Znt. J. Cancer: 49,696-703 (1991) 0 1991 Wiley-Liss, Inc.

Publication of the International Union Against Cancer Publication de I'Union lnternationale Contre le Cancer

P-GLYCOPROTEIN EXPRESSION AND SCHEDULE DEPENDENCE OF ADRIAMYCIN CYTOTOXICITY IN HUMAN COLON CARCINOMA CELL LINES Gi-Ming LAI"*,Yi-Nan CHEN',Lyn A. MICKLEY',Antonio T. FOJO'and Susan E. BATES''3 'Medicine Branch, National Cancer Institute, Bethesda, MD 20892, USA. Four human colon cancer cell lines (SW620,LS 180, DLD- I, and HCT- 15) and Adriamycin-resistant sub-lines with varying degrees of P-glycoproteinexpression were studied t o evaluate the reversibility of Adriamycin resistance in human colon cancer. Two groups of cell lines were studied. In the first, including a series of Adriamycin-resistant SW620 and DLD- I sub-lines, and in parental HCT- I 5 cells, P-glycoprotein has a major role in Adriamycin resistance, as evidenced by a correlation between Adriamycin resistance, expression of the multidrug-resistance gene mdr- I and i t s product, P-glycoprotein (Pgp), decreased drug accumulation and reversibility by verapamil. In these cell lines, increasing doses of verapamil are required t o fully reverse increasing levels of resistance. In the second group, including parental SW620, DLD-I and LS 180 cells and Adriamycin-selected LS I80 sub-lines, P-glycoprotein does not have a major role in Adriamycin resistance. There was correlation between the schedule dependence of Adriamycin cytotoxicity and the role of P-glycoproteinin modulating reristance. In the cell lines in which P-glycoprotein was a major determinant of Adriamycin resistance, the drug exposure (defined as the product of the concentration and the time of treatment) needed t o achieve a given percent cell kill was reduced as much as 9-fold when cells were treated for 7 days as compared with 3 hr. By comparison, in cell lines in which P-glycoprotein played a lesser role, the drug exposure necessary to achieve a given percent kill increased under conditions of continuous treatment. In some human colon carcinoma cell lines Pgp appears t o play a significant role in resistance t o Adriamycin, and this can be overcome by the use of competitive inhibitors of Pgp. The increased sensitivity with continuous treatment observed in cell lines with P-glycoprotein-mediated resistance suggests that administration of drugs by continuous infusion may be valuable in reversing clinical drug resistance mediated predominantly by P-glycoprotein.

Human colorectal cancer is resistant to most chemotherapeutic agents. At the present time, 5-FU and its analogs are the most effective drugs but have a low overall response rate ranging from 15 to 20%. No significant response occurs with Adriamycin, vincristine, or other natural products. The mechanisms responsible for this intrinsic drug resistance are unknown, although over-expression of the cytotoxic drug efflux pump, P-glycoprotein (Pgp), and alterations in the glutathione redox cycle have been proposed (Fojo et al., 1987; Kramer et al., 1988). Most studies of drug resistance use cell lines selected in vitro by exposure to chemotherapeutic agents. One resistance mechanism identified by this approach is over-expression of the multidrug-resistance gene rndr-1 and its protein product, Pgp. Pgp is a 130-to-180-kDamembrane phosphoglycoprotein which mediates drug resistance by increasing drug efflux, with a resultant decrease in intracellular drug accumulation (Fojo et al., 1985; Kartner et al., 1983; Biedler and Riehm, 1970; Ling and Thompson, 1974; Beck et al., 1979). Pgp serves this transport function as an energy-dependent drug efflux pump, similar to that seen in bacterial transport systems (Chen et al., 1986). Elevations in Pgp levels can be detected in cell lines selected in vitro with hydrophobic natural products such as Adriamycin, actinomycin D, or the Vinca alkaloids (Fojo et al., 1985; Kartner et al., 1983; Biedler and Riehm, 1970; Ling and Thompson, 1974; Beck et al., 1979), and can also be found in tumor samples obtained from patients (Fojo et al., 1987; Bell et

al., 1985; Dalton et al., 1989). Some cancers, including colon, adrenal, liver, and kidney cancers, express rndr-1 before treatment (Fojo et at., 1987). These tumors arise from tissues which normally express mdr-1. The physiologic role of Pgp in normal tissues remains unclear, but it is associated predominantly with cells possessing secretory functions, or with cells which are exposed to naturally occurring toxins (Fojo et al., 1987; Fairchild et al., 1987; Thorgeirsson et al., 1987). Such findings suggest a role for Pgp in the excretion of toxins in some normal cells. Expression of Pgp in colon cancer may account in part for resistance to agents derived from natural products: a pump normally used for exporting toxins is subverted in cancer cells for exporting chemotherapeutic agents. Calcium channel blockers such as verapamil can compete with cytotoxic drugs for Pgp, thereby blocking their efflux and raising intracellular drug levels (Cornwell et al., 1987; Safa et al., 1986; Tsuruo et al., 1983). Reversal of decreased drug accumulation by verapamil in drug-resistant cells is a hallmark of Pgp-mediated drug resistance. Through the modulation of drug cytotoxicity with verapamil one can infer a role for Pgp in resistance to chemotherapy. In the present study, we sought to examine the reversibility of drug resistance mediated by Pgp in human colon carcinoma cell lines. To do this, we initially selected the SW620 cell line because it is as Adriamycin-sensitive as any we have examined in the laboratory, and compares with drug-sensitive parental lines utilized in other studies (Akiyama et al., 1985; Cowan et al., 1986). We reasoned that by beginning with the SW620 cell line, we could isolate sub-lines expressing Pgp as the predominant mechanism of Adriamycin resistance. In order to extend some of the observations io other in vitro systems, 3 other human colon carcinoma cell lines were subsequently examined: LS 180, HCT-15 and DLD-1. All 4 parental lines were derived from tumors obtained from patients who had never received chemotherapy (Leibovitz et al., 1976; Tibbetts et al., 1977; Tom et al., 1976). These lines represent a spectrum of drug sensitivity to Adriamycin, ranging from the most sensitive SW620 to the most resistant HCT-15 line. In addition, sublines selected in vitro for resistance to Adriamycin were obtained from 3 of these cell lines. We used the cell lines and sub-lines to ask the following questions: (1) What is the contribution of Pgp to resistance in unselected and Adriamycinselected human colon carcinoma cell lines? ( 2 ) How effectively can resistance mediated by Pgp be modulated? (3) How effectively can resistance mediated by Pgp be modulated by changing the schedule of drug administration?

