[ RESEARCH 55. 1261-1266. March 15. ITO] Malignant and Nonmalignant Brain Tissues Differ in Their Messenger RNA Expression Patterns for ERCC1 and ERCC2

Meenakshi D. Dabholkar, Mitchel S. Berger, Justine A. Vionnet, Charles Egwuagu, John R. Silber, Jing Jie Yu, and Eddie Reed1

Medical Ovarian Cancer Section, Clinical Pharmacology Branch, National Cancer Institute, Bethesda. Maryland 20892 ¡M.D. D., J. A. V., J. J. Y., E. R.¡:Department of Neurological Surgery, University of Washington Medical Center. Seattle, Washington 98195 ¡M.S. B.. J. K. SJ: and Laboratory of Immunologi. National E\e Institute. Belhesda. Maryland 20X92 ¡C.E.I

ABSTRACT Studies in UV repair-deficient Chinese hamster ovary cells (5), clinical human ovarian cancer tissues (10), and human ovarian cancer Perturbation of the DNA repair process appears to be responsible for cells in tissue culture1 suggest that mRNA expression levels of the occurrence of a number of human diseases, which are usually associ ERCCÌmay represent the relative activity of the excision process ated with a propensity to develop internal and/or disorders within NER, if no other specific NER deficit exists. It is unclear of the central nervous system. We have been interested in the possibility that a subtle abnormality in DNA repair competency might be associated whether the selection of one gene from the helicase complex such as with the transformation of nonmalignant cells to the malignant state. To ERCC2 may represent activity of the helicase function of NER. study this question, we assayed malignant and nonmalignant brain tissues However, since ERCC2 appears to be an essential component of from 19 individuals for m UN Yexpression levels of the human DNA repair normal helicase function (6, 8), we have used ERCC2 as a tool to genes ERCC1, ERCC2, and XPAC and for differential splicing of the assess relative helicase activity within NER. ERCCÌand ERCC2 ERCC1 transcript. We separately compared expression levels of these are closely linked on chromosome 19ql3.2-13.3, along with DNA genes in the following situations: concordance of expression within ma ligase and XRCCÌ(11), suggesting a close linkage in nature of the lignant tissues; concordance of expression within nonmalignant tissues; four most basic functions in repairing DNA damage that occurs in concordance between malignant and nonmalignant tissues within individ the environment. uals of the cohort; and concordance of between two An abnormality in normal DNA repair appears to be responsible for nonmalignant tissue sites within a single individual. Linear regression analyses of mRNA values obtained suggested orderly concordance of these the occurrence of human nonpolyposis colon cancer, as recently three DNA repair genes in nonmalignant tissues within the patient cohort reported by Parsons et al. (12). Two recent reports have suggested that and an excellent concordance of these genes between two separate biopsy abnormalities of chromosome 19q are a common occurrence in human sites from the same individual. In contrast, malignant tissues showed brain cancer and may specifically involve region 19ql3.2-13.4 (13, disruption of concordance between the full-length ERCC1 transcript and 14). Concurrent with these studies, we have assessed the relationships ERCC2, which have excision and helicase functions, respectively. Further between two genes involved in DNA repair that are located in this more, within the same individuals, malignant tissues were discordant with area on chromosome 19, ERCCÌandERCC2. Our concern has been nonmalignant tissues for ERCC1 and ERCC2, although concordance for whether comparatively subtle abnormalities in DNA repair may con XPAC was preserved. These data suggest that one molecular characteristic tribute to the process of malignant transformation. of human may be the disruption of the normal relationship In this report, we have measured relative mRNA levels of ERCCI between the excision and the helicase functions of the nucleotide excision repair pathway. and ERCC2 in malignant and nonmalignant brain tissues taken from the same individuals. In nonmalignant brain tissues, mRNA levels of ERCCI and ERCC2 are tightly coordinated, particularly with respect INTRODUCTION to the full-length transcript of ERCCI, which is associated with NER2 is responsible for the repair of UV-induced DNA damage (1), efficient DNA repair. In malignant brain tissues, this coordination is completely lost. Nonmalignant brain tissues are similar in this respect DNA damage from polycyclic aromatic hydrocarbons, and DNA to four other situations where we have studied nonmalignant tissues damage induced by some chemotherapeutic agents such as cisplatin (15). Malignant brain tissues are dissimilar with nonmalignant tissues (2-5). The process is complex, with several models proposed in recent but very similar to malignant ovarian cancer tissue and chronic years (6, 7). There is agreement that the first step of NER (DNA lymphocytic leukemia, with respect to this type of analysis (16-18). damage recognition and excision) is rate limiting to the process (7). This dissimilarity between malignant brain tissues and nonmalignant There is also agreement that at least two protein complexes are brain tissues from the same individuals, as assessed by linear regres involved in the first step of NER. One of these complexes has helicase sion analysis, is strongly statistically significant. The possible biolog activity, may link DNA repair with DNA , and has as one ical implications of this observation are discussed. component of the complex, ERCC2 (along with ERCC3 and other transcription proteins; Refs. 6 and 8). The complex that performs incision of the DNA strand is distinct from the helicase complex and MATERIALS AND METHODS has as one component, ERCC1 (along with ERCC4, ERCC11, XPF, XPAC, and possibly other proteins; Refs. 6 and 9). Procurement of Tissue. Malignant and/or nonmalignant tissues were ob tained from 19 patients undergoing surgery for suspected primary or recurrent Received 6/27/94; accepted 1/18/95. brain neoplasms at the University of Washington Medical Center (Seattle, The costs of publication of this article were defrayed in part by the payment of page WA). Table 1 shows information regarding the antitumor therapy received by charges. This article must therefore be hereby marked advertisement in accordance with each patient studied. For patients receiving only surgery as their anticancer 18 U.S.C. Section 1734 solely to indicate this fact. 1To whom requests for reprints should be addressed, at Clinical Pharmacology therapy, this specimen was obtained at the initial surgical procedure. For Branch. National Cancer Institute. NIH. 9000 Rockville Pike, Building 10. Room 12N226, patients receiving radiation therapy and/or chemotherapy, this specimen was Bethesda, MD 20892. taken at the time of disease recurrence. Medications that were used to control - The abbreviations used are: NER. nucleotide excision repair; ERCC1. ERCC2. and edema and/or seizure activity are included under the category of "noncancer ERCC6. excision repair genes cross-complementing CHO mutant cell lines of comple mentation groups 1, 2. and ft, respectively; RT-PCR, reverse transcription-PCR: bp. base pair; XPAC. human xeroderma pigmcntosum A correcting gene. 3 E. Reed, unpublished observations. 1261

