CCAAT/Enhancer-binding Protein δ Mediates Tumor Necrosis Factor α-induced Aurora Kinase C Transcription and Promotes Genomic Instability*

Epidemiologic and clinical research indicates that chronic inflammation increases the risk of certain cancers, possibly through chromosomal instability. However, the mechanism of inflammation-dependent chromosomal instability associated with tumorigenesis is not well characterized. The transcription factor CCAAT/enhancer-binding protein δ (C/EBPδ, CEBPD) is induced by tumor necrosis factor α (TNFα) and expressed in chronically inflamed tissue. In this study, we show that TNFα promotes aneuploidy. Loss of CEBPD attenuated TNFα-induced aneuploidy, and CEBPD caused centromere abnormality. Additionally, TNFα-induced CEBPD expression augmented anchorage-independent growth. We found that TNFα induced expression of aurora kinase C (AURKC) through CEBPD, and that AURKC also causes aneuploidy. Furthermore, high CEBPD expression correlated with AURKC expression in inflamed cervical tissue specimens. These data provide insight into a novel function for CEBPD in inducing genomic instability through the activation of AURKC expression in response to inflammatory signals.

Chronic inflammation increases the risk of normal cells to become tumorigenic (1,2), and the immune system is thought to be the cause of chronic inflammation-associated tumorigenesis and cancer progression. Pro-inflammatory cytokines are thought to play a pathogenic role in age-related diseases, including cancer (3). The inflammatory cytokine TNF␣ promotes inflammation-associated carcinogenesis by activating nuclear factor-B (NF-B) (4) and signal transducers and activators of transcription (STATs) signaling (5). In addition, TNF␣ causes oxidative stress, enhances the population of Ͼ4N cells, and contributes to tumorigenesis (6). However, the acti-vation of oncogenes such as NF-B or STATs cannot fully explain why inflammation can kill tumor cells and yet promote tumor growth. This discrepancy further implies that there may be a molecule (or molecules) that plays a dual role in both killing and promoting tumor cells in inflammation.
The transcription factor CEBPD 3 was suggested to be a tumor suppressor because it can induce growth arrest and apoptosis in cancer cells (7) and loss of Cebpd increases the tumor incidence in a mouse mammary tumor model (8). On the other hand, CEBPD can also support tumor progression, most likely through a role in hypoxia adaptation (8). In addition, opposite to CEBPD up-regulated proapoptotic and growth arrest genes (9), some intriguing downstream targets of CEBPD, including cyclooxygenase 2 (10), the Fos/Jun family (11), and superoxide dismutase 1 (12), suggest that the CEBPD biology is complex and requires further studies, particularly in the pathogenesis that links inflammation and tumorigenesis.
In mammals, the aurora kinase (AURK) family consists of three members: aurora kinase A (AURKA), aurora kinase B (AURKB), aurora kinase C (AURKC). AURKs play important roles in the control of centrosome and spindle function, kinetochore-microtubule interactions, cytokinesis, and cell division (13)(14)(15)(16). Abnormal chromosome segregation can result in aneuploidy, which is associated with the majority of early embryonic loss in human, when it occurs in the germline (17). Aneuploidy is also characteristic of cancer cells. The dysregulation of AURKs is correlated with chromosomal instability and clinically aggressive cancer (18 -21). AURKC is localized at chromocenters in diplotene spermatocytes and centromeres in metaphase I and II and persists to a lesser extent in round spermatids (22). Similar to AURKB, AURKC can phosphorylate histone H3 at Ser-10 (23,24) and form a complex with inner centromere protein (INCENP) and survivin (24 -26). The expression of AURKC is very low in normal somatic tissues and higher in testis (27). Recently, AURKC was found to be overexpressed in primary colorectal cancers and thyroid carcinoma as well as in several cancer cell lines (28,29). Therefore, the deregulation of AURKC expression during tumorigenesis appears to be highly relevant and important.
