Regulatory circuits controlling Spx levels in Streptococcus mutans

Summary

Spx is a major regulator of stress responses in Firmicutes. In Streptococcus mutans, two Spx homologues, SpxA1 and SpxA2, were identified as mediators of oxidative stress responses but the regulatory circuits controlling their levels and activity are presently unknown. Comparison of SpxA1 and SpxA2 protein sequences revealed differences at the C-terminal end, with SpxA1 containing an unusual number of acidic residues. Here, we showed that a GFP reporter becomes unstable when fused to the last 10 amino acids of SpxA2 but remained stable when fused to the C-terminal acidic tail of SpxA1. Inactivation of clpP or simultaneous inactivation of clpC and clpE stabilized the GFP::SpxA2_tail fusion protein. Addition of acidic amino acids to the GFP::SpxA2_tail chimera stabilized GFP while deletion of the acidic residues destabilized GFP::SpxA1_tail. Promoter reporter fusions revealed that spxA1 transcription is co-repressed by the metalloregulators PerR and SloR while spxA2 transcription is largely dependent on the envelope stress regulator LiaFSR. In agreement with spxA2 being part of the LiaR regulon, SpxA2 was found to be critical for the growth of S. mutans under envelope stress conditions. Finally, we showed that redox-sensing is essential for SpxA1-dependent activation of oxidative stress responses but dispensable for SpxA2-mediated envelope stress responses.

Graphical Abstract

Abbreviated Summary

Streptococcus mutans encodes two Spx homologues. SpxA1 is resistant to ClpP proteolysis due to the presence of C-terminal acidic residues, transcriptionally regulated by the metalloregulators PerR and SloR, and relies on its redox sensing motif to activate an oxidative stress response. SpxA2 levels are controlled by the LiaFSR signal transduction system and posttranslationally by ClpP proteolysis. In addition, SpxA2 was critical for the growth of S. mutans under envelope stress conditions regardless of its redox status.

Introduction

An occasional inhabitant of the dental plaque, Streptococcus mutans is regarded as the major bacterial agent responsible for the initiation and progression of dental caries (Bowen, Burne, Wu, & Koo, 2018; K. Nakano, Nomura, Matsumoto, & Ooshima, 2010). Upon oral colonization, S. mutans metabolizes dietary carbohydrates to produce an acidic biofilm matrix that challenges the survival of less aciduric bacteria, which includes members of the mitis streptococci group such as S. sanguinis and S. gordonii (Lemos & Burne, 2008). Epidemiological studies have shown that high numbers of S. mutans in saliva and plaque are closely associated with active caries whereas high numbers of mitis streptococci are normally associated with oral health (Banas & Drake, 2018; Mira, Simon-Soro, & Curtis, 2017). It follows that S. sanguinis, S. gordonii and a few other streptococci associated with oral health are net producers of H₂O₂ that is inhibitory to the growth of S. mutans (Chen, Chakraborty, Zou, Burne, & Zeng, 2019; Zhu & Kreth, 2010). In addition to peroxigenic oral competitors, reactive oxygen species (ROS) in dental biofilms can derive from metabolic reduction of oxygen by the resident oral flora, including S. mutans’ own metabolism, and from the use of oral health and tooth bleaching products (Marquis, 1995).

In gram positive bacteria, activation of oxidative stress responses is largely dependent on the activity of the global transcription regulator Spx (Galvão et al., 2015; Kajfasz et al., 2012; Kajfasz et al., 2010; Whiteley, Ruhland, Edrozo, & Reniere, 2017; Zuber, 2004). The spx gene was first discovered in the context of reverting phenotypes associated with clpP and clpX mutants in the soil organism Bacillus subtilis, thus given the name Spx for suppressor of ClpP and ClpX phenotypes (M. M. Nakano, Hajarizadeh, Zhu, & Zuber, 2001). Subsequent studies revealed that B. subtilis Spx is subjected to ClpXP proteolytic control in vivo, and that toxic accumulation of Spx was responsible for many of the phenotypes associated with the clpP and clpX deletion strains (Garg, Kommineni, Henslee, Zhang, & Zuber, 2009; S. Nakano, Zheng, Nakano, & Zuber, 2002; Zhang & Zuber, 2007). The products of the clpP and clpX genes code for a serine peptidase and an ATPase, respectively. These subunits interact to form a barrel-shaped complex responsible for recognizing proteins that are captured and then subjected to ClpP degradation (T. A. Baker & Sauer, 2012). Spx proteolysis via ClpXP in B. subtilis was later shown to be greatly enhanced by the YjbH adapter protein, which interacts with the Spx C-terminal tail to facilitate ClpXP-mediated proteolysis (Engman & von Wachenfeldt, 2015; Frees, Savijoki, Varmanen, & Ingmer, 2007). More contemporary studies confirmed previous observations (S. Nakano et al., 2002) that the ClpCP protease also participates in Spx turnover, particularly under conditions that appear to destabilize ClpX (Rojas-Tapias & Helmann, 2019).

Previously, we identified two Spx homologues in S. mutans and showed that inactivation of either one of the Spx-encoding genes alleviated phenotypes of clpP and clpX mutants (Kajfasz et al., 2010). These genes were initially named spxA and spxB but later renamed spxA1 and spxA2, respectively, to avoid confusion with a pyruvate oxidase found in other streptococci that was also called spxB (Regev-Yochay, Trzcinski, Thompson, Malley, & Lipsitch, 2006; Xu, Itzek, & Kreth, 2014). Transcriptional and phenotypic characterizations of the S. mutans ΔspxA1 strain revealed that SpxA1 plays a major role in mediating oxidative stress survival by serving as the primary transcriptional activator of genes involved in cell detoxification (e.g., ahpCF, sodA, tpx), DNA repair (e.g., smn, smxA), and thiol homeostasis (e.g., trxA, trxB, gor) (Galvao et al., 2015; Kajfasz, Ganguly, Hardin, Abranches, & Lemos, 2017; Kajfasz et al., 2015). Not surprisingly, the ∆spxA1 strain was highly sensitive to oxidative stresses (Kajfasz et al., 2010). Though the ΔspxA2 strain did not show an oxidative stress tolerance phenotype, the ΔspxA1ΔspxA2 double mutant strain was hypersensitive to oxidative stresses when compared to its ΔspxA1 counterpart (Kajfasz et al., 2010; Kajfasz et al., 2015). In addition to what appears to be a secondary/ back-up role in the activation of oxidative stress responses, global transcriptional analysis suggested that SpxA2 may participate in the regulation of processes associated with cell envelope homeostasis (Kajfasz et al., 2010; Kajfasz et al., 2015). After the initial discovery of two Spx-like regulators in S. mutans, several studies revealed the presence of two Spx paralogs in other streptococci as well as in Listeria monocytogenes and Bacillus anthracis (Barendt, Birch, Mbengi, & Zuber, 2016; Barendt et al., 2013; Chen, Ge, Wang, Patel, & Xu, 2012; Port, Cusumano, Tumminello, & Caparon, 2017; Turlan, Prudhomme, Fichant, Martin, & Gutierrez, 2009; Zheng et al., 2014). In addition to the conserved role in activation of oxidative stress responses, Spx regulation has been more recently linked to activation of other types of stress, including antibiotic, osmotic and cell envelope stresses (Baek et al., 2015; Jousselin, Kelley, Barras, Lew, & Renzoni, 2013; Nilsson, Jakobsen, Givskov, Twetman, & Tolker-Nielsen, 2019; Renzoni et al., 2011; Rojas-Tapias & Helmann, 2018a; Villanueva et al., 2016). It is noteworthy that while the Spx protein of B. subtilis has been implicated in the regulation of multiple types of stress, including oxidative, heat and envelope stresses (Rojas-Tapias & Helmann, 2018a; Schafer et al., 2019; Zuber, 2004), the B. subtilis genome encodes a Spx paralogue, MgsR, that is part of the σ^B regulon and appears to participate in the ethanol stress response (Reder et al., 2008).

