Antioxidant-induced Nuclear Translocation of CCAAT/Enhancer-binding Protein β

Alterations in intracellular oxidative status activate several signal transduction pathways resulting in distinct patterns of gene expression. Treatment of colorectal cancer cells with antioxidants can lead to apoptosis by induction of p21 through a mechanism involving CCAAT/enhancer-binding protein β (C/EBPβ). Herein, we demonstrate that the antioxidant pyrrolidinedithiocarbamate activates cAMP-dependent protein kinase (PKA) in a colorectal cancer cell line DKO-1. Activation of PKA phosphorylates Ser299 within C/EBPβ, which is essential for protein translocation to the nucleus. Pharmacological inhibition of PKA and mutation of Ser299 to alanine blocks C/EBPβ nuclear translocation and induction of p21. Our results indicate that a cAMP-dependent phosphorylation of C/EBPβ at Ser299 is critical for nuclear translocation of this protein and its subsequent transactivation of genes in response to antioxidant treatment.


Alterations in intracellular oxidative status activate several signal transduction pathways resulting in distinct patterns of gene expression. Treatment of colorectal cancer cells with antioxidants can lead to apoptosis by induction of p21 through a mechanism involving CCAAT/enhancer-binding protein ␤ (C/EBP␤). Herein, we demonstrate that the antioxidant pyrrolidinedithiocarbamate activates cAMP-dependent protein kinase (PKA) in a colorectal cancer cell line DKO-1. Activation of PKA phosphorylates Ser 299 within C/EBP␤, which is essential for protein translocation to the nucleus. Pharmacological inhibition of PKA and mutation of Ser 299 to alanine blocks C/EBP␤ nuclear translocation and induction of p21. Our results indicate that a cAMP-dependent phosphorylation of C/EBP␤ at Ser 299 is critical for nuclear translocation of this protein and its subsequent transactivation of genes in response to antioxidant treatment.
Antioxidants have been associated with a diminished risk of cancer at various anatomical sites, including the colon (1,2). The primary mechanism of chemoprevention by antioxidants is through the reduction of DNA-damaging free radicals (3). We have reported that two antioxidants, pyrrolidinedithiocarbamate (PDTC) 1 and vitamin E, induce G 1 cell cycle arrest and apoptosis in various human cancer lines including breast, colon, and lung (4). These cell cycle perturbations were mediated by induction of p21, a powerful inhibitor of the cell cycle, through a mechanism involving activation and binding of C/ EBP␤ to the p21 promoter.
C/EBP␤ is a member of a diverse group of nuclear transcription factors that contain a leucine zipper motif required for dimer formation and a basic DNA-binding domain that facilitates interactions between these factors and the regulatory domains of promoters and/or enhancers of target genes (5)(6)(7)(8)(9)(10)(11).
C/EBP␤ activates several acute phase protein genes through the NF-IL6 responsive elements (8,12) and also has been implicated in adipocyte differentiation and inflammatory and immune responses (13). Thus, C/EBP␤ is a pleiotropic transactivator involved in a myriad of signal transduction and cell differentiation events.
Control of C/EBP␤ expression and activity is complex and poorly understood. It is known that C/EBP␤ gene is transcriptionally activated by IL-1 and lipopolysaccharide (5), whereas in other instances its binding to cognate DNA sequences is enhanced by cytokines (5,11). Additionally, C/EBP␤ can be a target for post-translational modification. Various kinases, including cAMP-dependent protein kinase (PKA) (14), protein kinase C (PKC) (15), a Ras-dependent MAP kinase (16), and a calcium calmodulin-dependent kinase (17) have been shown to phosphorylate C/EBP␤ in vitro. These phosphorylation events modulate DNA binding and transcriptional activity of C/EBP␤. However, it is unknown if the phosphorylation status of C/ EBP␤ influences its subcellular compartmentalization. Elevation of cAMP levels in PC-12 cells or activation of tumor necrosis factor receptors in hepatocytes leads to a redistribution of C/EBP␤ from the cytosolic to nuclear compartment (9,18). As a first step toward understanding the antioxidant-mediated increase in C/EBP␤ DNA binding activities, we evaluated the effect of the antioxidant PDTC on the post-translational modification and subcellular localization of C/EBP␤ in DKO-1 cells, a human colorectal cancer cell line. Our results indicate that PDTC induces a rapid and sustained translocation of C/EBP␤ protein from the cytoplasm to the nucleus, resulting in induction of p21. Moreover, we demonstrate that these antioxidantmediated events are regulated by PKA-mediated phosphorylation of Ser 299 in C/EBP␤. RNA Isolation and Northern Blot Analysis-RNA was extracted by the method of Schwab et al. (19). Poly (A) ϩ mRNA was separated by electrophoresis through 1% (w/v) agarose-formaldehyde gels, and Northern blotting was performed as described previously (20). A human p21 cDNA probe was labeled with [␣ 32 P]dCTP by the random primer extension method. Hybridization and post-hybridization washes were carried out at 43°C. IB15 was used as a control for equivalent loading and transfer (21).
