Contrasting effects of an Mdm2 functional polymorphism on tumor phenotypes

Abstract

MDM2, an E3 ubiquitin ligase, is a potent inhibitor of the p53 tumor suppressor and is elevated in many human cancers that retain wild type p53. MDM2 SNP309G is a functional polymorphism that results in elevated levels of MDM2 (due to enhanced SP1 binding to the MDM2 promoter) thus decreasing p53 activity. Mdm2^SNP309G/G mice are more prone to spontaneous tumor formation than Mdm2^SNP309T/T mice, providing direct evidence for the impact of this SNP in tumor development. We asked whether environmental factors impact SNP309G function and show that SNP309G cooperates with ionizing radiation to exacerbate tumor development. Surprisingly, ultraviolet B light or Benzo(a)pyrene exposure of skin shows that SNP309G allele actually protects against squamous cell carcinoma susceptibility. These contrasting differences led us to interrogate the mechanism by which Mdm2 SNP309 regulates tumor susceptibility in a tissue-specific manner. While basal Mdm2 levels were significantly higher in most tissues in Mdm2^SNP309G/G mice compared to Mdm2^SNP309T/T mice, they were significantly lower in Mdm2^SNP309G/G keratinocytes, the cell type susceptible to squamous cell carcinoma. The assessment of potential transcriptional regulators in ENCODE ChIP-seq database identified transcriptional repressor E2F6 as a possible negative regulator of MDM2 expression. Our data show that E2F6 suppresses Mdm2 expression in cells harboring the SNP309G allele but not the SNP309T allele. Thus, Mdm2 SNP309G exhibits tissue-specific regulation and differentially impacts cancer risk.

INTRODUCTION

A wealth of evidence exists associating single nucleotide polymorphisms (SNPs) and disease susceptibility (1–3); however, the detailed mechanisms behind this susceptibility are relatively unexplored. Functional SNPs are positioned within regulatory or protein coding sequences and can alter gene expression or protein structure, thus altering function and contributing to disease risk (4–10). Functional SNPs in tumor suppressor pathways exist and mechanistic studies to determine how these genetic variants destabilize cell homeostasis and contribute to cancer risk may provide insight for prevention and treatment of human cancer.

The TP53 gene is a tumor suppressor and guardian of the genome (11). In response to stress signals, p53 transactivates many genes involved in processes that are critical for maintaining cell homeostasis including cell cycle arrest, DNA repair, and apoptosis (12–15). Thus, TP53 is the most highly altered gene in human cancers (16). The importance of p53 function is further exemplified by the observation that slight differences in activity can lead to significant effects on cell function, tumor susceptibility, and survival (17, 18); consequently, p53 levels and function are tightly regulated. In addition to TP53 mutations and deletions, alternate mechanisms also disrupt the p53 pathway (19). MDM2 is an E3 ubiquitin ligase that ubiquitinates p53 targeting it for proteasomal degradation (20). Previous studies have illustrated the interrelationship between these two proteins; p53 transcriptionally activates the Mdm2 gene, which in turn produces a protein that binds and inhibits p53 activity creating a negative feedback loop (21). In addition, mouse models with Mdm2 deletion exhibit embryonic lethality, a phenotype that is completely rescued by p53 deletion (22, 23). Moreover, tumor studies revealed a mutually exclusive relationship between MDM2 and p53; tumors harboring MDM2 amplification seldom carry TP53 mutations (19, 24).

MDM2 SNP309G is a functional polymorphism in the p53 pathway that associates with increased cancer risk in many cancer types, including lung, colon, pancreas, endometrium, and head and neck (25–29). Furthermore, genetically engineered mice harboring the homozygous Mdm2 SNP309G allele are more susceptible to spontaneous tumor formation and exhibit a significantly lower overall survival compared to homozygous Mdm2 SNP309T mice (30). Mechanistically the SNP309G sequence creates a stronger binding site for the transcription factor SP1 in the MDM2 P2 promoter, which results in elevated MDM2 expression leading to dampened p53 activity (26, 31). However, recent studies have identified a protective role of SNP309G in some cancers (28, 32–34). For example, multiple studies associate the SNP309G allele with a decreased risk and late onset for prostate cancer compared to patients harboring the SNP309T allele (28, 34). Thus, SNP309 appears to differentially impact cancer susceptibility in different cancers for reasons that are currently unknown.