In the bacterial species studied to date, Spx levels were shown to be under ClpP proteolytic control (Chan, Garg, Lin, & Zuber, 2012; Chan, Hahn, & Zuber, 2014; Engman, Rogstam, Frees, Ingmer, & von Wachenfeldt, 2012; Garg et al., 2009; Kajfasz et al., 2009; S. Nakano et al., 2002; Pamp, Frees, Engelmann, Hecker, & Ingmer, 2006). The core genome of S. mutans encodes one copy of the clpP peptidase and five Clp ATPase-encoding genes named clpB, clpC, clpE, clpL and clpX. As the products of clpB and clpL lack the recognition tripeptide that mediates interaction with ClpP, only three of these S. mutans Clp ATPases interact with ClpP to form the ClpCP, ClpEP and ClpXP complexes (Frees et al., 2007). Using antibodies raised against the B. subtilis Spx, we showed that Spx accumulates in ΔclpP and ΔclpX strains of S. mutans but not in ΔclpC or ΔclpE strains (Kajfasz et al., 2009). However, due to the high degree of conservation at the protein level and nearly identical molecular weight of SpxA1 and SpxA2, it was not possible to discern if only one or both Spx proteins were accumulating in the ΔclpP and ΔclpX strains at that time. In addition to posttranscriptional control, the levels of Spx may also be regulated at the transcriptional level. In B. subtilis, spx is transcribed through five different promoters recognized by either the housekeeping sigma factor (σ^A) or by the alternative sigma factors σ^B, σ^M and σ^W. In addition, two transcriptional regulators, PerR and YodB, repress transcription from the major σ^A promoter such that spx promoters may be turned on or off in response to different environment inputs (Leelakriangsak, Kobayashi, & Zuber, 2007; Rojas-Tapias & Helmann, 2018a).

With the exception of B. subtilis, the transcriptional and posttranscriptional mechanisms controlling Spx levels have not been thoroughly investigated in other bacteria, particularly in organisms such as S. mutans that encodes two Spx homologs. In this study, we sought to identify the regulatory networks controlling the cellular levels and activity of the S. mutans SpxA1 and SpxA2 regulators. Using molecular genetics approaches, we showed that the cellular levels of SpxA1 and SpxA2 are controlled by distinct mechanisms, which includes the participation of different transcriptional regulators governing transcription of each gene and the unexpected observation that ClpP-mediated posttranslational control is restricted to SpxA2. In addition to the characterization of the regulatory circuits controlling Spx levels, we also demonstrate that the primary function of SpxA2 in S. mutans is to mediate an envelope stress response.

Results

Streptococcal Spx homologues differentiate mostly at their C-terminal end

Spx proteins belong to the ArsC_Spx sub-family (cd03032), which is divided into three sub-groups based on the presence of a redox-sensing CXXC motif in the amino terminus and a conserved glycine residue, crucial for the interaction of Spx with the RNAP α-CTD domain, located around the center of the protein sequence (Turlan et al., 2009). Two sub-groups in this sub-family are very closely associated consisting of 131 to 137 amino acid proteins with the conserved glycine residue always located at position 52 (Gly⁵²) (M. M. Nakano et al., 2010). The other sub-group of the ArsC family contains relatively shorter proteins (ranging from 117 to 120 amino acids) with a glycine residue at the 50^th amino acid position or absent in some cases. Previously, we showed that both S. mutans SpxA1 and SpxA2 have a Gly⁵² residue, and therefore belong in the first two sub-groups, whereas a third Spx paralog with a Gly⁵⁰ residue failed to suppress phenotypes associated with clpP gene inactivation and therefore was not considered a true Spx protein (Kajfasz et al., 2010). Here, we focused our analysis on the two groups that contain the conserved Gly⁵² motif, often termed as SpxA1 and SpxA2 (Chen et al., 2012; Turlan et al., 2009). The length of SpxA1 proteins among streptococcal species range from 133 to 137 amino acids, whereas all streptococcal SpxA2 proteins are 132 amino acids long (Fig. 1). The difference in the length of streptococcal SpxA1 proteins is largely due to the presence of acidic amino acid residues at the protein C-terminal end in some species, including S. mutans and Streptococcus agalactiae (Fig. 1 and Fig. S1). On the other hand, the last four residues (RAAL) of all streptococcal SpxA2 proteins sequenced to date are 100% conserved (Fig. 1).

Figure 1:

ClustalW alignment of SpxA1 (A) and SpxA2 (B) proteins from selected gram-positive bacteria. SpxA1 protein length varies from 131 to 137 amino acids whereas SpxA2 from all streptococcal species are 132 amino acids long. Identical residues are shown in dark shades and similar residues in light shades. A consensus is shown below the sequences. Black boxes indicate the conserved CXXC motif and glycine 52 residue. Red boxes indicate acidic residues present within the last 10 amino acids of each protein.

SpxA2 but not SpxA1 is subjected to ClpP proteolysis

In B. subtilis, Spx levels, herein Spx_Bsu, are kept in check by cooperative degradation by the ClpXP and ClpCP proteolytic systems (Chan et al., 2012; Engman et al., 2012; S. Nakano et al., 2002; Rojas-Tapias & Helmann, 2018b). Though both ClpCP and ClpXP can degrade Spx_Bsu, early evidence indicated that regulation of Spx_Bsu levels in vivo is primarily mediated by ClpXP (S. Nakano et al., 2002). Previously, we showed that Spx accumulated in both ΔclpP and ΔclpX strains of S. mutans, though at the time we were unable to determine if it was SpxA1, SpxA2, or both proteins that were accumulating in those mutant strains (Kajfasz et al., 2009). Attempts to monitor individual Spx protein stability in ΔclpPΔspxA1 or ΔclpPΔspxA2 strains by Western blotting were also unsuccessful, either because the detectable Spx band in the ΔclpP and ΔclpX cell extracts was a result of the accumulation of both proteins, or because inactivation of one spx gene affected transcriptional or posttranslational mechanisms controlling the levels of the other Spx protein (Kajfasz et al., 2009). A logical next step was to demonstrate degradation of recombinant S. mutans SpxA1 and SpxA2 proteins, herein SpxA1_Smu and SpxA2_Smu, by ClpXP in a reconstituted in vitro system (S. Nakano et al., 2002). However, under the conditions tested, both rSpxA1_Smu and rSpxA2_Smu remained fairly stable for 2 hours (Fig. S2). After that, both proteins started to degrade, though this degradation also occurred in the absence of purified ClpP (data not shown). In B. subtilis, the in vitro degradation of Spx_Bsu by ClpXP was very inefficient in the absence of the YjbH adaptor protein (Chan et al., 2014; Garg et al., 2009). Because streptococcal genomes do not encode YjbH homologs, we suspect that the S. mutans ClpXP system depends on the presence of a yet-to-be-identified adaptor protein for efficient degradation of Spx.

Previous studies from the Zuber lab showed that a motif comprised of the last 12 C-terminal amino acid residues of Spx_Bsu is recognized by the YjbH adaptor protein, facilitating Spx degradation by ClpXP (Chan et al., 2012). In addition, the addition of two aspartate residues to the C-terminal end was shown to render Spx_Bsu resistant to proteolysis. As mentioned above, SpxA1_Smu and SpxA2_Smu differ significantly at their C-terminal end, with SpxA1_Smu displaying 5 acidic amino acid residues within the last 10 amino acids (Fig. 1). Considering that the addition of acidic residues to the protein C-terminus stabilized Spx_Bsu (Chan et al., 2012), we suspected that SpxA2_Smu but not SpxA1_Smu is naturally subjected to ClpP proteolysis. To circumvent the hurdles and shortcomings of in vivo detection and in vitro proteolysis reconstitution, we engineered recombinant proteins containing the last 10 amino acids of either SpxA1_Smu or SpxA2_Smu fused to an otherwise stable green florescence protein (GFP) (Jana, Tao, & Biswas, 2016), and then used these GFP::Spx_tail chimera to investigate if the C-terminal of SpxA1_Smu or SpxA2_Smu served as recognition sites for ClpP-dependent proteolysis upon its expression in S. mutans (Fig. 2A). The stability of GFP over time was monitored in the S. mutans UA159 (parent) or ΔclpP strains by measuring fluorescence decay and Western blot before and after protein synthesis was halted with chloramphenicol. When expressed in the UA159 strain, the GFP protein alone remained stable for the duration of the experiment (Fig. 2B–C). Also in the UA159 strain, the GFP::SpxA1_tail fusion protein remained stable for up to 12 hours whereas GFP::SpxA2_tail was degraded by ∼ 80% within the initial 2 hours (Fig. 2B–C). Importantly, the GFP::SpxA2_tail fusion protein was stabilized in the ΔclpP strain strongly indicating that SpxA2_Smu is under ClpP proteolytic control (Fig. 2B–C).

Figure 2.

