Differential Regulation of Cyclin D1 Expression by Protein Kinase C α and ϵ Signaling in Intestinal Epithelial Cells*

Background: Tight control of cyclin D1 expression is critical for intestinal homeostasis. Results: Whereas PKCα suppresses cyclin D1 expression, PKCϵ up-regulates cyclin D1 via an ERK and NF-κB/CRE-mediated transcriptional mechanism. Conclusion: Cyclin D1 levels in intestinal cells reflect a balance between PKCα and PKCϵ signaling. Significance: The opposing effects of PKCα and PKCϵ on cyclin D1 accumulation reflect their contrasting contributions to intestinal tumorigenesis. Cellular accumulation of cyclin D1, a key regulator of cell proliferation and tumorigenesis, is subject to tight control. Our previous studies have identified PKCα as a negative regulator of cyclin D1 in the intestinal epithelium. However, treatment of non-transformed IEC-18 ileal crypt cells with PKC agonists has a biphasic effect on cyclin D1 expression. Initial PKCα-mediated down-regulation is followed by recovery and subsequent accumulation of the cyclin to levels markedly higher than those seen in untreated cells. Using protein overexpression strategies, siRNA, and pharmacological inhibitors, we now demonstrate that the recovery and hyperinduction of cyclin D1 reflect the combined effects of (a) loss of negative signals from PKCα due to agonist-induced PKCα down-regulation and (b) positive effects of PKCϵ. PKCϵ-mediated up-regulation of cyclin D1 requires sustained ERK stimulation and transcriptional activation of the proximal cyclin D1 (CCDN1) promoter, without apparent involvement of changes in protein stability or translation. PKCϵ also up-regulates cyclin D1 expression in colon cancer cells, through mechanisms that parallel those in IEC-18 cells. Although induction of cyclin D1 by PKCϵ is dependent on non-canonical NF-κB activation, the NF-κB site in the proximal promoter is not required. Instead, cyclin D1 promoter activity is regulated by a novel interaction between NF-κB and factors that associate with the cyclic AMP-response element adjacent to the NF-κB site. The differential effects of PKCα and PKCϵ on cyclin D1 accumulation are likely to contribute to the opposing tumor-suppressive and tumor-promoting activities of these PKC family members in the intestinal epithelium.


Cellular accumulation of cyclin D1, a key regulator of cell proliferation and tumorigenesis, is subject to tight control. Our previous studies have identified PKC␣ as a negative regulator of cyclin D1 in the intestinal epithelium. However, treatment of non-transformed IEC-18 ileal crypt cells with PKC agonists has a biphasic effect on cyclin D1 expression. Initial PKC␣-mediated down-regulation is followed by recovery and subsequent
accumulation of the cyclin to levels markedly higher than those seen in untreated cells. Using protein overexpression strategies, siRNA, and pharmacological inhibitors, we now demonstrate that the recovery and hyperinduction of cyclin D1 reflect the combined effects of (a) loss of negative signals from PKC␣ due to agonist-induced PKC␣ down-regulation and (b) positive effects of PKC⑀. PKC⑀-mediated up-regulation of cyclin D1 requires sustained ERK stimulation and transcriptional activation of the proximal cyclin D1 (CCDN1) promoter, without apparent involvement of changes in protein stability or translation. PKC⑀ also up-regulates cyclin D1 expression in colon cancer cells, through mechanisms that parallel those in IEC-18 cells. Although induction of cyclin D1 by PKC⑀ is dependent on non-canonical NF-B activation, the NF-B site in the proximal promoter is not required. Instead, cyclin D1 promoter activity is regulated by a novel interaction between NF-B and factors that associate with the cyclic AMP-response element adjacent to the NF-B site. The differential effects of PKC␣ and PKC⑀ on cyclin D1 accumulation are likely to contribute to the opposing tumorsuppressive and tumor-promoting activities of these PKC family members in the intestinal epithelium.
Cyclin D1 (CCND1) is a proto-oncogene and critical regulator of cell proliferation, acting as a major mitogen sensor that links cellular signaling networks to the cell cycle machinery (1,2). The best characterized role of cyclin D1 is as the regulatory subunit of cyclin-dependent kinases 4 and 6. Cyclin D1 both facilitates cyclin-dependent kinase-mediated phosphorylation and inactivation of the tumor-suppressive pocket proteins (retinoblastoma protein pRb, p107, and p130) and sequesters CIP/KIP cyclin-dependent kinase-inhibitory proteins (p21 Cip1 , p27 Kip1 , and p57 Kip2 ) away from cyclin⅐cyclin-dependent kinase 2 complexes (2). In addition, cyclin D1 promotes cell proliferation through interaction with transcription factors such as the estrogen receptor and Sp1 (1,3). Because of these effects, cell proliferation is extremely sensitive to alterations in the levels of cyclin D1, and even modest changes in its expression can have appreciable effects on cell cycle progression (4 -7). Thus, cellular accumulation of cyclin D1 is under tight control, and its expression is regulated at multiple levels, including gene transcription, mRNA transport and stability (8 -10), translation (11), and protein degradation (9). Disruption of these regulatory mechanisms is a common feature of multiple cancer types, with overexpression of cyclin D1 being among the most frequent alterations observed in tumors (1). Many oncogenic and tumor-suppressive signals converge on cyclin D1, altering its expression by enhancing transcription through distinct DNA sequences in its promoter as well as through changes in translation and protein degradation (1,2,8).
The intestinal epithelium is one of the most rapidly proliferating tissues in the body, and maintenance of mucosal homeostasis is dependent on precise control of cellular proliferation within intestinal crypts. Increased proliferation leads to crypt elongation and, eventually, adenoma formation and tumor progression. Overexpression of cyclin D1 is an early event in intestinal tumorigenesis (12), and robust expression of this cyclin is important for intestinal cell proliferation and maintenance of the transformed phenotype in mouse models and human colon cancer cells (13)(14)(15). Although gene amplification drives cyclin D1 overexpression in many cancer types (2), gains and amplifi-cations of CCND1 are rare in colon cancer (16), pointing to disruption of signal transduction pathways upstream of cyclin D1. Thus, the identification of signaling pathways that impact cyclin D1 expression in the normal intestine and in colon cancer cells is critical for understanding of mechanisms underlying intestinal homeostasis and colon cancer development.
