Strains and plasmids used in this study are shown in Table 1. Streptococcus pneumoniae strains were routinely grown in M17 agar containing 0.5% glucose and 1% (vol/vol) defibrinated sheep blood (Probiológica, Portugal) or as standing cultures without aeration in M17 broth (DifcoTM) supplemented with 0.5% glucose, at 37°C. Stocks and working stocks of isolated strains were prepared as described in (Carvalho et al., 2013b). When appropriate, 2.5 μg mL⁻¹ chloramphenicol, 0.25 μg mL⁻¹ erythromycin, 15 μg mL⁻¹ trimethoprim or 2.5 μg mL⁻¹ tetracycline were added to the medium. E. coli EC1000 was grown in Todd-Hewitt broth in a shaking incubator at 37°C; erythromycin was used at a concentration of 120 μg mL⁻¹. Lactococcus lactis NZ9000 was grown as standing cultures in TSB medium at 30°C; chloramphenicol and erythromycin were used at concentrations of 5 and 4 μg mL⁻¹, respectively.

The D39 spontaneous mutant (D39SM) was isolated in our laboratory by plating out a GM17 liquid grown working stock of strain D39 on GM17-agar plates that resulted in the appearance of two colony phenotypes: large and opaque (encapsulated D39) and small and transparent (non-encapsulated D39SM), in the proportion of circa 100:1, respectively, after an overnight growth at 37°C and 5% (vol/vol) CO₂. Both isolated strains were then purified by two sequential passage steps; in each step, one single colony of each isolate was selected and streaked onto GM17-agar plates. Genomic DNA was extracted from each strain and their whole genome was sequenced as described below.

Streptococcus pneumoniae was grown in the chemically defined medium (CDM) described in (Carvalho et al., 2013b), which contains 10 mg L⁻¹ uracil (82 μM). S. pneumoniae cultures were started at an initial optical density at 600 nm (OD₆₀₀) of 0.05–0.06 by addition of preculture cells harvested in exponential phase (OD₆₀₀ 0.8–1.0), centrifuged and suspended in fresh CDM without uracil. Cultivations were performed in static rubber-stoppered bottles at 37°C without pH control (initial pH 6.5). Glucose 1% was used as the carbon source. Growth was monitored by measuring OD₆₀₀ every hour. Maximum specific growth rates (μ_max) were calculated through linear regressions of the plots of ln(OD₆₀₀) versus time during the exponential growth phase. The effect of uracil deprivation on growth of the pneumococcus strains was assessed by cultivating cells as above, except that uracil was omitted from the medium. For each growth condition at least three independent experiments were performed. The plotted growth curves throughout the manuscript are from a representative experiment and the error in each point was in all cases below 15%.

For the complementation studies S. pneumoniae cells were grown in 250 μL CDM supplemented or not with uracil (10 mg L⁻¹) in 96-well microtiter plates, at 37°C. Cultures were started at an initial OD₆₀₀ of 0.1, by addition of an exponentially growing preculture suspended in fresh CDM without uracil, and monitored every 30 min at 600 nm with a MultiSkan GO plate reader (ThermoScientific). Optical densities at 600 nm were registered by the SkanIt Software 3.2 (ThermoScientific). Nisin (0 and 2 ng mL⁻¹) was added at time zero.

D39 and D39SM were grown in CDM supplemented or not with uracil (10 mg L⁻¹) in monoculture or co-culture. Both monocultures and co-cultures were started at an initial OD₆₀₀ of 0.06 ± 0.03 by addition of exponentially growing precultures suspended in fresh CDM without uracil, as described above. In the co-culture experiments, with and without uracil, D39 and D39SM were inoculated at a cell ratio of 4:1, 1:1, and 1:4. For each condition at least three independent growth experiments were performed. The plotted growth curves are from a representative experiment and the error in each point was in all cases below 15%. Colony forming units per mL (CFUs mL⁻¹) of D39 and D39SM were determined at time-point zero of inoculation (OD₆₀₀ of 0.06 ± 0.03) as follows: suspensions were diluted 1,000 times in CDM without uracil and plated (2 μL) in M17 blood agar containing 0.5% glucose. For the determination of strain relative abundance, samples in mid-exponential phase (OD₆₀₀ of 0.48 ± 0.05) were collected, diluted 100 times and plated (100 μL) in M17 blood agar containing 0.5% glucose. Plated cells were always grown at 37°C in a 5% (vol/vol) CO₂ incubator. The estimation of the relative abundances of the D39 and D39SM populations was determined by cell counting (CFUs mL⁻¹). Colonies of each strain were selected on the basis of their phenotypic appearance on plate, i.e. D39 colonies are opaque and bigger than D39SM colonies, which are also transparent.

