A Role for Ectophosphatase in Xenobiotic Resistance

Abstract

Xenobiotic resistance in animals, plants, yeast, and bacteria is known to involve ATP binding cassette transporters that efflux invading toxins. We present data from yeast and a higher plant indicating that xenobiotic resistance also involves extracellular ATP degradation. Transgenic upregulation of ecto-ATPase alone confers resistance to organisms that have had no previous exposure to toxins. Similarly, cells that are deficient in extracellular ATPase activity are more sensitive to xenobiotics. On the basis of these and other supporting data, we hypothesize that the hydrolysis of extracellular ATP by phosphatases and ATPases may be necessary for the resistance conferred by P-glycoprotein.

INTRODUCTION

Multidrug resistance (MDR) is a biological phenomenon common to many organisms in which cells have become resistant to various toxins. Past studies of MDR have generally emphasized the role of the chemical transporters involved in the efflux of toxins. One set of transporters upregulated in the MDR response are P-glycoproteins, and the genes encoding P-glycoproteins are capable of conferring drug resistance to drug-sensitive cell lines (Kolaczkowski et al., 1996). Competition assays involving reconstituted P-glycoprotein–containing vesicles have shown that toxins are preferentially removed from the cell by this class of transporter (Sharom et al., 1993). Members of the ATP binding cassette (ABC) superfamily, P-glycoproteins are encoded by the MULTIDRUG RESISTANCE 1 (MDR1) genes and are now known to exist in numerous organisms. P-glycoproteins transport various heterocyclic, aromatic, nitrogen-containing com-pounds (Gros and Hanna, 1996), and they do so in an ATP-dependent fashion (Sharom et al., 1993).

Since the MDR1 gene was first cloned in 1985, several models have been proposed to explain the mode of action of P-glycoprotein. Most of these mechanisms suggest that ATPase activity is essential for moving drugs across the plasma membrane. In the most widely proposed model, MDR1 acts as a drug pump, utilizing the free energy of ATP hydrolysis to actively transport a toxic substrate outside the cell (Gros and Hanna, 1996). This same mode of transport has been suggested for a few different types of MDR proteins (Chang et al., 1998; Decottignies et al., 1998). Other models suggest, however, that drug resistance involves more than the MDR1 protein acting as a pump. Crystallographic evidence suggests that eukaryotic P-glycoproteins may be a chimeric form of pump channel (Welsh et al., 1998) that uses ATP hydrolysis to gate the drug conduit.

Some evidence indicates that P-glycoprotein and possibly other ABC proteins can mediate the release of ATP from cells (Abraham et al., 1993; Roman et al., 1997) and that ATP release from yeast cells (Boyum and Guidotti, 1997a) possibly occurs through indirect action of an ABC transporter (Boyum and Guidotti, 1997b). We present data for yeast and plants to indicate that a plant MDR1 homolog may transport ATP and that extracellular ATP concentrations are kept at a low steady state level by ecto-ATPases. Destroying cell surface ATPase activity, through either chemical inhibition or genetic deletion, results in the loss of xenobiotic resistance both in yeast and in plants. Conversely, upregulating extracellular ATPase activity confers the MDR state. Our results indicate that ecto-ATPases are a second component of xenobiotic resistance and thus present a novel class of targets for understanding and manipulating resistance.

RESULTS

A Plant P-Glycoprotein Gene Expressed in Yeast Promotes ATP Release into the Culture Medium

To test previous claims that MDR1 is itself involved in release of ATP from cells, we used an MDR1 homolog cloned from Arabidopsis and known as atPGP1 (Dudler and Hertig, 1992). We chose this gene because the protein it encodes resembles mammalian P-glycoproteins and has been successfully manipulated in previous genetic studies involving Arabidopsis (Sidler et al., 1998).

When we introduced the Arabidopsis MDR1 cDNA into yeast by transformation, the ATP concentration in the extracellular fluid produced by the MDR1-transformed INVSC1 strain was greater than in the untransformed yeast, as seen in Figure 1A. Although untransformed yeast cells have been shown to release ATP into the extracellular fluid (Boyum and Guidotti, 1997a), presumably through protein-mediated translocation (Boyum and Guidotti, 1997b), our P-glycoprotein transformants had twice the steady state extracellular ATP concentration of wild-type cells, in both the presence and absence of glucose. Previous evidence has indicated that ATP translocation is mediated by ABC proteins in some cell types, such as the ocular epithelium (Mitchell et al., 1998), whereas other data suggest that P-glycoprotein functions as an ATP efflux channel in mammalian systems (Abraham et al., 1993; Roman et al., 1997). Our data suggest that the plant homolog functions in ATP efflux in nonmammalian systems as well.

Figure 1.

Luciferase Assays of Extracellular ATP and Adenosine Pulse–Chase in Experiments with Yeast.

(A) The steady state concentration of ATP in the extracellular fluid was doubled in wild-type yeast overexpressing the Arabidopsis P-glycoprotein, denoted MDR1. Data shown are the average of four experiments in which samples were taken every hour for 5 hr and averaged (±sem). Data are reported as an ATP concentration calculated from a relative light units–based standard curve. xATP, ATP found in the extracellular matrix; + or – glucose, the presence or absence of glucose in the assay solution, respectively; Mock, the INVSC transformed with vector alone.

(B) The yeast ectophosphatase mutant YMR4 accumulated ATP in the extracellular fluid. At bottom, the same yeast strains placed in acid-buffered NBT/BCIP show differential extracellular acid phosphatase activity. Activity is indicated by the development of a dark color along the side of the microcentrifuge tube in which the cells are pelleted. Wt, wild type.

(C) The yeast phosphatase mutant YMR4 accumulated six times the amount of extracellular ATP when the plant P-glycoprotein, denoted MDR1, was overexpressed. pvt101 is a mock transformant. Averages (±sem) of four experiments are shown.

(D) Early differential efflux of ATP in cells overexpressing MDR1. The graph plots the ratio of labeled extracellular adenosine to labeled intracellular adenosine during a pulse–chase in wild-type (INVSC) or mutant (YMR4) yeast cells transformed with either Arabidopsis MDR1 or pvt101 (vector only). Cells were labeled with ³H-adenosine for 20 min before the chase was started. Experiments were performed three times with similar results.

