Thrombomodulin Is a Clock-controlled Gene in Vascular Endothelial Cells*

Thromboembolic events such as pulmonary embolism (PE),³ cerebral infarction, and acute myocardial infarction (AMI) are major causes of death in developed countries. The onset of PE, cerebral infarction, and AMI shows apparent circadian variation and frequently occurs in the morning (1–3). Thus, elucidation of the mechanisms of circadian variation of these disorders may lead to the development of methods that prevent the onset of thromboembolic events.

The circadian variation of the onset of thromboembolic events is thought to be related to the biological clock. Recently, the molecular mechanisms of the biological clock have been elucidated (4). This system is composed of several clock genes, including clock, period, bmal, and cry. The heterodimer of CLOCK and BMAL binds to the E-box sites upstream of the period and cry genes and transactivates these genes. The PERIOD and CRY proteins inhibit their own transactivation by CLOCK and BMAL, resulting in the formation of a negative feedback loop. This negative feedback loop persists for ∼24 h, which corresponds to circadian rhythm (5). The biological clock regulates the expression of downstream target genes, clock-controlled genes, and modulates the cellular responses to these downstream genes. The central biological clock is located in the suprachiasmatic nucleus in the hypothalamus (6). Furthermore, we and other researchers have reported that clock genes are also expressed in peripheral organs and cells and that their expression shows circadian oscillation, suggesting the existence of a peripheral clock (7–9). Approximately 8–10% of the total number of genes expressed in mouse heart and liver were found to show a circadian expression pattern (10). The peripheral clock is thought to regulate the expression of organ-specific, clock-controlled genes directly or indirectly through the expression of core clock genes such as clock and bmal (10). The circadian expression of clock genes in the cardiovascular system has been demonstrated as well (11–16). However, the biological significance of this expression remains to be elucidated.

Previously, we identified a clock gene in vascular endothelial cells, CLIF, which is also termed BMAL2 (12). Subsequently, we found that CLOCK and BMAL2 in peripheral tissue may directly regulate the circadian expression of the plasminogen activator inhibitor-1 (PAI-1) gene, contributing to the morning onset of AMI (17–20). Thus, in this study we screened for clock-controlled genes in vascular endothelial cells using microarray technology. A total of 229 genes were up-regulated by CLOCK and BMAL2. Among them, we focused on the endothelial membrane protein, thrombomodulin (TM), since TM plays an important role in the regulation of blood coagulation processes by exerting anti-coagulant effects through the activation of protein C (21–23). Our results demonstrate that TM mRNA and protein expression have a circadian pattern and that a heterodimer of CLOCK and BMAL2 binds to the E-box upstream of the TM promoter and transactivates promoter activity. In addition, these results suggest that the circadian rhythm of TM gene expression is under the control of circadian clock molecules.

EXPERIMENTAL PROCEDURES

Animals—This study was approved by the Animal Committee of University of Tokyo. Clock mutant mice in BALB/c and C57BL/6J backgrounds were supplied by J. S. Takahashi (Northwestern University, Evanston, IL) (24). Clock mutant mice carry an internal deletion of 51 amino acids in the C-terminal activation domain of the CLOCK protein, which results in a dominant negative mutation. Thus, Clock homozygotes fail to express persistent circadian rhythms when maintained in constant darkness. A breeding colony was established by further backcrossing with Jcl:ICR mice as described previously (25). Male BALB/c mice of 8–12 weeks of age were maintained under a 12-h:12-h light-dark cycle. Zeitgeber time (ZT) 0 refers to light on, and ZT 12 refers to light off. ZT 0–12 is the subjective light phase, and 12–24 h ZT is the subjective dark phase. Under the restrictive feeding condition, mice were allowed access to food for 4 h from ZT5 to ZT9 for 10 consecutive days (days 7–16) (26).

Construction of Plasmids and Recombinant Adenoviruses—The TM reporter, pGL3TM1562 (–1562 to +140), was constructed as described previously (27, 28). The fragment from –1370 to 958 was removed by SmaI, yielding pGL3TM1562 (deletion 1370–958). Successive deletion of the 5′-flanking region was performed using restriction enzymes KpnI and Blp1, which yielded pGL3TM1379 (–1379 to +140), or KpnI and SmaI, which yielded pGL3TM958 (–958 to +140). pGL3TM1562mut is identical to pGL3TM1562, except that the E-box site (bp –1473 to –1468) was mutated from CACGTG to CTCGAG. The mammalian expression vectors for BMAL1 (phBMAL1), CLIF/BMAL2 (phBMAL2), and CLOCK (phCLOCK) were constructed as described previously (12). Recombinant adenoviruses were prepared as described previously (29), and those expressing green fluorescent protein (GFP), CLOCK, or BMAL2 were designated as AdCMV.GFP, AdCMV.CLOCK, or AdCMV.BMAL2, respectively.

