Turnover of the Acyl Phosphates of Human and Murine Prothymosin α in Vivo *

Prothymosin α is a small, highly acidic, abundant, nuclear, mammalian protein which is essential for cell growth. Our laboratory has recently shown that primate prothymosin α contains stoichiometric amounts of phosphate on the glutamyl groups of the protein and that in vitro the phosphate undergoes rapid hydrolysis or transfer to a nearby serine residue. Here an assay for the presence of acyl phosphates in vivo has been developed by measuring stable phosphoserine and phosphothreonine in vitro. The assay was used to determine the half-life of the acyl phosphates on prothymosin α in vivo by pulse-labeling HeLa cells with [32P]orthophosphate and chasing using three different techniques: permeabilization with digitonin to allow extracellular ATP to equilibrate with the intracellular pool; electroporation in the presence of ATP to reduce the specific activity of [32P]ATP by expansion of the pool; and incubation with inorganic phosphate. Regardless of the method, the phosphate turned over with a half-life of 75–90 min. The ability of cells to phosphorylate old prothymosin α molecules was established by demonstrating equivalent labeling of the protein with [32P]orthophosphate in the presence and absence of cycloheximide. The half-life of the acyl phosphates was also studied in resting and growing NIH3T3 cells, with measured values of 30–35 and 70 min, respectively. Our data suggest that the “activity” of prothymosin α involves the turnover of its acyl phosphates and that it participates in a function common to all nucleated mammalian cells regardless of whether they are quiescent or undergoing rapid proliferation. This is the first measurement of the stability of protein-bound acyl phosphates in vivo.

Prothymosin ␣ is a small, highly acidic (1, 2), nuclear (3)(4)(5)(6) protein found in virtually all mammalian tissues (7)(8)(9)(10)(11). Under conditions of rapid growth, a cultured cell accumulates upwards of 17 million molecules, a number roughly equivalent to that of histone cores (12). When cells are disrupted with detergents, the protein readily leaks out of the nucleus, suggesting that stable interactions with nucleosomes or with the nuclear matrix are not an inherent part of its activity (2,3). Its precise function is unknown. Nevertheless, a role in cell proliferation has been proposed based on the following observations: prothymosin ␣ mRNA is plentiful only in rapidly dividing cells (13)(14)(15); the level of the protein declines 10-fold in cells forced to subsist in stationary phase (12); the amount of prothymosin ␣ is directly proportional to the proliferative activity of the tissue from which it is isolated (13); and the uptake of antisense oligodeoxyribonucleotides directed toward various locations in prothymosin ␣ mRNA prevents synchronized human myeloma cells from entering mitosis (16). Hence, a deficiency in prothymosin ␣ is associated with failure to complete the cell cycle.
Prothymosin ␣ has several unusual features. The human protein, which is almost identical to that of all other mammals (17), has 109 amino acids, nearly 50% of which are acidic (1,2). A potent nuclear localization signal consisting of five basic amino acids has been identified near the carboxyl terminus (3,6), while a second cluster of five basic residues near the amino terminus has no unambiguous role (3,18). The absence of all aromatic residues renders the protein transparent at 280 nm; it also lacks methionine, cysteine, and histidine. Based on biophysical data, it is believed to have an unfolded structure (19), a conclusion consistent with the presence of only seven widely dispersed hydrophobic residues in the human protein. Prothymosin ␣ does not bind SDS and stains anomalously with silver stain (2). The protein and the peptides derived from it exhibit poor immunogenicity. However, due to yet another aberrant property, the ability to partition quantitatively into the aqueous phase of a phenol extraction, prothymosin ␣ can easily be obtained as a single band in a Coomassie Blue-stained gel.
Recently, we have shown that prothymosin ␣ contains phosphorylated glutamic acid residues (20). Acyl phosphates occur in proteins when a glutamic or aspartic acid residue undergoes a posttranslational modification producing a mixed anhydride of a carboxylic acid with phosphoric acid. Such phosphates are highly unusual, energy-rich, and easily transferred to any convenient hydroxyl group. In the prokaryotic world, in yeast, and in plants, they are found as aspartyl phosphates on the response regulator proteins of two component systems (reviewed in Refs. 21 and 22). These reactive acyl phosphates are acquired from phosphorylated histidine residues on the sensor components; they readily hydrolyze, with half-lives from 6 s to 3.5 min when measured in vitro (23)(24)(25). In higher eukaryotes, aspartyl phosphates occur as catalytic intermediates of P-type ATPases (reviewed in Ref. 26) with ATP as the physiological donor. Glutamyl phosphate has been identified on only one mammalian protein (prothymosin ␣), one amphibian protein (nucleoplasmin from Xenopus laevis), 1 and one avian protein (bone collagen of the chicken (27)). Little is known about the origins of glutamyl phosphates, and virtually nothing is understood about their contribution to cellular metabolism.
