Proc. Nati. Acad. Sci. USAVol. 87, pp. 7160-7164, September 1990Medical Sciences

Quantitative analysis ofMDRI (multidrug resistance) geneexpression in human tumors by polymerase chain reaction

(P-glycoprotein/,2-microglobul/RNA measurement/solid tumors/leukemias)

K. E. NOONAN*, C. BECK*, T. A. HOLZMAYER*, J. E. CHIN*, J. S. WUNDERt, I. L. ANDRULISt,A. F. GAZDAR§, C. L. WILLMAN1, B. GRIFFITH¶, D. D. VON HOFFII, AND I. B. RONINSON*'tt*Department of Genetics, University of Illinois at Chicago, Chicago, IL 60612; tDivision of Orthopedics and tDepartment of Medical Genetics, SamuelLunenfeld Research Institute, Mount Sinai Hospital, Toronto, ON M5G1XS, Canada; National Cancer Institute-Navy Medical Oncology Branch,Naval Hospital, Bethesda, MD 20814; IDepartments of Cell Biology and Pathology, University of New Mexico, Albuquerque, NM 87131;and I"University of Texas Health Science Center, San Antonio, TX 78284

Communicated by Robert T. Schimke, June 14, 1990

ABSTRACT The resistance of tumor cells to chemothera-peutic drugs is a major obstade to successful cancer chemo-therapy. In human cells, expression of the MDR1 gene, encod-ing a transmembrane efflux pump (P-glycoprotein), leads todecreased intracellular accumulation and resistance to a vari-ety of lipophilic drugs (multidrug resistance; MDR). The levelsof MDR In cell lines selected in vitro have been shown tocorrelate with the steady-state levels of MDR1 mRNA andP-glycoprotein. In cells with a severalfold increase in cellulardrug resistance, MDR] expression levels are dose to the limitsof detection by conventional assays. MDR1 expression has beenfrequently observed in human tumors after chemotherapy andin some but not all types of clinically refractory tumorsuntreated with chemotherapeutic drugs. We have devised ahighly sensitive, specific, and quantitative protocol for mea-suring the levels of MDR1 mRNA in clinical samples, based onthe polymerase chain reaction. We have used this assay tomeasure MDRI gene expression in MDR cell lines and >300normal tissues, tumor-derived cell lines, and clinical specimensofuntreated tumors ofthe types in whichMDRI expression wasrarely observed by standard assays. Low levels of MDRIexpression were found by polymerase chain reaction in mostsolid tumors and leukemias tested. The frequency of sampleswithout detectable MDRI expression varied among differenttypes of tumors; MDRI-negative samples were most commonamong tumor types known to be relatively responsive tochemotherapy.

The ability of tumor cells to survive exposure to variousanticancer drugs presents the greatest obstacle to successfulcancer chemotherapy (1). A particularly important type oftumor drug resistance, termed multidrug resistance (MDR),is manifested by cross-resistance to a number of structurallyand functionally unrelated lipophilic drugs. This type ofresistance has been shown to be the result of decreased drugaccumulation in resistant cells due to energy-dependent drugefflux (2). Acquisition of the MDR phenotype in human cellsis the result of expression of the MDR] gene (3-6), whichencodes a 170-kDa membrane glycoprotein, called P-glycoprotein, believed to function as an efflux pump forvarious hydrophobic compounds (7-9). MDR] gene expres-sion is frequently observed in different human tumors, bothuntreated and treated with chemotherapeutic drugs, as wellas in some normal tissues (10-15). Some tumor types knownto be intrinsically unresponsive to chemotherapy, such asrenal cell or colon carcinomas, frequently show high levels ofMDRJ expression prior to drug treatment. In contrast,equally resistant classes of tumors show little if any expres-

sion ofMDR] prior to treatment, when analyzed by conven-tional RNA hybridization or immunological techniques, al-though the frequency ofMDR] expression in these tumors isoften increased after chemotherapy (13). This category oftumors, which includes lung and breast carcinomas, adultsarcomas, and acute leukemias, may display primarily P-glycoprotein-unrelated mechanisms of drug resistance; it isalso possible, however, that low-level or heterogeneousexpression of MDR] in these tumors, undetectable by stan-dard assays, may contribute to their clinical unresponsive-ness.

