Enhanced ribosomal association of p27(Kip1) mRNA is a mechanism contributing to accumulation during growth arrest.

p27Kip1 regulates the decision to enter into S-phase or withdraw from the cell cycle by establishing an inhibitory threshold above which G1 cyclin-dependent kinases accumulate before activation. We have used the HL-60 cell line to study regulation of p27 as cells withdraw from the cell cycle following treatment with 12-O-tetradecanoylphorbol-13-acetate (TPA). We found that the amount of p27 is maximal in G0 cells, lower in G1 cells, and undetectable in S-phase cells. In contrast to the protein, the amount of p27 mRNA was the same in these populations, suggesting that accumulation of p27 during the cell cycle and as cells withdraw from the cell cycle is controlled by post-transcriptional mechanisms. In S-phase cells, the degradation of p27 appears to predominate as a regulatory mechanism. In G0 cells, there was an increase in the synthesis rate of p27. Our data demonstrate that, in G0 cells, accumulation of p27 is due to an increase in the amount of p27 mRNA in polyribosomes.

The activation status of the cyclin-dependent kinases (CDKs) 1 regulates progression through G 1 (1). The phosphorylation of the Rb protein correlates with transition through the restriction point when cells commit irreversibly to S-phase and subsequent cell division (2). Rb is phosphorylated by cyclin D⅐CDK complexes early in G 1 , and cyclin E⅐CDK2 complexes further phosphorylate and maintain the hyperphosphorylated state through the remainder of G 1 (3). A class of proteins that interact with both G 1 CDKs, called Kips (CDK inhibitory proteins), might coordinate these two classes of kinases and the start of S-phase. p27 Kip1 might directly bridge the activation of cyclin E⅐CDK2 complexes to expression of cyclin D⅐CDK complexes (4,5). p21 Kip1 might regulate cyclin⅐CDK2 complexes at a checkpoint following DNA damage or nucleotide pool perturbation (6 -8). The function of p57 Kip2 is unclear at this time.
Many observations suggest that p27 is a key regulator controlling entry into and exit from the cell cycle. Regardless of cell type or condition used to achieve growth arrest, an increase in the amount of p27⅐CDK2 complexes is a common feature of non-proliferating cells (9). In addition, ectopic expression of p27 in cells leads to G 0 /G 1 arrest (10,11), and antisense vectors that block p27 lead to an increase in the percentage of cells in S-phase (12). Furthermore, the phenotypes of Kipl mice suggest that the loss of p27 leads to an alteration in the balance between proliferating and non-proliferating cells (13)(14)(15). The stoichiometric nature of p27-mediated inhibition has led to the proposal that p27 functions to establish an inhibitory threshold above which CDKs accumulate before their activity can drive the cell cycle (16).
There are multiple mechanisms to regulate the amount of p27 available for interaction with cyclin⅐CDK2 complexes, dependent both on the cell type and the condition that leads to growth arrest. First, in proliferating cells, cyclin D⅐CDK complexes sequester the CDK2-inhibitory activity of p27. In proliferating mink lung epithelial cells (Mv1Lu) p27 associates with cyclin D2⅐CDK4 complexes (4). Exposure of Mv1Lu cells to transforming growth factor-␤ or growth to confluence leads to a displacement of p27 from cyclin D⅐CDK4 complexes, allowing it to associate with and inhibit the cyclin⅐CDK2 complex (17). In proliferating MANCA cells, p27 associates with cyclin D complexes in a non-inhibitory manner (5). Second, changing the amount of p27 mRNA can control accumulation of p27 protein; increased accumulation of p27 mRNA contributes to the increase in p27 protein following exposure of U937 cells to vitamin D3 (18). Third, ubiquitin-mediated degradation contributes to p27 accumulation in quiescent IMR-90 and MG-63 cells (19). We now demonstrate a fourth mechanism of p27 regulation that involves translation control, regulating association of p27 mRNA with polyribosomes.