Stability of GFP::SpxA1_tail and GFP::SpxA2_tail fusion proteins in S. mutans. (A) Graphical representation of GFP fused to the last 10 C-terminal amino acids of either SpxA1_Smu or SpxA2_Smu. (B) Fluorescence decay of GFP::SpxA1_tail and GFP::SpxA2_tail expressed in UA159 (parent) or ΔclpP strains after addition of chloramphenicol. Asterisks indicate time points showing statistically significant differences (p ≤ 0.01, one-way ANOVA) in decay of GFP expression in the SpxA2_tail construct when hosted in UA159 compared to the ΔclpP strain. (C) Western blot analysis of UA159 or ΔclpP expressing GFP::SpxA1_tail and GFP::SpxA2_tail probed with anti-GFP polyclonal antibody. The images shown are representative of 3 or more independent experiments.

To investigate whether C-terminal acidic residues can generally protect Spx against ClpP-mediated proteolysis, we next added 2 aspartate residues to the last 10 amino acids of the SpxA2_Smu C-terminus, generating the GFP::SpxA2_DDtail fusion protein. As anticipated, addition of aspartate residues stabilized the GFP:SpxA2_DDtail fusion protein to levels that were almost identical to those described for the GFP:SpxA2_tail in the ΔclpP strain (Fig. 3A and 3C). To further demonstrate the importance of the C-terminal residues for Spx stabilization, we engineered GFP::SpxA1_tail fusion proteins with either 5 or 7 of the last C-terminal amino acids removed such that the acidic amino acids shown to stabilize SpxA1_Smu were absent in these fusion proteins. In this case, the SpxA1_Smu C-terminal residues lacking the last 5 (GFP::SpxA1_-5tail) or 7 (GFP::SpxA1_-7tail) acidic amino acids but still containing 10 C-terminal amino acids of SpxA1_Smu was rapidly degraded (Fig. 3B–C), resembling the degradation kinetics observed for the SpxA2_Smu C-terminal tail in the UA159 background that is shown in Figure 2. Collectively, these results reveal that the levels of SpxA2_Smu are maintained by ClpP proteolysis, and that acidic residues at the SpxA1_Smu C-terminus confers protection against proteolysis.

Figure 3.

Stability of GFP::Spx_tail fusion proteins with acidic residues added to the C-terminus of SpxA2 (SpxA2_DDtail) or with acidic residues deleted from SpxA1 (SpxA1_-5tail and SpxA1_-7tail). (A) Fluorescence decay of GFP::SpxA2_tail or GFP::SpxA2_DDtail when expressed in UA159 or ΔclpP host strains. Asterisks indicate time points showing statistically significant differences (p ≤ 0.01, one-way ANOVA) in GFP::SpxA2_tail decay compared to GFP::SpxA2_DDtail when expressed in UA159. (B) Fluorescence decay of GFP::SpxA1_tail, GFP::SpxA1_-5tail or GFP::SpxA1_-7tail expressed in UA159. Asterisks indicate time points showing statistically significant differences (p ≤ 0.01, one-way ANOVA) in decay of the GFP::SpxA1_tail construct as compared to GFP::SpxA1_-5tail or GFP::SpxA1_-7tail. (C) Western blot analysis of S. mutans UA159 expressing GFP::SpxA2_DDtail, GFP::SpxA1_-5tail or GFP::SpxA1_-7tail probed with anti-GFP polyclonal antibody. The images shown are representative of 3 or more independent experiments.

Both Spx proteins of S. pneumoniae are subjected to ClpP degradation

As mentioned above, the C-terminal acidic residues of SpxA1_Smu are not widespread in other Spx proteins, seemingly limited to S. agalactiae and a small number of poorly characterized oral streptococcal species (Fig. S1). For a broader perspective of the posttranslational regulatory mechanisms controlling Spx levels in streptococci, we engineered GFP::Spx_tail fusion proteins containing the last 10 C-terminal amino acids of the S. pneumoniae SpxA1 (SpxA1_Spn) and SpxA2 (SpxA2_Spn) proteins, neither of which possess acidic residues at their C-terminal end (Fig. 1). Here, both GFP::SPNSpxA1_tail and GFP::SPNSpxA2_tail fusion proteins were efficiently degraded in the S. mutans UA159 background and stabilized in the ΔclpP strain (Fig. 4). Thus, the differences in the posttranslational regulation of the S. mutans Spx proteins appears to be the exception and not the rule in the Streptococcus genus. We predict that SpxA1 homologs from S. agalactiae and the few oral streptococci that contain multiple acidic residues (Fig. S1) are resistant to ClpP proteolysis.

Figure 4.

Stability of GFP::SPNSpx fusion proteins. Fluorescence decay of GFP::SPNSpxA1_tail and GFP::SPNSpxA2_tail expressed in UA159 or ΔclpP strains. Asterisks indicate time points showing statistically significant differences (p ≤ 0.01, one-way ANOVA) when the corresponding fusion protein is expressed in S. mutans UA159 or ΔclpP strains; a single asterisk indicates comparisons with the GFP::SPNSpxA1_tail construct, while two asterisks indicate comparisons with the GFP::SPNSpxA2_tail construct.

Simultaneous inactivation of clpC and clpE stabilized the GFP::SpxA2_tail fusion protein

Next, we sought to identify the Clp ATPase(s) that interact with ClpP to degrade SpxA2_Smu. First, we assessed the stability of the GFP::SpxA2_tail fusion in single clp ATPase deletion strains (ΔclpC, ΔclpE and ΔclpX) generated in a previous study (Kajfasz et al., 2009). While our initial results indicate that SpxA2_Smu is primarily under ClpXP proteolytic control (Kajfasz et al., 2009), single inactivation of clpX (or clpC) did not increase stability of the GFP::SpxA2_tail (Fig. 5). Surprisingly, inactivation of clpE significantly increased GFP::SpxA2_tail stability (Fig. 5). Due to the precedent of functional redundancy among different ClpP-Clp ATPase systems, including evidence that Spx_Bsu levels are controlled by both ClpCP and ClpXP (S. Nakano et al., 2002; Rojas-Tapias & Helmann, 2019), we next examined the stability of GFP::SpxA2_tail in the 3 possible clp ATPase double mutant combinations. When compared to the partial stability observed in the ΔclpE strain, the GFP::SpxA2_tail was further stabilized in the ΔclpXΔclpE strain, and almost completely stabilized in the ΔclpEΔclpC strain (Fig. 5). The simultaneous inactivation of clpC and clpX, the two primary Clp ATPases involved in degradation of Spx_Bsu (S. Nakano et al., 2002; Rojas-Tapias & Helmann, 2018b, 2019), did not increase GFP::SpxA2_tail stability.

Figure 5.

Fluorescence decay of GFP::SpxA2_tail fusion proteins expressed in S. mutans UA159 and a panel of clp single and double mutant strains. Time points showing statistically significant differences (p ≤ 0.01, one-way ANOVA) when GFP::SpxA2_tail expressed in S. mutans UA159 is compared to GFP::SpxA2_tail expressed in ΔclpP (@), ΔclpE (*), ΔclpEΔclpC (#), or ΔclpXΔclpE (&) are indicated in the figure.

Transcriptional analyses of selected SpxA1- and SpxA2-regulated genes strengthen the GFP reporter studies

To obtain additional evidence that ClpP-mediated degration of SpxA2_Smu but not of SpxA1_Smu occurs in its native context, we used quantitative real-time PCR to compare the expression profile of SpxA1- and SpxA2-regulated genes in the UA159, ∆clpE and ∆clpP strains as a readout of SpxA1 and SpxA2 stability in vivo. Transcription of smu1412c, a gene coding for a hypothetical protein previously shown to be under SpxA2 positive regulation (Kajfasz et al., 2010), was significantly inceased (~ 3 fold) in the ∆clpP strain (Fig. 6A), a finding that supports the GFP::SpxA2_tail protein chimera studies. This increase was not observed in the ∆clpE strain, possibly due to the redundant role of different ClpP systems (ClpEP, ClpCP and ClpXP) in SpxA2 degradation. On the other hand, transcription of sodA, a direct target of SpxA1 regulation (Kajfasz et al., 2017; Kajfasz et al., 2010; Kajfasz et al., 2015) was not altered in the ∆clpP and ∆clpE strains when compared to UA159 providing further evidence that SpxA1_Smu is not a naturally subjected to ClpP proteolysis (Fig. 6B).