To expand our grasp of the function of prothymosin ␣, we have determined the half-life of its glutamyl phosphates in vivo. To do so, we made use of the properties established by Trumbore et al. (20), who showed that these phosphates are unstable in cell lysates and undetectable, except for a small fraction that apparently transfer to serine 1 of the human * 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  protein (12) or to unspecified threonine residues located within the first 14 amino acids of the murine protein (28). The degree of accumulation of the stable component depends on the conditions experienced by prothymosin ␣ at the moment of cell lysis. Here we demonstrate that the fortuitous, albeit inefficient, migration of the labile glutamyl phosphate to positions of stability on serine or threonine on the same molecule can form the basis of a quantitative assay. When this technique was used together with three independent pulse-chase methods, we found that both new and old prothymosin ␣ molecules were indistinguishable targets for phosphorylation on glutamic acid and that all acyl phosphates were rapidly lost in vivo with half-lives slightly in excess of 1 h. We believe that turnover of prothymosin ␣'s phosphates might reflect a role that is continuously required by all cells, regardless of their metabolic state, and we present evidence for associating prothymosin ␣ with processes required both for the maintenance of cells and for their growth.

EXPERIMENTAL PROCEDURES
Cells, Culture Conditions, and Harvest Procedures--HeLa S3 cells, African green monkey kidney cells (COS-1), and NIH3T3 cells were grown in Dulbecco's modified Eagle's medium (DMEM) 2 from Life Technologies, Inc. (catalog number 15240-039) or Biofluids (catalog number 172; Rockville, MD). Human and monkey cells were cultured with 10% heat-inactivated fetal calf serum (Life Technologies, Inc. or HyClone (Logan, UT)), whereas the murine cells grew in 10% calf serum. All culture fluids contained 2 mM glutamine, 90 units/ml of penicillin, 90 g/ml of streptomycin, and 0.22 g/ml amphotericin B, and all cells were maintained in an environment of 5% CO 2 at 37°C. Quiescent NIH3T3 cells were obtained by incubating cells for 48 h in 0.25% calf serum. Cells were harvested from culture flasks by washing with Puck's saline and treating with 0.05% trypsin in Hanks' balanced salts containing 0.53 mM EDTA.
Development of an Assay for Glutamyl Phosphate on Prothymosin ␣-A two-step approach was employed for relating 32 P found in isolated prothymosin ␣ on serine or threonine to the amount of glutamyl phosphate existing on the protein inside the cell. First, the relationship between the amount of 32 P recovered stably on prothymosin ␣ and the amount of prothymosin ␣ was determined by transfecting 10 6 COS cells in 60-mm dishes with 0 -3 g of pRSV PTMA, a vector containing the prothymosin ␣ gene (29), using the DEAE-dextran method (30). These cells, expressing varying amounts of prothymosin ␣, were labeled 44 h post-transfection with 100 Ci/ml of carrier-free [ 32 P]orthophosphoric acid for 4 h. Prothymosin ␣ was isolated by means of a phenol extraction (2), further purified electrophoretically, and quantified both as a Coomassie Blue-stained band of protein and as a radioactive band on an autoradiograph (see below). Direct comparisons were carried out only with samples analyzed on the same gel and subjected to the same conditions of staining and autoradiography.
The relationship between the amount of prothymosin ␣ and the amount of glutamyl phosphate was determined by resuspending ϳ4 -10 ϫ 10 7 COS cells in 2 ml of dimethyl sulfoxide containing 30 mM [ 3 H]NaBH 4 (NEN Life Science Products; specific activity, 1000 mCi/ mmol) at a specific activity of 47 mCi/mmol (low specific activity method) and allowing the reaction to proceed for 10 min at room temperature (20,31). Alternatively, COS cells were lysed in 1 ml of 5 mCi of [ 3 H]NaBH 4 (222 mCi/mmol) in dimethyl sulfoxide (high specific activity method), and reactions were carried out for 1-3 h. In both cases, the insoluble material was washed in dimethyl sulfoxide, centrifuged, and suspended in water. Prothymosin ␣ was recovered from the soluble component with the aid of phenol (2), and purified using high pressure liquid chromatography (20). As a result of the borohydride reaction, tritium is incorporated into hydroxynorvaline, which is recovered as [ 3 H]proline after acid hydrolysis, derivatization, and amino acid analysis (20,32). The reaction is specific, but even in well defined solutions yields are low (31), and in crude solutions yields are quite poor (20). Radioactivity incorporated into prothymosin ␣ was measured in a Pack-ard Tri-Carb model 1500 liquid scintillation analyzer in Hydrofluor (National Diagnostics), and protein was quantified from the amino acid analysis. The two sets of data, 32 P and 3 H found in measured quantities of prothymosin ␣, form the basis of the assay.
Determination of the Cell Volume-Cell volume was ascertained using two techniques. In the first case, the volume occupied by 1.9 ϫ 10 6 or 3.8 ϫ 10 6 HeLa cells in a total volume of 20 or 25 l of medium was measured in a calibrated capillary tube. An average value of 1.65 l was obtained. The second method is a dilution assay; 10 6 cells in quadruplicate were allowed to take up 1 or 2 Ci/ml [ 32 P]orthophosphoric acid for 1.5 h. They were washed free of external radioactivity, resuspended in 1 ml of phosphate-buffered saline (PBS), and used for a determination of Cerenkov radiation to measure the amount of label imbibed. The cells were then removed from scintillation vials and recovered by centrifugation. The pellets from two samples were resuspended in 10 l of PBS, while the pellets from the remaining two samples were lysed with 10 l of 10 N NaOH. It should be understood that, at this point in the experiment, the total volume is 10 l of PBS plus the volume of the cell or cell lysate and that the amount of radioactivity, but not its concentration, is known. To determine the concentration, aliquots of 2 and 4 l of the cell suspension or lysate were subjected to scintillation counting. If the cells had contributed nothing to the volume, these samples would have contained 20 or 40%, respectively, of the initial radioactivity imbibed by the cells. However, because the intact cells or the lysed cells do occupy space, their volume can be determined by measuring the reduction in radioactivity caused by the dilution of the buffer or base by the cells or their contents. These measurements resulted in average values of 1.49 and 1.52 l/10 6 cells for intact and lysed cells, respectively; they were essential for determining the degree of equilibration of ATP across a permeabilized plasma membrane (see below).