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1995 American Association for Cancer Research. EKCCl. XPAC, AND ERCC2 IN MALIGNANT BRAIN TISSUES

Table 1 Tumor-related therapy received by palients prior lo the procurement of the RT-PCR-based analysis of ERCCI and ERCC2 expression were synthesized biopsy specimen studied in this report on a DNA synthesizer (Biosearch, Inc., San Rafael, CA) and purified by Patient polyacrylamide gel electrophoresis (25). Primers and probes for XPAC and no.therapy"1 Anti-cancer therapyDilantinTegretol, ß-actinPCR were synthesized by Lofstrand Laboratories (Gaithersburg, MD). Surgery2 RNA from the human T-lymphocytic cell line, H9 (26), was similarly pro Surgery3 phénobarbitalNoneDilantinDecadron,cessed and used as an internal control. Surgery4 Aliquots of 3 ¡¿\ofthe cDNA preparation from all samples were analyzed Surgery5 Surgery6 DilantinTegretol for alternative splicing of ERCCI. Primers flanking exon VIII, which is 72 SurgeryNon-cancer-related bases long, were used to detect the presence and relative proportions of Surgery for brain cancer full-length and alternatively spliced ERCCI mRNA in tissue samples (27). Two years prior to dx of brain cancer, patient was treated for with DNA segments of 196 and 268 bp were obtained on amplification of the Cytoxan, Adriamycin, and 5-FU ERCCI cDNA sequence extending from bases 692 to 959 (27). RT-PCR was Surgery Decadron conducted for 30 cycles. Southern blots of amplified segments were hybridized Surgery Carbamazepine, with a probe 26 bases long extending from bases 764 to 789, 5' of exon VIII, levothyroxine, estrogen, which detects amplified products representing full-length ERCC1 mRNA (with lorazepam, nortriptyline 10 Surgery. XRT Dilantin, dexamethasone the 72-base long exon VIII) and an alternatively spliced species of ERCCI 11 Surgery Decadron, Dilantin, Tegretol, mRNA (without exon VIII). phenobarbitol Numerical values for the expression of the ERCCÌ,ERCC2, and XPAC 12 Surgery, XRT Dilantin, Proventil, Premarin, genes and of full-length and alternatively spliced ERCCI mRNA were ob Vanceril, phenobarbitol, Feldene, Tegretol, Intal tained as follows. Densitometric readings of autoradiographic signals were 13 Surgery, XRT. cisplatin, BCNU Dilantin, Depakote, obtained using the Collage analytical program on the Foto/Analyst II image Tranexene, Decadron analysis system (Fotodyne, New Berlin, WI). For each sample, the densito- 14 Surgery, XRT Prednisone, Tegretol. Dilantin, Zovirax, Premarin metric readout of the autoradiographic signal generated by the RT-PCR- 15 Surgery, XRT Phenobarbitol, Dilantin amplified DNA, when hybridized to the respective 32P-labeled ERCCI, 16 Surgery, XRT, hydroxyurea, 5-FU, Decadron, Dilantin, Zantac, ERCC2, or XPAC probe, was divided by the densitometric reading for ß-actin. 6-thioguaninine, procarbazine, CCNU Dibromodulcitol For purposes of comparison, the "repair gene:actin" value for H9 was assigned 17 Surgery the value of "1," and all other samples were expressed relative to that value. 18 Surgery Comparisons were made between "low-grade" and "high-grade" malignan 19 Surgery Tegretol " dx, diagnosis; 5-FU, 5-fluorouracil; XRT, radiation therapy; BCNU, bisfchloro- cies for the DNA repair genes studied. In the low-grade category (seven tumor ethyl)-nitrosourea; CCNU, A'-(2-chloroethyl)-A'' -cyclohexyl-A'-nitrosourea. specimens), we included tumors with the histological designation of astrocy- toma and oligodendroglioma. In the high-grade category (eight tumor speci mens), we included tumors with the histological designation of glioblastoma related therapy." Some of these medications are known to affect P450 isoen- and anaplastic astrocytoma. One tumor specimen was not clearly designated as zymes in the liver, and their effects on human glial cells generally have not high grade or low grade. been studied. Statistical Analyses. mRNA levels of excision repair genes in brain tumor Sixteen specimens of histologically tumor-free brain tissue, at least 2-3 cm tissues and overlying normal tissues were plotted against one another to from the tumor margin, were also obtained from 15 patients. The specific types determine the possible evidence of coordinated gene expression (15). Statis of tumor were determined by neuropathology at the University of Washington. tical significance was assessed using Student's t test with the Statworks Matched samples of malignant and nonmalignant tissue from 12 patients, 4 program (Cricket Software, Inc., Philadelphia, PA) on a Macintosh SE com unmatched malignant tissues, and 4 nonmalignant tissues (2 of which were puter (Apple Computers, Inc., Palo Alto, CA). Two-sided fs are shown in the obtained from the same patient) are included in this study. Informed consent tables and text. Curve-fitting analyses to obtain correlation coefficients were was obtained from each patient or from that individual's legal guardian. similarly conducted using the CricketGraph program (Computer Associates PCR Analyses. A RT-PCR-based assay system was used to determine the International, Inc., Islandia, NY). Statistical significance of the correlations level of expression of XPAC, ERCC1, ERCC2, and ß-actin.Tissues were between levels of expression of repair genes in normal and tumor tissues were stored at -80°C and extracted for total RNA by hot phenol/chloroform determined using regression analyses with the Statworks program. The SDs extraction (19). cDNA was obtained from 10 ;u.g of total RNA, by reverse listed represent the first SD from the mean. transcription using oligo-dT primers (Reverse Transcription System; Promega, Madison, WI). cDNAs were washed and concentrated by ultrafiltration (Ami- con, Beverly, MA) and resuspended to 100 p.1 in low TE buffer [10 HIMTris RESULTS (pH 8.0)-0.1 mM EDTA]. For XPAC, primer and PCR conditions were optimized for amplification of Fig. 1 shows a representative set of autoradiographs of RT-PCR- a 531-bp segment from bases 164 to 694 (20), which spans a region that amplified mRNA for the excision repair genes ERCCI, ERCC2, and extends from within exons I to V (21). For ERCC1, primers and RT-PCR conditions were selected to effect amplification of a 481-bp segment, from XPAC from malignant brain tissues and from overlying nonmalignant tissues. The range, median, mean, and SD from the mean for the bases 245 to 725 of the ERCCÌcDNA nucleotide sequence (18), and includes exons III to VI of the ERCCI gene (22). For ERCC2, a 500-bp segment from relative mRNA levels of the DNA excision repair genes are included bases 253 to 752 of the cDNA sequence (23) was amplified. Primers chosen for in Table 2. Data obtained from a total of 32 brain tissues (16 malig ß-actinspanned a 731 -bp segment of the coding region of the ß-actingene and nant and 16 nonmalignant) indicate that malignant brain tissues and extended from base 269 of exon II to base 1535 in exon IV (24). overlying nonmalignant tissues exhibit similarities in the range of Aliquots of 7.5 p.\ of the cDNA preparation from each sample were thus levels of total ERCCI, XPAC, and ERCC2 mRNA. Malignant and amplified by RT-PCR for 30 cycles for ERCCI; 35 cycles for ERCC2; 40 nonmalignant brain tissues also did not differ with respect to the range cycles for XPAC; and for 25 cycles for ß-actin.The GeneAmp PCR reagent kit of levels of full-length ERCCÌmRNAand of the alternatively spliced with AmpliTaq DNA polymerase (Perkin-Elmer Cetus, Norwalk, CT) was ERCCI transcript without exon VIII. used for each gene. Aliquots of amplified DNA were electrophoresed through a 1.5% agarose gel and transferred to Hybond N+ membrane (Amersham, We compared mRNA levels of ERCCI, ERCC2, and XPAC in Buckinghamshire, United Kingdom). Oligonucleotides (26-mer) from the cen malignant tissues from patients who received surgery only to tumor tral region of each amplified sequence were end-labeled with 32P-rATP (Am tissues from patients who also received radiotherapy ±chemotherapy ersham) using T4 polynucleotide kinase (Stratagene, La Jolla, CA) and were (Table 1). No significant differences were found with respect to total used as the respective probes. Oligonucleotides used as primers and probes for ERCCI (P = 0.283), full-length ERCCI (P = 0.842), alternatively 1262

