The Multiple Roles of Polyphosphate in Ralstonia eutropha and Other Bacteria

The presence of insoluble phosphate-containing inclusions in microorganisms has been known for a long time. The first descriptions go back to Babes [Babes 1885] and Liebermann [Liebermann 1888] and were first summarized by A. Meyer in 1904 [Meyer 1904]. PolyP is also known as metaphosphate or volutin granules (identification of polyP in Spirillum volutans) because of the metachromic shift of basic dyes upon binding to polyphosphate [Wiame 1947]. PolyP was regularly found in yeast [Widra 1959] and in several bacterial species such as Aerobacter aerogenes [Smith et al. 1954] and in particular in Mycobacterium and Corynebacterium species [Sall et al., 1958; Knaysi 1959]. It became evident that volutin (polyP) granules are widespread in microorganisms (for early reviews see [Widra 1959; Harold 1966; Kulaev and Vagabov 1983], and meanwhile it is well accepted that polyP is present in all kingdoms of life [Kornberg et al., 1999; Kulaev and Kulakovskaya 2000; Rao et al., 2009] and presumably is present in every species. For the most recent reviews on polyP see [Albi and Serrano 2016; Jiménez et al., 2016; Xie and Jakob 2018].

Many functions have been ascribed to polyP. The most evident one is that of a storage compound for phosphorus and the counter-ions to the phosphate anions such as Ca²⁺, Mg²⁺, K⁺. However, additional functions have been addressed to polyP: polyP can be a reservoir for energy due to its energy-rich phosphor-anhydride bonds and polyP can substitute ATP in polyP-dependent phosphorylation (kinase) reactions. PolyP is directly or indirectly involved in pathogenicity [Parks and Hobden 2005; Peng et al., 2016; Kumar et al., 2016; Srisanga et al., 2019] and in motility [Rashid et al., 2000; Shi et al., 2004; Zhang et al., 2005; Fraley et al., 2007; Hossain et al., 2008]. PolyP can modulate the cellular responses to various stresses and is important for the adaptation to stationary phase conditions [Rao and Kornberg 1996; Nikel et al., 2013; Casey et al., 2013; Gray and Jakob 2015]. In Escherichia coli, polyP is part of the stringent response to nutrient downshifts [Kulaev and Vagabov 1983; Rao et al., 1998; Kornberg et al., 1999; Kuroda et al., 1999; Kulaev and Kulakovskaya 2000; Kuroda et al., 2001], recently updated by M. Gray [Gray 2019]. PolyP may be also involved in the conversion of (pathogenic) bacteria into the viable but non culturable (VBNC) state [Gangaiah et al., 2009].

Research of the last two decades showed that polyP is involved in and important for tolerance against stresses (heat, UV light, solvents, neutrophilic oxidants/hypohalous acids, heavy metals, antibiotics, survival in stationary phase and others) [Kim et al., 2002; Fraley et al., 2007; Seufferheld et al., 2008; Varela et al., 2010; Nikel et al., 2013; Tocheva et al., 2013; Alcántara et al., 2014; Gray and Jakob 2015; Bru et al., 2017; Groitl et al., 2017; Sultana et al., 2020]. Compounds that inhibit the formation of polyP in bacteria (mesalamine = 5-amino-salicylic acid) reduce the polyP content in intestinal bacteria in mammalians and in turn sensitize them to host defense by reactive oxygen species. This could explain the positive effect of mesalamine on patients suffering from ulcerative colitis [Dahl et al., 2017]. Recently published results provide evidence both for eukaryotes (yeast) and prokaryotes (Pseudomonas aeruginosa) that polyP is necessary for cell cycle control, cell cycle exit and repair of DNA damage [Bru et al., 2016; Racki et al., 2017; Bru et al., 2017]. PolyP has also a prominent role in bacteria that are part of the biological phosphate removal process (EBPR) in sewage sludge [Blackall et al., 2002; Seviour et al., 2003; Yang et al., 2017].

In proteins, polyP can be covalently bound to lysine residues in a process named polyphosphorylation. This posttranslational modification can modulate the protein activity similarly as other covalent modifications [Azevedo et al., 2015; Azevedo and Saiardi 2016]. Recently, it was shown that polyP can have a chaperone-like function by binding to partially denaturated/unfolded proteins and keeping them in a refolding competent state [Gray et al., 2014]. In mammalians, polyP is part of the blood coagulation system (pre-activation of Hageman factor XII by binding to polyP). Furthermore, polyP can interact with proteins present in the brain of mammalians. PolyP is able to bind to preforms of α-synuclein fibrils thereby preventing their uptake by neurones. α-Synuclein is a key protein in neurodegenerative diseases such as Alzheimer or Parkinson [Yoo et al., 2018; Lempart and Jakob 2019; Lempart et al., 2019] and can form amyloid-like aggregates that are typical indications occurring prior to the recognizable outbreak of such diseases. PolyP can protect the organism from the formation of pathogenic amyloid plaques by binding to partially unfolded α-synuclein proteins and preventing them from further misfolding. These findings suggest that polyP has a prominent function in neurodegenerative diseases.

