Identification of an Evolutionarily Conserved Family of Inorganic Polyphosphate Endopolyphosphatases

Abstract

Inorganic polyphosphate (poly-P) consists of just a chain of phosphate groups linked by high energy bonds. It is found in every organism and is implicated in a wide variety of cellular processes (e.g. phosphate storage, blood coagulation, and pathogenicity). Its metabolism has been studied mainly in bacteria while remaining largely uncharacterized in eukaryotes. It has recently been suggested that poly-P metabolism is connected to that of highly phosphorylated inositol species (inositol pyrophosphates). Inositol pyrophosphates are molecules in which phosphate groups outnumber carbon atoms. Like poly-P they contain high energy bonds and play important roles in cell signaling. Here, we show that budding yeast mutants unable to produce inositol pyrophosphates have undetectable levels of poly-P. Our results suggest a prominent metabolic parallel between these two highly phosphorylated molecules. More importantly, we demonstrate that DDP1, encoding diadenosine and diphosphoinositol phosphohydrolase, possesses a robust poly-P endopolyphosphohydrolase activity. In addition, we prove that this is an evolutionarily conserved feature because mammalian Nudix hydrolase family members, the three Ddp1 homologues in human cells (DIPP1, DIPP2, and DIPP3), are also capable of degrading poly-P.

Introduction

Phosphates can be found in three different forms within cells: free phosphate (or inorganic phosphate; P_i), phosphate conjugates of organic molecules, and phosphate polymers linked by high energy phosphoanhydride bonds (or inorganic polyphosphate (poly-P)²) (Fig. 1A). The latter is present in every organism, where it contributes to a number of biological processes (for reviews, see Refs. 1 and 2). Poly-P constitutes a phosphate repository and functions as a chelator of metal ions regulating cellular cation homeostasis. Furthermore, poly-P possesses signaling roles; for instance, in bacteria, it is responsible for pathogenicity (3). It has been proposed that the length of poly-P chains regulates fibrinolysis, or blood coagulation, in mammalian organisms (4). Poly-P metabolism has been studied primarily in bacteria. In prokaryotes, poly-P synthesis is catalyzed by poly-P kinases (PPKs), which employ ATP as a substrate, whereas degradation is mediated by several poly-P phosphatases (2). In eukaryotes, however, poly-P synthesis has been poorly characterized. A bacteria-like PPK1 has been identified in the amoeba Dictyostelium discoideum, although it is believed to have originated through bacterial gene incorporation (5). The same organism also shows an actin-like poly-P kinase (DdPPK2) (6). In the budding yeast Saccharomyces cerevisiae, the amounts of poly-P are atypically high compared with other eukaryotic organisms. In yeast, up to 10% of the dry weight may consist of poly-P molecules with chain lengths ranging from a few units to several hundred residues. In addition, most poly-P localize to the vacuole (1, 2). It has recently been found that Vtc4 (YJL012C; also known as Phm3), corresponding to subunit four of the vacuolar transporter chaperone complex, functions as a poly-P polymerase (7). Mammalian genomes do not seem to contain any genes homologous to those encoding PPK1, DdPPK2, or Vtc4; therefore, poly-P synthesis is still largely unknown and a matter of debate (8). Poly-P catabolism in budding yeast is controlled by two phosphatases: an exopolyphosphatase named Ppx1 (exopolyphosphatase; YHR201C), which removes P_i from the end of poly-P chains (Fig. 1B) (9), and an endopolyphosphatase, Ppn1 (endopolyphosphatase; YDR452W; also known as Phm5), which attacks internal phosphoanhydride bonds, hydrolyzing poly-P molecules into oligophosphates of smaller size (Fig. 1B) (10). Experimental evidence points to the existence of additional poly-P phosphatases because endopolyphosphatase activity has been detected in the cytosol of the double knock-out ppn1Δppx1Δ (11, 12).

FIGURE 1.

Schematic representation of poly-P and inositol pyrophosphate structures and inositol metabolic pathway. A, minimal structure of a poly-P molecule, with n from 1 to a few hundred. When n equals 1, the minimal poly-P is constituted by three phosphate residues; therefore, pyrophosphates (two phosphates) are not classified as poly-P. B, elongated representation of poly-P to show the bounds targeted by endo- and exopolyphosphatases. C, structure of inositol pyrophosphates. IP6Ks, such as Kcs1, phosphorylate position 5 of IP₆ to generate the isomer 5PP-IP₅ (21). PP-IP5Ks, such as Vip1, are able to phosphorylate position 1 or 3 to generate the isomer (1/3)PP-IP₅ of IP₇ (24). Note that the 1- and 3-positions (indicated in gray) are enantiomeric ring positions and therefore cannot be distinguished by NMR studies. IP6K, using IP₅ as substrate, is able to phosphorylate position 1/3, generating the isomer (1/3)PP-IP₄ (21). D, metabolic pathway of the synthesis and degradation of inositol polyphosphates in budding yeast. PIP₂, phosphatidylinositol 4,5-bisphosphate