Despite the fact that the yeast cells transformed with MDR1 and the plasmid-transformed wild type both efflux ATP to some degree, neither strain seems to accumulate increased ATP concentrations in the medium over time. It has been suggested that the concentration of extracellular ATP at steady state is kept very low through the action of ecto-ATPases and ectophosphatases (Boyum and Guidotti, 1997a). To test whether the very low extracellular ATP at steady state in yeast is the result of ATP hydrolysis outside the cell, we examined cell culture ATP accumulation in a yeast strain that is deficient in extracellular phosphatase activity. The yeast mutant YMR4 fails to produce the two secreted acid phosphatases encoded by the PHO3 and PHO5 genes (Vogel and Hinnen, 1990) and has almost no ectophosphatase activity at pH 5.0, as evidenced by the failure of nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) to precipitate on the surface of YMR4 cells (Figure 1B, bottom). If phosphatase activity regulates the concentration of extracellular ATP, then a yeast deficient in two extracellular acid phosphatases could be expected to accumulate more extracellular ATP than would the wild type.

Figure 1B (top) shows that the steady state extracellular ATP content for the INVSC1 cell line is <10 pmol per 3 × 10⁶ cells, whereas that for the YMR4 cells is as much as 10 times that amount. Although INVSC differs from YMR4 genotypically, we have used INVSC as a background for comparison with YMR4 because the amount of extracellular ATP seen in the INVSC cells, ∼3 × 10⁻¹⁸ mol per cell, is nearly the same as that previously measured for the parent cell line used to make the YMR4 null, 2.7 × 10⁻¹⁸ mol per cell (Boyum and Guidotti, 1997a). When the YMR4 strain is transformed with MDR1, the cell line accumulates extracellular ATP more quickly by an order of magnitude than does INVSC (Figure 1C). The YMR4/MDR1 transformants, releasing ATP at a greater rate through their P-glycoprotein, also fail to hydrolyze that ATP; with ATP turnover blocked, their extracellular concentrations of ATP rise.

To ensure that the measured extracellular ATP was not the result of cell death, we conducted viability assays of all cells used, and all cell cultures were found to contain at least 95% viable cells. Cell death itself is unlikely to have been the major source of ATP because variances in viability did not always correlate with extracellular ATP abundance. This observation seems counterintuitive because a simple calculation reveals that only 180 pmol of ATP could be generated per lysed cell. The 600 pmol of transport measured for the cells without extracellular phosphatase activity expressing the plant MDR1 is actually more ATP than would be available by lysis alone. Thus, implied in this observation is the potential to have more ATP outside of the cell than inside. That idea is easily accommodated if the cells are continuously generating ATP. A series of pulse–chase experiments seemed to validate this observation.

The previous experiments allowed us to measure external ATP over a period of hours; to learn about the flow of ATP through the plasma membrane over a much shorter period, we performed a series of pulse–chase experiments. Yeast cells were grown in a medium containing tritiated adenosine, and ATP efflux was measured as a function of the accumulation of labeled ATP in the cell supernatant over a 40-min chase period. When these pulse–chase experiments were performed in YMR4 cells transformed with MDR1, more than half of the labeled adenylate (presumably ATP) was effluxed in the first 30 min (Figure 1D). At the acidic pH of the media, the acid phosphatase mutants would have minimal ectophosphatase activity and would not be able to participate in the ATP dephosphorylation steps necessary for internalizing the purine (Che et al., 1992). The phosphatase mutants therefore would accumulate extracellular ATP—and hence label.

In the first 40 min of the efflux assay, more labeled adenosine is outside of the cell than inside. Given that the YMR4 cells are not energy or nutrient limited when growing in complete supplemented medium plus 2% glucose, and given that the extracellular ATP measured represents an hourly accumulation, perhaps it is possible to have more extracellular ATP than the amount that would be instantaneously available when lysis is complete. Similar calculations using the ATP efflux data in YMR4 from Boyum and Guidotti (1997a) yielded 3.4 × 10^–18 mol of ATP per cell, whereas we found 3.3 × 10^–17 mol of ATP per cell in YMR4/pvt101. The difference in our measurements is probably related to the difference in procedures used: their freon extraction followed by HPLC versus our luminometry of raw media. Irrespective of the differences in measurement, our data corroborate the large amounts of ATP in the extracellular matrix first reported by Boyum and Guidotti. In contrast to YMR4, wild-type lines are able to hydrolyze the ATP and internalize the adenosine; their extracellular ATP concentrations were therefore correspondingly lower in our pulse–chase experiments. Indeed, after chasing in media for 3 hr, we detected very little radioactive adenosine in the extracellular fluid of wild-type cells. The adenosine had apparently been recouped by the yeast because nearly all the label was detected in the cell pellets and had presumably been incorporated into nucleic acids. These results suggest that a plant MDR1 can efflux ATP across a yeast plasma membrane and that ectophosphatases dissipate the accumulation of ATP in the extracellular fluid.

ATP Efflux and Ectophosphatase Activity Correlate Positively with the MDR Phenotype

If ATP efflux is essential to the xenobiotic-extruding activities of cells, then we would expect a strong correlation between the ability of a cell to efflux ATP and its resistance to xenobiotics. Furthermore, we would expect that cells unable to regulate extracellular ATP accumulation might likewise show altered xenobiotic tolerance. For that reason, we would expect to see some correlation between the resistance phenotype and ectophosphatase activity.

To determine whether a cell's ability to efflux, and subsequently degrade, ATP is functionally related to xenobiotic resistance, we tested the various yeast cell lines for their resistance to cytotoxic compounds. Cycloheximide and nigericin were the toxic agents chosen, both being substrates of a functional MDR1 homolog in yeast (Kolaczkowski et al., 1996; Decottignies and Goffeau, 1997). Cycloheximide is a potent translation poison; nigericin is a membrane-targeted protonophore. Replication-targeting toxins were not used, which allowed us to obviate the difficulties encountered with mutagenic agents, such as topoisomerase poisons, that might increase mutation frequencies enough to make quantification of engineered MDR difficult. The results of experiments with these two toxins revealed a strong correlation between resistance and ATP efflux and subsequent ATP hydrolysis.

