Evidence for Isopentenyladenine Modification on a Cell Cycle-regulated Protein*

We have prepared antibodies that recognize isopentenyladenosine (PA), a modified nucleoside derived from mevalonic acid (MVA). In immunoblot assays, affinity-purified anti-i6A antibodies specifically bound to a 26-kDa protein (i6A26) in Chinese hamster ovary cells. Anti-ieA recognition of PA26 was blocked with i“A but not adenosine or isopentenol. Employing im- munoblot analysis we have quantitated the level of i6A26 in cells expressing various rates of DNA synthesis. The cellular content of PA26 was reduced 4-fold in quiescent cells cultured in the absence of serum. When serum-deprived cells were stimulated to enter the cell cycle, the amount of PA26 increased in the cells during the G1 phase. However, when synchronized cells were stimulated with serum-containing me- dium in the presence of mevinolin (an inhibitor of cellular MVA synthesis), we observed impaired expression of PA26 and delayed onset of S phase DNA synthesis. Mevinolin addition to asynchronously grow- ing cells resulted in low rates of cellular DNA synthesis and suppressed levels of PA26 which were reversed by coincubation with MVA. The ability of MVA to restore DNA synthesis and the cellular content of PA26 in mevinolin-treated cells showed similar MVA concentration and time dependences. Regenerating liver tis- sue also exhibited elevated levels of i‘A26. Thus, the expression of PA26 correlates with cellular prolifera- tion and growth. We speculate that ieA26 contains isopentenyladenine and mediates isoprenoid

We have prepared antibodies that recognize isopentenyladenosine (PA), a modified nucleoside derived from mevalonic acid (MVA). In immunoblot assays, affinity-purified anti-i6A antibodies specifically bound to a 26-kDa protein (i6A26) in Chinese hamster ovary cells. Anti-ieA recognition of PA26 was blocked with i"A but not adenosine or isopentenol. Employing immunoblot analysis we have quantitated the level of i6A26 in cells expressing various rates of DNA synthesis. The cellular content of PA26 was reduced 4-fold in quiescent cells cultured in the absence of serum. When serum-deprived cells were stimulated to enter the cell cycle, the amount of PA26 increased in the cells during the G1 phase. However, when synchronized cells were stimulated with serum-containing medium in the presence of mevinolin (an inhibitor of cellular MVA synthesis), we observed impaired G1 expression of PA26 and delayed onset of S phase DNA synthesis. Mevinolin addition to asynchronously growing cells resulted in low rates of cellular DNA synthesis and suppressed levels of PA26 which were reversed by coincubation with MVA. The ability of MVA to restore DNA synthesis and the cellular content of PA26 in mevinolin-treated cells showed similar MVA concentration and time dependences. Regenerating liver tissue also exhibited elevated levels of i'A26. Thus, the expression of PA26 correlates with cellular proliferation and growth. We speculate that ieA26 contains isopentenyladenine moieties and mediates isoprenoid regulation of DNA synthesis. Isopentenyladenylated proteins may also function in cytokinin regulation of proliferation and differentiation in plants.
Isoprenoids such as cholesterol, dolichol, and ubiquinone are essential components in a wide variety of mammalian cellular processes. MVA' is the central precursor for all cellular isoprenoids, and adequate levels of MVA are required for sustained cell growth (1). Employing both pharmacological (2-8) and genetic (9) manipulations, many reports have documented that MVA is also acutely required for DNA replication. No clear evidence has emerged implicating either cholesterol, dolichol, or ubiquinone in acutely regulating DNA synthesis, and the identity of the cellular isoprenoid required for DNA replication is unknown.
Recently much attention has focused upon isoprenylated proteins, i.e. macromolecules that are radiolabeled when cells are incubated with radioactive MVA (6,(9)(10)(11)(12)(13)(14)(15). Consistent with a potential regulatory role in cell proliferation, some isoprenylated proteins are localized in nuclear fractions (13,16). Also, Sinensky and Loge1 (9) observed a correlation between the decline in DNA synthesis in cells deprived of MVA and the rate of turnover of MVA-derived radioactivity in certain isoprenylated proteins.
Cellular ras and nuclear lamins are two types of cellular proteins which have been shown to contain isoprenyl groups (15). In ras, the carboxyl-terminal cysteine contains a farnesyl moiety (Fig. 1B) covalently attached by a thioether linkage (17,18). Farnesylation is required for proteolytic processing, palmitoylation, and tight binding of ras to cellular membranes (17,(19)(20)(21). In the absence of farnesylation, oncogenic forms of ras do not transform cultured cells (17)(18)(19)(20). Lamin B also contains farnesyl residues covalently attached to cysteine (22)(23)(24)(25). Farnesylation may promote binding of lamin B to the inner nuclear membrane and contribute to the integrity of the nuclear matrix. Also, prelamin A is post-translationally modified by an isoprenyl moiety (26). Although the exact chemical nature of the modification has not been elucidated, prelamin A contains an amino acid motif, identical to that in ras and lamin B, which is required for farnesylation (25).
Isoprenylation of prelamin A is required for proteolytic processing of the precursor to mature lamin A (26). Geranylgeranyl ( Fig. 1B) is another isoprenyl modification occurring in mammalian cell proteins (27,28). Given the presumed function of ras in regulating cell growth and the structural role of lamins in nuclear architecture, speculation has emerged that either of these isoprenylated proteins may regulate cell proliferation.