Inositol pyrophosphates are a eukaryote-specific class of phosphorylated molecules, in which the number of phosphate groups outnumbers that of carbon atoms (for reviews, see Refs. 13–15). The best characterized inositol pyrophosphates are diphosphoinositol pentakisphosphate, or PP-IP₅ (IP₇), and bis-diphosphoinositol tetrakisphosphate, or (PP)₂-IP₄ (IP₈), possessing seven and eight phosphate groups attached to the six-carbon inositol ring, respectively. Inositol pyrophosphate metabolism is highly dynamic, meaning that cells need to invest a considerable amount of ATP to keep their levels constant (16). Inositol pyrophosphates have important roles in the regulation of telomere length, vesicular trafficking, and apoptosis; in addition, a growing body of physiopathology evidence links them to important human diseases, such as diabetes, obesity, and cancer (13–15). Inositol pyrophosphates contain one or more highly energetic diphospho moieties that participate in phosphotransfer reactions (17). This ability has been hypothesized for poly-P molecules too (18, 19). Two distinct classes of evolutionarily conserved kinases are responsible for inositol pyrophosphate synthesis. The first, IP₆-kinases (IP6Ks in mammals and Kcs1 (YDR017C) in yeast), is primarily involved in synthesizing the isomer 5PP-IP₅ of IP₇ from IP₆ (Fig. 1C) (20, 21). The second, PP-IP₅-kinases (Vip1; YLR410W in yeast), converts IP₇ to IP₈ in vivo (22, 23) while also metabolizing IP₆ to the IP₇ isomer, (1/3)PP-IP₅, in vitro (Fig. 1C) (24). Both kinases are capable of converting the reciprocal IP₇ isomers to IP₈ in vitro (25). In budding yeast, the metabolic pathway leading to inositol pyrophosphate synthesis (Fig. 1D) starts with the hydrolysis of phosphatidylinositol 4,5-bisphosphate to IP₃ by phospholipase C (Plc1, YPL268W). Subsequently, IP₃ is converted to IP₄ and IP₅ through the action of the inositol polyphosphate multikinase (IPMK in mammals and Arg82 (YDR173C) in yeast). IP₅ is converted to the fully phosphorylated ring of IP₆ thanks to IP₅-2K (Ipk1; YDR315C in yeast) (Fig. 1D). Kcs1 metabolizes IP₅ to the pyrophosphate containing molecule PP-IP₄ (Fig. 1C) (26, 27). Under standard conditions, IP₅ does not accumulate in budding yeast; however, when IPK1 is knocked out, IP₅ can no longer be converted to IP₆ and thus accumulates in the cells, becoming the substrate for Kcs1 (Fig. 1D) (27). Inositol pyrophosphates are hydrolyzed by diphosphoinositol polyphosphate phosphohydrolases (DIPPs) (28). These phosphatases are also able to degrade nucleotide analogues, such as diadenosine hexaphosphate. Consequently, the yeast homologue has been named Ddp1 (diadenosine and diphosphoinositol polyphosphate phosphohydrolase) (YOR163W) (29, 30). DIPPs belong to the Nudix (nucleoside diphosphate-linked moiety X) hydrolase family and possess a 23-amino acid catalytic domain denominated the MutT motif or Nudix box (31). In mammals, four putative DIPPs, have been characterized: DIPP1 (NUDT3), DIPP2 (NUDT4), and DIPP3a/b (NUD10 and NUD11) (28, 29, 32).

A genome-wide screen of a yeast knock-out library indicated a link between inositol pyrophosphate metabolism and the synthesis of poly-P (33). One limitation of this study was the use of ³¹P NMR spectra to reveal the cellular presence of poly-P, which did not discriminate the true length of poly-P chains (34). A second genome-wide screen of the same yeast knock-out collection, employing direct extraction and colorimetric technology to analyze cellular poly-P levels, failed to identify any link between poly-P and the inositol pyrophosphate biosynthetic pathway (35). In the present study, we aimed at assessing the validity of the above observations. We have biochemically characterized poly-P levels and molecular species in a battery of yeast mutants with defects in inositol pyrophosphate metabolism. Our results point to a striking metabolic correlation between these two highly phosphorylated molecules, hence refining previous observations (33). More importantly, we have found that Ddp1 possesses a robust poly-P endopolyphosphohydrolase activity. In addition, we show that this is an evolutionarily conserved feature because human DIPP proteins also have the ability to degrade poly-P. Given that, to our knowledge, no equivalent of the yeast endopolyphosphatase Ppn1 has previously been characterized in mammals, our findings describe the first mammalian poly-P endopolyphosphatase class of enzymes.

EXPERIMENTAL PROCEDURES

Reagents

Polyacrylamide mix, TEMED, and ammonium persulfate were purchased from National Diagnostics; yeast synthetic media (SC) from Formedium; phytic acid from Calbiochem; and radiolabeled [³H]inositol from American Radiolabeled Chemicals. All other reagents were acquired from Sigma-Aldrich. Synthesis and purification of IP₈ and IP₇ isomers, 5PP-IP₅ and (1/3)PP-IP₅, have been described elsewhere (36).