All the plates were scanned with white light epi-illumination and stored as digitized image files with a Molecular Imager® ChemiDocTM XRS+ Imaging System (Bio-Rad). The Quantity one software package was used to adjust the contrast, thus enhancing the visualization of the colonies on the plates.

Chromosomal DNA isolation was performed according to the procedure of Johansen and Kibenich (1992) or alternatively, with the NZY Tissue gDNA Isolation kit from NZYTech. Plasmid isolation was carried out using plasmid isolation kits from Roche or Qiagen. Pwo DNA polymerase (Roche), Phusion DNA polymerase (Thermo Scientific), T4 DNA ligase (Biolabs, New England) and restriction enzymes (Biolabs, New England) were used according to the supplier's recommendations. Purification of the PCR products was performed using the High pure PCR product purification kit from Roche or alternatively, the QIAquick PCR purification kit from Qiagen. Purified PCR products or recombinant plasmids were introduced into S. pneumoniae by transformation as described in Kloosterman et al. (2006a). Positive transformants were selected in M17 blood agar or Brain Heart Infusion (BHI) agar containing 0.5% glucose with the appropriate antibiotic and confirmed by PCR and sequencing. L. lactis NZ9000 was transformed with plasmid DNA by electroporation as described before (Carvalho et al., 2011).

The genomic region comprising 1,453-bp before and 396-bp after the starting codon (ATG) of the cps2A gene was PCR amplified with primers P1_capsule to P5_capsule and sent for sequencing (Service XS, Leiden, The Netherlands). Streptococcus pneumoniae gDNA extracted according to the procedure of Johansen and Kibenich (1992) (Johansen and Kibenich, 1992) was sent for sequencing to the Genome Analysis Facility, Dept. of Genetics, University Medical Centre Groningen, The Netherlands.

The oligonucleotide primers used in this study are listed in Table S8. Thymine, located 30 nucleotides upstream of the initiation codon (ATG) of the cps2A gene, was replaced in the TATAAT binding box of the cps locus with a cytosine, thereby leading to TATAAC (Figure 2A), as follows. Up- and downstream DNA fragments of 654- and 635-bp, respectively, comprising the desired mutation were obtained by PCR amplification with the primer pairs Pcapsule-1/Pcapsule-mut2 and Pcapsule-2/ Pcapsule-mut1, using D39 wild-type DNA as a template. The internal primers, Pcapsule-mut2 and Pcapsule-mut1, comprise the point mutation leading to the TATAA (T→C) change. The resulting PCR products were fused by means of overlap extension PCR (Song et al., 2005) using the primers Pcapsule-1 and Pcapsule-2, leading to the formation of a DNA fragment (1,284-bp) containing the altered −10 binding box (TATAAC). This fragment was cloned into the EcoRI/BamHI sites of pORI280, forming the pORI[Pcps_T→C] construct. Escherichia coli EC1000 was used as cloning host. This construct was used to introduce the point mutation in the −10 promoter region of the cps locus of the chromosome of strain D39 following the procedure described previously by Kloosterman et al. (2006a). The resulting strain was designated D39Pcps_T→C.

The point mutation in carA was constructed by applying the same method as for Pcps. In brief, the primers pairs carA-1/carA-mut2 and carA-2/carA-mut1 were used to obtain up- and downstream fragments of 634- and 618-bp, respectively, by PCR amplification. These fragments, containing the desired mutation (C→A at position 710-bp of the carA gene), were fused and cloned into the EcoRI/BamHI sites of pORI280 in E. coli EC1000, yielding pORI[carA_C→A]. The carA point mutation in D39 (carA_C→A) was obtained as in Kloosterman et al. (2006a).