PolyP kinases (PPKs) catalyze the reversible formation of polyP by transfer of a γ-phosphate group from ATP (or another NTP) to a growing chain of polyP. The presence of an oligophosphate primer is not necessary to initiate the reaction. The first isolated and biochemically characterized PPK was that of E. coli [Ahn and Kornberg 1990; Kumble et al., 1996]. The PPK of E. coli is the prototype of so-called type 1 PPKs (PPK1s) which are proteins with molecular masses of ≈80 kDa and are composed of four domains (N-terminal domain (N), head domain (H), and two C-terminal domains (C1 and C2)) [Zhu et al., 2005]. A second type of PPK (so-called PPK2s) has been identified first in P. aeruginosa [Zhang et al., 2002; Ishige et al., 2002] and later in many other bacteria. Currently known PPK2s mostly have about half of the molecular masses of PPK1s (35–40 kDa) and are characterized by a so-called PPK2-domain. PPK2 of P. aeruginosa prefers the reverse reaction, i.e., the nucleotide dikinase reaction (NTP synthesis from polyP and NDPs) and works best with GDP compared to ADP and other NDPs. Therefore, PPK2 of P. aeruginosa might have a function to provide enough GTP for example for exopolysacharide biosynthesis [Ishige et al., 2002] and this might be one reason for polyP being important for virulence. Phylogenetic analysis of the amino acid sequences and comparison with biochemical properties of isolated PPK2s revealed that PPK2s can be divided into three subgroups [Motomura et al., 2014]: members of subgroup I and II catalyze the phosphorylation of nucleoside diphosphates and nucleoside monophosphates, respectively, while subgroup III PPK2s can catalyze both reactions.

Analysis of prokaryotic genome sequences revealed that bacteria can have either a ppk1 gene, a ppk2 gene or both a ppk1 and ppk2. Many bacterial species have even multiple copies of ppk1 and ppk2 genes. Ralstonia eutropha, for example, has two ppk1 genes (ppk1a, ppk1b) and five ppk2 genes (ppk2a – ppk2e) [Tumlirsch et al., 2015]. The presence of multiple genes for both types of PPKs suggests that they fulfill different functions some of which presumably have not yet been identified. Biochemically characterized PPK2s have a low substrate specificity and accept both purine and pyrimidine nucleotides [Motomura et al., 2014]. PPK2c of R. eutropha is the most unspecific PPK among biochemically characterized PPKs and accepts all natural ribo- and desoxyribonucleotides [Hildenbrand et al., 2020]. The purified protein is even able to form microscopically detectable polyP granules in vitro from NTPs [Hildenbrand et al., 2019]. Some PPKs, in particular PPK2s, not only form NTPs from polyP and NDPs but also catalyze the formation of oligo-phosphorylated nucleosides with more than three phosphate units. The first example was PPK1 of E. coli that catalyzed the formation of guanosine tetraphosphate from GDP and polyP [Kuroda and Kornberg, 1997]. In 2019, Mordhorst et al. showed that the PPK2s of Meiococcus ruber, Sinorhizobium meliloti, Francisella tularensis and Acinetobacter johnsonii formed tetra- and penta-phosphorylated adenosine from ADP and polyP [Mordhorst et al., 2019]. Very recently, another PPK2, PPK2 of Agrobacterium tumefaciens, was found to catalyze the formation of highly phosphorylated nucleosides up to the nona-phosphates from any NDP in the presence of polyP [Frank et al., 2020]. The physiological functions, that these oligophosphorylated nucleosides might have, are not known.

