EXPERIMENTAL

CELL RESEARCH

196, 323-329 (19%)

Laser Scanning and Confocal Microscopy of Daunorubicin, Doxorubicin, and Rhodamine 123 in Multidrug-Resistant Cells JAMES L. WEAVER,* P. *Diuision

PINE,* ADORJAN ASZALOS,*"PATRICIA V. SCHOENLEIN,-~STEPHENJ. CURRIER,? RAJIPADMANABHAN,~ANDMICHAEL M. GOTTESMAN~

SCOTT

of Research and Testing, CDER, HFD-471, Food and Drug Administration, 200 C Street Southwest, Washington TLaboratory of Cell Biology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland

The multidrug-resistant gene (MDRl) encodes an energy-dependent drug efflux pump (P-glycoprotein) for many anti-cancer drugs. We have studied the intracellular distribution of rhodamine 123 (R123), daunorubitin (DN), and doxorubicin (DOX) in cells expressing a human MDRl gene. The distribution of these fluorescent drugs was measured by laser scanning microscopy and confocal microscopy. We devised a new method for analysis of fluorescence line scan data to determine the intracellular distribution of fluorescent probes. This method and confocal microscopy showed that R123, DN, and DOX are localized to both plasma membrane and intracellular compartments in multidrug-resistant cells. When the cells are treated with verapamil, an inhibitor of the multidrug transporter, the amount of DOX, DN, and R123 associated with the cell rises. After inhibition, the relative distribution of DOX and DN between the cell surface and intracellular structures does not change dramatically. However, R123 tends to relocalize to intracellular sites from predominantly plasma membrane sites, indicating that this dye behaves differently than the anti-cancer drugs. These results show the subcellular distributions of R123, DN, and DOX in plasma membrane, cytoplasm, and intracellular membrane systems, but do not allow definitive distinctions among existing models of how P-glycoprotein affects the distribution of drugs. o 1991 Academic

D.C. 20204; and

have been proposed. These include the idea that the pump removes drugs directly from the plasma membrane [4] or relocalizes drugs intracellularly by pumping them into subcellular organelles, thus sequestering them from their cytotoxic target sites [5]. Several fluorescent model drugs have been used to study the activity of the multidrug transporter including daunorubicin [6], doxorubicin [4], and rhodamine 123 [7], and other fluorescent dyes [8]. The former two drugs bind to DNA in the nucleus while the later localizes to the mitochondria. Using laser scanning and confocal microscopy we have attempted to determine the distribution of these compounds in multidrug-resistant cells in the presence and absence of verapamil, an inhibitor of the multidrug transporter. The results confirm the ability of the multidrug transporter to remove anticancer drugs from both intracellular and membrane compartments. They do not show a significant redistribution between these compartments in response to treatment with verapamil. MATERIALS

AND

METHODS

The multidrug transporter is a 170,000-Da membrane glycoprotein, also known as P-glycoprotein [l-3], which uses the energy of ATP to pump anti-cancer drugs out of multidrug-resistant cells. Several hypotheses concerning the mechanism of action of this transporter

KB-3-1 cells were derived from a single Cell lines and cell growth. clone of the human KB adenocarcinoma cell line (American Type Culture Collection) and are drug-sensitive. KB-Vl is a multidrug-resistant derivative of KB-3-1, obtained by stepwise selection in increasing concentrations of vinblastine [9]. NIH 3T3 is a drug-sensitive mouse fibroblast cell line. The multidrug-resistant transfectants pHaMDR-Cl [lo] and pHaMDRGA-C2-1 were obtained by transfecting NIH 3T3 with the following plasmids: pHaMDR1 which contains an MDRl cDNA isolated from the cholchicine-selected cell line KB-C2.5 [ll] and pHaMDRGA which contains the wild-type MDRl cDNA [12]. The plasmid pHaMDR1 contains base pair mutations which result in a glycine + valine substitution at amino acid position 185 in the MDRl pump. This mutation confers preferential colchitine resistance [12,13], while pHaMDRGA has the wild-type glycine at this position. All cells were grown at 37°C in a humidified atmosphere containing 5% CO,. Cells were passaged as monolayer cultures in 75cm’ flasks in DMEM’ (GIBCO, Inc, Grand Island, NY) containing 10% fetal

1 To whom dressed.

2 Abbreviations used: DN, daunorubicin; DOX, doxorubicin; R123, rhodamine 123; DMEM, Dulbecco’s modified Eagles medium; Fluo-3-

Press, Inc.

