Expression in Normal Human Tissues of Five Nucleotide Excision Repair Genes Measured Simultaneously by Multiplex Reverse Transcription-Polymerase Chain Reaction¹

NER³ is one of the primary pathways by which mammalian cells remove DNA lesions caused by both endogenous and exogenous agents (1, 2, 3, 4). A wide spectrum of structurally unrelated lesions such as UV-induced photoproducts, bulky chemical adducts, and certain types of DNA cross-links are efficiently removed by the NER pathway (5, 6). In the process of repair, the products of more than a dozen genes are involved in damage recognition, incision, excision, elongation, and ligation and collectively restore the normal structure (1, 2, 6). Therefore, efficient DNA repair plays a central role in the survival of cells upon exposure to carcinogens that cause damage to DNA and the fidelity of subsequent replication of the genome (7, 8, 9, 10, 11).

Consequences of defective NER are well illustrated in several genetic disorders such as XP and CS (12). Although many genes participate in NER, there are overlaps in their functions. Cell-fusion studies have helped identify at least 11 DNA-repair complementation groups (each group having only one defective gene) in cell lines established from patients with these diseases and a large collection of UV-sensitive rodent cell mutants. A number of genes that correct defective human NER have been cloned and designated as ERCC genes (13, 14, 15). Some of these ERCC genes are involved in the human NER disorders. For example, ERCC2, ERCC3, and ERCC5 are identical to XPD (16), XPB (17), and XPG (18, 19), respectively, and mutations in these genes are involved in corresponding groups of XP. ERCC6 is identical to CSB, and mutations in this gene are involved in CS (20). However, ERCC1 has not been found to be involved in any XP, CSB, or trichothiodystrophy human syndromes (21), because defects in ERCC1 resulting from mutations or deletions of this gene cause early death before the symptoms develop (22).

It is technically feasible to measure the relative transcript levels of several genes at a time by a multiplex RT-PCR assay (23). Physiologically, a high level of expression should be correlated with enhanced DNA repair activities. Indeed, we have demonstrated that the level of ERCC1 expression correlates with NER activity (23). We also found that the expression levels of mismatch repair genes vary among individuals (24). To assess the variation in expression of multiple NER genes in different types of normal tissue, we conducted this study with the following aims: (a) to design a multiplex RT-PCR assay for measuring the relative mRNA expression of a panel of NER genes; (b) to describe the expression patterns or differences in various types of tissue; and (c) to see whether the expression in lymphocytes is comparable with that in other target organ tissues so that the lymphocytes can be used as a potential surrogate tissue. We report here that we simultaneously analyzed the relative expression of ERCC1, XPB/ERCC3, XPG/ERCC5, CSB/ERCC6, and XPC in 20 types of selected human normal tissue by using a new multiplex RT-PCR assay.

cDNAs from 18 types of human normal tissue were purchased from Invitrogen (Invitrogen Corp., San Diego, CA). The other two types of tissue were primary lymphocytes and skin. Additional cDNAs were synthesized from total RNA extracted from PHA-stimulated peripheral blood lymphocytes of 12 healthy subjects and snap-frozen skin biopsy samples (3 mm in diameter) from four healthy donors. All of the tissues used were histopathologically confirmed to be normal. Total RNA was extracted by using Tri-Reagent (Molecular Research Center, Cincinnati, OH) as described previously (25). The cDNA was synthesized by reverse transcriptase of 5 μg of the extracted total RNA in 20 μl of first-strand synthesis cDNA reaction mixture. Each reaction contained 0.5 μg of random primers (Promega Biotech, Madison, WI), 200 units of Molony murine leukemia virus reverse transcriptase (United States Biochemical, Cleveland, OH), 4 μl of 5× RT buffer [250 mm Tris-HCl (pH 8.3), 375 mm KCl, 50 mm DTT, 15 mm MgCl₂ (United States Biochemical), 0.5 mm each deoxynucleotide triphosphate, and 20 units of RNasin (Promega Biotech)] and diethylpyrocarbonate-treated water. The reaction mixture was then incubated at room temperature for 10 min and then at 42°C for 45 min, heated to 95°C for 10 min, and then quick-chilled on ice.

Four EBV-immortalized human lymphoblastoid cell lines from the Human Genetic Mutant Cell Repositories (Camden, NJ) were used: three apparently normal cell lines (GM00892B, GM03798, and GM00131A) and two XP cell lines, GM02246b (XP-C) and GM02345a (XP-A), with deficient nucleotide excision repair. All of the cells were cultured in T-25 flasks at 37°C in a 5% CO₂ atmosphere in the standard medium RPMI 1640 (Life Technologies, Inc., Grand Island, NY) supplemented with 15% fetal bovine serum (Life Technologies, Inc.) without antibiotics.