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1995 American Association for Cancer Research. ERCCI. XPAC. AND ERCC2 IN MALIGNANT BRAIN TISSU! S

Non- When the relationship between mRNA expression levels of these Malignant Malignant H9 genes was examined in paired malignant and nonmalignant tissues from 12 patients (Table 3), moderate to good concordance was seen between paired malignant and nonmalignant tissues for XPAC Total ERCC1 (r = 0.582; P = 0.004), total ERCCI (r = 0.464; P = 0.015), and ERCC2 (r2 = 0.346; P = 0.044). No concordance was observed between paired tissues for the full-length ERCCI transcript (r2 = 0.013; P = 0.723), although malignant and nonmalignant brain tissues were similar with respect to the range of mRNA levels of this gene (Table 2). Several factors suggest differences between malignant and nonma XPAC lignant tissues with respect to differential splicing of ERCCI mRNA. Based on data summarized in Table 3, the relationship between the full-length ERCCI transcript and total ERCCI is much stronger in nonmalignant tissues (P = 0.002) than in malignant tissues ERCC2 (P = 0.024). Also, there is a striking absence of concordance between malignant tissues and matched nonmalignant tissues from the same individuals, when comparing the mRNA levels of the full-length ß-ACTIN (functional) ERCCI transcript (r2 = 0.013; P = 0.723; Table 3). A second major difference between malignant and nonmalignant tissues appears to be that concordance of levels of the full-length ERCCÌ Patient # 1 234124 transcript with levels of ERCC2 is seen ¡nnonmalignant tissues Fig. 1. Representative autoradiographs of RT-PCR-amplified signals generated by total (P = 0.004) but is not present in malignant cells (P = 0.414; Table 3). ERCCI mRNA. the full-length ERCCI transcript (FL ERCCI), the alternatively spliced ERCCI transcript without exon VIII (ERCCI WE), ERCC2. XPAC, and ß-actin from Two nonmalignant tissue samples were biopsied at different sites malignant brain tissues and from overlying nonmalignant tissues. Signals generated by from one patient. Fig. 3 shows the comparison of relative expression mRNA from H9 cells are included as internal controls. levels of DNA excision repair genes in these two specimens. The correlation coefficients obtained with or without ERCC2 are strong (r2 0.646, with ERCC2, Fig. 3A; r 0.996, without ERCC2, Fig. 3ß). Table 2 Descriptive statistics for the relative mRNA alntndance of the excision repair genes ERCCI, XPAC, and ERCC2 in malignant ami nontnaUgnanìbrain tissues The slopes of the linear curve fits obtained, with or without ERCC2, mRNATotal ±SD1.07 are close to 1. This suggests that normal tissues from different sites ERCCIMalignant may have strikingly similar levels of expression of these genes. Thus, (n = 16) ±0.37 differences between malignant and nonmalignant tissues discussed 16)Full-lengthNonmalignant (n = 0.55-2.520.49-1.420.900.83 1.09 ±0.56 P -0.9430.88 above are probably not simply the result of the biopsy of different tissue sites. ERCCI transcript Table 4 shows data comparing high-grade and low-grade tumors Malignant (n = 16) ±0.25 16)AlternativelyNonmalignant (n = 0.47-1.730.02-0.200.880.08 0.95 ±0.41 for relative expression levels of all five transcripts studied. The P =0.5390.09 relative levels of full-length ERCCI transcript were higher in spliced ERCCI transcript low-grade tumors as compared to high grade tumors (P = 0.036). Malignant (n = 16) ±0.05 16)XPACMalignantNonmalignant (n = 0.03-0.290.03-1.740.120.92 0.13 ±0.07 P =0.1670.95 Table 3 Linear cun'e fir or simple regression analvsis of the relationship between relative mRNA abundance of the excision repair genes in malignant (n = 16) ±0.46 and nonmalignant brain tissues 16)ERCC2MalignantNonmalignanI («= 0.30-2.250.27-1.140.08-1.48Median1.040.780.610.60Mean0.99 ±0.49 P =0.8300.64 curve TranscriptsMalignantmRNA S0.9170.7550.5750.3130.4250.0480.0000.0850.9450.7050.7360.4940.4810.4530.0350.003=P•eO.OOl