Isopentenyladenosine (PA) is a modified base found in tRNA from a wide variety of eukaryotic and prokaryotic cells (29). The i6A moiety (Fig. 1A) is derived from 3-methyl-2buten-1-yl pyrophosphate, an intermediate in cellular isoprenoid biosynthesis. Isopentenyladenosine is found adjacent to the 3' end of the anticodon of tRNAs. This modified nucleoside appears only in tRNAs that bind to codons containing uridine as the first base. In bacteria, isopentenylated tRNAs have been implicated in the regulation of aromatic acid uptake (30) and aerobiosis (31). The precise role of the modified nucleotide in tRNA metabolism of eukaryotes is unknown.
Free, non-tRNA-associated i6A has been observed in the yeast strains Saccharomyces cerevisiae and Schizosaccharo- myces pombe (32). The cellular level of free i6A was not decreased in defective yeast strains possessing mutations that result in severely reduced amounts of isopentenylated tRNAs. This latter result indicates that free i6A may be derived via a synthetic pathway independent of isopentenylated tRNA degradation. In plants, i6A and related compounds are collectively referred to as cytokinins (33). Cytokinins are an important class of plant growth hormones which regulate cell division, gene expression, and differentiation in various tissues (34).
If an isoprenylated protein regulates DNA synthesis then we might expect that the cellular content of the modified protein may correlate with rates of DNA synthesis, i.e. the level of the isoprenylated protein would be low in cells expressing reduced rates of DNA replication resulting from MVA deprivation. Antibodies recognizing an isoprenylated protein could be used to quantitate the level of the modified protein in immunoblots performed on cell extracts. Within the context of MVA metabolism, we felt that the most meaningful quantitation would be achieved by using antibodies that are specific for the isoprenyl moiety in the modified protein. Knowing that adenylation and ADP ribosylation are covalent modifications occurring in mammalian cells (35) we speculated that isopentenyladenine may be present in a subset of isoprenylated proteins. We have prepared and characterized antibodies against i6A. In immunoblots, the anti-i6A antibodies specifically bind to a 26-kDa protein (termed i'A26) present in growing CHO cells and regenerating rat liver. We have quantitated the cellular content of PA26 in cells expressing various rates of DNA synthesis.
Preparation of L*A Conjugated to BSA, Immunization of Rabbits, and Affinity Purification of Anti-iGA Antibodies-Antibodies recognizing i6A were prepared by immunizing rabbits with i'A-BSA and affinity purifying the antisera on i6A-Sepharose. i'A-BSA was prepared at room temperature essentially as described by Erlanger and Beiser (36). Briefly, 100 mg of i6A was stirred for 20 min in 5 mi of 0.1 M sodium iodate. Excess iodate was decomposed by the addition of 0.3 ml of 1 M ethylene glycol. After 5 min the reaction mixture was added to a stirring aqueous solution (10 ml) of 280 mg of BSA (fraction V from Pentex) adjusted to pH 9.3 with 5% (w/v) potassium carbonate. The reaction was stirred for 45 min while maintaining pH 9.0-9.5 with 5% (w/v) potassium carbonate. Next, 10 ml of 15 mg/ml sodium borohydride was added and the reaction mixture incubated overnight. Five ml of 1.0 M formic acid was added followed 1 h later by adjustment to pH 8.5 with 2 ml of 10% ammonium hydroxide. Finally, the solution was dialyzed for 24 h against 6 liters of deionized water. The preparation yielded 144 mg of protein in 39 ml. The i6A concentration (as measured by absorbance a t 260 nm) was 0.57 mM.
As proposed by Erlanger and Beiser (36), periodate oxidation of i6A would cleave the ribose ring structure between carbons 2 and 3. In alkaline solution, the resulting dialdehyde would react with a primary amine on BSA, forming a cyclic hemialdal. Subsequent borohydride reduction produces a stable morpholine derivative in which a new six-member ring would form consisting of the original five atoms from the ribose ring plus the amino nitrogen attached to BSA. Thus, the attachment of i6A to the protein is via the ribose moiety and should not alter the atomic structure of isopentenyladenine.
Two New Zealand White rabbits were each immunized with 2 mg of PA-BSA emulsified with Freund's complete adjuvant on day 0. The rabbits received subsequent injections of 2 mg of i'A-BSA diluted 1:1 with incomplete Freund's adjuvant on days 8 and 21. On day 53 the rabbits were killed, and immune serum was isolated after clotting. The serum was stored at -70 "C.
Coupling of i6A to AH-Sepharose 4B (Pharmacia LKB Biotechnology Inc.) was exactly as described (37). For affinity purification, one aliquot (8 ml) of immune serum was diluted to 30 ml with buffer A and precleared by incubation with 10 ml of w-aminooctyl agarose (2 h, room temperature) followed by centrifugation (1,000 X g, 10 min, room temperature). All subsequent procedures were performed a t 4 "C. The precleared supernatant was mixed with 5 ml of PA-Sepharose and agitated gently overnight. Next the slurry was percolated through a 10 X 100-mm plastic column. The column was washed with buffer A (75-100 ml) until absorbance at 280 nm was less than 0.05. Two fractions of anti-i6A antibodies were eluted from the column. Fraction 1 was eluted with 20 ml of buffer A containing 500 mM instead of 140 mM NaC1. Subsequently, fraction 2 was eluted with 20 ml of 10% (v/v) pyridine in buffer A. Both fractions were dialyzed against 6 liters of buffer A and stored at -70 "C.