Determination of the Specific Activity of the ATP Pool-The method of Lee et al. (33), based on the work of Sasvá ri-Székely et al. (34), was used to determine the specific activity of the ATP pool of HeLa cells. Briefly, the technique makes use of the fact that the number of AMP residues equals the number of UMP residues when RNA is polymerized using poly(dA-dT) as the template. The ATP, which is synthesized in vivo by cells in the presence of [ 3 H]adenosine, is supplied by the cell extract, whereas UTP of known specific activity is provided exogenously in amounts which are large relative to the amount of endogenous UTP in the same cell extract. Under these conditions, all of the ATP in the extract is used to generate polymer, with no free ATP remaining in the reaction mix. More specifically, an aliquot of 1% of a neutralized perchloric acid extract from 5 ϫ 10 6 cells labeled for 2 h with 50 Ci/ml [2, H]adenosine (specific activity, 30.6 Ci/mmol; NEN Life Science Products) was used as a source of ATP for the polymerization of RNA in a 50-l reaction containing 0.2 g of poly(dA-dT) as template, 0.1 Ci of [U-14 C]UTP (609 mCi/mmol, NEN Life Science Products) 0 -5 nmol of nonradioactive UTP, and 3 units of RNA polymerase holoenzyme from Escherichia coli. The bacterial polymerase initiates synthesis of RNA at any location without the need for specific promoter sequences. Synthesis terminates when the limiting nucleotide, in this case ATP supplied by the extract, is exhausted. The specific activity of ATP given by the expression, (dpm in 3 H/dpm in 14 C) ϫ specific activity of UTP, was 6.3 ϫ 10 5 cpm/nmol. Thus, 10 6 cells with an average cell volume of 1.6 l contained 1.4 nmol of ATP at a concentration of 0.88 mM.
Pulse-Chase Experiments in Permeabilized Cells-Conditions for the use of digitonin were selected after the evaluation of several parameters. Each dish of 5-6 ϫ 10 6 HeLa S3 cells in a variation of a solution termed "cytomix" (35) (120 mM KCl, 150 mM CaCl 2 , 10 mM potassium phosphate at pH 7.6, 25 mM HEPES-KOH at pH 7.6, 2 mM EGTA, and 5 mM MgCl 2 ) or complete DMEM was treated with 20 -40 g/ml digitonin (Sigma) at either 0 or 37°C for periods ranging from 5 to 30 min. In some experiments, the cells were labeled with [ 3 H]glutamic acid before initiating the permeabilization treatment. After removal of digitonin by gentle aspiration and washing with Puck's saline, the chase conditions were simulated by incubating the cells in either [␥-32 P]ATP or [␥-32 P]UTP in either cytomix or DMEM without serum at 37°C for 0.5-4 h; here radioactive ATP or UTP served only as a marker for measuring the rate at which permeabilized cells could imbibe external substances. Optimal conditions appeared to be treatment with 5 ml of 20 g/ml of digitonin for 10 min in cytomix on ice. Such conditions allowed labeled ATP, placed outside the cell in either DMEM or cytomix, to equilibrate with the interior in less than 1 h.
For pulse-chase experiments, HeLa cells at 37°C were labeled in 60-mm dishes with 100 Ci/ml of [ 32 P]orthophosphoric acid (catalog number NEX-053, NEN Life Science Products) for 4 h in 2 ml of phosphate-free DMEM (Life Technologies, Inc.) containing the additions noted above. The cells were washed sequentially with warm Puck's saline followed by cytomix at 37°C, permeabilized using the optimized conditions noted, and chased in 5 ml of DMEM (without serum and additives) containing 10 mM ATP and 10 mM MgCl 2 at pH 7.0. Our methods were developed from published procedures (36). Permeabilized cells were maintained at 37°C in the incubator during the chase. A separate dish was used for each time point. At the stated time, the cells were harvested by aspirating the ATP-containing chase medium and by immersing them, still in the dish, in 1 ml of lysis buffer (10 mM Tris-HCl at pH 7.5, 5 mM EDTA, 12% sucrose, 1% Triton X-100, and 1 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride-HCl (AEBSF, Sigma)). Cells and debris were scraped free with a cell lifter (Costar), transferred to an Eppendorf tube, cooled in ice, mixed vigorously with a vortex mixer, and centrifuged at 4°C for 10 min at 15,000 rpm in a Tomy model MTX-150 centrifuge. The supernatant fluid was transferred to a 15-ml polypropylene tube, brought to 4 ml with cold filter-sterilized water, and extracted with phenol as described below. It is important to note that prothymosin ␣ in its entirety leaks out of the nuclei when digitonin-treated cells, as well as electroporated or normal cells, are lysed with nonionic detergents (2) and that prothymosin ␣ yields in all cases are virtually identical.