ERCCIFull-lengthE«CC//full-length spliced ERCCÌtranscript (P = 0.730), ERCC2 (P = 0.396), and ERCCI/XPACFull-length XPAC (P = 0.370). ERCCI/ERCC2ERCCI VUÕ/XPACERCCIwithout exon Table 3 is a summary of the linear regression relationships between VÌÌVERCC2Nonmalignantwithout exon ERCCI, ERCC2, and XPAC in malignant and nonmalignant brain 16)Total tissues (n = tissues. Correlation coefficients and Ps are used to indicate the relative ERCCI/XPACTotal ERCCI/ERCC2XPAC/ERCC2Total levels of concordance of expression of these genes. Fig. 2 shows a il niil0.0020.0030.0040.4860.8500.0150.7230.0040.0440.246 visual representation of this type of analysis. Fig. 2A shows the ERCCIFull-length£«CC//Full-lenglh example of excellent concordance between ERCCI and XPAC in ERCCI/XPACFull-length ERCCI/ERCC2ERCCI malignant tissues. Fig. 2B shows the example of poor concordance VUVXPACERCCIwithout exon between malignant and nonmalignant tissues for ERCCI. In general, VIII/EÄCC2Matchedwithout exon when malignant tissues are assessed as a group, the concordance of (nTotal malignant (n = 12) and nonmalignant 12)tissues0.4640.0130.5820.3460.132fit expression of ERCCI, ERCC2, and XPAC is good, with Ps that are ERCCIFull-length ERCCIXPACERCC2ERCCI strongly statistically significant. This is also true for nonmalignant tissues as a group. In all subsets, the nonfunctional truncated transcript of ERCCI (without exon VIII) did not show concordance with any gene. without exon VIIILinear 1263

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1995 American Association for Cancer Research. ERCCl. XPAC. AND ERCC2 IN MALIGNANT BRAIN TISSUES

Also, the shortened, alternatively spliced species of ERCCl was Table 4 Descriptive statistics for the relative mRNA abundance of the excision repair expressed to higher levels in high-grade tumors (P = 0.008). Total genes ERCCl, XPAC, and ERCC2 in low-grade tumors {astrocytoma and oligodendroglioma) and high-grade tumors (anaplastic astrocytoma and gliohlasloma) ERCCl mRNA did not differ between these two groups. Therefore, this suggests that greater disorder of ERCCl mRNA processing mRNATotal ±SD1.15 occurs as a cell moves from a less aggressive to a more aggressive ERCClLow-grade tumors (n = 7) ±0.49 state of malignancy. 8)Full-lengthHigh-grade tumors (n = 0.69-1.520.58-1.42 1.061.060.740.061.03 ±0.29 No differences were noted between high-grade and low-grade tu P =0.5741.03 mors for any of the other mRNA transcripts studied. Also assessed ERCCl transcript Low-grade tumors (n = 7) + 0.30 were the relationships between the various transcripts, comparing 8)AlternativelyHigh-grade tumors (n = 0.49-1.01transcript 0.75 + 0.15 high-grade and low-grade tumors. No differences were observed. P =0.0360.06 Fig. 4 shows variability data from replicates of PCR values for spliced ERCCl ERCCl and ERCC2, corrected for actin, in samples that gave Low-grade tumors («= 7) 0.02-0.13 + 0.04 8)XPACLow-gradeHigh-grade tumors (n = 0.07-0.200.03-1.74 0.130.93 0.13 + 0.04 detectable radiographie signals. The number of replicates per P =0.0081.05 formed for ERCCl, ERCC2, and ß-actinwasbased on the amount

tumors (n = 7) ±0.60 8)ERCC2Low-gradeHigh-grade tumors («= 0.45-1.400.27-1.14 0.920.61 0.90 ±0.35 Malignant Tissues P =0.5360.66

y = 1.18ÕX- 0.324 r1 = 0.! p < 0.001 tumors (n = 7) ±0.34 High-grade tumors (n = 8)Range0.36-1.730.41-0.83Median1.140.60Mean 0.61 ±0.15 P = 0.693

y= I36.521X-. 252.721* + 122.041 r2 =0.914

00 0-5 10 1.3 2.0 2.5 Relative Total ERCCl Expression

RelativeERCCl Levels Transcript of Full-Length

y = 0.082X + 0.820 0.013 p = 0.723

Gene Expression Corrected for Actin

Fig. 4. Analysis of variability of the values obtained in the PCR method used in this study. The percent variability of 24 data points for ERCCl and ERCC2 is shown.

