Individual Variability in the Detoxification of Carcinogenic Arylhydroxylamines in Human Breast

Abstract

Cytochrome b₅ (b5) and NADH cytochrome b₅ reductase (b5R) detoxify reactive hydroxylamine (NHOH) metabolites of known arylamine and heterocyclic amine mammary carcinogens. The aim of this study was to determine whether NHOH reduction for the prototypic arylamine 4-aminobiphenyl (4-ABP) was present in human breast and to determine whether variability in activity was associated with single nucleotide polymorphisms (SNPs) in the coding, promoter, and 3′untranslated region (UTR) regions of the genes encoding b5 (CYB5A) and b5R (CYB5R3). 4-ABP-NHOH reduction was readily detected in pooled human breast microsomes, with a K_m (280μM) similar to that found with recombinant b5 and b5R, and a V_max of 1.12 ± 0.19 nmol/min/mg protein 4-ABP-NHOH reduction varied 75-fold across 70 individual breast samples and correlated significantly with both b5 (80-fold variability) and b5R (14-fold) immunoreactive protein. In addition, wide variability in b5 protein expression was significantly associated with variability in CYB5A transcript levels, with a trend toward the same association between b5R and CYB5R3. Although a sample with a novel coding SNP in CYB5A, His22Arg, was found with low reduction and b5 expression, no other SNPs in either gene were associated with outlier activity or protein expression. We conclude that b5 and b5R catalyze the reduction of 4-ABP-NHOH in breast tissue, with very low activity, protein, and messenger RNA expression in some samples, which cannot be attributed to promoter, coding, or 3′UTR SNPs. Further studies are underway to characterize the transcriptional regulation of CYB5A and CYB5R3 and begin to understand the mechanisms of individual variability in this detoxification pathway.

Breast cancer kills more than 40,000 women per year in the Unites States alone (American Cancer Society, 2009). Breast cancer incidence is highest in industrialized nations, and women who move from low-risk to high-risk countries acquire a breast cancer risk of the host country within as little as two generations (Kelsey and Horn-Ross, 1993; McPherson et al., 2000). This supports the hypothesis that environmental factors play a significant role in the pathogenesis of breast cancer.

Two environmental factors, grilled meat consumption and cigarette smoking, have been associated with breast cancer risk. Well-done meats contain heterocyclic amines, such as 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP), which are mammary carcinogens in rodents (Ghoshal et al., 1994; Ito et al., 1991; Snyderwine et al., 1998). Most epidemiologic studies demonstrate a strong association between the consumption of well-done meats and breast cancer, regardless of menopausal state (Dai et al., 2002; De Stefani et al., 1997; Hermann et al., 2002; Shannon et al., 2003; Sinha et al., 2000; Steck et al., 2007). Tobacco smoke also contains PhIP (Manabe et al., 1991), as well as other mammary carcinogens such as the arylamine 4-aminobiphenyl (4-ABP) (Hecht, 2002; Shan et al., 2005). However, the association between smoking and breast cancer is less clear, with some studies finding a positive association (Bennicke et al., 1995; Brownson et al., 1988; Calle et al., 1994; Lash and Aschengrau, 1999; Marcus et al., 2000; Meara et al., 1989; Terry and Rohan, 2002) and other studies finding no association with cancer outcome (Baron et al., 1986; Braga et al., 1996; Chu et al., 1990; Egan et al., 2002; Field et al., 1992; Hunter et al., 1997; Millikan et al., 1998; Vatten and Kvinnsland, 1990). These discordant findings could be due, in part, to individual differences in the biotransformation of carcinogens after environmental exposure.

Both PhIP and 4-ABP form genotoxic DNA adducts that are thought to initiate cancer, and both compounds must be bioactivated to arylhydroxylamines in order to bind DNA (Bartsch, 1981; Fan et al., 1995; Turesky et al., 1991). Arylhydroxylamine metabolites are reduced back to the parent amines (4-ABP and PhIP), which are not directly mutagenic, by cytochrome b₅ (b5) and cytochrome b₅ reductase (b5R) in human liver (Kurian et al., 2006). This NADH-dependent microsomal pathway is up to 55 times more efficient than the forward oxidation reaction in human liver (King et al., 1999). We hypothesized that this reduction pathway also catalyzes the local detoxification of arylhydroxylamine carcinogens in cancer target tissues, such as breast. We further hypothesized that substantial genetic variability in local detoxification may be present among women, which could influence the risk of DNA adduct formation and breast cancer initiation after exposure to heterocyclic and aromatic amine carcinogens.

The overall aims of this study, therefore, were to determine the role of b5 and b5R in the local reduction of carcinogenic hydroxylamines (NHOHs) in human breast tissue using the NHOH of 4-ABP (4-ABP-NHOH) as a prototype compound and to evaluate individual variability in this pathway. To accomplish this, we performed kinetic and inhibitory studies for 4-ABP-NHOH reduction in breast and determined the correlation between b5 and b5R protein expression and reduction activities across individual breast samples. In addition, we resequenced the promoter, coding, and 3′untranslated region (UTR) regions of the genes encoding b5 (CYB5A) and b5R (CYB5R3) and evaluated the relationship between genotype and reduction phenotype in 70 human breast samples.

Materials and Methods

Breast samples.

