Cell cycle-coupled variation in topoisomerase IIalpha mRNA is regulated by the 3'-untranslated region. Possible role of redox-sensitive protein binding in mRNA accumulation.

Mammalian topoisomerase IIalpha (Topo II) is a highly regulated enzyme essential for many cellular processes including the G(2) cell cycle checkpoint. Because Topo II gene expression is regulated posttranscriptionally during the cell cycle, we investigated the possible role of the 3'-untranslated region (3'-UTR) in controlling Topo II mRNA accumulation. Reporter assays in stably transfected cells demonstrated that, similar to endogenous Topo II mRNA levels, the mRNA levels of reporter genes containing the Topo II 3'-UTR varied during the cell cycle and were maximal in S and G(2)/M relative to G(1). Topo II 3'-UTR sequence analysis and RNA-protein binding assays identified a 177-nucleotide (base pairs 4772-4949) region containing an AUUUUUA motif sufficient for protein binding. Multiple proteins (84, 70, 44, and 37 kDa) bound this region, and the binding of 84- and 37-kDa (tentatively identified as the adenosine- or uridine-rich element-binding factor AUF1) proteins was enhanced in G(1), correlating with decreased Topo II mRNA levels. The binding activity was enhanced in cellular extracts or cells treated with thiol-reducing agents, and increased binding correlated with decreased Topo II mRNA levels. These results support the hypothesis that cell cycle-coupled Topo II gene expression is regulated by interaction of the 3'-UTR with redox-sensitive protein complexes.

Mammalian topoisomerase II␣ (Topo II) 1 is a multifunctional protein involved in many cellular processes including replication, repair, transcription, recombination, chromosome condensation and segregation, and the G 2 cell cycle checkpoint pathway (1)(2)(3)(4)(5). Topo II levels (mRNA, protein, and activity) increase in late S phase, peak in G 2 /M, and rapidly decrease following mitosis (4, 6 -8). Consistent with these observations, inhibitors of Topo II have been shown to arrest mammalian cells before mitosis in the G 2 phase of the cell cycle (6, 9 -12), and deregu-lated expression of Topo II results in cell death (13). In HeLa cells synchronized by mitotic shake-off, Topo II mRNA levels increased 16-fold in late S to G 2 /M phases (14 -18 h after plating) relative to G 1 (6). This increase in Topo II mRNA level was associated with a significant increase in Topo II mRNA stability, but with only marginal changes in the transcription rate.
The regulation of mRNA stability has emerged as an important control mechanism of gene expression. Although the mechanisms that alter mRNA stability of different genes have unique features, in general, sequences located in the 3Ј-untranslated region (UTR) and their interactions with specific proteins regulate mRNA stability (14 -17). The most common 3Ј-UTR stability determinants are adenosine-or uridine-rich elements (AREs), which include AUUUA, AUUUUA, and AUUUUUA motifs (18 -22). Furthermore, a family of proteins, the AU-binding proteins, including AUF1 (23), Hel-N1 (24), AUH (25), HuR (26), and AUBF (27), have the capacity to bind with high affinity to mRNA containing ARE. Binding of AUbinding proteins (e.g. 15-, 18-, and 19-kDa AUBF and 37-, 40-, 42-, and 45-kDa AUF1) to AREs correlates with either mRNA stabilization (27,28) or destabilization (23,29,30). These observations suggest that ARE elements can be both positive and negative regulators of mRNA stability, depending on the sequence context in which the sequence elements are located and the precise protein composition of the RNA-protein complex. In addition, posttranslational modifications involving oxidation/ reduction as well as phosphorylation/dephosphorylation of AUbinding proteins have also been shown to affect ARE binding (23,27,31). Binding to cytokine and lymphokine mRNAs can be reversibly blocked by the thiol-oxidizing agent diamide but irreversibly inhibited by the thiol-alkylating agent N-ethylmaleimide (NEM) and enhanced by the thiol-reducing agent 2-mercaptoethanol (2-ME) (31,32). Thus, it is possible that subtle changes in redox potential during cell growth or in response to environmental stimuli may influence ARE-AUbinding protein binding and alter the stability of ARE-containing mRNA. We showed earlier that the cell cycle-coupled accumulation of Topo II mRNA levels is regulated by changes in mRNA stability (6). The present study was designed to test the idea that Topo II gene expression during the cell cycle is regulated by 3Ј-UTR via interactions with redox-sensitive protein complexes.

MATERIALS AND METHODS
Cell Culture and Extract Preparation-HeLa cell cultures, synchronization by selective detachment of mitotic cells, and flow cytometric analysis of cell cycle positions were performed following previously described procedures (6). Mouse NIH 3T3 fibroblast cells were grown in Dulbecco's modified Eagle's medium with 10% calf serum and antibiot-ics. Exponentially growing cells were synchronized by replacing the growth medium with medium containing 0.5% serum and serum-stimulated to reenter the cell cycle. Cytoplasmic extracts were prepared following the method described by Konarska and Sharp (33). In selected experiments, dithiothreitol (DTT) was omitted during the preparation of cytoplasmic extracts. For determining redox sensitivity in RNAprotein binding, protein extracts were treated with DTT, 2-ME, NEM, or diamide for 10 min at 25°C prior to RNA-protein gel shift analysis.
