FOXM1 regulates proliferation, senescence and oxidative stress in keratinocytes and cancer cells

Several transcription factors, including the master regulator of the epidermis, p63, are involved in controlling human keratinocyte proliferation and differentiation. Here, we report that in normal keratinocytes, the expression of FOXM1, a member of the Forkhead superfamily of transcription factors, is controlled by p63. We observe that, together with p63, FOXM1 strongly contributes to the maintenance of high proliferative potential in keratinocytes, whereas its expression decreases during differentiation, as well as during replicative-induced senescence. Depletion of FOXM1 is sufficient to induce keratinocyte senescence, paralleled by an increased ROS production and an inhibition of ROS-scavenger genes (SOD2, CAT, GPX2, PRDX). Interestingly, FOXM1 expression is strongly reduced in keratinocytes isolated from old human subjects compared with young subjects. FOXM1 depletion sensitizes both normal keratinocytes and squamous carcinoma cells to apoptosis and ROS-induced apoptosis. Together, these data identify FOXM1 as a key regulator of ROS in normal dividing epithelial cells and suggest that squamous carcinoma cells may also use FOXM1 to control oxidative stress to escape premature senescence and apoptosis.


INTRODUCTION
indicated that the expression of the transcription factor FOXM1 is strongly reduced in p63-depleted keratinocytes. FOXM1 is a member of the Forkhead domain protein family, which includes almost 100 different transcriptional factors involved in a broad range of processes [30][31][32], including stem cell expansion and renewal [33]. In particular, FOXM1 controls a network of genes necessary for G 2 -M phase transition and cell division [34]. FOXM1 is expressed during embryonic development, particularly in proliferative epithelia and mesenchymal cells [35]. In addition, FOXM1 has been shown to play an important role in oxidative stress regulation. Knocking down FOXM1 in fibroblast leads to increased intracellular levels of ROS. Catalase and superoxide dismutase are also known to be direct transcriptional targets of FOXM1 [36]. FOXM1 level is the most common differentially expressed gene in the majority of cancers, including oral, esophageal, lung, breast, kidney, and bladder cancer, as compared to normal tissues, suggesting a possible role in cancer initiation [37,38].
To date, studies of FOXM1 have been performed only in the context of cancer cell lines and/or immortalized cell lines in which the aberrant genetic background interferes with FOXM1 function. Herein, we sought to investigate the link between p63 and FOXM1 in normal human epidermal keratinocytes (NHEKs) by examining the effect of p63 deletion using specific siRNAs. We demonstrated that ΔNp63 indirectly regulates FOXM1 expression. Both transcription factors are important to maintain the proliferative potential of keratinocytes and to control the cellular redox state. FOXM1 depletion sensitizes normal keratinocytes and head and neck squamous carcinoma cells to apoptosis. Interestingly, FOXM1 knock-down induces replicative senescence, and skin biopsies of old subjects present reduced FOXM1 mRNA and protein levels compared with young subjects. Altogether, the data presented indicate that ΔNp63 and FOXM1 are important in maintaining and protecting skin progenitor cells to allow proper skin homeostasis and to counteract cellular senescence.

ΔNp63 indirectly regulates FOXM1 expression
Our previous RT-qPCR-based microarray and RNA-seq experiments [25,26] revealed that FOXM1 expression is strongly down-regulated in p63-depleted keratinocytes. We confirmed the results of the array in human keratinocytes ( Fig 1A) demonstrating that siRNAmediated knock-down of p63 in keratinocytes leads to decreased FOXM1 mRNA and protein levels. Next, we performed bioinformatic analyses to identify putative p63 response elements in the FOXM1 promoter. We identified one putative response element downstream of the transcription start site (TSS) and two that were upstream of the TSS (Fig 1B). We constructed vectors containing the sequences of these putative p63 response elements based on pGL3-vectors. Then, we carried out luciferase activity assays in H1299 cells overexpressing ΔNp63. As a positive control of ΔNp63 transcriptional activity, we used a pGL3 vector containing the keratin14 (K14) promoter sequence [9,20]. We did not observe any strong activation of luciferase expression with the FOXM1 promoter as compared to p63-target keratin14 (approximately 3-fold for RE2/3 versus approximately 15-fold for the K14 promoter, Fig 3 C-D). A mild activation of luciferase expression may indicate that some other transcriptional factors specific for keratinocytes are needed for p63-dipendent regulation of FOXM1 expression. We also confirmed this result performing chromatin immunoprecipitation and analyzing ChIP-seq data [21]: no p63 binding was detected on the indicated REs of the FOXM1 promoter ( Fig 1E and 1S). As a positive control, we used primers for ZNF750, which is known to be a direct target of p63 [5]. In addition, we overexpressed ΔNp63 in H1299 cells and analyzed FOXM1 mRNA expression, but we did not observe any changes (Fig 1 F-G). All these data indicate that p63 regulates FOXM1 indirectly.