Ninety-three breast tissue samples from women undergoing reduction mammoplasty or surgical lumpectomy, without a history of breast irradiation, were obtained through the Midwest Division of the Cooperative Human Tissue Network. Twenty-three samples were used for pooled experiments, and 70 samples were used for individual genotype-phenotype determinations. This latter population included 38 Caucasian women (median age 41 years, range 15–80) and 31 African-American women (31 years [19–86]), with race not recorded for one sample. All samples were histologically benign and were kept frozen at −80°C until processed. All participants provided informed consent through the Cooperative Human Tissue Network. Tissues samples were provided with anonymous ID numbers, and the Institutional Review Board of the University of Wisconsin-Madison approved the study protocol.

Subcellular fractionation.

Breast tissues (approximately 5 g each) were homogenized in Dulbecco’s PBS (Sigma Aldrich), pH 7.4, and subjected to standard ultracentrifugation to isolate nuclear pellets, cytosol, and microsomal fractions. Microsomes and nuclear pellet were stored at −80°C until use; microsomes were used for all activity experiments because arylhydroxylamine reduction is not present in cytosolic fractions (Kurian et al., 2006).

4-ABP-NHOH reduction in breast.

Standards for 4-ABP-NHOH (a gift from Dr Fred Kadlubar) and 4-ABP (Sigma Aldrich) were dissolved in 50% dimethyl sulfoxide (DMSO); for the NHOH, 3mM ascorbic acid was included to prevent autoxidation during sample preparation (Kurian et al., 2006). Personnel wore laboratory coats, gloves, and safety goggles when handling 4-ABP and 4-ABP-NHOH. Carcinogens were dispensed into pretared Eppendorf tubes in a fume hood, with careful precautions to prevent contamination of gloves, tube exteriors, balances, and bench tops.

Reduction of 4-ABP-NHOH to ABP was determined by incubating breast microsomes (90 μg) with 250μM 4-ABP-NHOH in PBS, pH 7.4, in the presence of 1mM NADH and 3mM ascorbic acid for 30 min at 37°C; these conditions approximate linear kinetics (Kurian et al., 2006). Human serum albumin (90 μg; Sigma), instead of microsomes, was included in negative control reactions and in all 4-ABP standards. 4-ABP product was detected using high performance liquid chromatography with ultraviolet detection, as previously described (Kurian et al., 2006); the assay was linear from 0.3μM to 1mM, with a limit of quantitation of 0.3μM (equivalent to 0.028 nmol/mg/min under these assay conditions).

Kinetics for 4-ABP-NHOH reduction were determined for breast microsomes pooled from 23 individuals under initial rate conditions using 50–1000μM 4-ABP-NHOH. Data were fit to the Michaelis-Menten equation and apparent kinetic parameters were determined using GraphPad Prism 5 (GraphPad Software, San Diego, CA).

To investigate the importance of b5 in 4-ABP-NHOH reduction in human breast, rabbit polyclonal antisera to human recombinant b5 were used for immunoinhibition of reduction (b5R antisera with active inhibitory activity were no longer available) (Kurian et al., 2004, 2006). Antisera (50 μl) were preincubated with pooled breast microsomes (90 μg) for 30 min at room temperature, followed by incubation with 4-ABP-NHOH (250μM) as described above. Preimmune rabbit serum (50 μl) was used for negative controls.

b5 and b5R protein expression.

In order to determine the relationship between enzyme expression and 4-ABP-NHOH reduction activities, immunoblotting for b5 and b5R was performed for individual breast microsomes using polyclonal antibodies to human b5 and b5R (Kurian et al., 2004, 2006). As a loading control, blots were also probed with anti-mouse β-actin monoclonal antibody conjugated with horseradish peroxidase (1:5000; Abcam, Cambridge, MA). In all cases, bound antibody was detected using chemiluminescence (Super Signal West Pico, Pierce) using a UVP Imager (UVP, Upland, CA) and quantified by ImageJ (http://rsb.info.nih.gov/ij/). Densitometries for bands corresponding to b5 and b5R were analyzed both directly and after normalization for β-actin densitometry.

Resequencing of CYB5A and CYB5R3 complementary DNAs.

Total RNA was isolated from individual breast samples (approximately 100 mg starting tissue) using the RNeasy Lipid Tissue Mini Kit (Qiagen, Valencia, CA). Reverse transcription PCR was performed using the RETROscript Kit (Ambion, Austin, TX) to obtain complementary DNA (cDNA). Primers were designed to amplify the entire coding regions of the microsomal forms of CYB5A and CYB5R3 (Table 1). For CYB5A, 0.125mM deoxyribonucleotide triphosphates (dNTPs) and Supertaq polymerase (Ambion) were added to each reaction, with the following thermocycler parameters: initial denaturation for 2 min at 94°C; followed by 35 cycles of 15 s at 94°C, 25 s at 51°C, and 30 s at 72°C; and a final extension at 72°C for 7 min. For CYB5R3, reactions were similar except for the addition of 2% DMSO and the use of a touchdown PCR method: initial denaturation for 2 min at 95°C; followed by 10 cycles of 30 s at 91°C, 30 s at 71°C–67°C (two cycles at each degree), and 1 min 15 s at 72°C; followed by 28 cycles of 30 s at 91°C, 30 s at the touchdown temperature of 66°C, and 1 min 15 s at 72°C; and by a final extension at 72°C for 7 min.

TABLE 1.

