Epigenome modulated xenobiotic detoxification pathways control DMBA-induced breast cancer in agouti A^vy/a mice

ABSTRACT

Environmental xenobiotics with genotoxic activity are carcinogenic. However, individual differences in the susceptibility to xenobiotic-induced breast cancer remain unclear. Since epigenetic modifications could control the expression of metabolic enzymes, our goal was to determine whether epigenome modulated metabolic networks determine susceptibility to xenobiotic-induced breast cancer. The effect of epigenetic background on predisposition to carcinogen 7,12-dimethylbenz(a)anthracene (DMBA)-induced breast cancer development and progression was assessed using the A^vy/a mouse model. In a randomized block design, 22 isogenic A^vy/a (8 yellow, 7 slightly mottled, 7 pseudoagouti) and 8 wild type non-agouti (a/a black) age matched female mice were subjected to DMBA (30 mg/kg per mouse weight) once a week for 6 weeks to induce breast cancer. Compared to pseudoagouti littermates, a significant decrease in tumour latency with increased tumour burden was observed in slightly mottled and yellow littermates (p ≤ 0.05). However, tumour latency and tumour burden were similar in non-agouti a/a mice and A^vy/a cohorts. Network analysis of differentially expressed liver genes identified altered metabolic gene networks among agouti phenotypes. Consequently, in HPLC analyses, DMBA metabolites were significantly increased in A^vy/a pseudoagouti mice (p ≤ 0.05). Relative to A^vy/a slightly mottled, A^vy/a yellow and non-agouti a/a black mice, DMBA metabolites increased nine-, eight-, and four-fold, respectively, in A^vy/a pseudoagouti mice. In agreement with this, seven phase 2 xenobiotic detoxification genes were significantly upregulated in A^vy/a pseudoagouti mice (p ≤ 0.05). The Results from this study suggest that epigenome modulation of xenobiotic detoxification pathways may control xenobiotic-induced breast cancer susceptibility in A^vy/a mice.

Introduction

Lifestyle and prolonged exposure to xenobiotic compounds such as pesticides, detergents, industrial chemicals, organic solvents, and polycyclic aromatic hydrocarbons (PAHs) can remodel the epigenome and alter gene expression to increase disease susceptibility [1]. PAHs are found in cigarette smoke, diet, and various industrial chemicals that pollute air and water [2,3]. Metabolic intermediates of PAHs are mutagenic and form DNA adducts to induce cancers [2], and hence exposure to PAHs is of particular concern. Hormonally active xenobiotic compounds are estrogenic and may induce the transcription of oestrogen target genes to promote breast cancer development and progression [4]. While the role of gene mutations in breast tumorigenesis and progression has been well established [5], the effects of the epigenome on predisposition to xenobiotic-induced breast cancer remain unclear.

The A^vy mouse model is commonly used to assess metabolic impact of the epigenome [6]. In the A^vy mouse model, the differential DNA methylation of the agouti gene promoter regulates coat colour and obesity among isogenic littermates [7]. Transient expression of agouti protein or agouti signalling peptide (ASP) promotes a spectrum of coat colour phenotypes in isogenic A^vy/a litters [6]. Upon agouti gene expression, hair shafts will contain both yellow and black bands, resulting in a brown-coated mouse [6]. In A^vy mice, the transposable element Intracisternal A Particle (IAP) carrying a promoter element is inserted in front of the agouti gene. Differential methylation of the IAP results in varied levels of ASP expression and produces a spectrum of coat colours among A^vy littermates [8]. Hypomethylation of IAP loci results in ASP expression to produce pheomelanin resulting in yellow coat colour. However, IAP hypermethylation suppresses ASP expression, resulting in eumelanin production and black coat colour, while varying methylation levels of IAP loci result in intermediary coat colours [9]. Since coat colour variations of isogenic A^vy littermates reflect differential DNA methylation of the IAP locus, we employed the A^vy mouse model to investigate the impact of the epigenome in predisposal to xenobiotic-induced cancer.

Oral exposure to the PAH 7, 12-dimethylbenz[a]anthracene (DMBA) was shown to produce mammary tumours in rodents [10]. Biotransformation of DMBA to its active metabolites occurs by action of the Cytochrome P450 1 (CYP1) family of enzymes and the resulting DMBA-3,4-diol-1,2-epoxide (DMBADE) was found to be the most carcinogenic [11]. DMBA metabolites were suggested to form covalent adducts with DNA, which depurinate DNA resulting in adenine to thymine (A-to-T) conversion [12]. As multiple pathways activate DMBA, its detoxification also involves several steps [13]. Therefore, induction of DMBA-metabolizing and detoxifying genes should have opposite effects on the carcinogenic effects of DMBA.

