Formation of Polyphosphate by Polyphosphate Kinases and Its Relationship to Poly(3-Hydroxybutyrate) Accumulation in Ralstonia eutropha Strain H16

Abstract

A protein (PhaX) that interacted with poly(3-hydroxybutyrate) (PHB) depolymerase PhaZa1 and with PHB granule-associated phasin protein PhaP2 was identified by two-hybrid analysis. Deletion of phaX resulted in an increase in the level of polyphosphate (polyP) granule formation and in impairment of PHB utilization in nutrient broth-gluconate cultures. A procedure for enrichment of polyP granules from cell extracts was developed. Twenty-seven proteins that were absent in other cell fractions were identified in the polyP granule fraction by proteome analysis. One protein (A2437) harbored motifs characteristic of type 1 polyphosphate kinases (PPK1s), and two proteins (A1212, A1271) had PPK2 motifs. In vivo colocalization with polyP granules was confirmed by expression of C- and N-terminal fusions of enhanced yellow fluorescent protein (eYFP) with the three polyphosphate kinases (PPKs). Screening of the genome DNA sequence for additional proteins with PPK motifs revealed one protein with PPK1 motifs and three proteins with PPK2 motifs. Construction and subsequent expression of C- and N-terminal fusions of the four new PPK candidates with eYFP showed that only A1979 (PPK2 motif) colocalized with polyP granules. The other three proteins formed fluorescent foci near the cell pole (apart from polyP) (A0997, B1019) or were soluble (A0226). Expression of the Ralstonia eutropha ppk (ppk_Reu) genes in an Escherichia coli Δppk background and construction of a set of single and multiple chromosomal deletions revealed that both A2437 (PPK1a) and A1212 (PPK2c) contributed to polyP granule formation. Mutants with deletion of both genes were unable to produce polyP granules. The formation and utilization of PHB and polyP granules were investigated in different chromosomal backgrounds.

INTRODUCTION

Ralstonia eutropha H16 is a facultative chemolithoautotrophic bacterium and has become famous because of its ability to grow autotrophically with hydrogen as the electron donor (the organism is referred to as Knallgasbakterium in German) and to accumulate large amounts of poly(3-hydroxybutyrate) (PHB). The PHB produced by R. eutropha or related species is, meanwhile, a commercially available biopolymer (1, 2). The formation of polyhydroxyalkanoates (PHAs) was studied in the past by many groups (for reviews, see references 3 to 9). Meanwhile, it is generally accepted that PHB granules are supramolecular complexes, and the designation as carbonosomes has been suggested for PHA granules (10). Carbonosomes are composed of a polymer core and have a surface layer of up to 16 proteins with different functions.

Inspection of the R. eutropha genome sequence (11) reveals the presence of the key enzymes for biosynthesis of biopolymers other than PHB, such as cyanophycin synthase (12) and polyphosphate kinase (PPK). While the synthesis of cyanophycin has not yet been demonstrated in R. eutropha, the formation of polyphosphate (polyP) is thought to be ubiquitous in all organisms (13–15). Indeed, the formation of polyP granules in R. eutropha is evident from early investigations in the labs of Hans G. Schlegel and others (16, 17) conducted more than 40 years ago and was recently demonstrated by cryotomography (18). PolyP is formed by the action of PPKs: two major types of PPKs are currently known: PPK1s are characterized by peptides of ∼80 kDa having 4 domains (the N domain, the H domain, and two C domains) (14). PPK of Escherichia coli is the prototype of PPK1s: it is the first PPK that was purified and biochemically characterized (19), and its 3-dimensional structure has also been solved (20). PPK2s differ from PPK1s in that their primary amino acid sequences are only approximately half the length of PPK1 amino acid sequences and their amino acid sequences are not related to the amino acid sequences of PPK1s. PPK2 of Pseudomonas aeruginosa (41 kDa) was the first PPK2 to be purified and biochemically studied (21). While most PPK1s use ATP as the phosphate donor and catalyze the synthesis of polyP, PPK2 of P. aeruginosa prefers GTP and works in the direction of GTP synthesis at the expense of polyP. Recently, PPK2s have been divided into three subgroups by bioinformatic characterization of their primary amino acid sequences (22). Bacteria differ in having either only PPK1 or only PPK2, or both types of PPKs (14).

Interestingly, polyP granules in R. eutropha are often located in the neighborhood of PHB granules if both biopolymers are present simultaneously. However, it is not known whether PHB and polyP granules are located in the same neighborhood simply because of limited space in the cell or whether there is a physical connection either between the two polymers or between the two polymers and a third common compound, such as the bacterial nucleoid. A physical connection of polyP with low-molecular-weight oligo(PHB) is well-known for Ca-polyP-PHB complexes (nonstorage PHBs) in biological membranes of E. coli (23–27). Remarkably, a physiological connection of PHB metabolism with the formation of polyP granules has also been known for a long time. The so-called enhanced biological phosphate removal (EBPR) process in modern wastewater treatment facilities takes advantage of this connection: carbon sources that are precursors of PHB, such as acetate or other fatty acids, are taken up by bacteria of the EBPR process and stored in the form of PHB granules in an oxygen-limited (anaerobic) first phase, and no or only a little polyP is accumulated. In a subsequent, aerobic phase, the PHB that accumulated in the anaerobic phase is mobilized and used for metabolism and growth. Simultaneously, large amounts of phosphate are taken up from the environment and stored intracellularly in the form of polyP granules. Interestingly, more phosphate than is necessary for cell growth is taken up (and is referred to as “luxurious phosphate uptake”; for reviews of the EBPR process, see references 28 to 31). The coupling of PHB synthesis in the first phase and the accumulation of polyP in the second phase during EBPR indicate that a metabolic link between the synthesis of PHB and the synthesis of polyP exists in bacteria that participate in EBPR. Unfortunately, bacteria that participate in EBPR cannot be cultivated in pure cultures in the lab, and molecular biological investigations are difficult to perform. In the study described in this contribution, we started to address the question of a potential link between PHB and polyP metabolism in R. eutropha, an easy-to-cultivate betaproteobacterium with a taxonomic relationship to “Candidatus Accumulibacter phosphatis” (29, 32) by two-hybrid analysis and by determination of key enzymes of polyP synthesis.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