Primers Used to Analyze the Coding, Promoter, and 3′UTR Regions of CYB5A and CYB5R3 in Human Breast Samples

Experiment	Orientation	Primer sequence (5′ → 3′)
CYB5A
Coding PCR and sequencing	Forward	GGCCTGGCTCGCGGCGAACCGAG
	Reverse	CTCCCGTGTCCAAAGCAGGC
Promoter PCR and sequencing	Forward	GGACTGCTGCCAATAGGAAA
	Reverse	TCTTGCTGTGGTTGTGCTTC
3′UTR PCR and sequencing	Forward	TTTTGAGTCCACCACAGTGC
	Reverse	CCTCTGTGGGTCTGGATGAG
qPCR	Forward	CCAAAGTTAAACAAGCCTCCG
	Reverse	TGTTCAGTCCTCTGCCATG
CYB5R3
Coding PCR	Forward	ACCATGGGGGCCCAGCTCA
	Reverse	ACTGGGTGAGCGTGAACAG
Sequencing	Forward	CTACCTCTCGGCTCGAATTG
	Reverse	TGTCTCCAATCTGCATGCTC
Promoter PCR and sequencing	Forward	GTACGGGACTTCAAACCA
	Reverse	CGTAAGTAGCGGTCACCA
Upstream portion of 3′UTR, PCR, and sequencing	Forward	AGCCTCTCCATTCTTCAGCA
	Reverse	CCTGGCACCTGCAGCTTT
Downstream portion of 3′UTR, PCR, and sequencing	Forward	CCACACACACTATAAGGCTGAGG
	Reverse	CAGACCCTCGAGGAGCTAGA
qPCR	Forward	CCCAGCTCAGCACGTTG
	Reverse	CAGCGGGTACTTGATGTCC

PCR reactions were cycle sequenced using BigDye Terminator v3.1 reagents (Applied Biosystems, Foster City, CA) and sequence-specific primers (Table 1); reactions were analyzed on an ABI 3700 DNA sequencer. Sequences were aligned, and variants from the published wild-type (WT) sequences (NM_148923 for CYB5A, NM_000398 for CYB5R3) were identified using the Staden Package (http://staden.sourceforge.net). All identified coding region single nucleotide polymorphisms (cSNPs) were confirmed by directly reviewing the sequencing chromatograms and by repeating the sequencing reaction. The first A in the translation initiation codon ATG was designated + 1 for both cDNAs.

Resequencing the promoter regions of CYB5A and CYB5R3.

Genomic DNA from individual subjects was isolated from 20 mg of the breast nuclear pellet obtained during subcellular fractionation using the DNeasy Tissue Kit (Qiagen). DNA was quantified using spectrophotometry and the 260:280 ratio. Primers were designed to amplify the promoter regions of CYB5A and CYB5R3 (Table 1). The 5′ flanking region of CYB5A has been characterized up to −1300 bp, and functional promoter regions have been mapped to −327 to + 15 bp relative to the transcription start site (Huang et al., 2005). For our study, the region from −480 to + 73 was amplified for CYB5A promoter resequencing. PCR conditions included 0.125mM dNTPs, 5% DMSO, and Supertaq polymerase, with initial denaturation for 2 min at 94°C; 35 cycles of 15 s at 94°C, 25 s at 56°C, and 45 s at 72°C; followed by a final extension at 72°C for 7 min.

The promoter region for the ubiquitous microsomal form of CYB5R3 has previously been evaluated to 3200 bp upstream from CYB5R3; functional promoter regions have been mapped, using deletion constructs, to within −559 bp relative to the transcription start site (Toyoda et al., 1995). To encompass this region, the promoter region for exon 1M of CYB5R3 (which is specific to the microsomal transcript of the gene) was resequenced from −673 through + 197 bp in intron 1. The components for the CYB5R3 promoter PCR reaction were as follows: 0.2mM deoxyadenosine triphosphate, 0.2mM deoxycytidine triphosphate, 0.2mM deoxythymidine triphosphate, 0.13mM deoxyguanosine triphosphate, 0.07mM deaza-GTP, 10% DMSO, and Go-Taq Hot Start Polymerase (Promega, Madison, WI). A modified touchdown PCR was designed to amplify the promoter of CYB5R3: initial denaturation for 5 min at 95°C; 30 cycles of 30 s at 95°C, 30 s at 60°C–46°C (two cycles per degree), and 60 s at 72°C; and an additional 30 cycles of 30 s at 95°C, 30 s at 50°C, and 60 s at 72°C, followed by final extension as before. The PCR reactions for CYB5R3 were purified using the Wizard SV PCR Clean-up System (Promega). PCR reactions for both promoters were cycle sequenced as for cDNAs, and variants from published genomic sequences (National Center for Biotechnology Information database ID 1528 for CYB5A and 1727 for CYB5R3) were identified using the Staden Package.

Resequencing the 3′UTR regions of CYB5A and CYB5R3.

Primers were designed to amplify the 3′UTR regions of CYB5A and CYB5R3 (Table 1) from breast genomic DNA. Supertaq polymerase and 0.125mM dNTPs were used for all reactions, with initial denaturation of 2 min at 95°C; 30 cycles of 15 s at 94°C, 30 s at 59°C, and 45 s at 72°C; and final extension for 7 min at 72°C for CYB5A. The 3′UTR region for CYB5R3 was split into two separate PCR reactions, one to amplify an upstream fragment of the 3′UTR and another for the downstream portion. The thermocycler conditions for the upstream reaction were as follows: 5 min at 95°C; 30 cycles of 30 s at 95°C, 30 s at 59°C, and 60 s at 72°C; and a final extension at 72°C for 7 min. The conditions for the downstream reaction were identical except for an annealing temperature of 56°C. 3′UTR PCR products were resequenced and aligned to CYB5A and CYB5R3 genomic reference sequences, as described for the promoter amplicons.