Figure 6.

qRT PCR analysis of (A) SpxA1-regulated sodA or (B) SpxA2-regulated smu.11412c genes in different the ΔclpP and ΔclpE mutant background. (*) p< 0.05

Transcriptional regulation of spxA1 and spxA2

While Spx proteins were initially thought to be primarily regulated at the posttranslational level and functionally activated by means of a redox-sensing CXXX motif (Schafer & Turgay, 2019), previous studies have shown that spx genes may also be under transcriptional control (Pamp et al., 2006; Rojas-Tapias & Helmann, 2018a; Shankar, Mohapatra, Biswas, & Biswas, 2015). Using RACE-PCR, a single transcription initiation site was identified 27 nucleotides 5′ to the translational initiation site of spxA1_smu that mapped to a putative σ^A-type promoter (TAGCCA-N₁₇-TATAAT) (Fig. 7A). Similarly, a single transcriptional start also located 27 nucleotides from the start codon and a σ^A-type promoter (TCTTTA-N₁₆-TAAGAT) were identified upstream of the spxA2 start codon (Fig. 7A). To determine the transcriptional profile of spxA1_smu and spxA2_smu, DNA fragments containing the promoter region of spxA1_smu (P_spxA1) and spxA2_smu (P_spxA2) were separately cloned in front of a promoterless chloramphenicol acetyl transferase (cat) reporter gene and integrated in the chromosome of UA159 using the one-step CAT integration vector pJL84 (Santiago, MacGilvray, Faustoferri, & Quivey, 2012). First, we measured CAT activity of P_spxA1 and P_spxA2 over the different phases of growth to find that CAT activity remained unaltered in both cases indicating that transcription from P_spxA1 or P_spxA2 is not growth phase dependent (Fig. S3). Based on this observation, the remaining experiments were performed with cells grown to an OD₆₀₀ ∼ 0.4 (mid-log phase). Next, we determined the CAT-specific activity of P_spxA1 or P_spxA2 in cultures exposed to 0.5 mM H₂O₂, 0.5 mM diamide, or pH 6.0 for up to 30 minutes. CAT activity from either P_spxA1 or P_spxA2 was not altered after those treatments, indicating that neither spxA1 nor spxA2 levels are regulated by peroxide or acid stresses (Fig. 7B–C).

Figure 7.

Transcriptional characterization of S. mutans spxA1 and spxA2. (A) Sequence of the P_spxA1 and P_spxA2 promoter regions. The σ^A-type −35 and −10 regions, mapped by 5’ RACE-PCR, are underlined. The transcriptional start site is shown in boldface and labeled with “+1”. The ATG start codon is also shown in boldface. Putative PerR- and SloR-binding sites in P_spxA1 and LiaR-binding site in P_spxA2 are shown inside boxes. (B-E) CAT activity driven from P_spxA1 or P_spxA2. (B-C) Cells were grown to mid-log phase and incubated in the presence of H₂O₂ (0.5 mM), diamide (0.5mM) or pH 6.0 for 10 or 30 minutes. (D-E) Mutant strains harboring the indicated CAT fusions were grown to mid-log phase. Asterisks indicates statistical significance (p value ≤ 0.01) when compared to control (B-C) or UA159 (D-E) by one-way ANOVA.

In addition to a previously identified SloR-binding motif (Crepps et al., 2016), in silico analysis revealed a potential regulatory site for PerR (NTANAANNATTNTAN) within the P_spxA1 region (Fig. 7A). Both PerR and SloR are metalloregulators shown to function primarily as negative regulators of genes associated with peroxide stress and metal uptake, respectively (Crepps et al., 2016; Makthal et al., 2013). While the role of PerR in spxA1smu gene regulation has not been explored, the Spatafora lab recently provided evidence that SloR interacts, albeit weakly, with the spxA1 promoter region (Crepps et al., 2016). To further assess the participation of SloR and begin to explore the role of PerR in spxA1 gene regulation, we compared activity of the P_spxA1-CAT reporter in the UA159 strain and its ΔsloR and ΔperR derivatives. Despite the high conservation (2 mismatches in 15 base consensus sequence) of the PerR-binding box identified in the spxA1 promoter region, loss of PerR resulted in a fairly small reduction in P_spxA1 activity that had no statistical significance (Fig. 7D). Though contrary to the identification of the PerR box operator, this result is consistent with the finding that that P_spxA1 was unresponsive to peroxide stress (Fig. 7B), the primary stress signal that alleviates PerR regulation. Despite previous evidence showing that SloR directly and specifically interacts with the spxA1 promoter region (Crepps et al., 2016), the CAT activity from P_spxA1 also did not change in the ΔsloR strain (Fig. 7D), even in the presence of a high concentration of manganese (250 µM) that is known to increase SloR activity (data not shown). One possible explanation for the unaltered activity of P_spxA1 in the ΔsloR or ΔperR strains is that repression of the spxA1 promoter by SloR and PerR is redundant such that one transcriptional repressor compensates for the absence of the other. It follows that the SloR-binding domain, known as SloR response element (SRE) (Spatafora et al., 2015) and PerR-binding domain are both AT-rich and one of the two putative SRE sites identified in the spxA1 promoter region is separated from the putative PerR-binding site by only two nucleotides (Fig. 7A). To test this possibility, we generated a double ΔsloRΔperR strain and found that P_spxA1 activity in this double mutant strain displayed a statistically significant increase (∼ 65%) when compared to the parent strain UA159 (Fig. 7D).

When in silico analysis was performed to search the P_spxA2 region for a conserved regulatory motif, only a previously identified LiaR-binding motif (Shankar et al., 2015; Suntharalingam, Senadheera, Mair, Levesque, & Cvitkovitch, 2009) was found. LiaR is the response regulator of the signal transduction system LiaFSR, which is responsible for orchestrating an envelope stress response against cell wall-targeting antibiotics and membrane damaging agents (Suntharalingam et al., 2009). Previously, Shankar and colleagues showed through a gel mobility shift assay that the S. mutans LiaR specifically binds to the P_spxA2 region, but further studies to determine whether LiaR functions as a repressor or activator of spxA2 were not pursued at that time (Shankar et al., 2015). Here, we found that P_spxA2 activity was strongly dependent on the LiaFSR system as P_spxA2::CAT activity was nearly undetectable in a ΔliaFSR strain (Fig. 7E). This finding prompted us to examine P_spxA2-CAT activity after exposure to a number of cell wall- or membrane-damaging agents that are sensed by LiaFSR in S. mutans or in closely-related organisms (Eldholm et al., 2010; Suntharalingam et al., 2009). In agreement with the strong dependence of LiaFSR for activation, CAT activity from P_spxA2 was significantly higher in cells exposed to bacitracin, chlorhexidine or SDS while ampicillin or daptomycin failed to induce P_spxA2 activity, at least under the conditions tested. (Fig. 8). As expected, the effect of bacitracin, chlorhexidine or SDS stresses on spxA2 transcription was sensed by the LiaFSR system as P_spxA2 CAT activity remained close to the detection limit in the ΔliaFSR strain after exposure to those envelope stress agents (Fig. S4).

Figure 8.

CAT activity driven from P_spxA2 under selected cell envelope stress conditions. Cells were grown to mid-log phase and incubated in the presence of bacitracin (64 µg ml⁻¹), daptomycin (2 µg ml⁻¹), ampicillin (0.4 µg ml⁻¹), chlorhexidine (0.5 µg ml⁻¹) or SDS (0.00125% w/v) for either 10 or 30 minutes. Asterisks indicate statistical significance (p value ≤ 0.01) when compared to untreated control by one-way ANOVA.