Functional SNPs may exacerbate cancer risk by exposure to environmental factors, yet these mechanisms are also poorly defined (35, 36). Our current study tested the hypothesis that Mdm2 SNP309G allele cooperates with environmental factors to exacerbate spontaneous cancer risk. Low dose ionizing radiation (IR), a source of reactive oxygen species (ROS) and DNA damage, exacerbates spontaneous tumor formation in the Mdm2^SNP309G/G mouse compared to Mdm2^SNP309T/T mice. Conversely, exposure to ultraviolet B (UVB) light or Benzo(a)pyrene (B(a)P), a carcinogen found in tobacco products, increased tumor risk in the skin of Mdm2^SNP309T/T mice compared to Mdm2^SNP309G/G mice. Here our investigations show that SNP309G directly alters Mdm2 basal expression in a tissue-dependent manner. In skin keratinocytes, E2F6 represses Mdm2 expression specifically from the SNP309G allele. These findings suggest that the SNP309G allele cooperates with environmental and cellular factors to alter cancer risk in a tissue-dependent manner.

RESULTS

SNP309G allele exacerbates tumor risk after low-dose IR

IR generates DNA double strand breaks and reactive oxygen species (ROS), direct mediators of p53 activation (37). Since SNP309G dampens p53 activity, we tested whether the SNP309G allele cooperates with low dose IR to increase spontaneous tumor risk in mice. First, we treated Mdm2^SNP309G/G and Mdm2^SNP309T/T mice in a C57BL/6 background with a single dose of 1Gy IR 2 days postpartum, a regimen previously shown to activate acute p53 response in C57BL/6 mouse tissues (38), and analyzed p53 activity. As expected, untreated spleen samples from Mdm2^SNP309G/G mice showed significantly higher Mdm2 basal levels (p=0.005) as compared to spleens from Mdm2^SNP309T/T mice. The spleens of Mdm2^SNP309G/G mice showed 7.8-fold increase in Mdm2 levels (p<0.001) 3 hours post IR compared to Mdm2^SNP309T/T spleens [Figure 1A] with concomitant decrease in p53 targets Cdkn1a and Bbc3 consistent with elevated Mdm2 levels and an attenuated p53 response [Figure 1B].

Figure 1.

Attenuation of the p53 pathway in an Mdm2^SNP309 allele specific manner after IR. A, Mdm2 expression in spleens of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice 3hrs post 1Gy IR (n = 5, mean ±SD). B, Expression of Cdkn1a and Bbc3 (p53 targets) in spleens of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice 3hrs post 1Gy IR. Expression levels are normalized to that of untreated T/T samples. C, D, Immunohistochemistry of spleens and thymi of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice with caspase-3 antibody. Graphs represent mean value, n = 3, mean ±SD. Statistical significance was determined by two-way analysis of variance (ANOVA). T/T, SNP309T, Mdm2^SNP309T/T mice; G/G, SNP309G, Mdm2^SNP309G/G mice. n= number of biological replicates.

To determine the effects on cell survival, we examined the onset of apoptosis post IR. Mdm2^SNP309G/G mouse lymphatic tissues exhibit a delay in activation of apoptosis compared to Mdm2^SNP309T/T after low-dose IR as measured by the number of cells staining positively for cleaved caspase-3 [Figure 1C, 1D]. At 6 hours post IR, spleen and thymi from Mdm2^SNP309T/T mice showed the presence of apoptotic cells compared to none in the same tissues of Mdm2^SNP309G/G mice; however, by 12 hours post IR, spleens and thymi from Mdm2^SNP309G/G mice showed increased apoptosis compared to Mdm2^SNP309T/T tissues, indicating a delayed response to IR. Next, we utilized the same regimen to examine long-term effects of low-dose IR on tumorigenesis. We generated a cohort of irradiated Mdm2^SNP309G/G and Mdm2^SNP309T/T mice in a C57BL/6 background and monitored for tumor development. Tumors developed significantly more quickly in irradiated Mdm2^SNP309G/G mice than in irradiated Mdm2^SNP309T/T mice (p<0.001) [Figure 2A]. Mdm2^SNP309G/G mice exhibited a median survival of 74 weeks; in contrast, the majority of Mdm2^SNP309T/T mice survived without tumor development and thus the median survival could not be calculated. Mdm2^SNP309G/G mice had a proclivity to develop lymphomas and sarcomas; however, a number of mice also developed other tumors, including mammary carcinomas, glioblastomas, and histiocytic sarcomas [Figure 2B and Table 1]. In addition, Mdm2^SNP309G/G mice treated with 1Gy IR had significantly increased tumor multiplicity compared to Mdm2^SNP309T/T mice (p<0.001) [Figure 2C]. Only irradiated Mdm2^SNP309G/G mice had three or more tumors per mouse. We further interrogated the lymphoma phenotype due to its high penetrance [Figure 2D]. Here, we observed that irradiated Mdm2^SNP309G/G mice developed lymphomas significantly faster (p<0.001) compared to Mdm2^SNP309T/T mice, of either T cell (CD3 marker) or B-cell (B220 marker) origin [Figure 2E]. These experiments demonstrate that the SNP309G allele impacts p53 activity and exacerbates spontaneous tumor development after lowdose IR compared to the SNP309T allele.

