p27 Suppresses Cyclooxygenase-2 Expression by Inhibiting p38β and p38δ-Mediated CREB Phosphorylation Upon Arsenite Exposure

Abstract

p27 is a cyclin-dependent kinase (CDK) inhibitor that suppresses a cell’s transition from G₀ to S phase, therefore acting as a tumor suppressor. Our most recent studies demonstrate that upon arsenite exposure, p27 suppresses Hsp27 and Hsp70 expressions through the JNK2/c-Jun- and HSF-1-dependent pathways, suggesting a novel molecular mechanism underlying the tumor suppressive function of p27 in a CDK-independent manner. We found that p27-deficiency (p27−/−) resulted in the elevation of cyclooxygenase-2 (COX-2) expression at transcriptional level, whereas the introduction of p27 brought back COX-2 expression to a level similar to that of p27+/+ cells, suggesting that p27 exhibits an inhibitory effect on COX-2 expression. Further studies identified that p27 inhibition of COX-2 expression was specifically due to phosphorylation of transcription factor cAMP response element binding (CREB) phosphorylation mediated by p38β and p38δ. These results demonstrate a novel mechanism underlying tumor suppression effect of p27 and will contribute to understanding of the overall mechanism of p27 tumor suppression in a CDK-independent manner.

1. Introduction

Arsenic is a metalloid chemical element that widely occurs in water, soil, minerals, and industrial products [1]. It mostly remains as sulfide or oxide in the environment. Some of the most common arsenic compounds are referred to as arsenite, which denotes any salt of arsenious acid comprised of trivalent arsenic [2]. Though arsenite in subtoxic doses has been developed for medical use, its toxicity [3] and environmental carcinogenic effect [4] still have to be considered. Occupational exposure occur in industries involving nonferrous metal alloys, electronic semiconductor manufacture, and wood preservation; and is a known cause of skin and epithelial cancers [5]. Generation of reactive oxygen species, genetic toxicity, and alteration of signaling pathways caused by arsenite are potentially responsible for arsenite-induced carcinogenic effects [6]. Our previous studies have shown that arsenite exposure initiates activation of several signal pathways and further alters downstream gene expression and cell transformation [7–9]. Cyclooxygenase-2 (COX-2) is known as a Prostaglandin endoperoxide synthase-2 (PTGS-2). COX-2 is upregulated during inflammatory responses, but is undetectable under normal conditions [10]. COX-2 is found to be highly overexpressed in various cancer tissues and cells. Further studies from both in vitro and in vivo show that COX-2 and its products, such as Prostaglandin E2 (PGE2) and H2 (PGH2), are involved in carcinogenesis at both the tumor promotion and progression stages, through multiple mechanisms that include anti-apoptosis and mediation of cancer cell proliferation [11, 12]. As an inducible enzyme, COX-2 expression is precisely regulated by its upstream pathways leading to alteration of transcription factors. Our previous studies have demonstrated that arsenite exposure induces COX-2 expression via the NFAT-dependent pathway in human bronchial epithelial Beas-2B cells, while COX-2 induction by arsenite in mouse epidermal Cl41 cell relies on NFκB activation [13]. Together with other findings indicating that COX-2 protein levels are also regulated at its mRNA stability, protein translation, and protein degradation, COX-2 expression is controlled by multiple pathways at various levels, which depend on cell types and stimuli.

p27, a member of the Cip/Kip family of CDK inhibitor proteins, acts as a CDK inhibitor, which affects the function of cyclin protein, leading to cell cycle arrest in G1 phase [14, 15]; thus, p27 is considered as a tumor suppressor [16, 17]. However, recent studies also reveal that p27 functions in a CDK-independent manner [18, 19]. Our most recent studies demonstrate that p27 suppressed Hsp27 and Hsp70 expressions at the transcriptional level specifically through JNK2/c-Jun- and HSF-1-dependent pathways upon arsenite exposure, which provides an additional important molecular mechanism for the tumor suppressive function of p27 [7]. In this study, we investigated the potential role of p27 in regulating the expression of COX-2, a key mediator involved in inflammation and carcinogenesis.