The carA complementation strain was constructed in S. pneumoniae using the NICE system described before (Carvalho et al., 2011). The carA mutant strain (D39carA_C→A) was transformed with D39nisRK chromosomal DNA, which contains the nisRK genes and a Trim^R gene in the bgaA locus. Clones resistant to trimethoprim (D39carA_C→AnisRK, Table 1) were selected and the presence of the carA mutation was confirmed by sequencing (STAB VIDA). The plasmid harboring carA under the control of the nisin-indusible promoter nisA (pNZ[carA], Table 1) was constructed as follows. The carA gene was PCR-amplified from D39 gDNA using the primer pair carA_fw and carA_rev and cloned as a BspHI/XbaI fragment into NcoI/XbaI digested pNG8048e. L. lactis NZ9000 was used as the cloning host. D39carA_C→AnisRK strain was transformed with pNZ[carA] and the empty plasmid pNZ8048. Transformed strains were selected for chloramphenicol resistant clones and confirmed by PCR.

An ectopic transcriptional fusion of the capsule locus promoter (Pcps) with the lacZ reporter gene was constructed using the pPP2 plasmid (Halfmann et al., 2007). A DNA fragment comprising Pcps was PCR-amplified from D39 chromosomal DNA using the primer pair Pcapsule-1/Pcapsule-2. This fragment (1284-bp) was cloned into the EcoRI/BamHI sites of pPP2 and the resulting plasmid (pPP2[Pcps]) was integrated via double cross-over into the bgaA locus of strains D39, D39SM, and D39carA_C→A, yielding strains D39pPP2[Pcps], D39SMpPP2[Pcps] and D39carA_C→ApPP2[Pcps], respectively. To construct a lacZ fusion with the mutated cps locus promoter (Pcps_T→C), the PCR fragment containing Pcps_T→C, generated as described above, was cloned into the EcoRI/BamHI sites of pPP2. The resulting plasmid (pPP2[Pcps_T→C]) was integrated via a double cross-over event into the bgaA locus of strains D39 and D39SM, leading to the formation of strains D39pPP2[Pcps_T→C] and D39SMpPP2[Pcps_T→C], respectively.

Cells containing the promoter-lacZ fusions were grown in CDM containing 1% glucose, with or without uracil, as described above. Culture samples of 2 mL were harvested during mid-exponential phases of growth and the activity of the promoters was assayed by measuring β-galactosidase activities as described before by Kloosterman et al. (2006a).

Strains D39SM and D39Pcps_T→C were compared to S. pneumoniae D39 by transcriptome analysis using whole-genome S. pneumoniae DNA microarrays, representing all TIGR4 open reading frames (ORFs), as well as R6 and D39 ORFs that are not in TIGR4 (Kloosterman et al., 2006b). Each strain was grown in triplicate in CDM containing 1% glucose and with or without uracil as described above. Cells for RNA isolation were harvested in mid-exponential phase of growth (Figure 1B). mRNA isolation, synthesis of cDNA and labeling of cDNA was performed as described in Carvalho et al. (2011). All samples were hybridized in duplicate in the spotted DNA microarrays. Scanning of the slides and the processing and analysis of the data was carried out as described before (Carvalho et al., 2011). Genes were considered to have significantly altered expression when the bayesian p-value was <0.0001 and the ratio (signal D39SM /signal D39 or signal D39Pcps_T→C /signal D39) was >3 and <-3. An in silico comparison was done between conditions with and without uracil in both D39 wild-type as well as D39SM. To this end, raw signals (i.e., slide images) of the experiments with uracil were compared with the corresponding signals of the experiments without uracil and analyzed as described above. The microarray data is deposited in GEO under the accession number GSE109129.

Samples for the determination of the capsular glucuronic acid amounts were grown in CDM supplemented with 1% glucose, with or without uracil, as described above, harvested during mid-exponential phases of growth (Figure 1B) and treated as in (Carvalho et al., 2011). In brief, cells were centrifuged (6,000 x g, 3-7 min, 4°C), resuspended in PBS and pelleted at 3000 x g, 4°C, for 20min. The pellet was resuspended in 500 μL of 150 mM Tris-HCl (pH 7.0) and 1 mM MgSO₄ and treated as described elsewhere (Morona et al., 2006). The supernatants were treated with 20% (wt/vol) trichloroacetic acid for protein precipitation and cold ethanol for exopolysaccharide precipitation as in Ramos et al. (Ramos et al., 2001). The glucuronic acid of capsule was quantified by the method for quantitative determination of uronic acids as described by Blumenkrantz and Asboe-Hansen (1973).