In this study, TNF␣ was found to stimulate CEBPD expression in non-hematopoietic cells. Evidence that links the increase of CEBPD and the induction of aneuploidy was established. Briefly, CEBPD induction resulted in increasing aneuploidy, and caused abnormal DNA segregation, and a prolonged mitotic phase. Moreover, we identified that CEBPD specifically induced expression of AURKC in response to TNF␣. The clinical evidence that correlates CEBPD with AURKC expression in inflamed cervical tissue was also demonstrated.
Immunofluorescence Assay and Time-lapse Microscopy Analysis-HeLa and HeLa#6 and HeLa#7 cells were grown either with or without ZnSO 4 (100 M) for 60 h and then 50 ng/ml of nocodazole was added for 12 h. The round-up cells were collected and plated onto a coverslip for 30 min incubation. The experimental cells were fixed by ice-cold methanol: acetone (v/v ϭ 1:1) for 20 min. The fixed cells were then individually probed with ␣-tubulin (DM1A-FITC, Sigma, 1:50) or CEBPD antibodies (GTX115047; Gene Tex, 1:100). For the immunofluorescence analysis, the primary antibodies of the reactive samples were probed with Alexa 568 FITC-conjugated goat anti-rabbit IgG (Molecular Probes, 1:200). After washing with 0.1% Tween 20 in PBS, coverslips were mounted and the images were observed by a laser scanning confocal system (FV1000; Olympus). For the time-lapse microscopy analysis, the round-up cells were plated on 35-mm glass bottom culture dishes containing 1 g/ml of Hoechst 33342, and incubated in a heated stage (37°C) with 5% CO 2 . Phase-contrast and Hoechst images were captured every 3 min for 4 h by a laser scanning confocal system.

CEBPD Level Determinates Long-term TNF␣-induced
Tumorigenesis-As mentioned above, TNF␣ causes enhancement of the population of Ͼ4N cells and contributes to tumorigenesis. In hematopoietic cells, CEBPD is induced by inflammatory stimuli including TNF␣ (31). To assess the potential involvement of CEBPD in TNF␣-induced aneuploidy of nonhematopoietic cells, we first determined if long-term TNF␣ treatment can promote aneuploidy in the HeLa cervical carcinoma cell line. As shown in Fig. 1A and supplemental Fig. S1, TNF␣ treatment indeed increased the Ͼ4N cell population in HeLa cells regardless of nocodazole treatment. We next examined if CEBPD induction contributed to TNF␣-induced genomic instability. We found that the TNF␣-induced Ͼ4N HeLa cell population was reduced when CEBPD expression was attenuated by two RNAi (Fig. 1B), suggesting that CEBPD plays a functional role in the TNF␣-induced genomic instability. Next, we examined the effect of CEBPD on TNF␣-induced transformation using an anchorage-independent growth assay in soft agar. We found that the insensitive induction of CEBPD upon TNF␣ treatment caused larger HeLa cell focus formation on soft agar (Fig. 1C, left panel), in agreement with the hypothesis that CEBPD acts as a tumor suppressor. Interestingly, when we examined the CEBPD expression in smaller foci, sensitive induction of CEBPD upon TNF␣ treatment (  12). This important discovery suggests that CEBPD, depending on its expression level, could play a dual role in determining long-term TNF␣-induced tumorigenic transformation.
CEBPD Activation Enhances Genomic Instability-As mentioned above, genomic instability is a hallmark of cancer cells and is suggested to play a critical role in shifting the cells from normal to the malignant stage. To further confirm whether increased CEBPD affects genomic integrity, a CEBPD-inducible system was introduced into HeLa cells. Consistent with the results presented above, the induction of CEBPD resulted in the increase of Ͼ4N cell population in synchronized cells ( Fig. 2A).
