Study subjects and data collection
A sample of 226 incident breast cancer cases (140 infiltrating ductal carcinomas, 27 infiltrating lobular carcinomas, four mixed infiltrating ductal and lobular carcinomas, one tubular carcinoma, three mucinous carcinomas, two metaplastic carcinomas and 49 unknown histological subtypes) diagnosed between 1995 and 1999 was identified through the Medical Oncology Clinics at the University of Pennsylvania Cancer Center (Table 1). Case status was confirmed by review of medical records using a standardized abstraction form. Women were excluded from this study if they were non-incident cases (i.e. those diagnosed more than 12 months prior to the date of study ascertainment) or had a prior diagnosis of cancer at any site except nonmelanoma skin cancer. The mean age of diagnosis was 59.8 years (standard deviation = 8.5 years) with a range of 43–80 years.
Information on risk factors, the medical history and breast cancer diagnostic information was obtained using a standardized questionnaire and a review of medical records. Information collected included the personal history of benign breast diseases, previous cancer diagnoses, demographic information such as race, and pathology data, including different tumor characteristics, such as the size, grade and ER status of tumors. All study subjects provided informed consent for participation in this research under a protocol approved by the Committee for Studies Involving Human Subjects at the University of Pennsylvania, and genetic samples and information were handled in accordance with the Helsinki Declaration. Genomic DNA for the present study was self-collected by each study subject using sterile cheek swabs (Cyto-Pak Cytosoft Brush; Medical Packaging Corporation, Camarillo, CA, USA).
Generation of recombinant SULT1A1 proteins
Purified 6X-histidine-tagged recombinant SULT1A1*1 and recombinant SULT1A1*2 allozymes were generated in a baculoviral/insect cell system. SULT1A1 cDNAs were cloned into the pBlueBacHis2A expression vector (Invitrogen, Carlsbad, CA, USA). The His2A/SULT constructs were co-transfected with BacVector 3000 viral DNA (Novagen, San Diego, CA, USA) through liposome-mediated transfection into Sf-9 insect cells. After incubation at 27°C, viral supernatant was removed and diluted for isolation of individual viral clones. Clones with the highest protein expression were selected for amplification to a high-titer viral stock (>5 × 108 pfu/ml). His-tagged proteins were purified from high-titer viral stock (>5 × 108 pfu/ml) using cobalt-immobilized metal-affinity chromatography (Talon resin; Clontech, Palo Alto, CA, USA). Histidine tags were removed using the EnterokinaseMax serine protease (Invitrogen). Purified untagged recombinant SULT1A1 protein was dialyzed with 5 mM phosphate buffer (pH 6.5), and the total protein concentration was determined using the Bradford assay (Pierce, Rockford, IL, USA). Several aliquots of the purified recombinant SULT1A1 allozymes were stored at -80°C in the presence of 0.75 mg/ml BSA until assay. Purified SULT1A1 has been shown to be stable for at least 5 months under these storage conditions .
SULT1A1 biochemical assay
Recombinant SULT1A1 allozymes were characterized with regard to biochemical activity toward E2 using a standard SULT1A1 radiometric assay [46, 47]. One hundred nanograms of purified recombinant SULT protein was incubated with 10 μM 35S-labeled 3'-phosphoadenosine-5'-phosphosulfate (35S-PAPS), the biological sulfate donor, and varying concentrations of E2 (0–250 μM; Sigma, St. Louis, MO, USA) dissolved in dimethylsulfoxide (DMSO). The final reaction volume was 30 μl and included the standard reaction buffer containing 50 mM potassium phosphate (pH 7.4), 0.75 mg/ml BSA, 13 mM dithiothreitol and 7.5 mM MgCl2. The final concentration of DMSO in the reaction mixture was 3%. The reaction was initiated with the addition of 10 μM 35S-PAPS (NEN, Boston, MA, USA), and reactions were incubated for 15 min at 37°C and then quenched with the addition of 20 μl of a 1:1 mixture of 50 mM barium hydroxide/barium acetate. Unincorporated 35S-PAPS was precipitated from the reaction mixture by adding, sequentially, 10 μl of 0.1 M ZnSO4, 10 μl of 0.1 M (Ba)OH2, 10 μl of 0.1 M ZnSO4 and 80 μl H2O. The resulting precipitate was pelleted by centrifugation at 3220 × g for 10 min. The 35S-labeled reaction products were subsequently detected by liquid scintillation counting of 100 μl reaction supernatant. Each assay was performed in triplicate and 'blank samples' utilized DMSO as the vehicle control (final reaction, 3% DMSO).
