Endopolyphosphatases for Long Chain Inorganic Polyphosphate in Yeast and Mammals*

Whereas exo polyphosphatases have been purified from yeast and a variety of bacteria, this is the first report characterizing endo polyphosphatases that act on long chain inorganic polyphosphate (polyP). The activity from Saccharomyces cerevisiae , localized in vacuoles, has been purified to homogeneity from a strain that possesses vacuolar proteases. The endopolyphosphatase is a dimer of 35-kDa subunits. Distributive action on polyP 750 produces shorter chains to a limit of about polyP 60 , as well as the more abundant release of polyP 3 ; the K m for polyP 750 is 185 n M . Endopolyphosphatases have been identified in a wide variety of sources, except for most eubacteria tested. The activity has been par- tially purified from rat and bovine brain where its abundance is about 10 times higher than in other tissues but less than 1 ⁄ 10 that of yeast; the limit product of digestion of the partially purified brain enzyme is polyP 3 . polyphosphates (polyP) 1 linear polymers of by phosphoanhydride

Whereas exopolyphosphatases have been purified from yeast and a variety of bacteria, this is the first report characterizing endopolyphosphatases that act on long chain inorganic polyphosphate (polyP). The activity from Saccharomyces cerevisiae, localized in vacuoles, has been purified to homogeneity from a strain that possesses vacuolar proteases. The endopolyphosphatase is a dimer of 35-kDa subunits. Distributive action on polyP 750 produces shorter chains to a limit of about polyP 60 , as well as the more abundant release of polyP 3 ; the K m for polyP 750 is 185 nM. Endopolyphosphatases have been identified in a wide variety of sources, except for most eubacteria tested. The activity has been partially purified from rat and bovine brain where its abundance is about 10 times higher than in other tissues but less than 1 ⁄10 that of yeast; the limit product of digestion of the partially purified brain enzyme is polyP 3 .
Inorganic polyphosphates (polyP) 1 are linear polymers of orthophosphate residues linked by high energy phosphoanhydride bonds. Likely prevalent in prebiotic evolution (1) polyP has been found in all organisms ranging from bacteria to mammals (2). The ubiquitous occurrence of polyP suggests multiple roles depending on the species, cell, subcellular localization, and physiological state.
Our approach toward understanding the functions of polyP has been to identify and isolate the enzymes that synthesize and degrade polyP. Such enzymes have been identified in a variety of microorganisms. PolyP kinase, the enzyme that catalyzes the reversible transfer of the terminal phosphate from ATP to synthesize polyP, has been purified to homogeneity in several bacteria (3)(4)(5). Exopolyphosphatases that catalyze the hydrolysis of terminal phosphates from polyP have been purified from Escherichia coli (6), Corynebacterium xerosis (7), and Saccharomyces cerevisiae (8,9). Phosphotransferases that transfer a phosphate from polyP to AMP (10), NAD (11), glucose (12), and 1,3-diphosphoglycerate (13) have also been described. Endopolyphosphatases (PPN) also called polyP depolymerases or polyphosphorylases catalyze the non-processive cleavage of polyP to release intermediate-size chains during the course of the reaction. PPN are the least studied of the polyP-metabolizing enzymes and have been reported in species of Penicillium and Aspergillus and in S. cerevisiae (14 -16).
During the course of investigating polyP metabolism in a variety of cells, we discovered PPN activities in organisms from archae to mammals but little or none among eubacteria. In this report, we describe the purification and characteristics of PPN from S. cerevisiae, where it is most abundant, using a mutant strain in which a major exopolyPase activity has been deleted (17).

EXPERIMENTAL PROCEDURES
Cells and Tissues-S. cerevisiae CRX (17) was grown in YPD medium (1% yeast extract, 2% tryptone, 2% glucose) at 30°C to an A 600 of 14.5. The harvested cells were resuspended in an equal volume of 50 mM Tris-HCl, pH 7.5, 10% sucrose, frozen in liquid nitrogen, and stored at Ϫ80°C.
Tissues were obtained from 4 -6-week-old Fisher 344 rats. Brain tissues at various developmental stages were obtained from Sprague-Dawley rats.