R. eutropha strain H16 (alternative designation Cupriavidus necator) is a famous β-proteobacterium due to its ability to grow chemolithoautotrophically (H₂/CO₂) and to accumulate large amounts of the “bioplastic” poly(3-hydroxybutyrate) (PHB) [Pohlmann et al., 2006]. Another interesting property of R. eutropha is the presence of seven ppk genes in its genome. At least three PPKs, PPK1a, PPK2b and PPK2c, contribute to the biosynthesis of polyP granules and four of them (PPK1a, PPK2c, PPK2d and PPK2e) are associated with polyP granules in vivo as revealed by fusion analysis with the enhanced yellow fluorescent protein (eYFP). PPK1b and PPK2b are localized near one of the cell poles and form fluorescent foci (apart from polyP granules) when fused to eYFP, whereas eYFP-PPK2a is soluble in the cytoplasm [Tumlirsch et al., 2015]. Only one of the seven PPKs of R. eutropha, PPK2c, has been biochemically characterized so far [Hildenbrand et al., 2019; Hildenbrand et al., 2020] and showed a very broad nucleotide specificity. The main function of PPK2c presumably is to replenish GTP and other NTP pools during times of enhanced demand on the expense of previously accumulated polyP. The properties of the other PPKs in R. eutropha are currently determined in our laboratory. Several proteins are associated with polyP granules in vivo in addition to the four PPKs (Fig. 1). These are PptA, PptB, PPI18 and PPI27: PptA and PptB both have a so-called conserved histidine alpha-helical domain (CHAD) [Iyer and Aravind 2002; Tumlirsch and Jendrossek 2017]. Expression of fusions of pptA or pptB with eyfp and staining of the cells with 4′,6-diamidin-2-phenylindole (DAPI) resulted in cell-pole localized polyP granules with attached eYFP-PptA or eYFP-PptB in contrast to polyP granules residing in the middle region of the cell as usual. The same result was obtained when fusions of eyfp with CHAD-encoding genes of other proteobacteria were expressed in R. eutropha. Since eYFP-CHAD proteins were soluble in a polyP-deficient background (R. eutropha in which all seven ppk genes had been deleted), these results indicate that CHAD proteins specifically bind to polyP surfaces and can direct the polyP granules to the cell poles if they are overexpressed (Ppt stands for polyP targeting). The designation of polyP-associated CHAD-motif containing proteins as “phosins” in analogy to phasins, that are polyhydroxyalkanoate (PHA) associated proteins without obvious catalytic function, has been suggested for PptA and PptB [Tumlirsch and Jendrossek 2017]. Meanwhile, phosins were identified independently in other species and their binding specificity to polyP has been demonstrated [Lorenzo-Orts et al., 2019; Werten et al., 2019]. The functions of the two remaining polyP associated proteins in R. eutropha (PPI18 and PPI27; PPI stands for polyP interacting protein) is unknown. Both proteins co-localize with polyP granules (as fusion with eYFP) in some but not in all growth stages [Tumlirsch 2017].

The unusual high number of seven ppk genes in the genome of R. eutropha suggests that polyP has multiple, presumably more functions than only that of a storage material for phosphorus. We thought that the exceptionally high number of ppk genes predestinates R. eutropha as an ideal species to study the different functions of polyP by knocking out ppk genes and to study the phenotypic effects, and this was the basis of the present study. E. coli was included for comparison.

PolyP granules can be visualized by staining cells with DAPI and imaging at a DAPI-polyP-specific wavelength of ≈515 nm (in comparison to DAPI-DNA at ≈465 nm) [Klauth et al., 2006]. R. eutropha forms one or two DAPI-stainable discrete polyP granules (Fig. 2 of Tumlirsch and Jendrossek [2017]) when the cells are grown under optimal supply with nutrients (such as nutrient broth medium) and appropriate physical parameters (moderate temperature [30°C] and pH [≈7]). PolyP granules in R. eutropha cells are located in the central (nucleoid) region. In contrast, E. coli is known to synthesize polyP only under stress conditions such as nutrient downshift, elevated temperature or oxidative stress [Rao et al., 1998; Kornberg et al., 1999; Gray and Jakob 2015] but the formation of polyP granules in non-stressed E. coli wild-type strains has not been demonstrated so far. To determine how polyP formation in R. eutropha is influenced by stress conditions, we analyzed the formation of polyP before and after exposure to different stress conditions. For most experiments, E. coli was used as a control. The experiments were performed with both wild-type (WT) strains and the corresponding polyP-free mutant strains (R. eutropha ∆ppk-all in which all seven ppk genes had been deleted and E. coli ∆ppk; E. coli has only one ppk gene [Akiyama et al., 1992]).

R. eutropha and E. coli (WT and ∆ppk mutants) were grown on NB or LB medium at 30 and 37°C, respectively, and the polyP contents were determined by fluorescence microscopy and polyP extraction from lyophilized samples. PolyP determination was repeated after applying a short-time (0–20 min) heat stress (55°C). Figure 2a shows E. coli and R. eutropha cells stained with DAPI before and after a heat stress impulse. Discrete DAPI-polyP foci were not detected in E. coli cells under any condition. Occasionally, E. coli cells with diffuse DAPI signals in the cell periphery or near the cell poles were observed. When the cells were imaged in brightfield, heat-stressed cells often revealed dark bubble-like signals near the poles and/or the cell periphery. Some of these signals seemed to co-localize with the DAPI signal. However, the DAPI and the bright field signals were detected not only in the WT but were also found in the ∆ppk mutant to the same extent. Therefore, these signals are unlikely to indicate the presence of polyP. They might be a response of the cells to the heat stress and could represent aggregates of heat-damaged cell components. In contrast, R. eutropha WT cells formed one or two distinct, globular-shaped granule-like structures that were clearly detectable in the DAPI-polyP channel. This was independent of whether a heat stress impulse was applied to the cells or not. Globular, DAPI-stainable structures were not observed in the R. eutropha ∆ppk-all mutant under any condition being in agreement with the polyP-negative phenotype of this mutant in our previous work [Tumlirsch and Jendrossek 2017]. In summary, our data indicate that microscopically detectable polyP granules are not formed in E. coli either in heat-stressed or in non-stressed cells but that polyP granules in R. eutropha WT are formed independent of the application of a heat-stress impulse.