Analysis of MDR derivatives of the human epidermoidcarcinoma cell line KB (16) has shown that even very lowlevels of MDR] gene expression can confer a severalfoldincrease in the level of drug resistance, which may beclinically significant (4, 15). Detection of these levels ofexpression is close to the lower limit of sensitivity of con-ventional methodologies, even when large amounts of RNAor protein are used for the assays. These results have raisedthe question of whether existing methodologies are suffi-ciently sensitive to detect low levels ofMDR in clinical tumorsamples, where the amount of available tissue is oftenlimited. In addition, analysis ofRNA from clinical samples ishampered by the fact that such samples frequently havesuffered significant degradation. Reliable detection ofMDR]gene expression is further complicated by the existence of ahomologous gene, called MDR2 (sometimes referred to asMDR3) (3, 15, 17), whose products show cross-reactivitywith some MDR1-reactive antibodies and nucleic acidprobes; MDR2 expression, however, has not been associatedwith resistance to chemotherapeutic drugs.To improve the ability to detect low levels of MDR in

clinical samples, we have developed a quantitative assay forMDR1 mRNA expression based on the polymerase chainreaction (PCR) (18, 19) that can be used with small RNAsamples of varying quality. We have used this technique toanalyze MDR] gene expression in MDR cell lines, normalhuman tissues, and >300 clinical tumor samples and tumor-derived cell lines. Our results indicate that low-level MDR]expression, undetectable by conventional assays, frequentlyoccurs in human tumors before chemotherapy.

MATERIALS AND METHODSCell Lines and Tissues. The drug-sensitive KB-3-1 epider-

moid carcinoma cell line and its MDR derivatives (16) were

Abbreviations: MDR, multidrug resistance; PCR, polymerase chainreaction; P32m, 832-microglobulin; SCLC, small cell lung carcinoma;NSCLC, non-small cell lung carcinoma; ANLL, acute nonlympho-cytic leukemia.tTTo whom reprint requests should be addressed at: Department ofGenetics (M/C 669), University of Illinois at Chicago, 808 SouthWood Street, Chicago, IL 60612.

7160

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021

Proc. Natl. Acad. Sci. USA 87 (1990) 7161

obtained from M. M. Gottesman (National Cancer Institute).The small cell lung carcinoma (SCLC), non-small cell lungcarcinoma (NSCLC), and carcinoid cell lines have beendescribed (14). Other cell lines were obtained from theAmerican Tissue Culture Collection. Acute nonlymphocyticleukemia (ANLL) samples were harvested as bone marrow

aspirates, and blast cells were isolated on Ficoll/Hypaquegradients and frozen within 12-24 hr from aspiration. Theblast count in all ANLL samples was >90%. Solid surgicalspecimens of lung and ovarian carcinomas were minced into2- to 5-mm fragments, disassociated by mechanical disrup-tion, and frozen within 30 min to 4 hr after harvesting. Othersolid tumors were frozen within 1 hr after surgical excisionand pulverized while frozen before RNA extraction. Toexclude contaminating normal tissue, tumor samples were

selected in conjunction with histological evaluation of frozensections.RNA Preparation, Quantitation, and cDNA Synthesis. Total

cellular RNA was prepared either by standard methods (20)or by using a nucleic acid extractor (Applied Biosystemsmodel 340A). RNA yield was measured by reaction withorcinol (21). cDNA was synthesized with 1 ,g of total cellularRNA and 100 ng of random hexadeoxynucleotide primer(Pharmacia) (22) in 10 gl of a solution containing 50 mMTris-HCI (pH 8.3), 75 mM KCI, 3 mM MgCl2, 10 mMdithiothreitol, 500 ,xM each dNTP, and 10 units of Moloneymurine leukemia virus reverse transcriptase (Bethesda Re-search Laboratories). After 1 hr at 370C, cDNA was diluted1:5 with water and stored at -30°C until use. The efficiencyof cDNA synthesis was estimated by incorporation of [a-32P]dCTP to be 85-90%.PCR. PCR was carried out with cDNA derived from 50 ng

of RNA, 1 unit of AmpliTaq Polymerase and reaction kits(Perkin-Elmer/Cetus) in a final vol of 25 ,u. Each cycle ofPCR included 30 sec of denaturation at 94°C, 1 min of primerannealing at 55°C, and 2 min of extension/synthesis at 72°C.PCR primers (amplimers) were synthesized by using an

