Characterization of Adriamycin-resistant Human Breast Cancer Cells Which Display Overexpression of a Novel Resistance-related Membrane Protein*

Development of multidrug resistance due to overexpression of P-glycoprotein (Pgp), a cell membrane drug efflux pump, occurs commonly during in vitro selections with adriamycin (Adr). Pgp-mediated drug resistance can be overcome by the calcium channel blocker verapamil (Vp), which acts as a competitive inhibitor of drug binding and efflux. In order to identify other mechanisms of Adr resistance, we isolated an Adr-resistant subline by selecting the human breast cancer cell line MCF-7 with incremental increases of Adr in the presence of 10 microgram/ml verapamil. The resultant MCF-7/AdrVp subline is 900-fold resistant to Adr, does not overexpress Pgp, and does not exhibit a decrease in Adr accumulation. It exhibits a unique cross-resistance pattern: high cross-resistance to the potent Adr analogue 3'-deamino-3'-(3-cyano-4-morpholinyl)doxorubicin, lower cross-resistance to the alkylating agent melphalan, and a sensitivity similar to the parental cell line to vinblastine. The levels of glutathione and glutathione S-transferase are similar in the parental line and the Adr-resistant subline. Topoisomerase II-DNA complexes measured by the potassium-sodium dodecyl sulfate precipitation method shows a 2-3 fold decrease in the resistant subline. The MCF-7/AdrVp cells overexpress a novel membrane protein with an apparent molecular mass of 95 kDa. Polyclonal antibodies raised against the P-95 protein demonstrate a correaltion between the level of expression and Adr resistance. Removal of Adr but not verapamil from the selection media results in a decline in P-95 protein levels that parallels a restoration of sensitivity to Adr. Immunohistochemistry demonstrates localization of the P-95 protein on the cell surface. The demonstration of high levels of the protein in clinical samples obtained from patients refractory to Adr suggests that this protein may play a role in clinical drug resistance.

Resistance to chemotherapy remains a major obstacle in the treatment of cancer. To study the mechanisms of drug resistance, workers in many laboratories have isolated resistant cell lines in vitro by selecting with various agents including * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ** To whom correspondence should be addressed. adriamycin (Adr)' (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11). These cell lines display the multidrug-resistant phenotype (mdr) demonstrating cross-resistance to multiple structurally unrelated drugs including the anthracyclines, the vinca alkaloids, the epipodophyllotoxins, actinomycin D, and colchicine. Cells with this mdr phenotype have decreased drug accumulation associated with overexpression of a cell membrane glycoprotein, originally described by Ling (12) and referred to as P-glycoprotein (Pgp), which functions as an energy-dependent efflux pump. The resistance to these drugs can be reversed by various agents, which act as competitive inhibitors of Pgp-mediated drug efflux, including the calcium channel blocker verapamil (13)(14)(15)(16)(17)(18). Although other mechanisms have been implicated in the development of Adr resistance (19)(20)(21)(22)(23)(24)(25)(26), assessment of their significance has been hampered by the presence of Pgp overexpression.
In an attempt to identify other mechanisms of Adr resistance besides Pgp, we have isolated an Adr-resistant cell line, MCF-7/AdrVp, by selecting MCF-7 human breast cancer cells with incremental increases in Adr in the presence of 10 pg/ml verapamil to enhance selection for mechanisms other than Pgp. We hypothesized that the presence of verapamil would negate the selective advantage of Pgp such that cells with other mechanisms of Adr resistance or a mutated Pgp insensitive to verapamil would survive. In this study we report the characterization of an Adr-resistant cell line which does not overexpress Pgp and has increased levels of a 95-kilodalton protein which is associated with Adr resistance. at 3000 rpm for 10 min at 4 "C. The pellet was resuspended in 1 ml of a wash solution (100 mM KCl, 1 mM EDTA, 0.1 mg/ml salmon sperm DNA, 10 mM Tris-HCl, pH 8.0) and placed at 65 "C for 10 min with occasional mixing. After cooling on ice for 10 min, the sample was centrifuged and the pellet was washed again. The pellet was resuspended in 0.2 ml of 65 "C water and added to 10 ml of scintillation fluid to determine radioactive counts. mM EDTA, and 10 mM Tris-HCl, pH 8.0) and centrifuged at 3,000 rpm for 5 min. The cell pellet was homogenized in 3 ml of hypotonic buffer (10 mM NaCI, 1 mM EDTA, 1% aprotinin, 20 mM sodium phosphate, pH 7.0) using a tight-fitting Dounce homogenizer for 100 strokes and centrifuged at 1,200 X g for 5 min to obtain a low speed pellet, and then at 100,000 X g for 30 min to obtain a high speed pellet (HSP) and high speed supernatant (32,33