Dot Blotting-Serial dilutions of i'A-BSA and BSA were made in 1 Fg/ml ovalbumin, and aliquots (200 pl) were applied to nitrocellulose (Schleicher & Schuell) in a dot blotting manifold. After applying vacuum to deposit the proteins onto nitrocellulose, the sheet was cut vertically between the rows of dots producing replicate strips of nitrocellulose containing various amounts of i'A-BSA and BSA. The nitrocellulose strips were washed with buffer C and incubated with buffer D as described in the immunoblotting procedure below. Replicate nitrocellulose strips were incubated with sera or affinity-puri-fied anti-i'A antibodies in 3 ml of buffer D on a rotating platform for 3.5 h a t room temperature. The strips were then pooled together and washed as described in the immunoblotting procedure below. Next the nitrocellulose strips were incubated with a 1:2,000 dilution of alkaline phosphatase-linked goat anti-rabbit antiserum for 2 h a t room temperature on a rotating platform and washed again. The alkaline phosphatase color reaction was performed in 100 mM sodium bicarbonate, 1 mM magnesium chloride, pH 9.8, using the substrates nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate, as recommended by the supplier (Promega Biotec, Madison, WI).
Measuring Cell Protein and DNA Synthesis-As indicated in the f i p~r e legends, monolayers were labeled with ["Hlthymidine in a 37 "C COY incubator. The cells were harvested a t 4 "C by washing the monolayers four times with 10 ml of Tris/saline: once quickly, twice with agitation for 7.5 min, and once quickly. Cells were scraped from the dish with a rubber policeman in Tris/saline (1.5 ml/dish), vortexed gently, and divided into three portions. Thirteen hundred microliters of the cell suspension was pelleted in microcentrifuge tubes (15,000 rpm, 5 min, 4 "C). These cell pellets were stored a t -70 "C and processed for immunoblotting as described below. Duplicate aliquots (120 pl) of the scraped cell suspension were incubated (15 min, 4 "C) with 50 pl of 0.5% (w/v) sodium deoxycholate and precipitated with 1.5 ml of 15% (w/v) trichloroacetic acid (15 min, 4 "C). Precipitates were centrifuged (2,000 X g, 15 min, 4 "C), washed with 1.5 ml of 5% (w/v) trichloroacetic acid, and recentrifuged. The washed pellets were dissolved in 1.5 ml of Lowry reagent (2% (w/v) sodium carbonate, 0.02% (w/v) sodium tartrate, and 0.01% (w/v) cupric sulfate in 0.1 N NaOH). After 10 min a t room temperature, 600 pl of the dissolved cellular macromolecules was transferred to a scintillation vial containing 10 ml of Liquiscint (National Diagnostics, Somerville, NJ). After adding 10 pl of glacial acetic acid, tritium radioactivity was determined by liquid scintillation counting. Immunoblotting-Cell pellets were resuspended in gel loading buffer a t a protein concentration ranging between 0.75 and 1.5 mg/ ml. The cell suspensions were vortexed vigorously, incubated a t room temperature for 15 min, sonicated in a bath sonicator for 10 s, heated at 90 "C for 4 min, cooled to room temperature, and centrifuged (15,000 X g, 5 min, room temperature). Aliquots of equivalent cell protein were subjected to SDS-PAGE as described by Laemmli (39), using a 5% polyacrylamide stacking gel and a 10.5% polyacrylamide separating gel. The resolved proteins were transferred electrophoretically to nitrocellulose in 10 mM sodium bicarbonate, 3 mM sodium carbonate, 0.01% (w/v) SDS, 20% (v/v) methanol. The transfer was performed overnight at 4 "C in a Bio-Rad Trans-Blot cell a t 50-60 volts. After transfer, the nitrocellulose was washed for 30 min in buffer C and incubated for 2 h in buffer D. Next, the nitrocellulose blots were incubated overnight at 4 "C with 75 pl/ml each of affinitypurified fractions 1 and 2 in buffer E. Unbound antibodies were removed from the nitrocellulose by four successive washings (5 min, room temperature) with buffer C. Next the nitrocellulose was incubated with ""I-protein A (1,000-2,000 cpm/pl) in buffer E containing 6 mM sodium iodide and 0.02% (w/v) sodium azide for 2 h a t room temperature. The blot was washed five times over a period of 30 min with buffer C, air dried, covered with Saran Wrap, and exposed to Kodak X-Omat AR film a t -70 "C using a Cronex intensifying screen (Du Pont).
Competition for Immunoblotting-Cells were cultured as described above. On day 2 the monolayers were refed 6 ml of medium 1. On day 3 the cells were washed and scraped as described above. The cell suspensions from 10 100-mm dishes were pooled, and aliquots were taken for cellular protein determination. The remaining cells were centrifuged (2,000 X g, 15 min, 4 "C) and stored at -70 "C. The cell pellet was thawed in the presence of gel loading buffer (1.0 mg/ml protein concentration), sonicated, heated, and microcentrifuged as described above. Two ml of the solubilized cells was electrophoresed as described above except that the stacking gel was formed in the absence of a comb. This allowed for the solubilized cell sample to disperse evenly over the entire width of the gel. The entire gel was transferred to nitrocellulose as described above. After transfer, the nitrocellulose blot was cut vertically into 6-mm-wide strips. This procedure produces replicate strips of nitrocellulose containing equivalent amounts of resolved cellular proteins.
Preparation of Control and Regenerating Rat Liver Supernatants-On day 0 male Sprague-Dawley rats (100-150 g) were anesthetized, and an approximately 70% hepatectomy was performed. The liver tissue removed at this time (designated "control") was stored at -70 "C. During the recovery phase the rats were allowed free access to food and water. On day 2 the rats were killed, and liver tissue (designated "regenerating") was removed and frozen at -70 'C. Aliquots of liver were thawed on ice in the presence of 14 volumes of buffer B. The tissue was Dounce homogenized using 20 strokes with a loose pestle followed by 20 strokes with a tight fitting pestle. The homogenate was centrifuged (1,000 X g, 10 min, 4 "C) and the supernatant saved. The pellet was resuspended in 7 volumes of buffer B by Dounce homogenization 20 times with a tight fitting pestle. After centrifugation of the homogenate (1,000 X g, 10 min, 4 "C), the resulting supernatant was combined with the initial supernatant. The pooled supernatants were centrifuged (100,000 X g, 60 min, 4 "C) and stored a t -20 "C.