Since epigenetic effects are manifested through gene expression, we hypothesized that epigenome modulated metabolic networks determine predisposition to xenobiotic-induced breast cancer development and progression. To test this hypothesis, the effects of epigenetic variations on predisposition to DMBA-induced breast cancer development and progression were investigated in the A^vy mouse model. Our results identified that variations in DMBA-induced breast cancer susceptibility in isogenic A^vy/a littermates are due to differential expression of genes that regulate xenobiotic detoxification pathways.

Results

Epigenome-dependant weight gain in A^vy/a agouti mice is not affected by DMBA

In A^vy/a mice, hypomethylation of the agouti gene promoter predisposes them to obesity [14] that was in turn implicated in breast tumorigenesis [7,15,16]. Therefore, the effect of DMBA on weight gain among A^vy/a littermates and a/a mice was compared during the entire course of the study (52weeks). In a randomized block design, age matched pseudoagouti (A^vyPS), slightly mottled (A^vySM), yellow (A^vyYell), and non-agouti (a/aBlk) mice were treated with DMBA and percent weight gain was assessed. As reported previously [7], differences in weight among littermates were associated with variations in coat colour in agouti mice (p < 0.0001, Figure 1(a and b). In multiple comparison tests, weight differences between non-agouti a/a Blk (26.4 ± 1.6 g) and pseudoagouti (36.1 ± 4.0 g) littermates were insignificant. However, a significant weight difference was observed between non-agouti a/a Blk (26.4 ± 1.6 g) and slightly mottled littermates (46.4 ± 1.7 g; p < 0.0001) as well as between non-agouti and A^vy yellow littermates (44.7 ± 2.6 g; p = 0.0001). During the course of DMBA treatment, a 70.7 ± 44% weight increase was observed in each mouse, except in two yellow mice that lost weight (Figure 1(c)). Together, these results suggested that the weight gain among littermates was unaffected by DMBA.

Figure 1.

Association between weight gain and coat colour in A^vy/a mice. (a) F1 generation isogenic A^vy mice display varying coat colours. From left to right: Isogenic littermates at 8 weeks old; wild type non-agouti (a/a) mouse (black), A^vy/a pseudoagouti, slightly mottled and yellow mice. (b) Weight gain in A^vy/a mice is associated with coat colour. Average weight among littermates was significantly different (p < 0.0001). (c) DMBA treatment had no effect on epigenome-directed weight gain in A^vy/a agouti mice. Mice were weighed weekly from the day of the first treatment with DMBA (at 8 weeks of age) up to the day they were euthanized. Values are mean ± SEM of weight gain in mice and a p value of 0.05 was considered significant in ANOVA analysis.

Epigenome determines DMBA-induced tumour latency and survival rates in A^vy/a mice

Next, the effect of epigenetic background on tumour latency (number of days from the first day of treatment with DMBA to the first day on which any palpable tumours were detected) was determined. One-way ANOVA analyses identified that tumour latency was significantly different among agouti phenotypes (p = 0.0133, Figure 2(a)). The average tumour latency in slightly mottled and yellow littermates was 155.6 ± 13.3 days and 151.6 ± 10.6 days, respectively. Tumour latency in A^vy pseudoagouti littermates was significantly increased to 231.8 ± 24.0 days (p ≤ 0.05, Figure 2(a and b)). The non-agouti (a/aBlk) littermates had a tumour latency of 184.6 ± 19.8 days which was similar to that of A^vy/a littermates. During the study period, three mice (an a/aBlk, an A^vyPS, and an A^vySM) exhibited a deterioration in health and were euthanized before any palpable tumours had formed.

Figure 2.

Tumour latency and overall survival were significantly increased in pseudoagouti littermates. (a) Palpable tumours developed at a significantly slower rate in A^vyPs than their A^vy/a littermates. One way ANOVA analyses yielded significant differences in tumour latency/time to tumour among littermates (p = 0.013). (c) Overall Survival rates were significantly higher in A^vy pseudoagouti littermates. Kaplan–Meier survival plots of littermates subjected to carcinogen DMBA. Comparison using a log-rank (mantel cox) test indicated a significance in percentage survival among littermates (p = 0.0374). Pairwise analyses using log-rank (mantel cox) tests identified significantly increased survival in pseudoagouti littermates compared to slightly mottled and yellow littermates. Survival in non-agouti mice (275.1 ± 26.8 days) was similar compared to A^vy/a littermates. Each data point on graphs is mean ± SEM days and a p value of 0.05 was considered significant.