The bacterial strains and plasmids used in this study are shown in Table 1, and the primers used in this study are listed in Table 2. E. coli strains were grown in lysogeny broth (LB) medium supplemented with the appropriate antibiotics at 37°C. R. eutropha H16 strains were grown on nutrient broth (NB; 0.8% [wt/vol]) with or without addition of 0.2% (wt/vol) sodium gluconate at 30°C.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant characteristic(s)a	Source or reference
Strains
Escherichia coli JM109	Cloning strain
E. coli BTH101	Two-hybrid assay strain, F− cya-99 araD139 galE15 galK16 rpsL1 (Strr) hsdR2 mcrA1 mcrB1	73
E. coli S17-1	Conjugation strain	74
E. coli LJ110 Δppk (SK5721)	Chromosomal deletion of ppk1	This study
Ralstonia eutropha H16	R. eutropha wild-type strain	DSMZ 428
R. eutropha ΔpitA (SK5226)	Chromosomal deletion of pitA	This study
R. eutropha ΔphaX (SK5691)	Chromosomal deletion of phaX (A2274)	This study
R. eutropha ΔphaC1 (SK5690)	Chromosomal deletion of phaC1	75
R. eutropha ΔphaC1 ΔphaX (SK5692)	Chromosomal deletion of phaC1 and phaX	This study
R. eutropha Δppk1a (SK5033)	Chromosomal deletion of ppk1a (A2437)	This study
R. eutropha Δppk1b (SK5036)	Chromosomal deletion of ppk1b (B1019)	This study
R. eutropha Δppk2c (SK5628)	Chromosomal deletion of ppk2c (A1212)	This study
R. eutropha Δppk2d (SK5631)	Chromosomal deletion of ppk2d (A1271)	This study
R. eutropha Δppk1a Δppk1b (SK5116)	Chromosomal deletion of ppk1a and ppk1b	This study
R. eutropha Δppk1a Δppk1b Δppk2c (SK5630)	Chromosomal deletion of ppk1a, ppk1b, and ppk2c	This study
Plasmids
pUT18C	Bacterial two-hybrid analysis	76
pKT25	Bacterial two-hybrid analysis	76
pLO3	Deletion vector, Tcr sacB	77
pBBR1MCS2	Broad-host-range vector, Kmr	78
pBBR1MCS2-PphaC-eyfp-c1	Universal vector for construction of fusions C terminal to eYFP under control of the PphaC1 promoter	35
pBBR1MCS2-PphaC-eyfp-n1	Universal vector for construction of fusions N terminal to eYFP under control of the PphaC1 promoter	35
pBBR1MCS2-PphaC1-eyfp-ppk1aReu	N-terminal fusion of PPK1aReu to eYFP	This study
pBBR1MCS2-PphaC1- ppk1aReu-eyfp	C-terminal fusion of PPK1aReu to eYFP	This study
pBBR1MCS2-PphaC1-eyfp-ppk1bReu	N-terminal fusion of PPK1bReu to eYFP	This study
pBBR1MCS2-PphaC1- ppk1bReu-eyfp	C-terminal fusion of PPK1bReu to eYFP	This study
pBBR1MCS2-PphaC1-eyfp-ppk2aReu	N-terminal fusion of PPK2aReu to eYFP	This study
pBBR1MCS2-PphaC1- ppk2aReu-eyfp	C-terminal fusion of PPK2aReu to eYFP	This study
pBBR1MCS2-PphaC1-eyfp-ppk2bReu	N-terminal fusion of PPK2bReu to eYFP	This study
pBBR1MCS2-PphaC1- ppk2bReu-eyfp	C-terminal fusion of PPK2bReu to eYFP	This study
pBBR1MCS2-PphaC1-eyfp-ppk2cReu	N-terminal fusion of PPK2cReu to eYFP	This study
pBBR1MCS2-PphaC1- ppk2cReu-eyfp	C-terminal fusion of PPK2cReu to eYFP	This study
pBBR1MCS2-PphaC1-eyfp-ppk2dReu	N-terminal fusion of PPK2dReu to eYFP	This study
pBBR1MCS2-PphaC1- ppk2dReu-eyfp	C-terminal fusion of PPK2dReu to eYFP	This study
pBBR1MCS2-PphaC1-eyfp-ppk2eReu	N-terminal fusion of PPK2eReu to eYFP	This study
pBBR1MCS2-PphaC1- ppk2eReu-eyfp	C-terminal fusion of PPK2eReu to eYFP	This study
pBBR1MCS2- PphaC1-phaX	Plasmid for complementation of ΔphaX strain	This study
pBBR1MCS2- PphaC1-pitA	Plasmid for complementation of ΔpitA strain	This study

^aKm^r, resistance to kanamycin; Tc^r, resistance to tetracycline.

TABLE 2.

Oligonucleotides used in this study

Two-hybrid assay.

The adenylate cyclase-dependent bacterial two-hybrid assay (BACTH) was used as described previously (33) and as described in detail earlier (34) to screen for proteins that interact in vivo with PHB depolymerase PhaZa1 or phasin PhaP2 or PhaP3 and to quantify the interaction potential between two test proteins.

Construction of fluorescent fusion proteins.

The construction of fusion proteins with enhanced yellow fluorescent protein (eYFP) as a fluorophore was performed as described previously using pBBR1MCS2-P_phaC-eyfp-c1 or pBBR1MCS2-P_phaC-eyfp-n1 as the vector (35). These plasmids allow the constitutive expression of the respective fusion protein (in R. eutropha) from the constitutive phaC1 promoter. All constructs were transformed into E. coli by standard transformation and were subsequently transferred via conjugation from recombinant E. coli S17-1 to R. eutropha H16. Selection was achieved by plating on mineral salts medium supplemented with 0.2% fructose and 15 μg ml⁻¹ tetracycline.

Construction of chromosomal knockouts.

Precise chromosomal deletions of R. eutropha genes were constructed using the sacB-sucrose selection method (15% sucrose was used for selection) and pLO3 as the deletion vector. Deletion of the ppk gene of E. coli LS was performed as described elsewhere (36). The genotypes of all deletion mutants were verified by PCR of the respective genomic region and determination of its DNA sequence.

Microscopic methods.

The formation of PHB granules was followed by fluorescence microscopy using Nile red as the dye (1 to 10 μg/ml dimethyl sulfoxide or ethanol, Nile red solution at 5 to 40% [vol/vol]). PolyP granules were stained with 4′,6-diamidino-2-phenylindole (DAPI; concentration, 60 μg/ml) and detected with the aid of a DAPI-polyP-specific filter set (excitation, 415/20 nm; emission, 520/60 nm). Nile red-stained PHB granules and eYFP were visualized using standard filter sets (for Nile red, excitation was at 562/40 nm and emission was at 594 nm [longpass filter]; for eYFP, excitation was at 500/24 nm and emission was at 542/27 nm). PHB granules could be also visualized by phase-contrast microscopy or by differential interference contrast (DIC). PHB granules became visible as dark globular structures in phase-contrast images.

Enrichment of polyP granules.

A 200-ml NB culture of R. eutropha H16 was harvested after ∼18 h of growth at 30°C. At this stage, the cells usually contained one or two polyP granules per cell. The cells were centrifuged (10 min, 8,000 rpm [GS3 rotor], 4°C), and the pellet was washed with 10 mM HEPES, pH 7, buffer. The pellet was resuspended in 20 ml 10 mM HEPES, and the cells were broken by three repeated French press steps. The crude extract was centrifuged (5 min, 4,500 × g, 4°C), and the supernatant was discarded. The pellet was washed two times with 40 ml HEPES. The remaining pellet was resuspended in 8 ml HEPES and filtered (pore size, 0.45 μm) at 4°C. The filtrate was subsequently ultracentrifuged (35 min, 35,000 rpm [TFT65.13 rotor; Beckmann Coulter, Krefeld, Germany], 4°C). The supernatant was discarded, and the white-gray pellet containing the polyP material was suspended in 100 μl 2% (wt/vol) sodium dodecyl sulfate (SDS) denaturation buffer. For SDS-polyacrylamide gel electrophoresis (PAGE) analysis of polyP-attached proteins, the polyP-SDS fraction was heated to 95°C for 3 min and subsequently centrifuged in an Eppendorf centrifuge at 13,000 rpm at room temperature for 2 min. The supernatant containing the denatured proteins was used for SDS-PAGE and proteome analysis.

