Streptococcus pneumoniae R6, wt (Hoskins et al., 2001; Tettelin et al., 2001) and its derivatives were usually grown in AGCH medium (Lacks, 1968) with 0.3% sucrose and 0.2% yeast extract at 37°C. When necessary, the medium was supplemented with kanamycin (250 μg/ml), chloramphenicol (1 or 3 μg/ml for cells harboring plasmid pC194r (del Solar and Espinosa, 1992), or streptomycin (100 μg/ml).

DNA manipulations and other molecular biology techniques were performed according to standard protocols (Sambrook and Russel, 2001) or manufacturers’ instructions when commercial kits were used. Genomic DNA was isolated with Bacterial Genomic DNA Isolation Kit (Norgen Biotech, Corp.); whereas plasmid DNA was extracted with High Pure Plasmid Isolation kit (Roche), but the protocols were slightly modified to account for the low G+C content of the pneumococcal genome (Ruiz-Cruz et al., 2010). DNA fragments or PCR-amplified DNA-products were purified with QIAquick Gel Extraction Kit (Qiagen). For DNA sequence verification, samples were sent for automated Sanger sequencing in Secugen S.L., Centro de Investigaciones Biológicas, CSIC, Madrid, and then analyzed by the BioEdit Sequence Alignment Editor version 7.0.4.1 (Hall, 1999).

To determine the growth patterns and CFU of the S. pneumoniae R6wt and R6ΔPezAT mutant, cells were grown overnight until OD₆₅₀∼0.3, and then diluted to OD₆₅₀∼0.03. Cells were then allowed to continue growing in fresh medium at 37°C without shaking. OD₆₅₀ were take every 30 min to assess growth patterns and samples were taken at various OD₆₅₀ (0.2, 0.4, 0.6, and 0.8) and then plated on AGCH 1% agar to assess CFU/ml. Samples were also taken at OD₆₅₀∼0.3 and 0.6 for examination under phase-contrast microscope (Olympus CKX41). Pictures were taken at 100× magnifications, 200 ms.

Phase variation analyses were done as reported (Weiser et al., 1996) with slight modifications. In brief, S. pneumoniae R6wt and R6ΔPezAT were grown until OD₆₅₀∼0.3. Cells were then diluted and plated on tryptic soy (Conda Pronadisa) plates containing 1% agar onto which 100 μl of catalase (5,000 U/ml; Calbiochem) was added. Cells were grown at 37°C overnight in a 5% CO₂ incubator. Colony morphology was observed under stereo microscope (Leica).

Oxidative stress assays were done as reported (Bortoni et al., 2009) with small modifications: S. pneumoniae R6wt and R6ΔPezAT strains were grown as above until OD₆₅₀∼0.3. For each strain, 1 ml of cells were harvested and resuspended in 0.5 ml 1× PBS (pH 7.0). The cells (50 μl) were mixed with an equal volume of H₂O₂ (Merck), to give a final concentration of either 20 mM or 40 mM H₂O₂, followed by incubation at 37°C, 20 min. The control mixtures contained cells and 1× PBS (pH 7.0). Serial dilutions were made and the cells were plated on AGCH agar and incubated at 37°C, 15 h. Colonies were counted and the results were represented as percent of survival relative to the control. These assays were repeated four times and the average and standard deviation were calculated.

We used resazurin as color indicator to measure the MIC of S. pneumoniae R6wt and R6ΔPezAT mutant strains. Resazurin is a blue non-fluorescent dye that can be converted to the pink fluorescent dye, resorufin, by metabolically active cells (Tizzard et al., 2006; Mania et al., 2010; Khalifa et al., 2013), a method that has been used for S. pneumoniae (Patel et al., 2011; Vandevelde et al., 2014). The color changes can be also measured by light absorbance at 570 nm for resorufin or 600 nm for resazurin. MICs of various antibiotics (ampicillin, benzetacil, levofloxacin, and streptomycin) were determined by using twofold serial dilution method. Briefly, the cells were grown at 37°C until OD₆₅₀∼0.3. Cells were diluted to 1:100 into pre-warmed fresh medium and 100 μl were added into a flat-bottom BD Falcon 96-Well Cell Culture Plate. Different amounts of antibiotics and 200 μM resazurin were added to each well. The mixtures were incubated at 37°C and the color changes were measured at 600 nm every hour by employment of a Varioskan Flash Multimode Reader (Thermo Scientific). These assays were repeated at three times and the mean values and the standard deviations were calculated. t-test (with p-value < 0.05) was used to evaluate the significant differences of the results. The results were validated by the use of standard MIC determinations on microtiter plates (Andrews, 2001), without resazurin but determining the number of colonies formed by cultures grown in the absence or in the presence of ampicillin.