Strains

Saccharomyces cerevisiae strains used in this study are isogenic to BY4741 (MATa, his3Δ1; leu2Δ0; met15Δ0; ura3Δ0) and described in Table 1. The knock-out mutants plc1Δ, arg82Δ, ipk1Δ, kcs1Δ, and ipk1Δkcs1Δ have been described earlier (27); ppn1Δ, ddp1Δ, and vip1Δ were acquired from Euroscarf; kcs1Δ, harboring mammalian IP6K expression plasmids, has been described previously (23). Finally, ppx1Δ was generated by homologous recombination (37) using specific oligonucleotides. Correct genomic integration of the substitution cassette was confirmed by genomic PCR as described previously (37).

TABLE 1.

S. cerevisiae strains used in this study

Strain	Relevant genotype	Source/Reference
BY4741	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0	Ref. 27
arg82Δ	BY4741 arg82::kanMX4	Ref. 27
ddp1Δ	BY4741 ddp1::kanMX4	Euroscarf
ipk1Δ	BY4741 ipk1::kanMX4	Ref. 27
kcs1Δ	BY4741 kcs1::kanMX4	Ref. 27
kcs1Δipk1Δ	BY4741 ipk1::kanMX4 kcs1::kanMX4	Ref. 27
kcs1Δ + pIP6K1	BY4741 kcs1::kanMX4 + pIP6K1	Ref. 23
kcs1Δ + pIP6K1 (K/A)	BY4741 kcs1::kanMX4 + pIP6K1(K/A)	Ref. 23
plc1Δ	BY4741 plc1::kanMX4	Ref. 27
ppn1Δ	BY4741 ppn1::kanMX4	Euroscarf
ppx1Δ	BY4741 ppx1::kanMX4	This study
vip1Δ	BY4741 vip1::kanMX4	Euroscarf

Plasmids

S. cerevisiae DDP1 (YOR163W) was amplified by PCR from genomic DNA and subcloned into the PstI and EcoRI sites of the pTrcHisB expression vector (Invitrogen). The generation of a catalytically inactive Ddp1 mutant was obtained by mutating two highly conserved adjacent glutamate residues (E83A and E84Q) within the mutT region (28). The substitution was performed using site-directed mutagenesis PCR with the following oligonucleotides: 5′-CAA CGA GAA ACT TGG GCC CAA GCT GGT TGC ATA GGT A-3′ and 5′-TAC CTA TGC AAC CAG CTT GGG CCC AAG TT CTC GTT G-3′. Human DDP1 (NUDT3), DDP2 (NUDT4), and DDP3a (NUD10) were amplified by PCR from specific cDNA clones or from cDNA prepared from HeLa cells and subcloned into the PstI and EcoRI sites of the pTrcHisB expression vector. All plasmids were confirmed by direct sequencing.

Poly-P Extraction

Poly-P was extracted as described earlier (38) with a few modifications. In short, cells equivalent to 10 units of OD (optical density measured at 600 nm absorbance) of logarithmically growing yeast cells were centrifuged at 1000 × g for 4 min. The cell pellet was resuspended in 250 μl of Lets Buffer (0.1 m LiCl, 10 mm EDTA, 10 mm Tris, pH 8.0, 0.5% SDS) and mixed with 250 μl of acid phenol. Sufficient glass beads were added so that their level was right below that of the phenol fraction. Then cells were vortexed for 5 min at 4 °C, followed by centrifugation at 15000 × g for 5 min at 4 °C. The water phase was transferred to a new tube and subjected to chloroform extraction. The supernatant was precipitated by adding 2.5 volumes of ethanol followed by overnight incubation at −20 °C. Subsequently, samples were spun down for 10 min, the RNA/poly-P pellet was resuspended in 50 μl of buffer (0.1% SDS, 1 mm EDTA, 10 mm Tris-HCl, pH 7.4), and the nucleic acid concentration was determined by reading the absorbance at 260 nm.

Analysis of Inositol Polyphosphates after Modulation of Phosphate Concentrations in the Medium

Yeast cells were cultivated overnight in SC medium supplemented with [³H]inositol (5 μCi/ml). Exponentially growing cells were collected, washed once with water, and resuspended in 10 ml of SC phosphate-free medium supplemented with [³H]inositol to an A₆₀₀ of 0.5. After a 2-h incubation at 30 °C, potassium phosphate was added to a final concentration of 10 mm, and cells were incubated for 1 h more at 30 °C. Cells were harvested at the indicated times, washed with water, and stored at −20 °C until use. Inositol polyphosphates were extracted and analyzed as described earlier (39). Poly-P was extracted from parallel unlabeled yeast cultures.