Immunoprecipitations and Western Blot Analysis-For immunoprecipitations, cells were washed twice in ice-cold phosphate-buffered saline containing 1 mM Na 3 VO 4 and 100 g/ml phenylmethylsulfonyl fluoride and lysed in RIPA buffer (50 mM Tris-Cl, pH 7.4, 200 mM NaCl, 2 mM EDTA, 0.5% SDS, 0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml pepstatin, and 2 g/ml leupeptin). Alternatively, nuclear proteins were isolated as described (22). Resulting extracts were precleared with protein A-Sepharose (Pierce) for 15 min at 4°C. Following addition of the agarose-conjugated anti-FLAG (M2), antibody-immune complexes were allowed to form during a 2-h rotation at 4°C. Finally, beads were washed five times in RIPA buffer prior to SDS-PAGE.
Cellular extracts (50 -100 g, as determined by Bradford analysis) were resolved on 12% SDS-PAGE gels and transferred to 0.2-m pore nitrocellulose membranes (Schleicher & Schuell). The blots were probed with anti-p21 or C/EBP␤ antibodies, and immunoreactive proteins were visualized by chemiluminescense (Amersham).
Electrophoretic Mobility Shift Assay (EMSA)-EMSA was performed as described previously (22). Each 20-l reaction mixture contained 5 g of nuclear protein plus a ␥-32 P-labeled 20-base pair oligonucleotide probe containing the C/EBP␤-binding site in the p21 promoter (5Ј-GTACTTAAGAAATATTGAAT-3Ј). The reaction mixture was incubated at room temperature for 10 min and loaded directly onto a 6.5% polyacrylamide (49:0.6 acrylamide/bisacrylamide) gel in a buffer of 25 mM Tris borate, pH 8.0, 0.25 mM EDTA. In some experiments, antiserum specific for unique C/EBP isoforms or a mutant C/EBP oligonucleotide (5Ј-GTACAAAAGAAATATTGAAT-3Ј) was added to reaction mixtures , or ␦ (lane 15) antibodies as described under "Experimental Procedures." B, DKO-1 cells were treated with PDTC for the indicated times, and C/EBP␤ mRNA (3 g) and protein (100 g of protein) levels were evaluated by Northern and Western blot analyses, respectively. IB15 is shown as a control for equivalent mRNA loading and transfer. C, DKO-1 cells were treated with PDTC in the presence of [ 32 P]orthophosphate and, at the indicated times, C/EBP␤ protein purified from the cytosol and nucleus by immunoprecipitation. Treatment-related variations in C/EBP␤ localization were analyzed by SDS-PAGE followed by autoradiography. D, DKO-1 cells were grown in the presence of PDTC for 3 h and then processed for immunofluorescence (see "Experimental Procedures") to detect differences in the compartmentalization of endogenous C/EBP␤ protein (red). Cellular nuclei were counterstained with 4Ј,6-diamidino-2-phenylindole (blue). Pink represents co-localization of red and blue signals. In all experiments, parallel cultures treated with anti-C/EBP␤ antisera that had been preincubated with in vitro translated C/EBP␤ protein demonstrated no fluorescent (red) signal (left panel). 10 min prior to the addition of radiolabeled probe. In all experiments, proteins were separated by electrophoresis at 200 V for 2 h at room temperature. Gels were dried and exposed to Kodak XAR film.
Construction of FLAG-C/EBP␤ and Mutants-An 8-amino acid FLAG epitope tag was introduced at the N terminus of human C/EBP␤ in pBluescript by the polymerase chain reaction using 5Ј-CCCAAGCT-TCCACGATGGACTACAAAGACGATGACGATAAAATGCAACGCC-TGGTGGCCT-3Ј as the sense primer and an internal C/EBP␤ sequence (5Ј-GGCTCGTAGTAGAAGTTGGCCACTTCCA-3Ј) as the antisense primer. The 133-base pair polymerase chain reaction fragment was then exchanged with the native N-terminal C/EBP␤ sequence by digestion at the flanking NcoI/Msc I sites. The 1.7-kilobase pair FLAG-C/ EBP␤ coding sequence was subcloned from pBluescript into pCB6. Mutagenesis of the putative PKA recognition sites (Ser 277 or 299 to Ala 277 or 299 ) was performed using the Muta-Gene M13 In Vitro mutagenesis kit (Bio-Rad), and the presence of the desired base pair change was verified by double-stranded DNA sequencing.