Figure 2.

Exacerbation of tumor phenotypes after 1Gy IR. A, Kaplan-Meier survival curve of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice after 1Gy IR. B, Tumor incidence of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice. C, Tumor multiplicity. Statistical analyses for tumor incidence and tumor multiplicity was performed by Fisher’s exact test. D, Lymphoma-free survival curve of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice. Statistical analyses for Kaplan-Meier survival curves was performed by Log-rank test. E, Immunohistochemistry of Mdm2^SNP309G/G lymphoma tissues with CD3 & B220. n= number of biological replicates.

TABLE 1.

Tumor spectrum of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice after 1Gy IR.

Tumor Type	Mdm2SNP309T/T (IR) n=11	Mdm2SNP309G/G (IR) n=46
Lymphoma	4	17
Sarcoma NOS		2
Rhabdomyosarcoma		1
Hemangiosarcoma		3
Osteosarcoma	2	25
Medulloblastoma		1
Glioblastoma		1
Neuroblastoma		1
Adenoma	2	3
Alveolar carcinoma		2
Hepatocellular carcinoma	3	4
Mammary carcinoma		2
Renal tubule		1
Squamous cell carcinoma		1
Histiocytic sarcoma	1	4
Germ cell tumor	1	2
Total tumors	13	70

Abbreviations: Gy, Gray; IR, ionizing radiation; NOS, not otherwise specified

Mdm2^SNP309T/T mice exhibit increase risk of skin squamous cell carcinoma after exposure to UVB compared to Mdm2^SNP309G/G mice

Ultraviolet (UV) radiation is the major environmental cause of skin cancer. Exposure to UVB radiation induces intracellular reactive oxygen species, which elicit a robust p53 response (39, 40) and leads to skin cancer development (41). The FVB mouse strain is susceptible to insult-induced skin tumorigenesis; hence, this model is a powerful tool in studying progressive squamous cell carcinoma (42). To test if the SNP309G allele cooperates with UVB exposure to alter p53 activity in skin, we first backcrossed SNP309 mice to the FVB background for five generations. Isolated primary keratinocytes from Mdm2^SNP309G/G and Mdm2^SNP309T/T mouse epidermis were treated them with a single dose of UVB (100mJ/cm²), a dose previously shown to activate acute p53 response in skin tissue (43), and analyzed for p53 activity. In contrast to studies in the spleen [Figure 1A], keratinocytes isolated from Mdm2^SNP309T/T mice showed greater than two-fold higher Mdm2 levels (p=0.001) 24 hours post UVB treatment compared to Mdm2^SNP309G/G keratinocytes [Figure 3A]. To further examine p53 activity in skin tissue, we treated 6-week old Mdm2^SNP309G/G and Mdm2^SNP309T/T mice with UVB and analyzed p21 expression. Consistent with the elevated Mdm2 expression levels in keratinocytes from Mdm2^SNP309T/T compared to Mdm2^SNP309G/G mice after UVB treatment, Mdm2^SNP309G/G keratinocytes displayed significantly higher number of cells staining positive for p21 (p=0.019) compared to Mdm2^SNP309T/T keratinocytes [Figure 3B]. To examine skin cancer risk after UVB exposure, we generated a cohort of Mdm2^SNP309G/G and Mdm2^SNP309T/T mice in a FVB background and treated with an established UVB regimen (UVB radiation three times per week for 30 weeks) (43) and monitored for spontaneous tumor development for squamous cell carcinoma (SCC). Tumor-free survival analysis of UVB treated mice also showed that Mdm2^SNP309T/T mice died significantly sooner than Mdm2^SNP309G/G mice (p<0.001) [Figure 3C] with a median survival of 28 weeks compared to with 36 weeks. Thus, UV treatment led to a dampened p53 response and shorter tumor latency in Mdm2^SNP309T/T mice compared to Mdm2^SNP309G/G mice, in contrast to IR treatment. These data suggest that either the type of DNA damage or the cell type account for the differences in tumorigenesis. We therefore repeated the experiment using fibroblasts, a cell type that previously displayed higher basal Mdm2 levels in SNP309G than in SNP309T genotype (30). Fibroblasts were treated with the same regimen of UVB and the results showed 1.9-fold higher (p=0.032) increase in Mdm2 expression 24 hours post UVB treatment in SNP309G compared to SNP309T fibroblasts [Figure 3D], suggesting that the contrasting differences were cell-type dependent.