Presence of Anti-PA Antibodies in Immune Serum and
Isolation of Affinity-purified Antibodies-We prepared antibodies directed against i' A by immunizing rabbits with i6A linked to bovine serum albumin (PA-BSA). Affinity-purified anti-i'A antibodies were obtained by fractionation of the immune serum with i6A linked to Sepharose (iGA-Sepharose). Using a dot blot assay, we screened for the presence of anti-iGA antibodies in the sera and affinity-purified fractions. The dot blots were prepared by spotting various amounts of either bovine serum albumin (BSA) or i'A-BSA onto nitrocellulose in multiple vertical rows. Replicate strips were obtained by vertically cutting the nitrocellulose between the rows of dots. These replicate strips were used to evaluate the presence of anti-i6A antibodies in test solutions. Antibody binding to the immobilized proteins was detected with alkaline phosphatase conjugated to goat anti-rabbit immunoglobulins and subsequent color reaction. By analyzing the degree of the color reaction formed over the various amounts of i6A and i'A-BSA, we evaluated the relative concentration of antibodies in the test solutions. Fig. 2 shows that antibodies binding to either BSA or i'A-BSA were not detected in preimmune rabbit serum (strip I). However, immune serum contained antibodies that bound to both proteins (strips 2-4). Note that antibody recognition can be detected on spots containing 4 ng of i'A-BSA whereas a t least 64 ng of BSA was required to visualize antibody binding. This result indicates that the immune rabbit serum had a higher titer of antibodies directed against PA-BSA relative to BSA. Purification of the immune serum with i'A-Sepharose resulted in a substantial reduction in the concentration of anti-BSA binding antibodies (strips 5-7). Also, the concentration of anti-i"A-BSA antibodies did change appreciably as a result of the affinity purification procedure. This observation implies that the affinity-purified fractions contain antibodies that recognize i"A specifically.
Immunoblotting of CHO Cell Extracts with Affinity-purified Anti-?A Antibodies-Unfractionated CHO cells were analyzed by immunoblotting for macromolecules that bind affinity-purified anti-i'A antibodies (Fig. 3). As shown in lane 1, CHO cells growing in serum-containing medium contain two macromolecules (approximate molecular sizes, 56 and 26 kDa) that bind I2'I-protein A after incubation with anti-i'A antibodies. No autoradiographic bands were observed when affinity-purified anti-i6A antibodies were omitted from the immunoblotting procedure (data not shown). Also, when unfractionated immune serum was used in the immunoblotting procedure, we observed antibody binding to both the 26-and 56-kDa macromolecules. Preimmune serum gave no antibody recognition at those molecular masses (data not shown). These results indicate that antibodies specifically recognizing these two entities were produced by the rabbit in response to immunization with i6A-BSA. Although two of two immunized rabbits produced antibodies specifically recognizing PA-BSA, affinity-purified antibodies recognizing the 56-and 26-kDa macromolecules were present in only one rabbit serum (data not shown). Antibodies recognizing the 26-kDa macromole-  1 (lanes  1-3) or medium 2 (medium 1 containing 0.2% instead of 5% (v/v) newborn calf serum) (lanes 4 and 5 ) . Indicated additions of 20 p~ mevinolin (lanes 2 and 3 ) and 2 m M MVA (lanes 3 and 5 ) were made, and monolayers were harvested after either 20 h (lanes 1-31 or 31 h  (lanes 4 and 5 ) . Aliquots of 75 pg of total cell protein were subjected t.o SDS-PAGE, and immunoblotting was performed as described under "Experimental Procedures." Immunoblots were exposed to film for 3 days. Molecular weight standards are shown. cule were observed exclusively in the 0.5 M NaCl eluate of the affinity purification (see "Experimental Procedures"). Antibodies recognizing the 56-kDa entity were present in both affinity-purified fractions; however, the pyridine-eluted fraction contained the highest titer (data not shown).
What is the chemical nature of these immunoreactive macromolecules? Several data indicate that the 26-kDa macromolecule is a protein. Knowing that isopentenyl tRNAs are cellular macromolecules that could bind anti-i6A antibodies, we subjected purified rat liver tRNA to immunoblotting analysis. When aliquots greater than 100 pg of liver tRNA were analyzed we observed a very diffuse autoradiographic image in the molecular mass range between 68 and 97 kDa (data not shown). In subcellular fractionation studies, the 26-kDa macromolecule was found to localize exclusively with the 100,000 x g supernatant, not with cellular DNA which sediments at low centrifugation forces (data not shown). Also, when the 100,000 X g supernatant was pretreated with trypsin and subsequently analyzed by immunoblotting, the 26-kDa immunoreactive macromolecule disappeared. Trypsin digestion of the 26-kDa macromolecule was prevented with soybean trypsin inhibitor (data not shown).
Consequently, we have concluded that the 26-kDa macromolecule is a protein, and we will refer to it as i'A26.