In previous studies, we have determined that protein kinase C (PKC) is an important regulator of cyclin D1 expression in the normal intestine and during intestinal tumorigenesis (11,14,(17)(18)(19). PKC comprises a family of at least 10 isozymes that have emerged as key regulators of cell proliferation and tumorigenesis in multiple tissues (20). PKC isozymes have been grouped into subfamilies based on differences in structure and cofactor requirements. Classical PKCs (PKC␣, PKC␤I, PKC␤II, and PKC␥) require diacylglycerol and Ca 2ϩ for activity; novel PKCs (PKC␦, PKC⑀, PKC, and PKC) are activated by diacylglycerol but do not require Ca 2ϩ ; and atypical PKCs (PKC/ and PKC) are activated by protein-protein interactions rather than by diacylglycerol (cf. Ref. 21). Physiological activation of classical PKCs and novel PKCs occurs following receptor-mediated activation of phospholipase C, which cleaves phosphatidylinositol 4,5-bisphosphate to generate diacylglycerol and inositol 3-phosphate. This leads to recruitment of these enzymes to the plasma membrane, where they undergo conformational changes that result in activation. A number of pharmacological activators, such as phorbol esters and bryostatins, can also recruit classical PKCs and novel PKCs to the membrane through interaction with the diacylglycerol binding site in their C1 regulatory domain (21). Our previous studies have determined that PKC␣ signaling rapidly inhibits cyclin D1 expression in intestinal epithelial cells through both translational and transcriptional mechanisms. Translational suppression involves PP2A-dependent activation of the translational repressor 4E-BP1 4 (11,18), whereas blockade of transcription involves down-regulation of the transcriptional regulator inhibitor of DNA binding 1 (Id1) and is mediated by promoter elements between Ϫ163 and Ϫ1745 from the transcriptional start site (14,19). The physiological relevance of these effects is underscored by the fact that genetic knockout of PKC␣ in mice leads to increased cyclin D1 expression in intestinal crypts (19). Consistent with this effect, PKC␣ acts as a tumor suppressor in the intestine (14,22), and its ability to down-regulate cyclin D1 is an important aspect of its tumor-suppressive activity (14). Other members of the PKC family (e.g. PKC⑀ and PKC) appear to function as oncogenes in the intestine (23,24), pointing to potential positive regulation of cyclin D1 by PKCs in this tissue.
In the current study, we further analyze the regulation of cyclin D1 in non-transformed intestinal epithelial cells and colon cancer cells and identify PKC⑀ as a positive regulator of cyclin D1 accumulation in this system. Our findings demonstrate that the opposing effects of PKC⑀ and PKC␣ on cyclin D1 levels involve distinct mechanisms, with PKC⑀ promoting transcriptional up-regulation of the cyclin mediated by an interac-tion between NF-B and factors that bind to the cyclic AMPresponse element (CRE) in the cyclin D1 gene promoter.
Western Blot Analysis-Cells were lysed in 1% SDS, 10 mM Tris-HCl, pH 7.4, and equal amounts of cellular protein were separated by SDS-PAGE, transferred to nitrocellulose membranes, and subjected to immunoblot analysis as we have described (17,25). Evenness of loading and transfer was routinely confirmed by staining membranes with 0.1% Fast Green (Sigma). Band intensities were quantified by analysis of scanned blots using ImageJ software (National Institutes of Health), and data are presented relative to control (mean Ϯ S.E.).
Metabolic Labeling-Labeling of IEC-18 cells with [ 35 S]methionine/cysteine and analysis of cyclin D1 synthesis by immunoprecipitation, SDS-PAGE, and phosphorimaging were as we have described (11).
Plasmids and Promoter Mutation-The human cyclin D1 promoter-luciferase reporter constructs containing sequences corresponding to the proximal cyclin D1 promoter extending to 1745 bp (Ϫ1745CD1Luc), 963 bp (Ϫ963CD1Luc), or 163 bp (Ϫ163CD1Luc) upstream of the transcription initiation site were from Dr. Richard Pestell (Thomas Jefferson University). The NF-B activity reporter construct (NF-B-luciferase) containing three NF-B binding sites upstream of the minimal CMV promoter and the dominant active IB␣ expression and control constructs were obtained from Drs. Andrei Bakin and Andrei Gudkov (Roswell Park Cancer Institute).
Promoter Analysis-Transfection of IEC-18 and DLD1 cells using Fugene 6 and measurement of promoter activity using the Dual-Luciferase assay kit (Promega) were as we have described (14,29). pRL-TK and pRL-CMV plasmids expressing Renilla luciferase (Promega) were routinely included in the transfections to monitor transfection efficiency; however, the thymidine kinase and CMV promoters in these reporters are responsive to PKC agonists. Therefore, the effects of drug treatments were determined from the relative firefly luciferase activity in control and treated cells transfected with the same transfection mixture. When different transfection mixes were used in a single experiment (i.e. those involving promoter mutants or dominant active IB␣), promoter activity was normalized for transfection efficiencies using the respective Renilla luciferase readings for each transcription mixture measured in vehicletreated cells.
Statistical Analysis-Student's t tests and regression analysis were performed using Microsoft Excel software. Differences with p values of Ͻ0.05 were considered statistically significant.

Cyclin D1 Expression Is Subject to Both Negative and Positive Regulation by PKC Isozyme Signaling in Intestinal Epithelial
Cells-We have previously demonstrated that treatment of non-transformed IEC-18 rat ileal crypt cells with the PKC agonist PMA has biphasic effects on cyclin D1 expression (17). Fig.  1A further demonstrates that (a) the initial down-regulation of cyclin D1 is maximal at ϳ2 h (lane 3), (b) expression of the cyclin recovers to control levels by ϳ4 h (Fig. 1A, lane 4), and (c) subsequent accumulation of cyclin D1 to levels markedly higher than those seen in unstimulated cells is evident by 6 -8 h (lanes 5 and 6). Up-regulation of cyclin D1 is maintained for at least 24 h in PMA-treated IEC-18 cells (Fig. 1A, lanes 8 and 9). Hyperinduction of cyclin D1 expression is also seen following prolonged treatment with other PKC agonists, including the phorbol ester phorbol 12,13-dibutyrate (data not shown), the macrolide lactone bryostatin 1 (Fig. 1B, lanes 1 and 3), and the synthetic diacylglycerol DiC 8 (Fig. 1B, lanes 4 and 5), indicating that the effect is a general response to long term PKC activation and does not reflect a specific response to PMA. In previous studies, we determined that the initial down-regulation of cyclin D1 induced by PKC agonists is mediated by PKC␣ (11,14,(17)(18)(19); however, the mechanisms underlying the later hyperinduction of the protein remained to be characterized.