Next, we directly determined the amount of formed polyP in both species by polyP extraction and subsequent colorimetric phosphate quantification. Non-stressed E. coli cells (WT and the ∆ppk mutant) did not contain significant amounts of polyP (<0.3 µmol P_i/g of cell dry weight [cdw]) (Table 1). Very low amounts of polyP of 1.0 ± 0.4 µmol P_i/g cdw were detected in heat-stressed E. coli WT cells, but no significant polyP content was detected in the ∆ppk mutant after heat stress. Apparently, E. coli is able to synthesize trace amounts of polyP after a heat-stress impulse, but this amount is too low to be reliably detected by fluorescence microscopy. In contrast, much higher levels of ≈60 µmol P_i/g cdw were determined for R. eutropha WT that did only marginally increase to 65 (in NaCl) or 67 (in NB medium) µmol P_i/g cdw upon heat stress. In an independent repetition of this experiment, we even determined a slight decrease in the polyP content after heat stress (not shown). From these data, we conclude that the polyP levels in R. eutropha are generally much higher than in E. coli but that heat stress does not influence the polyP levels to a large extent in the two species. As expected, no polyP was detected in the ∆ppk-all mutant under any condition and confirmed the polyP-negative phenotype of the mutant strain.

To determine, whether polyP could have an impact on survival of the bacteria after exposure to a heat stress impulse, we determined survival of WT cells and of the polyP-free ∆ppk-all mutant cells after the heat stress (0–20 min 55°C). However, we could not find a significant difference in the survival of the WT versus the polyP-free mutant (suppl. Fig. s1). When the survival of E. coli cells after exposure to a heat stress impulse was tested analogously, E. coli WT and the ∆ppk mutant revealed a similar sensitivity (suppl. Fig. s2). Only if highly diluted cell suspensions of OD = 0.002 were heat-challenged, E. coli WT cells exhibited higher viability compared to the ∆ppk mutant. We do not know why the ∆ppk mutant was more heat sensitive at diluted cell concentrations versus high cell concentrations.

The polyP levels in R. eutropha did not change dramatically after application of a short time heat impulse as shown above. To evaluate whether polyP could be advantageous for growth at temperatures near the grtemperature maximum we compared growth of R. eutropha WT and the polyP-free ∆ppk-all mutant at different temperatures. As evident from Figure 3, no difference in growth was determined between R. eutropha WT and the polyP-free ∆ppk-all mutant. Both strains showed good growth at 30–42°C but could not multiply at 45 or 47°C. At 42°C, growth of both strains was slightly reduced compared to 37 or 30°C suggesting that 42°C is near the upper temperature limit of R. eutropha. However, the WT strain and the ∆ppk-all mutant showed the same growth characteristics at all tested temperatures. We conclude that polyP has no function in heat tolerance in R. eutropha and does not allow a better growth at high temperatures.

Many reports have previously described the beneficial effects of polyP for resistance of E. coli and other species against oxidative stress such as hydrogen peroxide or bleach [Casey et al., 2013; Dahl et al., 2017; Groitl et al., 2017; Sultana et al., 2020]. To test if oxidative stress has an impact on survival of R. eutropha, we exposed the cells with 2 mM bleach (HOCl). E. coli cells were also tested for comparison. To this end, the WT and polyP-free mutant strains of both species were incubated in MOPS-buffered medium supplemented with 1.3 mM phosphate and 0.2% fructose (R. eutropha) or 0.2% glucose (E. coli). After incubation for 3 h, the cells were challenged with 2 mM bleach for 15 min. Cell survival was estimated by spotting appropriate dilutions on agar media and was additionally monitored by the fraction of cells that were resistant to the uptake of propidium iodide (PI). Although the number of PI-positive R. eutropha cells increased upon exposure to bleach, we could not detect a significant difference between WT and ∆ppk-all cells (Fig. 4). This was in agreement with a comparable number of viable cells between the WT and the ∆ppk-all mutant in the agar-spot assay (not shown). For E. coli, however, we determined a significant difference between the WT and the ∆ppk-strain in sensitivity toward exposure to bleach: HOCl-treated cultures of E. coli ∆ppkdisplayed a large increase in PI-positive ∆ppk cells to ≈34%, whereas the WT showed only an increase from 5 to 9% PI-positive cells. The difference in survival between the WT and the ∆ppk mutant was confirmed by the agar-spot-assay (not shown).