Using immunoblotting analysis we have determined the relative concentration of the antibody-binding entities in CHO cells cultured in various media (Fig. 3). When CHO cells were incubated for 20 h with mevinolin, a potent inhibitor of MVA biosynthesis, the cellular content of i'A26 declined dramatically (lane 2). Incubation of mevinolin-treated CHO cells with MVA restored the cellular content of i'A26 to the uninhibited level (lane 3 ) . These observations indicate that the mevinolin-induced suppression of i'A26 is specifically due to inhibition of cellular MVA synthesis. Also, CHO cells incubated for 31 h in serum-depleted medium have a lower level of i"A26 (lane 4) which was not reversed when serumstarved cells were incubated with exogenous MVA (lane 5). This indicates that reduced availability of MVA was not mediating the serum-deprived suppression. By densitometric scanning of multiple autoradiographic images in several different experiments we determined that incubation in either the presence of mevinolin or in the absence of serum, separately, resulted in an average 4-fold decline in the cellular content of i'A26. Fig. 3 also shows that the levels of the 56-kDa antibodybinding macromolecule did not vary in CHO cells preincubated in the various media. The invariant level of this macromolecule serves as a convenient control for equivalent loading of the gel and uniform immunoblotting.
Competition for Antibody Binding to iGA26"Based on our observations that the PA26 binds antibodies present in i'A-BSA immune serum as well as fractions purified with PA-Sepharose, we speculate that the 26-kDa protein contains isopentenyladenine moieties that are mediating the binding of anti-i6A antibodies. We tested for the presence of PAspecific binding of antibodies by performing the immunoblotting incubations in the presence of potential competing agents (Fig. 4). Affinity-purified antibodies were preincubated with various amounts of i6A, adenosine, or 3-methyl-2-buten-1-01 (an alcohol derivative of the isoprenoid moiety in ?A). Then the antibodies were used to immunoblot replicate strips of nitrocellulose containing resolved CHO cell proteins. To avoid the possibility that the competitors may differentially bind to BSA and nonfat dry milk, thus lowering the effective concentration of the competitors, the preincubation and primary antibody binding reactions were performed in the absence of Competition for immunoblotting. Replicate nitrocellulose strips containing equivalent amounts of CHO cell proteins resolved on SDS-PAGE were prepared as described under "Experimental Procedures." Sixty p1 of fraction 1 affinity-purified antibodies were preincubated overnight at room temperature in 2 ml of buffer C containing 100 p1 of dimethyl sulfoxide and the indicated concentration of i"A (strips 2-4), adenosine (strips 5-7), or 3-methyl-2-buten-1-01 (strips [8][9][10]. Strips were added to the solutions and agitated for 2 h at room temperature. The replicate strips were pooled together, washed, incubated with goat anti-rabbit antibodies linked to alkaline phosphatase, and visualized as described under "Experimental Procedures" (dot blotting procedure). The locations of the 26-and 56-kDa macromolecules are shown. The increased number of immunoreactive bands in this experiment (compared with other blots) is due to the omission of BSA and nonfat dry milk during the incubations with the competitors and primary antibodies. These agents were omitted to avoid the possibility that the competitors may differentially bind to RSA and nonfat dry milk, thus lowering their effective concentrations.
these blocking agents. The lack of inert protein resulted in nonspecific antibody binding to an increased number of macromolecules on the nitrocellulose strips (compare Figs. 3 and   4).
Antibody recognition of i"A26 was partially inhibited by 6 mM i"A ( Fig. 4, strip 2). Further inhibition of antibody binding was observed with 12 and 20 mM i"A (strips 3 and 4 ) . Neither adenosine (strips 5-7) nor 3-methyl-2-buten-1-01 (strips 8-IO), a t concentrations up to 30 mM, affected antibody recognition of i"A26. Twenty mM adenosine 5'-diphosphoribose did not compete for antibody binding to PA26 (data not shown). Competition with trans-farnesol was difficult to assess because a t concentrations greater than 1 mM, transfarnesol was not soluble in the assay buffer. Nevertheless, antibody competition was not observed with trans-farnesol (data not shown). These results show that an i"A binding site in the antibodies is mediating the interaction between the immunoglobin and i"A26 and imply that the 26-kDa protein contains isopentenyladenine moieties.
Contrary to our observations on the 26-kDa anti-PAA-binding protein, we did not observe i"A competition for antibody recognition of the 56-kDa macromolecule. We discuss possible explanations for the lack of competition under "Discussion." Correlation between Time Courses of Meoinolin Inhibition and Meualonate Restoration of Cellular DNA Synthesis with the Cellular Content of i"A26"CHO cells cultured in the absence of serum growth factors exhibit low rates of DNA synthesis and are synchronized in a quiescent stage of the cell cycle. The results presented in Fig. 3 show that the reduced cellular content of i"A26 is apparent in serum-deprived as well as mevinolin-treated CHO cells. Is mevinolin suppression of the iGA26 also associated with low rates of DNA replication? Two reports have documented that mevinolin treatment of growing CHO cells led to a decline in DNA synthesis and cell cycle arrest which could be restored by the addition of MVA (8,9). Fig. 5 shows the time courses of mevinolin suppression (panel A ) and subsequent MVA restoration (panel B ) of DNA replication along with the cellular content of i"A26 (panel C). Mevinolin elicited a time-dependent decline in cellular DNA synthesis in CHO cells cultured in the presence of serum. After a 21-h incubation with mevinolin, DNA synthesis was reduced by 88%. When MVA was added to cells preincubated with mevinolin for 21 h, we did not observe an immediate restoration of cellular DNA synthesis (panel R ) . In fact, rates of DNA synthesis continued to decline for 6 h after the addition of MVA. However, by 12 h the rate of cellular DNA replication had increased dramatically, and by 15 h DNA synthesis exceeded that of untreated cells. These observations corroborate the earlier reports (8,9) and indicate that preincubation with mevinolin induced a nonproliferative state in CHO cells. Addition of MVA to mevinolin-treated cells releases the cells from their quiescent state and induced a synchronous progression through the cell cycle. The cells reached S phase a t approximately 15 h after MVA addition. Fig. 5 also shows that mevinolin incubation caused a timedependent decline in the cellular level of i"A26 (panel C, lanes 1-6). We note that the time courses of mevinolin inhibition of DNA synthesis (panel A ) and suppression of i"A26 (panel C, lanes 1-6) are very similar. When MVA was added to cells preincubated with mevinolin for 21 h, i"A26 continued to decline for an additional 6 h (lanes 7 and 8). By 9 h after MVA addition the cellular content of the i"A26 had rebounded and continued to increase during the remainder of the time course (lanes 9-12). Thus, the time courses of MVA restoration of DNA synthesis (panel B) and cellular content of i"A26 In this experiment, mevinolin suppressed DNA synthesis 85%. Addition of MVA to the culture medium reversed the mevinolin inhibition of DNA synthesis in a concentrationdependent manner with full restoration at 300 p~ MVA. Greater than 300 p~ MVA did not stimulate cellular DNA synthesis further. Fig. 6B shows the result of immunoblotting with affinitypurified anti-i"A antibodies on the same cells pretreated with mevinolin and various concentrations of MVA. Low concentrations of MVA partially prevented mevinolin suppression of the cellular i"A26. Full restoration occurred a t 300 pM MVA. Greater than 300 p~ MVA did not induce increased levels of i"A26 in CHO cells.