RESULTS
The MCF-7 cell line used in the present study had been frozen in its 66th passage. In the initial step of the selection, a population of MCF-7 cells was exposed to 5 rig/ml Adr in the presence of 10 pg/ml verapamil. After an initial period of time during which the cells continued to grow, only a few clones survived and grew as isolated colonies. One of these clones (MCF-7/AdrVp) was selected for further studies and was advanced as a population by stepwise increases of the concentration of Adr. The Adr-resistant sublines described in the present report, MCF-'I/AdrVp(lO), MCF-7/AdrVp (20), MCF-7/AdrVp(lOO), and MCF-'i/AdrVp(ZOO) were maintained at Adr concentrations of 10, 20, 100, and 200 rig/ml Adr, respectively, in the presence of 10 @g/ml verapamil. The partial revertant sublines MCF-7/AdrVp/R5 and MCF-7/ AdrVp/RG were obtained by growing MCF-7/AdrVp(lOO) in Adr-free medium that still contained various agents is shown in Table I. The MCF-7/AdrVp(lOO) cells were 800-900-fold resistant to Adr. In addition to resistance to Adr, the MCF-7/AdrVp(lOO) subline exhibits very high cross-resistance to the cyanomorpholino derivative of Adr, 3'-deamino-3'- (3-cyano-4-morpholinyl)doxorubicin (38)(39)(40), and the anthracycline, daunorubicin, lower cross-resistance to VM-26 and the alkylating agent melphalan, and a sensitivity similar to the parental cell line for vinblastine, methotrexate, 5-fluorouracil and actinomycin D. MCF-'I/ AdrVP(10) cells were 13-fold resistant to adriamycin while the partial revertant cells, MCF-7/AdrVp/R5, were 30-fold resistant to Adr compared to the parental MCF-7 cells. This cross-resistance pattern differs from that observed in multidrug-resistant cell lines which overexpress Pgp and provides indirect evidence that resistance in this cell line is not mediated by Pgp or a similar drug efflux pump (1,2,6,8,12).
To examine the possibility that an alternate mechanism for reduced drug accumulation was responsible for resistance to Adr we performed drug accumulation studies. Sensitive and resistant cells were incubated with radiolabeled Adr, and the time course of cellular drug accumulation was determined. No decrease in accumulation could be demonstrated through the entire incubation period as shown in Fig. 1. In fact, the MCF-7/AdrVp(lOO) cells showed slightly higher levels of Adr, compared with the uncloned parental MCF-7 cell line. These differences were minor and may be explained in part by differences in cell size. Similar results were obtained with vinblastine.
Although the cross-resistance pattern and the drug accumulation studies indicated that resistance in this cell line was not mediated by Pgp, we confirmed this by analyzing for expression of mdr-l/Pgp by Northern blot analysis using an antisense probe complimentary to the middle third of the mdr-l/Pgp cDNA, as shown in Fig. 2. Messenger RNA from the parental MCF-7 cells and the resistant MCF-7/ AdrVp(lOO) cells and total RNA from a colon cell line selected with Adr which exhibits the multidrug resistance phenotype were analyzed. The results in Fig. 2 show that mdr-l/Pgp mRNA levels in the MCF-7/AdrVp( 100) and in the parental MCF-7 cells are undetectable. In contrast, mdr-l/Pgp expression was easily detected in total cellular RNA isolated from the colon carcinoma cell line.
In addition to overexpression of Pgp we sought to examine the potential role of other putative mechanisms of Adr resistance in our MCF-7/AdrVp cells. The glutathione redox cycle and the enzymes of glutathione metabolism have been implicated as mediators of Adr resistance in some experimental systems (19)(20)(21). In comparing our Adr-resistant MCF-7/ AdrVp(lOO) cell line with the parental MCF-7 cell line, we were unable to find increases in the activity of the glutathione transferases (total glutathione S-transferase activity: MCF-7, 14 units/mg protein; and MCF-7/AdrVp(lOO), 7.7 units/mg protein) or in total GSH/GSSG content (MCF-7, 11.6 nmol/ mg protein; and MCF-7/AdrVp(lOO), 12.4 nmol/mg protein). In addition, depletion of glutathione for 24 h with 5, 10, and 25 PM buthionine sulfoxime, conditions under which GSH content falls to 5-10% of the base-line level, followed by an additional I2 h of buthionine sulfoxime exposure in the presence of varying concentrations of Adr did not alter the relative resistance of the MCF-7/AdrVp(lOO) cell line compared with the parental MCF-7 line (data not shown). These results argue strongly against a significant role for these mechanisms in our resistant cells.