To test whether exposure to bleach stimulates the formation of polyP, we used the cells of the experiment described above to follow the formation of polyP granules fluorescence microscopically and to quantify the amount of extractable polyP in cell samples. As shown in Figure 2b, most E. coli WT and R. eutropha WT cells formed one (rarely two) DAPI-stainable polyP granules during incubation in the MOPS-phosphate-sugar medium. Interestingly, the number of DAPI-polyP granules did not further increase upon exposure of E. coli or R. eutropha WT cells to bleach. As expected, the R. eutropha ∆ppk-all and ∆ppk E. coli mutant strains did not form DAPI-stainable polyP granules under any condition. When the amount of extractable polyP was determined, R. eutropha WT revealed a polyP content of 45 µmol P_i/g cdw that remained almost constant upon exposure to bleach (Fig. 5a). When E. coli WT was tested, a high polyP content of 41 µmol P_i/g cdw was determined in MOPS-phosphate-glucose medium that increased to 67 µmol P_i/g cdw after exposure to bleach (Fig. 5b). The formation of microscopically detectable polyP granules and the determination of a high value of 41 µmol P_i/g cdw before treatment of E. coli WT with bleach was unexpected as no polyP was detected in cells after growth in LB medium (Fig. 2a). We assume that incubation of E. coli cells in a phosphate- and carbon source-containing MOPS-medium that lacks other nutrients enables the cells to produce more ATP by respiration of glucose than can be consumed by metabolism. The surplus of ATP is transiently accumulated in the form of polyP granules. Presumably, the nutrient downshift from complex medium to a MOPS-phosphate-glucose solution provoked stress to the cells that lead already to Dsk-dependent [Gray 2019] and ppGpp-associated polyP formation (and/or inhibition of polyP degradation by exo-polyphosphatase) so that the addition of bleach at a later time-point only marginally further increased the formation polyP. These findings suggest that polyP has the function of a barrage for ATP during periods of a transient misbalance of ATP-producing and ATP-consuming reactions.

PolyP has an impact on motility of several bacterial species such as P. aeruginosa, Pseudomonas syringae, Bacillus cereus or Myxococcus xanthus [Rashid et al., 2000; Shi et al., 2004; Zhang et al., 2005; Fraley et al., 2007; Hossain et al., 2008]. To analyze if this is also true for R. eutropha, we determined motility of R. eutropha WT and the ∆ppk-all mutant in swimming and swarming assays. However, we could not find any difference between the WT and the ∆ppk-all mutant (suppl. Fig. s3). We conclude that polyP has no impact on motility in R. eutropha.

Surprisingly, the presence of polyP in R. eutropha WT had no detectable advantage for survival in our experiments, and we could not detect a phenotype of the ∆ppk-all mutant under any of the tested conditions shown above (heat shock/high temperature, reactive oxygen species, motility). Apparently, polyP has no obvious function in coping with these stressors. The findings suggest that polyP must have another function in R. eutropha. The most obvious one is the function as a reservoir for phosphorus. R. eutropha was originally isolated from the spring of the Wende river near Göttingen/Germany [Wilde 1962]. The concentrations of nutrients such as nitrogen or phosphorus sources are usually low in aquatic ecosystems and are even lower in the springs of most rivers. Therefore, it makes sense for R. eutropha to store an excess of phosphorus intracellularly in the form of polyP granules, and this might be one explanation why polyP is regularly found in stationary cells of R. eutropha cultures that are not limited by phosphorus. To determine whether the presence of polyP has an advantage for R. eutropha under conditions of limited nutrient supply, we tested growth, polyP content, cell numbers and cell viability in a mineral salts medium (MSM) with restricted supply of phosphorus. To this end, we cultivated R. eutropha WT and the ∆ppk-all mutant in a modified MSM medium with 0.8% fructose in which phosphate, which is usually present in high concentration as a buffer, was replaced by tris(hydroxymethyl)aminomethane (Tris). Phosphate was added at 0.02% (1.3 mM) (Fig. 6) or was completely absent (Fig. 7). At selected time points, samples were taken and analyzed for optical density (OD), polyP and PHB content. R. eutropha WT and the ∆ppk-all mutant grew comparably well on Tris-MSM in the presence of fructose and 0.02% phosphate and reached OD₆₀₀ values of ≈5 after 2 days (Fig. 6a). Only in the exponential growth phase between 12 and 36 h, the OD₆₀₀ values of the ∆ppk-all mutant were slightly lower than those of the WT but caught up with WT levels after ≥48 h. PolyP was detectable only in the WT. The polyP levels were relatively high at the beginning of growth (80 µmol P_i/g cdw) but then decreased to levels below 5 µmol P_i/g cdw) (Fig. 6b). No polyP was detected in the ∆ppk-all mutant and confirmed the inability of this mutant to produce detectable amounts of polyP. Both strains accumulated large amounts of PHB up to 70% PHB of cellular dry weight (cdw) (Fig. 6c). However, PHB contents of the ∆ppk-all mutant during growth were slightly lower than in the WT, and this could explain the lower OD values of the mutant versus WT. The determination of high levels of PHB indicated that the cells were not limited by the carbon source (0.8% fructose) but were limited by phosphorus (0.02%), and this explains why the amount of polyP decreased during growth of the WT. The high values for polyP at the beginning of the experiment are due to polyP formation from the pre-culture.