'Current address: Division of Hematology and Oncology, Lin-Kou Medical Center, Chang Gung Memorial Hospital, Taipei, Taiwan, Republic of China.

3To whom correspondence and reprint requests should be addressed, at Building 10, Rm. 12N226, National Institutes of Health, 9000 Rockville Pike, Bethesda, MD 20892, USA. FAX 301-402-0172. Received: December 18,1990 and in revised form June 26,1991.

ADRIAMYCIN RESISTANCE IN COLON CANCER CELL LINES

MATERIAL AND METHODS

Chemicals and reagents RPMI 1640, fetal bovine serum (FBS), penicillin/streptomycin, glutamine, Dulbecco’s modified Eagle’s medium (DMEM) with 1,000 mg D-glucose/liter without sodium bicarbonate, and 4-(2-hydroxyethyl)-l-piperazineethanesulfonicacid were obtained from GIBCO, Grand Island, NY. Type-VII agarose was from Sigma, St. Louis, MO. Verapamil hydrochloride was from QUAD Pharmaceutical, Indianapolis, IN. Adriamycin was supplied from Adria, Columbus, OH. Tween-20 and the protein assay kit were BioRad (Richmond, CA) products. Monoclonal antibody (MAb) C219 was obtained from Centacor (Malvern, PA). 3H-vinblastine (23 Ci/mmol), and ‘,,IProtein A were purchased from Amersham (Arlington Heights, IL). Uridine 5’-[alpha-32P]triphosphate(3000 Ciimmol) was obtained from NEN (Boston, MA). The riboprobe Gemini-I1 system was provided by Promega (Madison, WI). Cell lines SW620, LS 180, HCT-15, and DLD-1 were obtained from ATCC, Rochille, MD, and were grown in RPMI media containing 10% FBS, 2 mM glutamine, 100 U/ml penicillin and 50 pg/ml streptomycin at 37°C in a 5% CO, humidified atmosphere. The resistant sub-lines that were selected stepwisely in vitro for resistance to Adriamycin (Ad) were maintained in drug-containing media at concentrations of 20, 100, 300 and 1000 ng/ml for SW620 Ad20, Ad100, Ad300 and AdlOOO respectively. Similarly, LS 180 Ad50 and Ad150 and DLD-1 AdlOOO were maintained in 50, 150 and 1000 ngiml Adriamycin respectively. All the cells were out of drug for a minimum of one week prior to any studies. Clonogenicassays The cytotoxicity of Adriamycin in the parent cells and their adriamycin resistant sublines was measured by double-layer soft agarose clonogenic assay in triplicate plates as previously described (Louie et al., 1985). The colonies were counted on the Bausch and Lomb FAS-I1 analysis system. Each experiment was performed at least twice, and in most cases 4 times. The IC,, was the drug concentration at which 50% of colony formation was inhibited. Relative resistance was calculated by comparing the IC,, of a given cell line to that of the parental cell line from which it was derived (SW620, LS 180 and DLD-1). The dose-modifying factor (DMF) represents the ratio of the IC, of Adriamycin alone to the IC,, of Adriamycin plus verapamil.

Drug accumulation assay Drug accumulation studies were performed in 6-well dishes as previously described (Fojo et al., 1985). Hybridizationprobes Two mdr-1/Pgp probes were used: (1) for Northern blot analysis, a 1.4-Kb cDNA containing sequences from the middle third of Pgp in a pGEM vector was used for riboprobe synthesis; (2) for RNase protection experiments, a 1-Kb Pst I genomic fragment containing sequences of the mdr-1 promoter plus additional 5’ sequences (Ueda et al., 1987) was subcloned into pGEM and used for synthesis of a riboprobe. The predominant fragment protected by mRNA is 130 bp. Fragments of slightly varying length are protected by mRNA arising from alternate start sites.

RNA extraction, electrophoresis and filter preparation Total cellular RNA was extracted from cells by homogenization in guanidine isothiocyanate, centrifugation over a cesium chloride cushion, and electrophoresis in 1% agarose, 6% formaldehyde gels. Gels were stained with 2 pg/ml ethidium bromide to allow inspection of the quantity and quality of RNA. RNA gels were subjected to partial alkaline hydrolysis