Figure 3.

Tumor development after UVB treatment. A, Mdm2 expression in primary keratinocytes of Mdm2^SNP309T/T (T/T) and Mdm2^SNP309G/G (G/G) mice 24hrs post UVB treatment (100 mJ/cm2) (mean ± SD; n= 3). B, Immunohistochemistry analysis of p21 in skin tissue samples from 6-week old Mdm2^SNP309T/T and Mdm2^SNP309G/G mice (n=5) 24hrs post UVB treatment. Statistical analysis was performed using Student’s t-test. C, Kaplan-Meier survival curve of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice after UVB treatment (Log-rank test). D, Mdm2 mRNA expression in primary adult fibroblasts of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice 24hrs post UVB treatment (mean ± SD; n= 4). Expression levels are normalized to that of untreated samples. Statistical analyses were performed using ANOVA test. n= number of biological replicates.

Benzo(a)pyrene (B(a)P) is a tobacco-related carcinogen that causes DNA damage (44). To determine if the differences in skin tumor onset are limited to UVB insult, we next investigated the effects of B(a)P on p53 activity in the skin of our mice. The epidermis of Mdm2^SNP309G/G and Mdm2^SNP309T/T mice was treated with either acetone (control) or B(a)P and total RNA was isolated 24-hour post treatment. Mdm2 mRNA levels were higher in Mdm2^SNP309T/T mice compared to Mdm2^SNP309G/G mice after B(a)P treatment, however the differences were not statistically significant [S1]. To examine tumor risk following B(a)P exposure, we treated Mdm2^SNP309G/G and Mdm2^SNP309T/T mice to an established regimen of 317 nmol of B(a)P once a week for 30 weeks (44) and monitored mice for spontaneous tumor formation. Tumor-free survival showed that B(a)P treated Mdm2^SNP309T/T mice were more susceptible to squamous cell carcinoma than Mdm2^SNP309G/G mice (p<0.001) [Figure 4A] with a median survival of 22 weeks compared to 25 weeks, respectively. Further assessment of the tumor-free survival data showed that the difference was primarily due to gender [Figure 4B]. Tumor development in male Mdm2^SNP309G/G mice was significantly delayed (p<0.001) compared to the other groups. We therefore revisited the Mdm2 expression levels in epidermis of Mdm2^SNP309G/G and Mdm2^SNP309T/T male mice. Similarly, Mdm2 mRNA levels were significantly lower (p=0.030) in the epidermis of male Mdm2^SNP309G/G mice compared to male Mdm2^SNP309T/T mice 24-hour post B(a)P treatment, but there were no significant differences between female mice [Figure 4C, 4D]. Moreover, epidermal keratinocytes in skin samples from male Mdm2^SNP309G/G mice displayed significantly higher number of cells staining positive for p21 (p=0.049) as compared to Mdm2^SNP309T/T keratinocytes after B(a)P treatment [Figure 4E]. Altogether, the data show that keratinocytes from Mdm2^SNP309G/G mice exhibit significantly lower Mdm2 expression and increased p53 activity in Mdm2^SNP309G/G mice compared to Mdm2^SNP309T/T mice in response to insult. Thus, Mdm2^SNP309G/G mice, particularly males, are more resistant to developing skin SCC in response to B(a)P treatment.