If polyP has the function of a P-reservoir, R. eutropha WT should have an advantage over a polyP-deficient mutant in media with growth-limiting amounts of phosphorus. To test this assumption, R. eutropha was first grown in NB medium; under these conditions, the cells form polyP at high levels. These polyP-rich cells were then transferred to a fresh Tris-MSM with fructose but without any P-source. As shown in Figure 7a, R. eutropha WT showed reduced but still substantial growth compared to the culture with phosphate in the first 24 h. However, growth slowed down continuously after 12 h reaching a final optical density of OD₆₀₀ of ≈1 after 144 h. In contrast, the ∆ppk-all mutant grew slower than the WT in the absence of phosphate and reached an OD₆₀₀ value of only ≈0.6. The optical density of bacterial cultures can be also influenced by the formation of PHB. Since the high carbon concentration (fructose) in the absence of a phosphate source should support strong PHB accumulation, we also determined the PHB content of the cultures. As expected, high amounts of PHB up to 60% PHB of cdw were determined. Interestingly, the PHB contents of WT cells were about 10% higher (reaching a maximum of 60% after 72 h) than those of the ∆ppk-all mutant which reached only 48.5% at 72 h. Therefore, a substantial amount of the higher OD₆₀₀ values of the WT compared to the ∆ppk-all mutant presumably is caused by the higher PHB content of the WT. To investigate if the presence of polyP really enables better growth, we determined the total number of cells of the WT and the ∆ppk-all culture at 72 h. As evident from Figure 7d, the cell numbers of the WT culture were substantially higher (1.33 × 10⁸ ± 2.88 × 10⁶) compared to the ∆ppk-all culture (1.02 × 10⁸ ± 3.75 × 10⁵ [77% of WT level]), while 88% of the WT cells (n = 540 cells) and 89% of the ∆ppk-all mutant cells (n = 359 cells) were still viable as revealed by PI staining (not shown). The mean cell size (measured as cell length) was also similar for the WT and the ∆ppk-all mutant (1.7–1.8 µm) but was lower compared to the beginning of the experiment (≈2.4 µm, 1.7–3.4 µm). These data lead to the conclusion that both the higher number of cells in the WT culture as well as the stronger accumulation of PHB contribute to the higher OD₆₀₀ values of the WT compared to the ∆ppk-all mutant and also confirmed that the presence of polyP leads to a substantial growth advantage of the WT over the polyP-deficient mutant.

To verify whether the polyP material was consumed by the WT during growth, we determined the levels of polyP at selected time points fluorescence microscopically and by polyP quantification after extraction from the cells (Fig. 7b). R. eutropha WT cells were rod-shaped, and most of the cells harbored one prominent polyP granule at the beginning of the growth experiment in the phosphate-free medium. A similar result was observed after 6 h of growth. After 12 h and at later time points the number of cells with DAPI-stainable polyP granules decreased and the cells became shorter and shorter reaching an almost coccoid morphology at the end of the experiment (suppl. Fig. s4a). In accordance with fluorescence microscopical analysis, R. eutropha WT revealed polyP levels of ≈60 µmol P_i/g cdw at the beginning of the experiment, and these values constantly decreased to sub-detectable levels within 36 h. In contrast, polyP was not detected in the ∆ppk-all mutant at any time point either in the form of DAPI-stainable granules or in the form of extractable polyP (suppl. Fig. s4b). When 0.02% phosphate was added to the cultures after 144 h, growth of the WT and of the ∆ppk-all mutant immediately resumed and reached OD₆₀₀ values of ≥7 (Fig. 7a) thereby confirming that the absence of phosphate was the growth-limiting factor in the experiment. Large amounts of PHB were accumulated by both strains. Remarkably, the amount of PHB was substantially higher in the WT compared to the ∆ppk-all mutant at all time points (Fig. 7c).

In summary, our data showed that the formed polyP in the WT was re-used as a phosphate source and allowed poor growth and for some (unknown) reason supported PHB accumulation to a 10–15% higher extent than in the ∆ppk-all mutant. However, the growth advantage of polyP under phosphate starvation in the WT seems to be only marginal. If the main function of polyP would be that of a P-reservoir, why do the cells form only one tiny polyP granule (in the stationary phase of NB-grown cultures) and not several large (voluminous) polyP granules as this is, for example, the case for PHB granules in PHB-accumulating cells? Formation of large polyP granules in R. eutropha is possible and has been demonstrated for R. eutropha previously by constitutive overexpression of ppk2c [Hildenbrand et al., 2020]. However, a simultaneous accumulation of high amounts of PHB AND high amounts of polyP leads to growth defects and inhibition of cell division resulting in the formation of unusually long cells [Tumlirsch et al., 2015; Hildenbrand et al., 2019].