Poly-P Quantification by Colorimetric Assay

The assay to measure poly-P levels is based on the ability of malachite green, molybdate, and free orthophosphate to form a complex that absorbs at 620–640 nm (40). Before performing this assay, 2–5 μg of RNA/poly-P were converted to orthophosphate by the concomitant action of recombinant Ppx1 and Ddp1. Alternatively, acidic hydrolysis of the phosphodiesteric bonds was performed by incubating RNA/poly-P in 1 m perchloric acid for 30 min at 90 °C. The acid treatment leads to the release of phosphate from RNA; however, this does not affect the overall result because the amounts of orthophosphate released from RNA are negligible compared with those generated by poly-P. Both procedures provided similar results. 10–20% of hydrolyzed poly-P was diluted in 500 μl of 0.3 m perchloric acid, and 500 μl of malachite green reagent (15 mg of malachite green, 100 mg sodium molybdate, 25 μl of Triton X-100, 2.82 ml of 38% hydrochloric acid in 50 ml of water) were added. Absorbance was measured at 650 nm after a 30-min incubation at room temperature. The concentration of hydrolyzed orthophosphates was compared with a P_i standard calibration curve.

Recombinant Protein Purification

Protein purification was performed as described previously (41). Protein recovery and purity was confirmed by Western blot using NuPage gels (Invitrogen).

Polyphosphatase Reactions

Enzyme activity was assayed in a reaction mix containing 25 mm Hepes, pH 6.8, 50 mm NaCl, 10 mm MgSO₄, 1 mm DTT, 5–20 ng of recombinant purified Ddp1 or Ppx1, and 10–100 nmol of poly-P or 2–10 nmol of IP₆/IP₇/IP₈ substrates. Reactions were performed at 37 °C for the indicated times, stopped by the addition of 2 μl of 100 mm EDTA, and placed on ice. Samples were run on a polyacrylamide gel or stored at −20 °C until further use. The mammalian DIPP1, -2, and -3 recombinant proteins were assayed in the same way with reaction buffers set at pH 7.4, 8.0, 8.5, or 9.0 using Tris buffer.

Fractionation of Inositol Polyphosphates and Poly-P by PAGE

Inositol polyphosphates and poly-P were resolved using a 24 × 16 × 0.1-cm gel with 20 or 35% polyacrylamide in TBE. Before loading the samples, gels were prerun for 30 min at 200 V. Samples were run at 2 mA (20%) or 5 mA (35%) overnight at 4 °C, until the Orange G dye front reached 10 cm from the gel's bottom. Gels were stained with DAPI or toluidine blue as described previously (25).

RESULTS

Poly-P and Inositol Pyrophosphates Are Metabolically Associated

The metabolic relation between inositol pyrophosphates and poly-P was explored by two large scale analyses that presented contradictory results (33, 35). To resolve this incongruence, we performed a systematic quantitative and qualitative analysis of poly-P using budding yeast mutants deficient in inositol pyrophosphate metabolism. Thus, whereas extracts from wild type and the IP₅-2K mutant, ipk1Δ, display poly-P in an electrophoretic separation, as shown by an intense metachromatic (dark) smear, mutants unable to synthesize inositol pyrophosphate such phospholipase C (plc1Δ), inositol polyphosphate multikinase (arg82Δ), inositol hexakisphosphate kinase (kcs1Δ), and the double mutant kcs1Δipk1Δ did not exhibit detectable levels of poly-P (Figs. 1D and 2A and Table 2). It should be noted that ipk1Δ displayed a higher amount of poly-P than the wild type (Table 2), which correlated with a substantial increase in inositol pyrophosphates. In fact, in ipk1Δ, the levels of IP₅ generated by PP-IP₄ represented more than 30% of its precursor (Table 2) (38). Interestingly, poly-P levels in kcs1Δ were restored by the introduction of the mouse IP6K1 but not by the catalytically inactive mutant (IP6K1K/A). These data further emphasize the requirement for inositol pyrophosphate metabolism in order to maintain poly-P homeostasis. In addition, we observed a significant decrease of poly-P levels in ddp1Δ. No significant qualitative or quantitative difference in poly-P was observed in vip1Δ (Fig. 2A and Table 2).

FIGURE 2.

Inositol pyrophosphates and poly-P are metabolically interrelated. A, poly-P, represented by a dark smear, was extracted from logarithmic growing yeast cultures, and 40 μg of RNA were resolved on a native 20% polyacrylamide gel and visualized with toluidine blue. Knock-out mutants were as follows: plc1Δ (phospholipase C), arg82Δ (IPMK), ipk1Δ (IP₅-2K), kcs1Δ (IP6K), and vip1Δ (PP-IP₅K). The levels of poly-P in kcs1Δ were restored by the introduction of the mouse IP6K1 enzyme, but not by the catalytically inactive mutant (mIP6K1K/A). OrG, migration of the Orange G dye. B, analysis of inositol pyrophosphates (IP₇) and poly-P in wild type during a phosphate overplus condition. C, analysis of inositol pyrophosphates (PP-IP₄) and poly-P in ipk1Δ during a phosphate overplus condition.