Pulse-Chase Experiments with Electroporated Cells-HeLa S3 Cells were electroporated in the presence of [ 32 P]ATP in the medium to determine the degree to which molecules outside the cell had equilibrated with the internal environment. An aliquot consisting of 1 ml of cells was placed in a disposable electroporation chamber with a 0.4-cm gap (catalog number 11601-028, Life Technologies, Inc.); the chamber was cooled in ice for 10 min and shocked with an electrical discharge of 875 V/cm and 330 microfarads in a Life Technologies, Inc. Cell-Porator at room temperature with the low resistance setting. The amount of a known concentration of [ 32 P]ATP taken up by the cells, 70 -80% of which survived, was measured and found to represent 70% equilibration.
For pulse-chase experiments, HeLa S3 cells were labeled in 175 cm 2 flasks under the conditions noted above, trypsinized, washed in Puck's saline, and resuspended at 5 ϫ 10 6 cells/ml in warm cytomix to which 20 mM ATP and 20 mM MgCl 2 at pH 7.0 had been added. The chase was initiated by electroporating as noted above. These conditions resulted in a 14 -15-fold instantaneous decrease in the specific activity of intracellular ATP. Following electroporation, the chamber was again cooled to 4°C for 10 min and brought to room temperature for 10 min. Cells recovered from each chamber were washed thoroughly with Puck's saline, centrifuged free of the wash solution, resuspended in 5 ml of complete DMEM (DMEM containing serum and additives), seeded into a 60-mm dish, and maintained in an incubator as described earlier. The unphysiological concentration of ATP achieved inside the cells after electroporation did not significantly affect the viability of the cells for the duration of the chase.
For the harvest, the cells in each dish were scraped free with a cell lifter, transferred to a centrifuge tube, cooled to 4°C, washed with PBS, recovered as a pellet, and disrupted in 1 ml of cold lysis buffer. After removing the nuclei by centrifugation, the supernatant fluids were transferred to a clean polypropylene tube, diluted with water as detailed above for digitonin-treated cells, and subjected to a phenol extraction (see below).
Pulse-Chase Experiments with Sodium Phosphate-HeLa S3 cells were labeled in flasks in phosphate-free complete DMEM for 4 h as indicated for the electroporated cells. To initiate the chase, the cells were washed free of the labeling solution, trypsinized, washed again, resuspended in complete DMEM containing 40 mM sodium phosphate at pH 7.0, and seeded into dishes at a concentration of 6 ϫ 10 6 cells/dish. At the end of the chase, the cells were recovered and lysed using the methods for electroporated cells.
Purification and Analysis of Prothymosin ␣-Supernatant fluids from lysed cells containing virtually all of the prothymosin ␣ in the cell in 4 ml were made 0.5% in SDS and extracted at 65°C with 2 ml of phenol saturated with 2 ϫ ACE buffer (20 mM sodium acetate at pH 5.1, 100 mM NaCl, and 6 mM disodium EDTA). The aqueous phase was recovered by centrifugation and extracted twice with phenol using the same methods. The final aqueous phase was precipitated with 4 volumes of acetone in dry ice for 1 h or overnight at Ϫ20°C, and the sample was recovered by centrifugation at 16,000 ϫ g for 30 min, washed in 80% acetone, 20% 20 mM Tris-HCl at pH 7.5, and dissolved in 400 l of 20 mM Tris-HCl at pH 7.5. RNA was destroyed by adding 40 g of pancreatic ribonuclease A and incubating the sample at 37°C for 20 min. Ribonuclease was removed by two additional phenol extractions. The sample was recovered by precipitation in acetone, and the pellet was dissolved in water and analyzed in an 18% polyacrylamide gel (catalog number EC6506, Novex). A general description of our methods for purification of the protein and for electrophoresis has appeared (2,12). Prothymosin ␣ was visualized by staining with Coomassie Brilliant Blue, and, when labeled with 32 P, radioactivity was detected by exposing the dried gel to XAR x-ray film. Stained protein and radioactive protein were quantified by scanning wet gels and films, respectively, with a Molecular Dynamics ImageQuant scanning densitometer. Regression lines, from which half-lives were determined, were fit using linear least squares analysis.
Peptides were also used as substrates for phosphorylation by Inhibition of Protein Synthesis-HeLa cells (5 ϫ 10 6 ) were treated with cycloheximide (Sigma) at 100 g/ml for 1 h in complete medium. Cells, with the inhibitor present in the culture fluids, were labeled with 100 Ci/ml of [ 32 P]orthophosphoric acid for 4 h, harvested, lysed as noted above for electroporated cells, and subjected to a phenol extraction to obtain prothymosin ␣.