Streptococcus pneumoniae was grown in CDM with or without uracil, as described above, and cells were harvested at mid-exponential phase of growth for the determination of intracellular UTP. The UTP content was measured by one-dimensional thin-layer chromatography (TLC) as described in Jendresen et al. (2011).

The statistical analyses (p-values) were obtained by using the two tailed unpaired student's t-test of GraphPad Prism software (GraphPad version 5.01, San Diego, CA), with a confidence interval of 95%. Data is presented as mean ± standard deviation (SD).

Unexpectedly, streaking a single GM17 liquid grown working stock, which was started from a −80°C stock of D39, on GM17-agar plates resulted in the appearance of two colony phenotypes. One colony type displayed the typical characteristics of the D39 Lilly isolate (Lanie et al., 2007), i.e., large and opaque colonies. The other colony type, hereafter denominated D39SM (D39 spontaneous mutant), exhibited a substantially smaller colony size and higher transparency than D39 (Figure 1A, Image S1). A single colony isolate of both the typical D39 Lilly phenotypic strain and the phenotypic variant was selected for further characterization. Individual growth and restreaking of the two isolates, D39 and D39SM, in GM17 liquid medium and agar plates always gave rise to colonies exhibiting their parental phenotype. The phenotype of D39SM colonies, resembling that of strain R6 (Figure 1A, Image S1), could be due to a considerably lower growth rate or to a decreased capsule production (Kim et al., 1999; Carvalho, 2012). Both wild-type and D39SM strains were grown in CDM containing 1% glucose, as described in Material and Methods, and the amount of capsule was determined in the mid-exponential phase of growth (D39, OD₆₀₀ of 0.40 ± 0.00; D39SM, OD₆₀₀ of 0.48 ± 0.02; Figure 1B). The maximal growth rate of strain D39SM was 1.4-fold lower than that of strain D39 (Figure 1B). Interestingly, this decrease in the growth rate is in contrast with the 1.4-fold increase in the final biomass reached by D39SM strain relative to the parent strain (Figure 1B). Moreover, the amount of capsule in D39SM was 12.4-fold lower relative to D39 levels (unpaired t-test, p-value < 0.0001; Figure 1C). Considering these data, the small colony and high transparency phenotype most likely reflects the substantial reduction in capsule amount, rather than the slight decrease in growth rate. Interestingly, the amounts of glucuronic acid in D39SM were similar to those of R6, the non-encapsulated derivative of strain D39, showing that D39SM is also nearly or entirely devoid of capsule (Figure 1C). Further analysis of the original culture from which D39SM was isolated revealed a frequency of D39SM colonies approximately 100-fold lower than that of the wild-type colonies. Moreover, the growth profile of other D39SM siblings isolated from the original culture was similar to that of D39SM that we selected for further characterization (data not shown). The scarce number of mutant colonies suggests that their appearance was an exceptional event that occurred exclusively during the preparation of this working stock. Supporting this notion, the only phenotype recovered upon growth of any other of our working stocks in Glc-M17 agar was that of the progenitor D39 strain.

The appearance of a D39 spontaneous mutant, phenotypically similar (acapsular) to strain R6, without inflicted selection pressure or artificial manipulation, during an overnight growth on GM17, prompted us to understand the origin of D39SM. We envisaged two possible scenarios: (i) unknown factor(s) drove natural selection pressure on D39 causing spontaneous mutation(s) on this strain; (ii) a contamination of the D39 GM17 culture with strain R6, used routinely in our laboratory, led to extensive genome recombination between both strains. Extensive genome recombination between two bacterial strains generally results in recombination phenomena, such as DNA duplication and deletion, and in chromosomal rearrangements, namely DNA inversion. Thus, we performed whole-genome sequencing analysis of one single colony of strains D39SM and D39 to gain better insight into the origin of D39SM.