Preparation of Extracts-When microorganisms were the source of enzyme, the cells were suspended in 5 volumes of lysis buffer (0.25 M sucrose in 10 mM Tris-HCl, pH 7.0, and 1 mM EDTA) and sonicated (Branson Instruments) in an ice bath for three 15-s bursts with 30-s cooling between bursts. The homogenates were centrifuged at 25,000 ϫ g for 30 min and the supernatant used for analysis.
When tissue was the enzyme source, it was rapidly excised and washed with cold 0.25 M sucrose in 10 mM Tris-HCl, pH 7.0. The tissue was weighed and suspended in 5 volumes of lysis buffer and homogenized using a Polytron (Brinkmann Instruments) for three 10-s bursts with 30-s cooling intervals. The homogenate was centrifuged at 25,000 ϫ g for 30 min and the supernatant used for analysis.
To obtain shorter chains, polyP 750 was partially hydrolyzed near pH 2 by adding an equal volume of 20 mM HCl and heating for 2-6 min at 100°C. Partially hydrolyzed polyP 750 was fractionated by 6% ureapolyacrylamide gel and then eluted from gel slices into 50 mM ammonium acetate, followed by ethanol precipitation (4).
PPN Activity Assay-The reaction mixture (10 l) contained 100 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , and 350 M [ 32 P]polyP (as P i residues). The reaction was carried out at 37°C for various times, followed by the addition of 2.5 l of 5 ϫ electrophoresis sample buffer (50% sucrose, 0.125% bromphenol blue, 2.25 M Tris borate, pH 8.3, and 67.5 mM EDTA) to stop the reaction. Controls were incubated on ice or at reaction temperatures without addition of extracts. The reaction products were electrophoresed on 6% urea-polyacrylamide gel (21), exposed to a PhosphorImager screen (Molecular Dynamics), and analyzed. The assay is linear over a 60-min period under the described assay conditions at protein concentrations up to 0.2 g (when crude extracts are used as the enzyme source).
Units of PPN Activity-PPN activity was calculated by measuring the percentage decrease in counts at the origin with respect to total counts in the lane. The percentage was then converted to pmol of polyP utilized (as P i equivalents). A unit was defined as pmol of polyP (as P i residues) utilized per min; units/mg of protein gives the specific activity.
Gel Filtration of PPN-Gel filtration for determination of native molecular weight of PPN was at 4°C on a TSK-Gel G2000SW XL column (TosoHaas) equilibrated in buffer B (100 mM Tris-HCl, pH 7.0, 100 mM KCl, 1 mM dithiothreitol, and 0.1 mM EDTA). Elution was performed with buffer B at a flow rate of 0.7 ml/min. Marker proteins were detected by monitoring UV absorption (A 280 nm ). Fractions containing PPN activity were detected by assay. Molecular mass markers were * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Other Methods-Protein was estimated by the method of Bradford (22) using bovine serum albumin as a standard. SDS-PAGE of proteins was according to Laemmli (23) with size standards from Bio-Rad. Protein bands were visualized by silver staining (24).

RESULTS
PPN Activity Assay-Early assays for detecting PPN activity were based on measurement of changes in viscosity of polyP which decreased as the reaction proceeded (25). However, the main limitation of this method is its inability to detect cleavages in substrates with a chain length shorter than ϳ100 P i residues. A linked spectrophotometric assay using polyP glucokinase allows for calculating the number of polyP chains formed by PPN action and distinguishes between exo-and endopolyPase activity (26). An assay method based on the ability of polyacrylamide gels to separate polyP of different chain lengths was developed by Wieckowski and Wood (27) to measure PPN activity in crude yeast extracts. This allows separation of substrate polyP from medium and short chain products that are formed as a result of clevage of internal phosphoanhydride bonds. Activity is determined by measuring the decrease in concentration of the initial substrate. We used a modification of this method to assay PPN in various crude extracts as well as to follow the purification of yeast PPN. Sensitivity was increased over previously described assays by using [ 32 P]polyP having a length of 750 residues with small size variations (Ϯ50) compared with the previously used commercial substrates.
However, the main drawback of this assay is its inability to provide a quantitative measure of the number of cleavage events during the PPN reaction. However, when the assay is done under conditions where less than 50% of the substrate is removed, linearity with respect to both time and enzyme concentration was observed (data not shown). Also, by scanning across the length of each lane one can determine the relative amounts of products formed during the course of the reaction; this could be used as a basis for calculating the number of cleavages in the original substrate per unit time. However, for routine assays, the amount of substrate removed was determined to calculate the activity of the enzyme.