2. Materials and Methods

2.1. Cell Culture and reagents

Mouse epidermal JB6 Cl41 cells were cultured in Minimal Essential Medium (MEM) with 5% FBS. Wild type of MEFs and various lines of the corresponding knockout cells p27Δ51/Δ51(p27−/−) were described in our previous studies [7]; p38α−/−, p38β−/− and p38δ−/− were provided by Dr. Jiahuai Han from School of Life Sciences, Xiamen University [20]. All MEFs and their stable transfectants were maintained in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% Fetal Bovine Serum (FBS), 2 mM L-glutamine, and 25 μg/ml of gentamicin at 37°C in a 5% CO₂ incubator. The cultures were dissociated with trypsin and transferred to new 75-cm² culture flasks (Fisher, Pittsburgh, PA) twice a week. FBS was purchased from Nova-Tech (Grand Island, NE). Sodium arsenite was purchased from Aldrich (Milwaukee, WI). The inhibitor of JNK, SP600125, inhibitor of p38 kinase, SB202190, and actinomycin D were from Calbiochem (La Jolla, CA). The inhibitor for MEK1/2, PD98059, was from Cell Signaling Technology (Beverly, MA). Antibodies against non-phosphorylated c-Jun, CREB, JNK1/2, p38, MKK3 phosphor-c-Jun at Ser63/Ser73, phosphorylated CREB at Ser133, phosphor-JNK at Thr183/Tyr185, phosphor-ERK at Thr202/Tyr204 phosphor-p38 kinase at Thr180/Tyr182, phosphor-MKK3 at Ser 189/MKK6 at Ser 207 were purchased from Cell Signaling Technology (Beverly, MA); Antibodies specifically against Jun B, Fra-1, and p27 were purchased from Santa Cruz Biotechnology Inc (Santa Cruz, CA). Antibody against β-Actin was from Sigma; Antiserum against COX-2 was from Cayman Chemical Co (Ann Arbor, MI).

2.2. Construct and Transfection

Dominant negative mutant p38 (DN-p38) plasmid, were generously provided by Dr. Roger J. Davis (University of Massachusetts Medical School) [21]. The above plasmid, as well as the empty vector, was stably transfected into p27−/− cells together with puromycin-resistant plasmid. The stable transfectants of DN-p38 were established by puromycin selection (4 μg/ml) and named as p27−/−(DN-p38). The CREB shRNA and its nonsense plasmid control were purchased from Open Biosystems (Thermo Fisher Scientific, Huntsville, AL). p27−/− MEFs were transfected with these plasmids to establish the CREB knockdown cell line which was named as p27−/− (shCREB), as well as its control cell line, after selection by puromycin (2 μg/ml). The shRNA-p27 (targeting mouse CGC AAG TGG AAT TTC GAC TT) was from Open Biosystems (Thermo Fisher Scientific, Huntsville, AL), and was transfected stably into Cl41 cells. The stable transfectants were obtained by puromycin selection (8 μg/ml). COX-2 luciferase reporter containing human COX-2 promoter sequences from −327 to +59 relating to the transcription initiation was kindly provided by Prof Kotha Subbaramaiah (Weill Medical College of Cornell University) [22].

2.3. Recombinant Adenovirus Construct and Infection

The AdEasy vector system (Quantum Biotechnologies, Montreal, Canada) and p27 cDNA were used to construct the recombinant adenoviral (Ad) vector. Shuttle plasmids were firstly electroporated into the recombinogenic Escherichia coli BJ5183 strain and selected. The virus was then transfected into HEK293 cells. Supernatant of lysed HEK293 cells were saved at −80 °C until they were used for infection. Detailed information regarding the plasmid, as well as the processes used for virus amplification, screening, and infection were described in our previous study [7].

2.4. Cell proliferation assay

p27+/+ and p27−/−(Δ51) cells were suspended in DMEM supplemented with 10% FBS in a concentration of 5×10⁴/mL. Cells in 100 μL DMEM suspension was cultured to each well of 96 well plates. After 12 hours of incubation at 37°C in a humidified atmosphere of 5% CO₂, cells were exposed to arsenite in a concentration of 20 μM for 6 to 12 hours. The exposed cells were lysed with 50 μL lysis buffer, and the proliferation of the cells was measured using CellTiter-Glo Luminescent Cell Viability Assay kit (Promega, Madison, WI, USA) with a luminometer (Wallac 1420 Victor2 multipliable counter system; PerkinElmer Life and Analytical Sciences, Inc., Waltham, MA, USA). The results are expressed as proliferation relative to control medium (proliferation index).

2.5. Reverse transcription polymerase chain reaction (RT-PCR)

Total RNA was extracted from the cells using Trizol reagent (Invitrogen, Carlsbad, California). Total cDNAs were synthesized by ThermoScript^™ RT-PCR system (Invitrogen, Carlsbad, California). The mRNA amount present in the cells was measured by semi-quantitative RT-PCR. The primers for mouse cox-2 were 5′-TCC TCC TGG AAC ATG GAC TC-3′ and 5′-GCT CGG CTT CCA GTA TTG AG-3′. The control mouse β-actin mRNA was also detected by RT-PCR using the primers (5′-GAC GAT GAT ATT GCC GCA CT-3′ and 5′-GAT ACC ACG CT T GCT CTG AG-3′). The PCR products were separated on 2% agarose gels, stained with EB. The results were displayed with the Alpha Innotech SP image system (Alpha Innotech Corporation., San Leandro, CA) and the densitometric analyses of the product bands were conducted using the software of ImageQuant 5.2 (GE Healthcare).