Plasmid Construction and DNA Transfections-Three reporter constructs, (a) H⌬Topo2, a truncated form of Topo II cDNA with the entire 3Ј-UTR (base pairs 1-3013 deleted), (b) H␤galSV40, ␤-galactosidase reporter gene fused with the SV40 3Ј-UTR at its 3Ј-end, and (c) H␤galTopo, ␤-galactosidase reporter fused with Topo II 3Ј-UTR at its 3Ј-end, were engineered. All three constructs were fused at the 5Ј-end with a 1.6-kb fragment containing 650 nucleotides (nt) of the Topo II promoter, first exon (21 nt), intron (860 nt), and the second exon (150 nt) (34). The 1.6-kb fragment was polymerase chain reaction-amplified from human genomic DNA (CLONTECH) using primer pairs 5Ј-Ϫ562 G-GGGCGGGGTTGAGGCAGATGCCAGAATCT Ϫ532 -3Ј and 5Ј-ϩ150 AGG-TGTCTGGGCGGAGCAAAATATGTTCC ϩ121 -3Ј (34). The polymerase chain reaction-amplified fragment was cloned into TA-cloning vector (Invitrogen), digested with MfeI restriction enzyme, and ligated into SpeI-digested hTopo2 plasmid (plasmid DNA containing the entire human Topo II cDNA, a gift from Dr. J. C. Wang,Ref. 35) to generate plasmid H⌬Topo2. Plasmid H⌬Topo2 when expressed will represent approximately a 2.2-kb Topo II reporter mRNA containing Topo II 3Ј-UTR at its 3Ј-end. For the H␤galSV40 plasmid, the TA-cloning vector containing the 650-nt Topo II promoter, first exon, intron, and the second exon was digested with SpeI and XhoI restriction enzymes and directionally cloned into a SmaI-and XhoI-digested pNASS␤ reporter vector (CLONTECH). The plasmid DNA pNASS␤ lacks eukaryotic promoter and enhancer sequences and contains E. coli ␤-galactosidase as the reporter. The resulting reporter construct when expressed will transcribe approximately a 3-kb ␤-galactosidase reporter mRNA with the SV40 3Ј-UTR at its 3Ј-end. In the third reporter construct H␤galTopo, the SV40 3Ј-UTR in H␤galSV40 plasmid DNA was removed by SalI and BamHI restriction enzyme digestion and replaced with a XhoI-and BamHI-digested Topo II 3Ј-UTR from the H⌬Topo2 plasmid. The resulting reporter construct when expressed will represent approximately a 4-kb ␤-galactosidase reporter mRNA with Topo II 3Ј-UTR at its 3Ј-end. Mouse NIH 3T3 fibroblast cells were stably transfected with plasmid DNAs pSVneo (CLONTECH) and H⌬Topo2, H␤galTopo, or H␤galSV40 using LipofectAMINE (Life Technologies, Inc.) following a manufacturer-supplied protocol. Geneticin-resistant colonies were pooled and cultured in G418-containing medium.
Construction of Riboprobe Templates and in Vitro Transcription-cDNA sequences representing various regions of Topo II 3Ј-UTR were amplified by standard polymerase chain reaction techniques from a plasmid containing the entire Topo II 3Ј-UTR (pTopUTR). Polymerase chain reaction products representing the proximal (nt 4495-4948), middle (nt 4772-5298), distal (nt 5093-5595), and overlapping (nt 4772-4948) regions of Topo II 3Ј-UTR ( Fig. 2) were purified by agarose gel electrophoresis and subcloned into pGEM-T plasmid DNA (Promega). The fos-ARE plasmid (pfosARE) containing both AUrich domains I and II was constructed by subcloning the fos 3Ј-UTR from the pBBB plasmid (20) into pBS II SK(ϩ). The plasmid containing the nonspecific ␤-galactosidase competitor was constructed by subcloning the coding sequence within the HpaI and ClaI restriction sites in pCMV plasmid (CLONTECH) into pBS II SK(ϩ). Orientation and sequence of inserts were verified by dideoxy sequencing of both strands of DNA in each plasmid construct.
Riboprobes representing the sense strand of RNA from each plasmid were transcribed in vitro following the protocol from Promega. Labeled riboprobes were transcribed by inclusion of [␣-32 P]-UTP (800 Ci/mmol; PerkinElmer Life Sciences) in the transcription reaction.
RNA-Protein Binding and Gel Mobility Shift Assay-RNA-protein gel mobility shift assay was performed following previously published methods (36). Briefly, protein extracts (15 g) were incubated with radiolabeled riboprobes (1 ϫ 10 5 cpm; approximately 1 ng of RNA) in 12 mM Hepes, pH 7.9, 15 mM KCl, 5 mM MgCl 2 , 2 mM DTT (unless specified otherwise), 1 g/l tRNA, 1 g/l heparin, 1 unit/l RNasin, and 10% glycerol in a total volume of 20 l for 20 min at 25°C. For competition experiments, the protein extract was first incubated for 10 min at 25°C with unlabeled competitor RNA (specific or nonspecific) or RNA homopolymers (poly(A), poly(C), poly(G), and poly(U)) prior to the addition of the radiolabeled transcript. Control reactions without competitors were sham-treated under identical conditions. For immunodepletion of AUF1, protein extracts were incubated for 1 h at 4°C with or without polyclonal antibody to AUF1 (23), and immune complexes were separated by protein A-Sepharose. An immunoblotting of the supernatants showed that, while AUF1 is present in the supernatant from the control reaction, it was absent in the immunodepleted supernatant (data not shown). Supernatants from control and immunodepleted extracts were used for RNA-protein gel mobility shift assay. RNA-protein binding reactions were treated with RNase T1 (5 units/reaction) and separated by electrophoresis on 4.5% native polyacrylamide gel in 45 mM Tris, 45 mM boric acid, and 1.2 mM EDTA buffer, pH 7.4. The RNA-protein complexes were analyzed by a PhosphorImager (STORM 840, Molecular Dynamics).