The effects of substrate inhibition were largely avoided by measuring the initial reaction rates at very low substrate levels. To determine the appropriate working concentration range, a broad range of E2 concentrations were assayed initially (0–250 μM) and used to calculate initial estimates of Vmax (maximum reaction rate) and the Michaelis–Menten constant, Km (a measure of the affinity between substrate and enzyme measured as the substrate concentration at which half the maximal reaction rate is achieved). Assays were then repeated over a lower concentration range to fit data to the Michaelis-Menten equation to determine the apparent Vmax and Km values. Data were analyzed using GraphPad Prism 3.0b software (GraphPad Software, Inc., San Diego, CA, USA). Statistical significance was determined using the Mann–Whitney test.
Generation of stably transfected MCF-7 cells
Native SULT1A1*1 and SULT1A1*2 cDNAs were cloned into the pCR3.1 expression vector (Invitrogen). MCF-7 cells were cultured in RPMI 1640 media (Cellgro, Herndon, VA, USA) and were transfected with 5 μg pCR3.1 SULT1A1*1, SULT1A1*2 or control pCR3.1 plasmids using a standard calcium phosphate method, and were cultured for 48 hours. Cells were cultured in the presence of G418 over 8 days until clones became visible to the eye. Six clones for each transfection were isolated and expanded in complete RPMI 1640 media with 10% fetal bovine serum. Clones that expressed similar levels of SULT1A1 mRNA (SULT1A1*1 or SULT1A1*2 allele) were selected for comparison of allele-dependent differences in cell proliferation.
We have previously determined that the biochemical mechanism by which the SULT1A1*2 allozyme is associated with low activity includes both low intrinsic turnover of substrates by the enzyme (Fig. 2) as well as faster cellular degradation of the SULT1A1*2 protein such that SULT1A1*1 has a threefold longer cellular half-life than SULT1A1*2  (manuscript submitted). We have also confirmed that there is no difference in the cellular stability of the SULT1A1*1 versus SULT1A1*2 mRNA. We therefore selected clones for further proliferative analysis based on the expression of equal levels of mRNA to mimic the biological mechanisms contributing to pharmacogenetic SULT1A1 variability. Selected clones expressed equal levels of SULT1A1 mRNA, yet, as expected, western blot analysis revealed that the protein levels differed such that cells expressing SULT1A1*1 expressed threefold higher levels of SULT1A1 protein than cells expressing SULT1A1*2 (data not shown). Native MCF7 cells are heterozygous for SULT1A1*1/*2 and native levels of expression of the SULT1A1 protein are negligible compared with the levels expressed in the stable cell lines.
Cell proliferation assay
Cell proliferation was assessed by the alamarBlue assay (BioSource International, Camarillo, CA, USA). The alamarBlue assay incorporates a water-soluble fluorometric and colorimetric indicator that is nontoxic to cells and is biotransformed to a compound detectable at 570 nm at a rate dependent upon the cell number. Cells expressing equivalent amounts of SULT1A1*1 and SULT1A1*2 mRNA and a vector-control cell line (pCR3.1) were plated in triplicate in 96-well plates at 1000 cells per well. Cells were cultured in RPMI 1640 with 10% charcoal-stripped fetal bovine serum and were washed several times over 48 hours to remove endogenous estrogens. On day 2, cells were treated with varying concentrations of E2. Proliferation was assessed as the percentage reduction of alamarBlue on day 6 by spectrophotometric absorbance at 570 nm and 600 nm on a SpectraMax Plus (Molecular Devices, Sunnyvale, CA, USA). Data were analyzed using GraphPad Prism 3.0b software (GraphPad Software). Statistical significance was evaluated using the one-way analysis of variance test (GraphPad Prism 3.0b, GraphPad Software, Inc., San Diego, CA).