TABLE 2.

Inositol pyrophosphates and poly-P levels

Inositol pyrophosphate values represent average ± S.D. of three or more independent assays.

	Inositol pyrophosphate, percentage of precursora	Poly-P
	%	μmol/mg RNA
Wild type	4.1 ± 1.2	7.48 ± 0.88
arg82Δ	Undetectable	Undetectable
ddp1Δ	10.3 ± 1.8	6.11 ± 0.36 p < 0.05
ipk1Δ	31.2 ± 5.3	13.35 ± 0.84 p < 0.01
kcs1Δ	Undetectable	<0.03
kcs1Δipk1Δ	Undetectable	Undetectable
plc1Δ	Undetectable	Undetectable
ppn1Δ	3.9 ± 1.1	8.00 ± 0.67
ppx1Δ	3.8 ± 2.0	7.45 ± 0.74
vip1Δ	15.2 ± 3.1	8.11 ± 0.93

^a For wild type kcs1Δ, vip1Δ, kcs1Δipk1Δ, ddp1Δ, ppn1Δ, and ppx1Δ, the values represent the ratio of IP₇ over IP₆. For ipk1Δ and kcs1Δipk1Δ, the values represent the ratio of PP-IP₄ over IP₅. Poly-P values are expressed as average ± S.D. of two of more independent assays, each done in triplicate. A paired Student's t test was applied to calculate the statistical significance of the value of the mutant against wild type.

The turnover of poly-P molecules was measured, employing an assay in which the concentration of phosphates in the medium was manipulated. Yeast cultures were first shifted to phosphate-free medium for 2 h and then followed by phosphate addition. This led to a rapid increase in poly-P, which accumulated to concentrations higher than would have normally been obtained under normal growth conditions (this assay is sometimes referred to as overplus). In order to analyze the levels of poly-P and inositol pyrophosphates, we supplied [³H]inositol to wild type and ipk1Δ cultures and performed a phosphate overplus assay (Fig. 2, B and C). In accordance with a previous report (42), when cells were shifted to phosphate-free medium, poly-P was rapidly hydrolyzed and resynthesized to a higher concentration when normal phosphate levels were restored (Fig. 2, B and C). In this assay, inositol pyrophosphate showed a pattern similar to that of poly-P. Both in wild type and ipk1Δ, inositol pyrophosphate amounts exhibit a dramatic decrease, without dropping below detection levels in phosphate-free medium. Inositol pyrophosphate levels recovered rapidly upon phosphate restoration to the medium, reaching concentrations higher than normal. However, the variability in our analysis does not allow determination of statistical significance. Inositol pyrophosphates are metabolically regulated by phosphate concentration in the medium. On the contrary, the concentration of their precursor IP₆, relative to the overall radiolabeled inositol pool, is not affected by the removal of phosphate (data not show). This finding indicates that, at least in yeast, IP₆ cannot be considered a phosphate storage molecule. These experiments demonstrate that a similar metabolic fate is shared between inositol pyrophosphate and poly-P.

Ddp1 Possesses a Poly-P Endopolyphosphatase Activity

The lower levels of poly-P observed in ddp1Δ suggested its involvement in poly-P metabolism. We thus investigated whether Ddp1 could metabolize poly-P. We cloned, expressed, and purified both Ddp1 and the catalytically inactive form Ddp1EE/AQ, using a polyhistidine tag (His). Incubation of His-DDP1 with commercially available synthetic polyphosphates, poly-P25 and poly-P65, revealed a robust poly-P phosphatase activity (Fig. 3, A and B). On the contrary, the catalytically inactive His-DDP1EE/AQ protein was unable to hydrolyze them (Fig. 3, A and B). Ddp1 endopolyphosphatase activity was demonstrated by the increase in the intensity of staining of small poly-P oligomers obtained after incubation with His-Ddp1 (Fig. 3B). Given the caveats regarding the quality of commercial poly-P standards, we tested His-Ddp1 capacity to digest poly-P extracted from wild type, ipk1Δ, and vip1Δ. Again, we found that His-Ddp1 exhibited a strong polyphosphatase activity, whereas His-Ddp1EE/AQ failed to degrade them (Fig. 3C).

FIGURE 3.

Ddp1 possesses poly-P endophosphatase activity. 10 ng of recombinant Ddp1 and the catalytically inactive form (Ddp1EE/AQ) were incubated at 37 °C for the indicated times with 100 nmol of poly-P25 (A) and poly-P65 (B). C, 10 ng of recombinant His-Ddp1 and His-Ddp1 mutant were incubated for 30 min at 37 °C with poly-P extracted from wild type (WT), ipk1Δ (IP₅-2K), and vip1Δ (PP-IP₅K). The reactions were stopped by the addition of EDTA, resolved on a 30% polyacrylamide gel, and visualized with DAPI staining.