INTRODUCTION

correspondence

and reprint

requests

should

he ad-

323 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.

WEAVER

0

25

50

75

100

Pixel Number

125

150

175

200

ET AL.

I

-601 0

50

25

(1 pixel = 0.1 pm)

Percent

75

100

Translocation

FIG. 1. Calculated line scan profiles show large changes between all cell membrane and all intracellular distributions. (A) Calculated line scan profiles for a lo-pm cell with O-100% of the fluor on the cell membrane in steps of 25%; data smoothed 3 X 10 points. Long dashes (- - -), 0% membrane Auor; dash dot dot (-. * ), 25%; dots (. * *), 50%; dash dot (-e), 75%, solid line (-), 100% membrane fluor. (B) Relationship between the slope of a linear regression line for data between pixel 68 and 85 (vertical dotted lines) from A and the percentage fluor associated with the cell membrane. The line is the linear regression line for the data points; the correlation coefficient is 0.9974.

bovine serum (human cell lines) or 10% calf serum (mouse cell lines), streptomycin (50 pglml), penicillin (50 units/ml), glutamine (5 n&f), and the appropriate concentration of drug. Sixteen hours prior to analysis, l-2 X 10’ cells were seeded on two well chamber slides (Nunc, Inc., Naperville, IL) containing 2.0 ml DMEM; drugs were not present during this 16-h incubation. Reagents. Fluo-3-AM was obtained from Molecular Probes, Inc. (Eugene, OR). NBD-PC was purchased from Avanti Polar Lipids (Pelham, AL). DOX, DN, R123, and verapamil were purchased from Sigma (St. Louis, MO). Line scan profiles. A line scan collected by a laser scanning microscope contains information about the distribution of fluorescent molecules within that cell. Figure 1A shows the line scan profiles generated by a computer model [14]. The profiles are for 0% membrane bound fluor, 100% cytoplasmic fluor to 100% membrane bound fluor, 0% cytoplasmic fluor in steps of 25%. The area between the vertical dotted lines was chosen as showing the largest changes. A linear regression line can be calculated for data in this region, Fig. 1B shows that there is a linear relationship between the slope of this linear regression line and the distribution of fluor. This method is discussed in detail in [14]. The fluorescence profiles were collected by laser scanning microscopy using an ACAS 570 system (Meridian Instruments, Okemos, MI). KB-Vl cells grown in suspension were incubated with DOX (2 pg/ml), DN (2 pglml), or R123 (8 &f) for 2 h at 37°C + verapamil(6 j&f). Cells were washed and resuspended in PBS and introduced into Cunningham chambers [15] for scanning. Cells were loaded with Fluo-3 by incubation with the cell permeant form Fluo-3-AM as previously described [16]. Cells were labeled with NBD-PC (100 j&f) for 10 min at room temperature. The excitation wavelength was 488 nm, with laser power and neutral density filters set to keep photobleaching at ~10%. Line scan fluorescence profile data were collected (2050 profiles/cell) using the kinetics line scan routine of the ACAS software, version 2.0d, at steps of 0.25 pm. Profiles were collected for 20 cells per treatment condition. Using software written for this project (J. L. Weaver, unpublished), the 20-50 profiles for each cell were averaged. The data for all cells for each condition were averaged to produce a composite fluorescence line scan profile. As controls, we AM, Fluo-3-acetomethoxyester; * 1,3-diazol-4-yl)amino)-phosphatidyl

NBD-PC, 6-(N-(7-nitrobenz-2-oxacholine.

collected profiles for cells stained with the membrane-limited fluorescent probe NBD-PC and the cytoplasmic Ca*’ indicating dye Fluo-3. Using these control profiles, the specific region of the profile to be used in analyzing data from KB-Vl cells treated with DOX or R123 was determined. Data in this region were used to calculate a linear regression line, the slope of this line was determined by the intracellular distribution of the fluors [14]. Confocal microscopy. This technique provides an optical thin section through the sample. The instrument allows collection of both phase contrast and fluorescent confocal images simultaneously. For this study horizontal optical sections of tl pm thickness from the middle of the cells were collected on a Bio-Rad Microsciences (Cambridge, MA) Model MRC-500 Confocal imaging System. It was configured with a Zeiss Axiovert 35 microscope using a 63X objective. Cells grown on coverslips were loaded with DOX, DN, or R123 + verapamil as described above. RESULTS