Redox-sensing switch is essential for SpxA1_Smu activity but not for SpxA2_Smu

A common feature of Spx proteins is the presence of a conserved N-terminal CXXC disulfide switch (Fig. 1). In Spx_Bsu, oxidation of this motif results in a conformational change that promotes interaction of the Spx-RNAP complex with Spx-regulated promoters (M. M. Nakano et al., 2010). However, a recent study from the Helmann group revealed that redox-sensing is not essential for activation of Spx_Bsu-regulated genes during cell envelope stress (Rojas-Tapias & Helmann, 2018a). To determine the importance of the CXXC motif for SpxA1_Smu and SpxA2_Smu function, the first cysteine residue of the motif was replaced by a serine residue (SXXC) in each protein to create the SpxA1^SXXC and SpxA2^SXXC strains. In the absence of stress (i.e., plain BHI at 37°C), the ΔspxA2 strain displayed a slightly extended adaptation (lag) phase whereas the ΔspxA1 strain grew slower without reaching the same final growth yield of the parent and ΔspxA2 strains (Fig. 9A). The SpxA1^SXXC and SpxA2^SXXC strains grew almost as well as the parent strain in plain BHI indicating that redox sensing is largely dispensable for SpxA1_Smu and SpxA2_Smu activities in the absence of stress (Fig. 9A). As shown previously (Kajfasz et al., 2010), ΔspxA1 was hypersensitive to peroxide stress (Fig. 9B) and, in this environment, an intact CXXC motif proved essential for SpxA1 activity as the SpxA1^SXXC strain phenocopied the ΔspxA1 strain (Fig. 9B). When compared to growth in the absence of stress, the ΔspxA2 strain displayed an even longer adaptation phase in the presence of H₂O₂, which was not observed in the SpxA2^SXXC strain (Fig. 9B).

Figure 9.

Growth curves of S. mutans UA159, Δspx and Spx^SXXC strains in (A) BHI or (B) in the presence of sub-inhibitory concentration of H₂O₂ (0.4 mM).

Based on global transcriptional analysis, we have previously suggested that SpxA2_Smu may have evolved to control cell envelope homeostasis while retaining a residual ability to regulate oxidative stress genes, particularly when in the absence of SpxA1_Smu (Kajfasz et al., 2017; Kajfasz et al., 2010). The possible association of SpxA2_Smu with envelope homeostasis is further supported by the discovery that spxA2 is under LiaFSR control (Fig. 7 and Fig. S4), and by previous studies that have implicated Spx regulation with envelope stress responses in other gram positive bacteria including, B. subtilis, E. faecalis and Streptococcus suis (Kajfasz et al., 2012; Rojas-Tapias & Helmann, 2018a; Zheng et al., 2014). To probe the role of SpxA2_Smu and of its CXXC motif in cell envelope stress, we compared the ability of the parent, ΔspxA2 and SpxA2^SXXC strains to grow in the presence of sub-inhibitory concentrations of cell wall (ampicillin or bacitracin) and membrane (chlorhexidine, daptomycin or SDS) stress agents using norfloxacin (DNA gyrase/replication inhibitor) and chloramphenicol (protein synthesis inhibitor) as antibiotic controls that do not target the cell envelope. When compared to the parent strain, the ΔspxA2 strain was dramatically more susceptible to all five cell envelope targeting antibiotics tested while it grew as well as the parent strain in the presence of norfloxacin or chloramphenicol (Fig. 10). When analyzed in conjuntion with the transcriptional studies shown in Figures 7 and 8, these results indicate that SpxA2_Smu is an integral member of the LiaR regulon and a major player in envelope stress responses. Finally, the SpxA2^SXXC strain phenocopied the parent strain (Fig. 10) indicating that redox-sensing is not necessary for activation of the SpxA2_Smu regulon during envelope stress.

Figure 10.

Growth curves of S. mutans UA159, ΔspxA2 and SpxA2^SXXC strains in BHI or in the presence of sub-inhibitory concentrations of ampicillin (0.4 µg ml⁻¹), bacitracin (32 µg ml⁻¹), norfloxacin (1 µg ml⁻¹), chloramphenicol (1 µg ml⁻¹), chlorhexidine (125 ng ml⁻¹), daptomycin (2 µg ml⁻¹), or SDS (0.0005%).

Discussion

In several gram positive bacteria, activation of oxidative stress responses is primarily mediated by Spx, a transcription factor that physically interacts with the C-terminal domain of the RNAP to activate transcription of genes involved in thiol homeostasis and detoxification (Antelmann & Helmann, 2011; Barendt et al., 2013; S. Nakano, Erwin, Ralle, & Zuber, 2005; S. Nakano, Kuster-Schock, Grossman, & Zuber, 2003; Runde et al., 2014; Villanueva et al., 2016). In addition, Spx has been shown to participate in the regulation of other types of stress and in developmental processes such as competence and sporulation, and to repress transcription of ribosomal RNA and ribosomal protein genes (Galvao et al., 2017; M. M. Nakano et al., 2001; M. M. Nakano, Nakano, & Zuber, 2002; Schafer et al., 2019; Turlan et al., 2009). Previously, we identified two Spx homologs in S. mutans that, when individually deleted, suppressed phenotypes associated with clpP or clpX gene inactivations (Kajfasz et al., 2009). In follow-up studies, we found that most pathways linked to oxidative stress survival are regulated by SpxA1_Smu and, to a much less extent, SpxA2_Smu (Galvao et al., 2015; Kajfasz et al., 2017; Kajfasz et al., 2010). Subsequent work from other laboratories supported our initial findings by showing that the genomes of streptococcal species and other Firmicutes encode two Spx paralogues often exhibiting both overlapping and unique regulatory functions (Chen et al., 2012; Zheng et al., 2014).

The Spx regulator was first identified in B. subtilis and, to this day, most of the mechanistic understanding of how cellular Spx levels are controlled comes from investigations conducted with this gram positive soil organism (Leelakriangsak et al., 2007; S. Nakano et al., 2003; S. Nakano et al., 2002; Rojas-Tapias & Helmann, 2018a, 2019; Zhang & Zuber, 2007). At the posttranslational level, Spx_Bsu is controlled by both the ClpXP and ClpCP systems (S. Nakano et al., 2002; Rojas-Tapias & Helmann, 2018b, 2019). In our initial study, we showed that Spx accumulates in S. mutans ΔclpP and ΔclpX strains, but we were unable to determine if one or both proteins were specifically targeted by ClpXP for degradation (Kajfasz et al., 2009). The low expression levels typical of transcriptional regulators and the high degree of conservation between SpxA1_Smu and SpxA2_Smu proved to be major hurdles for studying the stability of these proteins in vivo. Upon stumbling on the inherent limitation of a reconstituted in vitro proteolysis assay and taking into account that the C-terminal residues of Spx_Bsu are determinant for ClpP-mediated proteolysis, we turned our focus onto the roles that the C-terminal amino acids may play in differentiating SpxA1_Smu and SpxA2_Smu. To accomplish this, we fused the last ten amino acids of either SpxA1_Smu or SpxA2_Smu to an otherwise stable GFP and expressed the corresponding GFP::SpxA1_tail and GFP::SpxA2_tail fusion proteins in S. mutans to monitor GFP stability over time. At the amino acid level, the greatest difference between SpxA1_Smu and SpxA2_Smu appeared exactly at the C-terminal domain, with SpxA1_Smu displaying several acidic residues (4 of the last 5 are acidic residues) compared to a complete lack of acidic residues within the last 10 amino acids of SpxA2_Smu (Fig. 1). Notably, substitution of the last two non-acidic amino acid residues of Spx_Bsu (alanine and asparagine) by two aspartic acid residues abolished ClpXP-mediated proteolysis without affecting Spx-YjbH interaction (Chan et al., 2014; S. Nakano et al., 2003). Because of the disproportionately high number of acidic residues present in the C-terminus of SpxA1_Smu when compared to other Spx proteins, we predicted that SpxA1_Smu is resistant to ClpP proteolysis, which we were able to confirm using the GFP::SpxA1_tail and GFP::SpxA2_tail reporters (Fig. 2). The significance of C-terminal acidic residues for Spx stabilization was further demonstrated by the addition of two acidic residues to the SpxA2 tail (GFP::SpxA2_DDtail), which stabilized SpxA2_Smu, or by the removal of the acidic residues of SpxA1_Smu (GFP::SpxA1_-5tail and GFP::SpxA1_-7tail), which rendered SpxA1 susceptible to ClpP degradation (Fig. 3). Because the GFP reporter alone was stable for up to 12 hours in the parent UA159 strain, this later observation suggests that the amino acidic residues within the last 10 amino acids of SpxA1_Smu could be masking a degron in the C-terminal tail of SpxA1 that may only be exposed under specific conditions.