Abundance in Various Organisms-The ability of extracts prepared from various organisms to cleave long chain polyP was tested. Various amounts of extracts were used to hydrolyze 350 M [ 32 P]polyP, and activity was determined based on lin- a Standard error of 5% or less; units were corrected for the appearance of P i in the assay presumably due to exopolyPase action.

TABLE IV Purification of PPN
A paste of CRX cells was suspended in an equal volume of 50 mM Tris-HCl, pH 7.5, 10% sucrose, frozen in liquid nitrogen, and stored at Ϫ80°C. All subsequent steps were at 0 -4°C. Frozen paste (150 g) was thawed, mixed with an equal volume of 0.25 M sucrose, 10 mM Tris-HCl, pH 7.5, 1 mM EDTA and sonicated in a Branson Sonifier in 100-ml aliquots (60% duty cycle, 70% power) for five 1-min pulses with 2-min cooling intervals. The sonicate was centrifuged at 30,100 ϫ g for 30-min, the supernatant (fraction I) decanted, and the pellet discarded. Ammonium sulfate was added to the supernatant (360 ml) to a concentration of 0.48 g/ml; the mixture was stirred for 2 h and centrifuged at 30,100 ϫ g for 1 h. The pellet was resuspended in 50 ml of buffer A (50 mM Tris-HCl, pH 7.5, 10 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, 10% glycerol) and dialyzed against 4 liters of buffer A (three times for 2 h each). The dialysate (fraction II) was spun at 30,100 ϫ g for 30 min, made to 1 mM with MgCl 2 , incubated for 12 h on ice with DNase I and RNase A, each at 5 g/ml, and applied to a 100-ml heparin-Sepharose CL-4B column equilibrated with buffer A. The column was washed sequentially with 2 bed volumes of buffer A and buffer A containing 0.6, 0.8, and 1 M KCl. To the PPN activity, most abundant in the 0.8 M KCl wash, was added cytochrome c to 0.1 mg/ml and concentrated in an Amicon ultrafiltration cell to 10 ml (fraction III). Fraction III was dialyzed against buffer A as described above and loaded on a phosphocellulose column equilibrated with buffer A. The column was washed with 2 bed volumes of buffer A followed by 2 bed volumes each of buffer A containing 0.8 M sodium phosphate, 0.1 M sodium pyrophosphate, 1 M KCl, and 1 M MgCl 2 . To the PPN activity in the MgCl 2 eluate was added cytochrome c to 0.1 mg/ml and concentrated by ultrafiltration to 1 ml. The enzyme was dialyzed against 1 liter of buffer A (three times for 2 h each, fraction IV) and used for all further analyses. earity with respect to protein concentration and time. Inasmuch as widely varying sources were examined, the standard assay of PPN in each may not represent the true levels. Table  I summarizes the results of the survey under the described assay conditions. Among eubacteria, only Synechococcus (a cyanobacterium), had PPN activity; none was detected in E. coli, Bacillus subtilis, and Thermus aquaticus. Among the archaea, species belonging to crenarcheota had moderate levels of activity while the euryarchaeota tested (Pyrococcus furiosus and Thermococcus litoralis) had no detectable activity. All eukaryotes tested, ranging from Giardia, a protozoan considered to be the earliest diverging member within the eukaryotic lineage (28), to bovine brain had significant amounts of PPN activity. Abundance in Rat Tissues-PPN activity was found in all tissues tested (Table II). Brain had 10 times the specific activity of heart and kidney and 30 times that of lung and liver. Brain tissue from 15-day rat embryos had higher activity than that of developed brain (Table III). The crude brain enzyme appears to act by a mechanism similar to the action of the yeast PPN as judged by product distribution over a period of time after PPN action on polyP 750 (see Fig. 4, C and D).
