Antimitogenesis linked to regulation of Skp2 gene expression.

Prostacyclin has many effects in the vasculature; one of the less well understood is the ability to block cell cycle progression through G(1) phase. We previously reported that the prostacyclin mimetic, cicaprost, selectively inhibits cyclin E-cyclin-dependent kinase-2 (Cdk2), and now we show that it acts by regulating the expression of Skp2, the F-box protein that targets p27(Kip1) for ubiquitin-mediated proteolysis. First, we show that cicaprost prevents the late G(1) phase down-regulation of p27(Kip1) and that the inhibitory effect of cicaprost on cyclin E-Cdk2 activity and S phase entry is eliminated by deleting p27(Kip1). Levels of the closely related Cdk2 inhibitor, p21(Cip1), are unaffected by cicaprost. Moreover, we show that cicaprost blocks the induction of Skp2 mRNA and that ectopic expression of a Skp2 cDNA overrides the effect of cicaprost on p27(Kip1) levels and S phase entry. Our data show that inhibition of F-box protein gene expression can underlie the effect of a potent antimitogen.

Prostacyclin has many effects in the vasculature; one of the less well understood is the ability to block cell cycle progression through G 1 phase. We previously reported that the prostacyclin mimetic, cicaprost, selectively inhibits cyclin E-cyclin-dependent kinase-2 (Cdk2), and now we show that it acts by regulating the expression of Skp2, the F-box protein that targets p27 Kip1 for ubiquitin-mediated proteolysis. First, we show that cicaprost prevents the late G 1 phase downregulation of p27 Kip1 and that the inhibitory effect of cicaprost on cyclin E-Cdk2 activity and S phase entry is eliminated by deleting p27 Kip1 . Levels of the closely related Cdk2 inhibitor, p21 Cip1 , are unaffected by cicaprost. Moreover, we show that cicaprost blocks the induction of Skp2 mRNA and that ectopic expression of a Skp2 cDNA overrides the effect of cicaprost on p27 Kip1 levels and S phase entry. Our data show that inhibition of F-box protein gene expression can underlie the effect of a potent antimitogen.
The endothelium regulates vascular tone and homeostasis in part by the release of small molecules that regulate medial smooth muscle cells. One of the major endothelial cell products is the prostanoid, prostacyclin (1), which inhibits platelet activation and smooth muscle cell proliferation (2). Injury to the endothelium or disturbances in laminar blood flow inhibits the production of cyclooxygenase-2 (3), the predominant source of prostacyclin biosynthesis in humans (4). The resulting decrease in production and release of prostacyclin is thought to contribute to both atherosclerosis and restenosis. In support of this notion, deletion of IP (the heterotrimeric G-protein-coupled receptor for prostacyclin) increases susceptibility to thrombosis (5), accelerates atherogenesis (6), and exacerbates injury-induced smooth muscle cell proliferation and neointima formation (7).
Prostacyclin inhibits the proliferation of vascular smooth muscle cells by blocking progression from the G 1 to the S phase of the cell cycle (8). G 1 phase progression is regulated by a sequential activation of the G 1 phase cyclin-dependent kinases (Cdks), 1 namely cyclin D1-Cdk4/6 and cyclin E-Cdk2. The ac-tivation of Cdk4/6 in mid-G 1 phase is typically regulated by the mitogen-dependent induction of a D-type cyclin, usually cyclin D1 in mesenchymal cells. In contrast, cyclin E is expressed throughout the G 1 phase, and the activation of Cdk2 is typically regulated by the rate at which Cdk2 inhibitors, p21 Cip1 and p27 Kip1 , are down-regulated (9). Once activated, cyclin D-Cdk4/6 and cyclin E-Cdk2 phosphorylate the pocket protein family of transcriptional regulators (pRb, p107, p130), which in turn modulate the activation of E2F-dependent genes such as cyclin A (10). cyclin A is induced at the G 1 /S interface, and the consequent formation of cyclin A-Cdk2 complexes marks entry into the S phase of the cell cycle.
