Analysis of MDR1 Expression in Normal and Malignant Endometrium by Reverse Transcription-Polymerase Chain Reaction and Immunohistochemistry1

The purpose of this study was to quantitate the expression of human MDRJ mRNA levels in normal endometrium and in endometrial carcinoma and to determine the association of MDRJ levels with prognostic indicators. Endometrial samples from 43 postmenopausal patients with endometrial carcinoma and 38 patients (controls) with benign disease undergoing hysterectomy were snap-frozen. MDRJ levels were determined by quantitative reverse transcription-PCR (RT-PCR) and compared to sensitive and resistant cell lines. Immunohistochemistry was done with MM4.i7, an anti-MDRJ antibody, on paraffin sections, and the results were compared to those obtained from RT-PCR. Data was analyzed using the Kruskal-Wallis and Bonferroni tests, setting the P value at 0.05. In both postmenopausal endometrial tissue and tumors, MDR1 expression was localized to the epithelial cell layer Comparison of immunohistochemistry and RT-PCR results demonstrated a correlation of 80%. In control patients, MDRJ expression was significantly higher in postmenopausal endometrium (n 15) than in the proliferative premenopausal endometrium (n 15; P = 0.0024). MDRJ expression in all tumors was lower than that measured in the postmenopausal controls. Between each tumor group, there was no significant difference in the MDR1 levels observed. MDRJ expression was significantly lower in patients with high nuclear grade (n = 18) tumors when compared to patients with low nuclear grade (n i4; P = 0.04) tumors. Comparison of MDRJ levels with multiple prognostic indicators for endometrial cancer was only significant for nuclear grade. The data indicate that MDRJ expression is not a major component of the drug resistance observed in primary endometrial tumors. Received 3/21/96; revised 9/4/96; accepted 9/18/96. I Supported in part by USPHS Grant CA39821 (S. B. H.) and 5P30CA13330. D. Y-S. K. and S. M. contributed equally to this study. 2 To whom requests for reprints should be addressed, at Department of Molecular Pharmacology, 1300 Morris Park Avenue, Bronx, NY 10461. Phone: (718) 430-2163: Fax: (718) 430-8922. INTRODUCTION The development of resistance to chemotherapy is one of the major obstacles to the successful treatment of many malignancies. In general, tumors treated with a single agent develop alterations in the drug target that result in clinical resistance. However, some tumors become resistant to a broad spectrum of antitumor drugs after treatment with only a single agent: this phenomenon is known as MDR3 (I , 2). Only certain drugs, particularly hydrophobic natural products such as Adriamycin, Taxol, and vinblastine, induce MDR. The development of MDR has been extensively studied, using highly drug-resistant cubtured cell lines as model systems. In characterizing these cell lines, it was observed that the highly resistant cell lines accumulated less drug intracellularly compared to the parental cell lines (3, 4). In addition, these cells overproduced a membrane glycoprotein, P-gp, whose level of expression correlates with the observed degree of resistance (5). P-gp is encoded by a small gene family, MDR/ (6) and MDR2 (7, 8) in humans and mdrJa, tndrJb, and ,ndr2 in mice (9, 10). MDRJ, ,ndrJa, and mdrJb genes, which confer drug resistance to a sensitive cell, are considered Class I genes, whereas MDR2 and tndr2, which do not confer drug resistance, are classified as Class II genes. It is believed that P-gp functions as an energy-dependent efflux pump, thus maintaining the intracellular drug concentration below cytotoxic levels ( 1 1). Although P-gp was originally described in drug-resistant cells, it is now known that it is present in many normal tissues in both mouse and human. There is a high level of P-gp expression in adrenal cortical cells, the kidney brush border, the eanalicular surface of hepatocytes, the small and large intestinal mucosal cells, and the pancreas (12-15). In mouse, the expression is isoform-specific, with mdrla highly expressed in the intestine, liver, and lung and rndrlb expressed at high levels in the adrenals, kidney, and, interestingly, in the uterus during pregnancy ( 16). However, the function of P-gp in normal tissues has not yet been elucidated. From studies in transgenic mice, it has been suggested that mdrJa functions in the blood-brain barrier to protect against environmental toxins ( I 7) and that ,ndr2 functions in the liver to transport phospholipids into the bile (18). Many tumors that originate from tissues with high levels of MDRJ expression are difficult to