It is known that abnormal chromosomal segregation during mitosis impair genome integrity. We therefore assessed the role of CEBPD in chromosomal abnormality. We observed that CEBPD induction caused an increase in the number of tripolar nuclei, DNA bridge occurrence, and micronuclei formation (Fig. 2B). These results indicate that CEBPD can affect genomic integrity through induction of abnormalities in chromosomes and centrosome. Cancer cells are known to contain extra chromosomes. Therefore one would expect that their mitosis are prolonged. We examined the possibility of prolonged mitosis by CEBPD-induced genomic instability by phase-contrast time-lapse microscopy. The results in Fig. 2C show that CEBPD-activated cells take an average of 103.7 min (n ϭ 91). The results show that CEBPD expression increased the dura-tion of cell division. In addition, to assess the effect of CEBPD on genomic integrity, GFP-stabilized wild-type Cebpd and GFP-stabilized Cebpd-deficient MEF cells were individually generated. After culturing for several passages, GFP expression was lost more readily in wild-type Cebpd MEF cells upon long-term TNF␣ treatment (supplemental Fig. S2), which indicated the greater ability to induce genomic instability. Taken together, these results show that CEBPD contributes to genomic instability.
CEBPD Mediates TNF␣-induced AURKC, Which Increases Genomic Instability-Deregulation of AURKA, AURKB, and AURKC has been demonstrated in many cancers (32). However, it is unknown whether AURKs are responsive to TNF␣ treatment. We found that in HeLa cells, TNF␣ induced the expression of CEBPD and AURKC, slightly reduced AURKA expression, and had no effect on AURKB expression (Fig. 3A). Consistent with these data, the results of reporter assays suggest that the TNF␣-induced AURKC expression was due to activation of the AURKC gene promoter (Fig. 3B). AURKC plays a functional role in mitosis and increased AURKC was observed in many cancers (27). However, the potential role of the increase in AURKC in disturbing genomic integrity is unknown. We found that induction of ectopic AURKC alone was sufficient to increase the Ͼ4N cell population (Fig. 3C, left  panel), and the incidence of tripolar nuclei, DNA bridges, and micronuclei (Fig. 3C, right panel). To further assess the effect of AURKC on genomic integrity, stable HeLa-off AURKC#11-GFP and HeLa-off GFP cells were individually generated. Over time, GFP expression was lost more readily in AURKC-overexpressing cells (Fig. 3D). Taken together, these results show that AURKC contributes to genomic instability.
Both CEBPD and AURKC activation showed a concomitant increase in genomic instability and responsiveness to TNF␣ treatment. However, CEBPD is not directly located on the condensed DNA during mitosis (supplemental Fig. S3), suggesting that the CEBPD-induced chromosomal abnormalities may be caused by a transcription-dependent manner. These observations led us to test whether AURKC is the downstream target of CEBPD. Transcripts of AURKC, but not AURKA or AURKB, increase in response to CEBPD induction not only in HeLa cells but also in the U373MG astrocytoma cell line (Fig. 4A). Moreover, this CEBPD/AURKC axis is also observed in other cells responding to TNF␣ treatment (supplemental Fig. S4). On the other hand, loss of Cebpd attenuated TNF␣-induced Aurkc transcription in MEF cells (Fig. 4B). To determine whether CEBPD and Cebpd activate AURKC and Aurkc gene expression through promoter activation, the AURKC and Aurkc reporters were cotransfected with human CEBPD and mouse Cebpd expression vectors, respectively. As shown in supplemental Fig.  S5A, Cebpd can transactivate the Aurkc reporter, whereas CEBPD transactivates the AURKC reporter. Moreover, activation of the AURKC reporter is specifically regulated by CEBPD but not LAP1, a C/EBP family member (supplemental Fig. S5B). Furthermore, the lentiviral-mediated expression of CEBPD shRNAs significantly attenuated the TNF␣-induced AURKC reporter activity (Fig. 4C). Western blot analysis verified that CEBPD knockdown abolished TNF␣-induced AURKC expression (Fig. 4D). Taken together, these results show that CEBPD plays an important role in TNF␣-activated AURKC promoter activity.