As an attempt to show that ClpP-mediated degration of SpxA2 _Smu but not of SpxA1_Smu occurs in its native context, we compared the transcriptional profile of an SpxA1- (sodA) and of an SpxA2-regulated (smu1412c) gene (Kajfasz et al., 2010) in the UA159, ∆clpP and ∆clpE strains. While no changes in expression of sodA was observed between these strains, smu1412c was significantly upregulated in ∆clpP compared to parent strain UA159 (Fig. 6) suggesting accumulation of SpxA2 but not of SpxA1 in the absence of ClpP proteolysis. Unfortunately, attempts to validate these observations by reproducing the SpxA2_Smu stabilization (SpxA2_DD strain) and SpxA1_Smu destabilization (SpxA1_-7AA strain) in S. mutans were not conclusive. If in one hand, replacing the original spxA1 by a gene copy expressing a truncated SpxA1 lacking the last 7 amino acids (SpxA1_-7AA strain) increased peroxide stress sensitivity, stabilization of SpxA2_Smu (UA-SpxA2_DD strain) did not increase S. mutans tolerance towards cell envelope stress agents (Fig. S5 and S6). These mixed results were unsurprising given that there could be other factors that ultimately control Spx levels and activity. Moreover, stabilization and the presumably accumulation of SpxA2_Smu (UA-SpxA2_DD strain) may not necessarily have to translate in the opposite phenotype of the ΔspxA2 strain.

Barring a handful of streptococcal species (Fig. S1), the C-terminal residues of SpxA1 proteins have none or very few acidic amino acids (one or two within the last 10 residues, none within the last 5 amino acids). In the lone Spx protein of S. aureus (Spx_Sau), the last C terminal amino acid is an aspartic acid; however, the Spx_Sau was shown to become stable upon clpP gene inactivation (Engman et al., 2012). To further demonstrate the importance of the C-terminal tail for Spx stability, we also showed that addition of the C-terminal tail of either SpxA1_Spn or SpxA2_Spn (the 2 pneumococcal Spx proteins), neither one containing acidic residues within their last 7–8 amino acids, destabilized GFP in a ClpP-dependent manner (Fig. 4). Thus, it appears that the scenario in which SpxA1 is stable, while SpxA2 is susceptible to ClpP degradation is limited to S. mutans and few additional streptococcal species. Because of the intimate relationship between streptococcal SpxA1 and SpxA2 proteins, seemingly ranging from cooperative to antagonistic (Chen et al., 2012; Port et al., 2017; Zheng et al., 2014), the different susceptibility of streptococcal SpxA1 proteins to ClpP degradation and how this may affect expression of the SpxA1 and SpxA2 regulons will demand analysis on a species-by-species basis.

In B. subtilis, Spx is degraded by both ClpCP and ClpXP complexes (S. Nakano et al., 2002; Rojas-Tapias & Helmann, 2019). It has been proposed that in the absence of disulfide stress, Spx levels are primarily controlled by ClpXP and the YjbH adaptor (Chan et al., 2012; Engman et al., 2012; Feng et al., 2013; Garg et al., 2009). When exposed to disulfide stress, the structural integrity of ClpX is severely compromised due to structural changes in its N-terminal Cys₄ zinc-binding domain (Zhang & Zuber, 2007). It has been proposed that once ClpX becomes inactive, ClpCP degradation takes over to prevent accumulation of Spx (Rojas-Tapias & Helmann, 2019). Of note, the Spx paralog MgsR_Bsu protein was also shown to be subjected to both ClpCP and ClpXP posttranscriptional control (Reder, Pother, Gerth, & Hecker, 2012). In the study that led to the identification of spxA1_Smu and spxA2_Smu, we showed that Spx accumulates in ΔclpP and ΔclpX strains (Kajfasz et al., 2009), though at the time we were unable to make a distinction between SpxA1 and SpxA2. In this work, we confirmed the central role of ClpP in SpxA2_Smu degradation using the GFP reporter. Further, we found that deletion of clpE partially increased the stability of the GFP::SpxA2_tail fusion protein whereas single deletion of clpC or clpX did not. Stability of the GFP::SpxA2_tail construct was further increased in the ΔclpEΔclpX double mutant and almost completely stabilized in the ΔclpCΔclpE strain (Fig. 5). We certainly did not expect ClpE, followed by ClpC, to be the major Clp ATPases involved in SpxA2_Smu degradation and acknowledge that this observation may reflect the limitation of using chimeric GFP-Spx_tail reporters. Nevertheless, our results indicate that all three Clp proteolytic systems, ClpCP, ClpEP and ClpXP, have the potential to degrade SpxA2_Smu in vivo. We should also note that the Biswas group showed previously that SsrA-tagged proteins of S. mutans can be degraded by all three Clp proteolytic complexes, with ClpCP and ClpEP playing a more prominent role during heat stress (Tao & Biswas, 2015). In a separate study, they also characterized a tripeptide motif LPF as a ClpX degradation signal that functions only in selected S. mutans strains, including the UA159 strain used in our study (Jana et al., 2016). However, neither SpxA1_Smu nor SpxA2_Smu encode the LPF motif. Further studies are necessary for a complete understanding of how the different Clp ATPases regulate SpxA2_Smu levels. Considering the absence of YjbH (ClpXP adaptor) and McsB (ClpCP adaptor) homologs in streptococci, subsequent studies should consider the use of genetic screens for the identification of adaptor proteins that may shed new light into the specific contributions of the different Clp ATPases to Spx degradation in vivo.

In addition to proteolytic control, B. subtilis spx transcription is coordinated through multiple promoters that are controlled by different sigma factors (σ^A, σ^B and σ^M), as well as two transcriptional regulators (Barendt et al., 2016; Eiamphungporn & Helmann, 2008; Leelakriangsak et al., 2007; Reyes & Zuber, 2008; Rojas-Tapias & Helmann, 2018a). While we cannot rule out that the spxA1_Smu and spxA2_Smu genes may also be transcribed by distal promoters, we focused our initial studies on the proximal σ^A-type promoters upstream the spxA1 and spxA2 coding regions. We found that the PerR and SloR metalloregulators have redundant roles as repressors of P_spxA1 as only the simultaneous inactivation of perR and sloR increased P_spxA1 activity (Fig. 7). Based on this evidence, we propose that full induction of spxA1 transcription may only occur in a low manganese/high H₂O₂ environment, when the DNA-binding capacity of SloR and PerR should be simultaneously impaired. This condition may be restricted to a few specific environments such as the initial stages of an infectious process when neutrophils and other professional phagocytes secrete large quantities of manganese-sequestering calprotectin and mediate an oxidative burst at the same time.

Recently, Spx_Bsu was shown to accumulate in response to cell wall stress and the B. subtilis Δspx strain displayed increased sensitivity to cell wall-targeting antibiotics (Rojas-Tapias & Helmann, 2018a). In support of the possible association of SpxA2_Smu in cell envelope stress, a previous study revealed that spxA2_Smu transcription is under LiaFSR regulation, a signal transduction system that activates a transcriptional response to cell envelope stress in Firmicutes (Shankar et al., 2015; Suntharalingam et al., 2009). In S. mutans, inactivation of liaFSR increased sensitivity to cell wall lipid II inhibitors and to cell membrane-disrupting agents (Suntharalingam et al., 2009). Here, we showed that transcription of spxA2_Smu is strongly dependent on the LiaFSR signal transduction system (Fig. 8E), and that transcription of spxA2_Smu is induced by some of the same cell wall and membrane-targeting agents that activate the LiaFSR system (Fig. 8). In addition, we provided conclusive evidence that SpxA2_Smu plays a prominent role in envelope stress responses as the ΔspxA2 strain was hypersensitive to all five cell wall and membrane-targeting agents tested (Fig. 10). During revision of this manuscript, Baker and colleagues came to a similar conclusion by showing that transcription of S. mutans spxA2 was highly dependent on the LiaFSR system and that SpxA2_Smu was required for growth under envelope stress conditions (J. L. Baker, Saputo, Faustoferri, & Quivey, 2020). Studies to identify new members of the SpxA2_Smu regulon as a means to reveal the scope of SpxA2 regulation and to identify novel envelope stress genes are currently underway.