treat using chemotherapeutic agents. This has led to the hypothesis that certain tumors are intrinsically resistant, whereas others acquire resistance only after exposure to drug (19-23). MDRJ mRNA levels in tumors and cell lines seem to correlate with the observed degree of Research. on July 16, 2017. © 1996 American Association for Cancer clincancerres.aacrjournals.org Downloaded from 1982 MDRJ Expression in Human Endometrium resistance (24), and increased expression of MDRJ is thought to be responsible for treatment failure in some tumors (25-27). MDRJ gene expression also has been implicated as a marker of tumor progression and aggressiveness. It would be beneficial to accurately predict whether a tumor will respond to chemotherapy before treatment. To this end, many tumor types have been evaluated for MDRJ mRNA and P-gp expression by RT-PCR and immunohistochemistry, respectively, to determine the correlation, if any, between these levels and patient outcome. To date, only in neuroblastoma, childhood sarcoma, and certain leukemias have the levels of MDR been predictive of poor outcome in these retrospective studies (25, 28). In the mouse, mdrlb is highly expressed in the endometrium during pregnancy. Expression is restricted to the secretory columnar epithelium, which forms the endometrium (29, 30). Very little is known about the expression of MDRJ mRNA in the human uterus. Because the human and mouse genes are structurally and functionally similar and share almost identical tissue distribution, it is anticipated that MDRJ is also expressed in the human uterus. Retrospective studies using immunohistochemistry have demonstrated that MDRI is expressed in tumors derived from the uterus (31-33) and in the normal uterus during pregnancy (34). To fully characterize the expression of MDRJ in the human uterus, we have analyzed biopsy samples from benign endometrium as well as from tumors derived from endometrium by quantitative RT-PCR and immunohistochemistry. We have determined the levels of MDRJ in benign tissue from both premenopausal and postmenopausal endometrium. In addition, we have correlated expression of MDRJ mRNA with estrogen and progesterone receptor levels, stage and grade of the tumor, and other established prognostic indicators (35). MATERIALS AND METHODS Cell Lines and Tissue Procurement The project was approved by the Internal Review Board at the Montefiore Medical Center. Samples were obtained over an 18-month period from a total of 91 patients, including 53 patients with endometrial hyperplasia or endometrial carcinoma undergoing hysterectomy and surgical staging and 38 preand postmenopausal patients with benign pathological conditions undergoing hysterectomy at the Weiler Division of Albert Emstein College of Medicine and Montefiore Medical Center. The endometrial tissue was divided into two sections; one was snap-frozen in liquid nitrogen and stored at -70#{176}C, the other was fixed in formalin, later blocked in paraffin, and sectioned for both immunohistochemical and H&E staining for histological confirmation. The drug-sensitive SKOV3 human ovarian cell line and its drug-resistant derivative, SKVLB (a gift from Dr. V. Ling, British Columbia Cancer Research Center, Vancouver, B. C., Canada), were used as negative and positive controls. OVCAR cells treated with Taxol were used as a weakly positive control. Quantitative RT-PCR Isolation of Total RNA from Specimens. Total RNA was prepared by the method of Chomczynski and Sacchi with minor modifications (36). Briefly, tissue was snap-frozen by immersion in liquid nitrogen and pulverized. Six ml of denaturing solution [4 M guanidinium thiocyanate, 25 msi sodium citrate (pH 7), 0.5% N-lauroylsareosine. and 0. 1 M 2-mereaptoethanol] were added. The mixture was extracted with 6 ml of water-saturated phenol and 1 .2 ml of chloroform-isoamyl aleohol (49: 1) and incubated on ice for 15 mm. After centrifugation for 20 mm at 4#{176}C in a Beckman J-6B centrifuge at 2000 X g (2800 rpm), the aqueous-phase solution was precipitated with an equal volume of isopropanol in a -20#{176}Cfreezer and centrifuged. The RNA pellet was resuspended in 75% ethanol, centrifuged at 2000 x g, and air-dried for at least I 5 mm. The pellet was resuspended in 60 p.1 of diethylpropylcarbonate-treated water. Approximately 1 p.g of RNA was then used for RT. RT of RNA. The RT-PCR quantitative procedure used is the method of Horikoshi et a!. (37) with minor modifications. Briefly, for RT, I .tg of RNA was added to 20 t1 of 5 X transcription buffer [250 mM Tris-HCI (pH 8.3), 375 ms KC1. and 15 msi MgC12], 10 t1 of 100 mst DTT, 10 il of 10 msi deoxynucleotide