CEBPD Directly Regulates AURKC Promoter Activation-There are five putative CEBPD binding motifs within 1000 base pairs of the 5Ј-flanking region of the AURKC gene (Fig. 5A) identified by the PCDBM program. We further shortened the AURKC promoter from Ϫ996/ϩ84 bp (AC1P) to Ϫ593/ϩ84 bp (AC2P) for the purpose of determining the CEBPD-responsive region. As shown in Fig. 5A, CEBPD-dependent transactivation of AC2P reporter activity is similar to AC1P reporter activity, suggesting that the AC2P reporter contains the CEBPD-responsive element(s). We next performed an in vivo DNA binding assay to determine whether CEBPD binds directly to the endogenous AURKC promoter. Both exogenous HA/CEBPD and TNF␣-induced endogenous CEBPD were found to bind to the proximal 5Ј-flanking region of the AURKC promoter that contains the CEBPD-I and -II motifs (Fig. 5B). The same approach was conducted to assess whether Cebpd binds to the Aurkc promoter region. This result showed that the binding of Cebpd is detectable on the Aurkc promoter region upon TNF␣ treatment (supplemental Fig. S6). Furthermore, to specify the function of CEBPD-I and -II an in vitro DNA binding assay was conducted. The results show that the CEBPD-I motif exhibited a higher CEBPD binding activity than CEBPD-II in a gel shift assay (Fig. 5C). To determine the role of these elements in TNF␣-induced AURKC promoter activity, the reporter assay was performed with AC2P, AC2PM1 (AC2P mutant on mutant CEBPD-I motif), or AC2PM2 (AC2P mutant on CEBPD-II motif). Mutation of the CEBPD-I site reduced the basal activity of the AURKC promoter (Fig. 5D, compare the first lane with the fourth and seventh lanes). However, both elements contribute to the response to TNF␣ in a CEBPD-dependent manner (Fig. 5D, compare the second lane with the fifth and eighth lanes). These data suggested that the CEBPD-I motif plays an important role in basal AURKC promoter activation and that both CEBPD-I and -II are critical for TNF␣induced AURKC promoter activity.
Expressions of CEBPD and AURKC Are Coincident in Inflamed Cervical Tissue-The evidence presented above demonstrated that CEBPD is an important mediator of TNF␣-induced AURKC transcription. To assess the clinical relevance of the correlation between CEBPD and AURKC expressions in inflamed tissue, immunohistochemical staining was conducted on a clinical cervicitis specimen compared with normal cervical tissue. The results showed a simultaneous increase in expression of CEBPD and AURKC, but not AURKA, in inflamed cervical tissues (Fig. 6). These data strongly support clinical relevance for CEBPD-induced AURKC expression.

DISCUSSION
Although it is clear that cell proliferation alone does not cause cancer, sustained cell proliferation in an inflammatory environment combined with accumulation of genetic abnormalities contributes to initiation and promotion of cancer (33). Evidence is accumulating that demonstrates the link between inflammation and cancer (34). However, the precise mechanism remains uncertain. Two inflammation-responsive transcription factors, NF-B and STATs, have been suggested to initiate tumor growth by interfering with genomic integrity (6). In this report, we provide evidence that an additional inflammation responsive factor, CEBPD, affects genomic integrity at least in part through the activation of AURKC transcription. . TNF␣ induces AURKC expression through its promoter activation, and the increase of AURKC enhances the numbers of tripolar nuclei, DNA bridges, and micronuclei. A, TNF␣ induces AURKC and expression, represses AURKA expression, and does not affect AURKB expression. RT-PCR analysis with total RNA or Western blot (WB) with cell lysates from untreated HeLa cells or cells treated with TNF␣ for 3 h are indicated. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. B, the AURKC reporter is activated by TNF␣ treatment. Following the description under "Experimental Procedures," a luciferase assay was performed using various AURK reporters, AURKA reporter (pAURKA), AURKB reporter (pAURKB), and AURKC reporter (pAURKC), after TNF␣ treatment. The figure shows representative data from one of three independent experiments performed in triplicate. C, AURKC increases the occurrence of tripolar nuclei, DNA bridges, and micronuclei. Following the procedure described under "Experimental Procedures," HeLa-off (Con.) and HeLa-off/AURKC-Myc#11 (#11) cells were harvested for immunofluorescence microscopy. The figure is plotted as the mean Ϯ S.D., n ϭ 3. Western analysis shows AURKC-Myc expression in clone number 11. D, the increase of AURKC enhances the loss of GFP from stable GFP-positive cells. Equal amounts of HeLa-off/GFP (Con-G) and HeLa-off/AURKC#11/ GFP (#11-G) cells were seeded for growth. The GFP-positive cells were calculated on the indicated days and normalized to their individual first day (as 100%). A very similar pattern was observed in two independent experiments. The figure is representative of one experiment performed in triplicate. Statistical analysis in this figure was performed with Student's t test (*, p Ͻ 0.05; **, p Ͻ 0.01).