of cDNA that was obtained from the sample. The percentage of variability in this analysis is the value of the first SD from the mean, expressed as a percentage of the mean value. For example, Non-Malignant Tissues a value of 0.8 ±0.2 is plotted on this curve as 25%. As shown, Fig. 2. Visual representation of the analyses for correlation of expression of NER genes when radiographie signals were low (mean values below 0.4), the in malignant and nonmalignant brain tissues. A. example of excellent concordance between ERCCl and XPAC in malignant tissues. B. example of poor concordance between variability in the PCR value obtained was high. When radiographie malignant and nonmalignant tissues for ERCCl. Correlation coefficients obtained from signals were of substantial intensity (above 0.4), the variability in linear curve fit analysis (CricketGraph) and Ps obtained from simple regression analysis the PCR value was low and averaged about 10%. As shown in (Statworks) are shown. Tables 2 and 4, median values in each group were well above the 0.4 level, with the exception of the alternatively spliced ERCCl 1.124« + O.OÎ2 1.286«* 0.047 r: = 0.996 transcript. XPAC o Total ERCCl O

ERCCl Full DISCUSSION -Length Malignant and nonmalignant brain tissues exhibit broad similarities with respect to the range, mean, and median mRNA levels of total fl 53 ERCCl, ERCC2, XPAC, and the full-length ERCCl transcript. How /ERCCl w/o ever, analyses of these data show that for these three DNA excision Exon VIH repair genes, malignant brain tissues show disruption of several mRNA expression patterns that are prominent in nonmalignant cells. Non-Malignant Brain Tissue Non-Malignant Brain Tissue This disruption appears to be related to differential splicing of ERCCl Palienl «14Sample A Patient «14Sample A mRNA and to the loss of the "normal" relationship between ERCCl Fig. 3. Comparison of relative expression levels of excision repair genes in two nonmalignant tissue samples biopsied at different sites from one patient. The correlation and ERCC2. coefficient obtained from linear curve fit analysis for a comparison of total ERCCl, Malignant and nonmalignant brain tissues show concordance of full-length ERCCl transcript, alternatively spliced ERCCl transcript, XPAC, and ERCC2 are shown in (A). The correlation coefficient and linear curve fit obtained without ERCC2 expression of total ERCCl with XPAC, of total ERCCl with ERCC2, are shown in (B). and of ERCC2 with XPAC levels. Malignant tissues differ substan- 1264