triphosphate solution ( 10 mrvt final coneentration of dATP, dCTP, dGTP, and dTl’P; Pharmacia), 2.5 p.l of 3 mg/mi of BSA, 0.5 il of random hexamers [50 optical density units dissolved in 0.55 ml of 10 mrvi Tris-HC1 (pH 7.5) and 1 ms EDTA], 2.0 p.1 RNAguard (Pharmacia), and 3 jjJ Moloney murine leukemia virus reverse transeriptase (Life Technologies, Inc.). The mixture was incubated in the thermocycler at 26#{176}C for 8 mm, 42#{176}Cfor 90 mm, and, finally, 95#{176}Cfor 5 mm. The eDNA synthesized was stored at -70#{176}C. PCR Primers for Reference and Target Genes. The PCR primers were synthesized by the Albert Einstein College of Medicine Oligonucleotide Facility. The primers were designed to span two adjacent exons to detect genomic DNA and amplify a region of 100-200 bases. Each 5’ primer also contained the T7 promoter sequence (the first 17 bases) followed by the transcription initiation sequence or a clamping sequence (the next 5-6 bases), which functioned to improve the yield of the transcription product. The exact sequence of this promoter is TAA TAC GAC TCA CTA TA (marked T7 in the sequences below). The initiation sequence is listed in quotation marks. The other Sequences are: (a) 5’ T7-”GGGAGA”GCGGGAAATCGTGCGTGACATT 3’ (sense primer; bases 656-677) and 5’ GATGGAGTFGAAGGTAGTTTCGTG 3’ (antisense primer: bases 864-887) are the primer sequences for 3-aetin. The fragment generated is 232 bases long; and (b) 5’ T7-”GGGA”CCCATCATTGCAATAGCAGG 3’ (sense primer; bases 3020-3039) and 5’ GTTCAAACTTCTGCTCCTGA 3’ (antisense primer; bases 3157-3176) are the primer sequences for MDRJ. The fragment generated is 157 bases long. PCR Reactions. A Perkin-Elmer Cetus 9600 thermocycler was used for the PCR reactions. The eDNA was diluted serially from 0 to I : I 06. Ten jil of each dilution of eDNA were added to 2.5 p.l of lOX Taq buffer [500 msi KCI and 100 msi Tris-HC1 (pH 8.3)], 3.75 il of 12.5 m i MgCI,, 0.5 pi of a 10 mM deoxynucleotide triphosphate solution (same as for RT), 1 p.1 each of the 5’ and 3’ primers at I 2.5 pmol/l, I 0 p.1 of the eDNA solution, and 4.25 jj.l of 10 mtvi Tris-HCL. These tubes were heated for 5 mm at 95#{176}Cin the thermocycler. Taq polymerase (Perkin Elmer Cetus) was diluted at 1 : 16, and 2 p.1 of the enzyme were then added to each tube. The PCR protocol was Research. on July 16, 2017. © 1996 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Clinical Cancer Research 1983 93.5#{176}Cfor I mm, 55#{176}Cfor 1 mm, and 72#{176}C for I mm for 30 cycles and 7 mm at 72#{176}Cwith storage at 4#{176}C. T7 Polymerase Transcription Reaction. Twenty p.1 of solution A (14.6 i.l of diethylpropylcarbonate-treated water, 2.5 il of lOx transcription buffer, 2.5 j.l of 10 msi ribonucleotide solution, 0.25 il of 1 M DTT, 0. 1 5 j.l of RNAguard, 0. 1 pA of 0.5 M spermidine, and 0.25 p.1 of [a-32P]CTP at 3000 Ci/mmol) were added to each tube containing 3.33 il of each PCR product. The reaction was started by adding 1.67 i.l of T7 RNA polymerase (EpiCenter), incubated for 1 h at 37#{176}C. A total of 0.80 p.1 of 0.5 M EDTA was added to stop the reaction. Gel Electrophoresis. Twenty p.1 of the gel loading buffer ( 10 M urea and 0.02% bromphenol blue) were added to each sample and incubated at 90#{176}C for 10 mm. Each sample was electrophoresed on a 6% denaturing polyacrylamide gel with 0.5 x Tris-borate EDTA buffer. The gels were dried and exposed to film for 1-2 h. The bands were excised using the autoradiogram as a guide and counted in a liquid scintillation counter. Statistical Analysis. The best fit line was determined using linear regression analysis. All of the data, including patients’ clinical information, were compiled and analyzed using the Statistical Products for Service Solutions data analysis package. Differences among the groups of tumor specimens, in terms of their PCR ratio for MDRJ gene, were assessed by the Kruskal-Wallis and Bonferroni tests. All tests of significance were carried out with a P value of 0.05. Immunohistochemistry Technique. Paraffin-embedded sections were immunostained following the procedure in the Vectastain ABC kit with minor modification. Briefly, after dewaxing in Hemo-De (Fisher) and rehydrating in ethanol dilution series, the slides were incubated with I % hydrogen peroxide in methanol for 30 mm and pretreated with 10% normal goat serum and 20% normal horse serum for 30 and 40 mm, respectively. Incubation with MM4. I 7, an IgG2a,K monoebonal antibody to the fourth extracellular domain of MDRJ, was done at a concentration of 5.3 p.l/ml for 18 h at 4#{176}C (49). MM4.17 was a generous gift from Dr. M. Cianfraglia (Laboratorio di Immunologia, Rome, Italy). The samples were then immersed in S p.g/ml of biotinylated horse anti-mouse IgG at room temperature for 30 mm, followed by further incubation with an avidin and biotinylated horseradish peroxidase complex at room temperature for 45 mm. The slides were developed in 3-3 ‘ diaminobenzidine tetrahydroehboride containing 0.022% hydrogen peroxide, counterstained with hematoxylin, and dehydrated and mounted. The slides were washed with 0.1% saponin in sodium PBS [10 mt i PBS (pH 7.5) and 200 mM NaCI] between each step, and 20% normal horse serum in 0. 