EXPERIMENTAL PROCEDURES
Cell Culture, mRNA, and Protein Analysis-HL-60 cells were maintained at 2-5 ϫ 10 5 /ml in RPMI plus 10% fetal calf serum. Under these conditions, all cells incorporated bromodeoxyuridine during a 27-h labeling period (data not shown). To achieve growth arrest, cells were treated with 33 nM TPA. To obtain synchronized cell cycle fractions, 4 liters of cells were grown and elutriated as we described (20). DNA content was determined by propidium iodide staining as described (20). RNA was isolated from cells using RNA-STAT 60 following the recommendations of the manufacturer, and 20 g of total RNA was used for Northern blotting as described (5). Proteins were isolated by sonicating cells in Tween lysis buffer as described (5). Immunoblot and kinase assays were performed as described previously (5,20). To quantitatively determine the amount of protein using enhanced chemiluminescence (ECL), we included a standard protein curve in all experiments. This allowed us to determine a suitable range for quantitation based on the linear relationship between amount of protein and the intensity of signal. Signal quantitation was performed by spot densitometry using an IS1000 digital imaging system (Alpha-Innotech).
Accumulation of p27 in Cells Exposed to LLnL-Either following elutriation of an asynchronous population into cell cycle phase synchronized fractions or 48 h after TPA-treatment of an asynchronous population of cells, the G 0 , G 1 , or S-phase cells were exposed to either 2 g/ml * This work was supported by grants from the National Institutes of Health and the Society of Memorial Sloan-Kettering Cancer Center (to A. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
LLnL (19) in Me 2 SO or an equivalent volume of Me 2 SO for 3 h. Extracts were prepared from treated and untreated cells, and proteins were either directly resolved by electrophoresis on a 10% SDS-polyacrylamide gel or immunoprecipitated using an excess amount of p27 antibody. The proteins were then transferred to Immobilon (Millipore Corp.), and p27 was detected by immunoblotting. Spot densitometry was performed, and the increase in p27 was measured.
Metabolic Labeling-G 0 , G 1 , and S-phase cells were isolated and treated with LLnL for 90 min in medium containing 1 mCi/ml Amersham Promix at a density of 4 ϫ 10 6 cell/ml. Cells were washed twice in phosphate-buffered saline and lysed in RIPA buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 5 mM EDTA, pH 8.0) containing 10 g/ml each aprotinin, leupeptin, soybean trypsin inhibitor, and pepstatin, and 1 mM phenylmethylsulfonyl fluoride. A small aliquot was removed, protein was precipitated with TCA, and the amount of radioactivity incorporated was quantitated in a scintillation counter. Immunoprecipitations were normalized for incorporated radioactivity, and proteins were resolved on a 12% SDS-polyacrylamide gel. After autoradiography, the p27 band was excised, and radioactivity was measured.
Differential Polysome Association-To examine polysome association of mRNA, we treated cells with 100 g/ml cyclohexamide for 15 min prior to harvesting, collected 4 ϫ 10 7 cells by centrifugation, and washed the pellet twice with ice-cold phosphate-buffered saline containing 100 g/ml cyclohexamide. Cells were lysed in 1 ml of lysis buffer (150 mM NaCl, 1.5 mM MgCl 2 , 150 g/ml cyclohexamide, 20 mM dithiothreitol) containing 0.5% Nonidet P-40 and 1 mM phenylmethylsulfonyl fluoride. Nuclei and cell debris were removed by centrifugation at 13,000 ϫ g for 5 min at 4°C. We added 1.5 mg of heparin to the supernatant and applied this to the top of a 10-ml 15-40% continuous sucrose gradient. The gradient was prepared in lysis buffer containing 1.5 mg/ml heparin. Gradients were resolved by centrifugation at 38,000 rpm in an SW41 rotor for 1.5 h at 4°C, and 1-ml fractions were collected. The fractionation of ribosomes was assessed by UV absorbance and ethidium bromide staining of agarose gels.
To quantitate the amount of p27 mRNA in each fraction, we blotted 400 l (ϳ40% of the total) to nitrocellulose and fixed the membrane in 10% formalin before UV cross-linking. Membranes were probed with a random primed human p27 cDNA. Quantitation was performed using a phosphoimager (Fuji MacBas 2000, Version 2.0).