The whole-genome sequencing analysis revealed that D39SM genome contains 105 mutations relative to the genome of D39 (Tables S1, S2). Moreover, 80% of the mutations in D39SM are non-synonymous, nonsense and frameshifts, thus altering amino acid sequences of protein products. Strikingly, ~60% of D39SM mutations match those observed for the non-encapsulated strain R6 (Table S2), including mutations in the genes encoding functions of the de novo synthesis of pyrimidines, namely carA, carB and pyrDa, previously linked to capsule production (Carvalho, 2012; Carvalho et al., 2013b), and one point mutation in the capsule locus promoter (Pcps). However, the first nine genes of the capsule cluster (cps2ABCDETFGH) of D39SM are completely intact, while in strain R6 cps2BCDETFG and cps2A/cps2H genes are fully and partially deleted, respectively (Iannelli et al., 1999), which indicates that: (i) the acapsular phenotype of D39SM is not caused by deletion(s) within the cps locus or point-mutations in the cps operon, particularly in the initial five genes cps2A-E (Yother, 2011; Schaffner et al., 2014; Shainheit et al., 2015; Wen et al., 2016) and (ii) D39SM is not strain R6. Moreover, about 87% of D39SM mutations are single-nucleotide polymorphisms (SNPs), while duplicated and inverted DNA is absent and only a few DNA deletions were detected (Table S2). Thus, most likely an unknown factor imposing selection pressure on D39 led to the appearance of D39SM rather than an overnight extensive recombination between strains R6 and D39. In fact, many other mutations (40%) of D39SM, presumably required for its proper functioning, are not present in strain R6.

As revealed by whole-genome sequencing analysis, D39SM presents an intact cps operon, thus we hypothesized that the diminished capsule production in this strain could be due to a point mutation in the promoter region of the capsule cluster (Pcps) (Moscoso and García, 2009; Yother, 2011; Schaffner et al., 2014; Shainheit et al., 2015; Wen et al., 2016). A correlation between decreased cps expression and polymorphisms in the promoter of the cps gene cluster in S. pneumoniae is not unprecedented (Moscoso and García, 2009; Yother, 2011). The whole-genome sequencing analysis revealed a single point mutation in the −10 promoter region localized 30 nucleotides upstream of the starting codon of cps2A, and leading to the transition of the consensus sequence TATAAT to TATAAC (Pcps_T→C) (Figure 2A), which was further validated by sequencing of the dexB-cps2A genomic region. Interestingly, the same base substitution is present in the Pcps of the non-encapsulated D39 derivative strain R6 (Hoskins et al., 2001; Moscoso and García, 2009). In addition, considering that strain R6 harbors a large deletion (7.5 kbp) in the cps locus, most likely the events leading to the arising of the two strains were distinct. To investigate whether the decreased capsule production was mediated at the transcriptional level by Pcps_T→C mutation, the activity of cps and cps_T→C promoters was measured. Fusions of these promoters with the lacZ reporter gene were generated in the pPP2 plasmid (Halfmann et al., 2007) and introduced into D39 and D39SM backgrounds. The D39pPP2[Pcps], D39pPP2[Pcps_T→C], D39SMpPP2[Pcps] and D39SMpPP2[Pcps_T→C] strains generated in this way were grown in CDM containing 1% glucose, harvested in exponential phase and β-galactosidase activities were determined. The activity of the cps_T→C promoter was decreased by 18-fold relative to the activity of the Pcps in D39 and D39SM strains (unpaired t-test, p-value = 0.001 and 0.0008, respectively) (Figure 2B). Even though data are only shown for mid-exponential phase of growth (Figure 2B), low promoter activity was observed throughout growth (data not shown). This suggests that the reduction in capsule production is caused solely by the Pcps_T→C mutation. To strengthen our hypothesis, the Pcps_T→C point mutation was generated in the chromosome of strain D39 by using the pORI280 plasmid as described in section Materials and Methods. The amount of capsule in the D39Pcps_T→C mutant, as estimated from the GlcUA method, was similar to that in D39SM and 8-fold lower than that of the parent strain D39 (unpaired t-test, p-value = 0.0001; Figure 1C). Furthermore, in Glc-M17 blood agar plates the morphology of D39Pcps_T→C colonies totally resembled that of D39SM (Figure 2C, Image S2). These data provide evidence that the Pcps_T→C mutation is responsible for reduction of capsule synthesis. The additional mutations in D39SM are responsible for its lower maximal growth rate and final biomass when compared to those of strain D39Pcps_T→C (Figure 1B), and the non-production of capsule is responsible for the higher biomass of D39Pcps_T→C and D39SM relative to D39, consistent with capsule being an energetic burden (Hathaway et al., 2012) (Figure 1B).