Purification of the Enzyme-PPN was purified from S. cerevisiae CRX, a strain mutant in the ppx (exopolyPase) gene. Soluble enzyme was obtained only when the cells were broken by sonication; breakage with glass beads (29) or by Dounce homogenization of spheroplasts yielded preparations that were largely insoluble. 2 Removal of nucleic acids by DNase I and RNase A was essential for obtaining PPN as a single fraction from heparin-Sepharose (Table IV). Activity could not be eluted from a phosphocellulose column with increasing concentrations of phosphate or KCl up to 1 M; however, elution with 1 M MgCl 2 was effective in yielding a homogeneous protein as visualized on SDS-PAGE after silver staining (Fig. 1). In essence, two affinity columns led to a 3,500-fold purification. Cytochrome c was added to stabilize the enzyme when protein concentrations were less than 0.01 mg/ml. The molecular mass of denatured PPN was estimated to be 35 kDa by SDS-PAGE (Fig. 1); the native mass was judged to be 80 kDa by gel filtration on a TSK-Gel G2000SW XL high performance liquid chromatography column (Fig. 2).
Requirements for Activity-The pH profile showed an optimum near 7.5 (Fig. 3A). Activity required a metal ion; 10 mM EDTA resulted in total inhibition. Mn 2ϩ was more active than Mg 2ϩ with optima near 2.5 mM (Fig. 3B); CaCl 2 and ZnCl 2 at 1 mM inhibited by 35%. NaF at 10 mM and (NH 4 ) 2 SO 4 and NaCl, each up to 50 mM, had no effect on activity (data not shown). The addition of P i resulted in a 50% inhibition at 20 mM, and pyrophosphate at 10 mM inhibited completely (Fig. 3C). Inas-2 P. Ramulu and A. Kornberg, unpublished observations.  (Table IV) were analyzed by SDS-PAGE (4 -15% gradient); proteins were visualized by staining with silver. Molecular size markers are in the left column. much as lysine and arginine are present in large amounts in the yeast vacuole along with almost 99% of the cellular polyP, these amino acids were tested up to 100 mM but without any effect on activity (data not shown).
Effect of Chain Length-The enzyme activity was determined for various sizes of polyP chains used as substrates. Initial rates were plotted against a series of substrate concentrations for each chain length, and the K cat /K m was determined. No significant interpretation could be drawn from the K cat /K m values, which ranged from 20 to 50 min Ϫ1 nM Ϫ1 . With lengths below 60 residues virtually no substrate was utilized (Ͻ1% at concentrations up to 12 M). PPN did not hydrolyze polyP 15  Analysis of Reaction Products-As observed on its action on polyP 750 , PPN action was nonprocessive, giving rise to polyP of intermediate chain lengths throughout the course of the reaction. The products, characterized by electrophoresis on 6% polyacrylamide gel (Fig. 4A), showed a progressive decrease in length along with an appearance of polyP 3 (Fig. 4B). The sequential increase in intermediate size products indicated a clear preference for the longer chains. Based on PhosphorImager data, hydrolysis virtually stopped when the chains reached about 60 residues. The ratio of polyP products (Ն60) to polyP 3 as polymer was rather constant with a value near 3 at each time point (Table V and Fig. 4, A and B). Thus, only one of four cleavages releases a large molecule of polyP, the other three release polyP 3 from the chain end. Additional polyP species of sizes between 3 and 60 made up ϳ10% of the total products. These could result from the random intramolecular cleavage of polyP chains during the course of PPN action.
Localization of PPN-PPN activity in extracts of S. cerevisiae CBX (17), a strain lacking vacuolar proteases (in addition to exopolyPase), was 20-fold lower than that found in extracts of CRX. PPN activities in vacuolar extracts prepared (30) from each of these strains showed a 14-fold enrichment in those from vacuoles compared with the whole cell lysates from CRX strains, suggestive of a vacuolar location for PPN. No difference in PPN activity was detected between vacuolar extracts and the whole cell lysates of protease-deficient mutant CBX. DISCUSSION Depolymerases of nucleic acids, proteins, and polysaccharides have proven to have important metabolic roles, acting internally (endo) or at the ends of chains (exo). With regard to inorganic polyP, exopolyPases have been identified and characterized in bacteria (6,7) and yeast (8,9), but not in animal cells. As for PPNs, little is known about them from any source.
In our attempts to discover enzymes responsible for the synthesis of polyP in animal cells and tissues, we encountered instead formidable PPN activity. This activity was present in all rat tissues, particularly in brain (Table II); during fetal development the levels in brain were higher still (Table III). Brain tissues enriched in glial cells (spinal cord and optic nerve) were further enriched. Even though bovine brain could be used as a source material, the relatively low abundance of the PPN activity would make purification of the enzyme rather difficult.