Half-life of RNA-In some experiments, the quality of the chase was monitored by measuring the survival of pulse-labeled [ 32 P]RNA in the cytosol of [ 32 P]orthophosphate-labeled HeLa cells. For this purpose, the trichloroacetic acid-insoluble material in 20 l of the aqueous phase of a phenol extraction was recovered on a GF/C glass fiber filter and assayed for Cerenkov radiation in a Packard Tri-Carb 1500 liquid scintillation analyzer (38). Such samples are composed almost entirely of labeled RNA, because the contribution of prothymosin ␣ is insignificant and DNA remains with the nuclear debris.

RESULTS
Provenance of Prothymosin ␣'s N-Acetylserine Phosphate-It has been shown that a maximum of 2% of prothymosin ␣ molecules, when purified to homogeneity from bovine thymus or from 32 P-labeled human myeloma cells, contain phosphate on the N-terminal acetylserine residue (12). It has also been shown that prothymosin ␣ at the moment of cell lysis bears stoichiometric amounts of highly unstable glutamyl phosphate, the majority of which succumbs to hydrolysis (20). The two types of phosphate are related; conditions which best supported the transient survival of unstable phosphate resulted in the highest yield of stable phosphate on serine or, to a lesser extent, threonine. By manipulating conditions in vitro, we were able to show that Ͻ10% of the stable component recovered under optimal conditions could provisionally be attributed to a phosphorylation event that occurred in vivo (20). To rule out the possibility of direct, in vivo phosphorylation of serine and threonine residues on any prothymosin ␣ molecule, we first searched for and failed to find [ 32 P]prothymosin ␣ that was synthesized from [␥-32 P]ATP during cell lysis (20). Apparently, phosphorylation directly on serine or threonine as an artifact in vitro had not contributed to the tally of stable phosphate.
We then focused on whether or not cells contain a kinase capable of phosphorylating prothymosin ␣'s N-terminal acetylserine residue. If this position is a phosphate acceptor in vivo, cell extracts could be capable of performing the reaction in vitro. Phosphorylation in vitro was tested with two peptides containing an N-terminal acetylserine residue as a target and six histidine residues attached to the carboxyl terminus for ease in purification: Ac-SAAAADAAVDHHHHHH, a control with the acetylserine group separated from the prothymosin ␣ N-terminal sequences by four Ala residues, and Ac-SDAAVDH-HHHHH, the test peptide with six consecutive N-terminal residues identical to prothymosin ␣. The putative kinase was supplied by cytoplasmic or sonicated whole cell extracts prepared at pH 7.5. When high concentrations of peptides, [␥-32 P]ATP (either with or without added Mg 2ϩ ), and the cell extracts were incubated together for 30 min at 4 or 37°C, phosphorylation of cellular proteins occurred (data not shown). In contrast, when the lysates were fractionated by nickel column chromatography to recover the specific peptides, no covalently bound phosphate was detected regardless of the conditions. As shown in Fig. 1 (top, [ 32 P-␥]ATP), the stained peptides could be visualized with Coomassie Blue dye after electrophoretic analysis, but no evidence for phosphorylation of either peptide was obtained (Fig. 1, bottom, right). Furthermore, the peptides did not become radioactive when added to the buffer in which cells labeled in vivo with [ 32 P]orthophosphate were lysed (Fig. 1, top, 32 P Labeled Lysates, and bottom). These data strongly suggest that direct phosphorylation on acetylserine is not readily achieved by cellular kinases and that the required enzymes may be absent from COS cells. The failure to phosphorylate serine or threonine residues directly on either prothymosin ␣ or related peptides under a variety of conditions or to transfer phosphate intermolecularly from 32 Plabeled molecules of all types to prothymosin ␣ during cell disruption (20) seriously weakened the case for the biosynthesis of phosphoserine or phosphothreonine. Having already demonstrated the intramolecular transfer of prothymosin ␣'s labile phosphates to stable positions and having accounted for 90% of the stable phosphate in this manner (20), we concluded that virtually all of the stable phosphate arises from glutamyl phosphate.
Use of Stable Serine Phosphate as an Assay for Unstable Glutamyl Phosphate-Quantitative correlation between glutamyl phosphate and serine (or threonine) phosphate on prothymosin ␣ was established in two steps. In the first step, determinations were made of the amount of stable [ 32 P]prothymosin ␣ recovered from cells labeled in vivo and the amount of prothymosin ␣ protein recovered from COS cells that were transfected with increasing amounts of the gene. The data, displayed in Fig. 2A, show that, in each experiment consisting of eight measurements with the protein resolved in the same gel, the relative amounts of stained prothymosin ␣ and of stable radioactive phosphate were proportional. (See open squares, triangles, or circles.) To consider three independent experiments at once, both the total intensity from the stained gel and the total intensity on the autoradiogram were obtained, averaged, and set equal to 100 arbitrary units for each data set. Data points are percentages of that value for their respective experiments. In this way, the axes in Fig. 2A are arbitrary, but the fit of the points to a line is not arbitrary. The data show that the specific activity of prothymosin ␣, measured as stable [ 32 P]prothymosin ␣ divided by total prothymosin ␣ in each experiment, is constant regardless of how much prothymosin ␣ the cell contained.