1% saponin/PBS buffer was used for all of the dilutions. Paraformaldehyde-fixed cell preparations from a HeLa cell line transfected with mouse tndrlb were used to demonstrate the specificity of MM4.17 to human MDRJ-encoded P-gp. As a negative control, mouse IgG (10 i.g/mb) was used instead of the primary antibody. Acetone-fixed cytospin cell preparations of a vinblastine-resistant human ovarian tumor cell line (SKVLB) and a tumor sample with a high RT-PCR score were used as positive controls. Quantitation. The intensity, percentage, and pattern of staining for each sample slide were recorded separately, and the results were summarized to yield a single score. The slides were reviewed and scored by a single pathologist (H-J. S.) who did not know the RT-PCR results. Two different staining patterns of P-gp expression were recorded: (a) a cytoplasmic staining pattern as noted by dense focal perinuclear staining and a diffuse staining pattern: and (b) a membrane staining pattern as noted by a cell plasma membrane and luminal surface staining pattern. Samples were scored as follows: (a) + , weak or moderate nonhomogeneous intracellular cytoplasmie staining with minimal discrete membranous staining; (b) + +, moderate to strong luminal surface or membranous staining and/or moderate and usually homogeneous diffuse intracellular cytoplasmic staining; and (c) + + + , moderate to strong homogeneous or dense focal perinuclear cytoplasmic staining and strong continuous or discrete cell membrane or luminal surface staining. RESULTS The quantitative PCR methodology used in this study has been well documented by Horikoshi et a!. (37). A schematic of the experimental procedures is shown in Fig. I . A modification of RT-PCR methodology was chosen to determine the levels of MDRJ expression in biopsy samples of benign and malignant endometrial samples. RNA was extracted from the tumor and converted into eDNA. The eDNA was serially diluted and amplified by PCR with primers for the gene of interest and for a gene, usually actin or 3-microglobulin, that was used as an internal standard. Each reaction was done in a separate tube. It is assumed that the internal standard is produced universally and equally in the tissue samples (38). A linear relationship exists between the initial amount of eDNA and the final amount of PCR product. The 5 ‘ primer of each set was modified to include the T7 phage promoter, allowing for subsequent transcription with T7 polymerase. After PCR was performed, the samples were transcribed into RNA using a radiolabeled CTP and analyzed by PAGE. The inclusion of the transcription step increased the sensitivity almost 500-fold and also allowed for accurate quantitation. The transcription products were excised and quantitated by scintillation counting, and the relative amount of MDRJ as compared to aetin was calculated. In general, error is most often introduced in the initial quantitation of tissue RNA. By using this methodology, it is not necessary to know how much RNA is initially present. The level of the gene of interest is compared to an internal standard, and a relative value is calculated. In addition, our study used two independent means to verify the MDRJ levels; each tissue block was divided in half, one section for PCR and the other for immunohistochemistry. Representative autoradiograms of the MDRJ and 3-actin PCR products from a tumor sample and from drug-resistant cell lines are shown in Fig. 2. For each sample, the actual autoradiogram along with the quantitative analysis and the ratio of MDRJ to aetin is included. The results indicate that MDRI expression in this tumor sample is similar to MDRI expression in the low-resistance cell line. The MDRJ:actin ratio of the samples ranges from 0.00008-0.34 (Table 1), with the MDRI: aetin ratio of the highly resistant cell line being 0. 1 7 and of the Research. on July 16, 2017. © 1996 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Reverse transcription with random hexamers Target gene /“ PCR Internal standard MDR-l,/ Amplication with a \ Aetin 4_f 11 promoter containing primer\