We recently reported that the prostacyclin mimetic, cicaprost, inhibits G 1 to S phase cell cycle progression in aortic smooth muscle cells by inhibiting CREB-and pocket proteindependent cyclin A gene expression (11). The inhibitory effect of cicaprost on pocket protein phosphorylation was associated with a selective inhibition of cyclin E-Cdk2 activity; the induction of cyclin D1 and the activation of cyclin D-Cdk4 were unaffected by cicaprost. Cicaprost did not affect the levels of cyclin E or Cdk2, but we were unable to define the molecular mechanism underlying the inhibitory effect of cicaprost on cyclin E-Cdk2 activity, probably because the smooth muscle cells used in that study (primary murine smooth muscle cells and the A10 smooth muscle cell line) do not cycle efficiently. In this study, we used established mouse embryo fibroblasts (MEFs) to show that cicaprost exerts its inhibitory effect on cyclin E-Cdk2 activity by specifically regulating the expression of p27 Kip1 . Furthermore, we link this effect on p27 Kip1 to a cicaprost-mediated block in expression of the Skp2 gene. The Skp2 gene encodes the substrate-targeting component of the E3 ubiquitin ligase complex that degrades p27 Kip1 (12,13). These data identify a new mechanism for antimitogenesis with potential implications for cardiovascular disease.

EXPERIMENTAL PROCEDURES
Cell Culture-Spontaneously immortalized embryo fibroblasts from wild-type and p27 Kip1 Ϫ/Ϫ mice were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS). In cell cycle experiments, near confluent monolayers of MEFs were G 0 synchronized by incubation in a serum-free Dulbecco's modified Eagle's medium supplemented with 1 mg/ml heat-inactivated fatty acid-free bovine serum albumin for 36 -48 h. The quiescent cells were trypsinized, reseeded in culture plates, and stimulated with 10% FBS as described (14) in the absence or presence of 200 nM cicaprost. For HA-Skp2 overexpression studies, MEFs were transfected as described (14) using LipofectAMINE Plus and 5 g of either pcDNA3.1 (vector control) or a pcDNA3.1-based HA-tagged mouse Skp2 expression vector (HA-Skp2). After an overnight recovery, the cells were serum-starved for 36 h. Serum-starved transfectants were trypsinized, reseeded (ϳ10 6 cells per 10 ml of 10% FBS-Dulbecco's modified Eagle's medium) into 100-mm tissue culture dishes, and incubated in the absence or presence of 200 nM cicaprost. Cicaprost was the generous gift of Garret FitzGerald.
To monitor entry into S phase, quiescent MEFs were added to 100-mm (ϳ10 6 cells) dishes containing autoclaved glass coverslips and incubated in 2 ml of Dulbecco's modified Eagle's medium, 10% FBS, 3 g/ml BrdUrd (Amersham Biosciences) in the absence or presence of * This work was supported by National Institutes of Health Grants GM51878 and HL62250 (to R. K. A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  1 The abbreviations used are: Cdk, cyclin-dependent kinase; CRE, cyclic AMP response element; CREB, cyclic AMP response element binding protein; MEF, mouse embryo fibroblast; FBS, fetal bovine serum; HA, hemagglutinin; ERK, extracellular signal-regulated kinase; BrdUrd, bromodeoxyuridine; SCF, Skp1-cullin-F box protein.
200 nM cicaprost. Formaldehyde-fixed cells were incubated with anti-BrdUrd as described (11). Approximately 200 cells in 3-4 fields of view were counted per sample to determine the percent of BrdUrd-labeled nuclei. Results shown are typically representative of at least three independent experiments.
Cyclin A promoter-luciferase assays were performed as described (11) using the p434/cyclin A promoter-luciferase expression vector (bases Ϫ270 to ϩ164 of the human cyclin A promoter cloned upstream of a firefly luciferase reporter (15)), the Renilla luciferase vector (pRL-CMV), and an E7 expression vector. Results shown are representative of three independent experiments.