To amplify the five NER genes, we used a modification of our previously described multiplex RT-PCR technique (23). To select the primers, we used GenBank Sequence Data Library mRNA or “cDNA” or genomic DNA sequences for ERCC1 (GenBank accession no. M13194), XPB/ERCC3 (M31899), XPG/ERCC5 (L20046), CSB/ERCC6 (L04791), XPC (X65024), and β-actin (M10277), which we accessed through the Genetic Computer Group Genetic Database (version 8.0). The optimal primers were selected with Oligo Software (version 3.4; National Biosciences, Inc., Plymouth, MN). The primers for the PCR were designed to minimize base-pairing interaction between the primers and to have similar annealing temperatures to optimize the reaction. All of the primers were commercially synthesized by Life Technologies, Inc. (Table 1). The genes were chosen because they were relatively easier to be coamplified in our hands than other genes we tested (data not shown). For instance, we were unable to design an effective pair of primers for coamplification of XPD/ERCC2; therefore, we did not include this NER gene in the study. To optimize the multiplex PCR reaction, the primers were chosen to prevent nonspecific annealing and primer dimerization and to have similar annealing temperatures.

Each 40 μl of PCR contained 3 μl of RT mixture, 1× PCR buffer [500 mm KCl, 100 mm Tris-HCl (pH 9.0), 1% Triton X-100, and 2.5 mm MgCl₂], 0.1 mm each deoxynucleotide triphosphate, 2 unit of Taq polymerase (Promega Biotech), 75 pm β-actin primers, 35 pm ERCC1 primers, 50 pm XPB/ERCC3 primers, 50 pm XPG/ERCC5 primers, 50 pm CSB/ERCC6 primers, and 25 pm XPC primers. The reaction mixtures were heated at 95°C for 5 min. Amplification was performed in sequential cycles of 95°C for 30 s, 59°C for 30 s, and 72°C for 45 s. After 29 cycles of amplification (chosen from the PCR kinetic analysis; see “Results”), all samples were incubated for an additional 10 min at 72°C. The β-actin PCR fragment was used to monitor genomic DNA contamination. Because all genes were amplified in the same test tube, using only one control for DNA contamination was sufficient. The 621-bp β-actin fragment spans exon 3 through exon 5, containing intron 4 (95 bp) and intron 5 (112 bp), which should not be present in the cDNA. If the cDNA was contaminated with genomic DNA, the PCR product would be a 828-bp (i.e., 621 + 95 + 112 bp) band, which would be distinguished by agarose gel electrophoresis.

The assays were performed three times for the purchased cDNA samples and performed in triplicate for the skin biopsies and lymphocytes. The PCR products were separated by 2% agarose gel electrophoresis, stained with 0.5 μg/ml ethidium bromide, and visualized with UV light. To confirm that the PCR products were copies of the target sequences, each target gene was amplified separately as well as in the multiplex reaction. The sizes of the multiplex RT-PCR products were determined by electrophoresis with size marker φX174RF DNA/HaeIII (Life Technologies, Inc.). Then, each product was purified with a Centricon Concentrator (Amicon, Beverly, MA) according to the manufacturer’s instructions. The sequence of each PCR product was then confirmed by direct sequencing with an automated Model 373A Sequencer (Applied Biosystems, San Francisco, CA).

To quantify the relative levels of gene expression, the PCR products were electrophoresed, stained with ethidium bromide, and scanned with a Digital Imaging System (Model IS-1000; Alpha Innotech Co., San Leandro, CA), and the areas of the peaks were calculated in arbitrary units. The expression level of the internal standard (β-actin) in each reaction was used as the baseline expression (100%) of that sample, and the relative value (percentage of baseline) was calculated for each of the target genes amplified in the same reaction.

The relative expression levels of five NER genes in tissues, the means, and SDs were calculated by using the SAS statistical software package (version 6.11; SAS Institute, Inc., Cary, NC).

Amplification of a specific fragment of a target gene is often compromised by nonspecific reactions resulting from sequence homology between targeted and stand-by sequences in the genome. As a result, multiple unidentifiable bands, in addition to the expected band, are generated by PCR. This is particularly true when multiple pairs of primers are used in a single PCR. We have demonstrated previously that in our hands, targeted sequences of five mismatch repair genes can be coamplified with β-actin gene in a single reaction (23). In this study, optimal conditions were achieved in the multiplex RT-PCR of five NER genes by comparing the PCR results of a single pair with those of multiple pairs of primers used in the PCRs. As shown in Fig. 1,A, each gene was amplified with the β-actin gene only (Lanes 1–5), and all five NER genes were also amplified with the β-actin gene in a single PCR (Lane 6). In addition, the expression of five genes was detectable and similar in several lymphoblastoid cell lines (Fig. 1 B), except for low expression of XPCC in XP-C cells. These results indicated that the primers designed and the PCR conditions were optimal, and the multiplex RT-PCR generated the same bands as did a single pair of primers in a single reaction.