To control for the specificity of our p27 probe, we also performed RT-PCR. Each fraction was adjusted to 1% SDS and 10 mM EDTA and 150 ng of proteinase K was added. The fractions were subsequently incubated for 30 min at 37°C, and then RNA was purified, following phenol/chloroform extraction, on Nensorb 20 nucleic acid purification cartridges as described by the manufacturer (Dupont NEN). RNA was reverse-transcribed using random 9-mer primers (Stratagene) and M-MLV reverse transcriptase. The RT product was subsequently amplified with specific p27 primers (5Ј-TTGCAGGAACCTCTTCGGCC-3Ј and 5Ј-GGTCGCTTCCTTATTCCTGC-3Ј) for 40 to 45 cycles using AmpliTaq (Perkin-Elmer). PCR reactions were initiated with a 94°C 5-min incubation and then cycled through 94°C for 90 s, 55°C for 60 s, and 72°C for 120 s. PCR was completed with a 15-min extension at 72°C. Following PCR, an aliquot of product was loaded onto an agarose gel, transferred to nylon membranes, and probed with random primerlabeled p27 cDNA probes.
For detection of GAPDH, one half of the RNA from each fraction was loaded directly onto agarose gels, and Northern blotting was performed with a random primer-labeled GAPDH probe. Data was collected directly either by autoradiography or by phosphoimager (Fuji Mac-Bas2000, Version 2.0).

Cell Cycle-dependent and Growth-dependent Post-transcriptional Regulation of p27 in HL-60 Cells-
The decision to either proliferate or withdraw from the cell cycle is made during G 1 phase and is affected by the relative amounts of cyclin⅐CDK complexes and CDK inhibitors (18,(21)(22)(23)(24)(25). We used HL-60 cells to study the mechanisms controlling cell cycle withdrawal during differentiation because we could isolate populations of cells in each phase of the cell cycle in sufficient quantities for biochemical analysis. HL-60 cells differentiate into monocytes in the presence of TPA (26). Proliferation in asynchronous cultures of HL-60 cells completely ceased within 36 h following exposure to TPA, and typically, greater than 90% of the treated cells arrested with a G 1 content of DNA (Fig. 1A). The significance of the 10% of cells arrested in G 2 /M is unclear. TPA treatment of enriched G 1 or S-phase populations obtained by centrifugal elutriation suggested that cells arrested in a single cell cycle (Table I) at the next G 1 phase. The 15% reduction of G 1 cells following TPA treatment of elutriated G 1 cells is consistent with a TPA restriction point subdividing this phase of FIG. 1. In HL-60 cells, activation of PKC leads to growth arrest and induction of p27. A, flow cytometry. Asynchronous cultures of HL-60 cells were exposed to 33 nM TPA, and aliquots were removed at 6-h intervals as indicated on the right and processed for flow cytometry. B, CDK2 kinase activity is diminished in TPA-treated cells. Cell lysates were prepared, and the ability of CDK2 immunoprecipitates to phosphorylate histone H1 was measured. The lanes are labeled to correspond with panel A. C, p27 protein increases without a change in the amount of p27 mRNA. Protein and RNA were isolated from cell lysates, and the amount of p27 was determined by immunoblot (top) and Northern blotting (bottom). The lanes are labeled to correspond with panel A. p27 protein was detected with an affinity purified p27-specific antibody. Northern blots were probed with either p27 or GAPDH cDNA as indicated on the right. the cell cycle. Concomitant with accumulation of cells with a G 1 content of DNA following exposure to TPA, there was a coordinated decrease in immunoprecipitable CDK2 kinase activity (Fig. 1B) and an induction of p27 protein (Fig. 1C). The amount of p27 mRNA remained the same (Fig. 1C). Similar results were obtained with the phase-enriched populations (data not shown).
Since TPA-treated cells arrest with a 2C DNA content, p27 might accumulate in these cells as an indirect consequence of G 1 arrest in a state where p27 protein is stabilized. To address this, we first determined whether p27 expression is cell cycle regulated. We obtained cell cycle phase-specific populations of HL-60 cells by centrifugal elutriation of asynchronous cultures and examined p27 protein expression (Fig. 2). This method does not induce perturbations in cell cycle-regulated protein expression often observed with drug or metabolic-induced cell cycle synchronization. The amount of p27 protein was maximal during the G 1 phase of the cell cycle (Fig. 2B). The amount of p27 protein detected was linearly dependent on the amount of cell extract subjected to immunoprecipitation (data not shown; but see Fig. 3A). The amount of p27 mRNA did not change during the cell cycle (data not shown). These data suggest that, during the unperturbed cell cycle, p27 levels are determined by a post-transcriptional mode of regulation. However, the amount of p27 in equal numbers of G 0 and G 1 cells was quite different (Fig. 3A). G 0 cells contained at least 3-4-fold more p27 than G 1 cells. We obtained similar results comparing the amount of p27 as a function of total protein rather than cell number (data not shown). The amount of p27 detected was linearly dependent on the amount of cell extract subjected to immunoprecipitation (Fig. 3A, compare lanes 1 and 2). Together, these data suggest that accumulation of p27 reflected post-transcriptional regulation during both the G 1 /G 0 and G 1 /S transitions.
