Distribution and Functional Consequences of Nucleotide Polymorphisms in the 3′-Untranslated Region of the Human Sep15 Gene

INTRODUCTION

Selenium is an essential trace element currently being considered as a candidate chemopreventive agent. Research spanning the last 25 years has established that selenium is effective in the reduction of cancer incidence when provided to animals as doses only 5- to 10-fold above the nutritional requirement (1) . Chemoprevention studies with selenium in animal model systems have indicated that this element is effective for most, if not all of the organ systems and is protective against the carcinogenic effects of a wide variety of insults (1) . Both epidemiological studies and supplementation trials have also supported its efficacy in humans (2, 3, 4, 5) . However, the mechanism by which selenium suppresses tumor development remains unknown. Many of the biological activities of selenium are attributable to its role as a constituent of selenium-containing proteins. In all of the known mammalian selenoproteins, selenium exists as the amino acid selenocysteine, which is inserted cotranslationally into the nascent protein in response to the UGA codon in selenoprotein mRNA (6) . In mammals, the molecular choice to insert selenocysteine into the elongating selenoprotein rather than terminate translation requires a recognition element within the 3′-UTR 3 (3) of the corresponding mRNA referred to as a SECIS element (7) . SECIS elements are present in the mRNAs of all of the selenocysteine-containing proteins and share structural features, including an approximately 80-nucleotide stem-loop structure containing both an internal and apical loop of unpaired nucleotides and a conserved 4-bp domain within the stem. Although the functions of several selenoproteins have been determined, to date there has not been definitive evidence associating any of these proteins with either cancer risk or development. A recent search for novel selenium-containing proteins resulted in the isolation and characterization of a new selenoprotein (8) . Examination of data contained in the dbEST databases resulted in the assembly of a 1268 nucleotide cDNA with an ORF of 162 amino acids. Within this ORF was an in-frame UGA codon corresponding to amino acid 93. This and the observation that this protein labels with ⁷⁵Se and contains a SECIS element in the 3′-UTR strongly support the conclusion that the M_r 15,000 selenoprotein (Sep15) contains selenocysteine at position 93. Further analysis of the Sep15 cDNA sequence indicated that it was well conserved through evolution; homologous genes exist in mouse, rat, Brugia malayi, Caenorhabditis elegans, and other animals. On the basis of the frequency of recovery of Sep15 cDNA sequences from particular libraries included in the dbEST database, it was concluded that Sep15 is expressed in a wide variety of human tissues, with the highest levels being observed in the thyroid and prostate (8 , 9) . Examination of the available cDNA sequences for this protein revealed two distinct forms of the Sep15 gene differing by polymorphisms located at position 811 (C/T) and at position 1125 (G/A) within the 3′-UTR (8) . Here, we demonstrate significant differences in Sep15 allele frequencies by ethnicity and that the identity of the nucleotides at the polymorphic sites influences SECIS function in a selenium-dependent manner. This, together with genetic data indicating LOH at the Sep15 locus in certain human tumor types, suggests that Sep15 may be involved in cancer etiology.

MATERIALS AND METHODS

Sample Sources.

The majority of the samples obtained from African Americans, 490 subjects, was recruited from the community of Maywood, IL, a suburb of Chicago. Unrelated individuals over the age of 18 years were identified through door-to-door canvassing and asked to participate in a study of diet and heart disease from an IRB-approved protocol, and signed consent was given from all of the individuals. Exclusion criteria included any known cancer, non-insulin-dependent diabetes mellitus, and heart disease. The lymphocytes from normal volunteers and colon tumor samples were collected anonymously from an IRB-approved protocol. All subjects were >18 years of age. Normal Caucasian volunteers were recruited from the University of Chicago. Colon tumor samples were obtained after surgical resection from patients with confirmed colon cancer. African-American patients with hematological malignancies were treated according to institutional IRB-approved treatment studies for acute myeloid leukemia, acute lymphoblastic leukemia, and chronic myeloid leukemia. All subjects were >18 years of age, were referred to the University of Illinois at Chicago for treatment, and consented to an IRB-approved study for the analysis of molecular abnormalities present in leukemia cells. Bone marrow and blood samples were obtained at specified time points before, during, and after completion of treatment. Breast carcinomas from African-American women were archived from the Department of Surgical Oncology Tissue and Sera Bank at the University of Illinois at Chicago. Originally, tumors were obtained from women with confirmed histopathological diagnosis of breast cancer who were undergoing lumpectomy or mastectomy. After surgical excision, tumor tissue was transported to the Department of Pathology, an aliquot of the tissue required for the diagnostic purpose was reserved, and the remaining tumor tissue, otherwise to be discarded, was transported on ice to the Surgical Oncology Laboratory. Tumor was freed from nonmalignant tissue, allocated, and frozen at −70°C. Only those women with accurate information of their race were included in the study. Women with more than one primary cancer were excluded.