Prolonged PKC agonist treatment has long been recognized to down-regulate PKC isozymes, and reversal of the growth inhibitory effects of these agents in IEC-18 cells correlates with loss of PKC␣ (e.g. see Refs. 11,17,and 18). Therefore, the contribution of loss of PKC␣ activity to PKC agonist-induced upregulation of cyclin D1 was tested using the classical PKC inhibitor, Gö6976, which is selective for PKC␣ in IEC-18 cells (17), as well as the general PKC inhibitors, BIM and Gö6983. In keeping with a restraining effect of PKC␣ activity on cyclin D1 accumulation (14), all three inhibitors led to increased steady-state levels of cyclin D1 expression in the absence of PKC agonist treatment (Fig. 1C, compare lane 1 with lanes 3, 5, and 7). However, this increase was not as robust as the hyperinduction seen following prolonged (8-h) exposure to PMA (Fig. 1C, compare lanes 3, 5, and 7 with lane 2). General inhibition of PKC activity with BIM or Gö6983 blocked the ability of prolonged PMA treatment to further up-regulate cyclin D1 (Fig. 1C, compare lane 5 with lane 6 and lane 7 with lane 8), confirming that the hyperinduction is PKC-dependent. In contrast, selective blockade of PKC␣ with Gö6976 failed to prevent PMA-induced upregulation of cyclin D1 at 8 h (Fig. 1C, compare lanes 3 and 4), indicating that a PKC isozyme other than PKC␣ must contribute positively to PKC agonist-induced hyperinduction of the molecule in intestinal epithelial cells.
PKC⑀ Mediates Positive Effects of PKC Agonists on Cyclin D1 Expression-IEC-18 cells express three PKC isozymes that are responsive to the pharmacological agonists PMA, bryostatin 1, and DiC 8 : PKC␣, PKC␦, and PKC⑀ (30). As shown in Fig. 2A, i, PMA down-regulates these isozymes with different kinetics (also see Ref. 31). PKC␣ is depleted between 3 and 6 h of PMA treatment (Fig 2A, i), correlating with reversal of cyclin D1 loss (see Figs. 1A and 2A (i and ii), and remains barely detectable at later times of treatment (12-24 h), when cyclin D1 hyperinduction is evident ( Fig. 2A, ii). PKC␦ is down-regulated more rapidly than PKC␣ ( Fig. 2A, i), indicating that it is unlikely to contribute positively to the hyperinduction of cyclin D1 seen at later time points. In contrast, down-regulation of PKC⑀ is relatively slow and incomplete ( Fig. 2A, i, compare lane 2 with lanes 10 and 12), with the protein remaining readily detectable, albeit at lower levels, even after 24 h of PMA treatment (31). Thus, comparison of the kinetics of PKC isozyme down-regulation and cyclin D1 hyperinduction points to PKC⑀ as the positive factor mediating the effect.
The requirement for PKC⑀ in cyclin D1 hyperinduction was tested directly by manipulating PKC isozyme expression levels in IEC-18 cells. Transduction of cells with adenovirus expressing PKC␣ or PKC␦ had no effect on cyclin D1 steady-state levels in untreated cells (Fig. 2B, left). In contrast, PKC⑀ overexpression led to up-regulation of the cyclin (Fig. 2B, right), indicating that this isozyme is limiting for cyclin D1 expression in IEC-18 cells. The role of PKC⑀ was further tested by siRNA-mediated knockdown. Transfection of cells with PKC⑀-targeted siRNA markedly reduced PKC⑀ steady-state levels compared with non-targeting siRNA transfection (Fig. 2C, lanes 1 and 3) while having no effect on PKC␣ or PKC␦ (27). In addition, whereas PKC⑀ remained readily detectable at 6 h of PMA treatment in non-targeting siRNA-transfected cells (Fig. 2C, lane 2), the enzyme was depleted from PKC⑀ siRNA-transfected cells by this time (Fig. 2C, lane 4). Notably, knockdown of PKC⑀ completely abrogated the hyperinduction of cyclin D1 promoted by PMA. Indeed, whereas hyperinduced cyclin D1 expression was evident following a 6-h PMA treatment in control cells (Fig. 2C, lanes 1 and 2), levels of the cyclin remained down-regulated at this time point in PKC⑀-silenced cells. Thus, PKC⑀ activity is required for both hyperinduction and efficient restoration of cyclin D1 expression following PKC␣-mediated down-regulation of the cyclin in PKC agonist-treated cells.
To further examine the contributions of PKC␣ and PKC⑀ to the biphasic effects of PKC agonists on cyclin D1 expression, the consequences of overexpressing these proteins on cyclin D1 down-regulation, recovery, and hyperinduction were tested in IEC-18 cells. Consistent with the inhibitory effects of PKC␣, overexpression of this isozyme enhanced PMA-induced downregulation of cyclin D1 and delayed its initial recovery (Fig. 2D, i, compare lanes 1-4 with lanes 8-11); however, it did not affect the eventual hyperinduction of the cyclin (compare lanes 6 and 7 with lanes 13 and 14). In contrast to PKC␣, PKC⑀ overexpression reduced the extent and duration of PMA-induced downregulation of cyclin D1 (Fig. 2D, ii, compare lanes 1-3 with lanes 8 -10) while markedly potentiating cyclin D1 hyperinduction by prolonged PKC agonist treatment (Fig. 2D, ii, compare lanes 5-7 with lanes [12][13][14].
Collectively, these data indicate that regulation of cyclin D1 expression by PKC agonists is dictated by opposing activities of two members of the PKC family, PKC␣ and PKC⑀. The initial down-regulation results from suppressive effects of PKC␣; as these negative effects wane due to agonist-induced down-regulation of PKC␣ signaling, positive effects of PKC⑀ can pre- dominate, leading to recovery and eventual hyperinduction of cyclin D1 expression.