In the second step, both the amount of tritium incorporated into prothymosin ␣ at sites of acyl phosphorylation as a consequence of reduction with [ 3 H]NaBH 4 and the total amount of prothymosin ␣ were directly determined. Fig. 2B compares the amount of nonexchangeable tritium found in proline after amino acid analysis and the absolute amount of prothymosin ␣ from which it came using two reduction protocols. The data, which reflect 20 completely independent determinations, show that the amount of tritium, a measure of glutamyl phosphate, and the amount of prothymosin ␣ were directly proportional. Since stable [ 32 P]prothymosin ␣ is proportional to prothymosin ␣ and since labile phosphate, measured as tritium incorporated into proline upon reductive cleavage, is also proportional to prothymosin ␣ protein, stable serine phosphate must be proportional to labile glutamyl phosphate. This relationship provides the basis for quantifying the relative amounts of [ 32 P]phosphoglutamate in vivo.
Half-life of [ 32 P]Prothymosin ␣ in Permeabilized HeLa Cells-Digitonin, which removes cholesterol from plasma membranes, leaving the cytoplasm of treated cells in direct communication with the external medium (39), was the agent chosen for permeabilization. Cells were pulse-labeled, permeabilized, and chased in the presence of 10 mM ATP, a large concentration relative to the measured intracellular concentration of 0.9 mM ATP (see "Experimental Procedures"). Under these conditions, the specific activity of [ 32 P]ATP decreased rapidly, and a pronounced decline in the radioactivity incorporated into prothymosin ␣ was observed. As shown in Fig. 3A, which displays an autoradiogram of electrophoretically purified prothymosin ␣, the chase appeared to be effective within 0.5 h, leaving less than 10% of the initial radioactivity associated with the protein after 8 h. These data, corrected for the recovery of prothymosin ␣ in each sample, are shown in Fig. 3B. Because the decay followed first order kinetics, we were able to determine a halflife of ϳ80 min for phosphoprothymosin ␣ from the curve. During the interval (0 -8 h), [ 32 P]inorganic phosphate continued to accumulate in cytoplasmic mRNA, the most prevalent polynucleotide in our preparations (data not shown). The observation suggests that these permeabilized cells, which are no longer viable, retain the ability to synthesize RNA and export it from the nucleus.
Half-life of [ 32 P]Prothymosin ␣ in Electroporated HeLa Cells-To measure the half-life of the phosphate covalently attached to prothymosin ␣ in living cells, we made use of electroporation as a means of diluting intracellular labeled ATP with an excess of its nonradioactive counterpart. When HeLa cells were pulse-labeled with [ 32 P]orthophosphoric acid and chased by electroporation in cytomix containing 20 mM ATP, the half-life of phosphoprothymosin ␣ was ϳ90 min (Fig.  4), a value in good agreement with the half-life measured in digitonin-treated cells. However, unlike the experiment with digitonin, in which vanishingly small quantities of label remained in prothymosin ␣ after 18 -24 h, the cells chased by electroporation always retained 10 -20% of the radioactivity incorporated during the pulse (data not shown). This observation is consistent with an experimental design in which the FIG. 2. The relationship between prothymosin ␣ protein and its labile glutamyl phosphate and stable serine (and threonine) phosphate. A, COS-1 cells were transfected with varying amounts of pRSV PTMA, the prothymosin ␣ gene, and labeled after 2 days with [ 32 P]orthophosphate for 4 h. Prothymosin ␣ protein was isolated and purified electrophoretically. After staining the gel with Coomassie Blue, prothymosin ␣ was quantified by scanning the prothymosin ␣ band; the gel was dried and exposed to film. Radioactivity in prothymosin ␣ was quantified from the intensity of the signal on film. Each symbol represents a different set. For each experiment, the total intensity of stain or radioactivity for the eight points in the set was summed and averaged; that value was set equal to 100 arbitrary units. Each point is a percentage of the average value for its set. The amount of prothymosin ␣ ranged from 0.08 to 1.6 nmol, and the linear fit gave a residual (R) value of 0.93. B, COS cells were lysed in dimethyl sulfoxide containing [ 3 H]borohydride using either the low (inset) or high specific activity methods, prothymosin ␣ was purified, and the amount of tritium incorporated into proline was determined upon amino acid analysis. The amino acid analysis provided a measure of the corresponding amount of prothymosin ␣. Each point represents a completely independent determination. The linear fit R values are 0.96 for the data obtained at low specific activity shown in the inset and 0.98 for the high specific activity data. specific activity of ATP is reduced by an instantaneous 14 -15fold increase in the intracellular concentration of ATP rather than by continuous exchange of labeled and unlabeled precursors across the plasma membrane throughout the chase. The residual label might also reflect the existence of a subpopulation that failed to be perforated by electroporation or dead cells that ceased normal phosphate turnover on prothymosin ␣. Although all of the label was not fully eliminated during the chase, the results are in accord with stable prothymosin ␣ molecules undergoing phosphorylation and rapid dephosphorylation.