Determination of the Nucleotide Identity at Positions 811 or 1125 within the Human Sep15 Gene.

DNA from either lymphocytes or tissue was prepared using the protocols and procedures included in the Puragene DNA Isolation Kit (Gentra). The buffy coat plus red cells were isolated from a 10-ml draw into EDTA-containing vacutainers. Tubes were immediately placed in a refrigerator and centrifuged between 30 min and 2 h postdraw. Plasma was drawn off and stored, and white cells plus 0.5–1.0 ml of red cells were stored in 2-ml cryovials and frozen at −80°C until DNA extraction. Remaining red cells were discarded. From 0.1 to 1.0 μg of DNA was used as template for PCR amplification using the GeneAmp PCR Amplification Kit. Amplification of a 413-bp product was achieved with the Sep15 forward primer (5′-CAGACTTGCGGTTAATTATG-3′) and the Sep15 reverse primer (5′-GCCAAGTATGTATCTGATCC-3′). PCR was performed at 95°C for 3 min, followed by 33 cycles of 95°C for 30 s, 55°C for 60 s, and 72°C for 90 s. Successful amplification was indicated by the appearance of a DNA band of ∼400 nucleotides on a 1% agarose gel. Radioactive primers for use in the determination of the nucleotide at position 811 (5′-GGCATAGTAATCATCTGTCTTGTT-3′) or 1125 (5′-GTATGTATCTGATCCACACAAATCC-3′) were obtained by 5′-end labeling with γ-labeled ATP and polynucleotide kinase. Labeled primer and DNA were mixed in a solution of 40 mm Tris-HCl (pH 7.5), 20 mm MgCl₂, and 50 mm NaCl, heated to 95°C for 10 min, and then transferred to a 37°C water bath for 1 h. Extension was accomplished by adding 5 μl of a mix containing 100 mm DTT, 1 mm each of 3 dideoxynucleotide triphosphates, 5 mm dideoxynucleotide triphosphate, and 5 units of reverse transcriptase. The mixture was incubated for 15 min at 42°C, ethanol precipitated, and resuspended in formamide-loading buffer, and the extension products separated on a 10% polyacrylamide gel. Visualization of the extension products was accomplished by autoradiography. Nucleotide identity at positions 811 and 1125 was also determined by restriction enzyme digestion of the PCR amplification product described above. Amplified DNA was digested with DraI (Pharmacia) to evaluate the nucleotide identity at position 811 or with BfaI (New England Biolabs) to identify the nucleotide at position 1125 using buffers and conditions provided by those respective vendors. DNA digestion was evaluated by gel electrophoresis in 2.5% agarose. Statistical analysis was performed using Statistical Analysis Systems (SAS version 6.0). As all of the data are categorical, and differences between cases and controls for the three genotypes were measured using χ² statistics. Genetic analysis at the D1s207 and D1s223 loci was performed by Research Genetics, Inc. (Huntsville, AL).

In Vitro Mutagenesis.

The plasmid containing the Sep15 cDNA (AF051894; Ref. 8 ) was mutated with two pairs of primers, where the boldface nucleotide indicates the site of the mutation, as follows: 1125F, 5′-CACAGAAAACCTTTCTAAGGATTTGTGTGGATCAG-3′; 1125R, 5′-CTGATCCACACAAATCCTTAGAAAGGTTTTCTGTG-3′; and 811F, 5′-CCAGTTTTACGAACAACAGATTTTTAAATTAGAGAGGTTAAC-3′; 811R, 5′-GTTAACCTCTCTAATTTAAAAATCTGTTGTTCGTAAAACTGG-3′, using the QuickChange site-directed mutagenesis kit (Stratagene). Mutagenesis with primers 1125F/1125R yielded the CA (811:1125) variant of the Sep15 SECIS element, and mutagenesis with 811F/811R, the TG (811:1125) variant.

Reporter Constructs and Assays.

To generate pBPLUGA plasmids (10) containing a portion of the Sep15 3′-UTR, the 1535 CPTX human prostate cell line (11) and the STSAR90 human sarcoma cell line (12) were selected because their DNA contained either the CG or TA haplotype determined as described above for human samples. DNA from these lines was prepared, and the 3′-UTR was amplified with primers containing the recognition sequences for either SpeI or PstI (5′-AAAACTAGTGCTTTGTAACAGACTTGCGGTTAATTATGC-3′ or 5′-AAAGTGCAGGGTCTTACAAATGATCACTTTTAAATGGAC-3′, respectively). Amplification was as described above, and the products were digested with SpeI and PstI and directionally cloned into the pBPLUGA plasmid. Transfection of mouse NIH 3T3 cells was performed using the lipofectin reagent at a ratio of 1:3 (v/v) as indicated by the vendor (Life Technologies, Inc.). The cells were cotransfected with the reporter construct and the pSV₂Neo plasmid at a 10:1 molar ratio, and transfectants were selected with the antibiotic geneticin G418. G418-resistant colonies were pooled for further analysis. Selenium was added to the culture media in the form of sodium selenite for 5 days, at which time the cells were collected and lysates prepared. Cells were resuspended in Reporter Lysis Buffer (Promega) and lysed by a single freeze-thaw cycle. Lysates were generated by suspension in Reporter Lysis Buffer and were either used immediately or stored at −70°C. Luciferase and β-galactosidase activities were determined using the corresponding assay systems from Promega. Luciferase activity was quantified using a Fentomaster FB12 Luminometer from Zylux. Readthrough efficiencies were analyzed by standard t test performed using SYSTAT version 8.0 software.

