Physiological Roles of the Dual Phosphate Transporter Systems in Low and High Phosphate Conditions and in Capsule Maintenance of Streptococcus pneumoniae D39

Unlike most bacteria, Streptococcus pneumoniae (pneumococcus) has two evolutionarily distinct ABC transporters (Pst1 and Pst2) for inorganic phosphate (Pi) uptake. The genes encoding a two-component regulator (PnpRS) are located immediately upstream of the pst1 operon. Both the pst1 and pst2 operons encode putative PhoU-family regulators (PhoU1 and PhoU2) at their ends. This study addresses why S. pneumoniae contains dual Pi uptake systems and the regulation and contribution of the Pst1 and Pst2 systems in conditions of high (mM) Pi amount and low (μM) Pi amount. We show that in unencapsulated mutants, both pst1 and pst2 can be deleted, and Pi is taken up by a third Na+/Pi co-transporter, designated as NptA. In contrast, either pst1 or pst2 is unexpectedly required for the growth of capsule producing strains. We used a combination of mutational analysis, transcript level determinations by qRT-PCR and RNA-Seq, assays for cellular PnpR~P amounts by SDS-PAGE, and pulse-Pi uptake experiments to study the regulation of Pi uptake. In high Pi medium, PhoU2 serves as the master negative regulator of Pst2 transporter function and PnpR~P levels (post-transcriptionally). ΔphoU2 mutants have high PnpR~P levels and induction of the pst1 operon, poor growth, and sensitivity to antibiotics, possibly due to high Pi accumulation. In low Pi medium, Pst2 is still active, but PnpR~P amount and pst1 operon levels increase. Together, these results support a model in which pneumococcus maintains high Pi transport in high and low Pi conditions that is required for optimal capsule biosynthesis.


INTRODUCTION
Phosphorus is an essential element in all cells because of its structural and metabolic roles in nearly all biological processes, including the composition of nucleic acids, phospholipids, and energy intermediates. A preferred source of phosphorous for bacterial cells is environmental inorganic orthophosphate (PO − 4 ; P i ). The mechanism of extracellular P i uptake has been studied intensively in Escherichia coli and Bacillus subtilis as model organisms (Hulett, 1993;Takemaru et al., 1996;Wanner, 1996;Qi et al., 1997;Lamarche et al., 2008;Hsieh and Wanner, 2010;Botella et al., 2011Botella et al., , 2014, and recently in other bacterial species (Braibant et al., 1996;Gonin et al., 2000;Zaborina et al., 2008;Rifat et al., 2009;Shi and Zhang, 2010;Burut-Archanai et al., 2011;Cheng et al., 2012;Wang et al., 2013;de Almeida et al., 2015;Lubin et al., 2016). Generally, bacterial highaffinity P i uptake systems consist of an ATP-binding cassette (ABC) transporter, designated as Pst (for phosphate-specific transporter), which contains at least four component subunits: an extracellular P i binding protein (PstS), two transmembrane channel proteins (PstCA), and a cytoplasmic ATPase (PstB) (see Figure 1; Hsieh and Wanner, 2010). The expression of most bacterial Pst transporters is regulated at the transcriptional level by a two-component regulatory system (TCS), which has different designations in different bacteria (Hulett, 1993;Novak et al., 1999;Throup et al., 2000;Howell et al., 2006;Glover et al., 2007). Many bacteria also regulate P i uptake by an ancillary negative regulatory protein, designated PhoU (Steed and Wanner, 1993;Botella et al., 2011Botella et al., , 2014de Almeida et al., 2015;Lubin et al., 2016). In E. coli and related enterobacteria, the histidine kinase (HK) and response regulator (RR) that mediate P i transport are designated as PhoR and PhoB, respectively, and the regulation of P i uptake involves a PhoU regulator (Hsieh and Wanner, 2010;Gardner et al., 2014Gardner et al., , 2015. Briefly, when [P i ] > 4.0 µM, the expression of the phoB-phoR regulator and pst transporter operons is inhibited by PhoU by a mechanism described below (Hsieh and Wanner, 2010). When the [P i ] is depleted to <4.0 µM, PhoU releases inhibition of the PhoR HK and the PstB subunit of the transporter, allowing autophosphorylation of the PhoR HK, phosphoryl transfer to the PhoB RR, and activation of transcription by PhoB∼P of operons in the phosphate (pho) regulon, including the pst transporter, the phoB-phoR regulator, and other operons involved in the uptake and assimilation of phosphorous-containing compounds (Wanner, 1996;Hsieh and Wanner, 2010). PhoB∼P activates transcription by binding to the Pho box sequence upstream from the promoters of the regulon operons, including phoB-phoR, which provides autoregulation of the TCS proteins (Wanner, 1996;Martin, 2004;Lubin et al., 2016). Usually, the sequence of the consensus Pho box is different between Gram-negative (e.g., E. coli) and Grampositive (e.g., B. subtilis) bacteria (Martin, 2004). Since the Pst transporter is not activated in E. coli at high P i concentrations, this system is considered as a high-affinity transporter that works predominantly at low P i concentrations (Wanner, 1996). PhoB/R, the Pst transporter, and members of the Pho regulon are important for virulence in many pathogenic Gram-negative bacteria, including E. coli, Vibrio cholerae, Proteus mirabilis, and Pseudomonas aeruginosa (Jacobsen et al., 2008;Lamarche et al., 2008;Zaborina et al., 2008;Pratt et al., 2010;Chekabab et al., 2014a,b). In P. aeruginosa, the PstS P i binding protein also plays roles in adhesion and a P i -independent role in biofilm formation (Zaborina et al., 2008;Neznansky et al., 2014;Shah et al., 2014).
PhoU is a negative regulator of Pho regulon expression in E. coli and many other bacteria (Muda et al., 1992;Wanner, 1996;Hsieh and Wanner, 2010;Gardner et al., 2014;de Almeida et al., 2015). Although, PhoU is an important regulator in many bacteria, it is notably absent from certain Gram-positive bacteria, including B. subtilis (Qi et al., 1997;Moreno-Letelier et al., 2011). phoU deletion in E. coli, P. aeruginosa, and Mycobacterium marinum leads to growth defects (Steed and Wanner, 1993;Wanner, 1996;Rice et al., 2009;Wang et al., 2013;de Almeida et al., 2015). In E. coli, this growth defect is reversed by deletion of the pst transporter operon or the phoBR TCS operon (Steed and Wanner, 1993;Wanner, 1996). These observations suggest that the growth defect of phoU mutants is caused by unregulated function of the Pst transport system, leading to excess P i accumulation (Wanner, 1996;Rice et al., 2009). phoU mutants also accumulate increased amounts of poly-orthophosphate (poly-P i ) in E. coli, M. marinum, P. aeruginosa, and Caulobacter crescentus (Morohoshi et al., 2002;Hirota et al., 2013;Wang et al., 2013;de Almeida et al., 2015;Lubin et al., 2016). Poly-P i also accumulates in E. coli K-12 cells in stationary phase in high P i medium, and this accumulation is correlated with inhibition of biofilm formation mediated by PhoB∼P RR phosphorylation with acetyl-phosphate acting as donor (Grillo-Puertas et al., 2016). The rate of P i uptake was reported to increase in an E. coli phoU in one study (Rice et al., 2009), but not in another (Steed and Wanner, 1993). Besides defective growth, phoU mutants exhibit higher sensitivity to a diverse range of antibiotics in E. coli, Mycobacterium tuberculosis, M. marinum, and P. aeruginosa (Li and Zhang, 2007;Shi and Zhang, 2010;Wang et al., 2013;de Almeida et al., 2015) and a defect in mutagenic DNA break repair in E. coli (Gibson et al., 2015).