Cell Cultures—Human umbilical vein endothelial cells (HUVECs) were obtained from BioWhittaker (Walkersville, MD). A hemangioendothelioma cell line was obtained from ATCC (ATCC CRL-2586, Manassas, VA). Bovine aortic endothelial cells (BAECs) were obtained as primary cultures. Hemangioendothelioma cell lines and BAECs were cultured in Dulbecco's modified eagle's medium (Sigma).

Northern Blot Analysis—Total RNA was isolated from HUVECs using the RNeasy method (Qiagen KK, Tokyo, Japan) or from frozen tissues using ISOGEN (Nippon Gene Co., Tokyo, Japan). Northern blot analysis was performed as described previously (29).

Western Blot Analysis—Whole tissue protein was extracted using extraction buffer (50 mm Tris, pH 8.0, 150 mm NaCl, 0.5% sodium deoxycholate, 1% Nonidet P-40, 0.1% SDS) under nonreducing conditions and subjected to SDS-PAGE. The primary antibody used was rabbit anti-rat thrombomodulin polyclonal antibody (1:200, American Diagnostica Inc, Stamford, CT). Rabbit anti-β-actin polyclonal antibody (1:200, Santa Cruz) was used to correct for differences in protein loading.

Transient Transfection Assay—Transient transfection was performed using Lipofectamine reagent (Invitrogen) as described previously (12). Data for each construct are presented as the means ± S.E.

Microarray Analysis—Microarray analysis was performed using ∼12,000 human cDNAs (human 1 cDNA Microarray, G4100A, Agilent Technologies, Palo Alto, CA) as described previously (30, 31). HUVECs were infected with AdCMV. CLOCK and AdCMV.BMAL2 or AdCMV.GFP. Forty-eight hours after infection total RNA was isolated and applied to microarray analysis. Fluorescence labeling and hybridization were performed according to the manufacturer's instruction. The intensity of each signal was quantified by sequential excitation of the two fluorophores with a scanning laser read at an appropriate wavelength for each emission (GMS417 Array scanner, Affimetrix Santa Clara, CA). Expression data were analyzed using Imagene ver4.1 (Biodiscovery, El Segundo, CA). Differential expression values were demonstrated as the log ratios of the dye-normalized green and red channel signals. This hybridization experiment was performed in duplicate with dye swap design method. A threshold of a 2.5-fold change was used for comparison.

In Vitro Transcription and Translation/Gel Mobility Shift Assay—In vitro transcription and translation and gel mobility shift assays were carried out as described previously (32). The sequence of the double-stranded oligonucleotide containing the E-box site (5′-GGTGCGTGTCTGTCGCACGTGGCAGACGCCATACTC-3′) was derived from the sequence of the TM promoter corresponding to positions –1488 to –1453. For supershift analysis, 1 μg of anti-Myc antibody (Invitrogen) was used for the binding reaction.

Chromatin Immunoprecipitation (ChIP)—ChIP assays were carried out as described previously after administrating adenovirus expressing CLOCK and Myc-tagged BMAL2 (10 multiplicity of infection) into HUVECs (33). To precipitate Myc-tagged BMAL2, anti-c-Myc-agarose (Santa Cruz, sc-40, 9E10) was used. PCR of the TM promoter around the E-box (–1525∼–1364) was performed using immunoprecipitated chromatin with the following pair of oligonucleotide primers: 5′-TGCAGGTCAGTCCAGTCCAGCCCGGCCCAC-3′ and 5′-GCACAGAGTCCTCCTTCTGGGGTTGGAAGC-3′. For the negative control, PCR of the 3′-flanking region of the TM gene (+4062∼+4298) was performed with the following pair of oligonucleotide primers: 5′-AAGAAGTGTCTGGGCTGGGACGGACAGGAG-3′ and 5′-AGCAGTCGTGCTCGACGCACTGGCTGCCAC-3′. A 1:1000 dilution of input DNA served as a positive control for PCR amplification.

Quantitative Real-time Reverse Transcription-PCR—Total RNA was used for quantitative real-time reverse transcription-PCR as described previously (16). Primers and probes were obtained and used according to the manufacturer's protocol (Applied Biosystems, Foster City, CA). The level of 18 S rRNA was quantitatively measured in each sample to control for sample-to-sample differences in RNA concentration.