697

before Northern transfer to nitrocellulose by capillary blot. All filters were baked for 2 hr at 80°C under vacuum. Hybridization Synthetic riboprobe was prepared by SP6 polymerase transcription, using uridine 5’-triphosphate, tetra (triethylammonium) salt [al~ha-~’P] (specific activity, 3,000 Ci/mmol). Filters were prehybridized for 4 hr at 55°C in 50% formamide, 5~ Denhardt’s, 5x SSC, 0.1% SDS, and 200 pg/rnl salmon sperm DNA and hybridized for 16 to 18 hr in 10 ml of the same buffer with 2 x lo6 cpm/ml labelled riboprobe ( l x SSC = 0.15 M NaCI/l5 mM sodium citrate). Filters were washed twice for 30 min with 1x SSC, 0.1% SDS heated to 6 8 T , and once for 60 rnin with 0 . 1 SSC, ~ 0.1% SDS in a 68°C water bath. Autoradiography was performed using Kodak XAR-5 film exposed for 2 to 4 days at -70°C. m u s e protection assay Protection of 1 X lo6 cpm labelled riboprobe from ribonuclease digestion by hybridization with 5 or 50 pg total RNA was performed. Samples were denatured at 85°C for 5 min, followed by overnight incubation at 45°C in 80% formamide, 0.4 M NaCI, 40 mM Pipes, 1 mM EDTA, 2 mM LiCI,. The samples were then digested for 1 hr at 30°C with 40 pgiml RNase A and 28 U/ml RNase T,, and treated for 15 rnin at 37°C with 150 pg/ml proteinase K. Following extraction, samples were separated on a 6% polyacrylamide gel at 1,500V for 2 to 3 hr. Westernblotting Cells were harvested in STE buffer (20 mM Tris-HC1, pH 7.3; 0.15 M NaCI; 1 mM EDTA) and pelleted by centrifugation at 1,000 rpm for 5 min. Crude membranes were prepared by the method of Riordan and Ling (1979) without separation on sucrose gradients. Protein concentrations were determined by the BioRad protein microassay and equivalent amounts of membrane protein from each sample were separated on SDSiPAGE gels. The gels were then electroblotted onto nitrocellulose membranes. The blots were pre-incubated in Blotto buffer (10% Carnation instant non-fat dry milk in Dulbecco’s phosphate-buffered saline (DPBS)) at 24°C for 1 hr to block non-specific antibody binding. The blots were then incubated in Blotto buffer with 2 pg/ml C219 monoclonal antibody at 4°C for 24 hr. After the nitrocellulose had been washed 3 times, 5 min each time, in DPBS with 0.5% Tween-20, the blots were incubatedwith ‘251-ProteinA (5 X lo5 cpm/ml) for 90 rnin at 24°C in Blotto buffer plus 0.3% Tween-20. The wash procedure was repeated and autoradiography was performed at -70°C. RESULTS

Adriamycin cytotoxicity assays The original goal of this study was to study the role of Pgp in mediating multidrug resistance in colon cancer. In order to study this in a model where other mechanisms of resistance would not interfere, we chose SW620 cells as the parental line for the selection of several drug-resistant sub-lines. As shown in Figure 1, SW620 cell lines isolated following stepwise increments in Adriamycin were characterized. These sub-lines were 28-, 53-, 76-, and 108-fold more resistant than the parental line. The resistance to Adriamycin shown in Figure 1 for each of the selected cell lines was determined by clonogenic assay under conditions of continuous Adriamycin exposure. As in other systems, the fold of acquired resistance does not increase at the same proportion as the level tolerated in tissue culture (Inaba et aL, 1979; Kartner et al., 1983; Biedler and Riehm, 1970; Ling and Thompson, 1974; Becket al., 1979). For instance, the SW620 AdlOOO cells are cultured in 10 times more Adriamycin than the SW620 Ad100 cells, but the increase in relative resistance by clonogenic assay is 2-fold

W ET AL.

698

% SW620

100 IC,,

Relative

(ngiml) Resistance A SW620

80

0.5 14.0

A Ad20 0 Ad100 Ad300 Ad1000

-

+

60

26.3 38.0 54.0

'

*

O

F

28x 53x 76x lOBx

0 c

E

40

a 20

0 0.01

0.1

1

10

100

1000

Adriamycin, ngiml

FIGURE 1- Evaluation of Adriamycin cytotoxicity in the SW620 human colon cancer cell line and Adriamycin-resistant sub-lines. Cytotoxicity curves were obtained by clonogenic assay. Colonies were counted after 7 days of treatment. The IC, values and relative resistance to Adriamycin are shown on the right. The IC,, is the concentration of drug at which colony formation was 50% of control. The relative resistance compares the IC,, of Adriamycin of the resistant sub-line to the IC,, of parental SW620 cells.

rather than 10-fold. These lines were used as models for examining Adriamycin resistance in human colon cancer. In order to determine the contribution of a classical Pgpmediated resistance mechanism to the Adriamycin resistance of these cells, the sub-lines were further characterized. Enblastzne accumulation studies Pgp transports Adriamycin plus a variety of other natural products. Vinblastine was chosen for the drug-accumulation assays, since the labelled vinblastine compound is very stable, has high specific activity and gives reliable and reproducible results. A broad cross-resistance pattern due to Pgp would result in decreased 'H-vinblastine accumulation as well as Adriamycin resistance. The solid bars in Figure 2 show the level of drug accumulation in the SW620 parental line and its sub-lines. 3H-vinblastineaccumulation decreased in a stepwise manner with increasing Adriamycin resistance. Drug accumulation was reduced to 27% of the parental level in the SW620 Ad20 sub-line and fell to 21%, 7%, and 3% of the parental value in SW620 Ad100, Ad300, and AdlOOO sub-lines. Verapamil increased vinblastine accumulation in all the cell lines from 1.1- to 8.8-fold, as shown by the hatched bars in this figure. However, verapamil at this dose (10 pglml) and duration (1 hr) failed to return vinblastine accumulation to the SW620 level in any of the lines. The stepwise decrease in vinblastine accumulation seen in these cell lines, and its reversal by verapamil, suggests a role for Pgp in mediating some or all of the drug resistance observed. Correlation with mdr-1 lPgp expression To confirm that increased mdr-I/Pgp expression was present, as inferred from the decreased drug accumulation in the SW620 resistant sub-lines, Northern and Western analyses were performed. As shown in Figure 3a, the levels of mdr-1 mRNA rose stepwise as resistance increased. The SW620 Ad20 cells expressed 34-fold more mdr-1 mRNA than the SW620 parental cells, and the increase in the SW620 AdlOOO was 249-fold over that of the SW620 parental cells. An ethidium bromide stain of the comparably loaded Northern gel is shown in Figure 3b. The Western blot analysis in Figure 3c demonstrates stepwise increases in Pgp which are commensu-

u-

SW620

Ad20

Ad100

Ad300

AdlOOO

SW-620 series FIGURE2 - Studies of vinblastine accumulation in SW620 and Adriamycin-resistant sub-lines. Exponentially growing cells were incubated with 3H-vinblastinefor 1 hr at 37°C in the presence (a) or absence (W) of 10 pglml verapamil. Drug retained in cells after washing in cold PBS was scintillation counted. The results are normalized to cell number and are shown as percentage of parental SW620 level, which was assigned a value of 100%. Accumulationswere performed in triplicate, with variations of less than 10% of the total counts.