Sustained ERK Activity Is Required for PKC⑀-mediated Upregulation of Cyclin D1-The MEK-ERK signal transduction pathway is an important downstream mediator of PKC func-tion and can act as a positive or negative regulator of cyclin D1 expression (27,32,33). We have determined that sustained ERK activation during the first hour of PKC agonist treatment is required and sufficient for the growth-inhibitory functions of PKC␣ in IEC-18 cells (27). However, the activation of ERK induced by PKC agonists extends well beyond 1 h (27), pointing to the possibility that later ERK activation contributes to the PKC⑀-dependent up-regulation of cyclin D1 expression. To test this possibility, the duration of PKC agonist-induced ERK activity was manipulated by the addition of the MEK inhibitor, U0126, at various times after the initiation of PMA or vehicle treatment (Fig. 3A). In control cells, treatment with U0126 for 2-3 h had minimal effects on basal expression of cyclin D1 (Fig.  3B, lanes 1, 7, and 9); however, with longer ERK inhibition, expression of the cyclin dropped off and became barely detectable following 6 -8 h of U0126 treatment (lanes 3 and 5) (data not shown). Based on these findings, a 6-h PMA treatment was selected for these experiments; 6 h was sufficiently long to observe hyperinduction of cyclin D1 (Figs. 1, 2A, and 3B, lanes 1 and 2) while allowing for its detection at the longest times of MEK inhibitor treatment (5 h, lane 3). The addition of U0126 after 4 h of PMA treatment, a time when levels of cyclin D1 have recovered to control levels (Fig. 1A), blocked the hyperinduction of cyclin D1 seen at 6 h of PMA treatment (Fig. 3B, compare lane 1 with lane 2 and lane 9 with lane 10). Interestingly, when U0126 was added at times when cyclin D1 was downregulated (1-3 h), cyclin D1 failed to recover by 6 h, with levels remaining lower in cells treated with PMA and U0126 than in cells treated with U0126 alone (Fig. 3B Similar results were observed with the structurally unrelated MEK inhibitor PD98059 (data not shown), indicating that the effects were specific for inhibition of ERK signaling. Thus, ongoing ERK activity is required both for efficient recovery of cyclin D1 expression and for its eventual hyperinduction in PKC agonist-treated IEC-18 cells. Notably, these findings mirror the effects of PKC⑀ knockdown, which also blocked both recovery and hyperinduction of cyclin D1 expression following a 6-h PMA treatment (Fig. 2C). Taken together, these data indicate that ERK signaling mediates opposing effects of PKC isozymes on cyclin D1 expression in PKC agonist-treated cells. Early (Յ1 h) PKC␣mediated ERK activation negatively regulates cyclin D1 expression, whereas later ERK activity is required for the positive effects of PKC⑀ on expression of this cyclin.
PKC⑀ Up-regulates Cyclin D1 mRNA-Cyclin D1 expression is controlled at transcriptional, translational, and post-translational levels. Analysis of the rate of disappearance of the protein following cycloheximide treatment excluded a role for alterations in protein stability in the hyperinduction of cyclin D1 observed in long-term PMA-treated cells (Fig. 4A). These studies confirmed that cyclin D1 is a relatively unstable protein in IEC-18 cells (11), with a half-life of about 15 min (Fig. 4A). However, although levels of cyclin D1 are elevated in cells exposed to PMA for 6 h (compare lanes 1 and 5), long term PMA treatment did not significantly change the rate of cyclin D1 protein disappearance following the addition of cycloheximide (p ϭ 0.41).
Having excluded protein stabilization as the mechanism underlying PMA-induced hyperinduction of cyclin D1, the role of alterations in protein synthesis was explored. Metabolic labeling experiments demonstrated increased incorporation of [ 35 S]methionine into cyclin D1 in IEC-18 cells treated with PMA for Ͼ6 h (Fig. 4B). Our previous studies identified inhibition of cap-dependent translation as a major mechanism underlying PKC␣-mediated down-regulation of cyclin D1 in intestinal epithelial cells (11,14). This effect is the result of PKC␣-induced accumulation of the active, translationally repressive, dephosphorylated ␣ and ␤ forms of 4E-BP1, as can be clearly seen in IEC-18 cells at 1-4 h of PMA treatment (Fig.  4C, arrows). However, although levels of these faster migrating repressive forms of 4E-BP1 begin to fall by 6 h, they remain above control levels up to 10 h of PMA treatment (Fig. 4C,  compare lane 1 with lanes 5 and 6). Therefore, changes in the efficiency of cap-dependent translation are unlikely to account for the hyperinduction of cyclin D1 expression seen at these times.
To determine whether enhanced translation of cyclin D1 reflects alterations in levels of its mRNA, quantitative RT-PCR analysis was performed on cells treated with PMA for various times (Fig. 5A, i). As noted in our previous studies (11), PMA treatment led to an initial down-regulation of cyclin D1 mRNA (Fig. 5A, i, black bars, 2 h). However, as with the protein, expression of cyclin D1 mRNA subsequently recovered and rose to levels higher than those seen in control cells by 6 h of PMA treatment (Fig. 5A, i, black bars, 6 and 8 h). This effect was maintained for at least 24 h, as confirmed by Northern blot analysis (Fig. 5A, ii). The PKC dependence of these effects was confirmed by the ability of Rö32-0432 to block both the down-regulation and the subsequent hyperinduction of cyclin D1 mRNA in PMA-treated cells (Fig. 5A, i, gray bars). Notably, as seen at the protein level, overexpression of PKC⑀ enhanced the accumulation of cyclin D1 mRNA, whereas overexpression of PKC␣ or PKC␦ had no effect (Fig. 5B). Thus, PKC⑀ positively regulates cyclin D1 expression in IEC-18 cells through up-regulation of its mRNA.