Recent papers demonstrate that E. coli PhoU interacts with the PAS domain of the PhoR HK and with the PstB ATPase protein, in support of the dual inhibition of Pho regulon transcription and Pst mediated transport in E. coli (Gardner et al., 2014(Gardner et al., , 2015. Three crystal structures of PhoU-like proteins have been reported from Aquifex aeolicus, Thermotoga maritima, and P. aeruginosa (Liu et al., 2005;Oganesyan et al., 2005;Lee et al., 2014), showing that PhoU consists of two symmetric, threealpha-helix bundles. However, these PhoU proteins showed several quaternary structures in crystals, including monomer, dimer, or hexamer packing (Liu et al., 2005;Oganesyan et al., 2005;Lee et al., 2014). Gel filtration shows that purified E. coli PhoU forms a dimer in solution (Gardner et al., 2014). In addition, divalent cation binding of magnesium and manganese is required for E. coli PhoU binding to the cytoplasmic side of the inner membrane and may play a role in formation of a ternary regulatory complex containing PhoU, PhoR, and PstB (Gardner et al., 2014). On the other hand, a recent study suggests that a different paradigm operates in C. crescentus, where PhoU does not modulate PhoR HK activity directly (Lubin et al., 2016). Instead, PhoU may negatively regulate the activity of the Pst transporter in response to P i availability in C. crescentus (Lubin et al., 2016).
Besides Pst ABC transporters, P i is taken up by symporter secondary transport systems. In E. coli, two additional P i transporters, PitA and PitB, have been identified that are symporters of divalent cations, such as Mg 2+ and Ca 2+ (van Veen FIGURE 1 | Model for regulation of the dual Pst1 and Pst2 ABC P i transporters in Streptococcus pneumoniae D39. Genomic sequencing and transcriptome analyses show that pnpRS, pst1-phoU1, and pst2-phoU2 are organized into three distinct operons that are transcribed from separate promoters, indicated by Ps followed by arrows (see text). The Pst2 ABC P i transporter is constitutively expressed in both high and low P i conditions. PhoU2 inhibits phosphorylation of the PnpR response regulator (RR) and negatively regulates the transport activity of Pst2. In the absence of PhoU2, PnpR is phosphorylated by the PnpS histidine kinase (HK). PnpR∼P binds to the Pho box upstream of the pst1-phoU1 operon and activates transcription and expression of the genes encoding the Pst1 transporter and PhoU1 regulator; however, PnpR∼P does not autoregulate its own (pnpRS) transcription. In contrast to PhoU2, PhoU1 negatively regulates the transport activity of Pst1, but does not regulate phosphorylation of PnpR. In the absence of the Pst2 transporter or in low P i conditions, the PnpRS two component system (TCS) activates the expression of the genes encoding the Pst1 transporter and PhoU1 regulator. Either the Pst1 or Pst2 P i transporter is required in encapsulated cells. Therefore, the regulated Pst1 and constitutive Pst2 P i transport systems may act as a redundant failsafe to ensure that capsule biosynthesis is maintained during variations in P i conditions. See the text for additional details. et al., 1994;Wanner, 1996;Harris et al., 2001;Jackson et al., 2008). PitA and PitB have been considered as low-affinity P i transporters that predominantly function in high P i environments (Wanner, 1996;Hsieh and Wanner, 2010). Moreover, expression of pitA is induced by Zn 2+ addition (Jackson et al., 2008), suggesting that PitA may act primarily as a metal ion transporter instead of a P i transporter (Beard et al., 2000).
Unlike E. coli, B. subtilis, or C. crescentus, which contain only one Pst transporter, Streptococcus pneumoniae (pneumococcus) encodes two evolutionarily distinct P i ABC pump transporters, Pst1 and Pst2 (Figure 1; Lanie et al., 2007;Moreno-Letelier et al., 2011). The multigene pst1 and pst2 operons are located at different locations in the pneumococcus chromosome (Lanie et al., 2007). Only one phoBR-like TCS, pnpRS, encoding the PnpR RR and the PnpS HK, is encoded upstream of the pst1 operon, and both the pst1 and pst2 transporter operons encode separate PhoU-family regulators, designated PhoU1 and PhoU2 (Figure 1). The PnpRS TCS, Pst1 transporter, and PhoU1 were initially studied in unencapsulated S. pneumoniae laboratory strain R6x (Novak et al., 1999), before the discovery of the second Pst2 transporter and PhoU2 regulator. This work indicated that pnpRS operon expression was not regulated by P i amount, and that mutants deficient in the PstB1 ATPase subunit seemed to show decreased P i uptake in certain growth media (Novak et al., 1999). Subsequent work suggests that upregulation of pst1 operon expression is correlated with increased β-lactam antibiotic resistance in low-affinity pbp2x mutants and some clinical isolates of S. pneumoniae (Soualhine et al., 2005;Engel et al., 2014).
In this report, we studied the transcriptional and functional regulation of the pneumococcal Pst1 and Pst2 P i transporters under growth conditions containing high (≈18 mM) or low (≈100 µM) concentrations of P i . Our results show that pst2 operon transcription is constitutive, but Pst2 transporter activity is negatively regulated by PhoU2. In addition, PhoU2 negatively regulates PnpR RR phosphorylation and transcription of the pst1 operon at high concentrations of P i . Therefore, PhoU2 resembles E. coli PhoU in that it functions in regulating the level of RR phosphorylation, besides modulating Pst2 transporter activity. In contrast, PhoU1 resembles C. crescentus PhoU in that it only regulates Pst1 transporter activity and does not modulate PnpRS TCS function. Our results also indicate that encapsulated S. pneumoniae requires the function of Pst1 or Pst2 for growth, whereas a symporter, named NptA, can provide sufficient P i to allow the growth of unencapsulated mutants in high P i conditions.

Bacterial Strains and Growth Conditions
Strains used in this study are listed in Table S1. Encapsulated strains were derived from parent strain IU1781 (D39 cps + rpsL1), and unencapsulated strains were derived from parent strains IU1945 (D39 cps), IU1824 (D39 cps rpsL1), and IU3309 (D39 cps2E rpsL1), which are derivatives of serotype 2 S. pneumoniae strain D39 IU1690 (Lanie et al., 2007). Strains containing antibiotic markers were constructed by transformation of competent pneumococcal cells with linear DNA amplicons synthesized by overlapping fusion PCR (Ramos-Montanez et al., 2008;Tsui et al., 2010). Strains containing markerless alleles in native chromosomal loci were constructed using allele replacement via the Pc-[kan-rpsL + ] (Janus cassette; Sung et al., 2001). Primers used to synthesize different amplicons are listed in Table S2. All constructs were confirmed by DNA sequencing of chromosomal regions corresponding to the amplicon region used for transformation. Bacteria were grown on plates containing trypticase soy agar II (modified; Becton-Dickinson) and 5% (vol/vol) defibrinated sheep blood (TSAII-BA). Plates were incubated at 37 • C in an atmosphere of 5% CO 2 . For selections of transformants, TSAII-BA plates contained 250 µg/mL kanamycin, 0.3 µg/mL erythromycin, or 250 µg/mL streptomycin. Strains were cultured statically in Becton-Dickinson brain heart infusion (BHI) broth at 37 • C in an atmosphere of 5% CO 2 , and growth was monitored by OD 620 as described before (Tsui et al., 2016). Transformants were single-colony-isolated on TSAII-BA plates containing antibiotics twice before growth in antibioticcontaining BHI broth for storage (Tsui et al., 2016). All mutant constructs were confirmed by DNA sequencing of chromosomal regions corresponding to the amplicon region used for transformation.