Streptococcus pneumoniae strains R6wt and R6ΔPezAT were made competent by preparing cultures from stocks and diluting 1:1,000. Cells were grown at 37°C, and when an OD₆₅₀∼0.3 was reached, two 1:40 successive dilutions were made until OD₆₅₀∼0.3 was reached again. Cultures received 10% glycerol; aliquots of 250 μl were made as pre-competent cultures, frozen, and stored at -80°C. When transformation of cultures was tested, 200 μl of competent cells were added into 4 ml AGCH medium supplemented with 0.2% sucrose and 0.001% CaCl₂ and incubated at 30°C, 20 min. After incubation, 1 ml of competent cells received 400 ng of DNA (see Results), and to this mixture, 100 ng of CSP-1 was added (no CSP-1 was added to control cultures). Transformation mixtures were incubated at 30°C, 30 min and then transferred to 37°C, 90 min to allow phenotypic expression. Serial dilutions were made and different amounts of cells were plated on 1% AGCH agar supplemented with 100 μg/ml streptomycin (chromosomal transformants) or not (total cell counts). Plates were incubated 16 h at 37°C, and the number of transformants was determined. These assays were repeated three times.

Competence development was tested by using a transcriptional fusion of the luc gene, which encodes luciferase, to the ssbB gene that is specifically induced at competence. The ssbB::luc fusion reports development of competence by luciferase light emission (Prudhomme and Claverys, 2007). Competence assay were performed as described (Caymaris et al., 2010). Briefly, S. pneumoniae strains R6luc and R6ΔPezATluc were inoculated (1:40 dilutions) in C+Y medium (pH 7.0) supplemented with trypsin (2 μg/ml) and allowed to grow at 37°C until OD₅₅₀∼0.2. Cells were harvested and concentrated to OD₅₅₀∼0.8 in fresh medium containing 10% glycerol and kept at -80°C. To monitor the development of spontaneous competence, the cells were thawed and inoculated (1:50 dilutions) in C+Y medium prepared at various pH. For each sample, 182.7 μl of diluted cells were mixed with 13.3 μl firefly D-luciferin (Thermo Fisher Scientific; 10 mM) in a 96-well clear bottom white plate (Corning), followed by incubation at 37°C in a Varioskan Flash Multimode Reader (Thermo Scientific). RLU and OD₄₉₂ were measured every 7 min intervals. Values correspond to individual cultures representative of three independent experiments.

The genome of S. pneumoniae R6wt (Tettelin et al., 2001) harbors a single copy of the pezAT operon (spr0951–spr0952; NCBI accession no. NC_003098), located within the PPI1 (Khoo et al., 2007). Other pneumococcal strains, like CGSP14, ATCC 700669, and P1031 have two copies of the same operon (Chan et al., 2012). The presence of the pezAT operon has been related to pneumococcal virulence (Brown et al., 2004). Like the typical Type II TAs, the pezA antitoxin gene is placed upstream of pezT (Figure 1), and both genes overlap by one nucleotide, suggestive of coupled translation (Chan et al., 2012). Co-transcription of both genes is directed by a single promoter that is located upstream of pezA. This region also includes a long inverted repeat which spans the -35 and -10 regions of the promoter and that is the region where PezA/PezT proteins bind to control their own synthesis: binding of the PezA:PezT to their target would hinder binding of the host RNA polymerase to the promoter, leading to transcriptional repression of the operon (Khoo et al., 2007; Chan et al., 2016). We have studied the influence of pezAT on the pneumococcal lifestyle without recurring to previous general approaches that consist of the overexpression of the toxin gene, either in the homologous or in a heterologous host, followed by analysis of the resulting effects (Christensen and Gerdes, 2003; Khoo et al., 2007; Nieto et al., 2007). Instead, we constructed a mutant strain (R6ΔPezAT) in which the entire pezAT operon, including the upstream intergenic region was deleted and replaced by a gene encoding kanamycin (Figure 1), confirming the observation that the operon is not essential for S. pneumoniae (Brown et al., 2001). Under these conditions, lack of the operon could be studied on a number of pneumococcal responses.