Statistical Analysis—Results are expressed as mean ± S.E. Two-tailed Student's t test for unpaired samples or one-way analysis of variance was used for comparison of parameters. p < 0.05 was considered to indicate statistical significance.

RESULTS

Circadian Oscillation of the Expression of Clock Genes Was Induced in Cultured Vascular Endothelial Cells—First, we confirmed the existence of an intrinsic clock system in endothelial cells. Highly concentrated serum stimulates the circadian expression of clock genes in cultured fibroblasts (7). Thus, we added a 50% concentration of serum to cultures of hemangioendothelioma cells and analyzed the expression of clock genes. Serum stimulation induced the circadian expression of clock genes, suggesting that vascular endothelial cells have an intrinsic clock system (Fig. 1). Although similar results were obtained with HUVECs (supplemental Fig. S1), the amplitude of the oscillation was smaller. This reduced amplitude may be due to the high concentration of serum and several growth factors required for maintenance of HUVECs.

Downstream Clock-controlled Genes in Vascular Endothelial Cells—To identify the downstream target genes of the peripheral clock in endothelial cells, we screened for genes whose expression was up-regulated by CLOCK and BMAL2 using cDNA microarray techniques. We infected HUVECs with AdCMV.Clock and AdCMV.BMAL2 and isolated RNA after 48 h. As a reference, we used RNA from HUVECs that had been infected with AdCMV.GFP. Differential hybridization was performed on microarray slides containing ∼12,000 spots of human cDNA using labeled RNA from HUVECs. The expression of a total of 229 genes was up-regulated including Period2, Rev-erbα, wee-1, and Pai-1, which have been identified previously as clock genes or clock-controlled genes. The 229 identified genes included 14 genes that encode secreted proteins, 20 genes that encode membrane proteins, and 20 genes that encode transcription factors. These 229 genes are listed in the supplemental table. Among these genes we observed that TM expression was up-regulated by CLOCK and BMAL2. TM is expressed primarily in vascular endothelial cells and plays an important role in regulating the blood coagulation process (23). We further examined the molecular mechanisms by which TM is up-regulated by Clock and BMAL2. The analysis of other genes will be described elsewhere.⁴

Thrombomodulin mRNA and Protein Show Circadian Expression That Is Regulated by CLOCK and BMAL2—To confirm that TM expression is under the direct control of CLOCK and BMAL2, we performed Northern blot analysis using RNA from infected HUVECs and demonstrated that the overexpression of CLOCK and BMAL2 resulted in an increase in TM mRNA levels (Fig. 2, A and B). This induction was determined to be specific because infection by the same titer of control adenovirus, AdCMV.GFP, did not result in an increase the TM mRNA levels. These results indicate that TM mRNA is up-regulated by CLOCK and BMAL2 in vascular endothelial cells. To examine whether TM mRNA expression shows circadian oscillation, we performed quantitative real-time reverse transcription-PCR using total RNA extracted from the mouse heart and lung. TM mRNA in the mouse lung and heart exhibited apparent circadian expression with the highest expression levels occurring at 18 ZT and the lowest expression levels occurring at 6 ZT (Fig. 2, C and D). TM protein expression in the lung also showed circadian oscillation (Fig. 2E). We did not detect TM protein expression in the heart by Western blotting, which may be due to the reduced abundance of TM mRNA in the heart compared with that of the lung.

Co-expression of CLOCK and BMAL1 or -2 Transactivates the Thrombomodulin Promoter—To further elucidate the mechanism by which CLOCK and BMAL2 increase the TM mRNA level, we used a reporter plasmid, pGL3TM1562, that contains a 1.6-kilobase fragment of the human TM promoter. This reporter plasmid construct was co-transfected into BAECs together with the human CLOCK expression plasmid, phCLOCK, and the human BMAL1 or BMAL2 expression plasmid, phBMAL1 or phBMAL2. CLOCK, BMAL1, or BMAL2 alone did not induce TM promoter activation. However, the co-expression of CLOCK and BMAL1 or of CLOCK and BMAL2 increased TM promoter activity by about 3-fold (Fig. 3A). BMAL1 and BMAL2 showed similar activity with regard to TM promoter activation. We found two CACGTG-type E-box elements (Distal E-box at –1468 and Proximal E-box at –1009 from the transcription start site) within the TM promoter sequences, which act as putative binding sites of CLOCK and BMAL1 or BMAL2. Deletion of bp –1562 to –1379 in the TM promoter markedly diminished the induction of TM promoter activity by CLOCK and BMAL2, suggesting that the Distal E-box at –1468, but not the Proximal E-box at –1009, is important for transactivation of the TM promoter (Fig. 3B). To determine whether CLOCK and BMAL2 transactivate the TM promoter by binding to Distal E-box site, we constructed a mutant TM promoter with 2-bp mutations at –1472 and –1469 (CACGTG to CTCGAG). A reporter plasmid harboring this mutated site (pGL3TM1562mut) was transfected into BAECs. Mutation of the Distal E-box site abolished the transactivation of the TM promoter by CLOCK and BMAL2 (Fig. 3C).