Quantitative PCR for CYB5A and CYB5R3 messenger RNA expression.

To determine whether variability in b5 and b5R protein expression was associated with differences in messenger RNA (mRNA) expression, breast samples with outlier b5 or b5R densitometries (< 10th percentile and > 90th percentile for each protein) were screened for CYB5A and CYB5R3 expression by quantitative PCR (qPCR) using five breast samples with median b5 protein expression and five breast samples with median b5R protein expression as reference controls. One microgram of breast tissue total RNA from each sample was reverse transcribed into cDNA using the high-capacity reverse transcription kit (Applied Biosystems). Five microliters of this cDNA, diluted 1 in 25, was mixed with 12.5 μl Power SYBR mix, 0.25 μl uridine phosphate (UDP)-N-glycosylase, and 1 μl each of 5μM forward and reverse primers (Table 1) in a total volume of 25 μl. qPCR cycle conditions were 50°C for 2 min and 95°C for 10 min and 40 cycles of 95°C for 15 s and 60°C for 1 min (ABI StepOne Plus cycler, Applied Biosystems). Data were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) expression using the $2^{-\Delta \Delta {\mathit{C}}_T}$ method (user manual #2, ABI Prism 7700 SDS).

Statistical analyses.

Activities and expression were compared between samples with and without given SNPs and between Caucasian and African-American women using a Mann-Whitney U-test, with p < 0.05 considered significant. Spearman’s rank correlation was used to examine the relationship between age and either reduction activities or protein expression. These analyses were performed using Graphpad Prism (GraphPad Software). In addition, interactions among age, race, and tissue of origin (lumpectomy or reduction mammoplasty) were evaluated by analysis of covariance (ANCOVA) using R software (www.r-project.org/).

Haplotype analyses, including testing alleles for Hardy-Weinberg equilibrium, testing the significance of D′ between two loci, and association of haplotypes with reduction activities, were carried out using the Haploview 4.1 software package (Barrett et al., 2005). To predict the potential effects of cSNPs on protein function, two computational predictive software programs were used: PolyPhen (http://genetics.bwh.harvard.edu/pph/), which predicts the effect of SNPs on protein 3D structure and function, and pmut (http://mmb2.pcb.ub.es:8080/PMut/), which combines evolutionary information with structural data.

RESULTS

4-ABP-NHOH Reduction Kinetics and Immunoinhibition

Pooled breast microsomes catalyzed the reduction of 4-ABP-NHOH to 4-ABP, with kinetics that fit to a one-site Michaelis-Menten equation. The apparent K_m of 280μM was similar to that previously observed for 4-ABP-NHOH reduction in a recombinant system containing only b5 and b5R (220μM K_m) (Kurian et al., 2006). The V_max for 4-ABP-NHOH reduction in pooled breast was 1.12 ± 0.19 nmol/min/mg protein or about one-fourth of that previously observed in human liver (Kurian et al., 2006). 4-ABP-NHOH reduction in breast was inhibited by b5 antisera by > 97% compared with preimmune sera (mean of two experiments each performed in duplicate).

4-ABP-NHOH Reduction Protein Expression across Individuals

Reduction activities for 4-ABP-NHOH (250μM) in individual breast samples varied approximately 75-fold from ≤ 0.03 (the limit of quantitation of the assay) to 2.24 nmol/min/mg protein, with a median activity of 0.61 nmol/min/mg protein. Protein expression varied more than 80-fold for b5 and 14-fold for b5R (Fig. 1). Reduction activities were positively correlated with both b5 and b5R immunoreactive protein (r = 0.64, p < 0.0001 for b5; r = 0.44, p = 0.0002 for b5R; Fig. 1). Similar to what we have observed in human liver microsomes (Sacco and Trepanier, 2010), expression of b5 and b5R were also highly correlated with one another across individuals (r = 0.69, p < 0.0001).

FIG. 1.

Association of b5 and b5R protein expression with 4-ABP-NHOH reduction activities in 70 human breast samples. Protein expression was significantly correlated with 4-ABP-NHOH reduction activities for both b5 (top panel; r = 0.64, p < 0.0001) and b5R (bottom panel; r = 0.44, p = 0.0002).

Initially, we normalized all protein expression data to β-actin as a loading control. However, β-actin densitometries themselves varied 83-fold across breast microsomes, which was much greater than the narrower range (< 5-fold) that we have observed in human liver microsomes (data not shown) (Sacco and Trepanier, 2010). This variability in β-actin among samples could not be attributed to inaccurate gel loading because the median coefficient of variation for actin densitometries between tissue replicates was only 13%. In addition, when b5 and b5R densitometries were normalized to β-actin, the observed variability in protein expression became much higher (> 500- and > 100-fold, respectively). Therefore, we chose to express all protein expression data in direct densitometry units. This did not change the direction of the findings for any analyses but led to more conservative estimates of variability in enzyme expression than did data normalized to β-actin.