Quantitative Real Time Reverse Transcriptase-PCR-Quiescent MEFs were trypsinized, replated at subconfluence (ϳ5 ϫ 10 5 cells in 5 ml of medium per 60-mm culture dish), and incubated in 10% FBS in the absence or presence of 200 nM cicaprost for selected times. Cells were washed with phosphate-buffered saline and lysed in 1 ml of TRIzol (Invitrogen) per 60-mm dish to extract total RNA. Total RNA (ϳ400 ng) from each sample was used for cDNA synthesis using reverse transcription reagents from Applied Biosystems Inc. Each sample was analyzed in duplicate, and mRNA expression was quantified using ABI Prism 7000 sequence detection system software. Mean quantities were calculated from duplicate samples. Skp2 mRNA expression was normalized to Cdk4 mRNA expression, which did not vary reproducibly during serum stimulation or cicaprost treatment. Results shown are representative of three independent experiments.

Cicaprost Inhibits S Phase Entry in MEFs by Selectively
Regulating p27 Kip1 -We reported previously that cicaprost selectively inhibited the activation of cyclin E-Cdk2 in vascular smooth muscle cells. Yet levels of cyclin D1, the activation of Cdk4, and the levels of cyclin E and Cdk2 were unaffected (11). However, inefficient cycling of those cells prevented us from characterizing the responsible mechanism in further detail. Therefore we tested the effect of cicaprost in established MEFs, which efficiently re-enter G 1 phase from quiescence (Fig. 1A,  control). Cicaprost strongly inhibited S phase entry in MEFs (Fig. 1A) with a dose response that was similar to that of vascular smooth muscle cells (IC 50 ϳ30 nM; data not shown). Moreover, cicaprost blocked the phosphorylation of pRb and the induction of cyclin A in MEFs (Fig. 1B) in a manner very similar to that seen in vascular smooth muscle cells (11). However, unlike its effects in vascular smooth muscle cells (11), cicaprost did not regulate CREB-dependent cyclin A gene ex-pression in MEFs and did not inhibit the phosphorylation of CREB or the binding of CREB and phospho-CREB to the cyclin A CRE (supplemental Fig. 1, A and B, respectively). Moreover, the inhibitory effect of cicaprost on wild-type cyclin A promoter activity in MEFs was overcome by ectopic expression of human papillomavirus E7, a protein that sequesters pocket proteins and mimics inactivation by the G 1 phase Cdks (Fig. 1C). E7 expression fails to overcome the inhibitory effect of cicaprost on cyclin A promoter activity in vascular smooth muscle cells unless the CRE-inhibitory effect of cicaprost is first eliminated by the use of a CRE-deleted or mutated promoter (11). Thus, the antimitogenic effect of cicaprost on pocket protein phosphorylation is conserved in these mesenchymal cells whereas the antimitogenic effect on CREB is not.
As with vascular smooth muscle cells, cicaprost did not affect the levels of cyclin E and Cdk2 in MEFs ( Fig. 2A) but strongly inhibited the late G 1 phase activation of cyclin E-Cdk2 as determined by both the gel shift of Cdk2 (a hallmark of cdkactivating kinase mediated Cdk2 activation; Fig. 2A) and in vitro kinase assays (Fig. 2B). Cicaprost also blocked the downregulation of p27 Kip1 but did not affect the G 1 phase levels of p21 Cip1 (Fig. 2A). The absence of an effect on p21 Cip1 suggested FIG. 1. Cicaprost blocks S phase entry through a pocket proteindependent pathway. Quiescent MEFs were trypsinized, reseeded at subconfluence into 100-mm dishes containing coverslips, and incubated with 10% FBS in the presence of BrdUrd with and without 200 nM cicaprost for the indicated times. A, fixed coverslips were incubated with anti-BrdUrd, and the percent of BrdUrd-labeled nuclei was determined by immunofluorescence microscopy. B, collected cells were lysed and analyzed by Western blotting for the expression of pRb and cyclin A. Upper and lower arrows represent the hyper-and hypophosphorylated forms of pRb, respectively. C, cyclin A promoter activity was determined in the absence and presence of cicaprost and E7. Cyclin A promoter-luciferase activity is plotted relative to Renilla luciferase activity; promoter activity at 18 h in the absence of E7 is defined as 1.0. that the antimitogenic effect of cicaprost was not caused by mere inhibition of mitogenic signaling, and indeed we found that cicaprost did not inhibit activation of several well established mitogenic targets, including ERK1/2, AKT, and cyclin D1 (Fig. 2C). Finally, the effects of cicaprost on p27 Kip1 levels were causally related to its antimitogenic effect; cicaprost strongly blocked cyclin E-Cdk2 activity, cyclin A induction, and S phase entry in wild-type MEFs but not in MEFs lacking p27 Kip1 (Fig. 3, A and B). Overall, these results show that the selective effect of cicaprost on the levels of p27 Kip1 is causally related to its antimitogenic effect on S phase entry.