PKC⑀-mediated Up-regulation of Cyclin D1 Involves Transcriptional Activation through Proximal Promoter Sequences-
To determine the mechanism(s) underlying PKC agonist-induced up-regulation of cyclin D1 mRNA, the involvement of transcriptional alterations was tested using the RNA polymerase inhibitor, actinomycin D. As expected, actinomycin D led to a reduction in the overall levels of cyclin D1 expression in IEC-18 cells (Fig. 5C, compare lanes 1-3 with lanes 4 -6). Actinomycin D also abrogated PMA-induced up-regulation of cyclin D1 at 6 and 8 h (Fig. 5C, compare lanes 8 and 9 with lanes  11 and 12). Thus, the PKC⑀-dependent up-regulation of cyclin D1 expression seen following sustained PKC agonist treatment is dependent on new transcription. The main regulatory elements in the cyclin D1 promoter are located within a region 1745 bp upstream of the transcriptional start site (8). To test whether the transcriptional dependence reflected direct activation of the cyclin D1 promoter, cells were transfected with reporter constructs in which luciferase expression is driven by 1745 bp of the cyclin D1 promoter (Ϫ1745CD1Luc) or by a promoter truncated at Ϫ963 bp (Ϫ963CD1Luc) or Ϫ163 bp (Ϫ163CD1Luc). Treatment of transfected cells with PMA for 22 h led to a ϳ2-fold increase in the activity of all three of these promoter constructs (Fig. 5D). These data show that long term PMA treatment induces cyclin D1 transcription through promoter sequences within 163 base pairs upstream of the transcriptional start site.
PKC⑀ Up-regulates Cyclin D1 in DLD1 Colon Cancer Cells-To determine whether PKC⑀ signaling also up-regulates cyclin D1 in colon cancer cells, the effects of long term PMA treatment were tested in FET, GEO, and DLD1 cells. In contrast to IEC-18 cells, colon cancer cells lack functional PKC␣ signaling (14) and do not repress cyclin D1 levels in response to PKC agonists (Fig. 6A, compare lanes 1 and 2 with lanes 3 and 4; see also Ref. 14). However, longer PMA treatments led to a sustained increase in cyclin D1 levels in these cells (Fig. 6A, compare lanes 1 and 2 with lanes 4 -7). Notably, consistent with the absence of repressive PKC␣ signaling in colon cancer cells, this induction occurred more rapidly than in IEC-18 cells, with increased accumulation observed as early as 4 h after the PMA addition (compare lanes 1, 2, and 4 in Figs. 1A and 6A).
Having established that PKC agonists up-regulate cyclin D1 in colon cancer cells, the molecular mechanisms underlying this effect were further characterized using DLD1 cells. Pro- longed DiC 8 treatment also led to hyperinduction of cyclin D1 in these cells (Fig. 6B), and the PKC dependence of the effect was further confirmed by the ability of the general PKC inhibitor, BIM, to block the PMA-induced increase in cyclin D1 expression (Fig. 6C). Thus, as seen in IEC-18 cells, cyclin D1 hyperinduction is a result of PKC activation and not a specific response to phorbol esters. Also paralleling the induction in IEC-18 cells, the ability of PKC agonists to hyperinduce cyclin D1 in DLD1 cells was dependent on ERK activity because it could be blocked with U0126 or PD98059 (Fig. 6D) (data not shown). DLD1 cells lack expression of PKC␣ and PKC␦ but express PKC␤II in addition to PKC⑀ as the only two PMAresponsive isozymes (Fig. 6E) (14). Therefore, the contribution of PKC␤II to the effects of PKC agonists was tested using the PKC␤-selective inhibitor 3-(1-(3-imidazol-1-ylpropyl)-1H-indol-3-yl)-4-anilino-1H-pyrrole-2,5-dione (34). Even at concentrations as high as 1 M, this inhibitor failed to affect either the basal levels of cyclin D1 or the ability of PMA to induce its expression (Fig. 6F), arguing that PKC␤II activity is not required for PMA-induced up-regulation of cyclin D1 in these cells. Because PMA treatment failed to down-regulate PKC␤II in DLD1 cells, the efficacy of the inhibitor could not be tested by its ability to block activation-dependent down-regulation of the enzyme (cf. Ref. 25). However, even at concentrations as low as 100 nM, the PKC␤ inhibitor potently blocked down-regulation of PKC␤II in SK-UT-1B cells (data not shown), confirming the efficacy of this compound as an inhibitor of this PKC isozyme. Because PKC␤II and PKC⑀ are the only PMA-responsive isozymes in DLD1 cells, these findings indicate that PKC⑀ can positively regulate cyclin D1 expression in both colon cancer cells and non-transformed IEC-18 cells.
Consistent with findings in IEC-18 cells, PKC agonist-induced up-regulation of cyclin D1 in DLD1 cells is dependent on new transcription, as confirmed by the failure of PMA to increase cyclin D1 levels in the presence of actinomycin D (Fig.  7A). Transcriptional activation in DLD1 cells occurs through the same proximal promoter region responsible for PKC⑀-mediated up-regulation of cyclin D1 in IEC-18 cells, as demonstrated by the ability of PMA to enhance the activity of the Ϫ963CD1Luc and Ϫ163CD1Luc constructs in transient trans- fection assays (Fig. 7B). Collectively, these data indicate not only that PKC⑀ signaling is able to induce cyclin D1 expression in colon cancer cells but that it utilizes the same mechanisms for this up-regulation in both non-transformed and transformed intestinal epithelial cells.

PKC Agonists Activate NF-B in Intestinal Epithelial Cells-
The NF-B site at Ϫ38 bp in the PKC⑀-responsive cyclin D1 proximal promoter region (Fig. 8A) is a strong regulator of cyclin D1 transcription in multiple contexts (35). Because PKC signaling can elicit sustained NF-B activation in various systems, including colon cancer cells (e.g. see Refs. [35][36][37][38][39], the ability of PMA to stimulate NF-B in IEC-18 cells was tested using an NF-B reporter construct. PMA treatment increased NF-B activity about 2-2.5-fold in IEC-18 cells (Fig. 8, B and C (i)). Increased NF-B activity was first seen at 2-4 h, a time that slightly precedes the up-regulation of cyclin D1 mRNA levels, and the activity remained elevated for at least 24 h (Fig. 8B). This level of induction is close to that seen in DLD1 cells, which also occurred by 3 h of PMA treatment (Fig. 8C, ii) (data not shown).