Generation of Stably Transfected C/EBP␤ Clones-DKO-1 cells were transfected with 3 g of pCB6-C/EBP␤ or selected mutants using CELLFECTIN according to manufacturer's instructions (Life Technologies, Inc.). After 24 h, cells were shifted to medium supplemented with 1 mg/ml Geneticin (Life Technologies, Inc.) to select for transfected clones. Expression of C/EBP␤ protein was determined by FLAG immunoprecipitation and subsequent Western blot analysis.
Transient Transfections and Reporter Assays-Transfection of DKO-1 cells and subsequent reporter assays were carried out as described previously (4). Total DNA transfected was kept constant with the addition of pCMV-basic. pCMV-CAT was used as an internal control for gene expression (Promega).
In Vivo Phosphorylation, Tryptic Peptide Mapping, and Phosphoamino Acid Analysis-Prior to labeling, parental DKO-1 cells or stable transfectants containing an epitope-tagged C/EBP␤ expression construct were grown to 50% confluence on 100-mm polysine-coated plates and washed three times with serum-free, phosphate-free minimum essential medium (Life Technologies, Inc.). Cells were radiolabeled for 3 h in the same medium containing [ 32 P]orthophosphate (5 mCi/ml) in the presence of PDTC (70 M), mPKI (1 M), or forskolin (50 M). Epitope-tagged proteins were immunoprecipitated from total cell or nuclear extracts using agarose-conjugated FLAG antibodies, resolved by SDS-PAGE, and visualized by autoradiography.
Following identification of the C/EBP␤, bands were excised, and protein was eluted and analyzed as described elsewhere (24). Phosphotryptic peptides were analyzed in the first dimension by thin-layer electrophoresis at pH 1.9 and in the second dimension by ascending chromatography in n-butanol/pyridine/acetic acid/water (75:50:15:60). Phosphopeptides eluted from thin-layer plates were hydrolyzed in 6 N HCl at 110°C for 1 h, mixed with phosphoamino acid standards, and separated electrophoretically at pH 3.5 (25).
Measurement of cAMP Levels and PKA Activity-DKO-1 cells were grown until 80% confluence and incubated with PDTC or forskolin for up to 24 h. At the desired time points, cells were extracted with 4 mM EDTA and boiled for 5 min, and the amount of cAMP in the cell lysates was determined using the [ 3 H]cAMP radioimmunoassay kit (Amersham). PKA activity assays were performed on total cell lysates using the SignaTECT cAMP-dependent protein kinase assay system according to the manufacturer's instructions (Promega). Total protein was determined as described before (26).

RESULTS AND DISCUSSION
C/EBP␤ is a member of a diverse group of nuclear transcription factors that contain a leucine zipper motif and a basic DNA-binding domain (5). We previously have shown that p21 expression is induced by antioxidants through a mechanism involving activation of C/EBP␤ (4). To determine the mechanism by which this occurs, we performed EMSA with a 32 Plabeled oligonucleotide containing the p21-NF-IL6 cis element and nuclear extracts from the colorectal cell line DKO-1, treated over a 24-h period with PDTC. As shown in Fig. 1A  (lanes 1-9), DNA binding activity was increased 5-fold in cells treated with PDTC for 30 min. Shifted complexes were competed by a 50-fold molar excess of an unlabeled oligonucleotide containing a consensus NF-IL6 sequence (lane 11) but not by an oligonucleotide containing a mutated NF-IL6 consensus sequence (lane 12), indicating that the induced complex was specific for the NF-IL6 cis element. Supershift experiments confirmed that this was predominately due to increased C/ We next sought to determine if the observed PDTC-induced increase in C/EBP␤ DNA binding activities was due to increased C/EBP␤ expression. As shown in Fig. 1B, both C/EBP␤ mRNA and whole cell protein concentrations remained relatively constant over the 24-h treatment with PDTC. These results suggest that the increased DNA binding activities are unlikely to be mediated by increased synthesis of C/EBP␤. However, in vivo [ 32 P]orthophosphate-labeling of C/EBP␤ revealed the rapid appearance and sustained elevation of a phosphorylated form of C/EBP␤ in nuclear extracts following PDTC treatment (Fig. 1C). Immunocytochemical analysis confirmed that under these conditions, PDTC induces a rapid redistribution of C/EBP␤ to the nucleus (Fig. 1D), suggesting a potential mechanism for the observed increase in DNA binding activity. These findings implicate protein phosphorylation of C/EBP␤ as a possible mechanism for its nuclear translocation. Previous studies have shown that C/EBP␤ translocates to the nucleus following elevation of cAMP levels in PC-12 cells (9) or activation of tumor necrosis factor receptors in hepatocytes (18). PDTC does not, however, induce nuclear translocation or increase DNA binding activity of the transcription factor NF-B in DKO-1 cells (data not shown).