Applied Biosystems DNA synthesizer (model 380A); theprimer yield and quality were tested by UV spectroscopy andgel electrophoresis. Amplimers used for amplification ofB32-microglobulin (82m)-specific sequences were ACCCC-CACTGAAAAAGATGA (residues 1544-1563; sense strand)and ATCTTCAAACCTCCATGATG (residues 2253-2262and 3508-3517; antisense strand) (23); PCR using theseamplimers yields a 120-base-pair (bp) product. MDRI-specific sequences were amplified by using the sense-strandprimer CCCATCATTGCAATAGCAGG (residues 2596-2615) and the antisense-strand primer GTTCAAACTTCT-GCTCCTGA (residues 2733-2752) (8), which yield a 167-bpproduct. Each primer was added at 37.5 pmol per reaction.For quantitation, 2 ,uCi (1 Ci = 37 GBq) of [a-32P]dCTP was

added to each reaction mixture, performed in triplicate. PCRwas carried out in a DNA thermal cycler (Perkin-Elmer/Cetus); this apparatus provided a constant yield of PCR,independent of the position of the sample tube within therack. PCR products were separated on 12% polyacrylamidegels and stained with ethidium bromide. Bands correspond-ing to each specific PCR product were excised from the gelsand the amount of incorporated radioactivity was determinedby liquid scintillation counting. Several negative controlreactions were included in each experiment. Some of thenegative controls contained water instead of cDNA, andother control reaction mixtures contained aliquots of cDNAreaction mixtures that were prepared without the addition ofRNA. All the necessary precautions against contamination ofPCRs (24) were rigorously observed.

RESULTS

MDRJ-specific cDNA sequences were amplified using a pairof amplimers (22, 25) derived from different exons of the

MDR] gene, separated by a long intron (26), to preventamplification of genomic DNA sequences that may contam-inate cellular RNA preparations. Only a small (167 bp)segment of MDRJ cDNA was amplified, so that even signif-icantly degraded RNA could be used as template. TheMDRJ-specific amplimers were selected so they would notamplify the closely related MDR2 cDNA sequence; theirinability to amplify MDR2 was confirmed by using an MDR2cDNA clone (data not shown). Total cellular RNA was usedto synthesize cDNA with a random hexadeoxynucleotideprimer (22), which allowed us to use the product of a singlereverse transcription reaction for amplification of multiplemRNA sequences, regardless of the distance of the amplifiedsegment from the 3' end of mRNA.The yield of PCR product is proportional to the starting

amount of the template only under conditions in which PCRamplification proceeds exponentially at a constant efficiency.To establish conditions for exponential amplification ofMDRJ-specific sequences within the expected range ofmRNA concentrations, we performed kinetic analysis (27) ofMDRI-specific amplification in RNA preparations from aseries ofMDR KB cell lines selected for increasing levels ofresistance to colchicine (16) (Table 1). The experiment in Fig.LA shows that at higher cycle numbers (>25), the yield of theMDRI-specific product in cells expressing high levels ofMDRJ (KB-8-5-11 or KB-8-5-11-24) approaches the plateau,while the slopes of the reactions using cDNA from lower-level MDRI expressors are still in the exponential range.Before saturation at the plateau, however, MDRI amplifica-tion proceeds with the same efficiency in all the samples,which makes it possible to compare the initial amounts ofmRNA template in different cDNA samples.

Since differences in the extent of RNA degradation andpurity between different preparations, as well as samplingand measurement errors, can have a great effect on the yieldof PCR product, the target sequence must be amplified inparallel with an internal control sequence corresponding toanother mRNA expressed in the same cells. An endogenousmRNA sequence should be preferable to an exogenouscontrol, used by other investigators (28), since the yield ofendogenous cDNA sequences reflects the extent of degra-dation and purity of the tested RNA preparation. The controlsequence should be present at similar levels in different typesof cells and at an abundance comparable to that of the targetsequence, so that it will have similar amplification kinetics.In the case of MDR], the levels of cellular drug resistancecorrelate with the density of P-glycoprotein in the cell mem-brane, rather than with the absolute number of P-glycoprotein molecules per cell (K. Choi, T. 0. Frommel,and I.B.R., unpublished data); therefore, mRNA for a mem-brane protein expressed in a cell-surface-dependent mannerwould provide the best internal control. For these reasons,

Table 1. MDR1 mRNA expression in MDR KB cell lines

MDR1 mRNARelative resistance to expression

Colchi- Vinblas- Adria- Slot PCR,Cell line cine tine mycin blot copies per cell

KB-3-1 1 1 1 <0.1 0.06 ± 0.04KB-8 2.1 1.2 1.1 1 1.09 ± 0.21KB-8-5 3.8 6.3 3.2 3 5.15 ± 0.57KB-8-5-11 40 51 23 80 32.17 ± 1.62KB-8-5-11-24 128 20 26 160 74.12 ± 2.08

Derivation of MDR KB cell lines and determination of theirresistance relative to the parental KB-3-1 cell line are described inref. 16. mRNA levels determined by slot blot hybridization (refs. 4and 6; data not shown) are expressed in arbitrary units. The copynumber of MDR1 mRNA in different KB cell lines was estimatedafter 20 cycles of PCR as described in the text.