We next tested for associations between reduction activities or protein expression and age, race, or method of tissue procurement (lumpectomy vs. reduction mammoplasty) using ANCOVA. Only age at the time of breast sample procurement was significantly correlated with both reduction of ABP-NHOH (r = 0.34, p = 0.004) and expression of b5 in breast (r = 0.32, p = 0.007; Fig. 2), with a trend toward a positive association of age with b5R protein expression (r = 0.22, p = 0.07).

FIG. 2.

Correlation between increasing age and 4-ABP-NHOH reduction activities (main panel; r = 0.34, p = 0.004) and expression of b5 in human breast samples (inset panel; r = 0.32, p = 0.007).

Coding Polymorphisms in CYB5A and CYB5R3

The presence of SNPs in the coding regions of CYB5A and CYB5R3 was investigated in 70 breast samples; in all, three cSNPs were identified in CYB5A and four cSNPs in CYB5R3 (Table 2, Fig. 3). No amplicon was of an unexpected size when cDNA products were viewed by agarose gel electrophoresis. A novel nonsynonymous cSNP in CYB5A, 65A>G (His22Arg), was found in the heterozygous state in one African-American woman. His22Arg is located in the amino terminal hydrophilic heme/steroid-binding domain that is conserved in all proteins in the b₅ superfamily (Marchler-Bauer et al., 2011). In silico analysis by PolyPhen predicted that His22Arg would not be damaging, whereas pmut analysis predicted that His22Arg was likely to be pathologic. 4-ABP-NHOH reduction activity for this subject was in the lower 10th percentile of samples with a WT CYB5A coding sequence, with b5 protein expression in the lower 15th percentile (Fig. 4). Direct functional characterization of this allele, as well as further evaluation in a larger number of subjects, is underway.

TABLE 2.

Sequence Variants in the cDNAs for CYB5A and CYB5R3 Prepared from 70 Histologically Normal Breast Samples, Obtained from Women Undergoing Lumpectomy or Reduction Mammoplasty

Gene	Sequence change	Zygosity	Amino acid change	Reference SNP IDa	Number of subjects	Minor allele frequency
						AA	CA
CYB5A	36C>T	Het	None	rs1051236	14	0.031	0.184
		Hom			1
	65A>G	Het	His22Arg	rs74339771 (novel)	1	0.016	0.000
	288G>A	Het	None	rs7238987	12	0.078	0.118
		Hom			1
CYB5R3	132G>A	Het	None	rs5996200	18	0.125	0.132
	350C>G	Het	Thr117Serb	rs1800457	10	0.250	0.000
		Hom			3
	472G>A	Het	Ala158Thr	rs75478217 (novel)	1	0.000	0.013
	898G>A	Het	Val300Ile	rs61743746	2	0.031	0.000

Note. AA, African-American; CA, Caucasian; Het, heterozygous for cSNP; Hom, homozygous for cSNP.

^ahttp://www.ncbi.nlm.nih.gov/snp.

^bEquivalent to previously published T116S (Jenkins and Prchal, 1997).

FIG. 3.

Gene maps of promoter, coding, and 3′UTR SNPs in CYB5A and CYB5R3 found by cDNA and genomic DNA resequencing of 70 individual human breast samples. Novel SNPs are underlined.

FIG. 4.

Scatter plot of microsomal 4-ABP-NHOH reduction activities (panel A) and expression of b5 protein (panel B), in association with CYB5A coding polymorphisms, in human breast. Each point represents an individual sample; horizontal lines represent the medians.

For CYB5R3, a novel nonsynonymous cSNP, 472G>A (Ala158Thr), was found in the heterozygous state in one Caucasian woman (Table 2). This sample showed b5R expression in the upper 95th percentile of the samples that had WT CYB5R3 coding sequences (Fig. 5B), but reduction activities were closer to the population median (Fig. 5A). Two previously reported nonsynonymous cSNPs in CYB5R3, 350C>G (Thr117Ser) (Jenkins and Prchal, 1996, 1997) and 898G>A (Val300Ile; rs61743746), were identified in 13 and 2 samples, respectively all African-American but were not associated with outlier activity or protein expression.

FIG. 5.

Scatter plot of the association of CYB5R3 coding polymorphisms with reduction of 4-ABP-NHOH (panel A) and expression of b5R protein (panel B) in human breast microsomes. Each point represents an individual sample; horizontal lines represent the medians.

Promoter and 3′UTR Regions of CYB5A and CYB5R3

Because cSNPs in CYB5A and CYB5R3 could not account for the observed wide variability in 4-ABP-NHOH reduction activities and protein expression in human breast, the known promoter and 3′UTR regions of both genes were also resequenced for all samples. Seven SNPs, six of them novel, were found in the promoter region of CYB5A (Table 3). Five of these SNPs, −414T>G, −399G>T, −389G>A, −382C>T, and −363G>A (Fig. 3), were located in a previously identified possible silencer region of the CYB5A promoter (−450 to −300 bp) (Li et al., 1995); however, none of these CYB5A promoter SNPs was associated with outlier b5 expression or reduction activities (data not shown).

TABLE 3.