A survey of microbial and other sources for PPN activity revealed a particularly high level in yeast (Table I); remarkably little or none was detected in E. coli, in which exopolyPase activities are abundant (6). For the PPN activity to be manifest in yeast, it is essential that strains possess vacuolar proteases, indicative of a need for proteolytic processing of the enzymes. For ease of assay and purification, a strain was chosen in which a potent exopolyPase was eliminated by mutation (17). The PPN activity, as judged by subcellular fractionation, was concentrated in the vacuoles, although purification of the enzyme proceeded with extracts prepared from sonication of the whole cell.
Purification of the PPN activity entailed a 3,500-fold enrichment with a 23% yield (Table IV). The procedure was based largely on affinity to polyanionic exchange resins: heparin-Sepharose and phosphocellulose. A noteworthy feature of the chromatography on phosphocellulose was the rather novel use of 1 M MgCl 2 as the eluant after a high concentration of a monovalent salt (1 M KCl) had failed. The final preparation appeared as a single polypeptide of 35 kDa on SDS-PAGE ( Fig.  1) but gave a molecular mass of 80 kDa on gel filtration, indicating its possible existence as a dimer.
The divalent metal ion required for activity can be supplied by Mg 2ϩ at 5 mM or by Mn 2ϩ at a lower concentration (Fig. 3B); Ca 2ϩ and Zn 2ϩ were inhibitory. Both P i and PP i reduced activity (Fig. 3C); 50% decrease was observed with P i at 20 mM and with PP i at 2 mM. Sodium fluoride, ammonium sulfate, and sodium chloride did not influence activity, nor did arginine or lysine, regarded as vacuolar counterions (31).
The affinity of polyP 750 , measured by K m , was 185 nM. The K cat /K m values were similar (between 20 and 50 min Ϫ1 nM Ϫ1 ) over a chain length range of 100 -750 residues. No activity was detected with polyP 60 , one of the two end products of the reaction. With regard to the course of the reaction, a nonprocessive mode of action produces chains of decreasing size until the final accumulation of predominantly polyP 60 and polyP 3 (Fig. 4A). Additional evidence supports the inability of PPN to hydrolyze polyP Յ 60; no hydrolytic products were observed when polyP over a range of 3-60 were used as substrates. Kowalczyk and Wood (32) have reported that PPN activity in partially purified S. cerevisiae extracts decreased with decreasing chain lengths (range of 30 -700 P i residues) and dramatically dropped with polyP of lengths below 100 residues. With every chain cleavage internally, it appears that three polyP 3 molecules are cleaved from the chain ends until a length of 60 residues is reached (Fig. 4B).
Discussion of the role of PPN would seem too speculative with little known about the metabolism of polyP in yeast, its biosynthesis, utilization, and potential multiple functions (33)  The reaction was performed as described under "Experimental Procedures" and the products analyzed on PhosphorImager after electrophoresis on 6% urea-polyacrylamide gel. The amounts of polyP products formed during the course of the reaction were calculated in terms of polymer concentration. The ratios of long chain products to polyP 3 are given.  /polyP products  3  3  3  3  3 FIG. 4. Products of PPN action. The reaction was performed as described under "Experimental Procedures," and the hydrolysis products electrophoresed on 6% urea-polyacrylamide gel. For reference, unlabeled polyP of various lengths were electrophoresed and stained with toluidene blue. Panel A, products of yeast PPN were determined by autoradiography of a 6% gel. Panel B, PhosphorImager analysis of the gel pattern with three chain lengths given for each lane as indicated. Panel C, autoradiogram of products of rat brain PPN. Panel D, PhosphorImager analysis of gel pattern as described for the yeast enzyme.
in the vacuole and other cellular locations. Were the fate of polyP hydrolysis by PPN simply to furnish P i , then the polyP 60 and polyP 3 end products could be degraded further by known exopolyPases (8,9) and tripolyPase (34). Were there specific functions assigned to these products, few clues exist from studies of other cells and known from the yeast cell. Clearly, more information is needed about the metabolism of a polymer that in yeast may account for 20% of its dry weight (20) and the vast bulk of its phosphate.