2.6. Luciferase reporter assay

Wild type MEF cells, p27−/− MEFs, p27−/−(DN-p38) cells, p27−/−(shCREB) cells and their corresponding p27−/− MEFs transfected with control vector were transfected or co-transfected with COX-2 promoter luciferase reporter. Puromycin selection for establishment of stable transfectants for p27−/−(DN-p38) cells, p27−/−(shCREB) cells and their corresponding p27−/− control MEFs is described in the “Constructs and Transfection” section. Cellular lysates were prepared after arsenite treatment and luciferase activities were determined using a luminometer (Wallac 1420 Victor 2 multilabel counter system). For wild type MEFs and p27−/− MEFs, the arsenite treatment was performed 24 hours after the transient transfection.

2.7. Nuclear/cytosol fractionation

For preparation of cell nuclear/cytosol protein fractionations, p27+/+ and p27−/− MEFs were cultured in 10 mm dishes till 80% confluent. Cells were then exposed to arsenite (20 μM) for 9 hours, and were harvested and processed with a Nuclear/Cytosol kit (BioVision, Mountain, CA) to extract the nuclear and cytosol protein as described in our previous studies [23].

2.8. Western blotting

MEFs and their transfectants were plated in 6-well plates and cultured in normal 10% FBS DMEM to 70–80% confluence. After treatments indicated, cells were extracted and total protein was quantified with a DC protein assay kit (Bio-Rad, Hercules, CA). Western Blotting was carried out as previously described [24]. Primary antibody-bound proteins were detected using an alkaline phosphatase-linked secondary antibody and an ECF Western blotting system (Amersham, Piscataway, NJ). The results shown were representative of three independent experiments.

3. Results

3.1. p27 suppressed COX-2 induction by arsenite

Since skin is a well-known major target of arsenite, we determined COX-2 induction in mouse epidermal Cl41 cells by arsenite at either low dose (0.5–2 μM) chronic exposures or relative high dose (20 μM) acute exposure. The results showed that an acute exposure to 20 μM arsenite or repeated chronic exposure to 1 μM of arsenite two months induced a similar level of COX-2 protein expression in Cl41 cells (Figs. 1A and 1B). Thus, we used 10–20 μM arsenite in current study. Genetic knockout of a specific gene is a very unique molecular approach to address the function of such gene in a biological function. Spontaneous immortalization of cell line from a specific gene knockout mouse has long been considered as one of the most efficient and available approach for establishing gene knockout MEFs. To determine the potential role of p27 in the regulation of COX-2 induction by arsenite, we employed p27−/− MEFs and its accordingly wild type (p27+/+) MEFs that has been used in our previous studies [7] and was also indicated in the results in Western Blot and RT-PCR assays (Figs. 1C and 1D). Cell treatment with 20 μM arsenite for 9 hours resulted in COX-2 induction in p27+/+ cells. And this induction was markedly increased in p27−/− cells (Fig. 1E), suggesting that p27 expression provided a strong inhibitory effect on COX-2 induction by arsenite. To confirm that increased COX-2 expression in p27−/− MEFs is specifically due to p27 deficiency, adeno-p27 virus was applied to p27−/− MEFs to recover p27 expression. As shown in Fig. 1F, wild type p27 protein was detected in p27−/− cells infected with the adeno-p27 virus. Restoration of p27 expression in p27−/− MEFs dramatically inhibited COX-2 protein expression upon arsenite exposure compared to that in its parental p27−/− MEFs (Fig. 1F). The results obtained from ATPase assay showed that treatment of cells with 20 μM for 12 hours did not lead to significant cell death in both p27+/+ and p27−/−(Δ51) cells (Fig. 1G). To confirm this result in Cl41 cell, we used specific mouse p27 shRNA to knockdown p27 expression in Cl41 cells to determine whether the inhibitory effect of p27 on COX-2 induction by arsenite could be verified in Cl41 cells. As expected, introduction of p27 shRNA into Cl41 cells led to over 80% reduction of p27 protein expression in comparison with that in nonsense shRNA transfectant Cl41 sh-p27 (Fig. 1H). Consistently, knockdown of p27 expression in Cl41 (Cl41 sh-p27) cells also increased COX-2 protein induction upon arsenite treatment compared to Cl41-nonsense transfectant (Fig. 1I). These results demonstrate that p27 expression inhibits COX-2 induction via arsenite exposure.

Figure 1. p27 inhibited arsenite-induced COX-2 expression.