The YMR4 mutant that was mock-transformed without the plant MDR1 had the lowest rates of ATP efflux and extracellular ATP hydrolysis; it was also the most sensitive to cycloheximide and nigericin. As shown in Figure 2, the ectophosphatase mutant took nearly 5 days of continuous culture in cycloheximide to develop resistance, compared with 2 to 3 days for the wild type. When the YMR4 cell line was transformed with MDR1, however, its resistance to the toxins increased, as did its measured rate of ATP efflux (Figures 2A and 1C). The MDR1-expressing YMR4 cell line was resistant whether the toxins were administered separately or together (data not shown). Indeed, its growth rate was most comparable to that of the wild-type yeast, for both reached stationary phase within 2 or 3 days of growth. When the wild-type cell line was transformed with MDR1, its ATP efflux rate increased (Figure 1A) and so did its resistance to toxins (Figure 2). The wild-type MDR1 transformant, with its ectophosphatase activity intact and its ATP efflux rate augmented, was the line most resistant to cycloheximide and nigericin. Thus, the cell lines were resistant in the same order as their increasing MDR1 abundance and ectophosphatase activity.

Figure 2.

Toxin Resistance in Yeast.

(A) Resistance to cycloheximide and nigericin in wild-type (INVSC) and in ectophosphatase-deficient yeast strains. At left, strains grown in the absence of any added toxin; at center, strains grown in the presence of nigericin; and at right, strains grown in the presence of cycloheximide. Clockwise, starting at the top: wild type/MDR1, YMR4/vector, YMR4/MDR1, and wild type/vector. Experiments were performed at least five times with the same result.

(B) Turbidity measurements of the growth in nigericin-containing medium of the yeast cell lines used in (A). Experiments were performed at least five times with similar results.

(C) Turbidity measurements of the growth in cycloheximide-containing medium of the yeast phosphatase mutant expressing the plant MDR1 as well as of the same strain transformed with the vector alone. Experiments were performed at least five times with similar results.

Because resistance seems to correlate with a cell's ability to efflux ATP and degrade it extracellularly, we wondered whether cells with increased MDR1 and normal ectophosphatase function would become sensitive if the toxin were coadministered with exogenous ATP. In other words, could the addition of extracellular ATP to the medium affect the ability of a cell to extrude toxins? Likewise, could the addition of inhibitors of ecto-ATPase activity result in a diminution of resistance?

Wild-type cells transformed with MDR1 showed an increased sensitivity to toxin when extracellular ATP was increased. In fact, all cell lines showed decreased resistance to cycloheximide in the presence of excess extracellular ATP (Table 1). No cell line was able to grow in the presence of 0.1 M ATP and cycloheximide (OD₆₆₀ of 0.00 after 4 days of continuous culture). However, when grown in 0.1 M ATP alone, all cell lines were able to reach stationary growth within 3 days, indicating that ATP alone, even at very high concentrations, is not toxic to the yeast. This concentration is at least 25 times greater than the intracellular ATP concentration. At one order of magnitude less ATP, 12 mM, the MDR1-transformed YMR4 cell line experienced 91% growth inhibition in the presence of cycloheximide (Table 1), whereas the same concentration of ATP and cycloheximide resulted in only 58% growth inhibition in the INVSC strain. A priori, it is difficult to estimate what would be a meaningful extracellular ATP concentration in the yeast periplasm, but the amount of extracellular ATP required to abrogate resistance conferred by MDR1 appears to be strictly dependent on the presence of extracellular phosphatases; the less ectophosphatase activity a cell line has, the more susceptible its resistance to extracellular ATP is. Nonetheless, the concentration required to reverse resistance exceeds the expected intracellular ATP concentration.

Table 1.

Cycloheximide Sensitivity Conferred by Phosphatase Inhibition or Increased Extracellular ATP^a

	Percentage of Growth Inhibition in Cycloheximide
Cell Line	+125 μM Vanadate	+12 mM ATP
YMR4/pvt101	60	57
YMR4/MDR1	85	91
INVSC/pvt101	6	65
INVSC/MDR1	2	58

^aYMR4 and INVSC cells were transformed with the pvt101 vector containing the Arabidopsis MDR1 or with the vector alone. Cell growth was measured at 96 hr in duplicate by optical density at 660 nm and expressed as a percentage of the difference in growth of the same cells in cycloheximide alone, where 100% is complete inhibition.

When cycloheximide was added in the presence of vanadate, an alkaline phosphatase inhibitor, cell lines became sensitized to the toxin (Table 1). Sensitivities corresponded to the ectophosphatase activity of each cell line: both YMR4 cell lines experienced marked reductions in growth, but the wild-type (INVSC) lines were practically unaffected. The high inhibition of growth in the acid phosphatase mutant is most probably the result of inhibition of the other ectophosphatases. The concentration of vanadate used in these studies is unlikely to have reversed MDR by directly inhibiting MDR1 because the INVSC cell lines transformed with MDR1 showed minimal growth inhibition in the presence of toxin and vanadate (Table 1). The wild-type lines were less sensitive to vanadate because they retained strong acid phosphatase activity. Like the trends for survival on nigericin and cycloheximide, growth inhibition seems to be a function of MDR1 abundance and ectophosphatase activity. Cell lines expressing MDR1 show a greater sensitivity to phosphatase inhibitors and exogenous applied ATP.

These experiments demonstrate that extracellular ATP concentrations affect the ability of yeast to extrude xenobiotics. Yeast cells must carefully regulate the ATP in their extracellular fluid to fully utilize either their endogenous MDR proteins (Kolaczkowski et al., 1996; Carvajal et al., 1997; Nourani et al., 1997) or those introduced by genetic manipulation.