Phase-dependent Changes in Synthesis of p27-The accumulation of p27 represents the sum of protein synthesis and protein degradation. Precedence exists for the regulation of protein half-life as a function of cell cycle phase during both the G 1 /S transition (27) and at the metaphase/anaphase transition (28). Determination of protein half-life during specific phases of the cell cycle is problematic because traversal of the cell cycle phases during the chase period will obscure the measured value. Consequently, to define the post-transcriptional mechanisms regulating p27, we measured the accumulation of p27 and the actual synthesis of p27 in G 0 , G 1 , and S-phase cells treated with LLnL (N-acetyl-leucinyl-leucinyl-norleucinal-H), an inhibitor that binds to the chymotryptic site on the proteasome. To measure accumulation of p27, we titrated each extract against recombinant p27 over a 16-fold range using 2-fold serial dilution. The resulting autoradiograph was analyzed by spot densitometry using an IS-1000 digital imaging system (Alpha Innotech), and we compared only those points that fell within the linear range (established by comparing signal intensity to amount of recombinant protein). Although the amount of p27 increased in all populations treated with LLnL, consistent with detection of newly synthesized p27 (shown below), the magnitude of the increase was cell cycle phase-specific (Fig. 3B). The increase in p27 levels was greatest in S-phase cells (about 160%) and more moderate in G 1 and G 0 cells (both about 50%).
To measure the amount of p27 synthesis in each population treated with LLnL, we labeled cells with 35 S-methionine and -cysteine for 90 min and immunoprecipitated p27. The short labeling period ensured that synchronized cells did not progress into the next phase of the cell cycle and inclusion of LLnL would prevent protein degradation; however, cells within G 1 phase might traverse the restriction point during this period. 2 After normalizing lysates for total protein synthesis, we isolated p27 with affinity purified p27-specific antibodies. We quantitated the amount of p27 synthesis by scintillation counting of excised bands, normalizing incorporation to 100% in G 1 cells. We found that the amount of newly synthesized p27 was greatest in G 0 cells and lower in G 1 and S-phase cells (Fig. 3C). We were unable to demonstrate if LLnL affected synthesis of p27 in a cell cycle phase-specific manner; however, similar increases in translation of p27 have been reported in other systems where LLnL was not included (29,30). This suggested that the actual rate of p27 synthesis increased in G 0 cells.
Increased Synthesis of p27 Occurs as a Consequence of an Increase in the Amount of p27 mRNA in Polyribosomes-To determine the mechanism responsible for accumulation of p27 in G 0 cells, we compared the association of p27 mRNA with polysomes in proliferating and TPA-treated cells. To accomplish this, we fractionated RNA on continuous sucrose gradients and monitored the UV absorbance of the ribonucleoprotein complexes at A 254 . In this assay, RNA equilibrates within the gradient as a function of the associated proteins; polysomal RNA equilibrates in the densest regions of the gradient. The  extent of polysomal association is a reflection of the rate of protein synthesis; increased association reflects either an increase in the rate of initiation or an increase in the rate of elongation. To quantitate the p27 mRNA, we dot-blotted each fraction of the gradient and probed the membranes with a p27 cDNA probe. In untreated cells, p27 mRNA distributed in the free mRNA fraction of the gradient and in fractions containing monosomes and small polysomes (Fig. 4B). In TPA-treated cells, p27 mRNA distributes throughout the gradient, with the greatest differences occurring in the densest region of the gradients. Treatment of cells with puromycin will disrupt polysomes and lead to an accumulation of mRNA in monosome and subunit fractions of the gradient. When we exposed TPA-treated cells to puromycin, the p27 mRNA shift was eliminated, confirming polysome association (Fig. 4B). We found approximately 60% more p27 mRNA associated with polysomes in TPA-treated cells (Fig. 4D). This increase appears to occur at the expense of the monosome and small polysome associated mRNA (Fig. 4D), suggesting that p27 synthesis might be regulated at the level of translation elongation. We were unable to detect a significant change in the distribution of p27 mRNA in polyribosomes isolated from G 1 and S-phase cells, consistent with our observation that p27 synthesis, in the presence of the proteasome inhibitor LLnL, was approximately equal in these populations (data not shown).