Static growth in BHI broth, which contains ≈18 mM P i , was used as a high P i condition. For growth curves, strains were inoculated into 3 mL of BHI broth, serially diluted, and grown overnight. The next day, cultures with OD 620 = 0.1-0.3 were diluted into 5 mL of fresh BHI broth to OD 620 ≈0.002, and growth was monitored hourly. C+Y medium (Lacks and Hotchkiss, 1960) was used for studies of moderately low P i condition. We determined that C+Y broth (no added P i ) already contains ≈1.5 mM P i (see Results). A modified chemically defined medium (mCDM) (Carvalho et al., 2013) was used for moderate and low P i conditions. To optimize growth, the concentrations of choline-HCl and all amino acids amounts were increased by 1000-fold and tyrosine was added to 100 mg/L compared to the CDM recipe in Carvalho et al. (2013). In addition, 40 mM MOPS buffer was added to mCDM, which was adjusted to a final pH = 7.4 with 10 M NaOH. mCDM contains 36.4 mM P i (Carvalho et al., 2013). mCDM with no P i was made by omitting KH 2 PO 4 and K 2 HPO 4 and adding KCl to 50.8 mM. mCDM media with 2 mM or 1 mM P i was made by mixing mCDM and mCDM lacking P i in a ratio of 2-34.4 or 1-35.4, respectively. mCDM with 1 mM P i was diluted 10 or 100-fold with mCDM lacking P i make mCDM with 100 µM and 10 µM P i , respectively. For growth in mCDM, 3 mL overnight cultures were grown as described above, and the next day, cultures with OD 620 = 0.1-0.3 were centrifuged (5125 × g, 5 min, 25 • C), washed with 3 mL mCDM lacking P i twice, and resuspended in 3 mL mCDM lacking P i . Cells were then diluted in 5 mL of mCDM with 2, 1 mM, 100, 10 µM, or no P i to OD≈0.005 and growth of static cultures in an atmosphere of 5% CO 2 was monitored hourly at OD 620.

RNA Preparation, qRT-PCR, and RNA-Seq Analyses
To study high P i conditions, overnight cultures were diluted and grown in 5 mL of BHI to OD ≈0.15. Cells were collected by centrifuging at 16,000 × g for 5 min at 4 • C. 1 mL of RNApro solution (MP Biomedicals) was added to resuspend cell pellets. The suspension was transferred to a Lysing Matrix B tube (MP Biomedicals), which was shaken 3X in FastPrep homogenizer (6.0 M/s for 40 s each). Cell debris and lysing matrix were removed by centrifugation at 16,000 × g for 5 min at 4 • C. 700 µL of supernates was transferred to a new microcentrifuge tube and incubated at room temperature for 5 min. 300 µL of chloroform was then added followed by incubation at room temperature for 5 min. Mixtures were centrifuged at 16,000 × g for 5 min at 4 • C. 280 µL of the upper, aqueous phase was collected and mixed with 140 µL of 100% Ethanol in a new microcentrifuge tube for RNA precipitation. RNA purification was done using miRNeasy minikit (Qiagen), including on-column treatment with DNase I (Qiagen), following the manufacturer's instructions. 5 µg of purified RNA was treated by DNase from a DNAfree DNA removal kit (Ambion). 125 ng of treated RNA was used to synthesize cDNA by a qScript Felex cDNA synthesis kit (Quanta Biosciences). Synthesized cDNA was diluted 1:6 in water and then serially diluted 1:5 in water three more times. qRT-PCR reactions contained 10 µL of 2 × Brilliant III Ultra-Fast SYBR Green QPCR Master Mix (Agilent), 2 µL of each 2 µM primers (Table S3), 0.3 µL of a 1:500 dilution of ROX reference dye, and 6 µL of diluted cDNA. Samples were run in an MX3000P thermocycler (Stratagene) with Program MxPro v. 3.0. Transcript amounts were normalized to gyrA mRNA amount and compared with transcript amounts of the wild-type parent strain by unpaired t-test in GraphPad Prism (Kazmierczak et al., 2009).
To study low P i conditions, bacteria from overnight BHI broth cultures were washed, diluted, and grown in 5 mL of mCDM medium containing 36.4 mM P i to OD 620 ≈0.15. Cells were collected by centrifugation, washed twice with 5 mL mCDM lacking P i , and resuspended in mCDM containing 36.4 mM P i or 10 µM P i . Cultures were incubated at 37 • C in an atmosphere of 5% CO 2 for 30 min. Lysis, RNA extraction, purification, and qRT-PCR reactions were performed as described above, except that 16S rRNA was used to normalize transcript amounts, because gyrA was down-regulated under low P i condition. All transcript amounts were compared with the wild-type parent strain grown in mCDM containing 36.4 mM.
RNA samples for RNA-Seq analyses were prepared from 30 mL cultures as described previously (Hoover et al., 2015) cDNA library construction, single-end, 100 bp-sequencing on a HiSeq 2000 sequencer (Illumina), and bioinformatic analyses were performed as described in Hoover et al. (2015). Falsediscovery rates (FDR) were calculated using Benjamini and Hochberg's algorithm (Benjamini and Hochberg, 1995) and a gene or region was defined as differentially expressed if it had an up-or down-fold change of 1.8 with a FDR < 0.05. RNA-Seq data were deposited in the NCBI GEO database under accession number GSE80637.

Phos-Tag SDS-PAGE and Western Blot
Phos-tag SDS-PAGE and standard Western blotting were carried out as described previously (Wayne et al., 2010(Wayne et al., , 2012Tsui et al., 2014). To study high P i conditions, overnight BHI broth cultures were diluted and grown up to OD 620 ≈0.2 in 30 mL of BHI. Cells expressing PnpR-L-FLAG 3 were lysed using a FastPrep homogenizer, and cell lysates were resolved by Phos-tag SDS-PAGE at 4 • C (Wayne et al., 2012). Cells expressing PstS2-HA were lysed by the same method, but resolved by standard SDS-PAGE (Tsui et al., 2014). PnpR-L-FLAG 3 and PstS2-HA were detected by Western blotting as described previously (Tsui et al., 2014) using anti-FLAG or anti-HA antibody as primary antibody. Chemiluminescent signal in protein bands was quantitated by using an IVIS imaging system as described in Wayne et al. (2010).