Medical Sciences: Noonan et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021

7162 Medical Sciences: Noonan et al.

210 5

E

0

0 20 40 60 80 100

MDR1, copies/cell120

10 15 20 25 30 35Number of cycles

FIG. 1. Quantitation ofMDR1 mRNA in MDR KB cell lines. (A) Kinetics of PCR amplification ofMDR] sequences using cDNA from MDRKB cell lines. PCR was performed on cDNA samples from KB-3-1 drug-sensitive cells and a series ofMDR cell lines derived by step selectionwith colchicine (16). KB-Cl is a subclone of the KB-8-5-11-24 cell line. MDRJ-specific bands were separated by PAGE and excised from thegel. The yield is expressed as radioactivity (cpm), determined by liquid scintillation counting. (B) Normalized yield of MDR1 PCR productsrelative to the amount of template. PCR was performed by using mixtures of KB-3-1 cDNA and increasing amounts of a plasmid containingMDR] cDNA diluted to reflect 1-100 copies of MDRI per cell. PCR amplification of MDRJ- and 82M-specific sequences was performed for20 cycles. Each data point represents the average of three assays; the values differed by <5%. (C) Normalized yield of MDRI-specific PCRfor MDR KB cells. The relative yield of MDR/1,82m for the KB series of cell lines is expressed as a function of PCR cycle number. "Blots"refer to relative signal intensities in slot blot hybridization experiments performed with 50 ng of total cellular RNA.

we chose 832m mRNA as an internal control for MDR]expression. The ,f2m gene is ubiquitously expressed (29) andits expression correlates with cell-surface area in most celltypes (30). In addition, we found that the kinetics of ampli-fication of the f3M-specific PCR product are similar to thekinetics of the MDR]-specific product in KB-8-5-11-24 cells(data not shown). The ratio of MDRI/132m-specific PCRproduct for all KB cell lines remained unchanged over a

50-fold range of cDNA concentration and did not change asa result of RNA degradation that occurred after repeatedfreezing-thawing of RNA preparations (data not shown).Amplification of the MDRJ- and /32m-specific sequences canbe done in either the same or separate tubes; under ourconditions, identical results were obtained in both cases. Itshould be noted, however, that some other pairs of PCR-amplified sequences are not proportionally amplified whenreactions are performed in the same tube; separate reactionsneed to, be used in those cases (data not shown).We have used coamplification of,82m- and MDRJ-specific

sequences to determine the copy numbers of MDR1 mRNAmolecules in MDR KB cell lines. The yield of the MDRJ-specific product relative to the ,82m product was compared toa standard curve constructed from relative yields of PCRproducts obtained under identical conditions with increasingamounts of cloned MDR] cDNA mixed with a constantamount ofcDNA from drug-sensitive KB-3-1 cells (Fig. 1B).Under these conditions, the relative yield ofPCR products islinear over the range of 1-100 copies ofMDR] cDNA per cell.The calculated amounts of MDR1 mRNA in MDR KB celllines are shown in Table 1. The PCR-derived estimates ofmRNA copy number are consistent with the relative amountsof MDR1 mRNA in these cell lines determined by filterhybridization.To determine the optimal conditions for detection and

quantitation of MDR] expression, the relative yield of PCRproducts for the KB cell lines was determined after variousnumbers ofPCR cycles, as shown in Fig. 1C. At 40 cycles ofPCR, the relative yields of PCR products are similar for allMDR]-positive cells examined, indicating that this number ofcycles corresponds to the plateau. Under these conditions,however, the difference in relative PCR yield between any ofthe multidrug-resistant cell lines and drug-sensitive KB-3-1cells is at a maximum. At 25 cycles or less, MDRI levels can

be clearly distinguished between all the KB cell lines. Theresults obtained at the lowest cycle number (19 cycles) arealmost indistinguishable from the relative signal intensitiesobtained by slot blot hybridization. These results suggest thata two-step assay may be the most efficient for analyzing alarge number of samples. Using 35 or more PCR cycles in thefirst assay allows for consistent discrimination betweenMDR]-positive and -negative samples. The MDRJ-positivesamples can then be analyzed quantitatively at the secondstep with 25 cycles or less. The levels ofMDR1 mRNA in thetest samples are determined relative to the standard set ofRNA preparations from the KB cell lines, processed simul-taneously with the test samples. The use of these standardsis intended to provide a rapid correlation between the resultsof the PCR assay and the corresponding level of drugresistance.The assay described above was used to determine the

levels of MDR1 mRNA in 28 samples of normal humantissues, 72 tumor-derived cell lines, and 243 surgical samplesfrom 202 different tumors. All cell lines except for 10 SCLClines were derived from untreated tumors. The clinical tumorsamples were obtained from tumors of the types that, ac-

