Background
The forkhead box (FOX) transcription factors have been shown to regulate cell growth, proliferation, differentiation, longevity and transformation and exhibit a diverse range of functions during embryonic development and adult tissue homeostasis [reviewed in [
1]]. FOXM1-null mouse embryos were neonatal lethal as a result of the development of polyploid cardiomyocytes and hepatocytes, highlighting the role of FOXM1 in mitotic division [
2]. More recently a study using transgenic/knockout mouse embryonic fibroblasts and human osteosarcoma cells (U2OS) has shown that FOXM1, regulates expression of a large array of G2/M-specific genes, such as Plk1, Cyclin B2, Nek2 and CENP-F, and plays an important role in maintenance of chromosomal segregation and genomic stability [
3].
A key intrinsic mechanism that determines cell survival and apoptosis is the ability to detect and respond to genotoxic insults such as chemical carcinogens, ultraviolet or ionising irradiation. Failure to regulate DNA damage response checkpoints and subsequent genomic stability in cells often leads to tumourigenesis [
4]. The forkhead protein FOXO3a has been shown to play a role in both DNA repair pathways and cell cycle checkpoint in response to DNA damage [
5]. Moreover, it has recently been reported that FOXO3a can be modulated by oncogenes such as MUC1 causing increased DNA repair and enhanced cell survival in response to oxidative stress [
6] and recently FOXM1 was shown in a cancer cell line to stimulate DNA repair genes following genotoxic stress [
7].
Basal cell carcinoma (BCC) accounts for up to 20% of all Caucasian carcinomas. We were the first to establish a link between FOXM1 and tumourigenesis when we demonstrated that FOXM1 is upregulated in BCC [
8]. Since then, FOXM1 has been implicated in the majority of solid human cancers [reviewed in [
9]]. We recently showed that FOXM1 expression precedes malignancy in a number of solid human cancer types including oral, oesophagus, lung, breast, kidney, bladder and uterus indicating its pivotal role in cancer initiation [
10]. The present study investigated the putative early mechanism of UVB and FOXM1 in skin cancer initiation. We have used a high efficiency long-term retroviral transduction system to express exogenous FOXM1B in both immortal and primary normal human epidermal keratinocytes (NHEK) to replicate oncogenic levels found in cancer cells. Using Affymetrix SNP microarray to profile genomic instability we show that upregulation of FOXM1B in epidermal keratinocytes results in genomic instability and that this is augmented by UVB, a major aetiological factor in BCC.
Discussion
Our previous studies showed that FOXM1B is upregulated in BCC [
8] but its role in the tumour initiation remains unclear. The present study investigated the effect of upregulating FOXM1B in primary and immortalised human epidermal keratinocytes. To avoid overexpression artefacts, we titrated retroviral supernatant to achieve levels of FOXM1B expression, similar to those found in various cancer cell lines. This study presents the first evidence that FOXM1B is dose-dependently activated by UVB through protein stabilisation and its upregulation alone induces genomic instability in primary human epidermal keratinocytes.
We found that UVB inhibited proteasomal proteolysis and dose-dependently upregulated FOXM1B protein levels resulting in acute (within 3 hours) FOXM1B protein stabilisation and accumulation in the absence of de novo mRNA/protein synthesis. This agrees with a previous study showing FOXM1 protein stabilisation, rather than de novo mRNA expression, following UV, ionizing irradiation and Etoposide treatment in a human osteosarcoma U2OS cancer cell line [
7]. However, it is important to note that whilst UVB was able to upregulate endogenous FOXM1B and that FOXM1 has been shown to induce its own expression [
30], other factors such as mutations in PTCH and SMO with subsequent upregulation of Gli transcription factors are most likely to be responsible for the initial upregulation of FOXM1 in BCC ([
8]). However, because BCC keratinocytes are very difficult to maintain in culture, it is not possible to investigate this in primary tumour cells. However, the direct activating effect of DNA damage on FOXM1B activity may also explain why genotoxic agents, such as ionising radiation, chemotherapy, intensive photochemotherapy and arsenic intoxication, increase the rate of BCC development [
31].