Other techniques.

Proteome analysis was performed by the Proteome Core Facility of the Life Science Center, University of Hohenheim (Stuttgart, Germany), as described recently (37). Quantitative analysis of the PHA content was done by gas chromatography (GC) after acid methanolysis of lyophilized cells as described previously (38). Determination of the inorganic phosphate content in isolated polyP granules was performed after treatment with 2.5 M H₂SO₄ at 95°C for 15 min. The phosphate liberated from polyP was allowed to react with ammonium-molybdate (1 g/25 ml) under mild reducing conditions (ascorbate, 1 g/60 ml) to give a molybdophosphate blue complex that was detected spectroscopically at 880 nm. Discontinuous SDS-PAGE was performed under denaturing (sodium dodecyl sulfate) and reducing (mercaptoethanol) conditions.

RESULTS

Identification of a protein (PhaX) as a putative link between PHB and polyP metabolism.

A bacterial two-hybrid screening approach was performed to detect proteins interacting with PHB granule-associated proteins. When the gene for PHB depolymerase PhaZa1 or phasin PhaP2 or PhaP3 was used as the bait gene in pKT25 and a R. eutropha genomic library was used in pUT18C, interactions with several already known PHB granule-associated proteins were (re)identified; for example, 2 clones had PhaZa1/PhaP5 interactions, 1 clone had a PhaZa1/PhaZa5 interaction, 26 clones had PhaP2/PhaP5 interactions, and 1 clone had a PhaP3/PhaP2 interaction. For details, see Table 3. While the interaction between already known PHB granule-associated proteins was expected, a remarkable and unexpected finding was the independent identification of the same A2274 protein for all three bait proteins used in eight clones (Table 3). This finding suggests that the A2274 gene product is somehow connected to PHB granule formation and PHB metabolism. We designated the A2274 gene product the PhaX protein, to indicate a putative but so far functionally unknown link to PHB metabolism. The complete coding sequence of phaX was cloned by PCR into pUT18C and pKT25, and the two-hybrid assay was repeated with all permutations of phaZa1, phaP2, and phaP3 with phaX in liquid cultures, as described in the Materials and Methods section and in a previous contribution (34). As shown in Fig. 1, PhaX showed a strong interaction with itself (530 Miller units [MU]) and with PHB depolymerase PhaZa1 (420 to 570 MU). This indicated that PhaX is able to form homo-oligomers and to interact with PhaZa1 in vivo. The interaction of PhaX with phasin PhaP2 was lower (150 to 280 MU) but was still ∼5-fold above the background level (∼40 MU). The interaction of PhaX with PhaP3, however, was only a little above the background level (∼60 Miller units), and an in vivo interaction of PhaX with PhaP3 at a significant level could not be confirmed. Our data show that full-length PhaX interacts in vivo with at least two proteins involved in PHB metabolism, namely, PhaP2 and PhaZa1. The A2274 (phaX) gene product is annotated as a hypothetical putative phosphate transport regulator, and this finding led to the hypothesis of a relationship of PHB metabolism with phosphate metabolism via PhaX.

TABLE 3.

Two-hybrid screening results using PhaZa1, PhaP2, or PhaP3 as bait^a

^aNote the frequent and independent identification of the A2274 protein (indicated by shading) with three different bait proteins. Protein-protein interactions between previously identified PHB granule-associated proteins are indicated in bold.

FIG 1.

Two-hybrid analysis of PhaX (A2274) with PhaZa1, PhaP2, and PhaP3. Two-hybrid experiments were performed as described previously (35). Experiments were performed in triplicate. Error bars show standard deviations. T18 and T25 refer to adenylate cyclase domains.

Deletion of phaX increases the polyP content.

A mutant with a precise deletion of the phaX gene was constructed and investigated for the formation and subsequent mobilization of PHB granules. Microscopic analysis of the cells after staining with Nile red revealed that ΔphaX cells produced PHB granules similar to those produced by the wild type (WT). However, in addition to PHB granules, we observed the presence of another type of inclusion in ΔphaX cells that was not stained by Nile red. Staining of the cells with DAPI and microscopic examination near the DAPI-polyP-specific emission maximum at 520 nm (39–41) revealed the presence of several prominent polyP granules in ΔphaX cells (Fig. 2). The identification of these granules as polyP granules was confirmed by classical Neisser staining, by staining with tetracycline as described by Müller and colleagues (42), by staining with JC-D7 (43), and by gel electrophoresis and subsequent staining with toluidine blue as described previously (44, 45) (images not shown). Costaining of the cells with Nile red and DAPI clearly showed that two types of granules, PHB and polyP granules, were produced by R. eutropha (Fig. 2). In contrast to the wild type, which produced one to two polyP granules per cell, the number of polyP granules per ΔphaX cell was increased, and most cells harbored several polyP granules. Occasionally, long cells with more than 10 polyP granules were detected (Fig. 2). Moreover, the fluorescence intensity and brightness of the polyP granules in the ΔphaX cells were stronger than those in wild-type cells. This suggested that the polyP granules were larger than those of the wild type. Another feature of the phenotype of ΔphaX cells during growth on NB-gluconate medium was that the cells became considerably longer than the wild-type cells. Complementation of ΔphaX cells with phaX cloned on a plasmid resulted in restoration of the wild-type phenotype with only a small number of polyP granules and a WT morphology of the cells.

FIG 2.

Formation of polyP granules in R. eutropha. PHB-free cells of R. eutropha WT (A) and of the ΔphaX mutant (B) were transferred to NB-gluconate medium, and the formation of polyP granules was examined after DAPI staining at the indicated times. Top rows, images taken with a polyP-DAPI filter; bottom rows, phase-contrast images. A separate double-stained (DAPI and Nile red) image taken at 6 h is shown. Bar, 2 μm.

Deletion of phaX prevents mobilization of the accumulated PHB in the stationary phase.

The ability of PhaX to interact with PHB depolymerase in vivo suggested that PhaX could somehow affect the ability of the cells to mobilize PHB during times of starvation. To test this hypothesis, we performed growth experiments on NB-gluconate medium in which PHB-free cells of R. eutropha accumulated PHB during exponential-phase growth and subsequently reutilized (mobilized) PHB in the stationary growth phase. Figures 3A and B show the growth and PHB content of samples taken during growth. Wild-type and ΔphaX cells grew similarly to final optical densities at 600 nm of 6 to 7. The PHB content of both strains increased from ≤3% of cell dry weight at zero time to ∼45% after 12 h, but the number of polyP granules was higher in ΔphaX cells, in particular in the exponential growth phase, than in wild-type cells (Fig. 2). The wild type reutilized the accumulated PHB in the stationary phase and became PHB free and almost coccoid again after 32 to 48 h (Fig. 3C). In contrast, the PHB content of ΔphaX cells decreased only a little after the end of exponential growth and remained constantly high at 30 to 40% of cell dry weight. Microscopic analysis showed that some ΔphaX cells became very long and harbored many PHB granules and several polyP granules. When the cells were stained with propidium iodide, only a few wild-type cells (<3%) were stained, but a considerable fraction of ΔphaX cells (∼10%) had taken up propidium iodide. This indicates that ΔphaX cells lost their membrane potential and were no longer viable. In particular, most of the unusual long cells with a high number of PHB granules were stained by propidium iodide. In conclusion, ΔphaX cells were strongly impaired in PHB mobilization in the stationary growth phase and had reduced viability. However, it was not clear whether the reduced viability was a consequence of reduced PHB mobilization or whether reduced PHB mobilization was a consequence of reduced viability.