The role of another putative mechanism, alterations in DNA topoisomerase II was examined by measuring the formation of topoisomerase II-DNA cleavable complexes (protein-linked DNA breaks) using a modification of the K-SDS assay. The results of this assay with the topoisomerase II anatagonists, VM-26 and Adr, are shown in Fig. 3. We found a 2-3-fold decrease in precipitable "cleavable complexes" stabilized by VM-26 in MCF-7/AdrVp(lOO) cells, but no difference could be demonstrated after exposure to Adr. Protein levels of DNA topoisomerase II in the parental MCF-7 cells and the resistant MCF-7/AdrVp(lOO) subline were measured by immunoblot analysis using a topoisomerase II-specific antisera (36). As shown in the inset in Fig. 3, there was no difference in topoisomerase II protein levels when exponentially growing cells from the parental MCF-7 cell line and the resistant MCF-7/AdrVp(lOO) subline were compared. The finding of a decrease in precipitable cleavable complexes after VM-26 treatment without alterations in protein levels can be explained by the slower growth rate of the MCF-7/ AdrVp(lOO) cells (doubling times: MCF-7, 22.4 h; MCF-7/ AdrVp(lOO), 79.9 h), and is discussed below. The lack of a difference after Adr treatment can be explained by the small amount of cleavable complexes precipitated in both the parent and the resistant cells, reflecting the fact that Adr performs poorly in this assay.
Having established that previously described mechanisms did not have a significant role in our Adr-resistant cell line, we began a search for other possible explanations. We used a modification of the technique of in-gel renaturation described by Roninson (59) to look for amplified sequences. In this approach, DNA digested with Hind111 is first separated on a 1% agarose gel and subjected to 2 cycles of in-gel renaturation and denaturation without Sl nuclease digestion. A region of the gel extending from about 2.8 to 4.2 kilobases is then excised. This region is chosen because no naturally occurring amplified sequences are present. The DNA in the excised region is electroeluted for cloning into a suitable vector (pGEM3 digested with HindIII).
Only double-stranded DNA can be successfully cloned, and since amplified fragments are more likely to be double-stranded they have a higher likelihood of being cloned, so that the procedure serves to "enrich" for these fragments. Individual clones are then isolated and following plasmid purification, the DNA is radiolabeled and used to probe Southern blots to look for amplification in the DNA from the drug-resistant subline. Using this approach, we were unable to detect amplified sequences in the resistant subline carried at 60 rig/ml Adr although 10% of the 90 clones examined recognized a sequence which was amplified only lofold in both the parental MCF-7 cells and the Adr-resistant subline, probably n-RAS (results not shown).  However, a comparison of the protein pattern of the parental MCF-7 cells and the resistant subline after SDS-PAGE demonstrated increased levels of a 95-kilodalton protein in the MCF-7/AdrVp cells. Although the protein differences could be visualized after Coomassie Blue staining of whole cell extracts, the differences could be enhanced by comparing the HSP fraction which consists primarily of membraneassociated proteins. Labeling with [""Slmethionine also could be used to demonstrate the difference as shown in the left panel of Fig. 4. This analysis was performed numerous times and the difference in the 95-kilodalton protein was consistently observed. Other minor changes (both increases and decreases in protein levels) were inconsistently demonstrated and could usually be explained by differences in loading. In order to prepare polyclonal antibodies against the 95-kilodalton protein, the band was excised from the gel after SDS-PAGE, soaked in water overnight, and then lyophilized.