Figure 4.

Tumor phenotype after B(a)P treatment. A, Kaplan-Meier survival curves of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice after B(a)P treatment. B, Kaplan-Meier survival curves of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice separated by gender. Statistical analyses for Kaplan-Meier survival curves were performed by Log-rank test. C, D, Mdm2 expression in epidermis of male and female Mdm2^SNP309T/T and Mdm2^SNP309G/G mice 24hrs post acetone or B(a)P treatment. Data were normalized to expression levels of Mdm2 in the epidermis tissue samples of acetone treated controls, (n = 3, mean ±SD). Statistical analysis was performed using Two-way ANOVA. E, Immunohistochemistry of skin tissues of 6-week old male Mdm2^SNP309T/T and Mdm2^SNP309G/G mice with p21 antibody, 24hrs post B(a)P treatment (Student’s t-test, n=3). n= number of biological replicates.

The SNP309G allele regulates Mdm2 expression in a tissue-dependent manner

To address the contrasting differences in vulnerability to cancer risk after exposure to environmental factors, we evaluated Mdm2 basal expression in different tissues harvested from Mdm2^SNP309G/G and Mdm2^SNP309T/T mice in a C57BL/6 background. Mdm2 basal levels vary in different tissues in a SNP-dependent manner [Figure 5A]. Mdm2 mRNA levels in the thymus (the origin of T-cell lymphomas) of Mdm2^SNP309G/G mice are statistically higher (p=0.040) than in Mdm2^SNP309T/T mice. Conversely, keratinocytes, the cell of origin for SCC, in Mdm2^SNP309T/T mice exhibit significantly higher Mdm2 mRNA levels (p<0.001) than keratinocytes in Mdm2^SNP309G/G mice. One other tissue showed significant differences in Mdm2 levels; the heart tissue exhibits significantly higher Mdm2 mRNA levels (p=0.006) in Mdm2^SNP309G/G mice than in Mdm2^SNP309T/T mice [Figure 5A]. Further investigation showed that other tissues exhibited significant differences in Mdm2 basal levels when evaluated by gender [S2]. To examine if mouse strain contributed to the contrasting differences, we examined Mdm2 basal expression in fibroblasts and keratinocytes from C57BL/6 and FVB mice. While basal Mdm2 levels vary somewhat between strains, keratinocytes from Mdm2^SNP309T/T mice have more Mdm2 than Mdm2^SNP309G/G keratinocytes and fibroblasts from Mdm2^SNP309G/G mice have more Mdm2 than Mdm2^SNP309T/T fibroblasts in both strains [S3]. Thus, these Mdm2 basal expression differences are independent of strain.

Figure 5.

SNP309 modulates Mdm2 expression in a tissue-dependent manner. A, Mdm2 expression in different tissues of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice. Statistical analyses were performed using Student’s t-test (mean ±SD, n=8; 4♂, 4♀). In this experiment, RplpO expression was used as a reference gene. B, Predicted SP1 & E2F binding sites in the MDM2 P2 promoter proximal to SNP309 site. C, Expression of Mdm2 and Fgf21 (a known SP1 target) in MEFs transfected with SP1. D, Mdm2 and Brca1 (a known E2F6 target) expression in MEFs transfected with E2F6. Statistical analysis was performed using Two-way ANOVA. E, Western blot analysis of SP1 and E2F6 in thymi and keratinocytes of SNP309 mice. GAPDH was used as loading control. F, E2F6 ChIP analysis of the MDM2 P2 promoter in keratinocytes of Mdm2^SNP309T/T and Mdm2^SNP309G/G mice. E2F6 ChIP assays for the Brca1 promoter in keratinocytes was used as positive control. A sequence downstream of the MDM2 promoter not thought to bind E2F6 was used as a negative control. Statistical analysis was performed using Two-way ANOVA. n= number of biological replicates.