FIG 3.

Growth (A) and formation and mobilization of PHB (B) in the R. eutropha WT and ΔphaX mutant on NB-gluconate medium. (C) Cells of the early (20 h) and late (32 h) stationary growth phase on NB-gluconate medium after Nile red staining. A phase-contrast image is shown at the bottom left. Cells in the image at the bottom right were stained with propidium iodide. Bars, 2 μm. Growth experiments and PHB determinations were repeated more than six times and two times, respectively (biological replicates). The results of a typical experiment are shown. Error bars in panel B indicate standard deviations for three technical replicates. μ, growth rate; t_d, doubling time; DCW, cell dry weight.

The phenotype of the ΔphaX strain (in which the mobilization of PHB and viability in stationary phase were reduced) was reversible by complementation of the ΔphaX mutant with pBBR1MCS2::phaX (Fig. 4A to C) The cells grew and accumulated and mobilized PHB similarly to the wild type. The number of polyP granules in the complemented strain was reduced to the number in wild-type cells (Fig. 4D).

FIG 4.

Complementation of the ΔphaX mutant and phenotype of the ΔphaX ΔphaC1 double deletion mutant. Complementation of the ΔphaX deletion mutant with phaX in trans restored the wild-type phenotype with respect to cell length and cell morphology (A, B), the mobilization of PHB (C), and the number of polyP granules formed (D). For panel C, the PHB contents are average data from two independent experiments performed in triplicate, and error bars show standard deviations. CDW, cell dry weight. (E) The number of polyP granules was increased in the ΔphaX ΔphaC1 double deletion mutant, similar to the findings for the ΔphaX mutant. (F) The increased cell length and reduced viability (determined by propidium iodide staining; bottom middle) of the ΔphaX mutant were dependent on the presence of PHB granule overproduction in NB-gluconate medium and were not seen in the absence of PHB in the ΔphaX ΔphaC1 double deletion mutant. Bars, 2 μm.

The reduced viability of ΔphaX cells depends on the accumulation of PHB in NB-gluconate medium.

The accumulation of a large number of PHB and polyP granules in the ΔphaX mutant resulted in a decrease in the viability of the cells. To investigate whether the reduced viability of the ΔphaX strain was dependent on the accumulation of PHB, the PHB synthase gene (phaC1) was deleted in the ΔphaX background, and the ΔphaX ΔphaC1 double deletion mutant was grown in NB-gluconate medium. The cells were not able to produce any PHB granules, as expected. Remarkably, the number of polyP granules was elevated in the double deletion mutant, similar to the findings for the ΔphaX mutant, but the length of the cells was not increased, and also, no loss of viability of the cells was observed by propidium iodide staining during a 50-h period of growth (Fig. 4E and F). When cell morphology and the viability of the cells were determined on NB medium (without addition of gluconate) or on a mineral salts medium with gluconate as a carbon source, no evidence for substantial cell elongation or a loss of viability of ΔphaX cells was detected (data not shown).

Proteome analysis of ΔphaX cells.

The proteomes of soluble cell extracts of R. eutropha wild-type and ΔphaX cells were determined and quantitatively compared. Forty-four proteins differed in abundance between wild-type and ΔphaX cells by a factor of more than 3 (see Data Set S1 in the supplemental material). Remarkably, the three most upregulated proteins in the ΔphaX mutant had a predicted function in the uptake of inorganic phosphate, namely, a phosphatase (B2398, 53-fold), the response regulator PhoB (A2439, 48-fold), and ABC transport protein PstS (A2444, 33-fold). This result is in agreement with the increased polyP content of the cells.

Identification of ppk genes in R. eutropha.

The results presented above suggested a connection between PHB metabolism and polyP metabolism. However, only very little information on the key enzymes of polyP metabolism in R. eutropha is available. Two putative polyP kinase genes (A2437 [ppk1] and B1019 [ppk2]) are present in the genome of R. eutropha H16. The ppk1 gene is located together with a polyphosphatase gene (A2436 [ppx] in the opposite direction from ppk1) adjacent to a pst operon for high-affinity phosphate transport (see Fig. S1 in the supplemental material) (11). The ppk2 gene is located as a single gene on the B chromosome (B1019).

Interestingly, a bioinformatic screening of the R. eutropha genome for putative ppk1 and ppk2 genes revealed the presence of five ppk2 candidate genes (A0226, A0997, A1212, A1271, A1979), in addition to the already known PPKs (A2437 and B1019). All newly identified genes code for hypothetical proteins and have a predicted PPK2 motif (Table 4). Such a high number of seven potential PPK genes has not yet been described for any species.

TABLE 4.

Properties of PPKs

Species and protein (gene designation in database)	Suggested protein designation	Molecular mass (kDa)	PPK motif	Calculated IEP	R. eutropha localization of eYFP-tagged protein	Conferment of polyP granule formation in recombinant E. coli by the gene (localization of eYFP-tagged protein in E. coli Δppk)
E. coli (ppk)	PPK1Eco	80.4	N, H, C1, C2	9.0	NDa	No
P. aeruginosa
PA5242 (ppk1)	PPK1Pae	83.2	N, H, C1, C2	7.3	ND	ND
PA0141 (ppk2)	PPK2Pae	34.5	PPK2	8.4	ND	ND
R. eutropha
A2437 (ppk1)	PPK1aReu	78.2	N, H, C1, C2	7.8	PolyP granule	Yes (at polyP granule)
B1019 (ppk2)	PPK1bReu	79.6	N, H, C1, C2	5.9	Near one cell pole apart from polyP	No (near one cell pole)
A0226, hypothetical protein	PPK2aReu	31.1	PPK2	6.7	Cytoplasm	No (protein is soluble)
A0997, hypothetical protein	PPK2bReu	36.2	PPK2	5.8	Near one cell pole apart from polyP	Yes (protein is soluble)
A1212, hypothetical protein	PPK2cReu	42.1	PPK2	9.4	PolyP granule	Yes (at polyP granule)
A1271, hypothetical protein	PPK2dReu	33.9	PPK2	9.6	PolyP granule	No (protein is soluble)
A1979, hypothetical protein	PPK2eReu	30.7	PPK2	9.1	PolyP granule	No (protein is soluble)

^aND, not determined.