Half-life of [ 32 P]Prothymosin ␣ and RNA in HeLa Cells Chased with Sodium Phosphate-An experimental design in which cells are labeled with inorganic phosphate and chased with ATP requires the use of permeabilized cells. Neither method, however, is ideal; digitonin kills the cells, and intracellular ATP concentrations, which are Ͼ10-fold higher than normal, may affect the very parameters one seeks to measure. To minimize both problems, we again labeled cells with orthophosphate and chased with a high concentration of sodium phosphate (Fig. 5, A and B). The technique makes use of living cells and avoids puncturing the cell membrane. Instead, a chase is initiated with a molecule (inorganic phosphate) that may not be an immediate precursor of phosphoprothymosin ␣. With the revised conditions, the half-life of phosphate on prothymosin ␣ was 75 min, and greater than 90% of the label turned over (Fig. 5B). Even after 24 h of chase, however, ϳ5% of the radioactivity initially incorporated into prothymosin ␣ was retained (Fig. 5A).
The pulse-chase kinetics of RNA were measured in the same cytosolic samples. From the point of view of RNA, the chase with sodium phosphate was not effective, since equivalent amounts of labeled mRNA were observed immediately following the pulse and for 18 h thereafter (data not shown). By 24 h of chase, a diminution in radioactivity in RNA was observed but, on the average, about 60% of the label incorporated into RNA during the 4-h pulse remained. These values serve to illustrate the difficulties usually encountered in chasing large nucleoside triphosphate pools with inorganic phosphate.

Effect of Inhibition of Translation on the Phosphorylation of
Prothymosin ␣-Our experiments indicate that the half-life of the phosphate on prothymosin ␣ (Ͻ1.5 h) is very much shorter than the half-life of the protein moiety, itself (ϳ24 h) (12). However, the experimental design does not allow one to distinguish transient phosphorylation of newly synthesized molecules from rapid reversible phosphorylation of preexisting protein. Toward this end, HeLa cells were either treated with cycloheximide to prevent the accumulation of newly synthesized molecules or incubated in the absence of the inhibitor as a control and pulse-labeled with [ 32 P]inorganic phosphate. Phosphorylation of prothymosin ␣ was not affected by cycloheximide (Fig. 6); the amount of [ 32 P]prothymosin ␣ and total prothymosin ␣ remained unchanged regardless of whether new synthesis of prothymosin ␣ was curtailed. From these data, it can be inferred that both new and old prothymosin ␣ molecules can be phosphorylated. Furthermore, there is no evidence to suggest that newly synthesized prothymosin ␣ molecules undergo preferential phosphorylation. The data are also consistent with the known stability of prothymosin ␣, because the total amount of prothymosin ␣ (stained protein) was essentially unaffected by a short hiatus in protein synthesis (data not shown).
Turnover of Prothymosin ␣'s Phosphates in Resting and Growing NIH3T3 Cells-The stability of the phosphates was determined in NIH3T3 cells. Cells, either quiescent or growing, were pulse-labeled and chased in the presence of phosphate. As shown in Fig. 7, the values differed. The stability of prothymosin ␣'s glutamyl phosphate in growing murine cells was very similar to that measured in growing HeLa cells; a value of 70 min in NIH3T3 cells is not significantly different from 75 min measured in HeLa cells using the same technique. In resting NIH3T3 cells, however, there was a marked change. The halflife was 30 -35 min, approximately half the value found in rapidly growing cells. The data suggest that the stability of the phosphates on prothymosin ␣ can be metabolically regulated.

DISCUSSION
In this paper, we have examined the kinetics of the loss of phosphate from prothymosin ␣ under a variety of conditions. Our assay depends on a proportionality between the amount of glutamyl phosphate on prothymosin ␣ and the amount of stable phosphate retained by the purified protein. We have shown that the amount of stable radioactive phosphate covalently attached to homogeneous prothymosin ␣ is proportional to the amount of prothymosin ␣ protein, despite manipulation of the amount of prothymosin ␣ in the cell by transfection of the gene. We have also shown that the amount of acyl phosphate in prothymosin ␣, measured as [ 3 H]proline after the reaction with [ 3 H]borohydride, is proportional to the amount of prothymosin ␣ protein. It follows that prothymosin ␣'s abundant phosphoglutamate in vivo is proportional to its stable phosphate in vitro; a fixed fraction of the phosphate on glutamate residues in vivo serendipitously transfers preferentially to serine and to threonine at the time of cell lysis, while the remainder is subject to hydrolysis. There is no evidence to suggest that phosphorylation directly on serine or threonine occurs in vivo ( Fig. 1 and Ref. 20).
The assay does have limitations. Many acyl phosphates must compete for few positions (perhaps only one position) near the amino terminus of each protein molecule. It is possible that our determinations distinguish only phosphorylated from unphosphorylated molecules, without consideration for the number of phosphates possessed by each protein molecule. Alternatively, it is possible that the probability of finding a stable phosphate depends on the total number of glutamyl phosphates available on each molecule of prothymosin ␣. According to the latter view, a molecule with few glutamyl phosphates would be unlikely to retain one as a phosphate ester, whereas a richly endowed molecule would have a heightened probability. In neither case is the assay ideal, because the fate of a specific glutamyl phosphate cannot be determined. Nevertheless, like the standard assay for ribonuclease, which responds to loss of trichloroacetic acid precipitable RNA rather than marking each catalytic event, an insensitive assay of either type noted above can still be used to obtain useful kinetic information.