(A and B), Cl 41 cells were repeatedly treated with 1 μM of arsenite twice a week for up to 2 months or with 0.5–2 μM arsenite for 24 hrs (A) or with 10–20 μM arsenite for 12 hrs, The cells were extracted and subjected to Western Blotting. (C and D), p27-deficient cells were identified using RT-PCR (C) and Western Blotting (D). β-Actin was used as a loading control. (E), p27+/+ and p27−/− MEFs were exposed to arsenite at the indicated doses for 9 hrs. COX-2 protein expression was determined by Western Blotting. (F), Twenty-four hrs after p27−/− MEFs were infected with adenovirus carrying the full length GFP-tagged p27 cDNA or GFP vector control, the cells were treated with arsenite for 12 hrs. COX-2 and p27 protein expression was determined by Western Blotting. (G), p27+/+ and p27−/− MEFs were exposed to 20 μM arsenite for indicated times. The exposed cells were lysed with 50 μL lysis buffer, and the proliferation of the cells was measured using CellTiter-Glo Luminescent Cell Viability Assay kit with a luminometer as described in Methods section. The results are expressed as relative to medium control. (H and I), knockdown of p27 expression in Cl41 cells stably transfected with p27 shRNA was identified by western blot (F) and the stable transfectants were exposed to arsenite for 12 hrs at doses as indicated. The COX-2 protein induction was determined by Western Blotting. The experiments were performed for at least three times.

3.2. p27 suppressed COX-2 expression at transcription level

To elucidate the molecular mechanisms underlying p27 inhibition of COX-2 induction by arsenite, we first determined the COX-2 mRNA levels in both p27+/+ and p27−/− MEFs upon arsenite treatment. Though basal level of COX-2 mRNA is higher in p27 +/+ MEFs, arsenite exposure only slightly upregulated its expression 9 hours after exposure. However in p27−/− MEFs arsenite treatment significantly increased COX-2 mRNA expression by folds at both 6 hours and 9 hours of exposure in p27−/− MEFs under the same experimental conditions, which is consistent with the COX-2 protein expression (Figs. 2A and 2B). Since above results showed that p27 inhibits cox-2 mRNA levels upon arsenite exposure, we evaluated the cox-2 mRNA stabilities in both p27+/+ and p27−/− cells. As shown in Fig. 2C, treatment of cells with actinomycin D (20 μg/mL) resulted in a cox-2 mRNA degradation in p27+/+ cells, and this degradation was markedly increased in p27−/− cells (Fig. 2C). The half life of cox-2 mRNA (T_1/2) in p27+/+ cells was 2.61 h, whereas it was reduced to 1.44 h in p27−/− cells (Fig. 2D). Those results suggest that RNA stability does not contribute to p27 inhibition of cox-2 mRNA induction upon arsenite exposure. We then tested cox-2 promoter activity in both p27+/+ and p27−/− MEFs through a luciferase reporter assay. The results showed that in p27−/− MEFs, COX-2 is induced more than 5 folds while in p27+/+ cells there was no induction observed, indicating that the reason for the difference in COX-2 levels in these cells is discrimination in transcription (Fig. 2E). The promoter region of this reporter contains the DNA sequence of NFκB, NF-IL6, and CRE binding site. This is consistent with the fact that NFκB, CREB, and AP-1 are major transcription factors involved in regulation of COX-2 transcription [25–28]. This is indicative of the potential involvement of such transcription factors in COX-2 induction via this process.

Figure 2. p27 inhibited COX-2 transcription in response to arsenite.

(A), p27+/+ and p27−/− MEFs were treated with 20 μM of arsenite for 3, 6, and 12 hrs. RT-PCR was carried out to determine the mRNA levels of cox-2 expression. The PCR products were separated over 2% agarose gels, stained with ethidium bromide. (B), The densitometric analyses of the product bands were conducted using the software of ImageQuant 5.2 (GE Healthcare). (C), After pretreatment with 10% FBS for 9 hrs, p27+/+ and p27−/− cells were treated with actinomycin D of 20 μg/mL for time periods indicated. RT-PCR was carried out to determine the mRNA levels of cox-2 expression. The PCR products were separated over 2% agarose gels, stained with ethidium bromide. (D), The relative RNA level of cox-2 was determined by ImageQuant 5.2 (GE Healthcare). Natural logarithm of ratio [cox-2]_t/[cox-2]₀ was plotted against the time and the half life of cox-2 mRNA was calculated via linear regression. (E), p27+/+ and p27−/− cells stably transfected with COX-2 luciferase reporter were treated with arsenite for 6 hrs. The luciferase activities were then evaluated, and the results are presented as luciferase activity relative to the medium control. Each bar indicated the mean ± S.D. of triplicate assay wells. “*” indicated significant difference (P < 0.05). The experiments were performed at least three times.