The effects of D39SM mutations and the Pcps mutation in D39 were analyzed by whole-genome transcriptome analysis comparing mRNA levels in the mutant strains to those in strain D39. The total number of genes that showed significant altered expression in D39Pcps_T→C versus D39 microarray, according to the cut-off criteria applied, was 5.8-fold lower than that observed for D39SM versus D39 microarray (Tables S3, S4), which confirms a major influence of the mutations other than Pcps_T→C in the D39SM phenotype. As expected from our targeted analysis, the cps genes (spd_315-18, spd_321-2) were strongly downregulated in both mutant strains (Tables S3, S4). In D39Pcps_T→C the transcriptional response was modest and besides capsule repression, the most significant alterations were the upregulation of the spd_0277-83 operon, coding for a 6-phospho-beta-glucosidase, a lactose-type phosphoenolpryruvate:carbohydrate phosphotransferase system (PTS)- transporter and a multi-domain transcriptional regulator, presumably involved in the transport of beta-glucosides (Bidossi et al., 2012), and the downregulation of three hypothetical proteins (spd_0486, spd_2001, spd_2038) and of the spd_1514-6 operon, which includes an ATP-binding cassette (ABC)- transporter and two hypothetical proteins (Table S4). The transcriptional response of strain D39SM affected circa 5% of the whole genome, i.e., 100 genes. Most of the genes (35 genes) showing altered expression belonged to the carbohydrate transport and metabolism category. In addition, many genes with unknown (7 genes) or unpredicted (21 genes) function, as well as a number of genes (9 genes) with general function prediction only, were among the differentially expressed genes (Table S3). Noteworthy, eight PTS uptake systems (spd_0066-9, spd_0502, spd_0559-61, spd_0661, spd_1047-8, spd_1057, spd_1959, and spd_1992-89), and the enzymes involved in the metabolism of the PTS-incoming sugars (spd_0063, spd_0065, spd_0070-1, spd_0503, spd_0562, spd_1046, spd_1958-7, and spd_1995-4) were all strongly induced in the mutant strain (Table S3). The spd_0277-83 operon induced in the D39Pcps_T→C mutant was, however, not among the upregulated genes in strain D39SM. The PTS transporters upregulated in D39SM are hypothetically involved in the uptake of sugars (galactose, beta-glucosides, lactose, galactitol, maltose, tagatose, fucose, L-arabinose) other than glucose (Table S3) (Bidossi et al., 2012). RegR, another global regulator of the family LacI/LacR, involved in carbohydrate metabolism, was upregulated in D39SM. In addition to the PTS transporters, the catabolic enzymes, and RegR, the altered expression of two other genes called our attention in strain D39SM. The first gene (spd_0267) encodes a permease that putatively transports xanthine/uracil and the second gene encodes a putative dihydroorotase (pyrC, spd_1030), which is known in other species to dehydrate carbamoyl aspartate (CAA) forming dihydroorotate (DHO) in the de novo pathway for pyrimidine synthesis (Kilstrup et al., 2005). pyrC is part of a transcriptional unit comprising two other genes, ung and mutT, also induced in the mutant strain, and encoding functions unrelated to the de novo synthesis of pyrimidines (Table S3). Dihydroorotate is an early precursor for the synthesis of UMP, which is subsequently converted to UTP via the salvage pathway for pyrimidine biosynthesis (Kilstrup et al., 2005). We have previously shown that growth of S. pneumoniae D39 is limited by the nucleobase uracil (Carvalho, 2012; Carvalho et al., 2013b). Since the un-encapsulated strain R6 was rather insensitive to uracil and only marginally affected when the nucleobase was omitted from the cultivation medium, we speculated that the effects on D39 were related to capsule synthesis, a process that consumes UTP (Carvalho, 2012; Carvalho et al., 2013b). In this context, the differential regulation of genes involved in the pyrimidine biosynthetic pathways, a putative uracil transporter and pyrC (Table S3), in D39SM, a strain impaired in capsule synthesis and mutated in the de novo synthesis of pyrimidines genes carA, carB and pyrDa, when uracil was present in the medium was intriguing. Furthermore, repression of the xanthine/uracil permease gene or induction of the pyrC gene was not observed in the D39Pcps_T→C strain (Table S4), which is also deficient in capsule production.