Next, the effect of epigenetic background on cancer-associated survival rates among DMBA-treated mice was assessed using Kaplan–Meier survival analyses. In log-rank (mantel cox) tests, survival rates among DMBA-treated cohorts were significantly different (p = 0.0168). Among Avy/a cohorts, A^vyYell cohorts had the shortest survival rate and A^vy/a pseudoagouti cohorts survived the longest (Figure 2(b)). The pseudoagouti cohorts survived significantly longer (328.1 ± 19 days) than slightly mottled (279.7 ± 12.8 days) and yellow (238.6 ± 26.8 days) littermates (p < 0.05). The non-agouti mice had similar survival rate (275.1 ± 26.8 days) to that of A^vy/a cohorts. Moreover, weekly weight measurement further suggested lack of correlation between the body weight and cancer associated survival (Supplemental Figure 1). Taken together, these results indicate that epigenetic background is a major determinant of overall survival among mice littermates with DMBA-induced cancer.

DMBA-induced tumour burden was significantly decreased in A^vy/a pseudoagouti mice

Since the cumulative incidence of tumours (tumour burden) could determine overall survival, the total number of palpable or visible tumours from the beginning of DMBA administration until the death of mice were enumerated. Palpable tumours were categorized as either skin or mammary tumours. The majority of mammary tumours appeared on the thoracic and abdominal mammary glands (data not shown). Total tumour burden and mammary tumour burden were significantly lower in A^vy pseudoagouti mice relative to A^vy/a slightly mottled, A^vy/a yellow, and a/a black cohorts (p < 0.05, Figure 3(a–c)). While 42.9% of A^vy PS cohorts developed at least one mammary tumour each, 85.8% of A^vy SM and 87.5% A^vy Yell cohorts developed one or more palpable tumours (Figure 3(a)). Moreover, the majority of A^vySM (7/8) and A^vyYell cohorts (5/8) had two or more palpable mammary tumours. The a/a Blk cohorts had a similar number of mammary tumours as that of A^vy/a littermates. Compared to mammary tumour incidence, the incidence of skin tumours was similar among A^vy cohorts (Figure 3(b), p = 0.1042). Histopathological analyses confirmed the induction of mammary tumours by DMBA (Figure 3(d)).

Figure 3.

DMBA-induced tumour incidence among A^vy/a littermates. DMBA-induced tumour incidence was significantly decreased in A^vy/a pseudoagouti mice. Number of palpable tumours in (a) mammary and (b) skin were enumerated and (c) combined number of mammary and skin tumours. One-way ANOVA analyses and Post hoc Tukey’s multiple comparison tests indicated that average tumour burden per mouse was significantly decreased among pseudoagouti littermates compared to slightly mottled littermates and yellow littermates. Each point on the graph is mean ± SEM of number of tumours. (d) DMBA treatment produced mammary tumours in A^vy mice. Representative mammary tissue sections from DMBA-treated tumour bearing mice were prepared to confirm the presence of mammary cancer.

Xenobiotic metabolism genes were upregulated in A^vy/a pseudoagouti mice

To determine whether differential DNA methylation among agouti phenotypes affects expression of genes involved in xenobiotic metabolism or just agouti signalling, liver gene expression profiles among A^vy/a mice were compared by analysing publicly available expression array data GSE28559 [17]. Hierarchical clustering identified that compared to pseudoagouti (A^vyPS), expression of 235 liver genes was altered 2-fold or more in mottled (A^vyM), 184 genes in heavily mottled (A^vyHM), 127 genes in slightly mottled (A^vySM) and 28 genes in yellow mice (A^vyYell) (Figure 4(a and b). Among 400 differentially expressed genes, 9 of the genes were common among agouti phenotypes (Figure 4(b)). The A^vyM vs A^vyPS had the highest number of unique genes (112 genes) followed by the A^vySM vs A^vyPS set with 83 unique genes (Figure 4(b)).

Figure 4.