Mevinolin Inhibition of S Phase DNA Synthesis and Impaired Expression of FA26 i n Synchronized CHO Cells-As noted above, CHO cells cultured in serum-deprived medium have low levels of i"A26 (Fig. 3, lane 4 ) which are not restored to normal by the addition of MVA (Fig. 3, lane 5). This result implies that cells cultured in low concentrations of serum are not compromised with respect to MVA. We speculate that the low cellular content of the i'A26 may be related to the quiescent state (Go) of CHO cells cultured in low serum medium. When quiescent cells are refed medium containing a high concentration of serum, the cells enter the cell cycle in synchrony. The synchronous progression of the cells to S phase gether. C, lunes 1-6 represents cells pretreated with mevinolin for the indicated times (see punel A above). Lanes [7][8][9][10][11][12] represent cells pretreated with mevinolin for 21 h and then incubated with MVA for the indicated times (see panel B above). Sixty-pg aliquots of total cell protein were subjected to SDS-PAGE and immunoblotting as described under "Experimental Procedures." The immunoblot was exposed to film for 15 h. The locations of the 26-and 56-kDa macromolecules are shown. of the cell cycle can be monitored by pulse labeling cells with ["Hlthymidine and quantitating DNA synthesis. We were interested in determining a t which stage in the cell cycle i6A26 appears when quiescent cells are stimulated. Also, we attempted to correlate the appearance of i'A26 with S phase DNA synthesis (Fig. 7).
CHO cells were preincubated in low serum medium and subsequently refed high serum medium in the absence and presence of mevinolin. In panel A , the cells' progression to S phase was determined by measuring the rate of cellular DNA synthesis. Increased rates of DNA synthesis were apparent in cells 9 h after the addition of serum-containing medium, and peak levels of replication were observed in the cells at the 15h time point. When quiescent CHO cells were stimulated with serum-supplemented medium plus mevinolin, S phase DNA synthesis was dramatically inhibited.
Panel B shows the cellular levels of i'A26 during cell cycle progression. In the absence of mevinolin (lanes 2-6) increased levels of the i6A26 were apparent in cells within 6 h after the addition of serum. The cellular content of i'A26 rose further with increasing time in the cell cycle. When cells were stimulated with serum in the presence of mevinolin (lanes 7-11 ), the amount of i'A26 remained low. It is only after 15 h of serum stimulation in the presence of mevinolin that we observed an increase in the cellular content of this protein.
Thus, S phase DNA synthesis is preceded by an increase in the cellular level of i'A26. Also, mevinolin inhibition of S phase DNA synthesis correlates with the lack of induction of i"A26.
Elevated Expression of PA26 in Regenerating Rat Liver-Normal rat liver exhibits low rates of DNA synthesis and cell division. However, partial hepatectomy dramatically stimulates the remaining liver cells to proliferate (40). By 48 h posthepatectomy, the liver mass has regenerated to its normal size. By comparing unstimulated versus regenerating liver, cell proliferation can be studied in an intact tissue. We sought to determine if i'A26 was present in liver and whether the expression of this macromolecule correlates with liver cell proliferation. Immunoblots were performed on 100,000 X g supernatants from unstimulated (lanes 2 and 4 ) and regenerating (lanes 3  and 5 ) liver obtained in two separate experiments (Fig. 8). A sample of unfractionated growing CHO cells (lane 1 ) was included in this analysis. We observed that regenerating liver supernatant has several immunoreactive bands. Quantitatively, antibody recognition is greatest for macromolecules migrating with apparent molecular sizes of 56, 39, and 26 kDa. Although the content of the 56-and 39-kDa macromolecules was unchanged in the unstimulated (control) liver samples, the expression of the i'A26 was greatly decreased.