When the R6wt and the R6ΔPezAT strains were tested for growth under normal growth condition (37°C, no aeration, AGCH medium pH7.7), no differences were found at the exponential phase (Figure 2A). These observations agree with a previous report (Brown et al., 2004) in which no differences were found in growth in laboratory broth, serum or blood when the pezT gene was deleted. In our conditions, however, differences between the two strains were evident when the cultures entered into the stationary phase: the R6ΔPezAT mutant showed a higher resilience to lysis compared to the R6wt (Figure 2A). We took this finding as a solid indication of the participation of the pezAT operon in the pneumococcal growth. Despite no differences in growth patterns during exponential phase for both strains were observed, the CFU counts were lower for the wt than for the mutant along the exponential phase, as determined by plating appropriate dilutions of cultures at various ODs and counting of the colony numbers (Figure 2B). Given that PezT targets polymerisation of pneumococcal cell walls, it was interesting to know whether this phenomenon was due to changes in the cell morphology. This postulation was assessed by examination of growing cells under phase-contrast microscopy (Figure 2C). Indeed, R6ΔPezAT mutant formed shorter chains than the R6wt strain at exponential phase. However, the effect was reduced at the beginning of stationary phase, and difficult to assess at prolonged incubation times (OD₆₅₀∼0.8), due to lysis and appearance of ‘ghosts’: cell walls, cell debris, etc. Longer cell chains in the R6wt than in the mutant should lead to formation of less CFU in the former than in the latter strains as one colony would result from a string of cells. One possibility could be that the levels of one or more pneumococcal lytic enzymes were slightly increased in the mutant strain. If this were the case, a good candidate would be LytB because, contrary to LytA, it is a non-autolytic murein hydrolase that allows localized peptidoglycan hydrolysis and separates daughter cells (Rico-Lastres et al., 2015); inactivation of lytB led to formation of long chains integrated by more than 100 cells (De Las Rivas et al., 2002). Thus, increases in the levels of LytB by the deletion of pezT, would result to a reduced number of cells per chain (Johnston et al., 2016). This hypothesis, however, needs further experiments.

Two more straightforward phenotypic tests were also performed: colony appearance to check phase variation (Weiser et al., 1996), and colony sizes and morphologies. However, no significant differences were found between strains R6wt and R6ΔPezAT.

Streptococcus pneumoniae is a facultative anaerobe that colonizes the human nasopharynx of up to 70% of healthy individuals (Chan et al., 2012; Gamez and Hammerschmidt, 2012). As a consequence, the bacteria are customarily exposed to an oxygen-rich environment under colonization conditions, and thus the most frequent stress found by pneumococcal cells in vivo is oxidation. Pneumococci also produce H₂O₂ in amounts that can exceed 1 mM (which is 1,000-fold higher than the concentration needed to inhibit growth of E. coli cells) under aerobic and rich-nutrient conditions. In this way, resident pneumococci will kill or inhibit growth of other respiratory tract-colonizing flora, but growth of S. pneumoniae was not impaired even at the high levels of endogenously produced H₂O₂ (Pericone et al., 2003). Further, it was shown that pyruvate oxidase, which is the enzyme responsible for production of endogenous H₂O₂, also contributed to H₂O₂ resistance in pneumococci (Pericone et al., 2003). Although the ability of S. pneumoniae to cope with oxidative stress is still not well-understood, we explored whether pezAT could be involved in the response to this stress. This was done by exposing R6wt and R6ΔPezAT mutant to high concentration of exogenously added H₂O₂. The results (Figure 3) did not show any significant survival differences between the two strains at any of the tested H₂O₂ concentrations (20 and 40 mM), indicating that the pezAT operon was not involved in protecting the pneumococcal cells from oxidative stress. Another set of experiments were designed to test whether differences in biofilm formation between the wt and the mutant strains could be observed, since it has been shown that some TAs influence biofilm formation (Gonzalez Barrios et al., 2006; Wen et al., 2014).

Biofilm formation was quantified by measuring the OD₅₉₅. The bacterial growth at 37°C was similar in all the tested strains, and no significant differences were found in biofilm formation between the wt and the mutant even at 34°C, that is the nasopharynx temperature (Supplementary Figure S1).

The decreased response to lysis at stationary phase of the pezAT-deficient strain (Figure 2A), combined with the role of PezT on cell wall synthesis (Mutschler et al., 2011), prompted us to test whether deletion of the pezAT influenced the resistance/sensitivity of the pneumococcal cells only to β-lactam antibiotics or it was a more general stress response effect (Nieto et al., 2010). We used REMA plates and color change of resazurin (blue) to resorufin (pink) to compare the MICs of R6wt and R6ΔPezAT to antibiotics acting at different levels. The MICs are defined as the lowest concentration of the antibiotic that prevented color change, i.e., no bacterial growth (Andrews, 2001). By examining these changes, we found that the MICs of both strains to ampicillin was similar, 125 ng/ml (Figure 4A). However, at 63 ng/ml of ampicillin, the color changes from blue to pink was more prominent for the mutant than for the wt strain, indicating that the former strain was more resistant to ampicillin than the latter (Figure 4A). This observation was corroborated by measurement of the OD readings in which reduction of resazurin absorbance by the mutant strain (∼40%) was more prominent than the wt (∼24%) at 63 ng/ml of ampicillin after 3 h, and the differences were statistically significant (p-value < 0.05; Figure 4B). Similar results were observed when the MICs were estimated by the dilution method (Andrews, 2001; repeated four times and with a p-value < 0.05; not shown). Also similar results were observed for other β-lactam antibiotics, like benzetacil and penicillin G (not shown). No differences between wt and mutant strains were found when other antibiotics inhibiting gyrase (levofloxacin), or protein synthesis (streptomycin, tetracycline) were tested (not shown). Taken these results together, we conclude that lack of the pezAT operon does not induce a general stress response in S. pneumoniae, but that this response in focalized at the cell wall level.