PMA has been shown to promote NF-B activation in colon cancer cells through the canonical pathway, which requires the ubiquitination and degradation of IB, and through mechanisms that are independent of IB degradation (38 -40). Therefore, the effects of expressing a mutant IB␣ protein that is refractory to ubiquitination and degradation (DA IB) were tested. When DA IB was expressed in the NF-B activity assay, basal NF-B activity was markedly reduced in both IEC-18 and DLD1 cells, confirming the activity of DA IB under the experimental conditions used (Fig. 8C). However, although overall NF-B activity was decreased, PMA treatment was able to promote induction of NF-B Ͼ3-fold in the presence of DA IB, indicating that PKC agonists activate NF-B through pathways that are both sensitive and resistant to IB.
Inhibition of NF-B Activity Blocks PKC Agonist-induced Upregulation of Cyclin D1-Because PKC agonists increased NF-B activity in IEC-18 and DLD1 cells through mechanisms that are not sensitive to IB, the requirement for NF-B in PKC⑀-mediated induction of cyclin D1 was evaluated using  PDTC. This compound blocks NF-B activation by inhibiting the SCF/␤-TrCP ubiquitin ligase (41) and IKK activity (42) and can thus suppress both the canonical and non-canonical NF-B activation pathways (42,43). Based on analysis of the concentration of this compound required to inhibit PMA-induced NF-B activity (data not shown) and on published data (44 -46), 100 M PDTC was used in these studies. At this concentra-tion, PDTC blocked PMA-induced up-regulation of cyclin D1 protein in both IEC-18 and DLD1 cells (Fig. 9, A (i) and B (i)). In addition to preventing the hyperinduction of cyclin D1 in IEC-18 cells, treatment with PDTC unexpectedly led to an increase in steady-state levels of cyclin D1 (Fig. 9A, i). This effect, which may relate to the antioxidant properties of PDTC (47)(48)(49), is posttranscriptional because the inhibitor decreased cyclin D1 mRNA expression in IEC-18 (Fig. 9C, i) and is thus unrelated to the transcriptional effects of PKC⑀. The involvement of NF-B was also investigated using CAPE, an NF-B inhibitor that differs from PDTC in structure and mechanism of action (50) and that also inhibits both the canonical and non-canonical pathways (e.g. see Ref. 51). Although CAPE was mildly toxic to both IEC-18 and DLD1 cells under the conditions used (see the legend to Fig. 9) and resulted in down-regulation of basal cyclin D1 expression in both cell types (Fig. 9, A-D), this inhibitor (10 -100 M) also prevented cyclin D1 upregulation by prolonged PMA treatment (Fig. 9, A (ii) and B (ii)). PDTC and CAPE also blocked PMA-induced up-regulation of cyclin D1 mRNA (Fig. 9C) and activation of the Ϫ163CD1 promoter (Fig. 9D). In contrast to the effects of PDTC and CAPE, activation of cyclin D1 transcription by PMA in intestinal epithelial cells was not affected by expression of dominant active IB (Fig. 9E), indicating that the effect is not dependent on the canonical pathway of NF-B activation.

PKC⑀-induced Up-regulation of Cyclin D1 Transcription Involves Cooperation between NF-B-and CRE-interacting
Factors-To identify cyclin D1 promoter elements that mediate the transcriptional effects of PKC⑀-NF-B signaling, transcription factor binding sites in the proximal 163-bp promoter region (Fig. 8A) were individually mutated, and the resulting constructs were tested in DLD1 cells. Mutation of the majority of these sites had little effect on either the basal level of cyclin D1 promoter activity or the extent of its induction following PMA treatment (Fig. 10A). Interestingly, despite the evidence that NF-B activity is required for PKC⑀-mediated up-regulation of cyclin D1 (Fig. 9), this was true even for the NF-B site, indicating that an additional promoter element(s) is involved.
Mutation of three sites, the Egr1 site at Ϫ137 bp (Egr1b), the TCF/LEF site, and the CRE, did affect activity of the promoter. Mutation of the Egr1b site or the TCF/LEF site decreased the basal level of cyclin D1 promoter activity in DLD1 cells (Fig.  10A) but did not significantly affect the relative induction of promoter activity by PMA (wild-type, 1.9 Ϯ 0.1-fold; Egr1b mutant, 1.6 Ϯ 0.2-fold; TCF/LEF mutant, 2.0 Ϯ 0.1-fold). In contrast, mutation of the CRE not only led to a reduction in basal activity (Fig. 10A), but also significantly reduced the ability of PMA treatment (4 h) to enhance activity of the promoter (-fold induction ϭ 1.4 Ϯ 0.1 for the CRE mutant versus 1.9 Ϯ 0.1 for the wild type, p ϭ 0.004, n ϭ 9). Thus, although the TCF/ LEF, Egr1b, and CRE sites support basal cyclin D1 promoter activity in DLD1 cells, factors that interact with the CRE are also involved in mediating the effects of PMA.
The CRE in the cyclin D1 promoter is adjacent to the NF-B site (Fig. 8A); therefore, the effect of simultaneous mutation of these sites was investigated (Fig. 10A). Mutating both sites had only a slight effect on basal activity of the cyclin D1 promoter in DLD1 cells compared with mutation of the CRE alone; however, double mutation of these sites abrogated the increased activity of the cyclin D1 promoter seen with prolonged PMA treatment (-fold induction ϭ 1.1 Ϯ 0.2, n ϭ 6). The added effect of mutating both sites confirms a role for NF-B at the promoter and indicates that cooperation between the CRE and NF-B is integral to the ability of PKC⑀ to enhance cyclin D1 expression in DLD1 colon cancer cells.