The two already annotated PPK proteins (A2437 and B1019) are designated PPK1 and PPK2 in the database, respectively. However, both proteins are very similar in amino acid sequence to each other (58% identity, 72% similarity) and to E. coli PPK1 (35% identity and 56% similarity for PPK1 and 35% identity and 57% similarity for PPK2) and to P. aeruginosa PPK1 (53% identity and 68% similarity for PPK1 and 54% identity and 72% similarity for PPK2). Moreover, the A2437 and B1019 gene products had approximately the same predicted molecular masses (78.2 kDa and 79.6 kDa, respectively) as E. coli PPK (80.4 kDa), and both had the same arrangement of N, H, C1, and C2 domains; therefore, they most likely represent PPK1 proteins. We assume that PPK2 (B1019) of R. eutropha has been misannotated and was named simply to differentiate two similar gene products. We therefore designate A2437 and B1019 R. eutropha PPK1a (PPK1a_Reu) and PPK1b_Reu, respectively, from here on. All five remaining PPKs of R. eutropha have a so-called PPK2 motif, and their deduced primary amino acid sequences have molecular masses much smaller than those of PPK1 and PPK2. Accordingly, the five putative PPK2 proteins of R. eutropha were named PPK2a_Reu (A0226), PPK2b_Reu (A0997), PPK2c_Reu (A1212), PPK2d_Reu (A1271), and PPK2e_Reu (A1979). For an overview of the PPKs, see Table 4.

Development of a procedure for isolation of polyP granules.

To determine which of the PPK proteins were expressed and which were attached to polyP granules in R. eutropha, it was necessary to develop a suitable method for the isolation of polyP granules. Cells from 18-h cultures of R. eutropha on NB medium were used for polyP isolation. At this time point the cells were in the stationary growth phase and contained one or two polyP granules, as revealed by staining with DAPI and fluorescence microscopy (Fig. 2). Unfortunately, all efforts to isolate polyP granules on the basis of density gradient centrifugation steps failed (cesium chloride, glycerol, and sucrose gradients were tested; details not shown). Therefore, we enriched polyP granules by filtration of highly diluted pellets obtained from cell extracts by low-speed centrifugation as described in the Materials and Methods section. Despite a substantial loss of polyP granules by the filtration step, we could enrich polyP material, as evidenced from microscopic analysis (after DAPI staining) and by determination of the inorganic phosphate content after chemical hydrolysis of polyP and conversion of liberated phosphate to molybdophosphate blue. The polyP material isolated after centrifugation turned out to be a white-gray pellet material.

Determination of the R. eutropha polyP granule-associated proteins.

The presence of seven potential ppk genes in R. eutropha raised the question of which of the seven ppk genes was responsible for polyP granule formation in R. eutropha. We speculated that polyP-synthesizing enzymes could be bound to polyP granules in vivo, just as PHB synthases are attached to PHB granules and like the polyP-bound glucokinase in Corynebacterium glutamicum (46). The proteins attached to the isolated polyP granule fraction were analyzed by SDS-PAGE. Unfortunately, a large number of bands was visible after Coomassie staining (image not shown), indicating that the polyP granule fraction was strongly contaminated by cellular proteins. In agreement with this, determination of the proteins of the wild-type polyP granule fraction by proteome analysis revealed a large number of proteins (n = 910) in this fraction. In order to eliminate false-positive polyP-bound proteins, we compared the proteome data for the polyP fraction with those that had been determined for R. eutropha in a previous study (37) for a soluble fraction (which contained 1,756 proteins, obtained after ultrafiltration of cell extracts), a membrane fraction (which contained 889 proteins, obtained from a membrane preparation), a membrane-associated fraction (which contained 439 proteins, obtained after an alkaline carbonate wash of a membrane preparation), and a PHB granule fraction (which contained 268 proteins, obtained after glycerol density gradient centrifugation). Interestingly, only 27 proteins that were absent in the four other cell fractions remained in the polyP fraction (Table 5). Three of the seven potential PPK proteins that were identified in the R. eutropha translated genome were present among the 27 proteins of the polyP proteome. These were PPK1a_Reu, PPK2c_Reu, and PPK2d_Reu. Remarkably, PPK1a_Reu and PPK2c_Reu were the 2 proteins most abundant among the 27 proteins of the polyP fraction, a finding that suggested that they could have a prominent function in polyP metabolism. PPK2d_Reu was present only in trace amounts in the polyP granule fraction, and the other PPK proteins were not identified in this fraction. Nevertheless, PPK2a_Reu was present in the soluble fraction.

TABLE 5.

PolyP-specific proteins of R. eutropha^a

Protein no.	Protein identified	Accession no.	Molecular mass (kDa)	Presence of protein in:
				ΔphaC mutant	ΔphaX ΔphaC1 mutant
1	Hypothetical protein H16_A1212 (PPK2c)	gi|113867232	42	x	x
2	Polyphosphate kinase (PPK1a)	gi|113868405	78	x	x
3	Catechol 2,3-dioxygenase	gi|116694497	35	x	x
4	Phenol hydroxylase P3 protein	gi|116694493	59	x
5	Export of exopolysaccharide I polysaccharide export protein, putative tyrosine-protein kinase	gi|116693972	87	x	x
6	Aminotransferase	gi|116693991	41	x	x
7	4-Hydroxy-2-ketovalerate aldolase	gi|116694503	38	x
8	Phenol hydroxylase P1 protein	gi|116694491	37	x	x
9	Exopolysaccharide export protein	gi|116693984	43	x	x
10	Hypothetical protein H16_A0407	gi|113866436	12	x	x
11	Phosphatase	gi|116696334	70		x
12	Glycosyltransferase	gi|116693985	45		x
13	AraC family transcriptional regulator	gi|116696223	34	x	x
14	Hypothetical protein H16_A1108	gi|113867128	25	x	x
15	Hypothetical protein H16_B0037	gi|116693997	38	x	x
16	Outer membrane protein (porin)	gi|116695022	39	x	x
17	Transcriptional regulator	gi|116695076	26	x
18	LysR family transcriptional regulator	gi|116696448	33	x	x
19	AraC family transcriptional regulator	gi|116696194	37	x	x
20	Hypothetical protein H16_B0055	gi|116694015	44	x	x
21	ABC transporter ATPase	gi|113866320	30	x	x
22	Dimethylaniline monooxygenase (N-oxide forming)	gi|113868087	49	x
23	Patatin-like phospholipase	gi|116695035	31	x	x
24	bb3-type cytochrome oxidase, subunit I	gi|116695997	65	x	x
25	Putative double-glycine peptidase	gi|116694017	26	x	x
26	Transcriptional regulator	gi|116693971	27		x
27	AsnC family transcriptional regulator	gi|116695309	17		x
28	Long-chain fatty acid–coenzyme A ligase	gi|116694665	58		x
29	Hypothetical protein H16_A1271 (PPK2d)	gi|113867290	34	x
30	Short-chain coenzyme A dehydrogenase	gi|116695635	26		x
31	LysR family transcriptional regulator	gi|116694486	33		x
32	LysR family transcriptional regulator	gi|116694435	33		x
33	Alkaline phosphatase	gi|113868158	48		x
34	l-Aspartate dehydrogenase	gi|116694687	28		x
35	Hypothetical protein H16_B2377	gi|116696313	19		x
36	Phosphate transporter permease subunit PstC	gi|113868411	34		x
37	Response regulator	gi|116694572	25		x
38	Transcriptional regulator	gi|116695138	37	x
39	ATP-dependent DNA ligase	gi|116696288	98	x
40	Transcriptional regulator	gi|116695723	37	x

^aProteins of a polyP granule fraction prepared from R. eutropha were separated by SDS-PAGE and identified by proteome analysis. From the total number of 910 proteins identified in the wild type (2-peptide threshold), those proteins that were also identified in the soluble, membrane, membrane-associated, or PHB granule fraction (n = 37) were removed. A1212 was the most abundant protein of the polyP fraction, and a few peptide fragments were also present in the soluble fraction. Proteins specifically present in the polyP granule fraction of polyP-overproducing ΔphaX ΔphaC1 strains are also indicated in the last two columns. Boldface indicates PPK proteins.

