly generated single mutant strains.

Growth and stress survival assays

To generate growth curves, strains were grown overnight under anaerobic conditions and diluted 1:20 in BHI or BHI containing 0.4 mM H2O2. The protective effects of α-ketoglutarate (KG), catalase, or selected amino acids on growth in the presence of H2O2 were tested by adding increasing concentrations of each reagent to the growth media. In all cases, cultures were incubated at 37 °C in a 5% CO2 atmosphere and the OD600 recorded at selected intervals. To test the sensitivity of S. mutans UA159 and its derivatives to H2O2 in disc diffusion assays, a uniform layer of exponentially-grown cells was spread using a sterile swab onto agar plates made of the chemically defined medium FMC49, or FMC lacking glycine, histidine, or glutamate. After spreading the bacterial cultures, Whatman filter paper discs (6 mm diameter) saturated with 20 μl of 0.5% H2O2 solution were placed on the agar and the diameter of the zone of growth inhibition measured after 24 hr incubation at 37 °C in 5% CO2. To test the ability of pre-loaded KG to protect S. mutans, BHI inocula were initiated in the presence or absence of 2 mM KG. Cells were grown to early log phase (OD600 = 0.3), then washed twice in phosphate-buffered saline to remove extracellular KG. These washed culture pellets were then resuspended in fresh BHI, either with or without 0.4 mM H2O2. Final growth yields were measured after 24 hr incubation. All stress survival assays were performed with quadruplicate culture replicates.

Competition on solid media

Growth inhibition of S. mutans by production of H2O2 by mitis-group streptococci was performed as previously described11. Briefly, overnight cultures of S. sanguinis SK150 were normalized to OD600 of 0.5 and 8 μl aliquots spotted on BHI agar plates. After 16 hr incubation at 37 °C, overnight cultures of S. mutans UA159 and its derivatives were normalized to OD600 of 0.5 and 8 μl aliquots spotted next to the S. sanguinis spot. Plates were incubated for an additional 16 hr before visualizing the ability of the S. mutans strains to grow in proximity of the H2O2-generating S. sanguinis. To ensure that the growth inhibition was due to H2O2 production, a control condition included the addition of catalase directly on top of the S. sanguinis spot. Competition assays were performed with quadruplicate culture replicates.

Metabolite profiling

Abundance of metabolic end products (lactate, ethanol, and formic acid) was measured using Megazyme enzymatic kits (Megazyme International, Wicklow, Ireland). Acetoin production was measured using a Voges-Proskauer test as described previously55. Cultures were grown in BHI medium to early log phase (OD600 = 0.35) and split into two equal volumes that were harvested by centrifugation after 60 additional minutes of growth: (A) unstressed control, (B) exposed to 0.5 mM H2O2. Culture supernatants were deproteinized by addition of 4 M perchloric acid, then neutralized with potassium hydroxide. The supernatants were then stored at −80 °C until detection of metabolic products according to the manufacturer’s protocol using a 96-well plate format. Results were normalized to colony-forming units recovered for each sample. All metabolite assays were performed with quadruplicate culture replicates.

RNA analysis

Total RNA was isolated from homogenized S. mutans cell lysates by repeated hot acid-phenol:chloroform extractions as previously described56. The RNA was precipitated with ice-cold isopropanol and 3 M sodium acetate (pH 5) at 4 °C before RNA pellets were dissolved in nuclease-free H2O and treated with DNase I (Ambion, Carlsbad, CA) for 30 minutes at 37 °C. Then, 10 μg RNA aliquots were subjected to a second DNase I treatment using the DNA-free kit (Ambion). RNA concentrations were determined with the NanoDrop 1000 spectrophotometer (NanoDrop, Wilmington, DE) and RNA quality assessed with the Agilent Bioanalyzer (Agilent, Santa Clara, CA). RNA deep sequencing (RNA-Seq) was performed at the University of Rochester Genomics Research Center (UR-GRC) using the Illumina platform. The TruSeq RNA Sample Preparation Kit V2 (Illumina, San Diego, CA) was used for next generation sequencing library construction following the instructions from the manufacturer. The libraries were hybridized to the Illumina single end flow cell and amplified using cBot (Illumina) at a concentration of 8pM per lane. Raw reads were demultiplexed using configurebcl2fastq.pl version 1.8.4, and quality filtering and adapter removal performed using Trimmomatic version 0.32. Processed/cleaned reads were mapped to the S. mutans UA159 genome with STAR_2.4.2a. Initial differential expression analysis was performed using Cufflinks version 2.0.2, and DESeq. 2-1.10.1 was used for data normalization and differential expression analysis with an adjusted p-value threshold of 0.0557. Additional details on cDNA library construction, amplification and data analysis can be found at the UR-GRC website (https://www.urmc.rochester.edu/research/for-researchers/shared-resource-laboratories-facilities/laboratories/rochester-genomics-center.aspx). Gene expression trends as determined by RNA-Seq analysis were validated by qRT-PCR. Reverse transcription and qRT-PCR were carried out on triplicate samples of S. mutans UA159 RNA (unstressed or exposed to H2O2 as described for the RNA-Seq analysis) according to protocols described elsewhere56, using gene-specific primers listed in Table S6. Student’s t test was performed to verify significance of the qRT-PCR results. Gene expression data have been deposited in the NCBI Gene Expression Omnibus (GEO) database (www.ncbi.nlm.nih.gov/geo) under GEO Series Accession number GSE98526.