It is known that oncogene expression in normal cells triggers DNA-damage checkpoint as a first anti-cancer barrier response to prevent proliferation of damaged cells [
4,
32,
33]. Our results indicate that acute upregulation of FOXM1B transiently activated CDK inhibitor p21
cip1 and stress kinase p38 in primary NHEK. In marked contrast to our study in primary NHEK, Wang et al [
34,
35] have shown in murine hepatocytes and human U2OS osteosarcoma cells that FOXM1B expression suppressed p21
cip1 and p27
kip1 and promoted cell cycle progression. One possible explanation for these differences may be the fact that Wang et al used mouse cells and human carcinoma cells presumably with diverse or abnormal cellular background. In support of this, a recent study investigating the interaction between p53 and FOXM1 showed that different cancer cell lines exhibit different responses to DNA damage-induced FOXM1 levels depending on the p53 expression status [
36]. Moreover, we have found that in the N/TERT immortal keratinocyte cell line with suppressed levels of p16
INK4A and compromised checkpoint mechanism [
11], FOXM1B expression downregulated the levels of p21
cip1 (data not shown), suggesting a clear difference between primary and cancer cell lines in terms of response to FOXM1B expression. Interestingly, our current study shows that upregulation of FOXM1B in primary NHEK triggered only a minor apoptotic response despite activation of p21
cip1 and p38. This suggests that upregulation of FOXM1B allowed cells to tolerate significantly higher levels of p21
cip1 and activation of stress kinase p38. Upregulation of FOXM1B in primary NHEK showed enhanced apoptosis following UVB exposure, which is in agreement with a report showing that DNA damage in c-Myc-overexpressing normal mammary epithelial cells, sensitizes cells to DNA damage-induced apoptosis [
37]. Despite sensitising cells to UVB-induced apoptosis, the pro-proliferation survival advantage provided by the upregulation of FOXM1B may result in a selection of cells that escape cell death.
The existence of DNA replication stress is a common feature in human pre-cancerous lesions [
38] and recently, it has been shown that chronic induction of low, but not high, levels of Ras oncogene activation predisposes cells to tumour formation without inducing permanent cell cycle arrest [
39]. Furthermore, a recent study showed that DNA damage upregulates FOXM1 in cells with defective p53 pathway [
36]. This may explain our hypothesis that upregulation of FOXM1 following UVB exposure occurs in cells with defective checkpoint mechanism. Therefore, FOXM1 upregulation may provide a mechanism whereby cells evade a checkpoint response which allows damaged cells to proliferate and accumulate genomic instability.
Activation of cellular senescence pathways via the activation of p21
cip1 or p16
INK4A causes defects in the DNA damage response resulting in increased sensitivity to genotoxic stresses [
40]. We propose that FOXM1B-induced activation of p21
cip1 or p38 in NHEK may be a result of genomic instability and increase sensitivity to subsequent genotoxic stress (such as UVB) thereby accelerating the selection of genetically unstable cells. We hypothesised that this may be a mechanism whereby upregulation of FOXM1 by UVB may initiate and expedite carcinogenesis.
Given the role of FOXM1B in maintenance of chromosomal segregation and genomic stability [
3] and our findings that FOXM1B triggered DNA-damage stress responses (p21
cip1 or p38) in primary NHEK following UVB exposure, we investigated whether FOXM1B upregulation may be inducing DNA damage in the form of genomic instability. Recent reports have shown that oncogenes such as Ras induces chromosomal instability to promote malignant transformation [
41]. Moreover, we have previously shown that genomic instability was widespread in BCC [
14]. We have used a well established and highly sensitive genome-wide Affymetrix SNP mapping technique to profile and quantify genomic instability in the form of LOH and CNV. To our knowledge, this study provides the first evidence that constitutive and acute (4 days) expression of FOXM1B alone in the absent of other stimuli is sufficient to induce LOH and CNV in primary normal human keratinocytes. Furthermore, the FOXM1B-induced genomic instability was accumulated when these cells were passaged in culture. In support of this hypothesis, N/TERT cells expressing FOXM1B, but not EGFP, showed gross chromosomal CNV and LOH following UVB exposure. Interestingly, numerous genes (including MAP3K7IP2, SUMO4, p34/ZC3H12D, LATS1, RAET1 cluster, ULBP cluster, AKAP12, ESR1, MYCT1, VIP, TIAM2, SOD2, WTAP, MAS1, SLC22A cluster, IGF2R, etc.; see see additional file
1) found within the FOXM1B-induced LOH at 6q25.1-6q25.3, have been previously linked to oncogenesis of various human cancers [
40,
42‐
53]. Furthermore, in support of our data, genes including EGFR and IGFB1-3 found within the UVB/FOXM1B-induced CNV gain in chromosome 7p12-22 were previously reported to be amplified in HNSCC [
54]. This strongly indicates that upregulation of FOXM1B synergises with oncogenic stress (UVB) to promote genomic instability which may help cells gain a survival advantage. In support of our findings in skin keratinocytes, we recently showed that FOXM1B upregulation directly induces genomic instability in primary human oral keratinocytes and that nicotine at a genotoxic concentration promoted FOXM1-induced malignant transformation in oral keratinocytes [
10]. Nevertheless, further experiments are required to establish whether the FOXM1B-induced genomic instability is responsible for generating oncogenic LOH and CNV involved in skin malignant transformation.