Line Scan Profiles Profiles were collected from KB-Vl cells loaded with the control plasma membrane probe NBD-PC or the cytoplasmic Ca2+ probe Fluo-3. Figure 2A shows clear and obvious differences between the profiles obtained from the two probes. These differences are consistent with those expected from the plasma membrane location of NBD-PC and the cytoplasmic location of Fluo-3. The profiles of cells loaded with DOX or R123 f verapamil are shown in Fig. 2B. These profiles show that, as expected [6], verapamil increased the total fluorescence of these cells. This has been confirmed by flow cytometry (not shown). Comparison of the DOX and R123 profiles with those in Fig. 2A shows that fluorescence from both DOX and R123 appears intracellularly. The calculated slopes for profiles collected under these conditions are shown in Table 1. They indicate that the relative distribution of DOX between the plasma membrane and

SUBCELLULAR

LOCATION

$0 Pixel # (1 pixel = 0.25 pm)

OF DRUGS

IN MDR

325

CELLS

0 0

25

50

75

100

125

Pixel # (1 pixel = 0.25 pm)

FIG. 2. Observed fluorescence line scan profiles are similar to calculated profiles and show that DOX and R123 have different intracellular distributions and responses to verapamil. (A) Observed composite line scan profiles for KB-Vl cells labeled with the membrane probe NBD-phosphatidyl choline, dashed line (--), or the cytoplasmic probe Fluo-3, solid line (-). (B) Composite profiles of KB-Vl cells treated with DOX (2 fig/ml) or R123 (8 like) kverapamil(6 pcM). Dotted line (. * +), DOX, solid line (--), DOX with verapamil; dashed line (- -), R123; dashed line with dot (- . -), R123 and verapamil.

the interior of the cell is not affected by verapamil treatment. The difference between these two values (32 vs 28) could be due to cell-cell variability. The calculations show that R123 also has a significant cytoplasmic component, although less than DOX, and a significant increase in the apparent intracellular content of R123 is observed following verapamil treatment. Confocal Microscopy

The confocal images of KB-Vl cells treated as above confirm the results of the calculations from the line scan profiles. DOX and DN are found predominantly in intracellular compartments at significant levels (Figs. 3A and 3C), and fluorescence can be seen in both the plasma and nuclear membrane areas. The addition of verapamil did not change the intracellular distribution of either drug (Figs. 3B and 3D). The absolute brightness of the verapamil-treated cells increased but the increase was not large. This increase was confirmed by measuring the fluorescence intensity of lo4 cells by flow cytometry (not shown). Since DOX and DN bind to DNA, where their fluorescence is quenched, these data do not allow us to calculate absolute amounts of DOX and DN within cells. KB-Vl cells labeled with R123 show weak but discernible cell membrane staining and a very low degree of cytoplasmic staining (Fig. 3E). Verapamil treatment caused the appearance of significant intracellular staining concentrated in the perinuclear region (Fig. 3F). This perinuclear localization is presumably staining of mitochondria since R123 is used as a mitochondrial stain [17]. The drug-sensitive parental cell line of the KB-Vl cells, KB-3-1, has an intracellular distribution of DOX or DN similar to that seen in KB-Vl cells (Fig. 4).

Treatment of these cells with verapamil causes no discernible change in either the distribution or absolute brightness of these cells (compare Fig. 4A with 4C and 4B with 4D). Thus, the changes seen in the KB-Vl cells after verapamil treatment appear to be due to inhibition of P-glycoprotein function. To confirm that these changes are due to the activity of the MDRl gene and not due to other physiological activities in the multidrug-resistant KB cells, we repeated these experiments in two different NIH 3T3 cell lines transfected with wild-type and mutant human MDRl genes. Figure 5 shows that in NIH 3T3pHaMDRGA cells expressing the wild-type P-glycoprotein, a fluorescence pattern similar to that seen in KBVl cells is observed. Cells treated with DOX (Fig. 5A) show a similar distribution of fluorescence compared to that of cells treated with DOX + verapamil (Fig. 5B) but the latter show an increase in total fluorescence. This same pattern of intracellular fluorescence is also seen in the NIH 3T3-pHaMDR1 cells, which express P-glycoprotein with a gly,, + val mutation (not shown). The parental NIH 3T3 cells treated with DOX show no change in intracellular fluorescence in response to verapamil treatment (Fig. 5C vs 5D). This is the same behavior that was observed with the parental KB-3-1 cell line. DISCUSSION