Heterodimer of CLOCK and BMAL2 Binds to the E-box of the TM Promoter—To examine whether the heterodimer of CLOCK and BMAL2 binds directly to Distal E-box, we performed gel mobility shift assays using in vitro translated CLOCK and BMAL2 proteins and a labeled probe containing the E-box element. In the presence of both the CLOCK and BMAL2 proteins, DNA binding activity was detected (Fig. 4A). Binding was attenuated by the addition of an unlabeled specific competitor containing endogenous E-box but not by an unlabeled competitor containing mutated E-box with 2-bp mutations. The in vitro translated BMAL2 protein was Myc-tagged. We performed supershift analysis using anti-Myc antibody. The addition of 1 μg of anti-Myc antibody in this reaction mixture interfered with binding of the heterodimer (Fig. 4A). To further examine the association of CLOCK and BMAL2 with the region around the E-box of the TM gene (–1525 ∼–1364) in vivo, we performed ChIP assays using HUVECs expressing CLOCK and Myc-tagged BMAL2 or β-galactosidase as a control (Fig. 4B). ChIP was performed using anti-Myc antibody and analyzed by PCR amplification of total DNA before immunoprecipitation, after nonspecific immunoprecipitation of DNA, or after specifically immunoprecipitating DNA. For the negative control, the 3′-flanking region of the TM gene (+4062∼+4298) that contains no E-box (-like) element was amplified. The region around the E-box of the TM gene was amplified in DNA recovered from a ChIP reaction with anti-Myc antibody performed on chromatin lysates from cells expressing CLOCK and Myc-tagged BMAL2 protein but not from control lysates. These data suggest that a heterodimer of CLOCK and BMAL2 associates with the region around the E-box of the TM gene in vivo.

Oscillation of Thrombomodulin mRNA Expression Is Attenuated in Clock Mutant Mice—To examine whether TM expression is truly regulated by CLOCK in vivo, we analyzed the level of TM expression every 6 h during a 24-h period in wild-type mice and clock mutant mice, which have an impaired clock system. In wild-type mice, the TM mRNA showed clear cyclic expression. However, in clock mutant mice, the oscillation of TM mRNA expression was attenuated, suggesting that TM gene expression is regulated by clock in vivo (Fig. 5).

Restrictive Feeding Affects the Circadian Oscillation of TM mRNA Expression in the Lung—Temporal feeding restriction was found to induce a phase-shift of circadian gene expression in peripheral tissues, leaving the phase of cyclic gene expression in the suprachiasmatic nucleus unchanged (34). To determine whether TM gene expression is regulated by the peripheral clock, we examined the effect of restrictive feeding on TM expression. Under the conditions where mice could access the food freely (ad libitum), the cyclic expressions of clock genes (Per2) and TM mRNA was sustained. However, under scheduled restrictive feeding, the phases of circadian expression of Per2 and TM were affected, suggesting that TM gene expression is regulated by the peripheral clock (Fig. 6).

DISCUSSION

In this study we searched for clock-controlled genes using microarray analysis and found that TM is a clock-controlled gene expressed in vascular endothelial cells. The central biological clock is located in the suprachiasmatic nucleus in the hypothalamus (6). Recent evidence demonstrated the existence of a peripheral biological clock in each peripheral tissue, including the heart and vasculature. The central clock regulates a peripheral clock in each organ or tissue through a neurohormonal network (35). However, the biological relevance of the peripheral clock remains to be elucidated. To elucidate the biological significance of the peripheral clock, genes whose expression is regulated by the peripheral clock need to be identified. In addition, we need to clarify whether the peripheral clocks are altered under pathological conditions and whether the impaired biological clock is associated with the development of diseases (36).