The stability of the glutamyl phosphates initially present on prothymosin ␣ was measured in vivo using cells that were pulse-labeled, and chased using three independent techniques: cells were electroporated in the presence of a high concentration of ATP with the goal of diluting the intracellular [ 32 P]ATP pool; cells were permeabilized with digitonin to allow extracellular ATP in the bathing fluids to equilibrate with the intracellular pool; and cells were incubated in the presence of inor-ganic phosphate. In all cases, the half-life of the glutamyl phosphate on prothymosin ␣ was ϳ80 min, considerably shorter than the 24-h half-life of the protein. In these experiments, the rate of disappearance of radioactive phosphate on prothymosin ␣ was determined by noting the change in the specific activity ([ 32 P]prothymosin ␣/total prothymosin ␣) of the isolated protein. This approach is valid because the decay of prothymosin ␣ itself is negligible, and new synthesis of the protein does not make a significant contribution to the denominator during the time frame required for the chase (12). It should also be noted that stable phosphate is apparently not subject to phosphatases in the lysates; the addition of phosphatase inhibitors had no effect on the recovery of prothymosin ␣'s phosphoserine/phosphothreonine (20).
A measured half-life depends not only on the inherent stability of the labeled molecule in question but also on the rate of loss of label from the precursor pools. When [ 32 P]prothymosin ␣ was chased using two of the conditions noted above (digitonin and electroporation), instantaneous dilution of the label at the commencement of the chase was achieved by violating the cells. However, with inorganic phosphate as the chase vehicle in physiologically maintained cells, the chase also became effective virtually immediately, and the half-life obtained was equivalent. Stated in different words, the values obtained at very short times of phosphate chase, including the zero time point, fell on the same line generated by the data obtained after several hours of chase. There was no indication that the data required correcting for the kinetics of isotope dilution in the labeling pool. Such corrections, when necessary, make calculated half-lives shorter than measured ones by mathematically eliminating a subpopulation that continues to become labeled during an inadequate chase.
The effectiveness of the phosphate chase indicates that the source of prothymosin ␣'s label is a small, rapidly equilibrating pool. This pool, however heretical the idea, does not seem to be the 0.9 mM ATP pool unless a small, isolated compartment of ATP, which readily equilibrates with the extracellular milieu, can be identified in the nucleus. Although ATP is the most common source of phosphate for protein kinases, and casein kinase II has been reported to phosphorylate prothymosin ␣ in vitro with ATP or GTP (28), it is not at all clear how prothy- mosin ␣ is phosphorylated in vivo. We have not identified the immediate precursor pool for acyl phosphorylation and, obviously, have not yet determined the mechanism in vivo by which the phosphate is removed; nor do we understand how the energy stored in glutamyl phosphate bonds is utilized. However, it should be evident that phosphorylation in vitro directly on serine and threonine in defined solutions will not be an aid to understanding acyl phosphorylation in vivo.
The kinetics of the acquisition and loss of 32 P in prothymosin ␣ provide additional insight. We have previously shown that, in synchronized cells, prothymosin ␣ acquires equivalent amounts of label during each interval, suggesting that prothymosin ␣ molecules do not become phosphorylated at specific stages in the cell cycle (12). We show here that when unsynchronized cells are pulse-labeled and chased, virtually all of the label disappeared with the same kinetics, suggesting a single class of molecules. Furthermore, the ability to label prothymosin ␣ was not affected by the presence or absence of cycloheximide, data that imply that both newly synthesized and aged prothymosin ␣ molecules form a single pool, members of which bear phosphate at any instant in time. Taken together, these observations argue for the equivalence of all prothymosin ␣ molecules. Our findings are not consistent with a model in which prothymosin ␣ discharges its phosphate as cells pass a strategic cell cycle check point. Such a model predicts that radioactivity acquired by pulse-labeled, unsynchronized cells would persist in significant amounts until one complete cell cycle had been traversed, a projection at odds with a half-life of ϳ80 min. As a consequence, we believe it is necessary to abandon the idea that prothymosin ␣ plays a role at a specific stage of the cell cycle.
Prothymosin ␣ is required for cell growth and is found in increasing amounts in highly proliferative cells and tissues (12)(13)(14)(15)(16). Here we show that the glutamyl phosphate borne by the protein is unstable and surmise that the continuous turnover of these phosphates is synonymous with prothymosin ␣'s activity. Since prothymosin ␣ has been closely identified with cell growth, we performed a simple experiment to determine whether noncycling cells, also, make use of the phosphoprotein. We compared the half-life of prothymosin ␣'s phosphates in quiescent and growing NIH3T3 cells and found the values to be 30 -35 and 70 min, respectively. Because the chase in both metabolic states appeared to be effective almost immediately, there is no reason to suspect that the difference is an artifact caused by an alteration in the rate of loss of label from the precursor pools; the disparity in stability seems to be genuine. We do not yet know the significance of the increase in phosphate turnover in quiescent cells (whether it reflects greater or lesser use of a meager amount of prothymosin ␣), but we recognize that the half-life in both cases is considerably shorter than that expected from the known chemical stability of acyl phosphate-bearing model compounds (40) and considerably longer than the in vitro half-lives of the aspartyl phosphates found in two component systems (23)(24)(25). All told, it is difficult to escape the view that prothymosin ␣ functions in quiescent cells, that its activity can be regulated (a subject we examine further), 3 and that the phosphoprotein is required not just for cell growth, but for cell survival as well.