The mutations and altered expression of genes involved in pyrimidine biosynthetic pathways further reinforces the assumption of a tight connection between uracil metabolism and capsule production.

Strains D39SM and D39Pcps_T→C are “quasi” or non-encapsulated (Figure 1C), and consequently should have a lower requirement for uracil/UTP, as shown for strain R6 (Carvalho, 2012; Carvalho et al., 2013b). The influence of uracil on capsule production was evaluated by cultivating strains D39, D39SM and D39Pcps_T→C in the absence of the nucleobase and comparing with the results obtained in complete medium. In the absence of uracil, strains D39SM and D39Pcps_T→C presented growth profiles and capsule amounts similar to those in medium with uracil (Figures 1B,C). In contrast, strain D39 was profoundly affected by the omission of uracil: growth was characterized by a lag of about 6 h, a growth rate of 0.42 ± 0.01 h⁻¹, 2-fold lower than in medium containing uracil, and a maximal biomass 1.8 ± 0.0, which represented a 20% increase as compared to the standard conditions (uracil 10 mg L⁻¹) (Figure 1B). Furthermore, strain D39 exhibited a significant 22% decrease (unpaired t-test, p-value = 0.0002) in capsule amount relative to that produced in medium containing uracil (Figure 1C). Our data show a subtle coupling between uracil metabolism, capsule production and growth of strain D39. In agreement, the growth dependency on uracil is lost in un-encapsulated strains.

To gain further insight into the effect of uracil in S. pneumoniae, we grew strains D39SM and D39 in medium lacking uracil as in Figure 1B, and harvested samples at mid-exponential phase for transcriptome analysis (Figure 1B). In the absence of uracil, the D39SM versus D39 transcriptional response affected 4.2% of the whole genome (2069 genes). Of the genes showing altered transcript levels, 87% were common in medium with and without uracil (Tables S3, S5). An in silico DNA microarray analysis comparing D39 grown in CDM with and without uracil revealed a strong derepression of genes involved in de novo synthesis of pyrimidines [spd_0608-9 (pyrFE) and spd_0851-2 (pyrKDb)] in the absence of uracil (Table S6). Consistent with activation of the genes of the de novo synthesis of pyrimidines, the gene encoding a putative PyrR regulatory protein (spd_1134), an homolog of the lactococcal PyrR protein known to activate the expression of pyrimidine biosynthetic genes (Kilstrup et al., 2005), was also upregulated in S. pneumoniae strain D39 when uracil was omitted (Table S6). Our in silico data are in good agreement with the accepted view for other Streptococcaceae (Kilstrup et al., 2005).

The in silico DNA microarray analysis comparing D39SM grown in CDM with and without uracil showed no significant altered expression of genes (a cut-off of >3 or <-3 was used for the fold change and a Bayesian p-value ≤ 0.0001 to consider fold change statistically significant). Similarly, the absence of uracil did not induce major changes in the transcriptome profile of strain D39Pcps_T→C when compared with its parent strain D39, exceptions being the operon spd_1514-6 and gene spd_0109 (Tables S4, S7). Overall, these results show that uracil (or pyrimidines) has no major effect on the transcriptome of strains D39SM and D39Pcps_T→C, while inducing a considerable transcriptional response in the capsulated strain D39.

In strain D39 the amounts of capsule decreased in medium lacking uracil. To investigate whether this effect was mediated at the transcriptional level we measured the activity of the native and mutated cps promoters in cells grown without uracil (Figure 2B). Interestingly, in the absence of uracil the activity of the native cps promoter measured in D39 cells collected at mid-exponential phase was 25% lower than in D39 grown in medium with uracil (Figure 2B), a decrease similar to that observed for the reduction of capsule in strain D39 in uracil depleted conditions (Figure 1C). As expected, uracil had no effect on the activity of the cps_T→C promoter, independently of the genetic background (Figure 2B). Thus, the observed reduction in capsule production is presumably controlled at the transcriptional level through modulation of the cps promoter activity by uracil or an intracellular derivative, such as UMP or UTP. In the capsulated strain D39, the UTP pool was higher in medium containing uracil (data not shown). The same pattern was observed for strain D39SM (data not shown), but since this variant produced little or no capsule, its requirements for UTP are lower, and thus growth is not affected.