DISCUSSION
A major focus of these studies is the identification of a protein whose expression in cells correlated with cell proliferation. Five experiments substantiate the relationship between the cellular content of i'A26 and cellular DNA synthesis. First, mevinolin induced parallel declines in the cellular content of i'A26 and rates of DNA synthesis (Fig. 5). Second, the time course of MVA restoration of mevinolin-inhibited DNA synthesis displayed a prominent lag phase. A similar  Aliquots (i5 pg) of total cell protein were subjected to SDS-PAGE and immunoblotted as described under "Experimental Procedures." Immunohlots were exposed to film for 14 h. Molecular weight standards are shown. lag was also observed in the expression of i'A26 (Fig. 5). Third, a strong correlation exists between the concentrations of MVA which prevent both mevinolin-inhibited DNA synthesis and mevinolin suppression of i"A26 (Fig. 6). Fourth, during synchronous progression of cells in the cell cycle, increased expression of i"A26 occurred in the G I stage of the cell cycle, preceding the cells entry into the DNA replicative phase (Fig. 7). Synchronized cells incubated with mevinolin, which impairs their transit to S phase, showed delayed expression of i"A26 (Fig. 7). Fifth, proliferating liver tissue contains elevated amounts of i"A26 when compared with quiescent liver (Fig. 8).
This paper describes two types of experimental evidence indicating that i"A26 contains isopentenyladenine: immunologic and metabolic. First, antibodies that recognize i"A26 are not present in the preimmune serum but are found in serum of the rabbit immunized with isopentenyladenine linked to BSA. Furthermore, the specific antibodies that recognize i"A26 also bind to isopentenyladenine linked to Sepharose. The competition experiment (Fig. 4) verified that an isopentenyladenosine-specific binding site on the antibodies is directly responsible for antibody recognition of i"A26 and suggests that the 26-kDa protein contains isopentenyladenine.  lanes 1-6) and presence (lanes [7][8][9][10][11] of mevinolin and harvested at the indicated times. Seventy-five-pg aliquots of total cell protein were subjected to SDS-PAGE and immunoblotting as described under "Experimental Procedures." The immunohlot was exposed to film for 28 h. The locations of the 26-and 56-kDa macromolecules are shown. Second, isopentenyladenine is derived metabolically from MVA. Therefore, the synthesis of this modified base is suppressed in cells incubated with mevinolin. If an isopentenyladenine-containing protein was degraded in cells or the modification was removed, then the steady-state level of isopentenyladenine-containing protein would decline in mevinolin-treated cells. Indeed, we show (Figs. 3, 5, and 6) that mevinolin suppresses the cellular content of i"A26, and the mevinolin effect is reversed with MVA. Incubating cells in the absence of serum also suppresses the level of i"A26, and the modified protein reappears when the cells are refed serumcontaining medium. Mevinolin (MVA deprivation) blocks the normal appearance of i"A26 in the synchronized cells (Fig. 7 ) . Since the anti-i"A antibodies are recognizing a modification on the 26-kDa protein, we cannot distinguish whether the regulation of i"A26 is due to changes in the amount of the protein or in the proportion of the protein modified by isopentenyladenine.
At present we have been unsuccessful in attempting to immunoprecipitate i"A26 from cells radiolabeled with either .000 X g supernatants of control and regenerating liver. Supernatants from control ( C ) and regenerating ( R ) liver were prepared as described under "Experimental Procedures." Regenerating liver tissue was taken 36 and 42 h after partial hepatectomy for experiments 1 and 2, respectively. Seventy-five-pg aliquots of each supernatant were diluted with an equivalent volume of SDS-gel loading buffer and processed for SDS-PAGE as described under "Experimental Procedures." Also, a 30-pg aliquot of total cell protein from CHO cells incubated in medium 1 was analyzed at the same time (lune 1 ). Immunoblotting was performed as described under "Experimental Procedures." Immunoblots were exposed to film for 30 h. Molecular weight standards are shown.
["Hlmevalonate, ["Hladenosine, ['Hlleucine, or Tran"S-label. Therefore, we are not able to confirm that i"A26 contains isopentenyladenine. Nevertheless, we have performed experiments comparing the electrophoretic mobility of i"A26 with macromolecules radiolabeled in cells incubated with ["Hlmevalonate and ["Hladenosine. By immunoblotting CHO cell extracts resolved on isoelectric focusing/SDS-PAGE two-dimensional gels, we have determined that i"A26 migrates with an isoelectric point of 5.3.' Previously published data (16,54,55) as well as our own' two-dimensional gel analysis of extracts from ["Hlmevalonate-labeled cells reveal a cluster of radiolabeled spots in the 26-kDa/pl 5.3 region of two-dimensional gels. Also, we have incubated CHO cells with ['HI adenosine and observed a radiolabeled band comigrating with i"A26 on single dimension SDS-PAGE.' Colocalization of i"A26 with both [''Hlrnevalonate-and ["Hladenosine-labeled proteins supports our hypothesis that i"A26 is modified by isopentenyladenine.
The affinity-purified antibodies used in these studies recognized several macromolecules in addition to i'A26. In immunoblots of CHO cells, the antibodies bound to a 56-kDa macromolecule. The cellular content of this macromolecule did not decline in cells deprived of MVA synthesis (Figs. 3  and 5-7). Also, antibody recognition of the 56-kDa macromolecule was not inhibited with i"A in solution (Fig. 4). Therefore, contrary to i"A26, there are no supporting data to indicate that the 56-kDa macromolecule contains isopentenyladenine. Perhaps the 56-kDa macromolecule contains an antibody-binding epitope that is structurally similar but distinct from isopentenyladenine. Alternatively, the 56-kDa macromolecule may contain a metabolically stable form of isopentenyladenine which binds anti-i"A antibodies with sufficient affinity to preclude competition by i"A in solution. We ' J . R. Faust and J. F. Dice, unpublished data. also observed antibody binding to several bands in the immunoblot of the liver supernatants (Fig. 8). Liver may contain additional isopentenyladenine-modified proteins not present in CHO cells. It is also possible that many cellular proteins contain isopentenyladenine moieties; however, our anti-ifiA antibodies may recognize only a subset. Antibody binding to small epitopes may be influenced by various chemical structures, e.g. amino acid side chains, in the vicinity of the epitope.