Analysis of the CRE and NF-B mutant promoters in IEC-18 cells also revealed cooperation between these sites, although differences in the roles of each site were noted in these cell lines. Mutation of the NF-B site increased basal promoter activity in IEC-18 cells (Fig 10B), pointing to a repressive role of this site in regulating basal expression of cyclin D1 in non-transformed intestinal epithelial cells. However, as seen in DLD1 cells, the NF-B site was dispensable for cyclin D1 hyperinduction because mutation of this site did not affect PMA-induced upregulation of promoter activity. In contrast to the effects in DLD1 cells, mutation of the CRE alone did not affect basal levels of cyclin D1 promoter activity in IEC-18 cells, although CRE inactivation did reduce PMA-induced stimulation of the promoter (Fig. 10B). Notably, mutation of the CRE converted the NF-B site from a repressive to an activating promoter element; whereas mutation of the NF-B site increased basal activity of the wild-type promoter, the activity of the CRE/NF-B double mutant was less than that of the promoter with only the CRE mutated. Thus, the repressive activity of the NF-B site in IEC-18 cells requires interaction between the NF-B site and the CRE. Collectively, these data indicate that, although there are subtle differences between the regulation of the cyclin D1 promoter in DLD1 and IEC-18 cells, cooperation between factors that bind to the CRE and NF-B site mediates the up-regulation of cyclin D1 transcription induced by prolonged PKC agonist treatment in both non-transformed intestinal epithelial cells and colon cancer cells.

DISCUSSION
The findings presented herein cast new light on the regulation of cyclin D1 by members of the PKC family. PKC␣ and PKC⑀ have opposing effects on intestinal tumorigenesis, with PKC␣ acting as a tumor suppressor (14,22) and PKC⑀ exhibiting tumor-promoting activity (24,52). Our previous studies have determined that PKC␣ potently down-regulates cyclin D1 in intestinal epithelial cells and that loss of cyclin D1 represents an important component of the growth and tumor-suppressive effects of this isozyme (11,14,17,19). The present study indicates that these negative effects of PKC␣ are opposed by PKC⑀, which up-regulates cyclin D1 transcription in both non-transformed IEC-18 cells and colon cancer cells. The localization of PKC⑀ activity in intestinal tissues supports this function. In contrast to other PKC isozymes, which are predominantly activated in postmitotic cells of the intestinal villus and colon surface mucosa, activated (membrane-associated) PKC⑀ is also detected in proliferating crypt cells (30). Although activation of PKC⑀ in villus/surface mucosa cells presumably reflects its role in differentiated functions, such as mucin expression and chloride transport (53,54), the PKC⑀ activity in crypt cells coincides with expression of cyclin D1 in this tissue (19). Furthermore, whereas several PKCs, such as PKC␣ and PKC␦, tend to be lost during intestinal tumorigenesis, PKC⑀ is retained in colon cancer cells (14). Collectively, these findings point to PKC⑀ as a physiologically relevant regulator of cyclin D1 both in intestinal homeostasis and during tumor development.
Analysis of downstream signaling pathways determined that efficient recovery of cyclin D1 expression from PKC␣-induced down-regulation as well as its eventual hyperinduction require sustained ERK activity (Fig. 3). ERK activation is a common downstream effect of PKC⑀ (24,55) that has been linked to its oncogenic effects in intestinal epithelial cells (52) and the prostate (56). Interestingly, PKC␣-mediated down-regulation of cyclin D1 is also ERK-dependent (18,27). However, the temporal requirement for ERK activation in the effects of these two isozymes differs; whereas PKC⑀-dependent hyperinduction of cyclin D1 requires ERK activity for Ͼ4 h following PMA addition (Fig. 3), 1 h of ERK activity is sufficient to elicit the full growth-inhibitory effects of PKC␣ (27). Divergent effects of ERK signaling can result from differential timing and localization of activated ERK (33). In this regard, it is noteworthy that PMA treatment of IEC-18 cells promotes nuclear and cytoplasmic accumulation of phospho-ERK (27). Nuclear accumulation of active ERK, which is linked to the growth-inhibitory effects of PKC␣, is short-lived. In contrast, cytoplasmic ERK activity is more prolonged, consistent with the duration of ERK activation required for PKC⑀-dependent induction of cyclin D1 expression. Because divergent effects of PKC␣ and PKC⑀ on the localization and timing of ERK activation have been noted in other cell types (e.g. see Refs. 57 and 58), differential regulation of ERK signaling is likely to be a common mechanism by which individual PKC isozymes elicit distinct responses.
Consistent with these differences in downstream signaling, PKC␣ and PKC⑀ are not directly antagonistic but appear to modulate cyclin D1 expression via distinct mechanisms (see Fig. 11). The inhibitory activity of PKC␣ involves both translational and transcriptional repression (11,14), whereas the stimulatory effects of PKC⑀ appear to be predominantly transcriptional (Figs. 4, 5, and 7). PKC␣ signaling blocks cyclin D1 translation through rapid (within 30 min) PP2A-dependent dephosphorylation of 4E-BP1 (11,18); however, PKC agonists do not elicit changes in 4E-BP1 phosphorylation or PP2A activity when PKC␣ activity is selectively inhibited, excluding a role for PKC⑀ in modulation of this pathway. At the transcriptional level, PKC␣ exerts its inhibitory effects through cyclin D1 promoter sequences between 1745 and 163 bp upstream of the transcriptional start site (14), whereas the effects of PKC⑀ are mediated by sequences in the proximal 163 bp of the promoter (Figs. 5 and 7).
The opposing signals from PKC␣ and PKC⑀ result in biphasic effects of PKC agonists on cyclin D1 expression in IEC-18 cells. The initial agonist-induced down-regulation indicates that the FIGURE 11. Model of the differential effects of PKC␣ and PKC⑀ signaling on cyclin D1 expression in intestinal epithelial cells. We have previously demonstrated that PKC␣ signaling mediates translational and transcriptional repression of cyclin D1 in intestinal epithelial cells via ERK-dependent mechanisms (14,18,27). Translational repression involves PP2A-mediated activation of 4E-BP1, and transcriptional inhibition is mediated by an element(s) between Ϫ163 and Ϫ1745 bp of the cyclin D1 promoter. The current study further demonstrates that PKC⑀ promotes ERK-dependent transcriptional activation of the cyclin D1 promoter, via positive cooperation between NF-B-and CRE-interacting protein(s) near the transcriptional start site. The effects of PKC␣ are dominant over those of PKC⑀ in these cells. Ongoing studies are investigating the mechanisms underlying the effects of PKC␣ on PP2A and the B subunit involved, and a role for ␣4 protein has been excluded.