The lyophilized gel was used to immunize rabbits. Serum obtained after several months contained antibodies not present in preimmunization serum that recognized high levels of the 95kilodalton protein in whole cell extracts and in the HSP fractions of the MCF-7/AdrVp cells and after longer exposures very low levels in the parental MCF-7 cells. The high levels of the 95-kilodalton protein in MCF-7/AdrVp( 100) cells is shown in the right panel of Fig. 4. Two different exposures are presented. Using the polyclonal rabbit antibodies, the levels of the 95kilodalton protein was assessed in various human cell lines and tissues by immunoblot analysis as shown in Figs. 5 and 6. For these studies, total or high speed membrane fractions (HSP) from cultured cells and tissue homogenates were separated on a 10% polyacrylamide gel, transferred electrophoretically to a nitrocellulose filter, and then examined for expression of the 95-kilodalton protein using the rabbit polyclonal antibody. For some samples, several dilutions of pro- Lower levels could be observed in MCF-7 cells after longer exposures.
tein extracts were analyzed to allow for comparative quantitation. Fig. 5 shows the levels of the 95-kilodalton protein in the parental MCF-7 cells and in the Adr-resistant sublines maintamed in increasing concentrations of Adr during the process of selection. In this exposure no signal is detected in the parental MCF-7 cell line although a small amount of protein can be detected when higher amounts of protein are loaded (see Fig. 6). The sublines maintained in increasing concentrations of Adr demonstrate increasing levels of the 95kilodalton protein, as seen in the stepwise increase in levels in the MCF-7/AdrVp(lO), , and MCF-7/AdrVp(200) cells. In addition, Fig. 5 also depicts the results seen in revertants isolated by maintaining the cells in medium free of Adr but with verapamil still present. After 5 and 6 months of growth in the absence of Adr one can see a marked fall in the level of the protein, which paralled a loss of resistance. This figure also demonstrates the results obtained when an MCF-7 cell line that was selected with Adr in the absence of verapamil was examined. This cell line, designated MCF-'I/ AdrMDR has high levels of P-glycoprotein, but no detectable levels of the 95-kilodalton protein. Similar results were obtained with a second MCF-7 subline which was also selected with Adr and expresses high levels of Pgp (not shown). Likewise, colon carcinoma cell line (SW620) selected with Adr which has high levels of Pgp (SW620/AdrMDR) is also negative. The major protein band detected by the polyclonal antibodies is 95-kilodalton in size. This was true in most preparations, as shown in the two lanes at the far right of Fig.  5. Lower molecular weight peptides were occasionally observed and most likely represent degradation products or precursors which as shown on the far right usually comprised only a small fraction of the total signal. Additionally, a 55kilodalton band not present in all lanes in this figure was inconsistently observed, and had no correlation with Adr resistance. In order to further confirm the relationship between expression of the 95-kilodalton protein and Adr resistance, we carried out a second selection and examined a population of cells at low levels of resistance for expression of the 95kilodalton protein. The right panel of Fig. 6 shows the results obtained when such a population was examined after it had been maintained at 2 rig/ml Adr in the presence of verapamil for several passages. Again, one can see increased expression of the 95-kilodalton protein even at these low levels of Adr in the selection medium. One can also see lower levels of this protein in MCF-7 cells when larger quantities of sample are analyzed. Also shown in Fig. 6 sitive revertants begun over 1 year after the original isolation, produced similar results, with disappearance of the g&kilodalton protein by 6 months.

DISCUSSION
In the present study we describe the selection and characterization of an Adr-resistant cell line which possesses a phenotype not previously described and overexpresses a novel surface membrane protein with an apparent molecular mass of 95 kilodaltons. The correlation between expression of the 95kilodalton protein and Adr resistance suggests that this protein may play a role in mediating tolerance to Adr.
Various mechanisms of Adr resistance have been proposed and most investigators agree that several of these may coexist in a given cell line (18,21,25,41,42). Among these, the evidence for overexpression of P-glycoprotein as a mediator of Adr resistance is strongest, with a role for the glutathione redox cycle also implicated in several systems. Previous selec-tions for Adr resistance have resulted in overexpression of Pglycoprotein in cases where the presence of this glycoprotein has been examined. Although other mechanisms have been implicated, the presence of P-glycoprotein overexpression in these cell lines has made interpretation of such results more difficult. In our cell line we could not demonstrate an increase in P-glycoprotein expression, nor decreased drug accumulation In addition, GSH content and activity of the glutathione transferases were not altered and depletion of glutathione with buthionine sulfoxime did not alter resistance.