RESULTS

Genotype of DNA from Human Samples.

To determine the distribution of the Sep15 genetic variations within the human population, we used PCR to amplify a region of ∼400 nucleotides of the human Sep15 3′-UTR that included positions 811 and 1125. In initial studies, we established the identity of the nucleotides at the polymorphic sites using primer extension. Subsequently, we took advantage of the observation that position 811 is potentially located within the recognition sequence for the restriction endonuclease DraI (recognition sequence 5′-TTTAAA-3′). C at position 811 results in the sequence TCTAAA, which is not cleaved by DraI, whereas T at position 811 results in the recognition sequence of the enzyme. To validate the use of DraI digestion, 100 samples were analyzed by both primer extension and DraI digestion of PCR-amplified DNA with perfect agreement. In addition, the results of the primer extension and DraI digestions indicated that the only nucleotides at position 811 were either a T or C. To extend the analysis of DNA samples to position 1125, we used differential digestion using BfaI (recognition sequence, 5′-CTAG-3′). G at position 1125 results in the sequence 5′-TTTCTAGCCTAA-3′, which is cleaved by BfaI, whereas A results in a sequence not recognized by this enzyme.

Using the techniques described above, we genotyped DNA samples from >700 individuals. All of the samples analyzed for nucleotide identity at positions 811 and 1125 fell into one of two categories referred to as either the TA or CG haplotype. This result was in agreement with previous observations made by the analysis of EST sequences (8) , although the possibility of compound heterozygosity was never addressed in these studies. Unlike the database analysis, the methodology used in this study permitted the assignment of DNA samples from individuals as either TA or CG homozygotes or heterozygotes, the later being indicated by the presence of bands representing both alleles after restriction enzyme digests of PCR amplification products (e.g., Fig. 1A ⇓ ). The genotypes for all of the samples analyzed are shown in Table 1 ⇓ and exhibit a classic 1:2:1 distribution, following the Hardy-Weinberg equilibrium. However, when the data were subdivided by ethnicity, it was apparent that there was a difference in allele distribution between DNAs obtained from Caucasians and African Americans (Table 1) ⇓ ; e.g., the allele frequency for the TA/TA homozygous haplotype was 31% for African Americans, in contrast to 7% for Caucasians. No statistically significant differences in genotype distribution were observed for sex or age (data not shown).

Nucleotide Identity at Position 1125 Influences SECIS Function.

Position 1125 resides within the apical loop of the Sep15 SECIS element and thus could potentially influence the efficiency of Sec incorporation (8) . Because of this and the allele distribution data described above, we used a reporter construct designed to quantify the ability of a SECIS element to direct the readthrough of an in-frame UGA codon positioned between a β-galactosidase (β-gal) gene into a downstream luciferase (luc) gene (10) . Reporter constructs containing the 3′-UTR sequences representing either the TA or CG haplotype were cotransfected into NIH 3T3 mouse fibroblasts along with the pSV₂neo plasmid; individual clones were isolated and both β-gal and luc activities were determined in cell extracts. Although both polymorphic sequences were indeed capable of functioning as a SECIS element, the TA sequence was approximately twice as efficient in stimulating the readthrough of the UGA codon, as measured by luc activity (Fig. 2) ⇓ . To determine whether both the 811 and 1125 nucleotide positions were contributing to the observed SECIS activity, two additional sequences (811:1125, CA or TG) were generated by in vitro mutagenesis and evaluated using the same reporter system. As seen in Fig. 2 ⇓ , the CG and TG pairs yielded similar activity as did the CA and TA pairs, indicating that the identity of the nucleotide at the 1125 position determines SECIS efficiency. It was also evaluated whether the naturally occurring haplotype sequences could influence SECIS efficiency as a function of increasing selenium availability in the tissue culture media. Stably transfected NIH 3T3 cells were incubated in sodium selenite-supplemented media at concentrations between 30 nm and 120 nm, and reporter gene activity was determined. As seen in Table 2 ⇓ , the SECIS element containing T,A:811,1125 was less responsive to added selenium than was the CG-containing version of this sequence.