Xenobiotic metabolism genes were upregulated in pseudoagouti liver extracts. (a and b) Phase 2 xenobiotic metabolism genes were upregulated >2 fold in A^vy/a pseudoagouti liver extracts. (a) Expression array data GSE28559 was reanalysed for liver gene expression profiles among agouti phenotypes. Compared to A^vyPS, 235 liver genes were altered 2-fold or more in mottled A^vyM, 184 genes in A^vyHM, 127 genes in A^vySM and 28 genes in A^vy Yell. (b) Venn diagram analysis to identify unique and common liver gene expression among A^vy littermates. (c) Xenobiotic metabolism genes were enriched in A^vy/a mice. Differentially expressed genes were subjected to network analysis by using Altanalyze software. Xenobiotic metabolism, drug metabolism, and pathways in cancer were enriched among agouti phenotypes. (d and e) Xenobiotic metabolism networks were enriched in A^vyPS. Compared to A^vySM and A^vyHM, xenobiotic metabolism networks involving Cyp genes were enriched in A^vyPS.

To identify pathways enriched in each of the agouti cohorts, differentially expressed genes were subjected to network analysis using Altanalyze software [18]. In agouti phenotypes, xenobiotic drug metabolism and detoxification pathways, and cancer pathways were enriched (Figure 4(c)). Compared to A^vySM and A^vyHM cohorts, xenobiotic detoxification networks involving Cyp genes were enriched in A^vyPS livers (Figure 4(d and e)). Since expression of genes involved in xenobiotic metabolism were altered in the liver, expression of genes involved in DMBA metabolism in breast tissues was investigated. Initial qRT-PCR screening of commonly used housekeeping genes; namely B-glucuronidase (Gus), Tubulin (Tubb4a), B actin (ActB) and Hydroxysteroid Sulfotransferase (Sult1B1) identified that Sult1B1 expression was consistent among A^vy tissue samples. Therefore, Sult1B1 was used as the endogenous control to normalize gene expression analyses. The expression of 7 genes (Gstp1, Gsta3, Gcnt1, Sult3a1, Uggt1, Ugcg, and EPHX1) involved in phase 2 xenobiotic detoxification were upregulated two-fold or more in A^vyPS mice compared to A^vy Yell and A^vySM littermates, respectively (Table 1). The genes Gstp1 and Uggt1, which detoxify DMBA-3,4-epoxide and DMBA-3,4-diol were upregulated more than 100-fold in pseudoagouti mice compared to yellow mice (Table 1). Additionally, expression of key genes involved in phase 1 xenobiotic metabolism and ASP was found to be unaltered among agouti phenotypes (Table 1).

Table 1.

Expression of xenobiotic metabolic enzymes in mammary tissue extracts of DMBA-treated tumour bearing mice (relative to pseudoagouti). Phase 2 xenobiotic detoxification genes were upregulated >2 fold in pseudoagouti mice compared to A^vy/a littermates. Values are mean ± SEM of representative data obtained from four independent sets of mice.

Gene	Phase	Avy Yell	Avy SM	a/a Blk	p value
AhR	Phase 1	3.0 ± 3.0	3.0 ± 1.6	1.9 ± 0.8	p = 0.687
CYP1A1		2.4± 1.5	1.6 ± 1.1	0.5 ± 0.4	p = 0.525
CYP1B1		1.3 ± 1.0	1.4 ± 1.3	1.1 ± 0.5	p = 0.982
Akr1c22		0.8 ± 0.4	0.7 ± 0.3	1.1 ± 0.5	p = 0.88
ASP	N/A	0.9 ± 0.3	1.6 ± 0.7	2.2 ± 0.8	p = 0.321
Gsta3	Phase 2	0.07 ± 0.04	0.14 ± 0.07	0.02 ± 0.009	p < 0.00001
Gstp1		0.002± 0.002	0.2 ± 0.05	0.5 ± 0.4	p = 0.034
Gcnt1		0.07 ± 0.009	0.5 ± 0.4	5.1 ± 5.0	p = 0.534
Sult3a1		0.06 ± 0.05	0.08 ± 0.07	0.2 ± 0.2	p = 0.00034
Uggt1		0.007 ± 0.004	0.12 ± 0.05	0.32 ± 0.27	p = 0.0034
Ugcg		0.09 ± 0.09	0.3 ± 0.3	0.3 ± 0.3	p = 0.099
Ephx1	Phase 1 or 2	0.1 ± 0.08	0.2 ± 0.2	0.7 ± 0.3	p = 0.031

DMBA biotransformation metabolic profiles were altered in A^vy/a mice

Since gene networks involved in xenobiotic metabolism were altered among agouti phenotypes, the effect of epigenetic background on DMBA-biotransformation was further investigated by assessing DMBA and its metabolites in A^vy/a littermates. DMBA and its metabolites were detected in liver extracts of DMBA-treated mice using reverse phase HPLC. Chromatogram of liver extracts identified differences in DMBA metabolic profiles among A^vy littermates at 24 h after oral administration of DMBA (Figure 5(a)). Two major DMBA metabolites present in the liver extracts had retention times (RT) of 2.8 min and 7.2 min, respectively (Figure 5(a)). Although the peak area of the DMBA metabolite with 2.8 min RT was similar among A^vy littermates, peak area with 7.2 min RT was significantly increased in A^vyPS mice compared to their A^vy and a/a littermates (p = 0.04, Figure 5(b)). This second peak area was increased 9-, 8-, and 4-fold respectively in A^vyPS mice compared to A^vySM, A^vy Yell, and a/aBlk mice.