We are unaware of any reports describing isopentenyladenine modification of proteins and have considered that i'A26 may not contain this new modification. Mevinolin suppression of i'A26 may be a consequence of the mevinolin-induced quiescent state in CHO cells and may not reflect decreased isopentenyladenine modification as a result of MVA deprivation. In fact, we show that cells rendered quiescent as a result of incubation in the absence of serum growth factors express reduced levels of i"A26. Also, in addition to i"A, our antibodies may promiscuously recognize another unknown modification or unique cluster of amino acids with a conformation similar to isopentenyladenine. Nevertheless, we speculate that i"A26 contains isopentenyladenine and that elevated expression of i'A26 requires both serum growth factors and MVA in concert.
Studies on the regulation of the mammalian cell cycle have shown that nuclei of cells in GI phase will rapidly undergo DNA replication when the GI cells are fused to S phase cells (41). The major conclusion from these experiments is that a limiting truns-acting inducer of S phase accumulates gradually throughout GI and stimulates S phase when it reaches a critical threshold level. Also, Pardee and co-workers reviewed evidence (42,43) that a specific labile protein must accumulate to a sufficient level (termed "restriction point") in order for Go cells to traverse S phase. Our studies show that i6A26 accumulates in GI phase prior to the onset of S phase. Therefore, i"A26 may participate in trans-induction and/or restriction point control of S phase.
Fairbanks et al. (4) studied the specific requirement for MVA during the cell cycle. By varying the time of addition and removal of both mevinolin and MVA to culture medium after platelet-derived growth factor stimulation, these investigators delineated a critical time period in GI phase when synchronized human fibroblasts require MVA for S phase DNA synthesis. They concluded that a MVA metabolite is required in GI for cells to enter S phase. We report in this paper that i"A26 accumulates in GI phase of CHO cells. Mevinolin treatment of synchronized CHO cells inhibited GI accumulation of i'A26 and coincidentally delayed the onset of S phase. Consequently, i"A26 is a candidate for the critical MVA metabolite required for S phase DNA replication.
If mevinolin inhibition of DNA synthesis is due to decreased availability of i' A for modification of the 26-kDa protein, we might expect that i"A, or an isopentenyladenine derivative, would replace MVA in restoring mevinolin-inhibited DNA replication. Two groups have reported regulation of mammalian cell proliferation by isopentenyladenine and PA. Gallo et al. (44) observed that i6A either stimulated or inhibited DNA synthesis, cell transformation, and mitosis in phytohemagglutinin-incubated human lymphocytes. The opposing effects of i' A were highly dependent upon the concentration and time of addition of the nucleoside.
Also, Siperstein and colleagues (45, 46) studied the role of MVA in regulating S phase DNA synthesis in baby hamster kidney cells. These workers reported that isopentenyladenine and its 4'-hydroxylated derivative, zeatin, could substitute for MVA in restoring DNA replication in compactin-treated cells (compactin, similar to mevinolin, is an inhibitor of cellular MVA synthesis). Isopentenyladenosine was ineffective. Restoration of DNA synthesis was only observed in baby hamster kidney cells incubated within a narrow concentration range of isopentenyladenine and only when the modified base was added to cells late in GI phase. Several groups have reported that isopentenyladenine and/ or i6A did not prevent mevinolin-inhibited DNA synthesis (3, [7][8][9]47). Apparently the ability of i6A to modulate DNA replication is highly dependent upon the concentration, time of addition, and chemical form of the isopentenyl derivative. Also, these isopentenylated compounds appear to have multiple effects in various cultured cells, i.e. inhibition of protein and RNA synthesis and stimulation/inhibition of DNA replication. Perhaps the disparate effects of the isopentenyladenine derivatives in various cells may be due to dissimilar metabolic capabilities of the specific cells. If a phosphorylated nucleotide form of i6A is responsible for modulating DNA synthesis, then the differential expression of purine salvage and catabolic pathways may explain the conflicting observations in various cell types incubated with either the modified base or nucleoside.
Cytokinins are i6A derivatives that serve as important plant hormones regulating cell division, gene expression, and differentiation in various tissues (48). Although isopentenyl tRNA degradation is a minor pathway contributing to i6A levels in plant tissue, plant cytokinins are mostly derived via direct modification of adenosine monophosphate. Direct evidence for the role of i6A in plant cell proliferation has come from studies on tumor induction by Agrobacterium tumefaciens in crown gall disease (49,50). Crown gall disease is a result of the bacteria transferring a gene coding for i6A biosynthesis into plant cells which leads to increased production of the hormone. The elevated levels of cytokinin are specifically responsible for uncontrolled tumor growth as well as suppression of root growth. The molecular mechanisms underlying cytokinin-controlled differentiation and growth induction/suppression in plants are unknown. Perhaps isopentenyladenylated proteins mediate cytokinin regulation in plants. Conjugates of cytokinins and the amino acids alanine (51) and the 2-aminobutyric acid moiety of methionine (52,53) have been identified in soybean and Dictyostelium discodieum, respectively. It is possible that these cytokinin-conjugated amino acids may arise from degradation of isopentenyladenylated proteins. Our present observations correlating the concentration of an anti-i6A-binding cellular component with mammalian cell replication may underscore a common mechanism controlling proliferation in both the animal and plant kingdoms.