These studies confirm that the primary effect of the multidrug transporter is to reduce the level of the fluorescent, anti-cancer drugs DOX, DN, and the mitochondrial dye R123 in resistant cells. The high-resolution techniques, scanning laser, and confocal microscopy, demonstrate association of these drugs with both plasma membrane and intracellular sites in cells ex-

WEAVER

ET AL.

FIG. 3. Confocal microscopic images of KB-Vl cells show different intracellular distributions of DOX and R123. The left photograph in each set is the pseudo-phase contrast confocal image, the right photograph is the confocal fluorescence image. (A) Cells treated with DOX. The scale bar indicates 10 pm. (B) Cells treated with DOX + verapamil. (C) Cells treated with DN. (D) Cells treated with DN and verapamil. (E) Cells labeled with R123. (F) Cells labeled with R123 and treated with verapamil.

pressing the multidrug transporter and in cells in which the transporter has been inhibited or is not present. The relative amounts of drugs in these two compartments cannot be accurately determined with these techniques because of differences in the quantum yield among the various cellular environments which result from binding to target sites. Quantitative extraction methods would probably not give additional information due to

uncertainties about redistribution during the extraction procedures. However, these results underscore the fact that agents which are substrates for the multidrug transporter partition between aqueous and lipid phases. They also show that the fluorescent compounds tend to localize in several cellular location. These are near or within the plasma membrane, close to the nuclear membrane, and within intracellular vesicles, probably lyso-

SUBCELLULAR

TABLE

LOCATION

1

The Effect of Verapamil on Calculated Line Scan Slopes in KB-Vl Cells Label NBD-PC Fluo-3 Doxorubicin Doxorubicin Rhodamine Rhodamine

Drug Verapamil 123 123

Verapamil

Calculated

slope

-15 49 32 28 23 47

somes. These locations cannot be accurately determined with conventional fluorescence microscopy. We have shown that R123 has a somewhat different intracellular distribution than DOX or DN in cells expressing the human MDRl gene. In response to inhibition of the multidrug transporter with verapamil, the intracellular distribution of R123 shows a large increase in intracellular content and a high concentration in the perinuclear region. In contrast, while the intracellular

OF DRUGS

IN MDR

CELLS

327

concentration of DOX and DN increases following verapamil treatment, no change is observed in the intracellular distribution of these drugs, presumably because fluorescence is quenched in the nucleus after binding to DNA. This behavior is seen in both KB-Vl and 3T3 cell lines transfected with human MDRl gene. One possible explanation for the internal localization of DOX and DN in multidrug-resistant cells is that these drugs have been endocytosed from the cell surface and are trapped within acidic endocytic vesicles. These results also show that R123 and perhaps other mitochondrial stains such as berberine [8] are inappropriate models for studies of MDR pump function. Despite the high resolution of these two techniques, we cannot use these studies to support a specific model of P-glycoprotein action. Plasma membrane localization could represent drug near the membrane, or in either the inner or outer leaflet of the membrane (or both). Therefore, in drug-resistant cell lines, the pattern of localization of DOX, DN, and R123 near the plasma membrane could represent binding to the P-glycoprotein, outer membrane-associated proteins, or asso-

FIG. 4. Confocal microscopic images of KB-3-1 cells. (A) Cells treated with DOX. The scale bar indicates DOX and verapamil. (C) Cells treated with DN. (D) Cells treated with DN and verapamil.

10 pm. (B) Cells treated with

328

WEAVER

ET AL.