Figure 5.

DMBA biotransformation metabolic profiles differ among A^vy/a littermates. (a) A^vy mice display different DMBA metabolic profiles. The representative HPLC chromatogram illustrates differences in DMBA metabolic profiles between a/a Blk, A^vyPS, A^vySM and A^vy Yell among littermates 24 h after oral exposure to DMBA. Major DMBA metabolite peaks were detected at a retention time (RT) of 2.8 and 7.2 mins. The chromatogram is a representative of two independent experiments. (b) A^vy/a pseudoagouti mice display increased conversion of DMBA to DMBA-diol. Each bar on the graph is a representative of Mean Area under the metabolite peaks located at RT of 7.2 min ± SEM. Results are representative of data obtained from 2 independent sets of mice and a p value of 0.05 was considered significant in ANOVA analysis. (c) Model to illustrate the role of upregulated Phase 2 xenobiotic detoxification genes to reduce mammary tumour development and progression in A^vy pseudoagouti mice. DMBA detoxification from A^vy pseudoagouti mouse mammary tissue occurs in three phases. Upon entry into the body, DMBA is metabolized to active metabolites mostly by action of the Cytochrome P450 1 (CYP1) family of enzymes. The initial steps in DMBA metabolism involve increased expression of the aryl hydrocarbon receptor (AhR) in the cytosol. Phase 1 metabolic enzymes mainly function in the oxidation, reduction, hydrolysis or hydroxylation of the parent compound to form activated reactive intermediates. Phase 2 metabolic enzymes mainly function in conjugating reactive intermediates to several different chemical groups that make the compounds more polar and more readily released from the body as waste products in urine. Finally in phase 3, conjugated xenobiotics are excreted out of the liver via exporting mechanisms including ABC transporter subfamily members, organic anion and cation transporters, and solute carriers.

Discussion

In this study, using the agouti mouse model and DMBA, the effects of epigenetic background on predisposition to xenobiotic-induced breast cancer was assessed. Weight differences and coat colour variation among isogenic A^vy/a mice are frequently associated with differential DNA methylation of the agouti locus [7]. As reported earlier, [18,19], isogenic A^vy/a littermates exhibited significant weight differences by 8 weeks of age (Figure 1(b)). In prior DMBA-induced breast cancer studies, the weight variation of A^vy/a littermates was not accounted for and the same amount of DMBA was administered to all cohorts [16]. To avoid a potential variability resulting in from weight differences, in this study, DMBA concentration was normalized to mouse weight. A^vy/a littermates irrespective of their epigenetic background received the same concentration of DMBA per body weight. To our knowledge, this is the first study that accounted for weight difference bias while comparing the influence of epigenetic background on breast cancer predisposal. Our results identified that epigenetic background plays a prominent role on DMBA-induced tumour latency and tumour burden and impacts overall survival of A^vy/a littermates. The A^vy/a pseudoagouti mice, with a hypermethylated agouti locus, were less susceptible to DMBA-induced breast cancer (p = 0.0133, Figure 2(a)). While pseudoagouti, yellow, and slightly mottled cohorts displayed similar percent of weight gain, pseudoagouti littermates had significantly lower tumour burden than that of slightly mottled (p = 0.032) and yellow (p = 0.0138) littermates (Figure 2(a)). Moreover, despite the non-agouti a/a littermates being significantly leaner than the A^vySM and A^vy Yell littermates (p < 0.0001), these cohorts had similar mammary tumour latency and burden (Figures 1(b) and 2(a)). These results suggest that obesity is not the primary determinant of DMBA-induced breast tumours in agouti mice.