In addition to transcriptional and posttranslational mechanisms, Spx activity is also determined by reversible oxidation and disulfide bond formation of a N-terminal CXXC redox switch (S. Nakano et al., 2005). It has been shown that oxidation of the CXXC motif results in a conformational change that unfolds the helix α4 of the Spx protein, which facilitates interaction between the Spx-RNAP complex and target promoters (M. M. Nakano et al., 2010). Substitution of one or both cysteine residues of the motif by alanine significantly impaired competence development in B. subtilis and S. pneumoniae, as well as disulfide stress responses in B. subtilis and L. monocytogenes (Gaballa, Antelmann, Hamilton, & Helmann, 2013; S. Nakano et al., 2005; Rochat et al., 2012; Turlan et al., 2009; Whiteley et al., 2017). More recently new evidence indicated that the redox switch is not always essential for Spx activity, particularly during cell wall stress (Gaballa et al., 2013; Rojas-Tapias & Helmann, 2018a). More specifically, activation of the Spx regulon during cell wall stress occurred with Spx_Bsu found primarily in a reduced state (Rojas-Tapias & Helmann, 2018a). These authors discovered that Spx levels increased during envelope stress due to upregulation of the distal σ^M-regulated P_M1 promoter, revealing that increased Spx_Bsu levels rather than its oxidation state was the determinant factor for activation of the Spx regulon during envelope stress (Rojas-Tapias & Helmann, 2018a). Here, we showed that redox-sensing via the CXXC motif is essential for SpxA1_Smu-mediated activation of oxidative stress responses but dispensable for SpxA2_Smu activity during envelope stress conditions (Fig. 9). Thus, it appears that the functions of a single Spx protein in B. subtilis is shared by two Spx proteins in S. mutans, with SpxA2_Smu primarily functioning as a regulator of cell envelope stress and SpxA1_Smu as an oxidative stress regulator.

In summary, we showed here that SpxA1_Smu and SpxA2_Smu levels are controlled by very distinct transcriptional and posttranslational mechanisms. Most notably, posttranslational control through targeted proteolysis appears to be restricted to SpxA2_Smu as acidic amino acid residues at the extreme C-terminus of SpxA1_Smu prevented ClpP-mediated degradation. On the other hand, SpxA1_Smu activity was found to be largely dependent on its oxidation status whereas redox-sensing played a negligible role for SpxA2_Smu activity. Finally, transcriptional and phenotypic characterizations provided unequivocal evidence that SpxA2_Smu evolved to regulate responses associated with cell envelope stress while retaining the ability to partially compensate for the loss of SpxA1_Smu during oxidative stress.

Experimental procedures

Bacterial strains and culture conditions

Strains and plasmids used in this study are listed in Table 1 and Table S1, respectively. All E. coli strains were routinely grown in Luria-Bertani (LB) media at 37°C. When required, kanamycin (100 μg mL⁻¹) or ampicillin (100 μg mL⁻¹) was added to LB broth or agar plates. Strains of S. mutans were routinely grown in brain heart infusion (BHI) at 37°C in a humidified 5% CO₂ atmosphere. When required, kanamycin (1000 μg mL⁻¹), erythromycin (10 μg mL⁻¹), or spectinomycin (1500 μg mL⁻¹) was added to the growth media. Due to the high background detected in BHI medium, the biofilm media (BM) supplemented with glucose (Loo, Corliss, & Ganeshkumar, 2000) was used to monitor GFP decay. For growth kinetics experiments, overnight cultures were diluted 1:20 into BHI and grown in a 5% CO₂ atmosphere to an OD₆₀₀ of 0.3, at which point 5 μl of the culture was used to inoculate wells of a 100-well plate containing 250 μL of the appropriate medium. The automated growth reader Bioscreen C (Oy Growth Curves Ab Ltd.) was used to monitor the ability of S. mutans UA159 and derivatives to grow in the presence of sub-inhibitory concentrations of H₂O₂, or the cell envelope stress agents ampicillin, bacitracin, chlorhexidine, daptomycin and SDS. To maintain an anaerobic environment, an overlay of 50 μL of sterile mineral oil was added to each well.

Table 1.

Streptococcus mutans strains used in this study.

Strains	Relevant characteristics	Source
UA159	wild-type	Lab stock
∆clpP	clpP::Kan	(Kajfasz et al., 2009)
∆clpC	clpC::Spec	(Kajfasz et al., 2009)
∆clpX	clpX::Kan	(Kajfasz et al., 2009)
∆clpE	clpE::Kan	(Kajfasz et al., 2009)
∆clpC∆clpX	clpC::Spec; clpX::Kan	This study
∆clpC∆clpE	clpC::Spec; clpE::Kan	This study
∆clpE∆clpX	clpE::spec; clpX::Kan	This study
∆spxA1	spxA1::Spec	(Kajfasz et al., 2009)
∆spxA2	spxA2::Erm	(Kajfasz et al., 2009)
∆perR	perR::Erm	This study
∆liaFSR	liaFSR::spec	This study
GMS584 (∆sloR)	sloR::Erm	(Rolerson et al., 2006)
∆perR∆sloR	sloR::Erm; perR::Spec	This study
UA159 SpxA1SXXC	SpxA1 with C-S substitution at position 10	This study
UA159 SpxA2SXXC	SpxA2 with C-S substitution at position 10	This study
SpxA1-7AA	SpxA1 with 7 amino acid C-terminal deletion	This study
UA-SpxA2DD	UA159 harboring pIB184::SpxA2DD, ErmR	This study

Recombinant protein purification and in vitro proteolysis

To obtain N-terminal 6x-His-tagged rClpP, rClpX, rSpxA1 and rSpxA2 proteins, the full-length genes were individually ligated into the expression vector pET16b (Novagen), then introduced into the E. coli expression strain BL21 (DE3) via electroporation. Expression of recombinant proteins was achieved by growing cells in LB to an OD₆₀₀ of 0.5 and inducing protein expression by adding 0.4 mM isopropyl-β-d-1-thiogalactopyrosinide (IPTG) to the exponentially-grown cultures for 16 h at 15°C. After cell lysis using the Avestin Emulsiflex C5 homogenizer (ATA Scientific), recombinant proteins were purified by affinity chromatography with Ni²⁺-nitrilotriacetic acid-agarose (Qiagen). Eluted recombinant proteins were dialyzed overnight in phosphate-buffered saline (PBS) at 4°C. Purity of recombinant proteins was analyzed by 12% SDS-PAGE followed by Coomassie blue staining. Concentration of purified proteins was determined using the bicinchoninic acid (BCA) assay (Thermo fisher). Aliquots of purified proteins were stored at −20°C in 10% glycerol for one-time use.

The in vitro proteolysis assay was carried out as previously described (S. Nakano et al., 2002) with slight modifications. Briefly, rClpP (4 μM) and rSpxA1 or rSpxA2 (4 μM) were incubated at 37°C in the presence of rClpX (2.5 μM) to a final volume of 50 μl of reaction buffer (25 mM MOPS [morpholinepropanesulfonic acid]-KOH [pH 7.0], 100 mM KCl, 5 mM MgCl₂, 0.5 mM dithiothreitol, 4 mM ATP, 2 mM phosphoenol pyruvate, 0.93 μM pyruvate kinase [Sigma]). Phosphoenol pyruvate and pyruvate kinase were used to regenerate ATP. At selected intervals, 10 μl aliquots were mixed with 5 μl of stop solution. The stability of SpxA1 or SpxA2 was analyzed by separating proteins on 15% SDS-PAGE followed by Coomassie blue staining.

Mutant construction and genetic manipulation

The ΔclpC, ΔclpE, ΔclpP and ΔclpX strains have been isolated previously (Kajfasz et al., 2009). To accommodate double deletions of the clp ATPases, we used a PCR ligation approach to create a new clpE deletion strain using a spectinomycin resistance cassette in place of the original kanamycin resistance cassette. The ΔclpCΔclpX and ΔclpEΔclpX strains were generated by transforming the ΔclpX strain with clpC::spec or clpE::spec ligation mixes. The ΔclpCΔclpE strain was generated by transforming the ΔclpC mutant with the clpE::kan ligation mix. To create the perR and liaFSR deletion mutants, the PCR ligation mutagenesis approach was again utilized (Lau, Sung, Lee, Morrison, & Cvitkovitch, 2002). Briefly, PCR fragments flanking the desired region to be deleted were ligated to a nonpolar spectinomycin (spec^r) or erythromycin (erm^r) cassette and the ligation mix used to transform S. mutans UA159. To create the ΔperRΔsloR double mutant, perR::spec ligation mix was used to transform a sloR single mutant strain (Rolerson et al., 2006), a gift from Dr. Grace Spatafora (Middlebury College). The deletions were confirmed as correct by Sanger sequencing of the insertion site and flanking region.