UV RNA-Protein Cross-linking-RNA-protein binding reactions as described above were UV-irradiated (2500 J for 3 min) and incubated with RNase A (0.2 g/l) and RNase T1 (10 units) at 37°C for 20 min. Protein samples were analyzed by electrophoresis on 10% SDS-polyacrylamide gel along with prestained protein molecular size markers (Life Technologies) and visualized by exposing the dried gel to a Phos-phorImager screen.
Measurement of Glutathione and N-Acetyl-L-cysteine Levels-Intracellular reduced and oxidized glutathione as well as N-acetyl-L-cysteine (NAC) levels were assayed following previously published protocols (39,40). Cell pellets were homogenized in 50 mM potassium phosphate buffer (pH 7.8) containing 1.34 mM diethylenetriaminepentaacetic acid. Total glutathione content was determined by the method of Anderson (41). Reduced (GSH) and oxidized (GSSG) glutathione were distinguished by the addition of 2 l of a 1:1 mixture of 2-vinylpyridine and ethanol per 30 l of sample. NAC levels in cells were measured following derivatization with N-(1-pyrenyl)maleimide using a 15-cm C18 Reliasil column (Column Engineering, Ontario, CA) coupled with high performance liquid chromatography with fluorescent detection (40). All biochemical determinations were normalized to the protein content of whole cell homogenates using the method of Lowry et al. (42).
Measurement of Intracellular Prooxidant Production during the Cell Cycle-HeLa cells synchronized by mitotic shake-off were replated and harvested at representative times for measurement of prooxidant production (39). Cells were spun down, and the pellets were resuspended in 1 ml of phosphate-buffered saline supplemented with 5.5 mM glucose at 37°C. Samples were labeled with 10 l of 1 mg/ml C-400, an oxidationsensitive fluorescent probe (Molecular Probes, Inc., Eugene, OR) and incubated for 15 min at 37°C and then placed on ice immediately. Labeled samples were analyzed by flow cytometry (excitation at 488 nm, emission at 535 Ϯ 10 nm); 20,000 cells from each sample were analyzed to obtain mean fluorescence intensity and corrected for autofluorescence obtained from samples of unlabeled cells. The variations in mean fluorescence intensity during the cell cycle were calculated relative to early G 1 cells (2 h after mitotic shake-off).

A Regulatory Role for 3Ј-UTR in the Cell Cycle-coupled Accumulation of Topo II mRNA Levels-
To determine if the 3Ј-UTR of Topo II mRNA has a role in cell cycle-specific increase in Topo II mRNA levels, reporter assays in stably transfected 3T3 cells were performed. Stably transfected cells containing H⌬Topo2, H␤galTopo, or H␤galSV40 reporter constructs were synchronized by serum starvation and serum stimulation to reenter the cell cycle. Cells were harvested at various times for analysis of cell cycle position and reporter mRNA levels. Representative univariant flow cytometric analysis of cell cycle positions in H⌬Topo2 transfected cells are presented in Table I. Results presented in Table I indicate that the majority of the cells were in G 1 (80%) at the time of serum stimulation (0 h), and 58% of cells were in S phase at 16 h after serum stimulation. The fraction of cells in S phase increased to 76% by 20 h, and approximately 20 -30% of the cells entered G 2 /M at 24 h after serum stimulation. Similar results were obtained for H␤galTopo-or H␤galSV40-transfected cells and were comparable with untransfected control cells (data not shown), demonstrating that expression of the reporter constructs did not affect cell cycle transit. When the total cellular RNA from the H⌬Topo2-transfected cell populations was analyzed by Northern blotting, an mRNA band of approximately 2.2 kb, representing the H⌬Topo2 reporter construct, was detected with a radiolabeled probe specific for the human Topo II cDNA (Fig.  1A). H⌬Topo2 reporter mRNA levels were low in G 1 (0 -12 h poststimulation) and increased 12-16-fold at mid-S to G 2 phases (16 -24 h poststimulation). These results were comparable with our previously published results obtained with mitotically selected HeLa cells (6) and indicated that the truncated Topo II cDNA reporter construct contains sequence element(s) regulating Topo II mRNA levels. Since the H␤galTopo reporter mRNA levels ( Fig. 1C) also showed similar cell cycle-coupled variations (as compared with H⌬Topo2), we concluded that sequences within the open reading frame of Topo II are not involved in the regulation of mRNA levels. Therefore, the increase in H⌬Topo2 and H␤galTopo reporter mRNA levels might be regulated either by changes in the Topo II promoter activity or by a combination of promoter activity and posttranscriptional regulation governed by the Topo II 3Ј-UTR. Results from the H␤galSV40 reporter construct (Fig.  1C), which contains the Topo II promoter but not the 3Ј-UTR, showed a 6 -8-fold decrease in H␤galSV40 mRNA levels compared with H␤galTopo mRNA levels during 20 -24 h poststimulation. The low level of H␤galSV40 mRNA during 20 -24 h poststimulation can be attributed to the 2-fold increase in Topo II transcription that has been reported previously to occur during late S phase by us (6) and other investigators (34,43). Results from a representative Northern blot measuring the variations in endogenous Topo II mRNA levels are shown in Fig. 1B. The kinetics of endogenous Topo II mRNA accumulation was comparable to reporter mRNA levels during the cell cycle. Overall, these results demonstrate that Topo II 3Ј-UTR has a regulatory role in accumulation of Topo II mRNA levels during the mammalian cell cycle.