3.3. p27 suppressed p38 phosphorylation by arsenite

To determine signaling involved in the upregulation of COX-2 induction in p27−/− MEF due to arsenite exposure, we first examined the potential effects of the constitutive expression of p27 on MAPK phosphorylation by infection of p27−/− cells with either GFP adenovirus or GFP-p27 adenovirus. As shown in Figs. 3A and 3B, arsenite treatment resulted in activation of p38, JNK and ERK to certain extents in p27−/− cells infected with control adenoviral vectors in various time points, whereas the activation was dramatically impaired in p27−/− cells infected with adeno-p27 virus. Consistently, increased COX-2 induction in p27−/−(vector) cells was also dramatically reduced in p27−/− cells infected with adeno-p27 virus (Figs. 3A and 3B). As p38 upstream kinases, MKK3/6 phosphorylation was also upregulated upon arsenite exposure in p27−/−(vector) cells, whereas this upregulation was impaired in the p27−/− cells infected with adeno-p27 virus (Figs. 3A and 3B), suggesting that p27 might downregulate p38 activation by targeting MKK3/6 or their upsream kinases. Further elucidation of p27 target for mediating MKK3/6-p38 activation is currently undergoing in our laboratory. To determine whether inhibition of any of these MAPKs is associated with p27-regulated COX-2 expression, we first employed MAPK chemical inhibitors. The results suggested that inhibition of p38 phosphorylation by its inhibitor SB202190 is positively associated with downregulation of arsenite-induced COX-2 expression in p27−/− cells, whereas either inhibition of JNKs by its inhibitor SP600125 or inhibition of ERK phosphorylation by its inhibitor PD98059 did not show any observable correlation with COX-2 expression (Fig. 3C). These results suggested that p27 might suppress p38 in cells. This notion was further supported by data showing that p38 phosphorylation by arsenite was remarkably increased in p27−/− cells and Cl41 p27 knockdown cells compared to their corresponding parental cells (Figs. 3D, E). Collectively, our results indicated that p27 expression suppressed p38 phosphorylation upon arsenite exposure in cells. However, chemical inhibitor merely provided a clue showing potentially responsibility of kinases. Also, p38 has different isoforms, some of which are not significantly suppressed by inhibitor we applied above. Therefore, to further prove that p38 is actually involved in COX-2 activation, we used the dominant negative p38 to interfere with the endogenous p38 function, which would be discussed in the next part of the results.

Figure 3. p27-mediated downregulation of p38 was associated with its inhibition of COX-2 expression.

(A, B), Twenty-four hrs after p27−/− cells were infected with adenovirus carrying full length GFP-tagged p27 cDNA or GFP vector control, the cells were treated with arsenite in different doses for different time periods, as indicated. Phosphorylation of MAPKs and expression of COX-2 were assessed by Western Blotting. (C), p27−/− cells were pretreated with indicated MAPK inhibitors for 30 min and were then exposed to arsenite for 12 hrs. The cell extracts were subjected to Western Blotting for MAPK phosphorylation and COX-2 expression using specific antibodies. (D), p27+/+ and p27−/− MEFs were exposed to arsenite at different doses and different time periods, as indicated. The phosphorylated p38 and total p38 protein levels were assessed by Western Blotting. (E), Cl41 cells stably transfected with p27 shRNA or non-silencing control shRNA were treated with arsenite with different doses for 6 hrs. p38 phosphorylation was determined by Western Blotting. The experiments were performed three times.

3.4. p38β and p38δ, but not p38α, was crucial for arsenite-induced CREB phosphorylation and COX-2 expression

Since the above results suggested that p27 inhibited arsenite-induced COX-2 expression at the transcriptional level, we determined which transcription factor is involved in p27 downregulation of COX-2 expression. Consistent with the p38 inhibitor on COX-2 expression, SB202190 treatment of p27−/− cells showed a specific inhibition of arsenite-induced CREB phosphorylation without affecting CREB protein expression, however it did not show an observable effect on c-Jun phosphorylation or Fra-1 expression, as well as ATF-2, which is known as a direct downstream transcription factor of p38 in many other studies [29] (Fig. 4A). Moreover, treatment of p27−/− cells with SP600125 or PD98059 did not show an inhibition on CREB phosphorylation, whereas these inhibitors inhibited c-Jun phosphorylation and Fra-1 expression (Fig. 4A). To further test the role of p38 in p27 regulation of CREB phosphorylation and COX-2 expression, dominant negative mutant p38 was co-transfected with COX-2-luciferase reporter into p27−/− MEFs to establish a stable transfectant, p27−/− (DN-p38). As shown in Figs. 4B and 4C, specific inhibition of p38 by ectopic expression of DN-p38 in p27−/− MEFs impaired CREB phosphorylation and COX-2 expression upon arsenite treatment compared with those in p27−/−(vector) transfectant. Consistently, elevation of cox-2 transcription due to arsenite exposure in cox-2 promoter-driven luciferase reporter assay was also attenuated in DN-p38 transfectant (Fig. 4C). Previous studies have shown that p38 isoforms have differential biological effects [30]. Thus, we compared arsenite-induced COX-2 expression between cells with a deficiency of p38α (p38α−/−), p38β (p38β−/−) and p38δ (p38δ−/−) (Fig. 4D). The results showed that knockout of p38α did not show observable inhibition of COX-2 expression (Fig. 4E), while deletion of either p38β or p38δ effectively attenuated COX-2 induction by arsenite (Figs. 4F and 4G), demonstrating that both p38β and p38δ, but not p38α, were crucial for COX-2 induction by arsenite. Though p38δ was usually not considered as a target of SB202190, its involvement in arsenite induced COX-2 expression was still confirmed here. Consistently, the deficiency of p38β or p38δ in p38β−/− or p38δ−/− cells also impaired CREB phosphorylation (Figs. 4H and 4I), demonstrating that both p38β and p38δ are crucial for arsenite-induced CREB phosphorylation. Taken together, our results indicated that p27 inhibited arsenite-induced COX-2 expression via attenuating activation of p38β/p38δ-CREB axis.