FOXM1 levels correlate with proliferation status in keratinocytes
To investigate the function of FOXM1 in keratinocytes, we studied FOXM1 expression in calcium-induced differentiation. qRT-PCR and western blot analysis showed a dramatic decrease in FOXM1 mRNA and protein levels (Fig 2A) during keratinocyte differentiation. As a positive control, we measured keratin10 (K10) expression. Interestingly, in human epidermis FOXM1 co-localizes with p63 in progenitor layer of epidermis ( Fig 2B). To evaluate the role of FOXM1 as down-stream mediator of p63, we performed p63 and FOXM1 knock-down and measured the percentage of cells in S-phase. EdU-incorporation assays in NHEKs and in spontaneously immortalized (HaCaT) keratinocytes ( Fig 2D and Fig S3) indicated a 25% decrease in EdU-positive cells (from 18.0±3.3% to 12.5±3.3% in NHEK) upon siFOXM1, whereas sip63 caused an 80% reduction (Fig 2C-D, Fig S2). This difference could also due in part to knock-down efficacy that may be explained by higher protein stability of FOXM1 as has been shown previously [39,40]. As a control, we performed qRT-PCR and western blot analysis for c-Myc, whose expression is in part controlled by FOXM1 [41], confirming a link between FOXM1 and c-Myc (Fig 2E,F). Thus, we propose the www.impactaging.com presence of a positive feedback loop between FOXM1, p63 and c-Myc in controlling keratinocyte proliferation.
Altogether, these data indicate that FOXM1 is necessary in normal dividing epithelial cells.  www.impactaging.com

FOXM1 regulates oxidative stress and ROSmediated cell death in keratinocytes
We predicted that FOXM1 might protect keratinocytes from oxidative stress and ROS-mediated cell death. In order to verify this prediction, we measured intracellular ROS following siRNA-mediated knock-down of FOXM1 in keratinocytes. We observed a 1.5-fold increase in ROS levels in FOXM1-knock-down cells ( Fig 3A). Then, we exposed cells to oxidative stress using doxorubicin, which is known to produce reactive oxygen species. We observed a significant increase in cell death in FOXM1-depleted cells compared with control cells (Fig 3B). In parallel, qRT-PCR analysis revealed the decreased expression of genes encoding proteins involved in oxidative stress responses (e.g., CAT, PRDX, SOD2, and GPX2) ( Fig 3C). Altogether, these data showed that FOXM1 participates in balancing intracellular ROS level and protects keratinocyte from oxidative stress.

FOXM1 levels decrease during keratinocyte senescence and during skin aging
Having shown that FOXM1 has a role in protecting cells from oxidative stress, we next evaluated whether FOXM1 protects cells against senescence. We used a previously described [42] in vitro cellular model of replicative senescence, based on serial passaging of primary keratinocytes (P1, P2, P3, and P4). In this model, at P4, the cells undergo replicative-induced senescence, as determined by morphology, expression markers, and SA-beta-galactosidase activity. We analyzed the levels of FOXM1 in proliferating (P1) and senescent (P4) cells and observed a decrease in FOXM1 mRNA and protein levels (Fig 4A-B) in senescentinduced cells. Senescence was marked using p16, phosphorylated Retinoblastoma protein (pRb) and p63 (Fig 4 A-B). To investigate whether FOXM1 depletion is sufficient to induce senescence in proliferating keratinocytes, we performed siRNA-mediated knockdown of FOXM1 in P1 keratinocytes and evaluated the senescence-associated-β-galactosidase activity. As indicated in Fig 4C, the results showed a two-fold increase in β-galactosidase-positive cells in knockeddown samples compared with controls. Interestingly, FOXM1 levels were also down-regulated during skin natural aging in humans. We performed analysis of keratinocytes obtained from biopsies of young (<10 years old) and aged (>60 years old) donors and observed a 50% decrease in FOXM1 mRNA levels, which was paralleled by a decrease in protein levels (Fig 4 D-E). p16 served as positive control. These results indicate that FOXM1 expression significantly decreases during senescence in keratinocytes both in vitro and in vivo and that FOXM1 activity counteracts replicative-induced senescence.