To investigate the mechanism by which the SNP309 alleles regulate Mdm2 expression in a tissue specific manner, we first interrogated potential transcription factor binding to the SNP309 allele. Analyses of the ENCODE ChIP-seq database (45) in genome browser (https://genome.ucsc.edu) showed that transcriptional repressors E2F4 and E2F6 were candidates because of their binding proximity to SNP309 [Figure 5B, S4]. Additionally, previous studies associated these repressors as key regulators of keratinocyte proliferation (46). The cell lines utilized in the ENCODE ChIP-seq experiments for E2F4 and E2F6 were MCF10A and K562, respectively. Sequence analysis of these cell lines showed that MCF10A cells harbor a homozygous SNP309T genotype and K562 cells harbor a SNP309T/G genotype [S5]. To examine if E2F4 and E2F6 regulate Mdm2 expression in a SNP309 allele dependent manner, we generated MEFs from Mdm2^SNP309G/G and Mdm2^SNP309T/T C57BL/6 mice and transfected them with plasmids expressing GFP and either E2F4, E2F6, or SP1, a transcriptional activator that increases Mdm2 expression in SNP309G cells. As expected, sorted cells expressing GFP and SP1 preferentially increased Mdm2 expression in SNP309G MEFs (p<0.001) [Figure 5C]. E2F4 transfection increased Mdm2 expression slightly, but there were no significant differences between genotypes [S6]. E2F6 transfection statistically repressed Mdm2 levels in SNP309G MEFs compared to SNP309T MEFs (p<0.001) [Figure 5D]. To test the efficiency of transfections, we performed RT-qPCR on Fgf21 and Brca1, known targets of SP1 and E2F6, respectively. No significant differences in expression of Fgf21 and Brca1 were observed in SNP309T MEFs compared to SNP309G MEFs, indicating similar transfection efficiencies. We also examined protein levels of SP1 and E2F6 in thymus and keratinocytes harvested from Mdm2^SNP309G/G and Mdm2^SNP309T/T mice and showed that SP1 is expressed in thymus and not detectable in keratinocytes and vice versa E2F6 is clearly expressed in keratinocytes but not in thymus samples [Figure 5E]. These differences were independent of genotype. We next examined the binding specificity of E2F6 to the MDM2-P2 promoter in keratinocytes harvested from homozygous SNP309G and SNP309T newborn pups by performing chromatin immunoprecipitation assays (ChIP). ChIP assays using an E2F6 antibody showed that E2F6 bound more significantly (p<0.001) to the MDM2-P2 promoter in keratinocytes of SNP309G compared to SNP309T [Figure 5F]. We also performed ChIP assays on the Brca1 promoter, an establish target of E2F6, and saw no difference in E2F6 binding to SNP309G or SNP309T samples. In addition, a sequence downstream of the Mdm2 promoter not thought to bind E2F6 showed no difference in E2F6 binding. Collectively, these data show that SNP309G regulates Mdm2 basal expression in a tissue-dependent manner utilizing alternative transcription factors.

DISCUSSION

Functional polymorphisms alter gene function, and disease risk. Environmental stress can have a synergistic effect on some SNPs and exacerbate disease; however cooperative mechanisms between functional polymorphisms and environmental stress have not been well defined. In this study, we utilize a well-known functional polymorphism involved in the p53 pathway, SNP309, to investigate gene-environment interactions in modulating tumor susceptibility. This study used relevant environmental stress models that affect the p53 pathway: humans are consistently exposed to low dose IR via cosmic rays and medical imaging procedures (47), UV from the sun, and the B(a)P carcinogen from smoking (48).

Our study shows that the SNP309G allele further increased Mdm2 expression and dampened p53 activity in spleen and thymus post IR treatment compared to SNP309T allele and compared to non-irradiated mice, resulting in exacerbated tumor risk in Mdm2^SNP309G/G mice compared to Mdm2^SNP309T/T mice. These results are consistent with previous data, which show that the Mdm2^SNP309G/G mice have increased susceptibility to tumors compared to Mdm2^SNP309T/T mice, affirming that SNP309G is an ‘at risk’ allele for spontaneous tumor formation (30).