Figure 4. p38β and p38δ, but not p38α, was crucial for arsenite-induced CREB phosphorylation and COX-2 expression.

(A), p27−/− cells were pretreated with indicated MAPK inhibitors for 30 min and then exposed to arsenite for 12 hrs. The cell extracts were subjected to Western Blotting to determine the various transcription factor protein phosphorylation and expressions. (B), Dominant negative mutant of p38 or vector control were stably transfected into p27−/− cells. After the cells were treated with arsenite (20 μM) for 6 hrs or 12 hrs, the cell extracts were subjected to Western Blotting using specific antibodies, as indicated. (C) p27−/− MEFs stably co-transfected with COX-2 luciferase reporter and DN-p38 or vector control were exposed to arsenite for 6 hrs. The luciferase activities were then evaluated. The results are presented as luciferase activity relative to the medium control. Each bar indicates the mean±S.D. of triplicate assay wells. “*” indicated significant difference (P<0.05). (D), p38α−/−, p38β−/− and p38δ−/− cells were identified by genome PCR assay. (E-I), p38α−/−, p38β−/− and p38δ−/− cells in addition to their control wild type cells were exposed to arsenite at different doses for 12 hrs. Cell extracts were subjected to Western Blotting for determination of COX-2 and CREB expression, as well as CREB phosphorylation. The experiments were performed at least three times.

3.5. p38-regulated CREB activation was essential for increased COX-2 transcription in p27−/− cells

Our above results strongly indicate that p27 suppresses arsenite-induced COX-2 transcription and CREB activation, both of which are regulated by p38. Due to the fact that NFκB has been reported to play an important role in COX-2 induction by nickel and Cr (VI) in BEAS 2B cells, we determined whether NFκB was also important in p27-regulated COX-2 expression following arsenite exposure [31, 32]. As shown in Fig. 5A, the distribution of p50 and p65 in nuclear and cytoplasm did not show observable difference between p27+/+ and p27−/− cells upon arsenite exposure, indicating that NFκB was not involved in p27 regulation of COX-2 expression. To evaluate the potential role of CREB activation in upregulating COX-2 in p27−/− cells, we compared CREB activation in p27+/+ and p27−/− MEFs, and found that CREB phosphorylation was also elevated in p27−/− cells in comparison to that in p27+/+ MEFs (Fig. 5B). To further evaluate whether CREB is involved in COX-2 expression from arsenite exposure in p27−/− cells, we applied shRNA to knockdown CREB to test whether COX-2 induction would be inhibited. As shown in Fig. 5C, CREB expression and phosphorylation were dramatically knocked down by transfecting shRNA CREB into p27−/− cells. Knockdown of CREB expression led to a significant inhibition of COX-2 expression at both the transcriptional level and protein level as a result of arsenite exposure in p27−/− cells (Figs. 5D and 5E). Our results strongly suggested that p27 suppressed COX-2 induction specifically through targeting the p38β/p38δ-CREB pathway. Together with our previous result, this finding has extended understanding underlying of p27 in suppressing arsenite- induced MAPK and downstream pathways and gene expression as well as (Fig. 5F).

Figure 5. Upregulation of CREB phosphorylation was crucial for evaluation of COX-2 expression in p27−/− cells.

(A), p27 +/+ and p27−/− cells were treated with arsenite for 9 hrs and the cells were then extracted for isolation of protein fractionation using nuclear/cytosol fractionation kit. The extracted proteins from nuclears or cytosols were subjected to Western Blotting. Lamin B protein was used as nuclear protein marker to verify the quality of fractionation. (B) p27+/+ and p27−/− cells were treated with 20 μM arsenite for 6 hrs. The cell extracts were subjected to Western Blotting to determine CREB phosphorylation and expression. (C), Efficiency of CREB specific short hairpin RNA transfection was identified by comparing CREB protein levels in p27−/− (shCREB) and p27−/− (si-vector) cells using Western Blotting. (D), Western Blotting was employed to test COX-2 expression in p27−/− cells stably transfected with shCREB or its nonsense vector control, 9 hrs after treatment of 20 μM arsenite. (E), p27−/− cells stably co-transfected with COX-2 luciferase reporter together with shCREB, or nonsense vector control, were treated with arsenite. The luciferase activities were then evaluated and the results are presented as luciferase activity relative to the medium control. Each bar indicates the mean ± S.D. of triplicate assay wells. “*” indicated significant difference (P<0.05). The experiments were performed at least three times. (F), Anticipate model of p27 regulation due to arsenite exposure.