The ASP is an agouti gene product and is a naturally occurring paracrine signalling factor that has been implicated in agouti-induced obesity [14,20,21]. The ASP mRNA was expressed at similar levels in mammary tissues of A^vy/a and a/aBlk littermates (Table 1). This further supports our hypothesis that ASP-induced obesity is not the primary cause of differential DMBA-induced breast cancer susceptibility observed among agouti phenotypes. Additionally, results of our gene expression studies suggest that besides coat colour and obesity determination, differential DNA methylation at IAP loci in A^vy/a mice has a pleiotropic effect and highlights the significance of xenobiotic metabolic pathways in DMBA-induced breast cancer predisposal. In agreement with these results, a Whole Genome Bisulfite Sequencing (WGBS) study identified methylation differences in A^vy mice outside of IAP long term repeat (LTR) [22]. Although our experimental results were all obtained from female mice, the differential xenobiotic metabolic network identified in this study using female mice is consistent with the gene expression data in male mice [17], suggesting that the differential expression of xenobiotic metabolism genes is not sex specific.

The initial steps in DMBA activation into carcinogenic metabolites (phase 1) involves the binding of DMBA to the aryl hydrocarbon receptor (AhR) [2,3]. The AhR-DMBA complex binds to xenobiotic response elements (XRE) and induces AhR response genes such as cytochrome p450 genes (CYPs) [2]. Cyp enzymes convert DMBA into toxic epoxides such as DMBA-3,4-epoxide and DMBA diols [12]. Phase 2 and 3 xenobiotic metabolism are associated with detoxification of DMBA metabolites. In phase 2, reactive xenobiotic intermediates are conjugated to polar chemical groups such as sulphates, glutathione, and glucuronides and are readily released from the body through urine in phase 3 of biotransformation [12]. In A^vySM and A^vy Yell cohorts gene networks involved in converting DMBA into toxic intermediates were upregulated (Figure 4). On the contrary, in A^vyPS mice, seven genes (Gstp1, Gsta3, Gcnt1, Sult3a1, Uggt1, Ugcg, and Ephx1) involved in phase 2 detoxification pathways were upregulated (Table 1). The epoxide hydrolase encoded by Ephx1 hydrolyses carcinogenic DMBA-3,4-epoxide into a less toxic DMBA-3,4-diol. Indeed, in the liver extracts of pseudoagouti mice, DMBA-3,4-diol was significantly increased (Figure 5(a)). Enzymes encoded by Sult3a1, Gsta3, and Gstp1, Gcnt1, Ugcg, and Uggt1 readily conjugate DMBA-3,4-diol to polar compounds such as sulphates, glutathione, and glucuronides for excretion [23–25]. Increased tumour latency and lower tumour burden of A^vyPS could be the net effect of underactive phase 1 with overactive phase 2 biotransformation pathways. In agreement with these results, phase 2 enzymes in mammary cells were implicated in drug resistance and survival in breast cancer [26–30]. Genetic polymorphism and decreased expression of phase 2 detoxifying enzyme UGT1A1 were associated with higher incidence of breast cancers [26,31,32]. Similarly, inhibiting CYP1A1 and CYP1B1 activities, which biotransform DMBA into its active metabolite, protected the MCF-7 genome [28]. However, overexpression studies in MCF-7 cells suggested that UGT1A1 but not CYP1A1 has a protective effect against DMBA [27]. Conforming to our in vivo results, in cell line studies, methylation status of the epigenome was shown to impact the metabolism of xenobiotic compounds including DMBA by regulating expression of both phase 1 and phase 2 enzymes such as CYP and GSTP1, respectively [29,30]

In summary, ASP-mediated obesity factors are suggested to cause breast cancer development and progression in A^vy/a mice [7,15,16]. However, results of this study suggest that epigenetic modulation outside the agouti locus may control DMBA-induced breast cancer susceptibility in A^vy/a mice. This conclusion was reached with the following results. Firstly, ASP expression was similar in mammary tissue extracts obtained from all four DMBA-treated cohorts. Secondly, there was no difference in tumour burden or tumour latency between non-agouti mice and A^vy/a littermates despite a/a mice being significantly leaner than their A^vy/a yellow and A^vy/a slightly mottled littermates. Thirdly, phase 2 DMBA detoxification pathway genes were upregulated in A^vy/a pseudoagouti littermates resulting in higher levels of non-toxic, intermediate DMBA detoxification metabolites. Concurrent with our hypothesis, we conclude that the observed variations in breast cancer susceptibility in isogenic A^vy/a littermates is due to differential expression of genes that regulate xenobiotic metabolism.