Sequence Variants in the Promoter Regions of CYB5A and CYB5R3, Found during Resequencing of Genomic DNA from 70 Histologically Normal Breast Samples (Some Samples Had More than One SNP, so that Subject Totals Exceed 70)

Gene	Sequence change	Zygosity	Reference SNP IDa	Number of subjects	Allele frequency
					AA	CA
CYB5A	WT	Hom		44	NA	NA
	−458C>T	Het	rs79278115 (novel)	5	0.063	0.013
	−414T>G	Het	rs77569097 (novel)	2	0.031	0.000
	−399C>T	Het	Rs113933995 (novel)	1	0.016	0.000
	−389G>A	Het	rs77005399 (novel)	1	0.016	0.000
	−382C>T	Het	rs76631379 (novel)	1	0.016	0.000
	−363G>A	Het	rs3764506	2	0.016	0.013
	−214C>T	Het	rs77442034 (novel)	15	0.047	0.184
		Hom		1
CYB5R3	WT	Hom		13	NA	NA
	−386C>A	Het	rs5996205	25	0.597	0.684
		Hom		31
	−374C>T	Het	rs28365957	22	0.242	0.145
		Hom		2
	−322G>C	Het	rs28365958	22	0.274	0.171
		Hom		4
	−300C>*	Het	rs8190363	20	0.258	0.132
		Hom		3
	−272G>A	Het	rs8190364	1	0.018	0.000
	−251G>T	Het	rs73888347 (novel)	4	0.071	0.000
	−244G>A	Het	rs8190365	2	0.036	0.000
	−216A>G	Het	rs8190366	17	0.150	0.132
		Hom		1
	−137C>A	Hom	rs8190368	24	0.400	0.316
	I1 + 6C>T	Het	rs8190370	2	0.032	0.000

Note. Allele frequencies and genotype frequencies are listed for the African-American (AA) and Caucasian American (CA) populations. Het, heterozygous for SNP; Hom, homozygous for SNP; NA, not applicable.

^ahttp://www.ncbi.nlm.nih.gov/snp.

The CYB5R3 promoter showed more genetic variability than for CYB5A, with nine different promoter SNPs and one intronic SNP identified (Fig. 3). Clean sequences of the full CYB5R3 promoter region were not obtained in four samples; therefore, unequivocal genotypes for some SNPs were not determined in these subjects. Several promoter SNPs were observed at fairly high allele frequencies (Table 3); one of these, −137C>A, is located in an experimentally confirmed Sp1 transcription factor–binding site (Toyoda et al., 1995). Four less common SNPs (−272G>A, −251G>T, −244G>A, and I1 + 6C>T) were observed only in African-Americans. However, none of these CYB5R3 promoter polymorphisms was associated with outlier 4-ABP-NHOH reduction activities or b5R expression. Many of the SNPs found in the promoter region of CYB5R3 were identified in haplotype configurations; however, differences in b5R protein expression were still not observed when haplotype was considered (data not shown).

The 3′UTR region of CYB5A showed minimal genetic variability, with only three novel SNPs identified in five individuals (Table 4, Fig. 3). The 3′UTR region of CYB5R3 was more variable; 11 SNPs were identified, six of which were novel (Fig. 3). Several SNPs were found at high frequencies (Table 4), but none of the 3′UTR SNPs in either gene was associated with outlier activities or expression. The most common haplotypes for the CYB5R3 3′UTR were *63G/*287A (26 samples) and *63G/*287A/*392C (five samples), but no haplotype was associated with significant differences in b5R expression (data not shown).

TABLE 4.

Sequence Variants in the 3′UTR Regions of CYB5A and CYB5R3 in Breast Samples from 70 Women

Gene	Sequence change	Zygosity	Reference SNP IDa	Number of subjects	Allele frequency
					AA	CA
CYB5A	*21G>A	Het	rs111303909 (novel)	1	0.016	0.000
	*24G>C	Het	ss252841185 (novel)	1	0.000	0.013
	*186G>T	Het	rs115752542 (novel)	3	0.047	0.000
CYB5R3	*45C>G	Het	ss252841187 (novel)	1	0.016	0.000
	*62C>T	Het	ss252841188 (novel)	2	0.031	0.000
	*63A>G	Het	rs137124	23	0.672	0.237
		Hom		19
	*74C>T	Het	rs76582006 (novel)	1	0.000	0.013
	*191C>T	Het	ss252841189 (novel)	3	0.016	0.026
	*278G>T	Het	ss252841190 (novel)	2	0.031	0.000
	*287C>A	Het	rs137123	25	0.625	0.250
		Hom		17
	*392G>C	Het	rs7284807	6	0.156	0.000
		Hom		2
	*713G>A	Het	rs79192059 (novel)	2	0.016	0.013
	*851G>C	Het	rs6002821	1	0.016	0.000
	*929A>G	Het	rs6002820	5	0.063	0.013

Note. AA, African-American; CA, Caucasian American; Het, heterozygous for SNP; Hom, homozygous for SNP.

^ahttp://www.ncbi.nlm.nih.gov/snp.

CYB5A and CYB5R3 mRNA Expression

Five breast samples each with b5 protein expression in the lowest 10th percentile (median 1251 densitometry units, range 264–1647), top 90th percentile (median 15,605, range 14,524–23,801), and intermediate 45–55% percentile (median 4978, range 4175–7451) were evaluated for CYB5A mRNA expression. In breast samples with low b5 protein, CYB5A expression was also decreased (average fold difference in CYB5A expression of 0.6 [95% confidence interval {CI}, 0.3–0.9]), whereas in samples with high b5 protein, the average fold difference in CYB5A expression was variably increased (average fold difference 4.6 [95% CI, 0–9.7]) relative to the intermediate samples. In addition, ΔC_T for CYB5A expression (relative to GAPDH) was significantly associated with b5 protein densitometry across samples (r = −0.75, p = 0.002; Fig. 6; note that lower ΔC_T is equivalent to higher message level, so the relationship between message and protein is strongly positive).