rate with expression of the mdr-1 mRNA. This increase in mdr-1 mRNA and protein correlated quantitatively with the decrease in vinblastine accumulation and the increase in Adriamycin resistance in the SW620 Ad series. This observa;ion is depicted graphically in Figure 3d. The relative decrease in vinblastine accumulation (RDVA) expresses the fold decrease in accumulation relative to the SW620 parental line, calculated from the data presented in Figure 2. EfSect of verapamil on Adriamycin cytotoxicity As suggested by the drug accumulation and mdr-11Pgp expression data, resistance in SW620 Adriamycin-resistant cells appears to be mediated primarily by Pgp. Confirmation of the role of Pgp in the resistance of the SW620 sub-lines is provided by dose-response studies with verapamil performed by clonogenic assay. The results in the SW620 Ad series are shown in Figure 4.Resistance in both the SW620 Ad20 and the Ad100 cells was completely reversed with 10 pg/ml verapamil with DMFs of 33 and 45 respectively. That full reversal was achieved in this analysis, although vinblastine accumulation did not reach SW620 levels in Figure 2, is most likely explained by the different experimental conditions (1 hr vs. 7 days). In the SW620 Ad300 cells, 20 pglml verapamil was required to achieve nearly full reversal of resistance. Complete reversal was not possible in the highly resistant SW620 Ad1000 cells, even though 50 pg/ml verapamil was used (data not shown). Higher doses of verapamil could not be used because they were themselves cytotoxic. The most likely explanation for these results is that the level of Pgp is so high that the verapamil dose is not sufficient to compete effectively. Although it is also possible that Pgp in highly resistant lines has less affinity for verapamil than the selecting agent, this appears unlikely, since the primary structure of Pgp in these sub-lines is unchanged (data not shown).

ADRIAMYCIN RESISTANCE IN COLON CANCER CELL LINES

FIGURE 3 - Expression of mdr-1 gene and P-glycoprotein (Pgp) in the SW620 series. (a) 8 pg of total RNA from each cell line was hybridized with a synthetic antisense probe complementary to the middle 1/3 of the mdr-1 mRNA. A faint signal was seen in the parental cell line on longer exposures. Expression in SW620 Ad20 and SW620 Ad1000 was 34- and 249-fold higher than in the SW620 parent. No gene amplification was found by Southern analysis (data not shown). (6) Ethidium bromide stain of gel used in Northern analysis shows intact, comparably loaded RNA. (c) Western blot detection of Pgp by binding of "'I-labelled C219 monoclonal antibody to 100 pg crude membrane protein. Faint Pgp staining in the parental lane was observed with 150 pg protein in longer exposures. (d) For each SW620 Ad series sub-line, the levels of mRNA, Pgp, relative resistance (RR) and relative decrease of vinblastine accumulation (RDVA) are shown relative to the SW620 parent, which was assigned a value of one. The RR is obtained from Figure 1 and the RDVA is determined by dividing 100by the percent accumulation of the Ad series sub-line shown in Figure 2. For mdr-1 and Pgp expression, the OD values from densitometry of the Northern and Western blots in (a) and (c) are normalized to SW620 values.

699

Characterization of additional colon carcinoma cell lines In order to extend these observations, we examined 3 additional cell lines, LS 180, DLD-1 and HCT-15, also established from previously untreated patients. In addition, resistant sub-lines derived by selection in Adriamycin from LS 180 and DLD-1 cells, and designated LS 180 Ad50, LS 180 Ad150 and DLD-1 AdlOOO were also studied. LS 180 Ad50 and LS 180 Ad150 were 5.8- and 7.7-fold resistant compared with LS 180, while DLD-1 AdlOOO cells were 20.5-fold more resistant than parental DLD-1 cells. The resistance of these cell lines covered a similar range to that of the SW620 sub-lines. Vinblastine accumulation in these cell lines (expressed as percentage of SW620 accumulation) is shown in the solid bars in the upper panel of Figure 5 and is summarized in Table I. Parental DLD-1 and HCT-15 cells are 1.7 to 2.0 times more resistant to Adriamycin than SW620 Ad20 and have accumulation reduced to a level similar to that in SW620 Ad20 cells when compared to SW620.Zn vitro selection of DLD-1 cells for Adriamycin resistance resulted in a further reduction in accumulation to 3 to 5% of parental SW620 cells in the highly resistant DLD-1 AdlOOO sub-line, as in SW620 Ad1000 cells. No decrease in drug accumulation was observed in Adriamycinselected LS 180 Ad150 cells compared with parental LS 180. The levels of expression of mdr-1 mRNA in SW620, LS 180, HCT-15 and DLD-1 cells and their sub-lines were determined by RNase protection analysis of 5 or 50 p,g of total RNA and are shown in the lower panel of Figure 5 and quantitated in Table I. The results with the SW620 Ad sub-lines confirm those seen in Figure 3. DLD-1 and HCT-15 parental cells, also with resistance comparable to SW620 Ad300, demonstrate levels of mdr-1 mRNA less than the SW620 Ad20 cells. DLD-1 AdlOOO cells, like SW620 AdlOOO cells, have very high levels of mdr-1 expression. Both LS 180 and LS 180 Ad50 cells express low levels of mdr-1 mRNA (detected on longer exposures) in spite of resistance in the latter comparable to the SW620 Ad300. To further delineate the role of Pgp in Adriamycin resistance in these cell lines, cytotoxicity studies without and with verapamil were performed and are summarized in Table I. As shown earlier, verapamil increases Adriamycin cytotoxicity in a dose-dependent manner in the SW620 sub-lines. In contrast, although the LS 180 Ad50 and Ad150 cells are 6- to 8-fold more resistant than parental cells, Adriamycin cytotoxicity is not enhanced more than in parental LS 180 cells. In DLD-1 cells, verapamil increases Adriamycin toxicity by a factor (DMF) of only 3.4, so that SW620 sensitivity is not achieved. The explanation for this incomplete reversal by verapamil in DLD-1 is not simply a quantitative one in terms of mdr-1 expression, since full reversal could be achieved in cells with higher levels of Pgp, including SW620 Ad20 and SW620 Ad100. In the Adriamycin-resistant subline, DLD-1 Ad1000, however, verapamil increases sensitivity by S f o l d , a value similar to that in the SW620 Ad sub-lines, and consistent with the observations of decreased drug accumulation and increased mdr-1 /Pgp expression observed upon drug selection. Finally, in HCT-15 cells the dose-modifying factor for verapamil was 12. This value, which compares with the results observed in the SW620 Ad series cell lines with predominantly Pgp-mediated resistance, suggests that in this unselected cell line Pgp has an important role in mediating Adriamycin resistance. Thus, the cytotoxicity reversal by verapamil suggests that the cell lines can be divided into 2 groups based on the contribution of Pgp to Adriamycin resistance. The group expressing Pgp as a major component includes the SW620 Ad sublines, DLD-1 Ad1000 and HCT-15. The group in which Pgp plays a minimal or less important role includes SW620, LS 180, LS 180 Ad50, LS180 Ad150 and DLD-1 cells.