To study low P i conditions, strains were grown in 30 mL of mCDM to OD 620 ≈0.2 as described above. Cells were collected by centrifugation and washed twice with 30 mL of mCDM lacking P i . Cell pellets were resuspended in 30 mL of mCDM containing 36.4 mM P i or lacking P i and incubated statically at 37 • C in an atmosphere of 5% CO 2 for 40 min. Proteins samples were extracted and Phos-tag SDS-PAGE was performed as described above.

Qualitative Quellung Assay for Capsule
Overnight cultures were diluted and grown in 5 mL of BHI to OD≈0.15. 1 µL of culture was mixed with 1 µL of Type 2 pneumococcal antiserum (Statens Serum Institut) on a glass slide. A cover slip was placed on top of the mixture, which was viewed immediately with a 100X objective by a phase-contrast microscope. Cells surrounded by capsule appear enlarged or swollen.

P i Concentration Determination
The P i amount in C+Y broth was determined by the colorimetric method described in Katewa and Katyare (2003). Briefly, standards were prepared by dilution of a KH 2 PO 4 stock to give final P i concentrations of 2.5, 5.0, 10.0, 20.0, and 40.0 µM P i in 2.4 mL of water in glass tubes. 0.8 mL of 3N H 2 SO 4 was added to each standard tube. 0.4 mL of 2.5% (w/v) ammonium molybdate (prepared in 3N H 2 SO 4 ) was added in each tube. Last, 0.4 mL of reducing agents (20 mg of hydrazine sulfate and 20 mg of ascorbic acid dissolved in 1 mL of 0.1N H 2 SO 4 ) was added to each tube. After 2 h at room temperature, A 820 was determined and plotted to generate a standard curve. C+Y broth (no added P i ) was diluted 100X with water, P i content was assayed as described above, and P i concentration was determined from the standard curve.

P i Uptake Assays
To study high P i conditions, strains were grown in 5 mL of BHI broth to OD 600 ≈0.2. Cells were centrifuged at 5125 × g for 5 min at room temperature, washed twice with 5 mL of mCDM lacking P i , and resuspended at room temperature in 5 mL of mCDM lacking P i . K 2 H 32 PO 4 (10 7 -10 8 dpm; 8500-9120 Ci/mmole; Perkin Elmer) was added to a final concentration of 1 mM at t=0, and 100 µL of cells was collected by vacuum filtration (0.22 µm GSWP; 13 mm diameter; Millipore) at 1, 2, 4, 6, 10 min after addition of 32 P. Filters were washed 3X with 3 mL of roomtemperature 1 × PBS (Ambion). Washed filters were transferred individually into 20-mL glass scintillation vials to which 5 mL of a biodegradable counting cocktail was added. Dpm of each sample was determined using a TRI-CARB 2100TR Liquid Scintillation Counter (Perkin Elmer), P i amount incorporated at each time point was calculated.
To study low P i condition, strains were grown to OD 600 ≈0.2 in mCDM containing 36.4 mM Pi as described above. Cells were centrifuged at 5125 × g for 5 min at room temperature, and washed twice with 5 mL of mCDM lacking P i . Cell pellets were resuspended in 5 mL of mCDM lacking P i and incubated 1 h at 37 • C in an atmosphere of 5% CO 2 . 200 µM of K 2 H 32 PO 4 (10 7 -10 8 dpm) was added at t = 0, and samples were withdrawn, filtered, and counted as described above.

phoU2 Mutants Show a Growth Defect and Increased Sensitivity to a Range of Antibiotics That is Reversed by Inactivation of Pst2 Transport
The Pst transporter is not needed for growth of E. coli at high P i concentrations >4 µM, and phoU mutations lead to a severe growth defect that is reversed by inactivation of the Pst transporter (Steed and Wanner, 1993;Rice et al., 2009). To determine the roles of the Pst1 and Pst2 transporters and their regulation, we constructed a series of markerless deletion mutants [ phoU2 (IU6375); phoU1 (IU6377); pnpRS (IU6381); pst2 (IU6610); pst1 (IU6638)] in S. pneumoniae serotype 2 strain D39, which is encapsulated and virulent (Lanie et al., 2007; Figure 1, Table S1). Mutants were first grown in BHI broth, which contains a high P i concentration (≈18 mM). Only the phoU2 mutant showed a significant decrease in growth yield compared to the parent and other mutants (Figure 2A, Figure S1A, Table S4). Growth yield was restored when the phoU2 mutation was complemented by an ectopic copy of the phoU2 + gene expressed from the P ftsA promoter at the bgaA site (IU6397) (Figure 2A, Table S4). Similar results were obtained in mutants in unencapsulated D39 derivative strain K579 and E579 (data not shown).
Inactivation of phoU in E. coli generates higher sensitivity to various kinds of antibiotics than the parent strain (Li and Zhang, 2007). Since the pneumococcal phoU2 mutant shows a similar growth defect as the phoU mutant in E. coli (Figure 2A, Figure  S1A), we tested antibiotic sensitivity to several antibiotics. Of the Pho regulon mutants tested in the encapsulated strain, only the phoU2 mutant showed increased sensitivity to β-lactams and other classes of antibiotics, including glycopeptides and protein synthesis inhibitors, on plates that contain relatively high P i content ( Figure 2B, Figure S1B, and Table S5).
Previous work in E. coli (Steed and Wanner, 1993;Wanner, 1996;Rice et al., 2009;Hsieh and Wanner, 2010) and C. crescentus, in which PhoU is essential for growth (Lubin et al., 2016), indicated that the reduced growth of phoU deletion or depletion mutants could be reversed by inactivation of the Pst pump. This reversal was interpreted to mean that PhoU negatively regulates the Pst transporter itself, and in its absence, deleterious phosphate compounds accumulate that disrupt growth and metabolism. In the pneumococcal encapsulated D39 genetic background, a pst2-phoU2 deletion mutant lacking the Pst2 transporter and PhoU2 regulator grew similar to the parent strain and did not show increased sensitivity to antibiotics caused by the absence of PhoU2 alone (Figures 2C,D , Table S4). Likewise, single deletions of genes encoding each component of the Pst2 transporter restored growth yield of a phoU2 FIGURE 2 | Deletion of phoU2 leads to a lower growth yield and increased β-lactam antibiotic sensitivity that are reversed by a pst2 mutation. (A) Representative growth curves in BHI broth (≈18 mM P i ) of encapsulated parent strain (IU1781), a phoU2 mutant (IU6375), and a phoU2 mutant complemented by ectopic expression of PhoU2 + (IU6397). Strains were grown as described in Materials and Methods. A linear scale for OD 620 is used to emphasize differences in growth yields. Growth yields and rates are quantitated for multiple determinations in Table S4. (B) Cefotaxime sensitivity assays of encapsulated parent strain (IU1781), a phoU2 mutant (IU6375), and a PhoU2 + -complemented phoU2 mutant (IU6397). Cefotaxime disk sensitivity assays of bacteria grown in BHI broth were performed as described in Material and Methods. P-values were calculated by unpaired t-tests relative to the parent strain using GraphPad Prism; (n ≥ 3); ***P < 0.001. Increased sensitivity to other β-lactam antibiotics, vancomycin, gentamicin, and tetracycline of a phoU2 mutant compared to its isogenic parent strain is shown in Table S5. (C) Representative growth curves of encapsulated parent strain (IU1781) and phoU2 (IU6375), pst2-phoU2 (IU6550), and pst2 (IU6610) mutants in BHI broth. (D) Cefotaxime sensitivity assays for encapsulated parent strain (IU1781), and phoU2 (IU6375), pst2-phoU2 (IU6550), and pst2 (IU6610) mutants. (n ≥ 3); ***P < 0.001. mutant back to wild-type (data not shown). Western blot analysis showed that the phoU2 mutation did not change the cellular amount of the PstS2 transporter subunit fused to the HA epitope tag ( Figure S3), and by inference the amount of the Pst2 transporter. We conclude that the decreased growth yield and antibiotic sensitivity of encapsulated pneumococcus phoU2 mutants in high P i conditions are dependent on function of the Pst2 transporter, consistent with negative regulation of the Pst2 transporter by PhoU2.