A complete sequence analysis of the Topo II 3Ј-UTR was performed to determine if sequences known to regulate mRNA levels are present in the 3Ј-UTR. The hTopo2 plasmid (35) was digested with NdeI and XhoI restriction enzymes and cloned into pBS II SK(ϩ), and both strands were sequenced. Sequence analysis showed that the Topo II mRNA has a long (ϳ1.1-kb) 3Ј-UTR with several ARE motifs (four AUUUA, one AUUUUA, and one AUUUUUA; Fig. 2) present throughout the 3Ј-UTR. These results suggest that AREs in Topo II 3Ј-UTR could participate in the regulation of Topo II mRNA levels during the cell cycle.
Specific Sequences in Topo II 3Ј-UTR mRNA Bind to Proteins in Cellular Extracts-Because protein binding to 3Ј-UTR has been shown to regulate mRNA levels (23,(27)(28)(29), an in vitro RNA-protein binding assay was performed to determine if cellular proteins bind to the Topo II 3Ј-UTR. A series of overlapping [ 32 P]UTP-labeled riboprobes representing the proximal, middle, and distal regions of Topo II 3Ј-UTR mRNA were transcribed in vitro and incubated with cellular protein extracts prepared from asynchronously growing HeLa cells. RNA-protein complex formation was analyzed by gel mobility shift as-say. Results presented in Fig. 3 show that the mobility of all three transcripts was retarded in their migration when compared with the control reaction that did not contain the protein extract (compare lane 1 with lane 2, lane 6 with lane 7, and lane 12 with lane 13). The retardation in mobility was due to an RNA-protein interaction because proteinase K treatment of cellular extracts prior to the binding assay abolished the retardation (lane 11). To analyze the specificity of the RNA-protein complex, competition experiments were performed. The RNAprotein binding assay was carried out in the presence of an unlabeled specific competitor or a nonspecific competitor RNA (␤-galactosidase). The addition of a 1-fold excess of specific unlabeled competitor RNA had a minimal effect on RNA-protein complex formation (lanes 3, 8, and 14). However, a 10-fold excess of specific unlabeled proximal and middle transcripts decreased the RNA-protein complex formation by more than 90% (lanes 4 and 9). In contrast, a similar 10-fold excess of the unlabeled distal transcript did not compete with the labeled distal transcript for complex formation. A 10-fold excess of the nonspecific competitor RNA (␤-galactosidase) caused no change in complex formation (lanes 5, 10, and 16). These results demonstrate that protein binding to both the proximal and middle portions of the Topo II 3Ј-UTR is specific in contrast to nonspecific binding to the distal region. Alternatively, the protein(s) bound to the distal region could be highly abundant and therefore might not be competed away under these conditions.
To further localize the sequence region involved in Topo II 3Ј-UTR RNA-protein complex formation, cross-competition assays were performed in the presence of unlabeled transcripts from the proximal or the middle region. Cellular proteins bound to labeled proximal transcript could be competed more than 90% with a 10-fold excess of unlabeled proximal competitor RNA (compare lanes 2 and 3, Fig. 4A). Interestingly, a 10-fold excess of unlabeled competitor RNA representing the middle region of Topo II 3Ј-UTR mRNA also competed (lane 4), while a nonspecific competitor RNA (␤-galactosidase, lane 5) did not. Alternatively, the RNA-protein complex formed from the middle region of Topo II 3Ј-UTR mRNA (Fig. 4B, lane 7) could be competed away with 10-fold excess of unlabeled middle competitor RNA (lane 8) or unlabeled RNA from the proximal region of Topo II 3Ј-UTR (lane 9). These results indicate that the overlapping sequence spanning the proximal and middle regions of Topo II 3Ј-UTR may contain the motifs necessary for specific RNA-protein complex formation.
To more precisely define the binding site, a riboprobe representing the 177-nt (base pairs 4772-4948; Fig. 2) sequence overlapping the proximal and middle region of Topo II 3Ј-UTR was used in the RNA-protein binding assay. The 177-nt transcript, when incubated with cellular protein extract, formed an RNA-protein complex (Fig. 4C, compare lane 11 with lane 12). RNA-protein complex formation could be competed with unlabeled proximal and middle transcripts, indicating that this region of Topo II 3Ј-UTR mRNA is sufficient for RNA-protein complex formation (data not shown). To further examine RNA binding specificity, RNA homopolymers were tested for the ability to compete for binding to the 32 P-labeled 177-nt Topo II 3Ј-UTR RNA substrate. Protein extract was incubated for 10 min at 25°C with 1, 10, and 100 ng of poly(U) (lanes 13-15), poly(C) (lanes 16 -18), poly(A) (lanes 19 -21), or poly(G) (lanes [22][23][24] prior to the addition of the 32 P-labeled 177-nt transcript. While poly(C) and poly(G) RNA homopolymers had no effect on Topo II 3Ј-UTR RNA-protein binding (lanes 16 -18 and 22-24), both 10 and 100 ng of poly(U) inhibited more than 90% of the RNA-protein complex formation (lanes 14 and 15, Fig. 4C). Although competition assay with 1 and 10 ng poly(A) did not inhibit protein binding (lanes 19 and 20), 100 ng of poly(A) showed approximately 40% inhibition in complex formation (lane 21, Fig. 4C). The competition with poly(A) at higher concentration could be due to the formation of a doublestranded region with uridylate residues present within the 177-nt sequence. Furthermore, unlabeled c-fos ARE transcript containing the AUUUUUA motif also competed with 177-nt Topo II 3Ј-UTR for protein binding (data not shown). These results indicate that at least some of the proteins bound to the 177-nt overlapping transcript have poly(U) binding activity, suggesting a possible role for the AUUUUUA motif in complex formation.