Ddp1 Exhibits a Higher Substrate Affinity for Poly-P than for Inositol Pyrophosphates

The human homologue of Ddp1 (DPP1) has been shown to use IP₈ more efficiently than IP₇, suggesting a substrate preference for a specific pyrophosphate group position (43). We tested the ability of Ddp1 to dephosphorylate the two isomeric forms of IP₇ found in yeast: 5PP-IP₅ and (1/3)PP-IP₅. After 30 min of incubation, (1/3)PP-IP₅ was converted to IP₆, as judged by the observed band shift (Fig. 4A). Instead, 5PP-IP₅ remained intact, indicating that it is a poor substrate for Ddp1. We observed partial degradation of 5PP-IP₅ only upon incubation for extended periods of time or in the presence of an excess of Ddp1 (data not shown).

FIGURE 4.

Ddp1 prefers poly-P over inositol pyrophosphate as a substrate. A, recombinant His-DDP1 (10 ng) was incubated for 30 min at 37 °C with 100 nmol of biochemically synthesized IP₇ from IP6K1 (5PP-IP₅) or Vip1 ((1/3)PP-IP₅). B, His-Ddp1 (10 ng) was incubated for the indicated times at 37 °C with a mixture of 100 nmol of (1/3)PP-IP5 and 200 nmol of poly-P25. C, recombinant His-Ddp1 (10 ng) was incubated for 30 min at 37 °C in presence of different concentrations of NaF. D, His-Ddp1 (10 ng) was incubated for 60 min at 37 °C with 200 nmol of poly-P25 in the absence or presence of a 2 mm concentration of the indicated cations, supplied in chloride form. E, His-DDP1 (10 ng) was incubated for 30 min at 37 °C with 200 nmol of poly-P25 in the presence of 5 mm heparin.

We then compared Ddp1 substrate preference by incubating it with similar amounts of poly-P and (1/3)PP-IP₅ in the same reaction. We observed that poly-P was degraded very rapidly, whereas conversion of (1/3)PP-IP₅ to IP₆ took place later and at a slower rate (Fig. 4B). The strong preference for poly-P exhibited by Ddp1 can be explained by Ddp1 having primarily endopolyphosphatase activity and by the numerous phosphoanhydride bonds of poly-P. Human DPP1 has been reported to be very sensitive to inhibition by NaF (28). We tested whether that was the case for Ddp1 with poly-P as a substrate. We observed that Ddp1 polyphosphatase activity was highly sensitive to the presence of even small amounts of NaF, with total inhibition occurring at concentrations as low as 1 μm (Fig. 4C).

To verify whether Ddp1 could correspond to the partially purified endopolyphosphatase activity previously reported (11), we tested the effects of different cations and heparin. Ddp1 was active in the presence of divalent cations, such as magnesium or cobalt, but became inactive when incubated with heparin (Fig. 4E) or manganese or calcium (Fig. 4D). Our results indicate that Ddp1 shares a number of traits with the endopolyphosphatase activity identified by Lichko et al. (11), such as its molecular weight, the activity in the presence of magnesium and cobalt, the inability of calcium to activate the enzyme, and the sensitivity to heparin. Interestingly, it differed in two aspects: sensitivity to fluoride and inactivation by manganese.

Analysis of Poly-P in Polyphosphatase Mutants ppx1Δ, ppn1Δ, and ddp1Δ

In order to ascertain whether Ddp1 capacity to degrade inositol pyrophosphates is unique or whether it also extends to other polyphosphatases, we looked into the catalytic properties of Ppx1 (exopolyphosphatase) (9) and Ppn1 (endopolyphosphatase) (10). Purified recombinant Ppx1 was very active in degrading poly-P but failed to do the same with both isomers of IP₇ (Fig. 5). The inability to generate an active recombinant Ppn1 (44) prevented us from testing its phosphatase activity against inositol pyrophosphates. Therefore, we decided to investigate inositol pyrophosphate levels in ppx1Δ and ppn1Δ upon labeling with [³H]inositol. This analysis revealed that ppx1Δ and ppn1Δ possessed inositol pyrophosphate (IP₇ and IP₈) levels similar to those present in the wild type (Table 2). Similarly, the analysis of poly-P levels did not show any significant variation with respect to wild type (Table 2), confirming previous reports (45). Although the total amounts of poly-P in ppn1Δ were similar to wild type, the composition of poly-P species was different, showing a predominance of high molecular weight polymers (45). Thus, we investigated the size of poly-P molecules in ppx1Δ, ppn1Δ, and ddp1Δ by resolving extracts with gel electrophoresis. Unlike wild type, ppx1Δ and ddp1Δ, ppn1Δ displayed a clear reduction in low molecular weight poly-P (Fig. 6). In conclusion, this comparison reveals that only Ppn1 qualitatively affects a specific pool of poly-P, most likely corresponding to the vacuolar lumen fraction where Ppn1 itself localizes (46). Moreover, the quantitative differences in poly-P and inositol pyrophosphates levels are only observed in ddp1Δ.