cording to conventional assays, rarely express MDR1 mRNAprior to chemotherapy (13). The results of this analysis,expressed relative to the KB cell lines, are shown in Table 2.A representative autoradiogram of 32P-labeled PCR productsis shown in Fig. 2. Some MDR] expression was detectable in26 of 28 normal tissue samples, with the highest levelsobserved in the adrenal gland, kidney, colon, and liver, inagreement with previous results (11, 15, 31). In contrast,>25% of all tumors and tumor-derived cell lines were nega-tive for MDRI expression. The difference in the frequency ofnegative samples between the normal and tumor tissues ismost readily explained by suggesting that the normal tissueswith a low level of MDR1 mRNA express MDR] only in a

subpopulation of cells, whereas the absence ofMDR] expres-sion in some of the tumors may reflect their clonal origin froman MDRI-negative cell.MDR1 expression in the 58 SCLC, NSCLC, and carcinoid

cell lines used in this study was previously analyzed byconventional RNA slot blotting techniques (13, 14). Theresults of both studies showed very good agreement for allsamples, except that the use of PCR allowed us to detect a

7a)

Proc. Natl. Acad. Sci. USA 87 (1990)

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021

Proc. Natl. Acad. Sci. USA 87 (1990) 7163

Table 2. MDR1 expression in normal and tumor tissuesTumor type Total - + 2+ 3+ 4+ 5+

Normal tissuesAdrenal gland 1 0 0 0 0 0 1Kidney 4 0 0 0 0 4 0Colon 8 0 0 1 0 7 0Liver 2 0 0 0 2 0 0Lung 1 0 0 1 0 0 0Esophagus 1 0 0 1 0 0 0Spleen 1 0 0 1 0 0 0Stomach 1 0 1 0 0 0 0Muscle 3 1 2 0 0 0 0Bladder 1 0 1 0 0 0 0Breast 1 0 1 0 0 0 0Adipose tissue 2 0 2 0 0 0 0Bone marrow 2 1 0 1 0 0 0

Tumor cell linesLung carcinomaNSCLC 28 8 6 11 3 0 0SCLC 28 5 9 8 4 1 1

Carcinoid tumor 6 2 1 1 1 1 0Colon carcinoma 5 0 0 2 1 1 1Bladder carcinoma 2 1 0 1 0 0 0Breast carcinoma 1 1 0 0 0 0 0Leukemia 2 1 1 0 0 0 0

TumorsLung carcinoma 18 2 15 1 0 0 0Ovarian carcinoma 26 5 14 4 3 0 0SarcomaOsteosarcoma 31 (42) 7 11 10 3 0 0Chondrosarcoma 15 (21) 2 9 3 1 0 0Soft tissue sarcomas 40 (55) 7 24 6 3 0 0Ewing's sarcoma 6 (9) 6 0 0 0 0 0

Adamantinoma 1 (4) 0 0 0 1 0 0ANLL 51 24 13 12 2 0 0Lymphoma 1 1 0 0 0 0 0Lipoma 5 (7) 0 3 2 0 0 0Other benign tumors 9 0 6 3 0 0 0

Results are expressed relative to the KB series ofMDR cell lines.-, KB-3-1; +, intermediate between KB-3-1 and KB-8; 2+, KB-8;3+, KB-8-5; 4+, KB-8-5-11; 5+, KB-8-5-11-24. Numbers in paren-theses indicate total number of samples analyzed when different fromthe number of tumors.

number of cell lines with a low level ofMDR] expression thatwent undetected in the previous work (data not shown).There was no significant difference between SCLC andNSCLC cell lines, in spite of the different clinical behavior ofthe corresponding tumors. Since all the SCLC and NSCLCcell lines used in this study were drug resistant to an extentsimilar to or higher than KB-8-5-11 cells (ref. 14; data notshown), it appears that mechanisms other than MDR]expression must be primarily responsible for drug resistancein lung carcinoma cells.The RNA preparations from different sarcoma samples,

almost all obtained from untreated tumors, were tested forMDRJ expression by slot blot hybridization prior to PCRanalysis. Few samples in this group showed a detectablesignal by these assays (data not shown). In contrast, usingPCR we were able to detect low to moderate levels ofMDR1mRNA in the majority of samples of osteosarcoma, chon-drosarcoma, and soft tissue sarcomas. Although drugs withinthe MDR family have been used for adjuvant chemotherapyof patients with these tumors (32, 33), the complete remissionfrequency for these cancers is low. The major exception tothis pattern is Ewing's sarcoma, a disease of late childhoodand adolescence, which is routinely and successfully treatedchemotherapeutically (34). In agreement with this suscepti-