We confirmed the specificity of the dot-blot by isolating RNA from sucrose gradients and performing RT-PCR with p27-specific oligonucleotide primers. In this experiment, we assessed polysomal fractionation by ethidium bromide staining and detection of 28 and 18 S rRNA following agarose gel electrophoresis (Fig. 5). Consistent with our dot blot results, we found a shift in the distribution of p27 mRNA with a substantial amount fractionating with heavier polysomes in TPA-treated cells (Fig. 5). We confirmed that the PCR product was p27 by restriction endonuclease mapping (data not shown). Furthermore, addition of a control DNA encoding p27 but missing 100 base pairs of the sequence between the primer binding sites was specifically amplified (data not shown).
To ensure that the change of p27 migration into heavier polysomes did not represent a nonspecific change in bulk translation, we directly probed fractionated RNA with GAPDH. There was little change in the migration of GAPDH mRNA following TPA treatment, either in the RT-PCR analysis (Fig.  5) or dot-blotting procedure (Fig. 4C). This suggests that the changes in polysome distribution of p27 mRNA were specific and did not reflect general changes that cells might undergo during alternative developmental pathways. Together, these experiments demonstrate that increased expression of p27 pro-  , lanes 2-4). In the bottom panel, p27 was immunoprecipitated from 1 mg of each extract (lanes 2-4) and detected by immunoblotting. To ensure that excess antibody was used in the immunoprecipitation, 3 mg of G 0 extract was immunoprecipitated under the same conditons (lane 1). The amount of p27 is greatest in G 0 cells, minimal in S-phase cells, and intermediate in G 1 cells. B, accumulation of p27 in the presence of LLnL. Cells were treated with LLnL for 3 h prior to lysis. p27 was quantitated by immunoblotting. On the ordinate, we plotted the LLnL-dependent increase in p27. This experiment was repeated four times with comparable results, and the data presented is from a single representative experiment. Linearity was confirmed by including a titration of known amounts of recombinant p27 in the immunoblot. For quantiation of p27 accumulation in G 1 and S-phase cells, we compared the accumulation at 3 h in the presence and absence of LLnL. p27 accumulates to the greatest extent in G 0 phase cells. C, synthesis of p27 is maximal in cells treated with TPA. The p27 bands were excised from an SDS-polyacrylamide gel (shown below the graph), and the amount of radioactivity incorporated was measured by scintillation counting. Background was determined individually for each lane in a similar manner. On the ordinate, we plotted incorporation of radioactivity into p27 relative to the amount in G 1 cells. tein in cells exposed to TPA could be attributed to an increase in the density of ribosomes associated with p27 mRNA.

DISCUSSION
To study how the p27-mediated CDK inhibitory threshold is modulated during G 1 , we explored the mechanisms regulating p27 as HL-60 cells withdraw from the cell cycle or enter Sphase. We have analyzed synchronous cell populations in G 0 , G 1 , or S-phase. We have shown that the translation of p27 changes in a growth-dependent manner, the association of p27 mRNA with polysomes is a determinant of the accumulation of p27 protein. Furthermore, degradation of p27 is a major determinant of accumulation of p27 during the G 1 /S transition.