Investigations of the Nucleotide Salvage Model

Previous studies have also implicated extracellular nucleotide hydrolysis in the MDR phenomenon. For example, leukemia cells upregulate ecto-5′-nucleotidase activity when in the presence of doxorubicin, even before upregulating P-glycoprotein activity (Pietkiewicz et al., 1998). These ecto-5′-nucleotidases hydrolyze nucleotide monophosphates to their corresponding nucleosides, a step necessary for transporting nucleosides back into the cell. Because MDR tumor cells extrude ATP as part of their resistance mechanism, perhaps the increased abundance of the ecto-5′-nucleotidase allows the cells to replenish their ATP supply by salvaging adenosine (Ujhazy et al., 1994, 1996). If the intracellular ATP pool of the MDR cell is diminished enough to affect viability, then cells able to salvage their ATP losses in this way would have increased viability. This nucleotide salvage model thus couples ectophosphatase activity with stabilizing the intracellular energy charge of the MDR cell; the model is suggested by the observation that AMP added exogenously with MDR toxins decreases cell sensitivity to the toxins (Ujhazy et al., 1994). In this model, MDR cells would hydrolyze their exogenous AMP to remedy their adenylate deficit (Ujhazy et al., 1996).

According to the salvage model, YMR4 cells would be more sensitive to toxins because the cells are unable to replenish their intracellular energy charge. With diminished ectophosphatase activity, the YMR4 mutant would not be able to recoup as much of the ATP effluxed from its active MDR proteins (whether native or transformed), and the cells would then be less viable in xenobiotic than in wild-type cells. To test this alternative hypothesis, we cultured the YMR4 cell line and its MDR1 transformant in ATP hydrolysis products to determine whether the sensitivities of these cells would diminish. According to the salvage model, these hydrolysis products should obviate the need for ectophosphatase activity and render the YMR4 mutants able to participate in nucleotide salvage to the same extent as the wild-type strain.

When the cycloheximide-containing medium was supplemented with adenosine, however, the YMR4 mock transformant still showed diminished growth, and the YMR4 MDR1 transformants did not experience any growth enhancement (Table 2). Likewise, phosphate, whether added alone or in the presence of adenosine, did nothing to decrease the xenobiotic sensitivities of these two cell lines. Controls confirm that cells are able to grow in medium alone as well as in each of these ATP hydrolysis products. Therefore, it is unlikely that the ectophosphatase mutants experienced diminished growth in cycloheximide because of a deficiency in a nucleotide salvage pathway.

Table 2.

Yeast Drug Sensitivity in the Presence of High Levels of ATP Hydrolysis Products^a

Without Cycloheximide		With Cycloheximide
Product Added and Cell Line	Turbidity	Product Added and Cell Line	Turbidity
No addition		No addition
YMR4/MDR1	1.376	YMR4/MDR1	0.937
YMR4/pvt101	1.429	YMR4/pvt101	0.001
Phosphate		Phosphate
YMR4/MDR1	1.351	YMR4/MDR1	0.541
YMR4/pvt101	1.341	YMR4/pvt101	0.001
Adenosine		Adenosine
YMR4/MDR1	1.319	YMR4/MDR1	0.632
YMR4/pvt101	1.354	YMR4/pvt101	0.002
Adenosine and Pi		Adenosine and Pi
YMR4/MDR1	0.899	YMR4/MDR1	0.389
YMR4/pvt101	1.342	YMR4/pvt101	0.001

^aYMR4 cells were transformed with the pvt101 vector containing the Arabidopsis MDR1 and are designated YMR4/MDR1, or YMR4 cells were transformed with the vector alone and are designated YMR4/pvt101. Cell growth was assayed by optical density at 660 nm and measured in triplicate after 96 hr.

AtPGP1 Complements a Pdr5 Null

We have thus far demonstrated that the ectophosphatase null yeast strain cannot become multidrug resistant and that the Arabidopsis P-glycoprotein enhances the ability of the mutant to acquire drug resistance. We next wanted to determine whether AtPGP1 functions in yeast analogously to native yeast drug resistance mechanisms. We therefore transformed the cDNA encoding the plant MDR1 into a yeast strain deficient in an ABC transporter that contributes to the MDR phenotype in yeast. Pdr5p (for pleiotropic drug resistance protein 5) promotes the resistance phenotype when overexpressed in yeast cells (Egner et al., 1995). A plasma membrane–localized protein, Pdr5p has been characterized as a drug efflux pump responsible for sporides-min and cycloheximide resistance in yeast (Bissinger and Kuchler, 1994). However, like the human MDR1, the yeast Pdr5p confers resistance toward several structurally unrelated compounds, including anticancer drugs such as duanorubicin, doxorubicin, and tamoxifen (Kolaczkowski et al., 1996). Accordingly, pdr5 null mutants have increased sensitivity to several xenobiotics, including cycloheximide.

When we transformed the MDR1 cDNA into YKKA-7, the pdr5 null mutant, we found that the MDR1-expressing line gained resistance to cycloheximide, whereas the pdr5 null mock-transformed line remained sensitive to the toxin (Figure 3). However, the growth of the MDR1-transformed Δpdr5 line in cycloheximide was comparable to that of the wild-type parent strain YPH499. Unlike the phosphatase null mutants, the pdr5 null strain was never able to grow in the presence of cycloheximide unless complemented with the plant MDR1 gene. Similar growth trends were seen in the presence of nigericin (data not shown). When grown in media alone, however, the pdr5 null mock-transformed strain grew at least as well as the MDR1-transformed counterparts, and both null mutants grew comparably to their parent strain (Figure 3).

Figure 3.

Complementation of the Yeast Mutant Δpdr5 by the Plant MDR1.

YKKA-7 is the Δpdr5 strain; YPH499 is the wild-type isogenic strain used to make YKKA-7; pvt101 is the empty expression cassette; and MDR1 is the plant P-glycoprotein.

(A) The growth of strains on 500 ng/mL cycloheximide. Clockwise, starting at the top: YKKA-7/pvt101, YPH499/pvt101, YPH499/MDR1, and YKKA-7/MDR1.

(B) Turbidity measurements of the same cell lines grown in cycloheximide. Pdr5p-deficient cells were sensitive to cycloheximide, whereas cells expressing the Arabidopsis P-glycoprotein were resistant. Experiments were performed at least five times with similar results.