SULT1A1 genotyping assay
The SULT1A1 genotype was determined using a pyrosequencing-based assay. A 268 base pair (bp) fragment of the human SULT1A1 gene was PCR-amplified with the primers I6F395 (5'-biotin-GTTGAGGAGTTGGCTCTGCAGGGTC-3') and R733 (5'-GGGGACGGTGGTGTAGTTGGTCATAG-3'). Six nanograms of genomic DNA were mixed with TaqPlus Precision buffer (Stratagene, La Jolla, CA, USA) in a reaction mixture containing 50 mM dNTPs, 10 pmol PCR primers and 1 U TaqPlus Precision DNA polymerase in a 50 μl reaction volume. Cycling conditions included initial denaturation for 5 min at 94°C followed by 25 cycles of 94°C for 1 min, 68°C for 1 min initially with an incremental stepdown of 0.5°C on each cycle, and 72°C for 1 min. This was followed by 20 cycles of 94°C for 1 min, 56°C for 1 min and 72°C for 1 min. A final 10-min extension at 72°C completed the amplification. The presence of the PCR product was confirmed by agarose gel electrophoresis.
Amplicons were prepared for automatic pyrosequencing single nucleotide polymorphism (SNP) analysis on the PSQ 96 system (Pyrosequencing Inc., Westborough, MA, USA). Twenty-five microlitres of the double-stranded biotinylated amplicons were incubated with 100 μg streptavidin-coated DynaBeads in binding buffer (5 mM Tris, pH 7.6, 1 M NaCl, 0.5 mM ethylenediamine tetraacetic acid, 0.05% Tween 20) in a shaking thermal plate at 65°C for 15 min. Dynabeads and bound DNA were transferred to a 0.50 M NaOH solution to denature the DNA. The beads were washed and transferred to an annealing buffer (20 mM Tris acetate, pH 7.6, 5 mM Mg(OAc)2) for 1 min and then transferred to a sequencing primer solution containing annealing buffer and 10 pmol appropriate pyrosequencing primer.
For the G638A SNP the pyrosequencing primer was 5'-CCTCTGGCAGGGAG-3', and for the A667G SNP the primer was 5'-GAACGACGTGTGCTGAA-3'. The nucleotide dispensation sequences for the G638A and A667G SNPs were G TC AGCAC and ACATC AGAG, respectively (underlined nucleotides representing the negative control and bold nucleotides representing the polymorphic sites). The incorporation of homozygous nucleotides generated a luciferase signal with a peak height of 2X, while heterozygous nucleotides generated peak heights of 1×. The SULT1A1 genotype was assigned as follows: SULT1A1*1, G638 A667; SULT1A1*2, A638 A667; and SULT1A1*3, G638 G667. DNA samples with known SULT1A1 genotype were also evaluated as positive control samples.
UGT1A1 genotyping assay
UGT1A1 genotyping was performed using a PCR-based Genescan® (Applied Biosystems, Foster City, CA, USA) method. A segment of the UGT1A1 gene was amplified from genomic DNA by PCR using the primers F144887 (5'-TATCTCTGAAAGTGAACTC-3') and R175122 (5'-TAGTTGTCATAGAAGGGTC-3'). These primers flank the polymorphic TA locus in the promoter region of the UGT1A1 gene, and amplify a 256 bp fragment when a (TA)6TAA allele is present, a 258 bp fragment when a (TA)7TAA allele is present, a 260 bp fragment when a (TA)8TAA allele is present and a 254 bp fragment when a (TA)5TAA allele is present. The forward primer was labeled with a fluorescent dye (6-carboxyfluorescein) at its 5'-end to permit detection of the amplified product. The amplification reaction (25 μl reaction volume) included 20 ng total human genomic DNA as template, 0.2 μM each primer, 50 μM each dNTP, 1.5 mM MgCl2 and 0.5 U TaqPlus Precision DNA polymerase. Reaction conditions included initial denaturation for 4 min at 95°C, followed by 35 cycles of 95°C for 30 s, 52°C for 30 s and 72°C for 30 s, followed by a final extension at 72°C for 2 min. The products were visualized by agarose gel electrophoresis.