Four out of seven putative PPK proteins are bound to polyP granules in vivo in R. eutropha.

To verify that the polyP-granule-bound PPK proteins identified in vitro were also attached to polyP granules in vivo, N- and C-terminal gene fusions of all seven putative ppk genes with the gene for eYFP (eyfp) were constructed, cloned under the control of the constitutive phaC1 promoter, and transferred to R. eutropha via conjugation. Expression of eYFP alone (without fusion) resulted in uniform labeling of the cells, as shown previously (47) (see Fig. S2 in the supplemental material). Fusions of PPK1a_Reu (A2437), PPK2c_Reu (A1212), PPK2d_Reu (A1271), and PPK2e_Reu (A1979) formed fluorescent foci that colocalized with DAPI-stained polyP granules (Fig. 5A to D). Fluorescent foci colocalizing with polyP granules were obtained for both types of fusions (N- and C-terminal fusions and N-terminal-only fusions are shown in Fig. 5). This confirmed that PPK1a_Reu, PPK2c_Reu, PPK2d_Reu, and PPK2e_Reu are polyP granule-bound proteins in vivo. Interestingly, expression of eYFP fusions with PPK1b_Reu (Fig. 5E) and with PPK2b_Reu (Fig. 5F) also resulted in the formation of fluorescent foci. However, eYFP-PPK1b_Reu and eYFP-PPK2b_Reu foci were detected exclusively close to one of the cell poles at a position clearly different from the localization of polyP granules. N- and C-terminal fusions of PPK2a_Reu with eYFP were homogeneously distributed in the cytoplasm, and a colocalization with polyP granules was not observed (Fig. 5G; Table 4). In summary, four of the seven potential PPK_Reu proteins (PPK1a_Reu, PPK2c_Reu, PPK2d_Reu, and PPK2e_Reu) were bound to polyP granules, and three of these were detected in the polyP granule fraction by proteome analysis. PPK1b_Reu and PPK2b_Reu localized close to the cell poles, and PPK2a_Reu was apparently a cytoplasm-located enzyme. We also determined the localization of the eYFP-PPK fusion proteins in a ΔphaX background (with additional deletion of phaC1 to remove the bulky PHB granules). In principle, results that were the same as those obtained for the wild type were obtained. However, the cells were longer and had more polyP granules than the wild type (see Fig. S3 in the supplemental material). Interestingly, microscopic analysis of eYFP-PPK1a_Reu overexpressed in a ΔphaX ΔphaC1 background resulted in the detection of several polyP granules, but not all of them were labeled by the eYFP-PPK1a_Reu protein.

FIG 5.

PolyP formation and localization of fusion of eYFP with PPK candidate proteins in the R. eutropha wild type. (A to D) Colocalization of eYFP-PPK fusion proteins with DAPI-stained polyP granules (artificially in red) for eYFP-PPK1a_Reu (A), eYFP-PPK2c_Reu (B), eYFP-PPK2d_Reu (C), and eYFP-PPK2e_Reu (D); (E, F) formation of fluorescent foci near cell poles that do not colocalize with polyP granules for eYFP-PPK1b_Reu (E) and eYFP-PPK2b_Reu (F); (G) localization of eYFP-PPK2a_Reu fusion protein in the cytoplasm (soluble) and no colocalization with DAPI-stained polyP granules. Bars, 2 μm and 1 μm (enlargements in the insets).

PPK1a_Reu, PPK2b_Reu, and PPK2c_Reu confer the ability to synthesize polyP granules in E. coli Δppk.

To determine which of the identified PPKs of R. eutropha has polyP granule-synthesizing activity in vivo, all seven PPK proteins of R. eutropha were expressed in the form of fusions with eYFP in E. coli LJ. To avoid background polyP-forming activity, the endogenous ppk gene of E. coli was deleted and the plasmid harboring the test eyfp-ppk fusion was transformed into E. coli LJ Δppk. Fluorescence microscopic analysis confirmed that E. coli Δppk did not form any polyP granules (see Fig. S4 in the supplemental material). The formation of polyP granules was detected for strains expressing eYFP-PPK1a_Reu and eYFP-PPK2c_Reu (Fig. 6A and B; Table 4). In both cases, the fluorescence of the fusion proteins colocalized with the polyP granules that were formed, and this finding confirmed that the fusion proteins represent polyP-bound proteins, similar to the findings for R. eutropha (Table 4). Interestingly, E. coli Δppk cells expressing the eYFP-PPK2b_Reu fusion formed several polyP granules per cell, as verified by DAPI staining, but the eYFP-PPK2b_Reu fluorescence was present uniformly in the cytoplasm (Fig. 6C). No formation of polyP granules was detected in E. coli Δppk cells expressing eYFP-PPK1b_Reu, eYFP-PPK2a_Reu, eYFP-PPK2d_Reu, and eYFP-PPK2e_Reu (Fig. 6D to G). The fusion protein eYFP-PPK1b_Reu formed fluorescent foci at the cell poles in E. coli Δppk cells, as was the case in R. eutropha. The other three fusions were soluble in E. coli (Table 4).

FIG 6.

PolyP formation and localization of fusions of eYFP with PPK_Reu candidate proteins in E. coli Δppk. (A, B) Colocalization of eYFP-PPK fusion proteins with DAPI-stained polyP granules (red) for eYFP-PPK1a_Reu (A) and eYFP-PPK2c_Reu (B); (C) polyP granule formation but no colocalization for eYFP-PPK2b_Reu; (D) formation of fluorescent foci near the cell pole for eYFP-PPK1b_Reu; (E to G) no formation of polyP granules and soluble eYFP-PPK fusion proteins for eYFP-PPK2a_Reu (E), eYFP-PPK2d_Reu (F), and eYFP-PPK2e_Reu (G). Bars, 2 μm.

PPK1a_Reu and PPK2c_Reu are responsible for polyP granule formation in R. eutropha.

Precise chromosomal deletions of ppk1a_Reu, ppk1b_Reu, ppk2c_Reu, and ppk2d_Reu were constructed, and polyP granule formation was determined for each of the individual deletion strains. The Δppk1b_Reu strain and the Δppk2d_Reu strain produced polyP granules indistinguishable in number and size from those produced by the wild type (images not shown). In contrast, the level of polyP granule formation in strains with a deletion either in Δppk1a_Reu or in Δppk2c_Reu was reduced. PolyP granules were occasionally found in some cells of the two mutants only in the late stationary phase. Apparently, both Δppk1a_Reu and Δppk2c_Reu contribute to polyP granule formation. We additionally constructed a Δppk1a_Reu Δppk1b_Reu double deletion mutant and a Δppk1a_Reu Δppk1b_Reu Δppk2c_Reu triple deletion mutant. The double deletion mutant still was able to occasionally form polyP granules similarly to the Δppk1a_Reu single deletion mutant, and that finding confirmed that ppk1b_Reu is apparently not essential for polyP granule formation. The triple deletion mutant, however, completely lost its ability to form polyP granules. We conclude that both ppk1a_Reu and ppk2c_Reu contribute to polyP granule formation and their products represent the polyP granule-forming enzymes in R. eutropha.