(C) and (D) Growth of each strain in the absence of cycloheximide to control for growth phenotypes related to expression of the plant MDR1 and failed expression of the yeast Pdr5p.

MDR1 Expression Affects Concentrations of Extracellular ATP

The MDR phenomenon is common to many different organisms, and MDR1 functional homologs are found in plants and mammals. If ATP efflux and hydrolysis are necessary for resistance, then they should be characteristic of the MDR state in other organisms as well. To determine whether the positive correlation between MDR1 overexpression and extracellular ATP accumulation was evident in other systems, we studied Arabidopsis transformants overexpressing MDR1 under a strong constitutive viral promoter (Sidler et al., 1998).

We first assayed the leaf surfaces of MDR1-overexpressing plants (Sidler et al., 1998) for extracellular ATP, using the wild-type Arabidopsis as a point of comparison. The assay of extracellular ATP takes advantage of the fact that Arabidopsis can be grown in sterile culture under conditions humid enough to suppress cuticle development. On such plants, a droplet of buffer added to the leaf surface mixes with the extracellular matrix fluid. When the droplet is removed, some fluid from the extracellular matrix is removed with it, and the droplet can be measured for its ATP content. After assaying the plants in this way, we found that MDR1 expression correlated with higher concentrations of extracellular ATP. Plants overexpressing MDR1 had as much as two to three times the amount of extracellular ATP on the leaf surfaces as did wild-type plants (Figure 4), and this amount was statistically significant in two of five different lines tested.

Figure 4.

ATP Accumulation on the Leaf Surfaces of Arabidopsis Plants.

Measurement of leaf surface extracellular ATP in wild-type (Wt) and two MDR1-overexpressing (OE) lines. Data are reported as an ATP concentration calculated from a relative light units–based standard curve. Error bars indicate se. xATP, extracellular ATP.

These data suggested that the Arabidopsis MDR1 gene, similar to the human MDR1 gene, promotes the efflux of ATP. Moreover, because both Arabidopsis and yeast lines only doubled their extracellular ATP concentrations when transformed with MDR1, we see that both organisms have several highly active extracellular phosphatases maintaining the steady state extracellular ATP content, albeit at twice the normal extracellular ATP concentration.

Overexpressed Apyrase or MDR1 Confers Xenobiotic Resistance in Arabidopsis

In yeast, we found that extracellular ATP regulation was essential to the xenobiotic-resistant state; cells unable to regulate their extracellular ATP concentrations lost their MDR. We wondered whether this relationship would be seen in Arabidopsis as well. Therefore, we examined for xenobiotic sensitivity the plants overexpressing MDR1, and we compared their toxin resistance with that in plants overexpressing ectoapyrase, an ecto-ATPase with strong hydrolytic capacity. Most eukaryotes, except for yeast, use ectoapyrases for the bulk of their ectophosphatase activity. Apyrases are characterized by their ability to hydrolyze both the γ-phosphate and β-phosphate on ATP and ADP by their need for divalent cations, and by their failure to be inhibited by any known class of phosphatase inhibitors (Plesner, 1995). Most apyrases are expressed as plasma membrane–associated proteins, with their hydrolytic activity facing out into the extracellular matrix (Wang and Guidotti, 1996). The apyrase we used is derived from pea and localizes in a manner consistent with other apyrases (Thomas et al., 1999).

When we transformed a wild-type Arabidopsis line with the pea apyrase, we observed an increased resistance to cycloheximide, an increase also seen in the MDR1-overexpressing plants (Figures 5A to 5D). Because neither set of transgenics had prior exposure to the toxin, their endogenous resistance mechanisms had not been upregulated. Therefore, the transgenic plants could have gained resistance only as a result of the gene product introduced, that is, through the addition of MDR1 or the ectoapyrase. These results with Arabidopsis are consistent with our earlier findings in yeast. The ATP efflux and ATP hydrolysis were seen to be functionally related to xenobiotic resistance. In yeast, mutants lacking ectophosphatase activity became sensitive to toxins; in plants, transgenics overexpressing ecto-ATPase activity became resistant to them.

Figure 5.

Toxin Resistance in Arabidopsis.

(A) to (C) Resistance to cycloheximide is conferred by the overexpression of either apyrase or Arabidopsis MDR1 in Arabidopsis. (A) Wild-type seeds failed to germinate in 250 ng/mL cycloheximide. (B) Seeds from lines that overexpress apyrase germinated in 250 ng/mL cycloheximide. (C) Seeds from lines that overexpress the Arabidopsis MDR1 germinated in 250 ng/mL cycloheximide.

(D) Histogram showing germination percentages for wild-type (Wt), apyrase-expressing (Apyrase OE), and P-glycoprotein–expressing (MDR1 OE) lines sown in 250 ng/mL cycloheximide. Approximately 80 to 100 plants were evaluated for each line.

(E) Effects of the ATPase inhibitor methyleneadenosine 5′-diphosphate on wild-type (Wt) plants and on plants overexpressing MDR1 (MDR1 OE) in the presence (+) and absence of cycloheximide. Ratios below the photographs denote the percentage of MDR1-expressing plants that germinated (numerator) relative to the percentage of wild-type plants that germinated (denominator) for each treatment.

When Arabidopsis seeds were exposed to cycloheximide, the germination rate of those overexpressing apyrase was almost double that of seeds overexpressing MDR1 (Figure 5D). If ATP efflux is a mechanism used by other, as yet uncharacterized, xenobiotic transporters, then overexpressing the ectoapyrase may have heightened the activity of these transporters, thus enhancing their efficacy. Conversely, disrupting the ecto-ATPase activity of plant cells served to abolish toxin resistance. When an extracellular ATPase inhibitor (α,β-methyleneadenosine 5′-diphosphate) was simultaneously administered with cycloheximide, MDR1-overexpressing plants lost their resistance and became as sensitive to the toxin as wild-type plants in a germination-based assay (Figure 5E). Because the ATPase inhibitor itself does not enter the cell, the sensitization of MDR1-expressing plants was not caused by an abrogation of the MDR1 protein ATPase activity. The reversal of resistance was consistent with the dissipation of native ecto-ATPase activity.