PCR fragments were subjected to gel electrophoresis on an ABI 377 DNA analyzer (Perkin-Elmer, Wellesley, MA, USA). Amplified products were diluted in water with the addition of formamide and dextran blue loading buffer combined with a size standard (GS-350; Perkin-Elmer), were denaturated at 95°C and were loaded onto a 4% denaturing polyacrylamide gel. Fluorescent bands were analyzed using GENESCAN 2.1 software (Applied Biosystems) to determine the fragment length. Genotypes were assigned as UGT1A1*1, UGT1A1*33, UGT1A1*28, and UGT1A1*34 for six, five, seven and eight TA repeats, respectively. DNA samples with a known UGT1A1 genotype were also evaluated as positive control samples.
The chi-squared test was used to evaluate differences in allele frequencies between Caucasians and African-American subjects. Allele frequencies were not evaluated for the rest of the group (Hispanics and Asians) because of the small size of that group (11 samples; Table 1). For genotype–phenotype association analyses, genotypes were grouped based on known biological function of the alleles. SULT1A1 was grouped as follows: high activity, SULT1A1*1/*1; low activity, SULT1A1*2/*2; and intermediate or unknown function, SULT1A1*1/*2, SULT1A1*1/*3, SULT1A1*2/*3 and SULT1A1*3/*3. UGT1A1 genotype groups were grouped as: high activity, UGT1A1*1/*1, UGT1A1*1/*33 and UGT1A1*33/*33; low activity, UGT1A1*28/*28 and UGT1A1*28/*34; and intermediate or unknown function, UGT1A1*1/*28, UGT1A1*1/*34, UGT1A1*33/*28 and UGT1A1*33/*34.
We evaluated the associations between genotypes and tumor characteristics, including tumor size, tumor grade and age at diagnosis. A stepwise approach was used to identify genotype groups with a statistically significant association with tumor phenotypes. For each phenotype of interest, the chi-squared test was applied first to all genotype groups of a single gene. Genotype groups were subsequently evaluated as dichotomous variables such that high-activity genotype groups and, separately, low-activity genotype groups were tested versus all other groups.
Phenotypes were analyzed categorically. The age at diagnosis was evaluated as <60 years versus ≥60 years. This cutoff point was selected based on the mean age at diagnosis of 59.8 years with a range of 43–80 years. Tumor sizes were categorized based on those values known to be critical for prognosis and treatment strategy. Tumor size categories were ≤2 cm versus >2 cm. The tumor grade was analyzed using categories commonly applied during pathological evaluation, including grades 1, 2 or 3. Statistical analysis for the association between the genotype and the dichotomized tumor grade was performed such that grade 1 tumors were evaluated versus grade 2 and grade 3 tumors.
Logistic regression was used to estimate the ORs and 95% CIs for association of SULT1A1 and UGT1A1 genotypes with categorized tumor phenotypes, followed by a chi-squared test. ORs for age at diagnosis were adjusted for race (Caucasian versus non-Caucasian) and those for other phenotypes were adjusted for race and age at diagnosis. We used race-adjusted and age-adjusted analyses rather then race-stratified and age-stratified analyses because of the sample size limitation. Analysis with simultaneous adjustment for age at diagnosis, tumor size and grade was not performed, again because of sample size limitations. ORs, CIs and chi-squared P values were estimated for all subjects, among the group of subjects with ER-positive tumors and among the group of subjects with ER-negative tumors. The chi-squared test was used to estimate whether phenotype variables were distributed independently of each other. All analyses were undertaken using Minitab™ Statistical Software (Release 13.20; Minitab Inc., State College, PA, USA).