Our data provide evidences for a link between uracil/pyrimidine metabolism and capsule synthesis. Furthermore, we show that a spontaneous mutant with non-producing capsule phenotype contains several mutations in genes of the de novo synthesis of pyrimidines, namely carA, carB, and pyrDa. To show the link between pyrimidine metabolism and capsule production we measured capsule promoter expression in mid-log grown cells of strain D39 mutated in one of the genes that code for the enzymes of the de novo synthesis of pyrimidines found to be mutated in the spontaneous mutant. The gene selected was carA, that codes the small subunit of carbamoyl-phosphate synthase, the first enzyme of the de novo synthesis of pyrimidines pathway, that in D39SM contains an adenine instead of a cytosine localized 707 nucleotides downstream the starting codon, thereby leading to the translation of a valine instead of a glycine. This point mutation found in the D39SM strain was constructed in D39 background leading to the formation of D39carA_C→A. Remarkably, carA mutation led to a significant 2-fold decrease in capsule promoter expression in comparison with the wild-type strain, during the exponential phase in medium without uracil (Figure 3). In contrast, in medium containing increasing concentrations of uracil, such as 5 and 10 mg L⁻¹, no significant differences in capsule promoter expression were observed between the wild-type and the carA mutant (Figure 3), which is consistent with the presence of uracil in the medium and thus with activation of the salvage pathways for the production of UMP/UTP. Surprisingly, as for strains D39SM and D39Pcps_T→C, the growth profile of the carA mutant revealed to be insensitive to the presence of uracil (Figure S1), suggesting that somehow the de novo synthesis of pyrimidines was already induced in the mutant strain. Importantly, growth of the D39carA_C→A mutant with expression in trans of carA under the control of the nisin promoter, resembled that of D39 in medium devoid of uracil (Figure S1). It should be noted that the growth profile of D39 carA_C→AnisRKpNZ8048 was similar to that of the D39carA_C→A mutant (data not shown).

Bacterial spontaneous mutations generally occur during adaptation to a stressful environment. In this context, it is likely that loss of the ability to produce capsule conferred the resulting phenotype with a competitive advantage. Considering the established interplay between uracil metabolism and capsule production we surmised that the mutant strain would perform better in uracil-devoid environments. To verify our hypothesis D39SM and its parent D39 were grown in co-cultivation in medium with and without uracil (Figure 4). Strain abundance in the co-cultures was assessed by the CFU counts of samples harvested in mid-exponential phase of growth (Table 2; Figure S2). In medium containing uracil, in mid-exponential growth the most abundant species was the one predominating at the time of inoculation (Table 2; Figures S2A,B, Images S3–S8). Moreover, the dominant strain in the co-culture dictated to a great extent the growth profile of the mixture (Figure 4A). Curiously, a D39:D39SM ratio of 1 at the beginning of growth resulted in a very slight prevalence of strain D39 at mid-exponential phase (ratio D39:D39SM of 1.34, Table 2); the growth rate of the co-culture was more similar to that of D39, but the maximal biomass reached values similar to those achieved by D39SM (Figure 4A; Figures S2A,B, Images S3–S8). A possible explanation resides on the ability of D39SM to grow well in the absence uracil, which at 10 mg L⁻¹ is limiting for strain D39. Notably, in uracil-free medium at mid-exponential growth strain D39SM prevailed over D39, independently of the inoculation ratio (Table 2; Figures S2C,D, Images S9–S14). The predominance of strain D39SM in uracil-deprived medium resulted in maximal biomass of about 2.1 (ODMax) in the different co-cultures, values similar to those reached by strain D39SM alone (Figure 4B). In uracil-free medium, the average growth rate of the co-cultures also reflected the predominance of strain D39SM (Figure 4).

Our data show that D39SM presents competitive fitness over strain D39 in uracil-free conditions. This advantage most likely derives from the low requirement of the mutant strain for uracil due to impairment in capsule production. In conclusion, our data indicate that in uracil poor media capsule production is a tremendous nutritional burden.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.