In separate experiments, both MDR1 and apyrase transformants showed increased resistance to toxic concentrations of plant growth regulators such as cytokinin. The cytokinin N⁶-(2-isopentenyl)adenine (2IP), which itself meets the loose chemical criteria for an MDR1 substrate, severely stunts plant growth at high concentrations. Relative to their growth on a hormone-free medium (Figure 6A), wild-type Arabidopsis plants showed stunted growth or failed to germinate in the presence of 1 mM 2IP (Figures 6B and 6D). Arabidopsis lines overexpressing either MDR1 or ectoapyrase, however, were much less stunted in their growth (Figures 6B and 6C). Furthermore, at concentrations that inhibited root elongation in wild-type plants, MDR1-expressing plants had long roots (Figures 6C and 6E). The resistance of the transformed plants to cycloheximide and 2IP was specific insofar as the overexpressing lines did not grow any more robustly than the wild type when toxins were not present.

Figure 6.

Resistance to Toxic Doses of Hormone in Arabidopsis Transformed with Ectoapyrase or MDR1.

(A) The growth of wild-type, apyrase-overexpressing, and MDR1-overexpressing plants on untreated culture medium.

(B) and (D) Growth of the three types of plants in a culture medium containing the plant hormone 2IP at 1000 μM (B). Toxic effects of 1000 μM 2IP on plant growth progression as indicated by leaf expansion (D).

(C) and (E) Growth of the same three plant types in culture medium containing 2IP at 100 μM (C). Measurement of toxicity based on root elongation averaged (±sem) for three experiments (E). Data on the left portion of the histogram were collected from plants grown on media alone and on the right from plants grown on 100 μM 2IP.

Apy OE, apyrase overexpressing; MDR1 OE, MDR1 overexpressing; Wt, wild type.

Taken together, these experiments demonstrate that extracellular ATP hydrolysis may contribute to xenobiotic resistance as does MDR1; indeed, an increase in either can confer toxin resistance in Arabidopsis. More importantly, studies in Arabidopsis confirm the findings in yeast, suggesting that ATP efflux and hydrolysis are part of a more general mechanism of xenobiotic resistance.

Phosphatase Activity Is Itself Crucial in the Development of MDR in Yeast

If extracellular ATP hydrolysis is necessary for establishing cellular toxin resistance, then ectophosphatase activity could be part of an endogenous mechanism utilized when cells are first introduced to MDR substrates. Using the NBT/BCIP assay for measuring ectophosphatase activity in living yeast cells (Thomas et al., 1999), we tested whether yeast cells showed increased phosphatase activity when first introduced to xenobiotics, as seen in Figure 7. All of the yeast strains were sequentially tested for phosphatase activity after being cultured in cycloheximide or nigericin for 3 days. Both INVSC mock transformants and INVSC MDR1-expressing cells showed increased extracellular acid phosphatase activity when cultured in nigericin and in cycloheximide (Figure 7A). The YMR4 mock transformants could not be tested because they were unable to grow in cycloheximide. Although the YMR4 MDR1-expressing cells could be cultured in several xenobiotics, they showed no observable phosphatase activity, whether cultured in plain medium or in the presence of toxin. Because the YMR4 line is a pho3 and pho5 null mutant, this absence of activity is not surprising. In addition to these cell lines, we tested the pdr5 null mutants overexpressing MDR1, their wild-type background strain, and the background strain overexpressing MDR1; each exhibited increased phosphatase activity after 3 days of culture in these drugs (Figure 7B). The Δpdr5 mock transformants could not be tested because they did not grow in drug cultures. When grown in medium alone, every strain had only slight acid phosphatase activity.

Figure 7.

Extracellular Phosphatase Activity Increased When Cells Were Exposed to Xenobiotic Agents.

(A) Acid phosphatase activity was visualized by the precipitation of blue chromogen on the cell surface of yeast incubated in acetate-buffered NBT/BCIP at pH 4.5. Transformed YMR4 and INVSC are shown in the presence (+) and absence of cycloheximide (CHX).

(B) Δpdr5 and wild-type cells grown in the presence of CHX or nigericin (Nig) and tested for extracellular acid phosphatase activity.

To ensure that the MDR substrates themselves did not affect the NBT/BCIP assay, we tested medium containing cycloheximide and obtained results identical to those acquired by the chromogenic assay of medium alone. To ensure that the results were not attributable to intracellular phosphatases released when the drug-grown cells lysed, we vortex-mixed yeast cells with glass beads to release their cellular contents before assay. These lysed cells showed no more activity than normal, medium-cultured cells (data not shown).

DISCUSSION

ATP Gradients

Our results reveal a new role for extracellular ATP in the phenomenon of MDR. We see a consistent relationship between the ability of an organism to degrade extracellular ATP and its ability to acquire toxin resistance. Yeast strains unable to modulate their extracellular ATP concentrations with ecto-ATPase activity do not grow in the presence of toxin. On the other hand, the yeast strains that do develop toxin resistance have markedly enhanced ectophosphatase activities. Phosphatase inhibitors, such as vanadate, further sensitize ectophosphatase null mutants to toxins and, in the case of YMR4/MDR1, make an otherwise multidrug-resistant organism unable to grow in the presence of toxin. If yeast cells are cultured in the presence of toxin and nonphysiological concentrations of ATP, then the cells experience diminishing resistance–losses that correspond to both the concentration of ATP added and the ectophosphatase activity of the cell. More than 50% growth diminution is seen in every cell line that is cultured in medium containing toxin and ATP at twofold the expected intracellular ATP concentration.