Formation and mobilization of PHB in polyP granule-deficient and polyP granule-overproducing R. eutropha mutants.

To test for an assumed connection between PHB and polyP metabolism, we followed the formation and utilization of PHB and polyP granules in the R. eutropha wild type and in mutants either that were unable to produce polyP granules (the Δppk1a Δppk1b Δppk2c triple deletion mutant) or that overproduced polyP granules (the ΔphaX mutant and the eYFP-PPK2c-overproducing mutant). A constructed chromosomal ΔpitA mutant was also investigated. Deletion of the pitA low-affinity phosphate transporter gene led to the overproduction of polyP granules (see Fig. S5 in the supplemental material) that was reversible by complementation of the mutant with a pitA-containing plasmid, similar to the findings for the ΔphaX mutant, presumably by constitutive expression of the high-affinity Pst phosphate transport system. All strains were first depleted of any residual PHB granules by two subsequent passages on NB medium. The cells were then transferred to fresh NB medium that had been supplemented with 0.2% sodium gluconate to increase the C/N ratio. The R. eutropha WT intermediately accumulates PHB to about 50% of the cell dry weight on this medium and subsequently degrades PHB in the stationary growth phase within 48 h (48). The formation of PHB and of polyP granules was followed microscopically by staining with Nile red and DAPI. The accumulation of massive amounts of PHB granules was observed for all strains within the first 10 h of growth. Notably, the eYFP-PPK2c-overproducing strain grew considerably more slowly than the other strains and formed unusually long, filamentous cells. The wild type and the triple ppk mutant (the Δppk1a Δppk1b Δppk2c mutant) formed normal-shaped cells with few (1 or 2 per cell) and no polyP granules, respectively. PolyP granules in the wild type were detected in the early stages of growth (0 to 4 h) and at the end of exponential growth and in the stationary growth phase (>8 h). In contrast, the ΔpitA mutant, the ΔphaX mutant, and the eYFP-PPK2c-overproducing strain formed considerably more polyP granules than the WT, and polyP granules were detected at all stages of growth (Table 6). The maximum amount of polyP granules was observed during exponential growth (4 to 8 h) in the ΔpitA mutant, the ΔphaX mutant, and the eYFP-PPK2c-overproducing strain. In comparison to the findings for the WT and the triple ppk mutant, the three polyP-overproducing mutants became substantially elongated during growth. In particular, the eYFP-PPK2c-overproducing strain formed very long cell filaments. A remarkable result was obtained when the mobilization of PHB granules was followed in the stationary growth phase: most cells of the three polyP-overproducing strains (the ΔpitA mutant, the ΔphaX mutant, and the eYFP-PPK2c-overproducing mutant) were impaired in their ability to degrade the PHB that accumulated in the stationary growth phase, while the number of polyP granules slightly decreased in the stationary growth phase in all three strains. Most of the mutant cells, in particular, cells of the eYFP-PPK2c-overproducing strain, remained very long, while WT cells and cells of the triple ppk deletion mutant became much shorter and almost coccoid in the stationary growth phase. Apparently, the forced overproduction of polyP granules by the mutant genotype (ΔpitA, ΔphaX, and the eYFP-PPK2c-overproducing mutant genotype) and the accumulation of massive amounts of PHB granules by application of C/N at a high ratio in NB gluconate medium are not compatible and led to an inability to utilize the PHB that accumulated in the stationary growth phase.

TABLE 6.

Formation of polyP and PHB granules in R. eutropha mutants during growth on NB-gluconate medium^a

Time and genotype or phenotype	Cell length	Presence of:
		PolyP granules	PHB granules
0 h
WT	+	+	−
Δppk1aReu Δppk1bReu Δppk2cReu	+	−	−
ΔphaX	+	+	−
ΔpitA	+	+	−
eYFP-PPK2cReu overproducer	++	+	−
2 h
WT	+	+/−	+
Δppk1aReu Δppk1bReu Δppk2cReu	+	−	+
ΔphaX	++	+	+
ΔpitA	++	+	+
eYFP-PPK2cReu overproducer	++++	++	+
4 h
WT	++	+/−	+
Δppk1aReu Δppk1bReu Δppk2cReu	++	−	+
ΔphaX	++	++	+
ΔpitA	++	++	+
eYFP-PPK2cReu overproducer	+++++	++	+
6 h
WT	++	+/−	+
Δppk1aReu Δppk1bReu Δppk2cReu	++	−	+
ΔphaX	++	++	++
ΔpitA	++	++	++
eYFP-PPK2cReu overproducer	+++++	++	+
8 h
WT	++	+/−	++
Δppk1aReu Δppk1bReu Δppk2cReu	++	−	++
ΔphaX	+++	+++	+++
ΔpitA	+++	+++	+++
eYFP-PPK2cReu overproducer	+++++	++	+++
12 h
WT	+	+	++
Δppk1aReu Δppk1bReu Δppk2cReu	+	−	++
ΔphaX	++	+++	++++
ΔpitA	++	+++	++++
eYFP-PPK2cReu overproducer	+++++	+++	++++
24 h
WT	+	+	+
Δppk1aReu Δppk1bReu Δppk2cReu	+	−	+
ΔphaX	+/++	+/−	+++
ΔpitA	+/++	+/−	+++
eYFP-PPK2cReu overproducer	+/++++	++	+++
49 h
WT	+	+	−
Δppk1aReu Δppk1bReu Δppk2cReu	+	−	−
ΔphaX	+/++	+/−	−/+++
ΔpitA	+/++	+/−	−/+++
eYFP-PPK2cReu overproducer	+/+++	++	−/++

^a+ to +++++, weak to extreme; −, not present; pairs of symbols separated by a slash (e.g., +/− or +/++) represent inhomogeneous cell populations.

DISCUSSION

PolyP is a ubiquitous compound that can be formed abiotically by heating of phosphate (e.g., by volcanism) or biologically by enzymatic condensation of phosphate groups using ATP or GTP as a phosphate donor. The phosphoanhydride bonds formed in polyP are rich in energy and can replace ATP in some biochemical reactions (49). Besides the function as a reservoir for phosphorus, several other functions of polyP have been discussed in the literature. In bacteria, multiple pieces of evidence that polyP is not essential for survival but enhances survival under several stress conditions have been published (14, 50–52). Recently, some evidence for a connection of polyP metabolism with PHA metabolism in Pseudomonas putida was shown (53). PolyP is usually deposited intracellularly in the form of polyP granules and can be specifically stained with basic dyes, such as methylene blue or toluidine blue (44, 54). PolyP granules (also designated voluntin granules in the literature) were frequently found in bacteria [for a selection of descriptions of polyP in different bacterial species, see references 46 and 55 to 59]. In prokaryotes, polyP can be present in soluble or insoluble forms. Depending on the concentration of polyP and on the presence and concentrations of cations, polyP forms insoluble metal (K⁺, Mg²⁺, or Ca²⁺) complexes that can be detected as granular inclusions in the cells by staining with appropriate dyes. Remarkably, polyP granules of two species (Agrobacterium tumefaciens and Rhodospirillum rubrum) are supposed to accumulate in the form of membrane-enclosed vesicles (60, 61). The intracellular compartments of polyP granules in these species are acidic due to the action of a vacuolar pyrophosphatase (PPase; H⁺-V-PPase) that is located in the polyP membrane (62). Acidic polyP granules in A. tumefaciens and R. rubrum have been designated acidocalcisomes, which are similar to the acidocalcisomes in yeasts and in other eukaryotic cells (63, 64).