A reduction in H 2 O 2 levels can activate adenylyl cyclase and hence elevate cAMP levels (27). Because PDTC has been shown to alter the cellular oxidative status, in part by decreasing endogenous H 2 O 2 levels (4), increased cAMP levels may initiate the observed post-translational effects of this antioxidant. Therefore, we measured the effect of forskolin (a known elevator of intracellular cAMP) or PDTC on cAMP levels and PKA activity in DKO-1 cells over a 24-h period. As shown in Fig. 2 (A  and B), both PDTC and forskolin were able to rapidly increase intracellular cAMP levels and PKA activity within 5 min. Interestingly, cAMP levels and PKA activity returned to base line following 24 h of treatment with forskolin, whereas PDTC induced a sustained elevation in these parameters. Cholera toxin was able to inhibit PDTC-induced PKA activity (data not shown), suggesting that PKA is activated via a cAMP-dependent pathway through G s stimulation of adenylyl cyclase. The sustained elevation in cAMP and PKA activity in response to PDTC and not in response to forskolin may be a result of modulation of phosphodiesterase activity by the antioxidant. In addition, a PDTC-induced alteration in phosphatase activity also may contribute to the observed persistence in PKA activity. We are presently exploring these potential mechanisms.
We utilized DKO-1 cells constitutively expressing epitope- FIG. 4. PKA phosphorylation of C/EBP␤ is required for nuclear translocation. (A) DKO-1 cells stably transfected with epitope-tagged WT C/EBP␤ were grown in the presence or absence of PDTC or PDTC and mPKI for 3 h. Cells were fixed with paraformaldehyde and FLAGtagged C/EBP␤ visualized by immunofluorescence. Parental DKO-1 cells were treated as above and immunostained for endogenous C/ EBP␤. Treatment of parental DKO-1 cells and/or stable transfectants with mPKI alone failed to induce nuclear translocation of C/EBP␤ (data not shown). B, parental DKO-1 cells or cells stably transfected with FLAG-tagged WT C/EBP␤ or the Ala 299 mutant were treated with PDTC (0 or 70 M) in the presence or the absence of mPKI (1 M) for 3 h. Treatment-related variations in p21 mRNA were evaluated by Northern blot analysis. IB15 is shown as a control for equivalent loading and transfer. C, DKO-1 cells were transfected with either a p21 reporter plasmid and a cytomegalovirus expression plasmid containing WT C/ EBP␤ (0.5 g each) alone or with the indicated amounts of pCMV-Ala 299 C/EBP␤. Luciferase activity was measured in relative light units after 24 h and normalized to chloramphenicol acetyltransferase activity, and results were reported as fold activation above basal levels. Values represent mean Ϯ S.E. of three transfections carried out in triplicate. tagged C/EBP␤ protein to further investigate whether posttranslational modification (phosphorylation) of C/EBP␤ is responsible for the observed increase in C/EBP␤ activity. In vivo labeling with [ 32 P]orthophosphate followed by immunoprecipitation revealed a 6-fold increase in epitope-tagged C/EBP␤ phosphorylation in response to PDTC or forskolin (3 h) with no change in protein levels (data not shown). Truncated versions of C/EBP␤ that contained only the 160 or 200 C-terminal amino acids were poor substrates for PDTC-induced phosphorylation, whereas mutant C/EBP␤ that contained the 305 C-terminal amino acids was phosphorylated by PDTC as efficiently as the full-length C/EBP␤ (data not shown). Closer inspection of the primary amino acid sequence between 236 and 305 revealed that this region contained a consensus PKA phosphorylation site (Arg-Xaa-Ser 299 -Xaa).
Comparison of the tryptic phosphopeptide maps of epitopetagged C/EBP␤ from PDTC-or forskolin-treated and untreated stable transfectants showed most of the inducible phosphorylation to occur on one distinct phosphopeptide (X 3 ; Fig. 3A). To confirm that both endogenous and FLAG-tagged C/EBP␤ are subject to similar changes in their phosphorylation state in a different cell type, we repeated these experiments in differentiated 3T3 cells stably expressing PKA that additionally express high levels of C/EBP␤. As observed in DKO-1 cells, PDTC stimulated site-specific phosphorylation of both endogenous and expressed C/EBP␤ (data not shown). Although the level of several phosphopeptides was increased after PDTC treatment, the only change common to both cell types was a higher level of the phosphopeptide, X 3 ( Fig. 3A and data not shown).