FOXM1 regulates oxidative stress in epithelial squamous cell carcinoma
In order to investigate whether p63 and FOXM1 are also linked in epithelial cancer, we performed siRNA-  The results showed a significant decrease in FOXM1 at both the mRNA and protein levels (Fig 5A), indicating that p63 also indirectly controls FOXM1 expression in cancer cells. To study the physiological role of FOXM1 in this type of cancer, we performed siRNA-mediated knock-down of FOXM1. We exposed the cells to oxidative stress and measured ROS levels after exposure. Analysis of ROS levels a short time after hydrogen peroxide addition showed a significant increase in ROS levels in knocked-down samples with respect to controls (Fig 5B). We also analyzed ROS levels 24 h after treatment with hydrogen peroxide, which showed the same effect ( Fig 5C). Furthermore, we observed an increase in cell death in knocked-down samples with respect to the control at 24 h after treatment with doxorubicin (2-fold, Fig 5D). These results indicated that the p63-FOXM1 axis also has a protective role in cancer cells. Because different cancer types express high levels of FOXM1, FOXM1 anti-oxidant activity could be a mechanism through which cancer cells escape premature senescence and apoptosis.

DISCUSSION
Senescence is characterized by a progressive decline of cellular and body homeostasis [43][44][45]. When premature senescence acts cellular systems are not able to adequately respond to stress stimuli resulting in agerelated diseases [46][47][48][49][50]. FOXM1 is an important transcription factor that controls genes directly involved in cell cycle control and in the successful execution of the mitotic program, as well as in the maintenance of chromosome stability [51]. Studies have demonstrated that FOXM1 is required for the expansion of epithelial progenitor cells and that UV-light and environmental factors, such us nicotine, can directly activate the FOXM1 transcriptional network in keratinocytes [34], suggesting that FOXM1 could be an oncogenic hit. In fact, FOXM1 is overexpressed in various human malignancies, including prostate, breast, lung, ovary, colon, pancreas, stomach, bladder, liver and kidney cancer [52,53], and its oncogenic activity is downstream of Ras [54] and cyclin D1 [38], which are often implicated in epithelial tumors. In addition, FOXM1 has been shown to counteract oxidative stressinduced premature senescence by stimulating Bmi [55]. Additional studies using immortalized MEFs demonstrated that FOXM1 expression is induced by oncogenic stresses requiring ROS and that up-regulated FOXM1 engages a negative feedback loop to counteract ROS increases and protect dividing and cancer cells from oxidative stress [36]. Here, we investigated the role of FOXM1 in normal human primary keratinocytes. Important role in primary epithelial cells. We found that under normal conditions, ΔNp63 indirectly controls FOXM1 expression and that both factors are important to maintain the high proliferation rate of epithelial progenitor cells. FOXM1 also protects keratinocytes from apoptosis-induced oxidative stress by directly inducing the expression of anti-oxidant genes. Interestingly, FOXM1-depletion is sufficient to induce cellular senescence in proliferating epithelial cells, partially because of increased intracellular ROS (Fig 6). These in vitro results nicely paralleled the FOXM1 down-regulation (at the mRNA and protein level) measured in keratinocytes isolated from young and aged human donors biopsies. The latter results indicate that FOXM1 plays an important role, similar to that of p63, in maintaining normal tissue homeostasis and to avoid premature cellular senescence and, possibly, organismal aging (Fig 6). Interestingly, FOXM1 also regulates cmyc expression which has been shown to prevent cellular senescence [56][57][58] suggesting that both p63 and c-myc contribute to counteract cellular senescence in FOXM1-dipendent fashion. Although FOXM1 overexpression in various human malignancies has been very well described [52,53], no information is available regarding the mechanisms that controls FOXM1 overexpression. Our data indicated that the p63-FOXM1 axis is also present in submaxillary salivary gland epidermoid carcinoma (A253) cells, suggesting that ΔNp63 amplification could, at least in part, be responsible for FOXM1 over-expression in epithelial cancers. Finally, our results also indicated that squamous carcinoma cells uses FOXM1 to control oxidative stress to escape premature senescence and apoptosis.
Cell proliferation. The incorporation of EdU during DNA synthesis was evaluated using the Click-iT EdU flow cytometry assay kit according to the manufacturer's protocol (Molecular Probes). The cell cycle was analyzed using an Accuri C6 flow cytometer (BD Biosciences). Fifteen thousand events were evaluated using the Accuri C6 (BD) software.
Measurement of ROS level and cell death. NHEKs were treated with H2O2 (Sigma) or doxorubicin (Sigma) at the indicated concentrations. Cells were collected 24 h after exposure, washed with PBS and then stained with 10 μM CM-H2DCFDA (Life Technologies) for 20 min at 37°C, followed by analysis using a BD FACSCalibur flow cytometer (BD Biosciences). For cell death measurements, cells were collected with culture medium, treated with 13 kU/ml RNAse (Sigma) for 15 min at 37°C and then with 50 μg/ml propidium iodide (Sigma) for 20 min, followed by flow cytometry analysis.