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1995 American Association for Cancer Research. ERCCÃŒ.XPAC. AND EKCC2 IN MALIGNANT BRAIN TISSl'l S tially from nonmalignant brain tissue in the relationship between data suggests that ERCCI gene copy number may be increased in ERCC2 and the full-length ERCC1 transcript in that concordance of some specimens but not ERCC2.4 This would suggest the expression is lost in malignant cells. Also, when paired malignant and possibility that the excision function may be proceeding more nonmalignant tissues from the same individuals are compared for the rapidly than the helicase function under these circumstances. levels of these mRNAs, no correlation is observed between these Hence, for some proportion of lesions, incisions would be made in tissues for the full-length ERCCI transcript. the DNA at sites flanking DNA lesions first, and the helicase may In contrast to ERCC2 and the full-length ERCCI transcript, both later effect separation of the damaged strand. This is the reverse of malignant and nonmalignant brain tissues exhibit excellent coordina what would be expected to occur based on one current model (1). tion of XPAC with the full-length ERCCI transcript. Thus, discor Therefore, at any given point in time, the repair process in an dance in expression of ERCC2 and the full-length ERCCÃŒtranscript in affected cell might exhibit an abnormal, dysfunctional situation malignant tissues may not merely represent a random dissociation of as an increased percentage of the total DNA repair activity of genes encoding for NER proteins but instead indicates a possibly the cell. specific dissociation of genes critical to the incision and helicase protein complexes. This discordance in malignant tissues is striking in REFERENCES the light of the close proximity of ERCCI and ERCC2 on chromo some 19ql3.2-13.3 (11) and the frequent occurrence of observed 1. Hoeijmakers, J. H. J. Nucleotide excision repair II: from yeasts to mammals. Trends Genet., 9: 211-218, 1993. genetic changes in this genomic area in brain tumors (13, 14). 2. Cleaver, J. E. It was a very good year for DNA repair. Cell, 76: 1-4, 1994. Discordance of the relative levels of ERCC2 mRNA with the levels 3. Hansson, J., Grossman. L.. Lindahl. T.. and Wood, D. Complementation of the of ERCCI mRNA has been reported in ovarian tumor tissues (16). xeroderma pigmentosum DNA repair synthesis defect with Escherichia coli UvrABC proteins in a cell free system. Nucleic Acids Res., 18: 35-40, 1990. Clinical resistance to platinum-based therapy was associated with 4. Myles, G. M., and Sanear, A. DNA repair. Chem. Res. Toxicol., 2: 197-226, increased expression of ERCCI, whereas the expression levels of 1989. ERCC2 observed in patients responding to platinum-based therapy 5. Parker, R. J., Poirier, M. C., Bostick-Bruton, F., Vionnet, J.. Bohr, V. A., and Reed, E. The use of peripheral blood leukocytes as a surrogate marker for cisplatin drug were comparable to levels in tumors resistant to such therapy (16). resistance—studies of adduci levels and ERCCI. ¡n:B. M. Sutherland and A. D. When those ovarian cancer data are analyzed as in the present study, Woodhead (eds.). Brookhaven Symposia in Biology 36, DNA Damage and Repair in the regression analysis shows r2 = 0.068 and P = 0.198 (data not Human Tissues, pp. 251-261. New York: Plenum Publishing Corp., 1990. 6. Drapkin, R., Sanear, A., and Reinberg, D. Where transcription meets repair. Cell, 77: shown). Discordance of ERCCI with ERCCI has also been shown in 9-12, 1994. malignant lymphocytes from patients with chronic lymphocytic leu 7. Shivji, M. K. K., Kenny, M. K., and Wood, R. D. Proliferating cell nuclear antigen is required for DNA excision repair. Cell, 69: 367-374, 1992. kemia (17, 18). 8. Friedberg, E. C. Xeroderma pigmentosum, Cockayne's syndrome, helicases, and In nonmalignant brain tissue, there is good concordance of full- DNA repair: what's the relationship? Cell, 71: 887-889, 1992. length ERCCI mRNA and ERCC2. This has also been observed in 9. van Vuuren, A. J., Appeldoom, E., Odijk, H., Yasui, A., Jaspers, N. G. J.. Bootsma. D., and Hoeijmakers, J. H. J. Evidence for a repair enzyme complex involving nonmalignant bone marrow obtained from four patient cohorts with ERCCI and complementing activities of ERCC4, ERCCI 1, and xeroderma pigmen different malignancies (15). The development of discordance of the tosum group F. EMBO J., 12: 3693-3701, 1993. full-length ERCCI transcript with ERCC2 in malignant brain tissues 10. Dabholkar, M., Vionnet, J., Bostick-Bruton, F., Yu, J. J., and Reed, E. Messenger RNA levels of XPAC and ERCCI in ovarian cancer tissue correlates with response may thus be indicative of the development of disruption in the to platinum-based chemotherapy. J. Clin. Invest., 94: 703-708, 1994. nucleotide excision repair machinery as cells transit from the nonma 11. Barnes, D. E., Kodama, K-I.. Tynan, K., Trask, B. J., Christensen, M., De Jong, P. J.. Spurr, N. K., Lindahl, T., and Mohrenweiser, H. W. Assignment of the gene lignant to the malignant state. encoding DNA ligase I to human chromosome 19ql3.2-13.3. Genomics, 12: Disruption of the DNA repair machinery may lead to genomic 164-166, 1992. 