Cicaprost Inhibits Expression of the Skp2 Gene-One of the most common mechanisms for regulating p27 Kip1 involves its phosphorylation at Thr-187 by cyclin E-Cdk2, an event that targets p27 Kip1 for degradation by the SCF ubiquitin-ligase complex (17,18). The substrate-targeting component of the SCF complex that is required for the ubiquitin-mediated degradation of p27 Kip1 is the F-box protein, Skp2 (12,13). Skp2 is poorly expressed in quiescent cells, and its expression is stimulated by mitogens during late G 1 and S phase (12, 19) (Fig. 4,  A and B; control). Cicaprost blocked the serum-stimulated expression of Skp2 (Fig. 4A), and again this effect was not caused by an overall inhibition of mitogenic signaling (Fig. 2C).
Recent reports have identified the anaphase promoting complex Cdh1 complex as a major regulator of Skp2 proteolysis as cells progress from mitosis into G 1 as well as when cells are serum-starved to enter quiescence (20,21). Although these results raised the possibility that cicaprost could regulate Skp2 levels post-transcriptionally by regulating anaphase promoting complex Cdh1 , quantitative real time PCR showed that cicaprost strongly inhibited the induction of Skp2 mRNA (Fig. 4C). Moreover, we could rescue the expression of Skp2, the down-regulation of p27 Kip1 , the induction of cyclin A, and entry into S phase in cicaprost-treated cells by transfecting cells with a Skp2 cDNA and enforcing the expression of ectopic Skp2 at near normal levels (Fig. 5, A and B). The rescue of ectopic Skp2 protein expression in cicaprost-treated cells would be highly unlikely to occur if cicaprost strongly stimulated Skp2 proteolysis. Thus, our data indicate that prostacyclin is antimitogenic because it inhibits gene expression of the F-box protein that controls the degradation of p27 Kip1 . Note that MEFs expressing HA-Skp2 have reduced levels of endogenous Skp2 (Fig. 5A) consistent with a study indicating that Skp2 itself is a target of a Skp2-containing SCF complex (22).
8-Bromo-cAMP Induces G 1 Arrest without Affecting Skp2-Cicaprost binds to the IP receptor, a heterotrimeric G-proteincoupled receptor that canonically couples to G s and increases cAMP (23). Because cAMP can increase p27 Kip1 levels and inhibit S phase entry (24), we asked whether the effect of cicaprost on Skp2 could be recapitulated merely by increasing were incubated with anti-cyclin E. The collected immunoprecipitates were used to assess in vitro kinase activity and then analyzed by Western blotting for associated Cdk2. C, collected cells were lysed and analyzed by Western blotting for phospho-ERK, phospho-AKT, cyclin D1, and Cdk4 (loading control).

FIG. 3. The antimitogenic effect of cicaprost requires p27 Kip1 .