CM

II

0. cm.~~~ .. . :

i'c

:

MDR1-

02m -

FIG. 2. Autoradiogram ofMDR1 and ,32M-specific PCR productsfrom normal human tissues and tumor cell lines. PCR was performedin the presence of [a-32P]dCTP for 35 cycles. After PAGE, the gelwas dried and exposed to x-ray film for 3 hr at room temperature.Arrows indicate the specific product bands for f32m (120 bp) andMDRI (167 bp) sequences.

bility to chemotherapy, we have found no evidence ofMDRIexpression in any of the six tumors of this type.No MDR1 mRNA expression or gene amplification was

detected by filter hybridization in any of the 51 untreated denovo ANLL samples used in this study (data not shown). ByPCR analysis, 27 of 51 samples were found to express MDR],mostly at low levels. ANLL is commonly treated with drugstransported by P-glycoprotein (35), with a 40-50% remissionfrequency, similar to the frequency of MDRJ-negative sam-ples in our study. The majority ofovarian carcinomas showedsome expression ofMDR], but this expression was very lowin most cases (13). Similarly, very low levels of MDR]expression were also observed in surgical samples of lungcarcinoma, in agreement with previous data (13).

DISCUSSIONWe have developed an efficient protocol for quantitation ofgene expression in clinical samples by PCR. The importantaspects of this protocol, which distinguish it from otherproposed strategies for PCR quantitation of mRNA levels(27, 28, 36-38), include the use of a random primer for cDNAsynthesis, determination of the optimal number of PCRcycles for quantitation of the initial amount of mRNA tem-plate within a defined range of concentrations, amplificationof the target sequences in parallel with an internal control,and the use of a calibration curve composed of a standard setof RNA preparations with known concentrations of targetmRNA. For the MDR] gene, f32m mRNA was selected as aninternal control. Since the amount of f82m in different celltypes is not constant but is proportional to the cell surfacearea, this may not be the best control for genes encodingintracellular proteins; genes for some ubiquitously expressedhousekeeping enzymes may provide better internal controlsin those cases. Some types of tumors, including SCLC, havebeen reported to decrease or eliminate expression off32m andother HLA-related genes (39). In our assays, however, thelevels of expression of02m in SCLC cell lines were similar tothose observed in other types of cells. It remains possible that132m expression may be significantly decreased in some tumortypes; if such types are identified, a different internal controlfor MDR] expression should be selected.The results of PCR-based measurements of MDR] gene

expression in MDR KB cell lines, various normal tissues, andtumor-derived cell lines show a good correlation with studiescarried out by traditional hybridization techniques (4, 6, 11,13, 14). The PCR-based assay, however, requires much lesstissue, and it is the only assay that ensures reliable detectionof MDR] gene expression in samples from cells with a lowlevel of drug resistance (KB-8 and lower). The need for highsensitivity in the analysis of clinical samples is obvious, since

Medical Sciences: Noonan et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021

7164 Medical Sciences: Noonan et al.

even a modest increase in the levels of drug resistance maybe sufficient to enable a clinically refractory tumor. Anotherconsideration is that MDR] expression in tumors is oftenheterogeneous, so that only a subpopulation of tumor cellsexpresses the gene (25, 40). Even if some cells in a tumorexpress MDR] at a relatively high level, heterogeneity maylead to a very low signal when total RNA or protein extractedfrom the tumor is analyzed. In addition, selection of MDRJ-specific primer sequences ensures that the PCR-based assaywill not lead to false positives due to cross-reactivity with theclosely related MDR2 gene, which is not associated withMDR.We have used the PCR-based assay to analyze MDR1

mRNA levels in clinical samples of 202 tumors that belongedto types in which MDR] expression has been rarely detect-able by conventional procedures. Our assay showed that ahigh proportion of these tumors did express some MDR1mRNA, in most cases at the levels of KB-8 (1 molecule percell) or lower. A critical question is whether these low levelsof expression, detectable almost exclusively by PCR, wouldbe predictive for clinical resistance to P-glycoprotein-transported drugs. This could be the case if low levels ofMDR1 mRNA reflected the presence of a subpopulation ofP-glycoprotein-expressing cells within the tumor; such cellswould presumably have a growth advantage in the course ofsubsequent chemotherapy. The observed differences in thefrequency of low-level MDRJ-expressing samples betweendifferent classes of tumors suggest that these levels may besignificant for some tumor types. Thus, the proportion ofANLL samples expressing low levels of MDR1 mRNA issimilar to the proportion of patients who fail to respond in thefirst round of chemotherapy. Furthermore, the absence ofMDR1 mRNA in Ewing's sarcomas, which are usually sus-ceptible to chemotherapy, is in sharp contrast to the preva-lence of low-level MDR1 expression in the other morerefractory types of sarcoma. Clinical correlative studiesshould establish which, if any, levels of MDRI expressionwould be prognostically significant for different types oftumors.