FIGURE 5.

Ppx1 does not hydrolyze inositol pyrophosphates. Recombinant His-Ppx1 (10 ng) was incubated for 30 min at 37 °C with 10 nmol of biochemically synthesized IP₇ (5PP-IP₅) or IP₈ and 100 nmol of poly-P25 that, to the contrary to the inositol pyrophosphate, is fully hydrolyzed. OrG, migration of the Orange G dye.

FIGURE 6.

Comparison of poly-P profiles from ppn1Δ, ppx1Δ, and ddp1Δ mutants. Poly-P was extracted from logarithmic growing cells, and fractions containing 40 μg of total RNA were resolved and visualized with toluidine blue. A, 20% polyacrylamide gel allows the observation of large polymeric species. B, 35% polyacrylamide gel to visualize small poly-P fragments. Knock-out mutants were as follows: ppx1Δ (exopolyphosphatase), ppn1Δ (endopolyphosphatase), and ddp1Δ (diadenosine and diphosphoinositol polyphosphate phosphohydrolase). OrG, migration of the Orange G dye.

Mammalian Nudix Hydrolases DIPP1, DDP2, and DDP3 Function as Poly-P Endophosphatases

The four human homologues to Ddp1 (DIPP1, DIPP2, and DIPP3a/b (encoded by two genes on chromosome 10 of identical amino acid sequence)) are better characterized than their yeast counterpart (32). They harbor a dual substrate activity that allows for the degradation of both inositol pyrophosphate and nucleotide analogues, such as diadenosine hexaphosphate (28, 29, 32). To test the ability of these human enzymes to process poly-P as a substrate, we expressed and purified them from Escherichia coli (Fig. 7A). The Ddp1 enzyme assays presented previously (Figs. 3 and 4) were performed at pH 6.8. Ddp1 is more active in conditions near neutral pH with an optimal activity at pH 6.8–8.0, whereas it is partially inhibited at pH 9.0 (Fig. 7B). At neutral pH, human DIPPs did not possess phosphatase activity. However, DIPPs have been reported to hydrolyze adenine dimers in alkaline conditions, with an optimal pH 9.0 (28), a characteristic typical of this class of enzymes (47). Consequently, we assessed poly-P hydrolysis by recombinant DIPPs at a higher pH. We observed that their activity is restricted to a narrow basic pH range of 8.5–9.0 and remained inhibited at neutral pH 6.8–7.4 (Fig. 7B). These results also showed that Ddp1 is far more active than its human counterparts, leading to complete digestion of the available poly-P. A similar amount (10 ng) of human recombinant DIPP1 and DDP3 was only partially able to digest poly-P. DIPP2 seems to have an even weaker activity, although this might simply reflect the poor quality of the recombinant enzyme. The latter was not only poorly expressed in E. coli but also contained degradation products (Fig. 7A). We further confirmed the polyphosphatase activity of DIPP2 by increasing the incubation times to 2–4 h (Fig. 7C). Physiologically, the general weaker activity of the human enzymes could be explained by the lower amount of poly-P observed in mammalian cells compared with budding yeast (2). Therefore, the enzymes might have adapted to lower poly-P levels. Finally, the increased intensity of the smaller poly-P species migrating below the IP₆ marker (Fig. 7B) demonstrated that, like Ddp1, the human DIPPs are poly-P endopolyphosphatases.

FIGURE 7.

DIPP1/2/3 enzymes possess poly-P endophosphatase activity. A, recombinant His-tagged Ddp1, DIPP1, DIPP2, and DIPP3 (100 ng) were resolved on a 4–12% NuPage gel and visualized by Coomassie Blue staining. B, recombinant enzymes (10 ng) were incubated for 60 min at 37 °C with 200 nmol of poly-P25 in a reaction buffered at the indicated pH. C, DIPP1, DIPP2, and DIPP3 (10 ng) were incubated for the indicated time at 37 °C with 200 nmol of poly-P25 in a reaction buffered at pH 9.0. OrG, migration of the Orange G dye.

DISCUSSION

Our systematic biochemical examination of poly-P levels using yeast mutants unable to synthesize inositol pyrophosphates reveals a striking correlation between the lack of inositol pyrophosphates and the decrease in poly-P polymers (Fig. 2). These results confirm the observations by Auesukaree et al. (33), reporting that the PP-IP₄ synthesized from IP₅ is responsible for poly-P cellular presence. However, our analysis of the ipk1Δkcs1Δ double mutant revealed undetectable levels of poly-P. Therefore, it is the presence of any form of inositol pyrophosphates (IP₇/IP₈ in wild type or PP-IP₄ in ipk1Δ) that determines the presence of cellular poly-P.