R. eutropha and E. coli strains (Table 2) were grown in nutrient broth (NB, 0.8%) and in LB medium (10 g/L Trypton, 5 g/L yeast extract, 10 g/L NaCl) at 30 or 37°C, respectively. For most experiments with R. eutropha, the cells were grown overnight in NB medium (20 mL in a 100 mL Erlenmeyer flask) under continuous shaking at 30°C. 0.1 volume of this first seed culture was used to inoculate a second seed culture on the same medium (20–100 mL in an appropriately dimensioned flask) and was grown for 24 h at 30°C. Cells of this culture were used either directly or were harvested by centrifugation, washed with and resuspended in 0.9% NaCl (heat shock experiments) or in phosphate-free tris(hydroxymethyl)aminomethane (Tris)-MSM (experiments ± phosphate), respectively (see below). Growth experiments with R. eutropha strains at elevated temperatures were conducted in microtiter plates with 250 µL wells containing 200 µL NB medium. Wells were inoculated with cells from an NB overnight seed culture to an optical density (OD₆₀₀) of 0.05. Growth was monitored by OD₆₀₀ determination at temperatures as indicated with orbital shaking in a microplate reader (BioTek, Bad Friedrichshall, Germany). Alternatively, R. eutropha was cultured in MSM with carbon source as indicated. A Tris-buffered MSM of the following composition (values per liter) with or without the presence of 0.02% KH₂PO₄ was used: 6.06 g Tris, 4.68 g NaCl, 1.49 g KCl, 1.07 g NH₄Cl, 0.2 g MgSO₄ × 7H₂O, 0.01 g CaCl₂ × 2H₂O, 0.005 g ferri-ammonium-citrate (C₆H₈O₇ × nFe × nNH₃), 1:10,000 (v/v) trace element solution SL6 (per 1L: 1 g ZnSO₄ × 7H₂O, 0.3 g, MnCl₂ × 4H₂O, 3 g H₃BO₃, 2 g CoCl₂ × 6H₂O, 0.1 g CuCl₂ × 2H₂O, 0.2 g NiCl₂ × 6H₂O, 0.3 g Na₂MoO₄ × 2H₂O), pH 7.4. Fructose was added separately after sterilization to 0.8%.

NB-grown R. eutropha cells were incubated in a morpholino-propane sulfonic acid (MOPS) MSM with 4.18 g/L MOPS, 0.23 g/L K₂HPO₄ (pH 7.1) and 0.2% fructose (0.05 volumes of NB culture as seed culture) for 6 h at 30°C. 10 mL portions were transferred to fresh flasks and challenged with 100 µL phosphate-buffered saline (PBS, 8 g/L NaCl, 0.2 g/L KCl, 1.42 g/L Na₂HPO₄, 0.24 g/L KH₂PO₄) (control) or with 2 mM NaOCl (in 100 µL PBS). After 15 min of treatment with bleach, the samples were centrifuged, the supernatant was removed and the cells were resuspended in 20 mM MOPS buffer containing 20 mM Na₂S₂O₃. Subsequently, the cells were diluted 1:50 with PBS. Cells were stained with 2 µg PI/mL H₂O) for 15 min in the dark and analyzed by flow cytometry using a Cytek NL-2000 (excitation 530 nm, emission 615 nm) or by fluorescence microscopy. E. coli cells were treated analogously with the exception that fructose was replaced by glucose.

For polyP quantification and viability tests, 200 mL MOPS MSM with 1.32 mM phosphate and 0.2% fructose or glucose were inoculated with an NB or LB overnight pre-culture of R. eutropha and E. coli to an OD₆₀₀ = 0.3, cultivated at 30°C or 37°C, respectively, for 3 h with continuous shaking. 50 mL portions were transferred to fresh flasks and treated with H₂O as control or with 1 mM NaOCl (500 µL of 100 mM NaOCl stock in H₂O). After incubation for 1 h under cultivating conditions, 1 mL portions were centrifuged, 800 µL supernatant was removed and replaced by 800 µL of 20 mM Na₂S₂O₃ in 20 mM MOPS to stop the reactions. For viability test, cells were serially diluted from 10⁰ to 10⁻⁴ in 0.9% NaCl (4°C pre-cooled for E. coli, room temperature (RT) for R. eutropha). 2 µL of each dilution was spotted on an LB or NB agar plate and incubated at 37°C (E. coli) or at 30°C (R. eutropha), respectively.