12. Parsons, R.. Li, G-M., Longley, M. J.. Fang, W., Papadopoulos, N.. Jen, J.. de la instability as a result of the inability to repair routine types of DNA Chapelle, A.. Kinzler, K. W.. Vogelstein, B., and Modrich, P. Hypermutability and insults over time. This subtle change in the NER process may be mismatch repair deficiency in RER* tumor cells. Cell, 75: 1227-1236, 1993. analogous to the abnormality in mismatch repair that is associated 13. von Deimling. A., Bender, B., Jahuke, R., el al. Loci associated with malignant progression in astrocytomas: a candidate on chromosome 19q. Cancer Res.. 54: with the development of human nonpolyposis colon cancer (12). We 1397-1401, 1994. hypothesize that the accumulation of DNA damage over time may 14. Rubio. M-P., Correa, K. M., Ueki, K., Mohrenweiser, H. W., Gusella, J. F., von lead to multiple subsequent genomic changes that result in neoplastic Deimling, A., and Louis, D. N. The putative glioma tumor suppressor gene on chromosome 19q maps between APOC2 and HRC. Cancer Res., 54: 4760-4763. transformation. The fact that this disruption occurs in the three ma 1994. lignancies studied to date (ovary, brain, and chronic lymphocytic 15. Dabholkar, M., Bostick-Bruton, F.. Weber. C, Egwuagu. C, Bohr, V. A., and Reed. E. Expression of excision repair genes in non-malignant bone marrows from cancer leukemia), but not in five other settings of nonmalignant tissue patients. Res., 29.?: 151-160, 1993. (Ref. 15 as well as data reported here), suggests that further study 16. Dabholkar, M., Bostick-Bruton. F., Weber. C., Bohr. V., Egwuagu. C.. and Reed, E. of the interaction between these two functions of the NER process ERCCI and ERCC2 expression in malignant tissues from ovarian cancer patients. J. Nati. Cancer Inst., 84: 1512-1517, 1992. is warranted. 17. Geleziunas, R., McQuillan. A.. Malapetsa, A., Hutchinson, M., Kopriva, D., Published data from our laboratory suggests that exposure to Wainberg, M. A., Hiscott, J., Bramson, J., and Panasci, L. Increased DNA DNA-damaging agents in vitro and/or in vivo may lead to increased synthesis and repair-enzyme expression in lymphocytes from patients with chronic lymphocytic leukemia resistant to nitrogen mustards. J. Nati. Cancer Inst.. expression of ERCCÃŒand possibly other genes of the NER path 83: 557-564. 1991. way. This up-regulation in mRNA expression is seen in malignant 18. Panasci, L., McQuillan, A., Malapetsa, A., and Bramson, J. Evidence for enhanced and nonmalignant tissues after exposure to platinum compounds DNA repair enzyme expression in resistant chronic lymphocytic leukemia patients. Proc. Am. Assoc. Cancer Res., 32: 422, 1991. (16). For patients who received radiation therapy and/or DNA 19. Wahl, G. M., Padgett, R. A., and Stark, G. R. Gene amplification causes overpro damaging chemotherapy, nonmalignant tissues obtained at surgery duction of the first three enzymes of UMP synthesis in /V-(phosphonacetyl)-t.- aspartate-resistant hamster cells. J. Biol. Chem., 254: 8679-8684. 1979. (and reported in this study) were exposed to the same treatment as 20. Tanaka, K., Miura, N., Salokata, I., Miyamoto, I., Yoshida, M. C., Satoh, Y., Kondo, malignant tissues. Therefore, the loss of concordance between S., Yasui, A., Okayama. H., and Okada, Y. Analysis of a human DNA excision repair malignant and nonmalignant brain tissues cannot be explained on gene involved in group A xeroderma pigmenlosum and containing a zinc-finger domain. Nature (Lond.), 348: 73-76, 1990. the basis of patient therapy. 21. Satokata, I., Tanaka, K., Yuba, S., and Okada, Y. Identification of splicing Dissociation of the DNA damage recognition/excision function of the last nucleotides of exons, a nonsense mutation, and a missense mutation of the from the helicase function could possibly lead to several different types of circumstances. In malignant brain tissues, preliminary 4 E. Reed, unpublished observations. 1265