We also observed that modulating phosphate concentrations in the medium has a dramatic effect on inositol pyrophosphate metabolism. Exposing budding yeast to phosphate-free medium resulted in a decrease of intracellular inositol pyrophosphate levels, without affecting IP₆ concentration. Hence, in the case of budding yeast, IP₆ cannot be considered a phosphate-storage molecule. An IP₇ increase in phosphate-deprived medium, has been reported by the O'Shea group (48). However, those findings are in disagreement with the observation that reducing the concentration of phosphate in the medium lowered the levels of ATP required for IP₇ synthesis (49). We tried to replicate the results of O'Shea and co-workers (48) using the same experimental conditions and yeast background but were unsuccessful and cannot offer a logical explanation for this discrepancy.

The more important finding comprised in this study is the identification of the Nudix hydrolase family (Ddp1 in yeast and DIPP1, DIPP2, and DIPP3 in humans) as poly-P endopolyphosphatases. Earlier reports have indicated the presence in S. cerevisiae of an additional endopolyphosphatase besides Ppn1 (12). More recently, a partial purification yielded a protein of ∼20 kDa (11) with endopolyphosphatase activity, which is very similar to the molecular mass of Ddp1 (21.5 kDa). However, unlike the partially purified endopolyphosphatase (11), Ddp1 is fluoride-sensitive and becomes inactive in the presence of manganese. This leaves open the possibility of the existence of a third enzyme.

An analogous activity to that of Ddp1 has been reported in extracts of rat liver and in embryos of Artermia salina. These extracts possess an enzyme of a molecular weight similar to Ddp1 that can degrade nucleotide dimers, such as diguanosine tetraphosphate, and that has been found to be inhibited by calcium (50).

Unexpectedly, lack of Ddp1 (ddp1Δ) resulted in a decrease in poly-P levels (Table 1). Budding yeast has unusually high levels of poly-P, which is stored in the vacuole (1, 2). The endopolyphosphatase Ppn1 localizes to the vacuole (46), where it influences the metabolism of the poly-P pool in this compartment, most likely buffering the demand for phosphate in the cell. On the contrary, Ddp1 is a cytosolic enzyme (51) and thus acts on the smaller pool of cytosolic poly-P molecules, probably with signaling properties, as in prokaryotes. We speculate that in ddp1Δ, an increase in the cytosolic pool of polyphosphates may have a negative feedback effect on the larger pool of vacuolar poly-P, leading to its decrease. Unfortunately, our poly-P analysis only accounts for global changes in poly-P levels, and thus, it represents mainly a read-out of the vacuolar poly-P pool. The wide use of phenol methods for the extraction of poly-P (38, 45), also employed in this study, did not allow for the analysis of specific cellular fractions. Our attempts to separate cellular components before poly-P extraction have been unsuccessful, mainly due to the rapid degradation of poly-P and to the contamination of other subcellular fractions by the vacuolar portion.

The large amounts of IP₇ that were unexpectedly found in kcs1Δddp1Δ (52) allowed for the identification of Vip1 (PP-IP₅K) as a novel inositol pyrophosphate-synthesizing class of enzyme (22). This surprising feature of kcs1Δddp1Δ can be explained by the fact that DIPPs hydrolyze more efficiently the diphosphate added by PP-IP₅Ks (Fig. 3) (15). This masks the rate of synthesis of one of the two IP₇ isomers involved in IP₈ production. Further work is needed to fully appreciate the metabolic flux and the exact physiological role played by the Ddp1 phosphohydrolase activity in controlling the metabolism of its many substrates. The discovery that Ddp1 is a poly-P phosphatase opens new avenues of research, tackling questions such as how inositol pyrophosphate regulates cellular poly-P levels. In S. cerevisiae, a poly-P synthesizing enzyme, VTC-complex subunit four, Vtc4, was identified in the vacuolar membrane, and the polymerase reaction was shown to be dramatically accelerated by pyrophosphate (7). Thus, we can hypothesize that the pyrophosphate moiety of IP₇ can substitute for pyrophosphate in stimulating poly-P synthesis. However, this explanation cannot be applied to mammalian cells, where the VTC-complex is absent. One of the mammalian IP₆-kinase enzymes responsible for IP₇ synthesis was initially identified as PiUS (phosphate inorganic uptake stimulator), a protein named after its role as a stimulator of inorganic phosphate uptake (20, 53, 54). This early observation suggested that inositol pyrophosphates possess an evolutionarily conserved role in regulating phosphate homeostasis, both in yeast and mammalian cells. The identification of yeast and human Nudix hydrolase family members possessing poly-P endopolyphosphatase activity further supports the evolutionarily conserved link between inositol pyrophosphates and cellular phosphate metabolism.

The inability to express recombinant active endopolyphosphatase Ppn1 (44) has limited the scope of research in the field of poly-P. Yeast Ddp1 and human DIPP recombinant enzymes are easy to produce in their active form and may allow the testing of several exciting hypothesis, such as whether poly-P may be covalently associated with proteins or with the inositol ring to prevent degradation by active exopolyphosphatases. The notion that poly-P is important in controlling platelet aggregation (4) makes the discovery of the first mammalian poly-P endopolyphosphatases even more relevant because its regulation may have further implications for human health.