To generate GFP-Spx_tail chimeras, the gfp gene was PCR amplified with reverse primers that carried sequence of the respective Spx C-terminal 10 amino acids and a forward primer that recognizes GFP N-terminal end using the pUG vector (Hwang et al., 2016) as template. Following digestion with BamHI and XhoI, the amplified PCR products were ligated onto plasmid pIB184 (Biswas, Jha, & Fromm, 2008), which carries a strong P₂₃ promoter (Tao & Biswas, 2015). The same approach was used to generate the S. pneumoniae GFP::SpxA1 and GFP::SpxA2 tails. To express SpxA2_Smu with two additional acidic amino acids, the spxA2 gene was amplified using a reverse primer encoding two additional codons for aspartic acid before cloning into pIB184. To generate the SpxA1 C-terminal 5 or 7 amino acid deletion tail, primers were designed to match either SpxA1 amino acids corresponding to C-terminal 15^th to 6^th (SpxA1_-5tail) or C-terminal 17^th to 8^th (SpxA1_-7tail) amino acids along with a few C-terminal GFP amino acids and used with GFP N-terminal forward primer and cloned as described above. To generate promoter CAT fusions, the promoter region upstream spxA1 or spxA2 were PCR amplified with their own ribosome binding site and cloned into the one-step cat integration vector pJL84 (Santiago et al., 2012) and transformed into S. mutans UA159 and derivative strains such that a single copy of the P_spxA1-cat or P_spxA2-cat reporter was integrated at the mannitol utilization locus.

Overlap extension PCR (OE-PCR) was utilized to alter a single amino acid in the conserved CXXC motif of SpxA1 or SpxA2 to generate, respectively, the SpxA1^SXXC and SpxA2^SXXC strains. Primer sets 5’ C10SF/5’ C10SR and 3’ C10SF/3’ C10SR were used to amplify 2-kb fragments of the DNA flanking the site of mutation. The 2 kb PCR fragments carried an overlapping 26 base sequence, which was used to anneal the two PCR products in a ligase-free PCR reaction followed by overlap-extension (OE) amplification using the 5’ C10SF and 3’ C10SR primer sets. The overlap PCR product containing the single amino acid substitution was purified and used to transform the S. mutans ∆spxA1 and ∆spxA2 strains. Here, we took advantage of the antibiotic resistances of the ∆spxA1 and ∆spxA2 strains strictly as a screening tool. The mutant colonies were allowed to grow on BHI, then were patched onto both BHI and BHI containing the appropriate antibiotic for the original deletion mutation (spectinomycin for ∆spxA1 and kanamycin for ∆spxA2). Candidate transformants were sorted by loss of the ability to grow on BHI agar containing the antibiotic appropriate to the original mutation, then screened by PCR and confirmed by sequencing. A similar PCR-based approach was used to isolate the SpxA1_-7AA strain. Primer sets 5’ C10SF/ SpxA1–7 5armrevERI and SpxA1–7 3armfwdERI/3’ C10SR were used to amplify 2-kb fragments of the DNA flanking the 7 amino acids deletion region. The PCR products were digested with EcoRI and ligated using T4 DNA ligase (NEB) before transforming into ∆spxA1 strain and colonies were selected based on loss of antibiotic resistance as described before. For generating stable SpxA2_DD strain, the spxA2 gene was amplified using primer set pIB184SpxA2Fwd and SpxA2DDRev and cloned into pIB184. The plasmid expressing SpxA2_DD was transformed into UA159 to generate strain UA-SpxA2_DD. All primers used in the genetic manipulations are listed in Table S2.

Western blot analysis

Whole-cell protein lysates were obtained by homogenization in the presence of 0.1-mm glass beads using a bead beater (Biospec). Equal amounts of protein extracts (usually 50 μg per lane) were separated by 12% SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes (Millipore) using standard protocols. GFP detection was performed using rabbit anti-GFP polyclonal antibody (Thermo Fisher Scientific) diluted 1:2000 in PBS containing 0.01% Tween 20 and anti-rabbit horseradish peroxidase (HRP)-coupled antibody (Sigma-Aldrich). Protein concentration of the lysates was determined using the bicinchoninic acid (BCA) assay and equal loading was confirmed by staining the blots with Ponceau S (Figure S7).

CAT assay

Chloramphenicol acetyltransferase (CAT) activity was measured from cell cultures grown in BHI to different growth phases (early-, mid- and late-log), or grown to an OD₆₀₀ of 0.4 and subjected to acid stress (pH 6.0), 0.5 mM H₂O₂, 0.5 mM diamide, or cell envelope stress (0.4 µg ml⁻¹ ampicillin, 64 µg ml⁻¹ bacitracin, 0.5 µg ml⁻¹ chlorhexidine, 2 µg ml⁻¹ daptomycin or 0.00125% SDS) for up to 30 minutes. Aliquots of the culture left untreated were used as controls. Cell-free lysates were prepared by homogenizing the cells in a bead-beater, and CAT activity determined by a spectrophotometric method (Chakraborty & Burne, 2017). Briefly, the CAT assay reaction mixtures were prepared fresh so that final concentrations of each component were as follows: 5,5-dithio-bis-nitrobenzoic acid (DTNB), 0.4 mg ml⁻¹; Tris-HCl, 100 mM; and acetyl-CoA, 0.1 mM. Chloramphenicol (5 mM stock) and cell lysates were pre-warmed to 37°C just before the reactions were initiated. Changes in absorbance were recorded at OD₄₁₂ in a 37°C spectrophotometer. Protein concentrations of cell lysates used in CAT assays were determined using the BCA assay. CAT activity was expressed as the amount of lysate needed to acetylate 1 nmol of chloramphenicol min⁻¹ mg of protein⁻¹.

qRT-PCR analysis

RNA was extracted from cultures grown to mid-exponential phase (OD₆₀₀ ∼0.4) as previously described (Galvao et al., 2015). Briefly, cDNA from 1 μg of RNA was synthesized using a high-capacity cDNA reverse transcriptase kit containing random primers (Applied Biosystems). Gene-specific primers for the sodA and smu1412c (Table S2) were designed using Beacon Designer (version 2.0) software (Premier Biosoft International) to amplify a region of around 120 bp in length. Quantitative real-time PCR (qRT-PCR) were performed in an iCycler apparatus (Bio-Rad).

GFP decay assay

S. mutans UA159 and Δclp derivatives harboring different pIB184::GFP-Spx fusion tails were grown overnight in BHI supplemented with erythomycin and sub-cultured in BM-glucose media (Loo et al., 2000). Cultures were grown on 96-well plates at 37°C in a spectro-fluorimeter (Synergy H1, Biotek) to an OD₆₀₀ of 0.25 when an inhibitory concentration of chloramphenicol (20 ug ml⁻¹) was added to stop protein synthesis. GFP fluorescence was monitored in a timely manner and, to determine GFP decay rate, fluorescence was normalized to the fluorescence immediately before addition of chloramphenicol. The results are presented as percent fluorescence remaining over time of atleast 6 replicates.

5’ RACE-PCR

5’ Rapid Amplification of cDNA Ends (RACE)-PCR was used to determine the transcription start site of spxA1 and spxA2 according to the manufacturer’s protocol (Invitrogen). Briefly, 2 μg of RNA subjected to reverse transcription with gene-specific primer 1 (GSP1, Table S2) and SuperScript II Reverse Transcriptase, followed by RNase treatment and 3’poly dC tail addition with terminal deoxynucleotidyl transferase. The dC-tailed cDNA was then PCR amplified using a nested gene specific primer 2 (GSP2, Table S2) and Abridged Anchor Primer (AAP). The transcription start site was determined by sequencing the amplified cDNA.

Statistical Analysis

All data were analyzed using GraphPad Prism 6.0 software. For the GFP-decay assay, data were analyzed at each time point using ordinary one-way ANOVA with Dunnett’s multiple comparison post-test. To determine the statistical significance for the rate of degradation, geometric mean values of percentage of GFP-fluorescence remaining for each strain at a given time point were plotted with 99% confidence interval. Each respective figure legend indicates the control used for multiple comparisons. For CAT assays and qRT-PCR analyses, similar ordinary one-way ANOVA with Dunnett’s multiple comparison post-test were used. To determine the statistical significance under stress conditions, geometric mean values of CAT activities were plotted with 99% confidence interval from a single CAT construct and compared to a stress-free control strain. To determine the CAT activity in a different strain background, ordinary one-way ANOVA with Dunnett’s multiple comparison post-test were used as mentioned before, and significance were analyzed using parent strain CAT activity as control, plotting geometric mean values of CAT activities with 99% confidence interval.