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1995 American Association for Cancer Research. ERCCI. XPAC. AND ERCC2 IN MALIGNANT BRAIN TISSUES

XPAC gene as causes of group A xeroderma pigmentosum. Mutation Res., 273: the human cytoplasmic /3-actin gene: interspecies homology of sequences in the 203-212, 1992. intron. Proc. Nati. Acad. Sci. USA, 82: 6133-6137, 1985. 22. Hoeijmakers, J. H. J., Wecda, G., Troelstra, C., van Duin, M., Westerveld. A., van der 25. Maniatis, T., Fritsch, E. F., and Sambrook, J. Molecular Cloning: A Laboratory Eb, A., and Bootsma, D. Molecular genetic dissection of mammalian excision repair. Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory. 1989. In: M. W. Lambert and J. Laval (eds.), DNA Repair Mechanisms and Their Biolog- 26. Popovic, M., Read-Connole, E., and Gallo, R. C. T4 positive human neoplastic ical Implications in Mammalian Cells, pp. 563-574. New York: Plenum Publishing cell lines susceptible to and permissive for HTLV-III. Lancet, 2: 1472-1473, Corp., 1989. 1984. 23. Weber, C. A., Salazar, E. P., Stewart, S. A., and Thompson, L. H. ERCC2: cDNA 27. van Duin, M., de Wit. J., Odijk, H., Westerveld. A., Yasui, A., Koken, M., cloning and molecular characterization of a human nucleotide excision repair gene Hoeijmakers, J. H. J., and Bootsma. D. Molecular characterization of the human with high homology to yeast RAD3. EMBO J., 9: 1437-1447, 1990. excision repair gene ERCC-I: cDNA cloning and amino acid homology with the yeast 24. Nakajima-lijima. S.. Mamada, H., Reddy, P., and Kakunaga, T. Molecular structure of DNA repair gene RADIO. Cell, 44: 913-923, 1986.

1266

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1995 American Association for Cancer Research. Malignant and Nonmalignant Brain Tissues Differ in Their Messenger RNA Expression Patterns for ERCC1 and ERCC2

Meenakshi D. Dabholkar, Mitchel S. Berger, Justine A. Vionnet, et al.

Cancer Res 1995;55:1261-1266.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/55/6/1261

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/55/6/1261. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1995 American Association for Cancer Research.