Besides these two mechanisms, alterations in the topoisomerases have been advanced as a means of resistance to . Cell lines made resistant to topoisomerase I and II antagonists have developed topoisomerase-associated changes (43)(44)(45)(46). The stabilization of the cleavable complexes between DNA topoisomerases and DNA by cytotoxic drugs, rather than the inhibition of enzymatic activity has been suggested as the essential step in cytotoxicity (22,23,30,47). Our Adrresistant cell line demonstrated decreases in the activities of topoisomerase II as measured by the decreased formation of cleavable complexes. This was observed although there was no difference in the levels of topoisomerase II protein measured by immunoblotting. The decreased formation of cleavable complexes can be explained by the slower growth rate of the resistant cells which results in a decreased rate of entry into late S-phase.* Although it is unlikely that this represents an acquired mechanism of resistance, we cannot exclude the possibility that this is an adaptive response that is beneficial to the cell, since rapidly proliferating cells are more sensitive to topoisomerase II specific poisons than slower growing cells (48)(49)(50).
A role for the 95-kilodalton membrane protein in Adr resistance is suggested by our experimental findings. A correlation exists between expression of the 95-kilodalton protein and Adr resistance as demonstrated by Western blot analysis. Increasing levels are observed with increasing Adr resistance. Decreased levels in partial revertants grown in the absence of Adr but in 10 pg/ml verapamil, strengthen the correlation between expression of the 95-kilodalton protein and Adr resistance and provides evidence that expression of the 95kilodalton protein is not related to verapamil. In addition, once acquired, the increased expression of the 95-kilodalton protein was maintained by continuous Adr exposure without verapamil (data not shown). Furthermore, performing the selection with a population of cells resulted in overexpression of the 95-kilodalton protein even at low levels of Adr in the selection medium.
Immunohistochemical localization with anti P-95 antiserum demonstrated the protein on the cell surface. Recently, a surface membrane protein with an apparent molecular mass of 85 kilodaltons was reported in Adr-resistant cell lines overexpressing P-glycoprotein, including the ovarian carcinoma cell line 2780Ad (41). The levels of this 85-kilodalton protein could be modulated by Adr with increases observed following acute Adr exposure. We could not detect the 95kilodalton protein in the 2780Ad cells and were unable to induce its expression by acute Adr exposure in the 2780Ad cells or the MCF-7 cells (data not shown). In addition, when we used the MRK-20 monoclonal antibody to examine our MCF-7/AdrVp cell line, we were not able to detect expression of the 85-kilodalton protein. These observations make it unlikely that these are the same proteins.
It is remarkable that despite its widespread clinical use there exists no consensus on a primary target for Adr. DNA was first thought to be the primary target of Adr, a hypothesis encouraged by the observation that Adr binds double-stranded DNA with a high affinity and the demonstration that drug fluorescence is most prominent in the nucleus following exposure to Adr. Although an attractive hypothesis, many reports in experimental biological systems have demonstrated interactions of Adr with multiple subcellular organelles and macromolecular targets, suggesting there are multiple cellular targets mediating drug activity, including the cell membrane (51-57). Because of its amphipathic structure, Adr has an affinity for plasma membrane bilayers which makes the cell surface a potential drug target. The observation that Adr immobilized on large polymeric beads can be cytotoxic without penetrating cells strongly suggests the cell surface as a target site for the cytotoxic effect of the drug (54). Coupling of Adr to microspheres increased its cytotoxicity against a resistant rat liver cell line providing additional support for the cell membrane as a possible target for Adr (57). Resistance to Adr could thus be mediated by overexpression of a surface membrane protein such as the 95kilodalton membrane protein that is overexpressed in our Adr-resistant cell line. Future work will hopefully elucidate the function of this protein and thus critically address the possibility that it may be a target for Adr.
The search for more potent anthracycline analogues to which cells are not cross-resistant often employs multidrug-resistant cell lines which overexpress Pgp in screening for efficacy. Our results with this cyanomorpholino derivative of Adr indicates that cross-screening with a cell line expressing Pgp may lead to identification of agents capable of overcoming Pgp-mediated Adr resistance, but not other mechanisms.
Although such agents will be potentially very useful clinically, our findings are proof of the value of understanding additional mechanisms of resistance and drug targets for rational drug design.
Finally, the demonstration of high levels of the 95-kilodalton protein in several clinical samples, including cells obtained from a malignant pleural effusion of a patient with breast cancer refractory to Adr, suggests that this protein may play a role in clinical drug resistance. Additional studies are in progress to examine the relationship between overexpression of the P-95 protein and Adr resistance in clinical samples. Identification of the gene encoding this protein will hopefully provide a greater understanding of its role in Adr resistance.