FIG. 6.

Relationship between CYB5A and CYB5R3 mRNA and protein expression in human breast samples. qPCR was performed on breast samples with outlier low, outlier high, or intermediate protein expression for b5 and for b5R. ΔC_T for CYB5A expression (normalized to GAPDH) was significantly associated with b5 protein densitometry (r = −0.75, p = 0.002; top panel), with a trend for an association between ΔC_T for CYB5R3 expression and b5R protein densitometry (r = −0.50, p = 0.07; bottom panel). Note that lower ΔC_T is equivalent to higher message level.

For CYB5R3 expression, five breast samples each with b5R protein expression in the lowest 10th percentile (median 1251, range 264–8873), the top 90th percentile (median densitometry units 32,677, range 31,044–42,210), and the intermediate 45–55% percentile (median 21,224, range 17,825–21,513) were also evaluated. In breast samples with low b5R protein, CYB5R3 expression was not demonstrably different (1.3-fold, 95% CI, 0.3–2.3) relative to the intermediate samples, whereas in samples with high b5R protein, the average fold difference in CYB5R3 expression was variably increased (9.3-fold, 95% CI, 0–31.0). There was a trend for an association between ΔC_T for CYB5R3 expression (relative to GAPDH) and b5R protein densitometry, although this did not reach significance (r = −0.50, p = 0.07; Fig. 6).

DISCUSSION

The carcinogens 4-ABP and PhIP must be bioactivated to their NHOH intermediates in order to form genotoxic DNA adducts (Bartsch, 1981; Fan et al., 1995; Turesky et al., 1991). This bioactivation is mediated by CYP1A2 in liver and by CYP1B1, CYP1A1, and lactoperoxidases in human breast (Crofts et al., 1998; Gorlewska-Roberts et al., 2004; Hellmold et al., 1998; Josephy, 1996; Williams et al., 1998). The generation and presence of reactive NHOHs is counteracted by two pathways: reverse reduction to the parent amine by the b5/b5R pathway and glucuronidation of the NHOH by UDP-glucuronosyltransferases (UGTs). 4-ABP-NHOH is glucuronidated by UGT1A6 (formerly UGT1.6) and UGT1A7 (formerly UGT1.7) (Orzechowski et al., 1994), and PhIP-NHOH is glucuronidated primarily by UGT1A1 and UGT1A4 (Malfatti and Felton, 2001). However, none of these UGTs is expressed in human breast (Lehmann and Wagner, 2008; Nakamura et al., 2008; Ohno and Nakajin, 2009), suggesting that glucuronidation is not likely to be an important local detoxification pathway for arylhydroxylamine carcinogens in breast.

On the other hand, we found that reduction of 4-ABP-NHOH was present in breast, with a V_max that was approximately one-fourth that observed in liver, and an apparent K_m similar to that previously found in a purified system containing recombinant b5 and b5R (Kurian et al., 2006). In addition, breast reduction activities were inhibited > 97% by b5 antisera and were correlated significantly with both b5 and b5R immunoreactive protein, which suggests that this detoxification pathway is catalyzed by b5 and b5R in breast as in liver (Kurian et al., 2006).

A wide range of individual variability in 4-ABP-NHOH reduction activities was found in this population (75-fold) along with corresponding variability in both b5 and b5R protein expression. This suggests that individual differences in the expression of either enzyme could influence detoxification capacity for arylhydroxylamines. Primary breast epithelial cultures from individual women show more than 75-fold variability in DNA adduct formation after exposure to standardized concentrations of arylhydroxylamines in vitro (Stone et al., 1998). These previous findings indicate that there are local differences that influence breast adduct formation that are independent of exposure dose. Our data suggest that individual differences in adduct formation could be due, in part, to wide individual variability in local arylhydroxylamine reduction by b5 and b5R.

We initially normalized immunoreactive protein to β-actin to control for possible differences in gel loading. β-Actin is expressed in breast tissue (Hellmold et al., 1998) and is commonly used as a loading control for microsomal fractions, including microsomes prepared from a breast cell line (Mahadevan et al., 2005). However, we found that β-actin densitometries in microsomes prepared from whole breast tissue varied much more widely than we have observed for liver microsomes. This wide range in breast β-actin may have resulted from the high adipose content of breast tissue relative to liver. Specifically, preadipocytes, which are present in normal breast (Ali et al., 2006), have high actin content, whereas mature adipocytes have about one-tenth as much actin (Spiegelman and Farmer, 1982), and this differentiation is under hormonal control. Therefore, β-actin may not have been an appropriate constitutive reference protein for native breast tissues. When analyzing protein expression data both with and without normalization for β-actin, we found similar qualitative results, but the range of protein expression was more conservative without normalization, which is how we chose to report the data.