Quiescent wild-type and p27 Kip1 Ϫ/Ϫ MEFs were trypsinized, reseeded at subconfluence into 100-mm dishes containing coverslips, and stimulated with 10% FBS in the presence of BrdUrd with and without 200 nM cicaprost for the indicated times. A, collected cells were lysed, and equal amounts of protein (150 g) were incubated with anti-cyclin E. The collected immunoprecipitates were used to assess in vitro kinase activity and then analyzed by Western blotting for associated Cdk2. Duplicate lysates were analyzed by Western blotting for p27 Kip1 , cyclin A, and Cdk4 (loading control). B, fixed coverslips were incubated with anti-BrdUrd, and BrdUrd incorporation into nuclei was determined by immunofluorescence microscopy. cAMP. Consistent with previous studies (24), the cAMP analog, 8-bromo-cAMP, increased p27 Kip1 levels, inhibited the induction of cyclin A, and blocked S phase entry in a dose-dependent manner (Fig. 6, A and B). However, this effect on p27 Kip1 was not accompanied by a decreased expression of Skp2 (Fig. 6A). Similar results were seen with forskolin (data not shown). These results show that the antimitogenic effect of cicaprost on Skp2 expression is not a general response to increased cAMP levels and indicate that the effects of prostacyclin and cAMP on p27 Kip1 are mechanistically distinct. They also emphasize that the inhibitory effect of cicaprost on the expression of Skp2 is not merely a secondary consequence of G 1 phase arrest; MEFs treated with 1 mM 8-bromo-cAMP are arrested in G 1 phase yet still express Skp2 (Fig. 6, A and B). DISCUSSION Our results show that prostacyclin exerts its antimitogenic effects by selectively regulating p27 Kip1 . We first showed that cicaprost acts through a pocket protein-dependent pathway to inhibit cyclin A promoter activity. Moreover, cicaprost inhibited the activation of cyclin E-Cdk2 and the down-regulation of p27 Kip1 that accompanies cell cycle progression through G 1 phase, and these antimitogenic effects of cicaprost on cyclin E-Cdk2 activity, cyclin A induction, and S phase entry are overcome by deleting p27 Kip1 . Interestingly, p27 Kip1 -null MEFs retain the ability to undergo G 1 arrest in response to contact inhibition, serum starvation, and ␥-irradiation (25). Thus, although the inhibitory effect of p27 Kip1 can be compensated for in these settings, p27 Kip1 is essential for the antimitogenic effect of cicaprost.
Our data also indicate that prostacyclin regulates p27 Kip1 by controlling the expression of the F-box protein, Skp2. Cicaprost blocked Skp2 gene expression, and forced expression of Skp2 rescued p27 Kip1 down-regulation and S phase entry in cicaprost-treated cells. In addition to p27 Kip1 , others have reported that Skp2 targets cyclin E, p21 Cip1 , p130, c-Myc, and p57 Kip2 for proteolysis (26 -30). We saw no effect of cicaprost on p21 Cip1 or cyclin E under conditions in which the induction of Skp2 was clearly inhibited. Although the basis for these different results is not completely clear, SCF Skp2 primarily targets free cyclin E for degradation. In our established MEFs (26), most of the cyclin E may be complexed to Cdk2. Moreover, the effect of SCF Skp2 on p21 Cip1 degradation was studied in S phase (28); our study focuses on G 1 phase regulation.
Ubiquitin-mediated proteolysis appears to be the most common mechanism regulating p27 Kip1 levels during G 1 phase progression (31), but p27 Kip1 levels can also be regulated by an FIG. 4. Antimitogenic effect of cicaprost linked to inhibition of Skp2 gene expression. Quiescent MEFs were trypsinized, reseeded at subconfluence into 100-mm dishes containing coverslips, and stimulated with 10% FBS in the presence of BrdUrd with and without 200 nM cicaprost for the indicated times. A, collected cells were lysed and analyzed by Western blotting for the expression of Skp2, cyclin A, and Cdk4 (loading control). B, fixed coverslips were incubated with anti-BrdUrd, and BrdUrd incorporation into nuclei was determined by immunofluorescence microscopy. C, total RNA was extracted from collected cells and analyzed by quantitative real time reverse transcriptase-PCR for the expression of Skp2 and Cdk4 mRNAs. Skp2 mRNA expression is plotted relative to Cdk4 mRNA levels.