The sensitivity of ectophosphatase null mutants, the increased ectophosphatase activity of drug-cultured yeast cells, and the effect of extracellular ATP on toxin resistance all suggest that the hydrolysis of extracellular ATP is necessary to maintain the resistant state. Our observations indicate that the need for this hydrolysis arises from ATP extrusion events associated with the upregulation of the plant MDR1 protein. Whether acting as a channel for ATP itself or by activating some other ATP pore, the plant MDR1 increases the concentrations of ATP both in the periplasm of yeast and in the apoplasm of Arabidopsis. Our measurements of extracellular ATP accumulation are conservative approximations of the ATP that might actually be found in the microenvironment of the plasma membrane. Indeed, most of the ATP we measured was that which had escaped ectophosphatase activity. Presumably, an important amount is hydrolyzed immediately on the membrane surface if the ectophosphatases are intact; conversely, a very large amount accumulates periplasmically if phosphatases are not there.

The possibility exists that the extracellular ATP we measured can be accounted for by nonspecific changes in membrane permeability caused by the overexpression of MDR1. We find this alternative unlikely. Nonselective increases in membrane permeability would not just allow ATP out but would also allow other small molecules in, molecules such as methylene blue, tritiated adenosine, and the phosphate analog NBT/BCIP. However, viability assays using methylene blue demonstrate no difference in the INVSC mock-transformed line and the MDR1-transformed line, a result that would not be obtained if the MDR1-expressing lines were more permeable to the dye. Similarly, the YMR4 MDR1-transformed strain exposed to NBT/BCIP showed no more phosphatase activity than its mock-transformed counterpart, a result that would not be possible if the now membrane-permeable NBT/BCIP were exposed to intracellular phosphatases in the MDR1-expressing cell line. Last, our tritiated adenosine experiments revealed that when the YMR4 and INVSC cell lines were incubated with tritiated adenosine for 40 min, their initial rates of nucleoside uptake were all comparable; the MDR1-expressing strains did not have a greatly reduced extracellular-to-intracellular ratio for label. Thus, we think it likely that the increased extracellular ATP concentrations in the MDR1-overexpressing lines result more specifically from the channel, or channel-activating, properties of AtPGP1 itself.

Work put forth in the early 1990s demonstrated that P-glycoprotein is capable of effluxing ATP (Abraham et al., 1993). In that study, MDR1 promoted the electrogenic efflux of ATP from transfected mammalian cells at a rate of 4 × 10⁶ molecules per sec per cell. More recent estimations have placed the efflux velocity at 1 to 10 molecules per channel per sec (Abraham et al., 1998). Our data suggest that a population of 3 × 10⁶ yeast cells in 1 mL of culture produces a steady state concentration that is 4 × 10⁵ times less than the intracellular ATP concentration of any single cell, given an extracellular ATP concentration of 10 nM (our measurement) and an intracellular ATP concentration of 4 mM (Ludin, 1989). If this gradient were to be used to cotransport xenobiotics, then MDR xenobiotics should interact directly with ATP. Indeed, the work of Abraham and colleagues, based on chemical electrophoretic mobility shift assays, shows that ATP actually has an affinity for doxorubicin (Abraham et al., 1997). Furthermore, the interaction of zwitterionic xenobiotics with ATP is specific to this nucleotide; similarly, there is a notable lack of affinity for several xenobiotics to which MDR cell lines are sensitive.

Our results are consistent with the possibility that ATP and xenobiotics might move together through MDR1. The speculation that drugs are symported with ATP also stipulates that a very low steady state concentration of extracellular ATP must be maintained to form a gradient of ATP across the plasma membrane. If the gradient is compromised either by knocking out the ecto-ATPase activity of the cell or, more simply, by artificially increasing the extracellular ATP to twice the intracellular concentration, the result for the resistant cell is the same–death in the presence of a xenobiotic. If the MDR1 channel cotransports toxins with ATP, then the ability of a cell to remain resistant would be contingent on its ability to maintain a steep ATP gradient. Our work with the YMR4 mutant has revealed that the ATP gradient could be diminished simply by destroying extracellular phosphatase activity. Moreover, the ATP symport model for MDR would predict that this loss of ectophosphatase activity should simultaneously decouple cellular resistance to toxins, a hypothesis borne out by our observation that the YMR4 cells were the most sensitive to xenobiotics.

If a steep ATP gradient were powering MDR1-mediated drug efflux, then cells on initial presentation with toxins would need not only to upregulate MDR proteins to efflux drugs but also to enhance their ATPase activity to maintain the gradient. Our studies revealed that cell lines cultured for 3 days in cycloheximide, nigericin, or both had markedly increased ectophosphatase activities. The fact that all resistant yeast strains cultured in toxin take 24 hr longer to reach growth saturation than do their counterparts cultured in non-toxin-containing media could be explained by the need to upregulate ectophosphatases to energize the available MDR1-like proteins. Both the 24-hr lag and the corresponding increased ectophosphatase activity in toxin are evident in the INVSC strain, the PDR5 null strain, and YKKA-7 and its parent strain, YPH499.

An ATP symport model would also allow us to explain how a plant MDR1 cDNA is able to complement both an ectophosphatase null mutant and a pdr5 null mutant. The data from the pdr5 complementation experiments demonstrate that the plant MDR1 protein is a functional homolog of Pdr5p. However, the fact that the MDR1 gene can also complement the drug resistance of a yeast ectophosphatase mutant is suggestive of a mechanistic complementation, a complementation that implicates ectophosphatases and drug extrusion pumps in a single and shared mechanism of achieving toxin resistance. We can imagine that ectophosphatase mutants, unable to maintain steep ATP gradients, have MDR-like pumps that extrude drugs slowly and inefficiently. The ATP symport model would suggest that by simply increasing the number of drug efflux pumps, the ectophosphatase yeast mutant could achieve the same result as a cell line with fewer but more efficiently energized MDR1-like proteins.

In some sense, the role of ectophosphatase activity in xenobiotic resistance was established with the observation that 5′-ectophosphatase upregulation is induced by MDR substrates even before MDR1 induction (Ujhazy et al., 1996). In a bacterial system, the LmrA gene, a prokaryotic functional homolog of MDR1, is found in the same operon as an alkaline phosphatase gene, which it abuts (van Veen et al., 1996). Although no evidence at present indicates that the phosphatase functions with LmrA in MDR, it is convergently transcribed with the LmrA gene, a fact highly suggestive of a role in MDR. In addition to the evidence from bacteria, results from yeast suggest that ectophosphatase activity increases resistance to the polygenic antifungal agent, amphotericin B (Ramanandraibe et al., 1998). To date, however, no studies have functionally linked ecto-ATPase activity and xenobiotic resistance.