In our population, b5 protein content varied more widely than did b5R expression, both when directly estimated from densitometries and when normalized to β-actin immunoreactive protein. We have also observed greater variability for b5 protein expression, compared with b5R, in human liver (Kurian et al., 2004, 2006). We hypothesized that individual variability in both protein expression and 4-ABP-NHOH reduction could be attributed, in part, to SNPs in CYB5A and CYB5R3. We have previously found two nonsynonymous cSNPs in CYB5A, Thr60Ala and Ser5Ala, both identified in African-American subjects, which were associated with unstable protein and low arylhydroxylamine reduction, respectively (Kurian et al., 2007; Sacco and Trepanier, 2010). In this breast survey, we identified another CYB5A cSNP, His22Arg, in a single African-American woman, for which both reduction activity and b5 expression were in the lower 10th–15th percentile compared with WT samples. This subject had no other nonsynonymous SNPs in the coding, promoter, or 3′UTR regions of either gene. However, we can conclude little from a single individual, and functional characterization of this cSNP is underway. No other cSNPs in CYB5A were observed in association with outlier expression or activity.

For CYB5R3, we found a high allele frequency for Thr117Ser, all in African-American subjects. This SNP was originally reported in an extended African-American family but without phenotypic characterization (Jenkins and Prchal, 1997). Thr117Ser was not associated with outlier reduction activities or b5R expression, either in breast or in our previous survey of 111 human livers (Sacco and Trepanier, 2010). We also identified a novel nonsynonymous CYB5R3 cSNP, Ala158Thr, with b5R protein expression in the 95th percentile; no coexisting promoter or 3′UTR SNPs were found in this subject, and this SNP requires further functional characterization.

Because polymorphisms in the coding sequences of CYB5A and CYB5R3 could not explain the observed wide variability in arylhydroxylamine reduction and protein expression, we next resequenced the regions encompassing the known promoter regions and 3′UTRs for both genes. However, none of the observed CYB5A or CYB5R3 promoter SNPs was associated with outlier protein expression using either normalized or nonnormalized densitometry data. The 3′UTR region of CYB5A showed minimal genetic variability; three novel SNPs were identified, but all were present at low frequencies. The 3′UTR of CYB5R3 was more variable, but neither SNPs nor haplotypes could be associated with outlier b5R expression.

Overall, the wide variability in arylhydroxylamine reduction and the corresponding variability in b5 and b5R expression could not be accounted for by polymorphisms in either CYB5A or CYB5R3. However, low b5 expression was significantly associated with low CYB5A message, with a trend for an association of variable b5R with CYB5R3 message as well, although the observed range of b5R protein expression was narrower than that seen for b5. We focused our gene resequencing on experimentally demonstrated functional promoter regions as well as in silico predictions of transcription factor–binding sites using MatInspector/MatBase (Genomatix). However, there may have been other important upstream or downstream regulator regions for either gene that we did not address.

In addition, there are likely to be environmental effects (genetic or epigenetic) on CYB5A and CYB5R3 expression that are not yet understood. For example, antioxidant deficiency in vivo decreases b5 expression and associated arylhydroxylamine reduction in liver (Bhusari et al., 2010), and the nonsteroidal antiinflammatory drug indomethacin increases cyb5a and cyb5r3 expression in rats (Naito et al., 2007). In addition, both CYB5A and CYB5R3 have putative Ah receptor-binding regions (Genomatix MatInspector), which could allow for induction of these two enzymes by environmental pollutants. This is supported by the finding that activity of this pathway is induced by benzo[a]pyrene (Anandakumar et al., 2009). The significant association in our population between increasing age and both b5 expression and arylhydroxylamine reduction in breast is intriguing and could reflect factors such as age group differences in hormonal status or cumulative environmental exposures. Little is known about the transcriptional regulation of these two genes; however, the strong correlation between their expression in both breast and in liver (Sacco and Trepanier, 2010) suggests that CYB5A and CYB5R3 may be coregulated. Studies on the transcriptional regulation of CYB5A and CYB5R3 are underway in our laboratory.

As for epigenetic regulation, both CYB5A and CYB5R3 have C-phosphate-G (CpG) islands in their 5′UTRs (http://genome.ucsc.edu; Rosenbloom et al., 2010) that may be subject to methylation and subsequent downregulation of gene expression. There is, in fact, experimental evidence that both genes undergo methylation. In particular, a proximal CpG island in CYB5A (I1 + 290 to −358), as well as one 23 kb upstream of the start codon, is strongly methylated in several human cell lines (Brunner et al., 2009; Meissner et al., 2008). CYB5R3 also has a proximal CpG island (I1 + 290 to −358) and a second distal site (−12806 to −13430) that are differentially methylated in human cell lines (Brunner et al., 2009; Meissner et al., 2008). However, individual differences in methylation have not been explored for these two genes. Experiments are underway to determine whether individual variability in expression of CYB5A and CYB5R3 in human breast can be associated with differences in gene methylation.

In summary, microsomal b5 and b5R catalyze the local reduction of carcinogenic 4-ABP-NHOH in human breast, with wide individual variability in the activity and expression of this pathway. These individual differences cannot, for the most part, be attributed to promoter, coding, or 3′UTR polymorphisms in CYB5A or CYB5R3. Investigations to characterize the transcriptional and epigenetic regulation of CYB5A and CYB5R3 are ongoing, with the overall goal of understanding possible environmental influences on this detoxification pathway and their potential role in cancer risk in association with arylamine and heterocyclic amine exposures.

FUNDING

Prevent Cancer Foundation; National Institute of General Medical Sciences, National Institutes of Health (R01 GM61753). University of Wisconsin-Madison Molecular and Environmental Toxicology Training Grant (T32 ES007015) to K.R.