700

LA1 ET AL.

(B)S W 6 2 0 Ad100

% Control

100

YPQE 10

0.1

100

1000

Adriamycin, ng/ml

Adriamycin, ng/ml % Control

10

1

46

(C) SW620 Ad300

100

100

(D) S W 6 2 0 Ad1000

% Control

!lRpbdE

5 10.5 10 24

80

80

60

60

40

40

20

20

0 0.1

1

10

100

1000

Ad riamycin, ng/ mI

0 0.1

I

I

l l l l l l l

1

I

I

l l l l l l l

10

I

I

b

I 1 1 1 1 1 1

100

I

I

I 1 1 1 1

1000

Ad r iamycin, ng/ mI

FIGURE 4 - Dose-dependent enhancement of Adriamycin cytotoxicity by verapamil (Vp) in the four SW620 Ad sublines. The cytotoxicity curve of each Adriamycin resistant subline (*-*) is compared with that of parental SW620 cells (O---e). Complete reversal of resistance by Vp 10 kg/ml x-x) is found in SW620 Ad20 (a) and Ad100 (b). The response to variable doses of Vp (pg/ml) at 5 (U-a), 10 ( e x ) and 20 (H&) is shown in SW620 Ad300 (c) and SW620 Ad1000 (d).

Drug schedule and Adriamycin resistance During the cytotoxicity studies, it became evident that sensitivity to Adriamycin was influenced by the schedule of Adriamycin exposure, independent of modifying agents. In these studies, cells were treated with Adriamycin for either 3 hr or continuously for 7 days. For the 3-hr experiments, cells

were incubated with varying doses of Adriamycin at 37°C for 3 hr before harvesting and plating in the double-layer soft agarose system. For the 7-day experiments, cells were plated in the double-layer soft agarose system in the presence of Adriamycin. As expected, a lower IC,, for the longer duration of exposure was noted in comparing 7 days of Adriamycin

701

ADRIAMYCIN RESISTANCE IN COLON CANCER CELL LINES TABLE I - ENHANCEMENT OF ADRIAMYCIN CYTOTOXICITY BY VERAPAMIL (Vp) IN CLONOGENIC ASSAY‘

Control

SW620 SW 620 Ad20 SW620 Ad1000 LS 180 LS 180 Ad50 LS 180 Ad150 DLD-1 DLD-1 AdlOOO HCT-15

Verapamil

IC5,Z

IC,”

DMF’

0.6 12.0

0.5 0.85

1.2 14.0

76.0

6.6 ..

i2.n

5.3 31.0 41.0 20.5 420.0 24.0

3.3 16.5 26.0 6.0 28.0 2.0

1.6

~~

1.9

1.6 3.4

15.0 12.0

‘Cellswere plated in soft agarose containingvariable doses of Adriamycin with 10 pgiml Vp, and counted after 7 days.-*The IC,, in ng/ml of Adriamycin was determined by clonogenic assay.”DMF (dose-modifying factor) is the ratio of the IC,, of Adriamycin alone to the IC,, of Adriamycin plus Vp.

FIGURE 5 - Vinblastine accumulation and expression of mdr-1 in parental colon cancer lines and Adriamycin-resistant sublines. (a) Vinblastine accumulation in the presence (B) or absence (m) of 10 p,g/ml verapamil. The values, as in Figure 2, were normalized to cell number, and are shown as percentage of SW620 accumulation. Accumulations were performed in triplicate, with variations of less than 10% of the total counts. (b) RNase protection analysis of 5 pg of total RNA (except 50 bg for SW620 and LS 180 cells) hybridized in solution with an mdr-1-specific antisense probe complementary to sequences of the mdr-1 promoter.

treatment with 3 hr of treatment. However, as shown in Table 11, the relative resistance of the cells with Pgp-mediated resistance increased with decreasing duration of exposure to Adriamycin. Thus, for example, the relative resistance of SW620 Ad1000 cells is 108 when drug is present for 7 days and 4190 following a 3-hr treatment with Adriamycin. Likewise, the relative resistance for DLD-1 Ad1000 cells changed from 20.5-fold after a 7-day incubation to 650-fold after a 3-hr treatment. These cell lines had high levels of mdr-I, and both the drug accumulation data and the cytotoxicity studies in the presence of verapamil were consistent with a predominant role for Pgp in modulating Adriamycin resistance. In contrast, in the Adriamycin-selected LS 180 sub-lines, no difference in relative resistance was observed between the 2 schedules of treatment. In these cell lines, although Pgp could be detected, the importance of other mechanisms of resistance was impli-