Materials and methods

Maintenance of agouti A^vy/a mouse colony and study design

IACUC approved procedures and protocols were used to maintain, treat, handle, and collect data from mice (AUP# P16-006). Male (A^vy/a) mice were bred with female non-agouti (a/a) mice to obtain 22 isogenic A^vy/a mice(8 yellow, 7 slightly mottled, 7 pseudoagouti) and 8 non-agouti (a/a) mice. F1 generation females were then allocated to different groups using a randomized block design. Mice received ear tags for individual identification and were housed at a maximum of four mice per cage. Mice were kept in a temperature and humidity controlled environment and were provided with regular feed (Prolab RMH 1800;CAT # 5LL2) and water ad libitum. Mice were observed daily and weighed once in a week for calculatingchanges in individual mouse weight.

Breast cancer induction in mice

DMBA diluted in sesame oil (Sigma Inc) was administered to the mice to induce cancer. Six doses of DMBA (30 mg/kg mouse weight) by weekly syringe feeding method beginning at 8 weeks of age were administered to virgin female agouti mice (a/a and A^vy/a) [33,34]. Sixty milligrams of DMBA were diluted in 10 ml sesame oil to make a stock (6 mg/ml) which was aliquoted and frozen at −80 °C. The required dosage for each mouse was calculated before each new dose using the formula: [(weight of mouse in grams ÷ 20) x 100 µl]. The mice were examined and inspected everyday, but weighed and palpated once a week. Mice were sacrificed when the tumour diameter reached 2 cm or when they exhibited an inadequate physical condition.

Extraction and processing of mouse tissues

Mice were sacrificed using carbon dioxide followed by cervical dislocation, after which livers, mammary tissue, and tumours were located, excised and prepared for analysis as described [35]. Immediately after removal, tissues were cut into small sections and placed in vials containing 10% neutral buffered formalin to begin the fixing process. Tissues were left in formalin for no more than 24 h and then transferred to 70% ethanol and stored at 4ºC until they were ready for processing. Tissues were transferred to labelled tissue cassettes and placed into Leica ASP 300S tissue processor overnight. Tissues were subjected to standard dehydration and embedding in paraffin wax in preparation for H & E staining as described below.

Haematoxylin & eosin staining

After processing, the tissues were trimmed at 25 microns using a Leica RM 2125 RTS rotary manual microtome. Tissues were left on ice for up to 30 min to hydrate and then sectioned at 5 microns and placed on glass slides. Slides were dried at 56°C overnight. The following day, the tissues were deparaffinized three times for 5 min each in xylene. The tissues were rehydrated twice for 3 min each in 100% ethanol, twice for 3 min each in 95% ethanol and finally in 70% ethanol for 3 min. The tissues were rinsed in distilled water for 5 min before being stained with filtered haematoxylin for 6 min. After being stained with haematoxylin, the tissues were rinsed in running tap water for 20 min. Following the rinsing step, the tissues were decolourized in 0.25% acid alcohol for 1 s and immediately rinsed in distilled water for 5 min. The tissues were counter-stained with eosin for 15 s. After retaining the eosin stain, the tissues were rehydrated by placing in 70% ethanol for 3 min, then twice for 3 min each in 95% ethanol, and twice for 5 min each in xylene. The slides were mounted with cytoseal in the fume hood and viewed under a light microscope. Images were captured using an Amscope 3.0 Digital Camera, version x86, 3.1.312.

Histological analyses of tumours

After images were captured, tissues and tumours were classified using the IARC Scientific Publication no. 11 and scientific web journals dedicated to visual and surgical pathology [35,36].

Gene expression and gene network reanalysis

Hierarchical clustering (Euclidean distance) of liver gene expression array data from GSE28559 [17] was performed using Arraystar expression analysis software version 15.0.1 (DNASTAR Inc.). Differentially expressed genes were selected and gene network analyses were conducted using the AltAnalyze open-source software project from Cincinnati Children’s Hospital Medical Center and the University of Cincinnati [18].

Quantitative RT-PCR

The mRNA levels in various tissues were assessed using quantitative RT-PCR (qRT-PCR) as described [37]. In brief, RNA extracted from mouse tissues was reverse transcribed into complementary DNA. qRT-PCR was performed using QIAGEN Rotor Gene series Q and Rotor Gene series Q software, with SYBR Green as the detecting reagent. Melt values were obtained to confirm correct amplicons. For qRT-PCR, the following primers were used (Table 2).

Table 2.

Primers used in qRT-PCR analysis.