repressive effects of PKC␣ are dominant over those of PKC⑀ at early time points. However, PKC␣-mediated repression of cyclin D1 is transient, and reversal of the effect correlates with agonist-induced down-regulation of the isozyme ( Fig. 2A) (17,31). Although the recovery and increased expression of cyclin D1 can be explained, in part, by loss of repressive effects of PKC␣, positive effects of PKC⑀ play an important role both in the timing of the recovery and in the extent of cyclin D1 up-regulation. Hyperinduction of cyclin D1 occurs at times when repressive effects of PKC␣ are still evident, albeit more weakly, in IEC-18 cells. For example, cyclin D1 is hyperinduced at 6 h of PMA treatment despite the continued presence of elevated levels of the active ␣ and ␤ forms of 4E-BP1 (Fig. 4C). The involvement of PKC⑀ in overcoming these persistent repressive effects is reflected in the fact that cyclin D1 remains below basal levels at 6 h of PMA treatment in cells in which PKC⑀ signaling is abrogated by siRNA-mediated knockdown or by inhibition of downstream ERK activation (Figs. 2C and 3). A requirement for PKC⑀ in cyclin D1 hyperinduction is also seen in (a) the ability of general PKC inhibitors to abrogate PMA-induced cyclin D1 up-regulation and (b) the fact that the effect is still seen when PKC␣ activity is selectively inhibited with Gö6976 and in DLD1 colon cancer cells that lack PKC␣ activity (14). Furthermore, whereas overexpression of PKC␣ enhanced and prolonged cyclin D1 down-regulation in agonist-treated cells, PKC⑀ overexpression attenuated PKC␣-induced cyclin D1 suppression while increasing the extent of its eventual hyperinduction. Thus, the balance of the relative strengths of signaling from these isozymes dictates the levels of cyclin D1 in intestinal cells. The consequences of disrupting this balance are evident in colon cancer cells, where the absence of functional PKC␣ signaling not only prevents the initial down-regulation of cyclin D1 in response to PKC-agonist treatment but is also associated with earlier cyclin D1 induction (Figs. 1A, 2A, and 6A). Restoration of PKC␣ in colon cancer cells, on the other hand, decreases basal expression of cyclin D1, re-establishes the ability of PKC agonists to down-regulate the cyclin (14), and delays its eventual PKC⑀-mediated hyperinduction.
Our data in IEC-18 and DLD1 cells demonstrate that NF-B activity is required for PKC⑀-mediated up-regulation of cyclin D1 in intestinal epithelial cells. Studies from a number of systems have established the ability of PKC signaling, and PKC⑀ in particular, to activate NF-B through the canonical IB-regulated pathway (e.g. see Refs. 59 -61). However, PKC agonistinduced up-regulation of cyclin D1 promoter activity in intestinal epithelial cells was insensitive to expression of DA IB, excluding involvement of this pathway in the effects seen here. This finding is consistent with the report that activation of NF-B by PMA can be independent of IB degradation in DLD1 cells (39). In contrast to the canonical pathway, the noncanonical pathway generally gives rise to prolonged NF-B activation (62), further supporting a role for the latter in PKC⑀-mediated hyperinduction of cyclin D1 transcription. Involvement of noncanonical activation of NF-B is also consistent with the finding that p52⅐RelB (63,64) and p52/p52⅐Bcl3 complexes (51,65) can activate the cyclin D1 promoter. Current efforts are determining the involvement of these mechanisms in the observed effects of PKC⑀ on cyclin D1 transcription.
Although PKC⑀-mediated transcriptional up-regulation of cyclin D1 is dependent on NF-B activity, the NF-B site in the proximal promoter is not required for the effect. Rather, induction of cyclin D1 transcription involves interaction between NF-B activity and the neighboring CRE. Although the precise mechanisms underlying this cooperation remain to be established, the finding that either the CRE or NF-B site can support the induction of cyclin D1 transcription (Fig. 10) points to a role for physical interaction between NF-B proteins and factors that bind to the CRE. This CRE binds multiple proteins that have been shown to physically interact with NF-B family members, including JunD, ATF, c-Jun, and c-Fos (e.g. see Refs. 66 -68). As with the cooperation seen here, interaction of NF-B with c-Jun or c-Fos can support synergistic activation of transcription through promoter sites for either factor (67). Furthermore, in immortalized human hepatocyte cells, c-Jun can activate cyclin D1 promoter activity through the NF-B site in a DA IB-insensitive manner, pointing to c-Jun as a potential mediator of the observed cooperation in intestinal cells.
In addition to further characterizing regulation of cyclin D1 by PKC isozymes, the current study has revealed aspects of its transcriptional regulation that may be of direct relevance to the up-regulation of cyclin D1 in colon cancer cells. Mutational analysis determined that the NF-B site in the cyclin D1 promoter is a repressive element in IEC-18 cells (Fig. 10B). Notably, this NF-B site is also repressive in mammary epithelial and smooth muscle cells (37,64), indicating that negative regulation of cyclin D1 transcription through the NF-B site might be common in non-transformed cells. However, this site is not repressive in colon cancer cells (Fig. 10A) (e.g. see Ref. 69); thus, loss of the repressive effects of the NF-B site would appear to contribute to the up-regulation of cyclin D1 expression seen during intestinal tumorigenesis. Conversely, the CRE gains positive effects in colon cancer cells (Fig. 10, A and B), indicating that changes in factors that bind this site also play a role in increasing cyclin D1 expression in these tumors. These two phenomena appear to be related because the negative effects of the NF-B site in IEC-18 cells require interaction with an intact CRE (Fig. 10B). Thus, the ability of PKC⑀ to promote positive cooperation between NF-B and CRE-interacting protein(s) at the cyclin D1 promoter positions this isozyme to be an important player in intestinal transformation and is likely to account, in part, for its oncogenic activity.
In conclusion, the current study complements our previous findings that cyclin D1 is an important target for the tumorsuppressive effects of PKC␣. Given the importance of cyclin D1 in intestinal homeostasis and tumorigenesis, the antagonistic effects of PKC␣ and PKC⑀ on expression of this cyclin would, at least partially, explain their contrasting effects on intestinal tumor development. These findings highlight a role for opposing actions of PKC family members, which converge on cyclin D1 expression, as important regulators of tissue homeostasis and tumorigenesis in the intestine (Fig. 11).