cated by the high resistance relative to the low level of Pgp and the incomplete reversibility after verapamil. These differences in relative resistance reflect changes in sensitivity to Adriamycin and can be quantitated by comparing the “drug exposure” under different conditions. Drug exposure (DE) is defined as the product of the drug concentration and the time of treatment and in the table is calculated by multiplying the IC,, in ng/ml by the number of hours of drug treatment. For example, for SW620 AdlOOO cells, the drug exposure after 3 hr equals 72900 ng . hr/ml, while that after 7 days is 9072, or about 1/8th the 3-hr value. That is, one eighth as much drug exposure achieved a comparable cell kill following 7 days’ incubation as with a 3-hr treatment. Similar values were obtained with the DLD-1 AdlOOO cells which, like the SW620 Ad1000 cells, have predominantly Pgp-mediated Adriamycin resistance. Likewise, the HCT-15 cell line, which had a high DMF with verapamil treatment, demonstrates decreased drug requirement with long exposure. The decrease of 1.7-fold observed, although not as large a value as that obtained with the SW620 AdlOOO and DLD-1 Ad1000 sub-lines, is similar to the result obtained with SW620 Ad20 cells, and may be explained by the lower levels of Pgp present in this cell line, or by the co-existence of other mechanisms of resistance which are adversely affected by longer drug exposures. In contrast, parental SW620, LS 180 and DLD-1 cells and the LS 180 Ad sub-lines require higher drug exposures in the 7-day incubation. For example, in DLD-1 cells, the drug exposure needed to achieve a comparable 50% cell kill (IC5,,) increased by 4.8-fold. DISCUSSION

Human colon carcinoma represents a complex model in which to study multidrug resistance. Resistance to chemotherapy, particularly the natural products, is not surprising, since this malignancy arises from cells which are constantly exposed to naturally occurring toxins. Despite the high prevalence of the disease, the mechanisms underlying resistance are poorly understood. The discovery that colon cancer cells had high levels of Pgp led to the attractive hypothesis that enhanced drug efflux could explain in part the refractoriness to therapy seen in this malignancy (Fojo et aZ., 1987). Thus, in colon cancer, Pgp could be responsible for giving tumor cells a phenotype well-characterized in vifro in cells selected for resistance to the natural products. Most models of colon cancer are complex, and the coexistence of additional mechanisms of resistance preclude precise conclusions regarding the role of Pgp. To overcome this, we sought initially to examine an in vitro model in which resistance to Adriamycin was mediated predominantly by Pgp in order to determine the extent to which Pgp could account

702

LA1 ET AL. TABLE I1 -VARIATION OF ADRIAMYCIN CYTOTOXICITY BASED O N THE DURATION OF DRUG TREATMENT IC,, in ngiml Adriamycin

Cell lines

3 hr

IC*,L

SW620 SW620 Ad20 SW620 AdlOOO LS 180 LS 180 Ad50 LS 180 Ad150 DLD-1 DLD-1 AdlOOO HCT-15

5.8 1,260 24,300 86 410 520 240 160,000 2,300

RR~

DEI

17 3,780 72,900 258 1,230 1,560 720 480,000 6,900

IC5;

4.6

24 hr RR’

(i%J!] 2,400 (520) {%] (650)

-

I day DE’

110 57,600

1c5;

0.5 14 54 5.3 31 41 20.5 420 24

RR’

(#:]

(20.5)

DE?

84 2,352 9,072 890 5,208 6,888 3,444 70,560 4,032

‘The values represent the IC,,s determined by clonogenic assay, in ngiml adriamy&.-’Relative resistance.-’Drug exposure is the concentration of drug times the time of exposure. Units are ng.hr/ml.-This table presents the results of a group of experiments which are representative of 2 to 4 experiments for each cell line.

for resistance in colon cancer. We used a cell line (SW620) established from a previously untreated patient which was very sensitive to Adriamycin and derived resistant cell lines by in vitro exposure to Adriamycin. The data supporting Pgp as the predominant mechanism of resistance in the SW620 sub-lines includes: the correlation between stepwise increases in Adriamycin resistance, mdr-l/Pgp expression, and reduced drug accumulation; and the observation that verapamil can reverse drug resistance in a dose-dependent manner. Taken together, the results suggest that Pgp is primarily responsible for Adriamycin resistance in these selected cell lines. Using similar correlates, additional parental lines and drug-selected sub-lines were evaluated. HCT-15 parental cells and Adriamycin-selected DLD-1 AdlOOO cells are similar to the SW620 Ad sub-lines in that antagonism of Pgp markedly enhances chemosensitivity, implying a predominant role for Pgp in mediating their drug resistance. In contrast, in DLD-1, SW620 and LS 180 parental cells and in the LS 180 Ad sub-lines, Pgp plays a minor role in Adriamycin resistance. DLD-1 cells, despite comparable levels of resistance to HCT-15 had only a 3.4-fold increase in Adriamycin sensitivity after verapamil. This suggests that in DLD-1 cells Pgp cannot fully account for Adriamycin resistance, and co-exists with other mechanisms of resistance. The LS 180 cell line has very low levels of mdr-1/Pgp and no increase in Adriamycin sensitivity after verapamil, suggesting little or no mediation of resistance by Pgp in these cells. In the LS 180 cells and sublines, there is circumstantial evidence to support a role for involvement of the GSH redox cycle in Adriamycin resistance (Lai ef al., 1991). Although some reversal of Adriamycin resistance by verapamil was achieved in all of the SW620 Ad sublines, full reversal by verapamil was achieved only in the lower levels of resistance. While the data suggest that incomplete reversal is due to inadequate concentrations of verapamil, the presence of other mechanisms in the higher levels of resistance is not absolutely excluded. The studies which demonstrate that increasing doses of verapamil are required to fully reverse increasing levels of resistance suggest that in clinical samples, higher levels of mdr-l/Pgp may not be reversed by the blocking agents currently available, since high levels of mdr-1/Pgp expression require levels of Pgp inhibitors not clinically tolerable (Ozols et al., 1987). This observation may have clinical relevance in tumors such as those arising from the adrenal gland. Fortunately, mdr-1 levels in tumors generally approximate to levels found either in unselected lines or in lines selected for low levels of resistance (Fojo et al., 1987b). The results with the unselected HCT-15 cell line, for example, provide evidence that Pgp can mediate resistance in colon cancer, and in some cases, reversal of its contribution to resistance may result in enhanced chemosensitivity. One caveat, however, is that neither 5 nor 10 pg/ml verapamil is achievable clinically. Indeed, early clinical studies demon-