We thank Drs. M. M. Gottesman and 1. Pastan for MDR KB celllines; Drs. A. Fojo and D. K. Werasinghe for kind gifts of humantissue RNA; Dr. S. Lai for RNA preparations from tumor cell lines;Drs. R. Bell, A. Czitrom, F. Langer, F. Sim, and L. Wold forsarcoma specimens; and Dr. J. Weick and M. Grever of the South-west Oncology Group for the ANLL samples. This work wassupported by National Cancer Institute Grants CA39365, CA40333(I.B.R.), and CA32102 supporting the Southwest Oncology GroupLeukemia Biology program (C.L.W.); Bristol Myers Drug Resis-tance Grants (I.B.R. and D.D.V.H.); a Faculty Research Awardfrom the American Cancer Society (I.B.R.); a grant from the MedicalResearch Council of Canada (I.L.A.); a grant from the CanadianOrthopedic Foundation and a fellowship from the Medical ResearchCouncil of Canada (J.S.W.); and a postdoctoral fellowship from theNational Cancer Institute (K.E.N.).

1. DeVita, V. T. (1989) in Cancer: Principles and Practice of Oncol-ogy, eds. DeVita, V. T., Hellman, S. & Rosenberg, S. A. (Lippin-cott, Philadelphia), pp. 276-300.

2. Dano, K. (1973) Biochim. Biophys. Acta 323, 466-483.3. Roninson, I. B., Chin, J. E., Choi, K., Gros, P., Housman, D. E.,

Fojo, A., Shen, D.-W., Gottesman, M. M. & Pastan, I. (1986) Proc.Natl. Acad. Sci. USA 83, 4538-4542.

4. Shen, D.-W., Fojo, A., Chin, J. E., Roninson, I. B., Richert, N.,Pastan, I. & Gottesman, M. M. (1986) Science 232, 643-645.

5. Ueda, K., Cardarelli, C., Gottesman, M. M. & Pastan, I. (1987)Proc. Natl. Acad. Sci. USA 84, 3004-3008.

6. Choi, K., Chen, C.-J., Kriegler, M. & Roninson, I. B. (1988) Cell 53,519-529.

7. Juliano, R. L. & Ling, V. (1976) Biochim. Biophys. Acta 455,152-162.

8. Chen, C.-J., Chin, J. E., Ueda, K., Clark, D. P., Pastan, I.,Gottesman, M. M. & Roninson, I. B. (1986) Cell 47, 381-389.

9. Gottesman, M. M. & Pastan, I. (1988) J. Biol. Chem. 263, 12163-12166.

10. Bell, D. R., Gerlach, J. H., Kartner, N., Buick, R. N. & Ling, V.(1985) J. Clin. Oncol. 3, 311-315.

11. Fojo, A. T., Ueda, K., Slamon, D. L., Poplack, D. G., Gottesman,M. M. & Pastan, 1. (1987) Proc. Natl. Acad. Sci. USA 84, 265-269.

12. Dalton, W. S., Grogan, T. M., Meltzer, P. S., Scheper, R. J.,Durie, B. G. M., Taylor, C. W., Miller, T. P. & Salmon, S. E.(1989) J. Clin. Oncol. 7, 415-429.

13. Goldstein, L. J., Galski, H., Fojo, A., Willingham, M., Lai, S.-L.,Gazdar, A., Pirker, R., Green, A., Crist, W., Brodeur, G. M.,Lieber, M., Cossman, J., Gottesman, M. M. & Pastan, I. (1989) J.Natl. Cancer Inst. 81, 116-124.

14. Lai, S.-L., Goldstein, L. J., Gottesman, M. M., Pastan, I., Tsai,C.-M., Johnson, B. E., Mulshine, J. L., Ihde, D. C., Kayser, K. &Gazdar, A. F. (1989) J. Natl. Cancer Inst. 81, 1144-1150.

15. Chin, J. E., Soffir, R., Noonan, K. E., Choi, K. & Roninson, I. B.(1989) Mol. Cell. Biol. 9, 3808-3820.

16. Akiyama S.-I., Fojo, A., Hanover, J. A., Pastan, 1. & Gottesman,M. M. (1985) Somat. Cell Mol. Genet. 11, 117-126.

17. Van der Bliek, A. M., Baas, F., deLange, T. T. H., Kooiman,P. M., Van der Velde-Koerts, T. & Borst, P. (1987) EMBO J. 3,3325-3331.