Due to a tight feedback control between proinflammatory and anti-inflammatory cytokines, normal inflammation, but not chronic inflammation, is a self-limiting process to prevent the persistence of immune responses (35). This raises the intriguing scenario of inflammation serving as a two-edged sword to both suppress and promote tumor formation and cancer progression. Previously, CEBPD was reported to act as an innate inflammation-responsive factor (36), which persists in high concentration in chronically inflamed tissues (9). However, unlike nuclear factor-␤ (NF-␤) and signal transducers and activators of transcription (STATs), the level of CEBPD is low during tumorigenesis, as confirmed in multiple cancers (7,9). We recently demonstrated that an epigenetic regulator, the polycomb group/DNA methyltransferases pathway, increases CEBPD gene hypermethylation, which results in CEBPD being resistant to external stimulation (7). Whereas these reports are in line with a tumor suppressor function of CEBPD, the results of this study suggest that TNF␣-induced CEBPD mediates in part the transforming activity of TNF␣.
Genomic instability is observed in most cancers, and although tumorigenesis has been attributed to the dysregulation of mitotic checkpoint proteins (37), the molecular mechanisms underlying inflammation-related chromosomal instability associated with tumorigenesis are poorly understood. For example, the mitotic regulators that control genomic stability in chronic inflammation-induced tumors remain relatively unknown. Increased cyclooxygenase 2 is known to induce genomic instability, and we previously reported that cyclooxygenase 2 is a CEBPD-responsive gene (38), which suggests that a sustained CEBPD expression may play a role in oncogenesis. Moreover, based on the current study, the association between aneuploidy and chronic inflammation can be linked to CEBPDmediated AURKC expression. This study not only highlights the role of chronic CEBPD expression promoting genomic FIGURE 4. AURKC is a CEBPD-responsive gene. A, AURKC transcripts respond to CEBPD induction. The transcript products and protein lysates were harvested from two CEBPD-inducible cell lines and then analyzed by RT-PCR or Western blot (WB) with specific primers or antibodies as indicated. B, Aurkc transcripts respond to TNF␣ treatment in Cebpd ϩ/ϩ MEFs. RT-PCR assays were performed with the total RNA products harvested from TNF␣-treated Cebpd Ϫ/Ϫ and Cebpd ϩ/ϩ MEFs. CEBPD inactivation attenuates TNF␣-induced AURKC reporter activity (C) and protein levels (D) in HeLa cells. The various reporters, as indicated, were transfected into the experimental cells expressing shLacZ or shCEBPDs (shD1 and shD2). Luciferase assays were performed using the lysates harvested from the transfectants after treatment with or without TNF␣ for 16 h. All of the reporter activities of control transfectants with shLacZ were set to 1 as the standard. The numbers of relative reporter activities were normalized with individual standard controls with the mean Ϯ S.E. Statistical analysis was performed with Student's t test (*, p Ͻ 0.05). The loss of CEBPD-attenuated TNF␣-induced AURKC reporter activity was represented in the right panel of C; the reporter activities of TNF␣-treated transfectants with shC were set to 1 and 100, respectively, as the standard. FIGURE 5. CEBPD binds to the AURKC promoter in vivo, and the identification of CEBPD-responsive motifs. A, the AURKC promoter Ϫ593/ϩ84 is important for activation by CEBPD. Following the description under "Experimental Procedures," a luciferase assay was performed using the AURKC promoter Ϫ996/ϩ84 reporter (AC1P) or Ϫ593/ϩ84 reporter (AC2P) and cotransfected with HA/CEBPD. The reporter activities of the transfectants expressing the empty vector (pCDNA3-HA) were set to 100 as the standard. The data represent three independent experiments. B, CEBPD binds to the AURKC promoter in vivo. ChIP assays were performed with HeLa cells by using the indicated antibody-immunoprecipitated products from the experimental