LOH at the Sep15 Locus.

The DNAs genotyped in Table 1 ⇓ were derived either from lymphocytes obtained from cancer-free individuals or from malignant cells. We, therefore, determined whether allele frequencies at this locus were different for tumors versus lymphocytes from cancer-free individuals of a particular ethnic background. These data (Table 1) ⇓ indicate that there was a significant reduction in heterozygosity in tumors in African Americans, and this was true for both classes of tumors. Although there appears to be a trend toward LOH in head and neck tumors in Caucasians as well, the sample size was too small to permit the statistical analysis to reach significance.

To evaluate whether LOH at the Sep15 locus could be detected during tumor development, we obtained lymphocytes from three patients whose tumor DNAs genotyped to an apparently homozygous TA/TA haplotype. As seen in Fig. 1B ⇓ , a single DNA pair demonstrated LOH at this locus. The Sep15 gene has been previously mapped to human chromosome 1 at the 121.9 interval (8) . To verify LOH determined by restriction enzyme analysis and to assess the extent of the sequences deleted in this tumor sample, we used highly polymorphic markers in the region of the Sep15 gene. LOH was verified using the D1s207 and D1s223 markers located at positions 117.6 and 137.3, respectively. Both markers were heterozygous in lymphocyte DNA, and only one allele was evident in the DNA obtained from the tumor.

DISCUSSION

The wealth of data indicating that selenium can reduce cancer incidence in animal models and humans has stimulated the search for selenoproteins that may contribute to the beneficial properties of this element. The observation that the Sep15 gene exists in humans as two allelic forms differing at nucleotide positions 811 and 1125 permitted the evaluation of allele frequencies using the PCR amplification of the polymorphic region, followed by diagnostic restriction enzyme digestion. The examination of a large number of DNA samples clearly indicated that there was a significant difference in allele frequencies between African Americans and Caucasians. Therefore, subsequent analysis of the distribution of polymorphic sequences within the Sep15 gene required the sorting of comparative genetic data on the basis of ethnicity. When this is done, differences in allele frequencies between DNAs obtained from Caucasian, cancer-free individuals and the three classes of cancer examined in this study failed to indicate a statistically significant difference. However, the relatively low sample sizes may have precluded the detection of small differences in allele distribution. In contrast, the compiled genetic data did indicate a difference in allele frequencies obtained from tumors of either breast or head and neck origin when compared with cancer-free controls among African Americans. Furthermore, LOH was detected in a single instance when DNA from circulating lymphocytes of an African-American patient with a T₃N_2C supraglotis tumor was compared with the DNA obtained from that tumor. LOH is a common genetic event in cancer development and is often associated with the loss of protective genes referred to as “tumor suppressors” (13, 14, 15) . The data indicating LOH at the Sep15 locus in these tumor types suggests that either the loss of the Sep15 gene or perhaps some other gene in its vicinity on chromosome 1 contributes to the development of certain malignancies. LOH in this region has been described for several tumors, including breast cancer as well as head and neck cancers (16 , 17) .

In addition to potentially being involved in tumorigenesis, allelic identity of the Sep15 gene may also be associated with cancer risk. Using a specialized reporter construct, it was determined that the 3′-UTR region including the polymorphic sequences can indeed function as a SECIS element, supporting the translation of an in-frame UGA codon situated upstream from a luc gene. Using sequences representing both the naturally occurring alleles as well as variants generated by in vitro mutagenesis, it was examined whether the nucleotides at positions 811 and 1125 could influence the efficiency of UGA readthrough. This analysis indicated that the nucleotide at position 1125 located within the apical loop of the SECIS element could in fact influence SECIS function. It is particularly interesting to note that an A at position 1125 resulted in the SECIS element being less responsive to the stimulation in activity observed when selenium was added to the tissue culture media compared with the same sequences containing a G at that position. Because tissue culture media is generally considered to be deficient in selenium compared with levels obtained in human tissues and plasma (18) , the differences observed in SECIS function under conditions of selenium supplementation might suggest that a Sep15 gene with an A at 1125 may produce less of its protein product. Additionally, individuals with that genotype might be less responsive to the protective benefits of dietary selenium. The development of antibodies capable of quantifying Sep15 in the relevant tissues will be required to quantify this protein in human tissues to address this possibility.

Although limited conclusions can be drawn from an uncontrolled retrospective study, the observed differences in allele frequencies and the single example of LOH at the Sep15 locus presented suggest that this protein may play a role in tumor evolution. In addition, the data presented indicating a functional consequence of the nucleotide identities at the polymorphic positions raise the possibility that the Sep15 genotype may influence translation of that protein in response to dietary selenium and whether individuals can benefit from the chemoprotective properties of this element.