18. Saiki, R. K., Scharf, S., Faloona, F., Mullis, K., Horn, G. T.,Erlich, H. A. & Arnheim, N. (1985) Science 230, 1350-1354.

19. Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R.,Horn, G. T., Mullis, K. & Erlich, H. A. (1988) Science 239,487-491.

20. Chirgwin, J., Przybyla, A., MacDonald, R. & Rutter, R. J. (1979)Biochemistry 18, 5294-5299.

21. Lin, R. I.-s. & Schjeide, 0. A. (1969) Anal. Biochem. 27, 473-483.22. Noonan, K. E. & Roninson, I. B. (1988) Nucleic Acids Res. 16,

10366.23. Gussow, D., Rein, R., Ginjaar, I., Hochstenbach, F., Seemann, G.,

Kottman, A. & Plough, H. L. (1987) J. Immunol. 139, 3132-3138.24. Kwok, S. & Higuchi, R. (1989) Nature (London) 339, 237-238.25. Roninson, I. B., Patel, M. C., Lee, I., Noonan, K. E., Chen, C.-J.,

Choi, K., Chin, J. E., Kaplan, R. & Tsuruo, T. (1989) in CancerCells, eds. Furth, M. & Greaves, M. (Cold Spring Harbor Lab.,Cold Spring Harbor, NY), Vol. 7, pp. 81-86.

26. Chen, C.-J., Clark, D., Ueda, K., Pastan, I., Gottesman, M. M. &Roninson, I. B. (1990) J. Biol. Chem. 265, 506-514.

27. Chelly, J., Kaplan, J.-C., Maire, P., Gautron, S. & Kahn, A. (1988)Nature (London) 333, 858-860.

28. Wang, A. M., Doyle, M. V. & Mark, D. F. (1989) Proc. Natl. Acad.Sci. USA 86, 9717-9721.

29. Morello, D., Duprey, P., Israel, A. & Babinet, C. (1985) Immuno-genetics 22, 441-452.

30. Lillehoj, E. P. & Poulik, M. D. (1984) in Molecular Immunology,eds. Atassi, M. Z., Van Oss, C. J. & Absolom, D. R. (Dekker, NewYork), pp. 201-214.

31. Thiebaut, F., Tsuruo, T., Hamada, H., Gottesman, M. M., Pastan,1. & Willingham, M. C. (1987) Proc. Natl. Acad. Sci. USA 84,7735-7738.

32. Chang, A. E., Rosenberg, S. A., Glatstein, E. J. & Antman, K. A.(1989) in Cancer: Principles and Practice of Oncology, eds. DeVita,V. T., Hellman, S. & Rosenberg, S. A. (Lippincott, Philadelphia),pp. 1345-1398.

33. Malawer, M. A., Link, M. P. & Donaldson, S. S. (1989) in Cancer:Principles and Practice of Oncology, eds. DeVita, V. T., Hellman,S. & Rosenberg, S. A. (Lippincott, Philadelphia), pp. 1418-1468.

34. Pizzo, P. A., Horowitz, M. E., Poplack, D. G., Hays, D. M. &Kun, L. E. (1989) in Cancer: Principles and Practice of Oncology,eds. DeVita, V. T., Hellman, S. & Rosenberg, S. A. (Lippincott,Philadelphia), pp. 1612-1670.

35. Wiernik, P. H. (1989) in Cancer: Principles and Practice of Oncol-ogy, eds. DeVita, V. T., Hellman, S. & Rosenberg, S. A. (Lippin-cott, Philadelphia), pp. 1809-1835.

36. Rappolee, D. A., Mark, D., Banda, M. J. & Werb, Z. (1988) Science241, 708-712.

37. Kashani-Sabet, M., Rossi, J. J., Lu, Y., Ma, J. x., Chen, J.,Miyachi, H. & Scanlon, K. J. (1988) Cancer Res. 48, 5775-5778.

38. Galliland, G., Perrin, S. & Bunn, H. F. (1990) in PCR Protocols: AGuide to Methods and Applications, eds. Innis, M. A., Gelfand,D. H., Sninsky, J. J. & White, T. J. (Academic, New York), pp.60-69.

39. Doyle, A., Martin, W. J., Funa, K., Gazdar, A., Carney, D.,Martin, S. E., Linnoila, I., Cuttitta, F., Mulshine, J., Bunn, P. &Minna, J. (1985) J. Exp. Med. 161, 1135-1151.

40. Weinstein, R. S., Kuszak, J. R., Kluskens, L. F. & Coon, J. S.(1990) Hum. Pathol. 21, 34-48.

Proc. Natl. Acad Sci. USA 87 (1990)

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021