Previously, as the first step to address these issues, we examined several genes as candidates of clock-controlled genes and found that PAI-1 expression is under the control of the clock genes, Clock and Bmal1/2 (12). Because PAI-1 plays a crucial role in controlling fibrinolytic activity, the circadian oscillation of plasma PAI-1 activity partially explains the predominate occurrence of thromboembolic events in the morning (17–20). Among cardiovascular functions, vascular endothelial function has been well documented to show circadian oscillation (37). For example, healthy individuals exhibit reduced endothelium-dependent vasodilation in the morning (38). Vascular endothelial cells prevent thrombogenesis, and the incidence of thromboembolic events, including pulmonary thromboembolism and AMI, exhibits morning peaks (1, 2). Moreover, the expression of the vascular endothelial growth factor gene shows circadian oscillation, implying the presence of a circadian rhythm in angiogenic processes in tumors (39). The internal biological clock is thought to play an important role in circadian oscillation of endothelial functions; however, the precise mechanisms are not fully elucidated. In this study we searched for clock-controlled genes in vascular endothelial cells and demonstrated that TM expression is under the control of the clock genes, Clock and Bmal1/2. The oscillation of TM mRNA expression was abolished in clock mutant mice, suggesting that TM expression is regulated by CLOCK in vivo. Moreover, the phase of circadian expression of TM was affected by scheduled restrictive feeding, implying cyclic TM gene expression was regulated by the peripheral clock.

BMAL2, as well as BMAL1, forms a heterodimer with CLOCK and transactivates the promoter activity of clock-controlled genes by direct binding to the E-box of their promoter sequences. The influence of either BMAL1 or BMAL2 on clock-controlled gene expression in vivo is thought to depend on the tissue or cell types.

Cardiovascular parameters, including heart rate and blood pressure, show circadian variation. At the molecular level, the cardiovascular system has its own internal clock system (12, 13). In the healthy state the clock system sustains cardiovascular function (15). However, in a pathological state, such as left ventricular hypertrophy, the circadian expression of several clock genes and genes encoding cardioprotective proteins, including atrial natriuretic polypeptide is attenuated (16). These data suggest that the peripheral clock is affected in damaged tissue. Furthermore, the clock system plays an important role in maintaining homeostasis. Mice carrying a mutation in the clock gene (clk/clk) show impaired circadian rhythm (24). In a recent report, clock mutant mice were shown to be susceptible to developing a metabolic disorder that resembled metabolic syndrome in humans (40). These results suggest that an impaired biological clock is associated with the development of diseases. Studies using genetically manipulated mice that have a tissue-specific impairment of the clock system will clarify the importance of the peripheral clock, separate from that of the central clock.

Regarding the circadian expression of the TM gene, the phase of circadian TM protein expression is similar to that of its mRNA level. TM protein is thought to be produced soon after transcription of the TM gene (41). However, the half-life of TM protein is ∼8 h in vitro (42). Therefore, in addition to the circadian regulation of its transcription, there may be other mechanisms such as degradation processes that account for the oscillation of TM protein levels.

One unresolved question concerns the biological significance of circadian TM expression. The times of the peak and trough expression of the TM gene were similar to those of PAI-1. Thrombin and PAI-1 act cooperatively to accelerate fibrin accumulation by inducing fibrin formation as well as inhibiting the degradation processes. TM exerts an opposite effect to PAI-1 by inhibiting the function of thrombin. Moreover, the complex of TM and thrombin activates protein C, which has an inhibitory effect on PAI-1 activity (21–23). One possibility is that TM may protect the endothelium against the thrombogenic activity of PAI-1. In clinical studies, members of families carrying a mutation in the TM gene showed an increased risk of thromboembolic diseases including PE and AMI (43, 44). Accordingly, impairment of TM function leads to an imbalance between fibrinolytic and thrombogenic activities and may induce thromboembolic events. In contrast, TM may promote blood coagulation through enhancement of the inhibition of thrombus lysis, since the protein functions as the activator of thrombin-activable fibrinolysis inhibitor (procar-boxypeptidase U) under the presence of a high thrombin concentration on the local vascular vessels (45, 46). Further studies are needed to elucidate the role of circadian expression of TM in cardiovascular diseases.

In conclusion, these data suggest that the peripheral clock in vascular endothelial cells regulates TM gene expression, and the circadian oscillation of TM expression may contribute to the circadian variation of cardiovascular events. Elucidation of the biological relevance of the peripheral clock of the heart and vasculature will provide new insights into the development as well as the prevention of cardiovascular disease.