It is known that FOXM1B plays an important role in the maintenance of genomic stability [
3,
55] and that FOXM1B is upregulated in majority of human cancers [
1]. Although FOXM1B at physiological level has been reported as a regulator of DNA repair [
7], its upregulation is likely to interfere with the normal DNA repair mechanism leading to enhanced genomic instability rather than enhanced DNA repair. This highlights the fact that a tight regulation of FOXM1B expression level is required during the cell cycle for proper maintenance of genomic stability. Hence, FOXM1B-induced genomic instability could be a result of aberrant mitotic division due to aberrant expression of mitotic spindle assembly genes such as CEPN-F, Aurora B and Plk1 [
3,
55] and genes involved in sister chromatids separation and cytokinesis such as CEP55 which we have recently shown to be a downstream target of FOXM1B [
10]. In support of our findings, numerous studies have demonstrated that proteins which are important in DNA repair and the maintenance of genomic stability, including mitotic spindle-associated proteins are often found amplified in human cancers, with centrosome amplification being a well characterized mechanism giving rise to genomic instability [
56]. Furthermore, consistent with our findings, a recent study has shown that upregulation of FOXM1 cells confer cisplatin resistance in breast cancer cells through deregulation of the DNA repair pathway causing genomic instability [
57]. CENP-F (mitosin) overexpression has also been linked to the generation of chromosomal instability in breast cancer patients [
58] as well as in head and neck squamous cell carcinomas [
59]. Upregulation of Aurora centrosome kinase has been associated with genomic instability in primary human non-small cell lung carcinomas [
60], pancreatic cancer [
61], and ovarian cancer derived cell lines [
62]. Furthermore, FOXM1B upregulation has been reported in majority of human cancers [
1], suggesting that gain of FOXM1B function is an important step in human carcinogenesis. In agreement, a recent study measured the levels of aneuploidy, as a marker for genomic instability in 6 different human tumours types, based on genome-wide gene expression pattern, the study found that FOXM1 was the third highest ranked gene with a consensus expression pattern significantly associated with genomic instability in diverse human malignancies [
63].
Whilst upregulation of FOXM1B alone can induce genomic instability, we have found that this mechanism alone is not sufficient to induce malignant transformation in NHEK because the rapid replicative exhaustion of NHEK in culture may not allow sufficient time for cells to acquire subsequent oncogenic hits necessary for malignant transformation. In support, FOXM1B overexpression alone did not induce malignant transformation in oral keratinocytes [
10]. Indeed, many studies have shown that normal human cells are highly resistant to single-oncogene mediated transformation which usually requires multiple oncogenic hits [
64,
65]. In line with our findings, in the presence of a second oncogenic pressure such as UVB, FOXM1B, but not EGFP, expressing cells acquired and accumulated definitive LOH and CNV loci, suggesting that upregulation of FOXM1B may predispose cells to malignant transformation. This notion is strongly supported by our previous finding that FOXM1B-expressing oral keratinocytes are highly predisposed to nicotine-induced malignant transformation [
10]. Our current study provided further evidence that upregulation of FOXM1B alone without UVB exposure in primary NHEK resulted in genomic instability which could be retained, accumulated and amplified in multiple cell culture passages thereby creating an oncogenic selection pressure prior to UVB exposure.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MTT conceived, coordinated the study, interpreted data, wrote and finalised the manuscript, performed retroviral transduction, UVB time-course and dose-dependent fluorescence microscopy, FACS, SNP microarray and SNP data analysis. EG performed cell culture, retroviral transduction, immunoblotting, SNP microarray, data interpretation, discussion and manuscript writing and editing. TC and BDY contributed to SNP chip scanning and data interpretation. MPP participated in the design, interpretation and discussion of the study, help editing, writing and finalising the manuscript. All authors read and approved the final manuscript.