Enzymes that catalyze the formation of polyP are polyP kinases (PPKs) (for an overview of PPKs, see references 14 and 22). In E. coli, PPK is present in the nonsoluble fraction and catalyzes the formation of soluble polyP from ATP (19). Microscopically detectable polyP granules are only rarely formed in E. coli; e.g., they are formed under oxidative stress conditions (65), but they can be very large in other bacteria and contribute to a substantial amount of the cellular dry weight. The starting point of this study was the multiple identification of the A2274 gene product (PhaX) as a prey protein in two-hybrid screening experiments with PHB depolymerase PhaZa1 and with phasin proteins PhaP2 and PhaP3 as bait proteins. PhaX is annotated as a putative phosphate transport regulator and forms an operon with pitA, the low-affinity phosphate transporter. Deletion of phaX resulted in a strong increase in the number of polyP granules (Fig. 2B and D), probably because of a direct or indirect release of inhibition of the high-affinity Pst phosphate transporter. This assumption is supported by proteome analysis of soluble proteins of the ΔphaX mutant that showed a strong (53-fold) upregulation of a phosphatase and 33-fold and 48-fold increases in the levels of the PstS and PhoB proteins, respectively, compared to the level for the wild type (see Data Set S1 in the supplemental material). PstS is part of the high-affinity ATP-dependent Pst phosphate transport system, and PhoB is the response regulator of the Pho regulon. PhaX is able to interact with PHB depolymerase PhaZa1 and with phasin PhaP2 (Fig. 1). One could speculate that the cellular availability of free PhaX at physiological (low) concentrations is influenced by the physiological state of the cell with respect to PHB synthesis and PHB mobilization. The binding of PhaX to PHB granules, to active PHB depolymerase, and/or to phasin PhaP2 would decrease the concentration of free PhaX and could then lead to the increased activity of the high-affinity Pst phosphate transport system, as observed for the ΔphaX mutant. However, so far we have no evidence for a different behavior of PhaX binding to PHB granules in different metabolic situations. We detected PhaX in an isolated PHB granule fraction of R. eutropha in only one previous proteome experiment (8).

An interesting observation was made when the formation and utilization of PHB in cells that produce either no polyP granules (the Δppk1a Δppk1b Δppk2c triple deletion mutant) or that overproduce polyP granules (the ΔpitA, ΔphaX, or eyfp-ppk2c-expressing strain) were investigated. Independently of the nature of the event that led to the overproduction of polyP granules, the simultaneous overproduction of PHB granules in NB-gluconate medium resulted in severe impairment of PHB utilization in the stationary growth phase and in defects of cell division, as revealed by the formation of unusually long or even filamentous cells. We speculate that the transit from C/N at a high ratio to C/N at a low ratio is a cellular signal to mobilize the accumulated PHB for maintenance metabolism and energy generation and to simultaneously accumulate P_i in the form of polyP as a reservoir for phosphorus. NB medium mainly consists of amino acids and oligopeptides. The end of exponential growth on NB medium is probably caused by the lack of amino acid availability. In E. coli, an intracellular decrease of the amino acid pool results in induction of the stringent response, for which the concentration of ppGpp is a cellular signal molecule. Since ppGpp has an impact on polyP formation (14), PHB metabolism (66), and the EBPR process (67), the overproduction of polyP and PHB might result in futile effects in R. eutropha. It will be necessary to determine the levels of ppGpp in R. eutropha WT and polyP-overproducing mutants in future.

Knowledge of the cell biology of polyP granules in prokaryotes is minimal. PolyP granules of A. tumefaciens and R. rubrum are related to the acidocalcisomes of eukaryotes and are surrounded by a membrane (60, 61). However, it is unlikely that polyP granules of R. eutropha are covered by a membrane. No evidence for a membranous surface layer can be detected by transmission electron microscopy of thin sections of R. eutropha cells (68) or by recent cryotomography results (18). Finally, the R. eutropha genome does not contain a gene for a vacuolar pyrophosphatase (H⁺-V-PPase) homologue like that present in A. tumefaciens and R. rubrum. We conclude that polyP granules in R. eutropha are not covered by a membrane. However, we have evidence that several proteins are bound to the surface of polyP granules in vivo. At least four PPK enzymes clearly colocalized with polyP granules. These were PPK1a_Reu, PPK2c_Reu, PPK2d_Reu, and PPK2e_Reu (Table 4). With the exception of PPK2e_Reu, all PPK enzymes were expressed in R. eutropha cells during growth on NB medium. Due to the polyP granule-negative phenotype of a Δppk1a_Reu ppk2c_Reu mutant, we conclude that PPK1a_Reu and PPK2c_Reu are the most important polyP-synthesizing PPK proteins of R. eutropha. These results are supported by the formation of polyP granules in E. coli Δppk in the presence of either ppk1a_Reu or ppk2c_Reu. Since expression of ppk2d_Reu or ppk2e_Reu in E. coli Δppk did not result in polyP granule formation, we speculate that these two enzymes function more in the utilization of polyP for ATP or GTP synthesis. The function of PPK1b_Reu or PPK2b_Reu is unclear. Remarkably, both proteins are located close to the cell poles in R. eutropha and may have special functions.

Calculation of the isoelectric points (IEPs) of PPK proteins revealed an interesting relationship (Table 4): all four R. eutropha PPK proteins that colocalized with polyP granules in vivo had basic IEPs (7.8 to 9.6). A positive charge of PPKs may facilitate binding of PPKs to the highly charged polyanion polyP. The PPK proteins of E. coli and P. aeruginosa also had IEP values above 7. In contrast, PPK1b_Reu, PPK2a_Reu, and PPK2b_Reu had acid IEPs, and none of the three proteins colocalized with polyP granules. Presumably, these proteins so far have an unknown function and apparently are not necessary for the synthesis of polyP granules. We predict that true PPK proteins with in vivo polyP granule-synthesizing activity in other species will generally have basic IEPs.

Recently, Henry and Crosson showed that polyP granules in Caulobacter crescentus are actively partitioned during cell division (69). The images of polyP granules shown in Fig. 2 suggest that stationary (resting) R. eutropha wild-type cells generally have one polyP granule and actively dividing cells have two polyP granules. These findings are compatible with the attachment of polyP granules to the (dividing) nucleoid, similar to the attachment of PHB granules via PhaM in R. eutropha (35, 70) or to PHA granules via PhaF in Pseudomonas putida (71, 72). We will analyze the cell biology of polyP granule formation in R. eutropha in the future.