FIG. 5. Confocal microscopic images of NIH 3T3 cells with or without transfection with the MDRl gene. (A) NIH 3T3-pHaMDRGA cells (transfected) treated with DOX. The scale bar indicates 10 pm. (B) NIH 3T3-pHaMDRGA cells treated with DOX and verapamil. (C) NIH 3T3 cells (parental) treated with DOX. (D) NIH 3T3 cells treated with DOX and verapamil.

ciation of these hydrophobic drugs with either or both leaflets of the membrane. Similarly, the intracellular localization could be unbound drug free in the cytoplasm, or drug associated with target sites such as mitochondria for R123, or acidic compartments such as lysosomes for DN and DOX [6]. Evaluation of the nuclear levels of DN or DOX is not feasible using fluorescent techniques since the fluorescence of these drugs is quenched after intercalation into the DNA helix. Due to these limitations, these methods lack the resolution to determine whether the drug molecules which are pumped out through the plasma membrane are taken from the cytosol or from within the membrane itself. However, the higher resolution afforded by laser scanning and confocal microscopy have yielded new information about the subcellular distribution of substrates for the multidrug transporter. Further, these studies clearly exclude simple hypotheses such as that the mul-

tidrug transporter excludes all DOX, DN, and R123 from all intracellular compartments. We thank Dan Towson of Meridian Instruments for assistance with the kinetics line scan file format. We are grateful to the Department of Anatomy at Uniformed Services University of the Health Sciences for use of the confocal microscope system and to Dr. Mark Adelman, L. S. M. coordinator, for his assistance with confocal microscopy.

REFERENCES 1. 2. 3.

Gottesman, M. M., and Pastan, I. (1988) J. Bill. Chem. 263, 12,163-12,166. Endicott, J. A., and Ling, V. (1989) Anna Rev. Biochem. 58, 137-171. Kane, S. E., Pastan, I., and Gottesman, M. M. (1990) J. Bioenerg. Biomembr. 22, 593-618.

SUBCELLULAR

LOCATION

OF DRUGS

4.

Raviv, Y., Pollard, H. B., Brugsemann, Gottesman, M. M. (1990) J. Biol. Chm.

E. P., Pastan, I., and 265, 3975-3980.

11.

5.

Schuurhuis, G. J., Broxterman, H. J., Cervantes, A., van Heijninggen, T. H. M., delange, J. H. M., Baak, J. P. A., Pinedo, H. M., and Lankelma, J. (1989) J. Natl. Cancer Inst. 81,18871892.

12.

6.

Willingham, M. C., Cornwell, M. M., Cardarelli, C. O., Gottesman, M. M., and Pastan, I. (1986) Cancer Res. 46, 5941-5946.

7.

Kessel, D. (1989) Cancer Commun.

1, 145-149.

8.

Neyfakh,

9.

Shen, D., Cardarelli, C., Hwang, J., Cornwell, M., Richert, N., Ishii, S., Pastan, I., and Gottesman, M. (1986) J. Biol. Chm. 261, 7762-7770.

10.

Currier, S. J., Ueda, K., Willingham, M. C., Pastan, I., and Gottesman, M. (1989) J. Biol. Chem. 264, 14,376-14,381.

Received April

13. 14. 15.

A. A. (1986) Exp. Cell. Res. 184, 168-176.

17, 1991

16. 17.

IN MDR

CELLS

329

Ueda, K., Cardarelli, C., Gottesman, M. M., and Pastan, I. (1987) Proc. Natl. Ad. Sci. USA 84,3004-3008. Kioka, N., Tsubota, J., Kakehi, Y., Komano, T., Gottesman, M., Pastan, I., and Ueda, K. (1989) Biochem. Biophys. Res. Commun. 162,224-231. Choi, K., Chen, C., Kreigler, M., and Roninson, I. (1988) Cell 53, 519-529. Weaver, J. L., Submitted for publication. Smith, P. F., Luque, E. H., and Neill, J. D. (1986) in Methods in Enzymology (Fleischer, S., and Fleischer, B., Eds.), Vol. 124, pp. 443-465, Academic Press, San Diego, CA. Weaver, J. L., Gergely, P., Pine, P. S., Patzer, E., and Aszalos, A. (1990) AIDS Res. Human Retrouir. 6, 1125-1130. Haugland, R. (1989) in Handbook of Fluorescent Probes and Research Chemicals, p. 125, Molecular Probes, Inc., Eugene, OR.