4. Discussion

p27 was first identified as a CDK inhibitor. It binds to cyclin E-Cdk2 complexes, suspending the cell cycle at the G1 phase [14]. By using affinity chromatography of cyclin E-Cdk2, this 27-kD protein was purified from cycling-arrest cells for the first time and is referred to as p27 [15]. Early research on p27 knockout mouse revealed that absence of the p27 gene, which causes excessive cell proliferation and physical dysfunction such as multiple organ hyperplasias, increased body size and several kinds of tumor formation [33, 34]. These facts further support the conclusion that p27 regulates cell proliferation as a CDK inhibitor. Later, p27 was proven to work on different CDKs in the cell nucleus and affect the entire cell cycle [35–37]. As nucleus localization of p27 is linked to its CDK-dependent effect, its relocalization can thereby release inhibitory effects on CDKs and promote cancer development [38]. However, other studies of the relationship between p27 and cancer exhibit evidence showing that relocalized p27 may also affect cancer development through a CDK-independent pathway. Studies of the CDK-independent role of p27 indicate that cytoplasma localized p27 are related to metastatic melanoma and other metastatic diseases [39]. This effect was later inferred to be accomplished via the interaction between p27 and Rho A via binding [40]. Though the above studies have revealed part of the cyclin/CDK regulatory independent cancer effect of p27, the CDK-independent effect of tumor suppression in p27 is still largely unexplored. Our previous work has demonstrated that with arsenite exposure, p27 can exhibit anticancer effects by suppressing JNK/c-Jun and HSF-1 pathway activation, thus reducing hsp27 and hsp70 expression [7]. In this study, we discovered that p27 exhibited an inhibitory effect on COX-2 induction at the transcriptional level upon arsenite exposure, specifically through inhibition of p38β and p38δ-mediated CREB-dependent pathway, based on the results obtained from utilizing p27−/− and p38 deficient MEFs as well as DN-p38 transfectants. These novel findings provide new mechanistic information for understanding of p27 tumor suppressor functions through CDK-independent mechanisms.

Cyclooxygnenase (COX) is the enzyme responsible for biogenesis of prostanoids, a signaling molecule with multiple functions in various biological processes, including inflammation and cancer development [41]. Genetic study of COX-2 knockout mice has revealed a strong correlation between COX-2 expression and cancer development, indicating that COX-2 contributes to carcinogenesis and oncogenesis [42, 43]. Our previous findings also suggest several cancer-related biological effects of COX-2, including anti-apoptotic effect, cell transformation, and cell cycle progression in different context [44]. Regarding cell transformation and cell cycle progression as key steps of carcinogenesis, and apoptotic as a cancer suppression process, we anticipate that COX-2 is a factor that is positively associated with cancer development. It has been shown that COX-2 expression is also regulated by various oncogenes [45]. Thus in our current study, it is reasonable to expect COX-2 induction by arsenite may partly contribute to the carcinogenic effect of arsenite. This hypothesis is supported by amounts of inflammation studies. It has been reported that cadmium exposure induces HO-1 and COX-2 expression with elevation of overall PGE-2 generation, indicating that COX-2 enzyme activity is increased in presence of HO-1 following cadmium exposure [46]. It has also been reported that arsenite exposure induces COX-2 expression with significantly increasing PGE2 generation and then leads to inflammatory effect in rat liver epithelial cells [47]. Previous studies have demonstrated that p27 knockout mice have more florid inflammation response, demonstrating the anti-inflammatory function of p27 [48]. Thus, we anticipate that p27-regulated COX-2 expression exhibits its inflammatory response following arsenite exposure although HO-1 may exhibit some suppressive effect on COX-2 enzyme activity and subsequently resulting in some counterbalance of its inflammatory response. This information provides a significantly insight into understanding anti-inflammatory effects of p27 that extend its anti-oncogene functions. In addition, elucidation of anti-inflammatory effect of p27 via inhibiting COX-2 expression extends our understanding of chronic inflammation-associated carcinogenic effects of arsenite exposure. Since many metals, such as arsenite, nickel(II), and chromium(IV), share a similar biological effect on COX-2 induction [31, 32], we will extend current studies to test whether p27 has a similar effect on COX-2 induction following exposure of other metals, such as nickel(II), and chromium(IV) in our future investigation.