To determine the linear range of simultaneous amplification of six genes, we performed a PCR kinetic analysis. The same amount of cDNA was aliquoted into each of seven tubes, one of which was then removed from the cycle at each of the following cycles: 25th, 26th, 27th, 28th, 29th, 30th, and 31st. A representative experiment was shown in Fig. 2, in which the relationship between the number of cycles and the product yield was approximately linear between 25 and 30 cycles, but PCR amplification was saturated at the 30th cycle. Therefore, we arbitrarily chose 29 cycles as the cycle number for later experiments, as described previously (23).

To determine the relationship between the initial amount of template and the yield of the PCR product of each gene, a serial dilution of cDNA containing reverse transcriptase mixture of lymphoblastoid cell line GM00131 (0.3125, 0.625, 1.25, 2.5, 5, and 10 μl) was prepared and amplified under the conditions described above. The dose-amplification curve for each gene indicated that 0.3125–10 μl of cDNA reaction mixture allowed the amplification of all genes simultaneously within an approximately linear range (Fig. 3). On the basis of these results, it is clear that by optimizing concentrations of primers used in the multiplex RT-PCR, we achieved similar efficiencies of amplification for all six NER genes. An approximately linear (dose-dependent) amplification of the six genes was also achieved with 29 amplification cycles, avoiding reaching the plateau of amplification. Given fixed 29 cycles and the amount of cDNA in a single PCR, the relative expression levels of the NER genes were then calculated by comparison with the level of the housekeeping gene, β-actin. It was noted that the relative expression level of each NER gene did not have a dose response as did the absolute level, i.e., they appear to be independent of the amount of cDNA used (Fig. 3). This feature is important for such relative quantification, because the measurement will be less likely biased by the variation in the amount of cDNA used for PCR amplification.

The expression levels of the five NER genes from 18 Invitrogen tissues plus the skin and PHA-stimulated peripheral blood T lymphocytes are listed in Table 2. The means and SDs were calculated from three repeated PCR assays performed on the same cDNA samples, which provided information on intraassay (individual) variation. The mean expression of PHA-stimulated T lymphocytes from 12 individual blood samples and the mean expression of four skin biopsies were used for comparison, which provided information on interassay (individual) variation. In general, the interassay variation, particularly for lymphocytes, was larger than intraassay variation. All 18 Invitrogen tissues plus skin had detectable expression levels for all five NER genes. Fig. 4 shows the results of one experiment with 13 Invitrogen tissues, plus peripheral blood T lymphocytes. It appears that PHA-stimulated and unstimulated lymphocytes had comparable expression levels of the five NER genes. Most tissues had similar patterns of expression levels, whereas the brain and spleen had relatively low expression levels, which may be due to the quality of the cDNA. Compared with the expression in PHA-stimulated T lymphocytes, ERCC1 had a similar or higher expression and XPB/ERCC3 had a lower expression in most of the tissues tested. NER gene expression was relatively higher in most of the proliferating tissues (such as skin, breast, intestine, liver, testis, ovary, placenta, and prostate) than in most of the nonproliferating or slowly proliferating tissues (adipose tissue, brain, hippocampus, muscle, spleen, and lung; Table 2). The Invitrogen cDNA samples were obtained from individuals of different ages and sex (Table 2), and statistical comparisons of these expression levels across different tissues suggested significant differences. However, the comparisons of expression levels between different tissues from different subjects as well as multiple tests with an arbitrary reference may not be valid and, therefore, were not presented.

We have developed a new multiplex RT-PCR assay for simultaneous amplification of five NER genes, ERCC1, XPB/ERCC3, XPG/ERCC5, CSB/ERCC6, and XPC. We found that the expression of these five NER genes was detectable but variable in the 20 types of normal human tissue including lymphocytes. These results provided evidence that the genes were actively transcribed in all of the tissues tested, which is consistent with the notion that the NER pathway is central to mammalian cells (1, 2, 26, 27). In addition, the expression level in lymphocytes may be used as a surrogate measurement for that in other tissues, if the expression level is genetically determined.

Because the output of the RT-PCR assay depends on the number of amplification cycles and starting copies of the templates in the cDNA sample, the arbitrary selection of 29 cycles and the unmeasured starting amount of the target gene transcript may render the quantification inaccurate. Therefore, the measurement may be considered as semiquantitative, and the assay may be used to approximate changes in gene expression relative to β-actin.