The migration position of phosphopeptide X 3 appeared identical to that of a phosphopeptide that contains Ser 277 and Ser 299 as its phosphoacceptors (data not shown), sites previously reported to be weakly phosphorylated by PKA in vitro (15). To determine whether Ser 277 and/or Ser 299 were indeed PDTC-responsive phosphorylation site(s), we substituted either amino acid with alanine. Vectors expressing FLAG-tagged wild type (WT) C/EBP␤ or C/EBP␤ (Ala 277 or Ala 299 ) were stably transfected into DKO-1 cells, and the resultant proteins were isolated after in vivo labeling with [ 32 P]orthophosphate. As shown in Fig. 3 (B and C), substitution of Ser 299 but not Ser 277 led to selective loss of PDTC-inducible phosphorylation and nuclear translocation of C/EBP␤. Interestingly, the low levels of Ala 299 C/EBP␤ detected by [ 32 P]orthophosphate labeling appeared as a doublet similar to the phosphorylated forms observed with endogenous protein (Fig. 1C and 3B). Furthermore, substitution of Ser 299 with alanine failed to alter the rate of C/EBP␤ migration during SDS-PAGE (Fig. 3B). These data suggest that these different C/EBP␤ species do not represent phosphorylation at Ser 299 . The appearance of this doublet form in the nucleus of DKO-1 but not HCT 116 or HCT 15 colon cancer cells (4) may represent additional post-translational modifications, independent of the PDTC-translocational signal.
Previous studies by our group have shown that overexpression of C/EBP␤ leads to increased p21 expression (4). Induction of p21 by DNA damage frequently relies on the tumor suppressor protein p53 (28), presumably through interaction with p53binding sites present in the p21 promoter region. However, many inducers of p21, including cytokines, prostaglandins, and genotoxic agents (29), operate to a large degree through poorly characterized p53-independent mechanisms. Therefore, we investigated the effect of PDTC-mediated phosphorylation on C/EBP␤ activity by comparing the ability of wild type C/EBP␤ and the Ala 299 mutant in stably transfected DKO-1 cells to activate p21 gene expression following nuclear translocation. As shown in Fig. 4 (A and B), wild type C/EBP␤ (endogenous and epitope-tagged WT) was able to translocate to the nucleus and induce p21 expression following PDTC treatment (Fig. 4, A  and B). PDTC treatment of these cells in the presence of a highly specific PKA inhibitor, mPKI, completely abolished these cellular changes (Fig. 4B). In contrast, co-administration of PDTC with the selective kinase inhibitors, tamoxifen citrate (PKC) or KN-93 (calcium-calmodulin-dependent), failed to inhibit nuclear translocation of C/EBP␤ or its ability to induce p21 expression (data not shown). As predicted, Ala 299 C/EBP␤ failed to translocate to the nucleus following PDTC treatment (Fig. 4A). Surprisingly, expression of this mutant acted as a dominant-negative in these cells, preventing the ability of endogenous protein to induce p21 expression (Fig. 4B). The dominant-negative effect of this mutation was confirmed by transient transfection analysis (Fig. 4C). These results demonstrate that translocation of C/EBP␤ to the nucleus is directly enhanced by the PKA-induced phosphorylation of Ser 299 , an effect inhibited by the presence of mPKI.
Previous reports have demonstrated that C/EBP␤ is phosphorylated by PKC (Ser 105 in vivo or Ser 248 , Ser 277 , and Ser 299 in vitro) (14,15), PKA (Ser 277 and Ser 299 in vitro) (15), a Ras-dependent MAP kinase (Thr 235 in vitro) (16), and a calcium-calmodulindependent kinase (Ser 276 in vivo and in vitro) (17), leading to alterations in DNA binding activities. Although these reports provide insights into the mechanisms by which the activity of C/EBP␤ could be potentiated by selected agents to affect activation of genes, specific biological end points were not explored. For example, transactivation of C/EBP␤ by PKC in hepatocytes may affect activation of acute phase protein synthesis, whereas phosphorylation of Thr 235 by MAP kinase may lead to cell growth or inhibition. We envision that differential phosphorylation of C/EBP␤ may contribute, at least in part, to the diverse signals mediated by this transcription factor. Collectively, our results demonstrate that in DKO-1 cells phosphorylation of C/EBP␤ at Ser 299 following activation of PKA is critical for its nuclear translocation and subsequent transactivation of genes in response to altered oxidative states.