Characterization of the Topo II 3Ј-UTR RNA-Protein Complex Formation-If protein binding to the Topo II 3Ј-UTR influences Topo II mRNA levels, then it is reasonable to postulate that variations in Topo II mRNA levels during the cell cycle may be associated with changes in proteins binding to the UTR. Specific protein binding to the 177-nt Topo II 3Ј-UTR RNA was analyzed by a UV cross-linking assay. Protein extracts prepared from G 1 (2 and 4 h after mitotic shake-off) and S (16 h after mitotic shake-off) phase cells were incubated with the 177-nt riboprobe and UV-irradiated, and unbound RNA was digested with RNases A and T1. Proteins were separated by SDS-gel electrophoresis and visualized by autoradiography. In the absence of UV irradiation, no cross-linking of labeled probe to protein was observed (Fig. 5, lane 1). In contrast, several polypeptide bands of apparent molecular masses of 84, 70, 44, and 37 kDa (Fig. 5, lanes 2-4) were detected in UV-irradiated reactions. Protein binding was specific and could be competed with 10 ng of poly(U) (lane 5). Furthermore, while the binding of the 70-and 44-kDa polypeptides was detected in extracts from both G 1 and S phase cells, binding of the 84-and 37-kDa polypeptides was specific to the G 1 phase. These results support the idea that differential binding of proteins to the Topo II 3Ј-UTR might regulate Topo II mRNA accumulation during the cell cycle.
Since the molecular mass of the 37-kDa polypeptide (Fig. 5A) is similar to that of the ARE-binding protein AUF1 (23), and the 177 nt of Topo II 3Ј-UTR contains an AUUUUUA motif, we determined whether AUF1 is present in the RNA-protein complex. Protein extract prepared from asynchronously growing HeLa cells was incubated with or without a polyclonal antibody to AUF1, and the immune complex was separated using protein A-Sepharose. Immunoblotting of the supernatants showed that, while AUF1 is present in the supernatant from the control reaction, it was not detected in the supernatant from the immunoprecipitation reaction performed with the AUF1 polyclonal antibody (data not shown). RNA-protein binding assays were then performed using the control or AUF1-immunodepleted supernatant. While the supernatant from the control reaction showed no change in complex formation, depletion of AUF1 from the protein extract inhibited complex formation by more than 90% (compare lanes 1 and 2, Fig. 5B). Binding reactions performed with bovine serum albumin alone (lane 3)

FIG. 2. Several AU-rich elements are present in Topo II 3-UTR.
Sequence analysis of Topo II 3Ј-UTR cDNA. Boldface letters represent AU-rich elements. Translation termination and polyadenylation signals are in boldface type and underlined. Uppercase letters represent coding sequence, and lowercase letters represent 3Ј-untranslated region. The putative protein-binding region is double underlined. showed no retardation in mobility of the transcript and served as a negative control for the assay. The presence of AUF1 in the Topo II 3Ј-UTR RNA-protein complex was further verified by comparing binding reactions with c-fos 3Ј-UTR transcript as a positive control. RNA-protein binding reactions with the 177-nt Topo II 3Ј-UTR transcript, c-fos 3Ј-UTR transcript, or a control reaction without any transcript were first resolved by RNAprotein gel shift assay and excised under autoradiographic guidance. In the control lane, an identical region of the gel compared with the radioactive band was also excised. Bound proteins in the excised bands were separated by SDS-gel electrophoresis, transferred to Immobilon-P membrane, and immunoblotted with a polyclonal antibody to AUF1. The control reaction without the transcript did not show any AUF1 immunoreactive polypeptide band (Fig. 5C, lane 1). In contrast, AUF1 was detected in both the 177-nt Topo II and c-fos 3Ј-UTR RNA-protein complexes (lanes 2 and 3). These results identify AUF1 as one of the proteins present in the Topo II 3Ј-UTR RNA-protein complex and suggest that the interaction of AUF1 with the AUUUUUA motif in Topo II 3Ј-UTR may have a regulatory role in determining Topo II mRNA levels during the cell cycle.