While the constitutive isoform of COX-1 remains in most tissues, the inducible isoform COX-2 is only detectable in certain conditions, such as tissue and cells with mitogenic stimuli in inflammation or in various types of cancer tissues [49, 50]. Activation on a transcriptional level is an example of one mechanism responsible for COX-2 overexpression in cancer tissues [51]. Transcription factors, such as NFAT, NFκB, and AP-1, have been verified to regulate COX-2 transcription in various experimental systems [52]. Several selective COX-2 inhibitors have been developed for pharmacologic applications, some of which are considered to have promising potential for cancer treatment [53, 54]. Among these inhibitors, some work at the transcriptional regulation of COX-2 expression [55]. In the present study, we found that the transcription factor CREB is required for COX-2 elevation in p27−/− cells, and that p38β and p38δ, as well as MKK3/6, act as major kinases responsible for the increased phosphorylation of CREB, whereas other transcription factors, such as AP-1, does not play a significant role in mediation of the increased COX-2 expression in p27−/− cells even though it is also activated following arsenite exposure. This information may provide the direction for future drug development targeting COX-2.

p38 is an MAPK family member and was first isolated as tyrosine phosphorylated in response to LPS [56]. Four isoforms of p38 have been identified so far: α, β, γ, and δ, each with a unique signaling pathway [57, 58]. Activation of p38 is related to phosphorylation status. Among its upstream kinases, MAP kinase kinase 6 can activate all p38 isoforms via phosphorylation, while MKK3 can only activate its α, γ, and δ isoforms [21]. Dephosphorylation of p38 involves phosphatases that include Protein Phosphatase 1 and Protein Phosphatase 2A [59, 60]. p38 can phosphorylate transcription factors such as ATF-2, and then activate downstream target gene expression [61]. Our previous studies also indicated the activation of types of MAPKs and thus their downstream target upon arsenite exposure [7, 62]. Porter et al. reported that Rac, Rho, MEKK3, and MEKK4 are required for JNK activation upon arsenite treatment in kidney 293 cells [63]. Moreover, reactive oxygen species (ROS) generation as an important consequence of arsenite treatment during cellular metabolisms has been studied in published studies [64, 65]. It has been reported that arsenite-induced generation of ROS mediates MAPK activation [6, 66]. Our unpublished studies also showed that p27 knockout increases MnSOD expression and hydrogen peroxide generation (Zheng, X, and Huang, C, unpublished data). Thus, potential contribution of MnSOD expression and hydrogen peroxide generation in p27 regulation of COX-2 axis is under investigation in our group. In the current studies, we have identified that p38β and p38δ, but not p38α, is regulated by p27 and have confirmed that p38β and p38δ are crucial for mediating CREB phosphorylation and activation upon arsenite exposure. Our results reveal the differential effects of p38 isoforms in relation to their various downstream transcription factors, such as CREB, as well as their specific cellular biological functions.

The results demonstrated in current studies figured out the signaling pathway leading to increased COX-2 induction by arsenite exposure in p27−/− cells. By evaluating the role of p27 in regulation of COX-2 induction upon arsenite response, we demonstrated the novel function of p27 modulation of p38 activation. p27 exhibits the inhibition of CREB phosphorylation and activation, and further regulates COX-2 transcription. Previous studies have reported much about the correlation between p27 and MAPK, including p38, but p27 is mostly considered merely as a CDK inhibitor that is regulated by MAPK [67, 68]. However, our study shows that p27 can reversely regulate p38 phosphorylation, revealing the crosstalk between p38 and p27 in controlling the complex anti-cancer mechanism of p27. The CDK-independent function of p27 has been studied frequently but studies are seldom done on inhibition of tumors via the MAPK pathways. Therefore, to fully elucidate how p27 affects phosphorylation of p38 isoforms will be highly significant for complete understanding this novel function of p27 in regulating COX-2 expression, as well as mediating inflammation and chronic inflammation-associated carcinogenic effects.

Highlights.

• p27 suppressed COX-2 induction by arsenite at transcription level

• p27 suppressed p38 activation by arsenite

• p38β and p38δ, but not p38α, was crucial for arsenite-induced CREB phosphorylation and COX-2 expression

• p38-regulated CREB activation was essential for increased COX-2 transcription in p27−/− cells

Abbreviations

CDK

cyclin-dependent kinase

COX-2

cyclooxygenase-2

MAPK

Mitogen-activated protein kinase

NFκB

nuclear factor κB

Hsp

heat shock protein

NFAT

nuclear factor of activated T-cell

JNK

c-Jun N-terminal kinase

MEF

mouse embryo fibroblast cells

DMED

Dulbecco’s modified Eagle’s medium

FBS

fetal bovine serum

MEM

Minimum Essential Medium Eagle

MEK

MAPK/ERK kinase

SP-1

specificity protein-1

DN

Dominant negative

AP-1

activator protein-1

CREB

cAMP responding element-binding protein