PhoU2 Negatively Regulates Transcription Activation of the pst1 Operon by the PnpRS TCS in High P i Conditions As noted in the Introduction, PhoU negatively regulates the PhoBR TCS in E. coli, but not in C. crescentus (Hsieh and Wanner, 2010;Lubin et al., 2016). Consequently, we tested whether phoU1 or phoU2 deletions affected transcript levels of the pst1 or pst2 operons under high P i conditions (see Figure 1). We first performed RNA-Seq analyses of phoU2::kanrpsL + and phoU2::kanrpsL + phoU1::P c -erm mutants growing in early-middle exponential phase in BHI broth, which contains a high (18 mM) concentration of P i (Table S6). In both strains, only transcript amounts of the pst1 transporter operon, including phoU1 in the single mutant, were strongly induced (≈22X). The transcript amounts of the pnpRS regulator and pst2 transporter operons were not induced, and the number of other genes in the Pho regulon appears to be limited in S. pneumoniae D39 (Table S6). Notably, pneumococcus encodes neither an alkaline phosphatase (phoA) nor a pathway for synthesis of teichuronic acids lacking phosphate (Wanner, 1996;Botella et al., 2011Botella et al., , 2014. Besides the strong induction of pst1 operon transcription, there were small (2-4X) changes in the relative amounts of only a handful of other transcripts, including some corresponding to metabolic and stress-responsive genes, possibly reflecting the defective growth of these phoU2 mutants. In both mutants, one of the stronger responses was a decrease in the relative transcript amounts of the genes encoding the glycerol facilitator (GlpF) and glycerol kinase (GlpK). A putative, somewhat degenerate Pho box is located -125 bp upstream of the glpK reading frame. Together, these results suggested that PhoU2 negatively regulates pst1 operon expression under high P i conditions, whereas the pnpRS and pst2 operons are constitutively expressed.
These conclusions were confirmed by qRT-PCR analysis of combinations of markerless deletion mutations in the pst and pnpRS genes ( Table 1). RNA-Seq transcriptome analysis indicates that the pnpRS, pst1-phoU1, and pst2-phoU2 operons are separately transcribed (Table S6, Figure 1, Figures S2A,B). Hence, we quantitated the relative amounts of the pnpR, pstS1, and pstS2 transcripts normalized to gyrA transcript amount by qRT-PCR to represent pnpRS, pst1-phoU1, and pst2-phoU2 operon expression (see Materials and Methods; Wayne et al., 2012). Consistent with the RNA-Seq results, the phoU2 mutations caused ≈16X increase in pst1-phoU1 operon transcript amount, but no change in expression of pst2 or pnpRS operon (Table 1, line 2). In contrast, a phoU1 mutation did not cause a significant change in the relative amounts of transcript from any of the three operons ( Table 1,   amount in the phoU2 mutant was detected in complementation experiments in which a wild-type copy of phoU2 + was expressed from an ectopic site (Table 1, line 4). We next determined that the increased transcription of the pst1-phoU1 operon in the phoU2 mutant is mediated by the PnpRS TCS through increased phosphorylation of the PnpR∼P RR. pnpR, pnpS, or pnpRS mutants showed slightly reduced pst1-phoU1 operon transcript amounts compared to the parent strain, with no change in pst2-phoU2 operon expression ( Table 1, lines 5-7). Likewise, pnpR phoU2, pnpS phoU2, and pnpRS phoU2 double mutants showed reduced pst1-phoU1 operon relative transcript amounts (Table 1, lines 8-10), instead of the sizable increase observed for the phoU2 single mutant containing an active PnpRS TCS (Table 1, line 2). To confirm directly that PhoU2 acts as a negative regulator of PnpRS function under high P i conditions, we performed Phos-tag SDS PAGE analysis to determine PnpR∼P RR phosphorylation levels (see Materials and Methods; Figure 3). In these experiments, we fused three tandem copies of the FLAG epitope tag to the Cterminus of the PnpR RR regulator expressed from its normal chromosomal locus (Figure 3, Table S1). The PnpR-L-FLAG 3 RR induced pst1-phoU1 transcription to a similar extent as wild-type (untagged) PnpR + in a phoU2 mutant (data not shown). In the phoU2 + strain, essentially no PnpR∼P (<1%) was detected in cells growing exponentially in high-P i BHI broth (Figure 3). In contrast, the phoU2 mutant contained ≈45% PnpR∼P, which accounts for the high (≈16X) increase in pst1-phoU1 transcript detected (Table 1, line 2). We conclude that PhoU2 negatively regulates PnpR∼P amounts and transcription of the pst1-phoU1 operon, but does not regulate transcription of the pnpRS or pst2-phoU2 operon, which are constitutively expressed.
Consistent with this interpretation, a putative PnpR∼P binding site (Pho-box) sequence (TTTACACAATCTTTACA; Martin, 2004) is located 92 bp upstream of pstS1 reading frame gene (Figure 1), but no recognizable Pho-box sequences can be found upstream of the pnpRS or pst2-phoU2 operon. Finally, we tested whether induction of pst1-phoU1 operon expression contributes to the growth defect and antibiotic sensitivity of a phoU2 mutant (Figure 2). A phoU2 pst1 or phoU2 pst1-phoU1 double mutant showed the same decrease in growth yield ( Figure 4A, Table S4) and antibiotic sensitivity (Figure 4B) as the phoU2 single mutant, indicating that these defects were caused primarily by misregulation of the Pst2 transporter in high P i conditions, instead of induced expression of the Pst1 transporter.