Conversely, both UVB and B(a)P skin carcinogenesis models in a FVB background demonstrated the opposite effect and highlight increased Mdm2 expression and attenuated p53 activity is observed in the epidermal skin cells of Mdm2^SNP309T/T mice compared to Mdm2^SNP309G/G mice. Consequently, Mdm2^SNP309T/T mice exhibit increased risk of skin squamous cell carcinoma compared to Mdm2^SNP309G/G mice after UVB or B(a)P treatment. In this study, one irradiated Mdm2^SNP309G/G C57BL/6 mouse presented with a squamous cell carcinoma, as well as histiocytic sarcoma and lymphoma; this squamous cell carcinoma represents only ~1% of all tumors present in irradiated Mdm2^SNP309G/G C57BL/6 mice. In comparison, a larger number of UVB and B(a)P treated mice showed both genotypes develop squamous cell carcinoma but Mdm2^SNP309T/T mice were more susceptible.

A further examination of basal Mdm2 expression from different tissues harvested from SNP309 mice in a C57BL/6 background confirms that SNP309 regulates Mdm2 basal expression in a tissue-dependent manner. The contrasting difference in basal Mdm2 expression from skin keratinocytes was observed in C57BL/6 and FVB backgrounds; thus, the skin tissue phenotype is not strain-dependent. Moreover, fibroblasts from FVB mice showed increased Mdm2 expression in SNP309G compared SNP309T fibroblast after UVB treatment, indicating that the phenotype is also not insult-dependent. Collectively, these data show that SNP309 contrastingly regulates basal Mdm2 expression levels and concomitantly affects p53 activity and cancer risk in a cell-type-specific manner.

To date, epidemiological studies do not associate MDM2 SNP309 with skin squamous cell carcinoma risk (49, 50). However these study cohorts are modest. Many meta-analysis associating MDM2 SNP309 to cancer risk are conflicting (25, 51), even in larger studies suggesting that other factors can contribute to tumor risk. We observed differences in expression when we stratified by gender, thus, supporting the hypothesis that other factors cooperate to alter gene expression in a SNP-dependent manner in different tissues. Moreover, a recent study demonstrated a decreased risk of developing oral squamous cell carcinoma (OSCC) for the MDM2 SNP309G group (52) with pronounced susceptibility in the MDM2 SNP309T group, suggesting MDM2 SNP309G may be protective in a cell-dependent manner.

Finally, our data show that Mdm2 expression is controlled by the SNP309G allele in a tissue-dependent manner. The SP1 overexpression experiment supports previous mechanistic studies, both in human and mouse tissues, showing SP1 preferentially binds to SNP309G allele to increase Mdm2 expression (26, 30). In contrast, the E2F6 overexpression experiments show preferential repression of Mdm2 expression in the SNP309G allele, suggesting that this allele will elicit a more potent p53 response and protect mice harboring this SNP from tumor risk. Protein expression of thymi and keratinocytes from SNP309 mice show a contrasting difference, where E2F6 expression is more pronounced in keratinocyte, thus explaining why E2F6 protein may be more active in this tissue type. The ChIP-seq data from ENCODE confirmed that E2F6 binds to MDM2-P2 promoter and our ChIP assay confirmed that E2F6 preferentially binds to the MDM2-P2 promoter in keratinocytes of SNP309G compared to SNP309T. Thus the strength of this study is that we define a specific regulatory mechanism that directly impacts the Mdm2 SNP309G allele resulting in an increased cancer risk in a tissue-dependent manner through discrete mechanisms.

In summary, our data illustrates how SNP309G cooperates with environmental stress to exacerbate or protect cancer development in a tissue-dependent manner. Our findings suggest that functional polymorphisms can be multidimensional gene regulators with varying disease risk.

MATERIALS AND METHODS

Mouse Models

Mouse experiments were conducted in compliance with MD Anderson Cancer Center Institutional Animal Care and Use Committee (IACUC). The Mdm2^SNP309G/G and Mdm2^SNP309T/T mice in C57BL/6 strain have been previously characterized (30). To generate a skin tumor prone FVB strain, the C57BL/6 strains were backcrossed for five generations to FVB.

Environmental stress experiments

Mdm2^SNP309G/G and Mdm2^SNP309T/T mice in C57BL/6 strain were treated with 1Gy IR 2 days after birth and monitored for tumor incidence and sacrificed upon tumor formation. Six-week old Mdm2^SNP309G/G and Mdm2^SNP309T/T mice in FVB strain were treated with skin carcinogenesis protocols, including B[a]P and UVB, as previously described (43, 53). Non-treated Mdm2^SNP309G/G and Mdm2^SNP309T/T mice were used as controls. The mice were monitored daily for tumor incidence and sacrificed upon tumor formation. Cohorts of C57BL/6 and FVB strain contain male and female mice. Number of animals used for each experiments are present in figures and figure legends.