Oxidation/reduction Sensitivity of Protein Binding to the Topo II 3Ј-UTR-Since proteins that bind AREs have been shown previously to demonstrate alterations in binding based on the oxidation/reduction (redox) state of the protein complexes (31,32), redox sensitivity of protein binding to the 177-nt Topo II 3Ј-UTR was investigated (Fig. 6). Protein extract was prepared from asynchronously growing HeLa cells using extraction buffer that lacks a reducing agent. Binding reactions were then performed with increasing concentrations of the thiol-reducing agent DTT, or 2-ME, prior to the addition of the 177-nt riboprobe and analyzed by RNA gel shift assay (Fig. 6A,  lanes 2-5). Binding of cellular proteins to the Topo II 3Ј-UTR increased 5-6-fold in protein extracts pretreated with the reducing agents DTT (lanes 3 and 4) or 2-ME (lane 5). The assay was then repeated in extracts supplemented with 2 mM DTT and increasing concentrations of a thiol-oxidizing agent (diamide) or a thiol-alkylating agent, NEM (Fig. 6B). A dose-dependent inhibition in protein binding was observed in extracts that were pretreated with the sulfhydryl-oxidizing agent, diamide (1, 5, 10, and 20 mM; lanes 6 -9). While 1 and 5 mM of diamide caused 40 and 90% decreases in protein binding ( lanes  6 and 7), both 10 and 20 mM of diamide completely abolished the complex formation (lanes 8 and 9). The inhibition in protein binding in 5 mM diamide-treated extract could be reversed by  lanes 1-5), middle (lanes 6 -11), or distal (lanes 12-16) region of Topo II 3Ј-UTR mRNA. Radiolabeled riboprobes (1 ϫ 10 5 cpm; approximately 1 ng of RNA) were incubated with 15 g of protein extract prepared from asynchronously growing HeLa cells. Binding reactions were performed in a total volume of 20 l and incubated for 20 min at 25°C. For competition experiments, the protein extract was first incubated for 10 min at 25°C with unlabeled competitor RNA (specific or nonspecific) prior to the addition of the radiolabeled transcript. Control reactions without competitors were sham treated under identical conditions. Binding mixtures were treated with RNase T1 prior to electrophoresis. C, RNA-protein complex; F, free probe; Comp., specific competitor RNA; NS, nonspecific ␤-galactosidase competitor; Prot.K, proteinase K; 1ϫ, 1 ng of RNA.  1, 6, and 11, binding reactions without the protein extract; lanes 2, 7, and 12, binding reactions with the protein extract prepared from asynchronously growing HeLa cells. Competition assays with a 10-fold (10 ng) excess of the specific unlabeled competitor (lanes 3 and 8), middle (lane 4) and proximal (lane 9) cross-competitor, and the nonspecific competitor ␤-galactosidase (lanes 5 and 10) were performed following methods described in Fig. 3. Lanes 13-24 represent competition assays with increasing concentrations of poly(U) (lanes [13][14][15], poly(C) (lanes 16 -18), poly(A) (lanes 19 -21), or poly(G) (lanes 22-24) RNA homoploymers. Protein extracts were incubated with RNA homopolymers for 10 min at 25°C prior to the addition of the radiolabeled transcript. M, middle; P, proximal Topo II 3ЈUTR RNA; NS, nonspecific competitor. treatment with the thiol-reducing agent, 2-ME (compare lanes 7 and 10). Treatment with NEM also showed a dose-dependent inhibition in protein binding to Topo II 3Ј-UTR (1 and 10 mM, lanes 11 and 12). Because thiol alkylation is an irreversible reaction, inhibition in RNA-protein binding by NEM could not be reversed by the addition of DTT or 2-ME to the reaction mixture (data not shown). Taken together, these results indicate that reduced thiol residues in Topo II 3Ј-UTR binding proteins participate in the RNA-protein complex formation and that oxidation of these thiol residues to the disulfide form abolishes binding.
If the above in vitro results are relevant to in vivo regulation, then it is reasonable to postulate that manipulations of intracellular redox status would influence Topo II 3Ј-UTR RNAprotein binding and mRNA levels. To address this question, asynchronously growing HeLa cells were treated with 20 mM NAC and assayed for NAC uptake, reduced and oxidized glutathione content, RNA-protein binding, and Topo II mRNA levels. By 2 h, intracellular NAC uptake was approximately 4 nmol/mg protein and increased to almost 6 nmol/mg protein by 6 h of treatment (Fig. 7A). Consistent with these results, the ratio of reduced to oxidized glutathione increased approximately 1.5-to 2.0-fold during this time frame, demonstrating that NAC treatment altered the intracellular redox state in favor of a more reducing environment. Topo II 3Ј-UTR mRNA-protein gel mobility shift assays performed with protein extracts from duplicate samples showed that protein binding increased 2-3-fold in NAC-treated cells compared with untreated controls (compare lanes 2-4, Fig. 7B). Results from Fig.  7B (inset) showed that the increase in protein binding was not due to changes in AUF1 protein levels following NAC treatment. Interestingly, Topo II mRNA levels under these conditions decreased dramatically to 10% of control at 2-6 h after NAC treatment (Fig. 7, C and D). These results provide in vivo evidence that a shift to a more reducing environment enhances protein binding to the 177-nt Topo II 3Ј-UTR, which correlates with a decrease in mRNA levels.
If these apparent redox-sensitive alterations in Topo II 3Ј-UTR mRNA-protein complex formation are a reflection of events occurring during the cell cycle, then a lower oxidation potential in G 1 (relative to S, G 2 , and M) would favor Topo II 3Ј-UTR mRNA-protein binding, resulting in decreased mRNA levels. In contrast, a relatively more oxidizing environment in late S, G 2 , and M would inhibit complex formation, resulting in increased mRNA levels. To determine if variations in redox potential occur during the cell cycle, synchronized HeLa cells collected by mitotic shake-off were replated and harvested at representative times for analysis of prooxidant production. Cells were stained with an oxidation-sensitive fluorescent probe (C-400) and analyzed by flow cytometry. Oxidation of the probe 1) was greatest in cells immediately following mitotic shake-off (0 h postmitotic shake-off); 2) was minimal in G 1 cells (2-7 h); and 3) increased 3-5-fold in late S/G 2 cells (14 -20 h) relative to G 1 (Fig. 8A). An analysis of Topo II mRNA levels at representative time points during the cell cycle showed that the lower oxidation potential during the G 1 phase (Fig. 8A) correlated with a decrease in Topo II mRNA levels (Fig. 8B). Similarly, an increase in oxidation potential during S/G 2 phase correlated with an increase in Topo II mRNA levels. These observations are consistent with the results from Figs. 6 and 7, suggesting that a relatively reducing environment during G 1 favors protein binding to the Topo II 3Ј-UTR and correlates with a decrease in Topo II mRNA levels. In contrast, a relatively greater oxidation potential during S/G 2 phase would be expected to inhibit protein binding to Topo II 3Ј-UTR, resulting in increased Topo II mRNA levels.