Gene	Forward Primer (5’ to 3’)	Reverse Primer (5’ to 3’)
ASP	TCTCTGGTGGGTGGGACTTC	TGATTTTAGCCTCCATTAGGTTTCC
CYP1A1	AGAATACGGTGACAGCCAGG	TTTGGGAGGAAGTGGAAGG
CYP1B1	GGACGCCTTCATCCTCTCTG	CTGAACATCCGGGTATCTGGT
AhR	CCAGGACCAGTGTAGAGCAC	CCATTCAGCGCCTGTAACAA
GUS	CCGACCTCTCGAACAACCG	GCTTCCCGTTCATACCACACC
Ephx1	GTTCTACATTCAAGGCGGCG	CAAACCTTTCACGTGGTTGGG
Akr1c22	GGCCACTTAATTCCTGCCCT	GTGACGAACAGGTCTTCTCTCT
Gcnt1	GGGATTGTGCCGAGGCTGT	TGCCAGTTTAACAGCGGGAC
Sult3a1	TCGCAAAGGTGTTGTTGGAG	AAGTACAACCTCTGGCTAGTACA
Sult1b1	CATTCGTTCTGATTGAAGCCTG	GGTGTGCTCTGGAACTCTTCA
Gstp1	GCGGCAAATATGTCACCCTC	TCGGCAAAGGAGATCTGGTC
Gsta3	TTGCCCTGGTTGAACTCCTC	CTGCTTCTCAGCGCTTTCAG
Ugcg	TGTGTGACGGGGATGTCTTG	TGAAAACCTCCAACCTCGGTC
Uggt1	TCTGGACCTAACAGCAAGCA	TTCCTTGGAGGACATTCCTTTT
Tubb4a	ATGAGGCCACAGGTGGAAAC	CTCGGTGTAGTGACCCTTGG

HPLC analysis of mouse tissues

To analyse DMBA metabolism in mouse liver samples, High Performance Liquid Chromatography (HPLC) was used. In brief, 8-week-old mice (total 8 mice: 2 from each experimental group) were given one dose of DMBA at a concentration of 30 mg/kg mouse weight and were sacrificed after 24 h. In preparation of HPLC analyses, necropsy was carried out and liver sections weighing approximately 100 mg were extracted from the mice. These sections were homogenized and prepared for HPLC as described [24]. In brief, tissues were homogenized thoroughly in 1 ml of homogenization buffer (10 mM Tris, 0.1 M K2HPO4, 0.1 M KCl, and 10 mM ethylenediaminetetraacetic acid, and pH7.4) using a dounce homogenizer. The homogenate was then centrifuged at 1000× g for 10 min and the supernatant was collected. The pellet was rehomogenized and supernatants were combined. Two millilitres of formic acid were added to the supernatant and mixed gently. Next, 100 μL of 5 μM DBP was added to the mixture to be used as an internal standard for HPLC. Two millilitres of hexane were added to the mixture and shaked for 15 min. Samples were then centrifuged and the organic phase (top layer) was collected and hexane was evaporated under a cold air stream. Sample residues containing DMBA metabolites were dissolved in 200 μl of methanol and separated using a C18 reversed-phase column [250 X 4.6 mm, 5 µm (Sigma Model #58,345)] with a guard protecting column. HPLC was carried out using Shimadzu HPLC analysis instruments: (system controller SCL-10AVP, degasser DGU-20A3, liquid chromatograph LC-20AT, UV/VIS detector SPD-20AV, fraction controller FRC-10A). HPLC was performed using an isocratic method with a flow rate of 1 mL/min and a mobile phase consisting of methanol: H2O (92:8) for 40 min. The effluent was detected by a fluorescence detector using an excitation wavelength of 276 nm and emission wavelength of 376 nm.

Statistical analyses

GraphPad Prism computer software and Microsoft Excel were used to perform statistical analyses on collected data. One-way ANOVA was the major statistical analysis performed on data. Survival of animals was assessed using Kaplan–Meier survival analysis. All data were determined statistically significant or otherwise at a set of p < 0.05.

Funding Statement

This work was supported in part by an NIH R03 grant from National Cancer Institute (NCI 1R03CA198630-01 and 1R03CA202427-01) to Venu Cheriyath.

Acknowledgments

A^vy/a and a/a mice were a kind gift from Dr. Robert Waterland (Baylor College of Medicine). We thank Dr. Lani Lyman-Henley and Dr. Sylvia Meyer, and other Animal Care Facility Staff for their assistance with the mouse care. We are thankful to Dr. Larry Lemanski, Anne Davenport at Texas A&M-Commerce, and Karthika Venugopalan at University of Texas at Dallas for critical reading of the manuscript. This study was supported in part by an NIH R03 grants NCI 1R03CA202427-01 and NCI 1R03CA198630-01 to Venu Cheriyath. Simbarashe Mazambani was a recipient of A&M-Commerce's graduate assistantship.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.