Transcription of the pst1 Operon is Further Induced by the Absence of the Pst2 Transporter Under High P i Conditions
We next tested whether absence of the Pst2 transporter affects pst1 operon expression in bacteria growing in high P i -BHI broth. Surprisingly, relative transcript amounts from the pst1 operon increased by ≈44 fold in the pst2 phoU2 + , pst2-phoU2, and pst2-phoU2 phoU1 mutant compared to the parent strain (Table 1, lines 12, 13, and 14). By contrast, pst2 operon transcription is unchanged in a pst1 phoU1 + mutant compared to the parent (Table 1, line 15). The similar induction in the pst2 phoU2 + and pst2-phoU2 mutants ( Table 1, lines 12 and 13) implies that negative regulation of the PnpRS TCS by PhoU2 depends on a functional Pst2 transporter system. Moreover, the similarity of pst1 induction in the pst2-phoU2 phoU1 and other mutants implies that PhoU1 does not directly regulate pst1 transcription. This conclusion was supported by the similar induction of pst1 operon transcription  Figure S4). (B) Quantification of 3 independent Phos-tag SDS-PAGE experiments. Less than 1% of PnpR was phosphorylated in the phoU2 + strain, whereas ≈45% of PnpR is phosphorylated (PnpR∼P) in the phoU2 mutant. P-value was determined by an unpaired t-test in GraphPad Prism; ***P < 0.001.  Table S4. (B) Cefotaxime sensitivity assays for encapsulated parent strain (IU1781), and phoU2 (IU6375), phoU2 pst1 (IU6664), and phoU2 pst1-phoU1 (IU6854) mutants. (n ≥ 3); ***P < 0.001. See Materials and Methods, the legend to Figure 2, and the text for additional details.
in the phoU2 phoU1 + and phoU2 phoU1 mutants ( Table 1, lines 2 and 11). Instead, the growth characteristics of these mutants imply that PhoU1 negatively regulates the activity of the Pst1 transporter, in parallel to the negative regulation of the activity of Pst2 by PhoU2 ( Figure 5A, Table S4). In BHI broth, the parent, phoU1, and pst2-phoU2 mutant show similar growth and antibiotic sensitivity (Figure 5; Table S4). In contrast, the phoU2 and phoU1 pst2-phoU2 mutants showed reduced growth yield and increased antibiotic sensitivity (Figure 5, Table S4), consistent with increased P i accumulation caused by misregulation of the Pst2 and Pst1 transporters, respectively. This interpretation was further supported by the reduced growth rate and yield and increased antibiotic sensitivity of the phoU2 phoU1 mutant compared to the phoU2 and pst2-phoU2 phoU1 mutants (Figure 5; Table S4).
Either the Pst1 or Pst2 Transporter is Required in the Encapsulated, But Not in the Unencapsulated, D39 Strain E. coli encodes an alternate low-affinity P i uptake (Pit) system that functions in the absence of the high-affinity Pst transporter in high P i conditions (Wanner, 1996;Harris et al., 2001;Hsieh and Wanner, 2010). However, we were unable to construct pst2 pst1 mutants in the encapsulated strain of D39 ( Table 2, top two sections), where the nptA amplicon was used as a positive control in the transformations. Likewise, we were unable to construct a pst2 pnpRS double mutant, where induction of the Pst1 transporter was negated by the absence of the PnpRS TCS (Table 1, top section; Figure 1). As expected, the cps + pnpRS-pst1 mutant could be constructed (IU6133), where the Pst2 transporter is functional.
In all pst2 pst1 cps + transformations, several colonies (<10) appeared upon prolonged incubation (Table 2, top two sections). These suppressor colonies had a rougher appearance than the smooth colonies of D39 cps + strains (data not shown), suggesting that capsule production was lost or reduced in these pst2 pst1 transformants. The Quellung test for serotype 2 capsule (see Materials and Methods) confirmed that a suppressor strain (IU6413) had lost its capsule (data not shown). This result suggested that unlike in cps + strains, we would be able to construct cps pst1 pst2 mutants in the D39 genetic background. This hypothesis was confirmed by transformation experiments (Table 2, bottom two sections), and growth experiments showing that cps pst1 pst2 triple mutants grew comparably to the cps single mutant in BHI broth ( Figure S5). Thus, we conclude that either the Pst1 or Pst2 P i transporter must be functional in encapsulated D39 strains and that the pst1 pst2 cps + double mutant is not viable.   Table S2).
Amplicons used for transformations contained the Pc-kanrpsL + or Pc-erm antibiotic cassette for selection. Transformations with the nptA amplicon or without DNA were the positive or negative control, respectively. d ≤10 colonies on plates indicates accumulation of unencapsulated suppressor mutants (see text).
NptA is the Third P i Uptake System That Functions in cps pst1 pst2 Mutants Normal growth of the cps pst1 pst2 mutant in high P i medium (Figure 5) implies that sufficient P i is being taken up by a third uptake system. BLAST searches did not reveal a close pneumococcal homolog of the Pit symporters of E. coli and B. subtilis. During these searches, we found another candidate gene, spd_0443, which encodes a putative Na + /P i -cotransporter IIlike protein. Spd_0443 homologs have been shown to act as P i transporters in mammalian intestines and kidneys (Katai et al., 1999) and in certain bacterial species, such as V. cholerae and Vibrio vulnificus (Lebens et al., 2002;Staley and Harwood, 2014). Consistent with a role in P i uptake, we could not delete spd_0443 in a cps pst1 pst2 mutant in high P i medium ( Table 3), but we could delete spd_0443 in the pst1 or pst2 single mutant ( Table 2). Thus, the Spd_0443 Na + /P i -cotransporter likely acts as a third Pi uptake system in S. pneumoniae. Because the iron transporter in S. pneumoniae is already named "Pit, " we renamed Spd_0443 as NptA (Na + -dependent phosphate transporter A), similar to Vibrio species (Lebens et al., 2002).   Table S2).
Amplicons used for transformations contained the Pc-kanrpsL + or Pc-erm antibiotic cassette for selection. Transformations with the pnpR amplicon or without DNA were the positive and negative control, respectively.

pst1-phoU1 or pst2-phoU2 Deletion Has No Effect on Growth Under Low P i Conditions
We next examined the roles of the Pst1 and Pst2 transporters and their regulators (Figure 1) under low P i culture conditions. A previous study used the semi-defined C+Y medium as a low P i condition (Novak et al., 1999). However, encapsulated strains grew in C+Y broth without P i addition ( Figure S6), and direct chemical assay (see Materials and Methods; Katewa and Katyare, 2003) showed that C+Y broth (with no added P i ) contains ≈1.5 mM P i ( Figure S7). Therefore, a modified chemically defined medium (mCDM) was used to study low P i culture conditions (see Materials and Methods; Carvalho et al., 2013). The growth yield of the wild-type encapsulated strain was dependent on P i concentration below 1 mM P i , with growth detectable down to ≈10 µM and no growth without P i addition ( Figure 6A). Growth rates and yields of the pst2-phoU2 and pst1-phoU1 mutants were similar to those of the parent strain (Figures 6B,C), implying that a functional Pst1 or Pst2 transporter is sufficient for optimal growth under low P i conditions.