Because the Invitrogen tissues were obtained from different individuals, the variation of the five NER genes in the tissues examined in this study may be due to either interindividual variation or intertissue variation. The inter- and intraindividual variations need to be investigated in future studies using the tissues from the same individuals. PHA-stimulated T lymphocytes were chosen in this study for comparison because studies have shown that although unstimulated lymphocytes do have detectable levels of DNA repair enzymes (28, 29, 30, 31), they have limited NER activity (32). We have confirmed that the expression (mRNA) of proliferating cell nuclear antigen, a cell cycle-dependent and essential element of DNA repair, was detectable in PHA-stimulated lymphocytes but not in unstimulated lymphocytes.⁴ This is consistent with the report that proliferating cell nuclear antigen is only expressed in proliferating cells and is required for normal NER activity (33).

The relative expression levels of ERCC1 were consistently higher (but not for XPB/ERCC3, XPG/ERCC5, CSB/ERCC6, or XPC) in most Invitrogen tissues and the skin. This may reflect the fact that ERCC1 plays an important role in NER. ERCC1 is involved in the repair of a wide range of DNA damage. Cells of complementation group 1 (which lack ERCC1 and ERCC4) are hypersensitive to many bifunctional cross-linking DNA-damaging agents, whereas the cells of other complementary groups are not (34, 35, 36, 37). The higher expression levels of NER genes observed in most of the proliferating tissues may reflect these tissues’ greater need for proofreading of newly replicated DNA strands and repair of replication errors as a result of rapid cell proliferation in these tissues.

Variation in gene expression among individuals may result from inherited germ-line mutations or deletions (38), methylation (39), or polymorphisms (40), which may affect gene transcription and stability of transcripts. Because humans are constantly exposed to environmental toxicants and endogenous metabolites that cause DNA damage, increases in cellular DNA repair activity in response to DNA damage should result from a concomitant increase in gene transcription and translation activities. Therefore, apparently normal individuals with low levels of expression of DNA repair genes may have low DNA repair capacity, which may result in genetic alterations involved in carcinogenesis. We showed recently that low expression of hMLH1 and hMSH6 in peripheral lymphocytes is associated with an increased risk of head and neck cancer (24). In addition, aberrant expression of hMLH1 and hMSH2 in tumor tissues is also associated with tumor progression (41). Therefore, this new multiplex RT-PCR assay described here may be a useful tool for evaluating the role of NER expression in the development of cancer and tumor progression.

It is difficult, if not impossible, to study simultaneously the expression of several NER genes by Northern blot hybridization analysis or the RNase protection assay, because the amount of tissue available is often small and the expression of the NER genes in some tissues is relatively low. Therefore, a more sensitive method is necessary, particularly in studies where blood is the only accessible tissue. We have demonstrated that the multiplex RT-PCR assay provides a fast and sensitive technique to simultaneously detect the levels of specific transcripts of several NER genes in the tissues tested. The ability of this assay to provide a measure of the relative levels of gene expression should help characterize variation in gene expression in humans, and the assay may be a useful tool in molecular epidemiology studies (24).

Because they are accessible and easy to collect, peripheral blood lymphocytes are commonly used as surrogates for target tissues to monitor molecular events, such as the level of DNA adducts (42, 43), that may be associated with development of cancer. Our data have demonstrated that in all of the tissues we examined, including PHA-stimulated T lymphocytes, the expression levels of the five NER genes were measurable. Although the tissues tested came from different individuals, the PHA-stimulated T lymphocytes had a comparable level of NER gene expression to that of other tissues. It is possible that the NER gene expression of the target tissues may be influenced by direct carcinogen exposure. Therefore, it is appropriate to use nontargeted tissues such as lymphocytes to assess genetically determined gene expression levels that have not been affected by tissue-specific mutagens or carcinogens. Taken together, we believe that the expression levels of the five NER genes in PHA-stimulated peripheral T lymphocytes may provide a reasonable estimate for that of target tissues that are not accessible for clinical trials or epidemiological studies.

Although in vivo gene expression can be modulated by many factors including exposure to carcinogens, diet, and medications, measuring the in vivo DNA-repair gene expression in humans may allow assessment of host factors that influence the gene expression level. This assay may give insight into the relationship between DNA damage and repair and an individual’s response to environmental carcinogens and resistance to chemotherapy and radiotherapy (27); therefore, it may be a useful tool for rapid screening for aberrant expression of NER genes in both normal and tumor tissues. The validity of this assay should be tested further. Studies are needed to determine the variation in NER gene expression between tissues and individuals, using different types of tissue obtained from the same subject and the same tissue from different subjects. The ultimate validation of the assay will rely on future pilot case-control studies to provide cancer risk estimates. Additional studies of the relationships among NER gene expression, phenotypic DNA repair capacity (44, 45, 46), and NER gene polymorphisms (47) will validate the usefulness of this assay.

We thank Drs. Reuben Lotan and Margaret Spitz for critical review of the manuscript, Dr. Maureen Goode for scientific editing, and Joanne Sider and Joyce Brown for manuscript preparation.