Natural competence is a state of some bacterial species (the best known being S. pneumoniae and Bacillus subtilis) that is characterized by two main traits: (i) it is a bistable phenomenon, in which competence is expressed stochastically only in a part of the population, and (ii) the competent cells show growth arrest (Claverys et al., 2006; Hahn et al., 2015). Although, there has not a report showing a direct relationship between TA functionality and transformability, both processes share the two above features (Chan et al., 2012). To assess whether the pezAT operon influenced pneumococcal transformability, the two strains, mutant and wt, were tested for their transformation frequencies with DNA. To this end, we used competent cultures treated or not with CSP-1, and two different types of DNA. The first one consisted total chromosomal DNA (Chr) isolated from a pneumococcal strain that harbors a point mutation conferring resistance to streptomycin (Lacks, 1966). The second type of DNA (Str41) consisted of a 2002 bp-homogeneous DNA fragment PCR-amplified from Chr with primer pair rpsL_3/rpsL_4 (Supplementary Table S1); this fragment also harbors the above point mutation (Caymaris et al., 2010). As controls for genetic complementation, we used the same strains but harboring an ‘empty’ low-copy number plasmid (pC194r) or the same plasmid in which the entire pezAT operon (including its own transcription/translation signals) was cloned (pC104rPezAT). Both plasmids confer resistance to chloramphenicol.

Under homogenous growing environment and transformation procedures, with Chr DNA, the parental strain R6wt showed transformation frequencies of 0.3% in the absence of CSP-1, and increased to 1.1% when CSP-1 was added (Figure 5). As expected for homogeneous DNA (López et al., 1982), employment of Str41 DNA led to a substantial increase in the frequencies of transformation: 1.7% (without CSP-1) and 6.2% (with CSP-1; Figure 5). In the case of the mutant strain R6ΔPezAT, transformation frequencies augmented significantly (p-value < 0.05) ∼1.4- to 1.8-fold compared to the R6wt strain at all four different conditions used. Genetic complementation was observed when the strain used carried plasmid pC194rPezAT, but not with the negative control, ‘empty’ plasmid (Figure 5).

The above results showed that the transformability of S. pneumoniae increased in the strain devoid of the pezAT operon. However, they do not allow us to conclude that genetic competence was different in both strains. Consequently, we made use of the observation that induction of spontaneous competence strongly relies on the initial pH of the cultures (Chen and Morrison, 1987). To determine whether PezAT plays a role in regulation/development of competence, we used a gene cassette with a transcriptional fusion of the luciferase gene to the late competence gene ssb (Caymaris et al., 2010). The cassette was inserted into the chromosome of strains R6wt and R6ΔPezAT, thus constructing the two isogenic strains R6luc and the mutant R6ΔPezATluc. Media with initial pH ranging from 6.4 to 7.9 were used to investigate the time of occurrence and the level of competence of both strains. Under the same conditions, we observed that both R6wt and R6ΔPezAT mutant strains developed spontaneous competence only from pH 7.7 onward, and no prominent development of spontaneous competence was observed from pH 6.4 to 7.6 (only data from pH 7.6 to 7.9 are shown; Figure 6). Even though the R6luc and the R6ΔPezATluc strains developed spontaneous competence at the same initial pH (i.e., 7.7), the time of competence development and the magnitude of the competence peaks were different. At initial pH 7.7, the R6luc strain developed spontaneous competence at 140 min and peaked at 154 min with RLU/OD₄₉₂∼8.9 × 10⁴, whereas the R6ΔPezATluc mutant stain started at 112 min and reached the peak at 140 min with ∼6.1-fold higher magnitude. Similar results were observed at medium with initial pH 7.8, where R6luc parent strain started to develop spontaneous competence at 119 min and peaked at 140 min with RLU/OD₄₉₂∼3.8 × 10⁵; whereas the mutant began at 98 min and peaked at 119 min with ∼2.7-fold higher magnitude. For medium with initial pH 7.9, spontaneous competence started for the parent and the mutant strain at 98 and 77 min, respectively; they reached the competence peak at 119 and 98 min, respectively. Further, the ratio RLU/OD₄₉₂ for the mutant was 1.3-fold higher than the parent strain (RLU/OD₄₉₂∼1.0 × 10⁶). In summary, the mutant strain developed spontaneous competence much earlier and with higher magnitude, even though the growth rate for both strains was similar in medium with initial pH 7.7–7.9. Combining all these observations, we conclude that the pezAT operon has a significant influence on the competence development and hence, transformability of S. pneumoniae.