strated that 1 pg/ml verapamil had significant cardiac effects (Ozols et al., 1987). In the sub-lines which demonstrated a predominant role for Pgp in resistance (the SW620 Ad20, SW620 AdlOOO and the DLD-1 AdlOOO cells), the relative resistance to Adriamycin was demonstrated to be schedule-dependent. Lower relative resistance was observed with 7-day drug incubation than with 3-hr treatment. In these cell lines, both the reversibility of vinblastine accumulation by verapamil and the increase in Adriamycin cytotoxicity in the clonogenic assay with verapamil supported a primary role for Pgp in modulating resistance. In the LS 180 sub-lines, relative resistance to Adriamycin was not schedule-dependent. These data suggested that enhancement of chemosensitivity by long exposure could be a characteristic of Pgp-mediated resistance. This schedule dependence is also apparent if one compares drug exposure, which is calculated by multiplying the drug concentration by the time of exposure. For SW620 Ad1000 sublines and DLD-1 AdlOOO cells, 7- to 8-fold less drug was required to achieve an IC,, with a 7-day incubation compared with the 3-hr drug treatment; evidence that with longer treatment times Pgp-mediated resistance was more readily overcome. With this approach, a similar schedule dependence could be demonstrated for HCT-15 cells. That the HCT-15 cells were the only parental cells in which continuous treatment proved more efficacious reflects the importance of Pgp to Adriamycin resistance in this cell line. In the SW620 Ad20 sub-line and in HCT-15 cells, which have lower levels of Pgp expression, the schedule dependence was not as marked as that in the sub-lines with high levels of Pgp. In contrast to the results in the cell lines with Pgp-mediated resistance, in the other 3 parental cell lines (SW620, LS 180 and DLD-1) and the other Adriamycin-selected sub-lines (LS 180 Ad50, LS 180 Ad150) the schedule dependence observed was the opposite: continuous treatment required 3.4- to 4.8-fold greater drug exposure to achieve a comparable cell kill. In this latter group, although Pgp could be detected, other mechanisms of resistance were felt to be more important, as evidenced by the small modulation by verapamil and the high IC,,s in the presence of verapamil. These results should not be interpreted as evidence that the continuous treatment was not able to modulate the contribution of Pgp to resistance in these cell lines. This modulation may have occurred, but was masked by the presence of other mechanisms of resistance. Although most studies of schedule dependence of cytotoxicity have examined the antimetabolites, few reports examining Adriamycin have been published (Yang and Famaan, 1988; Carmichael et al., 1987; Mergenthaler et al., 1987). In both cancer cell lines and normal cells, Adriamycin cytotoxicity has been found to be independent of the schedule of administration. The difference between these previous observations and

ADRIAMYCIN RESISTANCE IN COLON CANCER CELL LINES

the results described in the present study are probably the result of differences in cell types. Although normal cells and other cell lines may not demonstrate schedule dependence, this should not be extrapolated to cells possessing a resistance mechanism, such as Pgp. In summary, the present study demonstrates that in some human colon carcinoma cell lines, Pgp appears to play a significant role in resistance to Adriamycin and this can be overcome by the use of competitive inhibitors of Pgp. However, measurement of mdr-l/Pgp cannot predict the level of drug accumulation, the degree to which resistance is mediated

703

by Pgp, nor the reversibility by verapamil. Increasing doses of verapamil are required to fully reverse increasing levels of resistance, suggesting that in clinical samples, higher levels of mdr-1JPgp cannot be reversed by the blocking agents currently available. Finally, in these models, cells expressing Pgp are relatively more resistant than cells without Pgp when shorter drug exposures are used. Although the search for effective and safe competitors of Pgp continues, these results suggest that administration of drugs by continuous infusion may provide a means on which to begin to build effective regimens to overcome multidrug resistance mediated by Pgp.