Cells (10 µL) were stained with DAPI and PI for fluorescence microscopy (for details, see below). The remaining 49 mL sample was centrifuged at 5,087 g, resuspended in 20 mL 20 mM Na₂S₂O₃ in 20 mM MOPS and washed with 0.9% saline. Cells were lyophilized for polyP extraction and quantification.

NB (R. eutropha) or LB (E. coli) cultures were transferred to fresh medium and grown for 24 h at 30 and 37°C, respectively. Cells were harvested by centrifugation in an Eppendorf centrifuge (1 min, 5,087 g at RT) and resuspended in 0.9% NaCl. Cell density was adjusted to an optical density (OD₆₀₀) of 0.2 or 20. 10 µL of cells were added to 990 µL of pre-heated (55°C) 0.9% NaCl in a 2 mL microfuge tube so that the resulting OD₆₀₀ was 0.002 or 0.2. Cells were incubated at 55°C for 0–20 min with constant shaking (450 rpm in a thermo shaker, Biometra, Germany). 100 µL samples were taken in time intervals and serially diluted from 10⁰ to 10⁻⁴ in 0.9% NaCl (4°C pre-cooled for E. coli, RT for R. eutropha). 2 µL of each dilution was spotted on an NB or LB agar plate and incubated at 30°C for 2 days (R. eutropha) or 37°C for 1 day (E. coli), respectively. For determination of the polyP content after a heat shock, the experiment was scaled up to 50 mL, and the initial cell density was raised to OD₆₀₀ of 20. The cell suspension was diluted to a final OD₆₀₀ = 0.2 in 50 mL pre-heated (55°C) 0.9% NaCl, NB or LB in a glass flask with shaking. After 10 min (R. eutropha) or 20 min (E. coli), 50 mL stressed culture was centrifuged at 5,087 g for 8 min at 4°C. The supernatant was removed, cells were washed with H₂O, and the cell pellet was lyophilized and stored until polyP extraction.

PolyP was extracted from lyophilized cells by neutral phenol-chloroform extraction and subsequent precipitation in ethanol as described in [Bru et al., 2016; Frank and Jendrossek 2020]. Subsequently, polyP was digested with exo-polyphosphatase and inorganic pyro-phosphatase, and released phosphate was quantified using an Sb-tartrate-ascorbate colorimetric assay as described in [Christ et al., 2019].

Petri dishes were filled with 25 mL of heat-sterilized swimming or swarming agar (10 g/L Tryptone, 5 g/L NaCl, 0.3% standard agarose (Roth) for swimming assays and 8 g/L nutrient broth, 0.5% Bacto agar (Becton Dickinson, Sparks, USA) for swarming assays), and were allowed to solidify at RT for ≈1 h (swimming) or 3 h (swarming). Suspension of R. eutropha strains had been grown in 3 subsequent NB cultures each for 24 h at 30°C. OD_600, was adjusted to OD₆₀₀ of 0.1 with NB, and 5 µL of this suspension was spotted in five technical replicates in the center of a swimming agar plate. 1 µL of cell suspension was spotted on swarming plates. The plates were incubated at 30°C for 1 day on a horizontal underground without further movement of the plates.

The determination of cell density (total cell numbers) was performed with Neubauer counting chambers (DHC-N01, Digital Bio, Suwon, Korea) according to the manufacturer’s instruction.

PHB determination of lyophilized cell samples was performed by gas chromatography after acidic methanolysis as described in detail previously [Juengert et al., 2018].

Microscopy was carried out with a Nikon Ti-E fluorescence microscope (model number MEA53100) using a DAPI-polyP specific filter set (excitation, 387/11 nm; emission, 447/60 nm) or a long-pass filter (excitation at 562/40 nm and emission at 594 nm) for PI-stained cells. Formation of polyP granules was followed fluorescence microscopically after staining the cells with DAPI, final concentration 5 µg/mL (1:20 (v/v) 0.1 mg/mL stock solution) for 10–15 min. Live/dead staining of cells was carried out with 100 µg PI/mL (1:10 (v/v) 0.1 mg/mL stock) for 10 min. Cells were immobilized on agarose pads, and PI-positive and negative cells were counted from microscopic images.

The help and input of L. Blank and J. Christ (both RWTW Aachen) in establishing the antimony-tartrate-ascorbate assay for phosphate quantification as well as support by Eva Slaninova from Brno University during her stay in Stuttgart are greatly acknowledged.

Ethic approval is not required for this study (no experiments with humans or animals).

The authors have no conflicts of interest to declare.

This work was supported by a grant of the Deutsche Forschungsgemeinschaft (DFG) to the University of Tübingen with D.J. as an (external) PI and H.R. as doctoral student (RTG1708, Bacterial survival strategies).

H.R. and L.K. performed the experiments, prepared the figures and wrote parts of the manuscript. S.O. hosted HR during a 2-months stay at Brno University and helped H.R. in some experiments at Brno. D.J. designed the study and wrote most parts of the manuscript. All authors evaluated the data and read the manuscript.