We used unencapsulated ( cps) mutants to determine the effects of P i concentration when both the Pst1 and Pst2 transporters were absent. Similar to the encapsulated strains (Figure 6), growth yield of the parent, pst1-phoU1, and pst2-phoU2 single mutants decreased below 1 mM P i and was still detectable at 10 µM P i (Figures 7A-C). By contrast, the growth of the pst2-phoU2 pst1-phoU1 double mutant was not fully supported even by 2 mM P i , and growth yield showed apparent autolysis in 1 mM P i and no growth with ≥100 µM P i (Figure 7D). We conclude that in low P i conditions, the NptA transporter alone is not sufficient for growth of D39-derived S. pneumoniae, consistent with NptA acting as a low-affinity P i transporter compared to Pst1 or Pst2. Low P i Induces pst1 Operon Transcript Amount >100X via Phosphorylation of the PnpR∼P RR We determined the effect of low P i concentration on pst1 operon expression in the encapsulated D39 strain. Strains were grown in high-P i (36.4 mM) mCDM, washed, and resuspended in mCDM containing high (36.4 mM) or low (10 µM) P i (see Materials and Methods). After 30 min, samples were taken for qRT-PCR using 16S rRNA as the normalization standard (see Materials and Methods). In all strains tested, P i limitation to 10 µM induced relative pst1 transcript amount by >100X compared to the wild-type parent strain in high (36.4 mM) P i ( Table 4). Similar to strains grown in BHI (Table 1), phoU2 and phoU2 phoU1 mutants showed ≈25X induction of pst1 operon transcript amount when grown in mCDM containing high P i (Table 4), whereas a phoU1 mutant showed no increase ( Table 4). In these mutants, reduction of P i concentration to 10 µM induced pst1 transcript amounts by an additional ≈5-100X (Table 4). P i limitation of a pnpRS TCS mutant showed no increase in relative pst1 operon transcript amounts (data not shown), indicating a dependence on the PnpRS TCS for induction of pst1 operon transcription in low P i media. This conclusion was confirmed by Phos-tag SDS PAGE, which showed that phosphorylation of the PnpR RR (PnpR∼P) went from <1% in mCDM containing 36.4 mM P i to ≈80% following a shift to mCDM lacking P i for 40 min ( Figure S8A). We conclude that transcription of pneumococcal pst1 transporter operon is strongly induced by the PnpR∼P RR in low P i conditions.
Deletion of pst1-phoU1 and pst2-phoU2 Reduces the Rate of P i Uptake by ≈50% in an Unencapsulated Strain Previously, the relative rate of P i uptake was reported to be reduced by ≈30% in a pstB1 mutant of laboratory strain R6x grown in C+Y medium (no added P i ) (Novak et al., 1999), which turns out to contain a relatively high (1.5 mM) P i content (above; Figure S7). To the contrary, 32 P i uptake experiments in encapsulated and unencapsulated D39 mutants lacking the Pst1 ( pst1-phoU1) or Pst2 ( pst2-phoU2) transporter showed linear rates of 32 P i uptake for at least 10 min that were similar to those of the wild-type parent strains in mCDM containing 1 mM total P i (Figures 8A,B, Table S7). R6x is a derivative of R6 (Tiraby and Fox, 1973), which is an unencapsulated derivative of D39 (Lanie et al., 2007). The difference between this and the previous result (Novak et al., 1999) may partly reflect the large number of mutations in the R6 laboratory strain compared to the D39 strain progenitor genetic background (Lanie et al., 2007). In contrast to the single mutants, the unencapsulated pst1-phoU1 pst2-phoU2 double mutant showed a significant drop (≈50%) in the rate of P i uptake compared to the wild-type parent strain or FIGURE 7 | The NptA Na + /P i co-transporter alone cannot provide sufficient P i for long-term growth in mCDM containing moderately high (1 mM) and lower P i concentrations. (A) Representative growth curves of unencapsulated parent strain cps (IU1945) in mCDM with 36.4, 2, 1 mM, 100, 10 µM, and no P i . (B) Representative growth curves of unencapsulated cps pst2-phoU2 mutant (K583) in mCDM with 36.4, 2, 1 mM, 100, 10 µM, and no P i . (C) Representative growth curves of unencapsulated cps pst1-phoU1 mutant (E595) in mCDM with 36.4, 2, 1 mM, 100, 10 µM, and no P i . (D) Representative growth curves of unencapsulated cps pst2-phoU2 pst1-phoU1 mutant (IU5774), which only contains the NptA Na + /P i transporter, in mCDM with 36.4, 2, 1 mM, 100, 10 µM, and no P i . Growth curves were determined at least 3X independently. the single mutants in mCDM containing 1 mM P i ( Figure 8B, Table S7) or in C+Y broth (data not shown). This result implicates both the Pst1 and Pst2 transporters in P i uptake in S. pneumoniae D39. P i Uptake is Reduced in a pst1-phoU1 Mutant Limited for Total P i To determine P i uptake under low P i conditions, we incubated encapsulated parent, pst1-phoU1, or pst2-phoU2 mutants in mCDM lacking P i for 1 h, and then added 32 P i (total [P i ] = 200 µM), and sampled P i uptake with time ( Figure 8C). Unlike the constant rate of 32 P uptake observed at 1 mM total P i (Figures 8A,B), P i uptake leveled off after about 5 min in 200 µM P i (Figure 8C). The initial rate of P i uptake by the encapsulated parent strain was markedly greater (≈8X) in 200 µM P i than in 1 mM P i (Figures 8A,C, Tables S7,  S8). The initial rate of P i uptake was comparable for the parent strain and the pst2-phoU2 mutant ( Figure 8C, Table  S8). In contrast, the initial rate of P i uptake was reduced by ≈2X in the pst1-phoU1 mutant compared to the parent or pst2-phoU2 mutant ( Figure 8C, Table S8). These results support a primary role for the Pst1 transporter in low-P i conditions.

DISCUSSION
In this paper we demonstrate for the first time that regulation of the two evolutionarily distinct Pst P i transport systems is linked in several ways in the human commensal and pathogen, S. pneumoniae. Niches encountered by pneumococcus in human hosts contain different P i concentrations, including nasal fluid (≈5 mM), saliva (≈1 mM), and serum (≈1 mM) (Bansal, 1990;Wilson, 2005Wilson, , 2008. Several aspects of this regulatory network are unusual, compared to the usual negative regulation of P i uptake even at moderate P i concentrations in other bacterial species (Figure 1; Hulett, 1993;Hsieh and Wanner, 2010;Botella et al., 2011). The pst1-phoU1 and pst2-phoU2 operons are completely separated in the pneumococcal chromosome, and although upstream of the pst1-phoU1 operon, the pnpRS operon, which encodes the PnpRS TCS, is independently expressed and not autoregulated (Table 1, Figure 1). Deletion of gene spd_1614, which encodes a third putative PhoU-like protein, did not lead to growth phenotypes and was not studied further (data not shown).