The idea that the 3Ј-UTR containing the ARE motifs participates in the regulation of mRNA levels during the cell cycle was further investigated by measuring mRNA levels of two other ARE-containing cell cycle genes, p21 (37) and geminin (38). Both mRNAs contain AREs including the AUUUUUA motif in their 3Ј-UTRs. HeLa cells synchronized by mitotic shake-off were replated, and total cellular RNA isolated at 4, 12, 18, and 24 h after plating. Cells from duplicate dishes were assayed for cell cycle position by flow cytometric determination of DNA content. Approximately 85% of the cells were in G 1 at 4 h after plating, and by 12 h 76% of the cells were in S phase (Fig. 9A). At 18 h after plating, 42% of the cells entered G 2 , and by 24 h approximately 65% of the cells entered G 1 of the next generation. Consistent with previous reports, Northern blot analysis showed that Topo II mRNA levels were low in G 1 (lane 1) and increased 15-fold during late S and G 2 (lanes 2 and 3,  Fig. 9B). Following cell division, Topo II mRNA levels decreased to basal levels (lane 4) in G 1 of the next generation. Interestingly, both geminin and p21 mRNA levels were low in G 1 (lane 1) and increased approximately 2-(geminin) and 5-fold (p21) during late S and G 2 (lanes 2 and 3). The mRNA levels of both geminin and p21 decreased during G 1 of the subsequent generation (lane 4). These results support the idea that AREs Protein extract prepared from asynchronously growing HeLa cells was incubated with or without the AUF1 polyclonal antibody for 1 h at 4°C. Immune complexes were separated by protein A-Sepharose, and the supernatants were used in the RNA-protein binding assay. Lane 3 represents a negative control for the assay that was performed with 15 g of bovine serum albumin. C, immunoblot analysis of AUF1 protein in RNA-protein complexes from the 177-nt Topo II 3Ј-UTR (lane 2) and c-fos ARE (lane 3) riboprobes. Lane 1 represents a control binding reaction that was carried out without the radiolabeled transcript under identical conditions. RNA-protein binding reactions were carried out with protein extract isolated from asynchronously growing HeLa cells and resolved by RNA-protein gel shift assay; excised and bound proteins were separated by SDS-gel electrophoresis. Proteins were transferred to Immobilon-P membrane and immunoblotted with a polyclonal antibody to AUF1. Immunoreactive bands were visualized by the chemiluminescence method following the supplier's protocol (Amersham Pharmacia Biotech).
including the AUUUUUA motif present in the 3Ј-UTRs of Topo II, geminin, and p21 could possibly contribute to the regulation of mRNA accumulation during the cell cycle. DISCUSSION We have shown previously that Topo II mRNA levels increase 15-16-fold during late S and G 2 compared with the G 1 phase of the HeLa cell cycle and that variations in Topo II mRNA levels are associated with significant changes in mRNA stability (6). The present study investigates whether Topo II 3Ј-UTR has a regulatory role in cell cycle-coupled accumulation of mRNA levels. Using a reporter assay in stably transfected 3T3 cells, we have shown that the Topo II 3Ј-UTR can confer cell cycle regulation in reporter mRNA levels. A sequence analysis showed that Topo II mRNA contains a 1.1-kb 3Ј-UTR with several known mRNA instability elements (AREs) present throughout the entire UTR. Furthermore, results from in vitro RNA-protein binding assays demonstrate that cellular proteins bind to specific sequences within the Topo II 3Ј-UTR, and protein binding to a 177-nt Topo II 3Ј-UTR transcript containing a AUUUUUA motif varies during the cell cycle. These results indicate that the AUUUUUA motif within the 177-nt Topo II 3Ј-UTR transcript may have a regulatory role in cell cycle control of Topo II mRNA accumulation. This idea is further supported by results from Fig. 9, demonstrating that mRNA accumulation of two other AUUUUUA-containing cell cycle genes (geminin and p21) also varied during the cell cycle, and similar to Topo II their mRNA levels were low in G 1 . This is intriguing because, in general, the pentameric AUUUA motif(s) have been shown to regulate mRNA levels (18,27,29,30). Although the role of heptamer ARE motifs in the regulation of mRNA levels has not been reported previously, it is well known that the nonamer ARE motif (UUAUUUAUU) can act as a potent mRNA destabilizer (21). Our results suggest that the potential influence of ARE-mediated mRNA instability may extend beyond the control of cytokine and proto-oncogene gene expression and include other growth-regulatory genes. Future studies will determine if the AUUUUUA motif by itself or in combination with other AREs or non-AREs confers cell cycle regulation of Topo II, geminin, and p21 gene expression.