Real-Time RT-PCR

Fresh tissues were necropsied from mice and flash frozen. RNA isolation and Real-Time RT-PCR was performed as previously described (54). RT-PCR primers for E2f4 (forward 5’- GATCGCTGACAAGCTGATTG -3’ and reverse 5’- CTGTTCTGGACGTCCTCAGTG -3’), E2f6 (forward 5’- CGGAAGAGGCGAGTGTATG -3’ and reverse 5’- CAGTTCGATGCCATCCAAG -3’), Fgf21 (forward 5’- GGGTCTACCAAGCATACCCC -3’ and reverse 5’- GTACCTCTGCCGGACTTGAC -3’), Brca1 (forward 5’- GCTGCAGCCACGCTTTTC and reverse 5’-GGCGAAGAACGAGAGAATGAA-3’), Sp1 (forward 5’-GCGAAGCATCTTGGGTGTGT-3’ and reverse 5’- TCCTTTCCTCTTCAGTCGCTTT-3). The RT-PCR primers for Mdm2, Bbc3, Cdkn1a (p21), and Rplp0 have been previously described (30). Expression was normalized to Rplp0. Experiments were repeated three times for each sample.

Cell culture and transfections

Primary keratinocytes and early passage MEFs from Mdm2^SNP309G/G and Mdm2^SNP309T/T mice were generated as previously described (21, 55). Early passage MEFs were transfected with 7 µg of pCMV-GFP vector containing E2F4 cDNA, E2F6 cDNA, SP1 cDNA, or empty vector using Lipofectamine 3000 reagent (Sigma, L3000015). Forty-eight hours later, GFP positive cells were sorted by FACS and RNA was extracted for RT-PCR analysis.

Chromatin immunoprecipitation assay (ChIP)

The ChIP and Real-time PCR assays were executed as previously described (54). The E2F6 antibody (Santa Cruz, Dallas, TX, USA, E20) (6ug) was added overnight at 4°C. RT-PCR primers for MDM2-P2 promoter (forward 5’- GATCGCTGACAAGCTGATTG -3’ and reverse 5’- CTGTTCTGGACGTCCTCAGTG -3’), Brca1 (forward 5’- GCAGCCGCAATTACAATCTATC -3’ and reverse 5’- ACGTGTCTGGATCTTGTGTTC -3’), Negative (forward 5’- ACCTGAGGTCAGGAGTACAA -3’ and reverse 5’- AGCTGGGATTATAGGCATGTG -3’). Data was presented and calculated as percentages of the total.

Protein analysis

Protein lysates were prepared by lysing thymi or keratinocytes from SNP309 mice in NP-40 buffer. Protein estimation was carried out with BCA (Pierce, Waltham, MA, USA #23225). One hundred micrograms of lysate was resolved on 10% SDS-PAGE and immunoblotted with antibodies against SP1 (1:200; Sata Cruz, sc-59), E2F6 (1:200; Santa Cruz #E-20), and GAPDH (1:1000; Abcam, Cambridge, UK #ab9485). Western blots were repeated at least three times with biological replicates.

Immunohistochemistry

Samples were stained for p21 (Dako, Santa Clara, CA, USA #SX118), caspase 3 (R&D Systems, Minneapolis, MN, USA #AF835) CD3 (Abcam #Ab5690), or B220 (BD Pharmigen, San Jose, CA, USA #530286). Digital images of stained sections were captured using the Aperio ScanScope (Aperio Technologies, Vista, CA, USA). Experiments were repeated at least three times with biological replicates.

Statistical analysis

All comparisons were analyzed using GraphPad Prism version 6.00 (GraphPad Software, LaJolla CA, USA). For statistical analysis, experiments were repeated with at least three biological replicates. Differences between two or more groups were analyzed by the Student’s t-test or analysis of variance followed by the Dunnett’s multiple comparison post-hoc test, respectively. A statistical analysis for tumor incidence and tumor multiplicity was performed by Fisher’s exact test. A statistical analysis for Kaplan-Meier survival curves was performed by Log-rank test. All statements of significant differences showed a 5% level of probability.