We found that an increase in the rate of p27 protein synthe-sis modulates the amount of p27 as HL-60 cells withdraw from the cell cycle. First, the amount of p27 is higher in growtharrested cells than in G 1 cells although the amount of mRNA remained the same. Second, the p27 synthesis rate is increased in G 0 cells compared with either G 1 or S-phase cells when proteolysis was prevented by treating cells with LLnL. Third, the polysome distribution of p27 mRNA shifts to heavier fractions in G 0 cells, representing a greater density of actively translating ribosomes on each mRNA. This type of polysome distribution might represent more efficient utilization of the p27 mRNA template either by altering the initiation or elongation phases of translation. Translation control is not unique to either TPA or differentiating HL-60 cells. Hengst and Reed FIG. 4. p27 mRNA is distributed to heavier polysomes in TPA-treated cells. A, sucrose gradient tracings. Cells were cultured with (right panel) or without TPA (left panel) as described under "Experimental Procedures." Puromycin (150 g/ml) was added 45 min prior to harvesting. Fractions were collected from the bottom of the gradient, and the A 254 was measured during fractionation. Solid lines represent the absorbance in the absence of puromycin, and dashed lines represent the absorbance in the presence of puromycin. B, quantitation of p27 mRNA in fractions of a representative sucrose gradient. The amount of mRNA in each fraction was measured by determining the amount of probe hybridized to dotblots of each fraction on a phosphoimager and plotting the value as a percent of total p27 mRNA signal. In the left panel, control (white) and TPA-treated (black) gradients were compared. In the right panel, cells were treated with puromycin before collection and comparison of the p27 mRNA amounts. Results are representative of seven separate experiments. C, TPA increases the polysomal associated p27 mRNA 60%. Shown is the quantitation of GAPDH mRNA in each fraction of the sucrose gradients. Labeling is as described in panel B. D, the increase in polysomal mRNA occurs as a consequence of more efficient conversion of monosome and subunit-associated mRNA into polysomes. We compared the total amount of puromycin-sensitive p27 mRNA, the free mRNA (determined on ethidium bromidestained cells by absence of both 28 and 18 S rRNA species in the fraction), and the remaining fractions. These values are tabulated (mean Ϯ S.E.). (29) reported translation regulation of p27 in both lovastatinarrested HeLa cells and confluent human diploid fibroblasts, and Agrawal et al. (30) showed that platelet-derived growth factor treatment of BALB/c 3T3 cells represses translation of p27 mRNA. Our data expands this observation to events following the programmed withdrawal of a cell from the cell cycle into a terminally differentiated state.
There are many examples where mRNA interaction with polysomes occurs as cells re-enter the cell cycle from a nonproliferative state (31,32). In contrast, p27 mRNA appears to be a member of a small group of messages that interact more efficiently with the translation apparatus as cells withdraw from the cell cycle. The mechanism regulating association is unclear. Factors might either promote ribosome association with p27 mRNA in G 0 cells or antagonize the association in non-G 0 cells. There is precedence for factors operating at sequences located in non-translated regions that might control ribosome association (32)(33)(34)(35)(36) or subcellular localization (37,38). Alternatively, ribosomal proteins specific to non-proliferating cells might eliminate secondary structures in the p27 mRNA, allowing it to interact efficiently with ribosomes. The eIF-4E helicase facilitates unwinding and ribosomal association of mRNAs required to enter the cell cycle (38,39). A presumptive factor regulating p27 mRNA association might act in a similar fashion, but with opposite biologic effect, and selectively recruit G 0 -specific mRNA. However, we favor a model in which an inefficient repressor of translation is alleviated as cells withdraw from the cell cycle because there is translation occurring in proliferating cells, albeit less efficiently than in non-proliferating cells. Future experiments are underway to explore the mechanism regulating p27 translation.
The degradation of proteins is a key regulatory mechanism controlling protein expression as cells traverse the cell cycle (40). Substantial evidence has accumulated to support the crucial role of proteolysis at the metaphase/anaphase transition. Likewise, the involvement of a proteasome at the G 1 /S transition has been implicated in Saccharomyces cerevisiae (27,41,42). We have shown that the amount of p27 decreases precipitously at the G 1 /S transition although the synthesis of p27 remained the same. Furthermore, p27 accumulates to a similar extent when LLnL is added to either G 1 or S-phase cells. These observations suggest that protein degradation controls p27 accumulation during the cell cycle, and the amount of LLnL used in our experiments is sufficient to inhibit that proteasome activity. Confirmation of this hypothesis, however, awaits the ability to measure the half-life of p27 in cell cycle phase-specific extracts capable of degrading p27.