Cytokine and proto-oncogene mRNA decay are thought to involve an AU-rich binding protein AUF1, which complexes with heat shock proteins hsc70-hsp70, translation initiation factor eIF4G, and poly(A)-binding protein (44). Multiple proteins with apparent molecular masses of 84, 70, 44, and 37 kDa FIG. 6. Protein binding to the 177-nt riboprobe is sensitive to oxidation and reduction reactions in vitro. RNA-protein gel shift assay is shown of the 177-nt Topo II 3Ј-UTR riboprobe and HeLa cellular protein extracts pretreated with thiol-reducing or -oxidizing agents. A, protein extracts from asynchronously growing HeLa cells were prepared using extraction buffer without any DTT. RNA-protein binding reactions were carried out with extracts that were pretreated with varying concentrations of DTT (0, 2, and 10 mM; lanes 2-4) or 2-ME (1%, lane 5). B, protein extracts from asynchronously growing HeLa cells were prepared with regular extraction buffer containing 2 mM DTT. RNA-protein binding reactions were carried out in the presence of increasing concentrations of diamide (1, 5, 10, and 20 mM; lanes 6 -9) or NEM (1 and 10 mM; lanes 11 and 12). Lane 1 represents a probe without any extract, and lane 10 represents 2-ME treatment of a 5 mM diamide-treated extract. Asynchronously growing HeLa cells were treated with 20 mM NAC (pH 7.0) and assayed for NAC uptake and reduced and oxidized glutathione content (A). Open circles and solid squares represent NAC uptake and the ratio of reduced to oxidized glutathione levels, respectively. Dashed and dotted lines represent control and NAC-treated cells, respectively. B, RNA-protein gel mobility shift assay with the 177-nt Topo II 3Ј-UTR transcript and protein extracts prepared from cells in duplicate dishes. Proteins were extracted with extraction buffer without DTT. Inset represents immunoblotting of AUF1 in protein extracts isolated from control and 6-h NAC-treated cells. C, Topo II mRNA levels in replicate dishes were analyzed by Northern blot analysis. The blot was rehybridized with a radiolabeled glyceraldehyde-3-phosphate dehydrogenase (GAPD) cDNA probe for quantitation. D, quantitation of mRNA levels was performed using the PhosphorImager and plotted as a function of time of treatment with NAC. bind to Topo II 3Ј-UTR and identified AUF1 as one of the proteins present in the RNA-protein complex. Since the 37-kDa protein has a molecular mass similar to AUF1 and its binding was enhanced in G 1 (Fig. 5A), the possibility that enhanced binding could be related to changes in absolute quantities of AUF1 during the cell cycle was investigated. An immunoblot analysis of AUF1 in whole HeLa cell extracts prepared from cells in G 1 and S showed that the absolute quantities of immunoreactive AUF1 protein did not vary during the cell cycle (data not shown). Thus, the absolute amount of AUF1 protein does not appear to influence Topo II mRNA levels, but posttranslational modifications (e.g. phosphorylation and/or oxidation/reduction) might regulate AUF1 activity during the cell cycle. Since AUF1 consists of four protein isomers, our results do not eliminate the possibility that other isomers of AUF1 may be involved in the cell cycle regulation of Topo II mRNA levels. Nonetheless, these results indicate the potential role of AUF1 in ARE-mediated regulation of cell cycle gene expression.
Our results further indicated that the protein binding to Topo II 3Ј-UTR is sensitive to redox reactions both in vitro and in vivo. In vitro protein binding to the 177-nt Topo II 3Ј-UTR was found to be reversibly enhanced by thiol-reducing agents and inhibited by thiol-oxidizing agents. These results support earlier in vitro reports demonstrating redox sensitivity of protein binding to ARE-containing cytokine and lymphokine mRNAs (27). In vivo experiments in which cells were treated with a thiol-reducing agent showed that a shift to a more reducing intracellular environment resulted in increased RNAprotein binding as well as more than 90% reduction in Topo II mRNA levels (Fig. 7). Taken together, these results indicate that thiol residues (presumably cysteine residues) in AREbinding proteins (e.g. AUF1) participate in the binding to the Topo II 3Ј-UTR mRNA and that reduction of these thiol resi-dues is required for binding. Since AUF1 has three cysteine residues in its two RNA binding domains (23), it is possible that one or more of these cysteine residues are involved in redox sensitivity of RNA-protein binding during the cell cycle. Thus, subtle changes in redox states during cell growth or in response to environmental stimuli may influence Topo II 3Ј-UTR mRNA-protein interactions, thereby affecting mRNA levels. Consistent with this assumption, a lower oxidation potential during G 1 in synchronized cells was associated with more protein binding to Topo II 3Ј-UTR, correlating with decreased mRNA levels. Likewise, a relatively more oxidizing environment during S and G 2 was associated with less protein binding, correlating with increased Topo II mRNA levels. The variations in intracellular redox states (Fig. 8) are consistent with previous findings (45), suggesting that a relatively high intracellular oxidation potential is associated with progression toward mitosis. Furthermore, our results also support the idea that proteins regulating cell cycle progression are sensitive to the redox potential of the microenvironment (46,47).
In summary, results from this study support the hypothesis that a posttranscriptional mechanism regulated by the interaction of 3Ј-UTR mRNA with redox-sensitive protein complexes containing AUF1 may regulate Topo II mRNA accumulation during the cell cycle.
FIG. 8. A higher intracellular prooxidant production during S and G 2 correlates with increased Topo II mRNA levels. Cells synchronized by mitotic shake-off were replated and harvested at various times for measurement of prooxidant production, cell cycle position, and mRNA levels. A, cells were stained with the C-400 fluorescent probe and analyzed by flow cytometry. Cells from replicate dishes were analyzed for cell cycle position by flow cytometric assay of DNA content in propidium iodide-stained cells. The left y axis represents the fraction of cells at various hours after mitotic shake-off, and the mean fluorescence intensity (MFI) of the C-400 probe after normalization to autofluorescence and relative to early G 1 is plotted on the right y axis. B, Northern blot analysis of Topo II and glyceraldehyde-3-phosphate dehydrogenase (GAPD) mRNA levels harvested from replicate dishes under similar experimental conditions. FIG. 9. Messenger RNA levels of ARE-containing cell cycle genes varied during the cell cycle. HeLa cells synchronized by mitotic shake-off were harvested at the indicated times for flow cytometric determination of DNA content (A) and Northern blot analysis (B). Cells were stained with propidium iodide and analyzed for DNA content following a previously published procedure (6). DNA contents of G 1 and G 2 are marked with arrows. For Northern blot analysis, blots were hybridized with random-labeled cDNA probes for human Topo II, geminin, and p21. Ethidium bromide staining of 28 S ribosomal RNA was used for loading control.