Expression of the Pst2 transporter is constitutive under the conditions tested here, including media containing high P i concentrations ( Table 1, Table S6, and Figure S3). However, Pst2 uptake of P i is likely negatively regulated by the PhoU2 protein, whose absence causes a drop in growth yield and increased antibiotic sensitivity in mutants growing in high P i medium (Tables S4, S5, Figures 1, 2, and Figure S3). In addition, PhoU2 negatively regulates the transcription activation of the pst1-phoU1 operon by the phosphorylated PnpR∼P RR, such that a phoU2 deletion mutation leads to a substantial increase in pst1-phoU1 transcript amounts (Table 1, Figures 1, 3, and Figure S4). The PhoU1 protein, which shares only 34% amino acid identities with PhoU2, does not play a reciprocal role in negatively regulating pst1-phoU1 transcription ( Table 1, lines 3, 11, and 14), but PhoU1 likely negatively regulates P i uptake by the Pst1, but not the Pst2, P i transporter (Figures 1, 4, 5). Thus, the PhoU2 protein can regulate both PnpR phosphorylation level and Pst2 transporter function, similar to the PhoU homolog in E. coli (Hsieh and Wanner, 2010;Gardner et al., 2014Gardner et al., , 2015. In contrast, PhoU1 function is restricted to regulating transporter but not TCS function, similar to the PhoU in C. crescentus (Lubin et al., 2016). E. coli PhoU interacts with the PAS domain of the PhoR HK (Gardner et al., 2014(Gardner et al., , 2015. However, the pneumococcal PnpS HK lacks a recognizable PAS domain, and it remains to be determined whether PhoU2 regulation of PnpR∼P levels is through direct interactions with the PnpS HK. Likewise, it is unknown whether PhoU2 interacts with the PstB1 subunit of the Pst1 transporter to exert control over pst1-phoU1 operon transcription, by analogy to control in E. coli (Gardner et al., 2014(Gardner et al., , 2015. Interactions of PhoU1 or PhoU2 with subunits of the Pst1 or Pst2 transporters also remain to be determined. The requirement of a functional Pst1 or Pst2 for growth of the encapsulated serotype 2 strain provides a biological rationale for the constitutive expression and function of the Pst2 transporter at high P i concentrations ( Table 2) and why S. pneumoniae maintains dual P i uptake systems. Capsule is one of the most important factors required for pneumococcal colonization, carriage, and virulence in its human host (Briles et al., 1992;Morona et al., 2004Morona et al., , 2006Bentley et al., 2006;Hyams et al., 2010). Expression of the Pst1 transporter is strongly induced by low P i conditions (Table 4), and lack of PhoU1 does not change pst1-phoU1 operon expression in high P i media (Tables 2, 4). These results, and a reduced rate of P i uptake by mutants lacking Pst1 ( Figure 8C, Table S8) indicate that Pst1 functions mainly at low P i concentrations. However, encapsulated S. pneumoniae strains require high-affinity P i transport even in high-P i media ( Table 2), and this P i uptake cannot be provided by the lowaffinity NptA Na + /P i cotransporter that replaces Pst1 and Pst2 in unencapsulated strains ( Figure 8B, Table S7). The reason underlying this link between Pst-mediated P i transport and the maintenance of capsule is not currently clear. A recent report suggests that low P i conditions induce capsule biosynthesis in M. tuberculosis (van de Weerd et al., 2016). Overexpression of capsule could alter the metabolism of S. pneumoniae thereby inhibiting the growth of cps + pst1 pst2 mutants and leading to the appearance of spontaneous cps mutants ( Table 2). However, starvation of the wild-type D39 cps + strain for P i for 1 h did not reveal a qualitative change in capsule amount in the Quellung reaction (data not shown). Moreover, pneumococcal capsule biosynthesis is positively regulated by phosphorylation of regulatory protein CpsD (Yother, 2011), and this protein phosphorylation would likely be reduced during P i limitation.
The regulatory pathways that mediate capsule induction in M. tuberculosis involve sigma factors and poly-P i kinases (van de Weerd et al., 2016) that are absent from S. pneumoniae. As mentioned in the Introduction, phoU mutants of other bacteria accumulate poly-P i (Morohoshi et al., 2002;Hirota et al., 2013;Wang et al., 2013;de Almeida et al., 2015;Lubin et al., 2016), possibly because high intracellular Pi concentration disrupt metabolic homeostasis leading to defects in growth accompanied by general antibiotic sensitivity. RNA-Seq results reported here indicate that the pneumococcal Pho regulon contains a limited number of recognizable genes involved in P i accumulation (Table  S6). Notably, S. pneumoniae serotype 2 strain D39 as well as some other serotype strains, such as TIGR4, lack recognizable homologs of poly-P i kinases Ppk1 and Ppk2 (Zhang et al., 2002) and Mg 2+ -dependent poly-P i exopolyphosphatase Ppx (Akiyama et al., 1993), but do encode a degradative Mn 2+dependent, inorganic pyrophosphatase (PpaC) (Lanie et al., 2007); therefore, it is unclear whether these S. pneumoniae strains produces poly-P i . DAPI staining experiments that revealed poly-P i accumulation in C. crescentus phoU mutants (Lubin et al., 2016) were inconclusive and did not indicate any differences in staining of the phoU2 + parent and phoU2 mutant in high-P i BHI broth (data not shown). Similarly, DAPI-based assays of extracts did not show any difference between the parent and phoU2 mutant indicative of different poly-Pi amounts (data not shown).
Taken together, our results suggest that S. pneumoniae may maintain the regulated Pst1 and constitutive Pst2 P i transport systems as a failsafe to ensure capsule biosynthesis is maintained during variations in P i conditions. Coordination between the two Pst P i transport systems is coordinated by the PhoU2 protein that modulates transcription of the pst1-phoU1 operon by the PnpRS TCS and separately regulates the function of FIGURE 8 | The Pst1 or Pst2 transporter is sufficient for P i uptake under moderately high P i conditions (1 mM), but the Pst1 transporter is required for optimal P i uptake under low P i conditions (200 µM) following P i limitation. (A) P i uptake assays of encapsulated parent strain (IU1781) and pst2-phoU2 (IU6550) and pst1-phoU1 (IU6830) mutants in 1 mM total 32 P i in mCDM (n = 3). P i uptake was determined as the initial rate of 32 P i incorporation into cells in mCDM in pulse-labeling experiments described in Materials and Methods. See Tables S7 for initial rates of P i uptake (n = 3). (B) P i uptake assays of unencapsulated parent strain cps (IU1945) and cps pst2-phoU2 (K583), cps pst1-phoU1 (E595), and cps pst1-phoU1 pst2-phoU2 (IU5774) mutants in 1 mM P i mCDM. See Tables S7 for initial rates of P i uptake (n = 3). (C) P i uptake assay of encapsulated parent strain (IU1781) and pst2-phoU2 (IU6550) and pst1-phoU1 (IU6830) mutants in 200 µM total P i mCDM following P i starvation for 1 h as described in Materials and Methods (n = 3). Initial rates of Pi uptake were determined with the first min of labeling and are listed in Table S8 (n = 3).
the Pst2 transporter. Most bacteria encode a single Pst-type P i transporter as part of a Pho regulon that encode numerous phosphate accumulation and sparing proteins (Hulett, 1993;Wanner, 1996;Hsieh and Wanner, 2010). The limited Pho regulon of S. pneumoniae may reflect restriction to the human host, where P i is the primary source of phosphorus (Wilson, 2008). Even among the Streptococcus, dual Pst systems are limited to only a few species, including S. pneumoniae, S. pseudopneumoniae, S. dysgalactiae, S. equi (Group C), S. porcinus, S. agalactiae (Group B), and S. equinus, and are absent from the major Viridans and A Groups. Outside of the Streptococcus, multiple Pst P i transporters have only been reported in a very limited number of bacteria, including Synechocystis sp. PCC 6803 and M. tuberculosis (Braibant et al., 1996;Suzuki et al., 2004;Burut-Archanai et al., 2011;Tischler et al., 2013). In Synechocystis, low P i conditions activate expression of both of the Pst P i transporters that are present (Suzuki et al., 2004), whereas in M. tuberculosis one Pst transporter is activated by low P i and at least one other Pst transporter seems to be constitutively expressed, like Pst2 in S. pneumoniae (Tischler et al., 2013).

AUTHOR CONTRIBUTIONS
JZ, DS, KW, MW contributed to the conception and design of the work; JZ, DS carried out acquisition and analysis of data; JZ, DS, KW, MW interpreted data for the work; JZ, MW drafted and wrote the final version of the paper; JZ, DS, KW, MW approved the final version to be published.