REFERENCES AKIYAMA, S-I., FOJO, A., HANOVER, J.A., PASTAN, I. and GOTTESMAN, LAI, G.M., MOSCOW, J.M., ALVAREZ,M.G., FOJO,A.T. and BATES, M.M., Isolation and genetic characterization of human Kl3 cell lines S.E., Contribution of glutathione and glutathione-dependent enzymes in the reversal of Adriamycin resistance in colon carcinoma cell lines. resistant to multiple drugs. Somat. Cell Mol. Genet., 11, 117-126 (1985). Znt. J. Cancer, 49,688-695 (1991). A., STINSON,J.C., MCCOMBS111, W.B., McCoy, C.E., BECK, W.T., MUELLER,T.J. and TANZER,L.R., Altered surface LEIBOVITZ, membrane glycoprotein in Vinca alkaloid-resistant human leukemia MAZUR,K.C. and MABRY,N.D., Classification of human colorectal adenocarcinoma cell lines. Cancer Res., 3 6 , 4 5 6 2 4 6 9 (1976). lymphoblasts. Cancer Res., 39,2070-2076 (1979). L.H., Reduced permeability in CHO cells as BELL,D.R., GERLACH, J.H., KARTNER, N., BUICK,R.N. and LING,V., LING,V. and THOMPSON, Detection of P-glycoprotein in ovarian cancer: a molecular marker a mechanism of resistance to colchicine. J. Cell Physiol., 83, 103-116 associated with multidrug resistance. J. clin. Oncol., 3,311-315 (1985). (1974). B.C., KINSELLA, T.J., HAMILTON, T.C., GROTZBIEDLER,J.L. and RIEHM,H., Cellular resistance to actinomycin D in LOUIE,K.G., BEHRENS, Chinese hamster cells in vitro: cross-resistance, radio-autographic and INGER,K.R., MCKOY, W.M., WINKER,M.A. and OZOLS,R.F., Radiation survival parameters of anti-neoplastic drug-sensitive and -resiscytogenetic studies. Cancer Res., 30,3174-1184 (1970). tant human ovarian cancer cell lines and their modification by CARMICHAEL, J., DEGRAFF,W.G., GAZDAR,A., MINNA,J.D. and buthionine sulfoximine. CancerRes., 45,2110-2115 (1985). MITCHELL,J.B., Evaluation of a tetrazolium-based semi-automated H.G., BRUHL,P., EHNINGER, G. and HEIDEMANN, E., colorimetric assay: assessment of chemosensitivity testing. CancerRes., MERGENTHALER, Comparative in vitro toxicity of mitoxantrone and adriamycin in human 47,936942 (1987). granulocyte-macro hage progenitor cells. Cancer Chemother. PharmaCHEN,C.J., CHIN,J.E., UEDA,K., CLARK,D.P., PASTAN,I., GOTTES- col., 20,8-12 (19877. MAN, M.M. and RONINSON, I., Internal du lication and homology to R.E., KLECKER,R.W., JR., HAMILTON, T.C., bacterial transport proteins in the mdrl (!-glyco rotein) gene from OZOLS,R.F., CUNNION, OSTCHEGA,Y., PARILLO,J.E. and YOUNG,R.C., Verapamil and multidrug-resistant human cells. Cell, 47,381-389 8986). Adriamycin in the treatment of drug-resistant ovarian cancer patients. CORNWELL, M.M., PASTAN, I. and GOTTESMAN, M.M., Certain calcium J. clin. Oncol., 5,641-647 (1987). channel blockers bind specifically to multidrug-resistant human KB RIORDAN,J.R. and LING, V., Purification of P-glycoprotein from carcinoma membrane vesicles and inhibit drug binding to P-glycopro- plasma membrane vesicles of Chinese hamster ovary cell mutants with tein. J. biol. Chem., 262,2166-2170 (1987). reduced colchicine permeability. J. biol. Chem., 254, 12701-12705 COWAN, K.H., BATIST,G., TULPULE, A,, SINHA,B.K. and MYERS,C.E., (1979). Similar biochemical changes associated with multidrug resistance in SAFA,A.R., GLOVER,C.J., MEYERS,M.B., BIEDLER,J.L. and FELSTED, human breast cancer cells and carcinogen-induced resistance to R.L., Vinblastine photoaffinity labelling of a high-molecular-weight xenobiotics in rats. Proc. nat. Acad. Sci. (Wash.), 83,9328-9332 (1986). surface membrane glycoprotein s ecific for multidrug-resistant cells. J. DALTON, W.F., GROGAN, T.M., MELTZER,P.S., SCHEPER, R.J., DURIE, biol. Cheni., 261,6137-6140 ( 1 9 8 6 . B.G.M., TAYLOR,C.W., MILLER,T.P. and SALMON,S.E., Drug THORGEIRSSON, S.S., HUBER,B.E., SORRELL, S., FOJO,A.T., PASTAN, I. resistance in multiple myeloma in non-Hodgkin's lymphoma: detec- and GOTTESMAN, M.M., Expression of the multidrug-resistant gene in tion of P-glycoprotein and potential circumvention by addition of hepatocarcinogenesis and regenerating rat liver. Science, 236, 1120verapamil to chemotherapy. J. clin. Oncol., 4,415-424 (1989). 1122 (1987). FAIRCHILD, C.R., IVY,S.P., RUSHMORE, T., LEE, G., Koo, P., GOLD- TIBBEITS,L.M., CHU,M.Y., HAGER,J.C., DEXTER,D.L. and CALAE. and COWAN, K.H., Carcinogen- BRESI,P., Chemotherapy of cell-line-derived human colon carcinomas SMITH,M.E., MYERS,C.E., FARBER, induced mdr over-expression is associated with xenobiotic resistance in in mice immunosuppressed with antithymocyte serum. Cancer, 40, rat pre-neoplastic liver nodules and hepatocellular carcinomas. Proc. 2651-2659 (1977). nat. Acad. Sci. (Wash.),84,7701-7705 (1987). TOM,B.H., RUTZKY,L.P., JAKSTYS, M.M., OYASU,R., KAYE,C.I. and FOJO,A.T., AKIYAMA,S.I., GOTTESMAN,M.M. and PASTAN,I., KAHAN,B.D., Human colonic adenocarcinoma cells. I. Establishment Reduced drug accumulation in multiply drug-resistant human KB and description of a new line. In Vitro, 12,180-191 (1976). carcinoma cell lines. Cancer Res., 45,3002-3007 (1985). TSURUO,T., LIDA,H., TSUKAGOSHI, S. and SAKURAI, Y.,Potentiation FOJO,A.T., UEDA,K., SLAMON, D.J., POPLACK,D.G., GOTTESMAN, of vincristine and Adriamycin effects in human haemopoietic cell lines M.M. and PASTAN,I., Expression of a multidrug-resistance gene in by calcium antagonists and calmodulin inhibitors. Cancer Res., 43, human tumors and tissues. Proc. nat. Acad. Sci. (Wmh.), 84, 265-269 2267-2272 (1983). (1987). UEDA, K., CLARK,D.P., CHEN,C.J., RONINSON, I.B., GOTTESMAN, KARTNER, N., RIORDAN,J.R. and LING,V., Cell-surface P-glycopro- M.M. and PASTAN,I., The human multidrug resistance (mdr-1) gene: tein is associated with multidrug resistance in mammalian cell lines. cDNA cloning and transcription initiation. J. biol. Chem., 262,505-508 (1987). Science, 221,1285-1288 (1983). KRAMER, R.A., ZAKHER,J. and KIM,G., Role of the glutathione redox YANG.K.P. and FAMAAN.N.A.. Lethal efficacy of doxorubicin on cycle in acquired and de novo multidrug resistance. Science, 241, human medullary thyroid carcinoma cells in vitr;. Anticancer Res., 8, 245-248 (1988). 694-697 (1988).