cells expressing HA/CEBPD (left panel) or treated with or without TNF␣. The locations of the specific primers on the AURKC promoter region are shown in the upper panel. C, CEBPD binds to the C/EBPD-I and -II motifs. EMSA was performed with protein-DNA complexes formed by in vitro translated HA-CEBPD protein (I.V.T. D) and individual 32 P-labeled probes bearing putative CEBPD motifs, C/EBPD-I or C/EBPD-II. D, CEBPD-I and -II motifs are important for TNF␣-induced AURKC reporter activity. HeLa cells transfected with various luciferase reporters, as indicated, were stimulated with or without TNF␣. The experiment was performed as described in the legend to Fig. 4C. Similar results were obtained from three independent experiments, and the data are from one experiment performed in triplicate. The average fold-induction is shown as the mean Ϯ S.E. The mutated sequence in each individual construct is shown by a black oval.
instability but also suggests that CEBPD could account in part for the connection between inflammation and tumorigenesis.
The accurate expression of mitotic proteins, such as AURKs, plays a critical role in the maintenance of genomic integrity (32). Our results demonstrate that AURKA and AURKC are inactivated and activated, respectively, in response to TNF␣, and that AURKC is specifically up-regulated by CEBPD. Although the mechanism of AURKA reduction during inflammation remains unknown, TNF␣-induced CEBPD indeed plays a functional role in transcriptional activation of the AURKC gene, which subsequently causes aneuploidy and genomic instability (Figs. 3 and 4). Induction of AURKC was reported in the G 2 /M phase, which is coincident with or followed by, an increase in AURKB expression (25). However, following cell cycle progression we found that the expression of AURKC is comparable in asynchronized and synchronized HeLa cells (supplemental Fig. S7A). Furthermore, CEBPD showed a slightly shifted pattern, but not an increase, during the synchronized condition (supplemental Fig. S7A). In addition, CEBPD induction not only induces AURKC expression but also promotes the phosphorylation of histone H3 (supplemental Fig.  S7B), which is known to associate with chromosome condensation and segregation (39). These observations suggest that the promotion of genomic instability regulated by the CEBPD-mediated TNF␣-dependent induction of AURKC transcription uncouples mitotic control of cell cycle progression.
Cervical human papillomavirus infection is not believed to be inflammatory in nature (40). Nevertheless, there is some epidemiological evidence, albeit weak, to suggest that inflammation might be linked to cervical cancer, perhaps as an human papillomavirus cofactor (41,42). Moreover, populations with high rates of cervical inflammation are associated with high incidence rates of cervical neoplasia (43). In this report, we demon-strated that long-term TNF␣ treatment increased aneuploidy and anchorage-independent growth of the HeLa cervical carcinoma cell line, which was attenuated by blocking CEBPD expression (Fig. 1, B and C). Therefore, we suggest that inflammation-induced CEBPD expression may promote human papillomavirus-mediated tumorigenesis in cervical cells through inflammation-disrupted genome stability.
Accurate mitotic protein levels play a critical role in maintaining genomic integrity. Primary Cebpd-deficient MEFs also exhibit genomic instability and centrosome abnormalities (44). Thus, it may be important to precisely regulate the levels of CEBPD protein to maintain genomic integrity. In addition, cells that survive TNF␣-induced CEBPD-mediated genomic instability may be similar to cells that escape mitotic failure-induced apoptosis and contribute to the risk of tumorigenesis.
In summary, this study shows that CEBPD can cause genomic instability, which provides a new mechanism by which it may promote tumorigenesis. The dual role of CEBPD may in part underlie the dual functions of TNF␣ in both suppression and promotion of cancer progression and the complex role of inflammation in cancer.