FIG. 5. Ectopic expression of Skp2 rescues p27 Kip1 down-regulation and S phase entry. Wild-type MEFs transiently transfected with either empty vector or HA-tagged mouse Skp2 were serumstarved, trypsinized, reseeded into 100-mm dishes containing coverslips, and stimulated with 10% FBS in the presence of BrdUrd with and without cicaprost for 18 h. A, collected cells were lysed and analyzed by Western blotting for the expression of Skp2, p27 Kip1 , cyclin E, cyclin A, and Cdk4 (loading control). B, fixed coverslips were incubated with anti-BrdUrd, and BrdUrd incorporation into nuclei was determined by immunofluorescence microscopy.
AKT-dependent transcription of the p27 Kip1 gene (32) and a RhoA-dependent translation of p27 Kip1 mRNA (33). Although we have not studied these potential mechanisms in detail, they are unlikely to be major targets of prostacyclin because (i) AKT activation is unaffected by cicaprost, and (ii) cicaprost-treated MEFs do not show the morphological changes expected to accompany RhoA inhibition. Moreover, inhibition of RhoA would strongly increase the level of p21 Cip1 (34), and cicaprost did not affect p21 Cip1 levels. Previous studies have shown that Skp2 mRNA levels are also reduced when cells are detached from their extracellular matrix (19); the relationship between prostacyclin and integrin signaling remains to be determined.
Cicaprost binds to and activates the IP receptor, a heterotrimeric G-protein-coupled receptor that is best characterized as an activator of G s. (35,23). G s increases cAMP production, and we and others (24) find that cAMP-elevating agents (8-bromo-cAMP and forskolin) increase the levels of p27 Kip1 and lead to G 1 phase arrest. However in striking contrast to cicaprost, 8-bromo-cAMP and forskolin inhibited the down-regulation of p27 Kip1 and entry into S phase without affecting the levels of Skp2. These data indicate that prostacyclin-dependent inhibition of Skp2 gene expression is not mediated merely by increasing the levels of cAMP.
Our study establishes a selective and causal relationship between the effects of prostacyclin on p27 Kip1 and G 1 phase cell cycle progression. Others (36) showed a correlation between the antimitogenic effects of a different prostacyclin analog, beraprost, and increased levels of p27 Kip1 after vascular injury in vivo. Together, these two reports strongly indicate that p27 Kip1 mediates the antimitogenic effect of prostacyclin on smooth muscle cell proliferation. By comparing the effects of 8-bromo-cAMP and beraprost on p27 Kip1 levels in cultured smooth muscle cells, Ii et al. (36) also concluded that prostacy-clin affects p27 Kip1 expression through cAMP signaling. However, we show that prostacyclin (but not cAMP) is able to inhibit expression of the Skp2 gene. This result is not restricted to MEFs because cicaprost also blocks Skp2 mRNA induction in primary murine smooth muscle cells (data not shown). In summary, our data identify a new mechanism mediating antimitogenesis. Moreover, because prostacyclin is a likely inhibitor of smooth muscle cell proliferation in atherosclerosis as well as restenosis (6,7), Skp2 may be a potential therapeutic target in cardiovascular disease.

FIG. 6. 8-Bromo-cAMP does not block Skp2 induction in MEFs.
Quiescent MEFs were trypsinized, reseeded at subconfluence into 100-mm dishes containing coverslips, and stimulated with 10% FBS in the absence (control) or presence on increasing doses of 8-bromo-cAMP (100 M, 200 M, 500 M, and 1 mM) for the indicated times. A, collected cells were lysed and analyzed by Western blotting for the expression of p27 Kip1 , Skp2, cyclin A, and Cdk4 (loading control). B, coverslips collected from cells incubated with BrdUrd were analyzed using immunofluorescence microscopy to determine the percent of BrdUrd-labeled nuclei.