In Arabidopsis plants, the overexpression of a CD39 ectoapyrase homolog from pea (Wang and Guidotti, 1996) itself confers toxin resistance. This provides an example in which a multicellular organism has been genetically modified to have MDR addressed with a gene other than the one encoding an ABC transporter. The fact that increased ecto-ATPase activity can itself confer this seemingly complex phenotype suggests that the activity is pivotal to the translocation of xenobiotics across the plasma membrane. The fact that not all intracellular ATP is lost through the ABC protein attests to the fidelity of the scavenging mechanisms for phosphate (Thomas et al., 1999) and adenosine. It also suggests that the MDR1 channel is gated and that its ATP efflux activity is in some measure controlled.

The ABC proteins found in bacteria are structurally more reminiscent of pumps, whereas eukaryotic ABC proteins such as the cystic fibrosis transmembrane conductance regulator (CFTR) and MDR1 are channel-like (Welsh et al., 1998). However, the complementation of human MDR1 mutants with bacterial transporters (van Veen et al., 1998) unequivocally demonstrates that the pump model is biologically feasible insofar as a prokaryotic pump can functionally substitute for a putative eukaryotic channel. We realize that ATP hydrolysis by P-glycoprotein is real and critical to xenobiotic efflux. However, we speculate that ATP hydrolysis may not be directly coupled to the energetics of transport but instead could be required for the gating of the ATP channel. Inferences from structural data suggest that the nucleotide binding domains of the eukaryotic ABC transporters may participate in potentiating the MDR “pump-channel” (Welsh et al., 1998). The two pools of ATP, one involved in gating and the other in efflux, may be coordinately regulated.

Does Extracellular ATP Act Indirectly in MDR?

Because our experiments demonstrated that increased ATP efflux correlated with the xenobiotic resistant state in both yeast and Arabidopsis, we wondered whether extracellular ATP could act as a signaling molecule to promote MDR. Extracellular ATP is known to act as a signaling molecule in several systems, for example, as a substrate for P2X receptors—a class of ATP gated cation channels—and P2Y receptors—a class of heterotrimeric G protein receptors (North, 1996). In MDR, extracellular ATP has been known to influence sodium–proton exchange globally, which explains the effect of P-glycoprotein upregulation on plasma membrane potential (Weiner et al., 1986; Huang et al., 1992). Extrapolating from this idea, Wadkins and Roepe (1997) proposed that extracellular ATP may cause changes in membrane potential that specifically affect passive xenobiotic uptake. In this model, MDR would result mainly from xenobiotic exclusion.

There is good reason to suspect that ATP may have an indirect effect on membrane potential because such a mechanism has been demonstrated with another ABC ATP conduit, the CFTR. CFTR releases ATP into the extracellular fluid; once there, ATP functions as a signal to regulate outwardly rectifying chloride channels (Schwiebert et al., 1995). In this scenario, extracellular ATP is released by CFTR (Reisin et al., 1994) and binds to a G protein–coupled ATP receptor that ultimately activates a cAMP-dependent kinase; this process culminates in the opening of an outwardly rectifying chloride channel (Al-Awaqati, 1995). The activity of the chloride channel is dependent on the availability of ATP, which in turn is regulated by ecto-ATPase activity. If ATP efflux through MDR1 functions analogously, then a change of membrane potential would be expected with the onset of ATP efflux because this extracellular ATP would target purinergic receptors that themselves regulate polarization. As the membrane potential changes, the partitioning of hydrophobic xenobiotics into the membrane could be altered in such a way as to cause xenobiotic exclusion (Wadkins and Houghton, 1995).

Ecto-Apyrases as MDR Reversal Targets

The strongest known ATPases are apyrases, some of which have a turnover number of 10⁴ molecules per sec (Handa and Guidotti, 1996). Because ectoapyrases are known to exist in human cells (Wang and Guidotti, 1996), we speculate that they could be the major ATPases on the cell surface. The ectoapyrase from pea plants, which is encoded by a single gene, accounts for >60% of the total cell surface ATPase activity in roots (Y. Sun and S.J. Roux, unpublished data). Future studies will determine whether the human apyrase, CD39, is itself upregulated in MDR cells in a manner consistent with other ATPases. We hypothesize that ectoapyrase will be an important target for MDR reversal.

METHODS

Expression of AtPGP-1 in Yeast

A full-length 4-kb AtPGP1, or MDR1, cDNA was assembled from several pieces that were obtained by reverse transcription–polymerase chain reaction (RT-PCR). The 1.9-kb SacI-BamHI insert of pMDRPCR123 (Sidler et al., 1998) containing the 5′ half of the AtPGP1 cDNA was subcloned into the same sites of pUC12 to give pUCcMDR5. A 3′ part of the AtPGP1 cDNA that overlapped with the 1.9-kb 5′ part over 200 bp was obtained by RT-PCR with the oligonucleotides 5′-GGCTGCTCGAGTCGCAAATG-3′ (sense strand; sequence corresponds to nucleotide positions 2912 to 2931) and 5′-cgggatCCAAAGTAGTAAGTACTAAGC (antisense strand; sequence corresponds to positions 5822 to 5842 [Dudler and Hertig, 1992]); the nucleotides shown in lowercase were added to provide a BamHI restriction site. The fragment obtained was digested with XhoI and BamHI and was used to replace the ∼200-bp XhoI-BamHI fragment of pMDR5 with which it overlapped. The resulting clone was named pUCcMDR and contained the entire coding sequence of the AtPGP1 gene, which we verified by sequencing. The complete 4-kb cDNA fragment was cut out with StuI (which cleaves in the leader sequence) and SalI (which cuts in the polylinker of pUC12 downstream of the cDNA insert) and inserted into a EcoRV-XhoI–cleaved pcDNA I/Amp vector (Invitrogen, Carlsbad, CA); from this, it was cut out again as a BamHI fragment and finally inserted in the correct orientation into the BamHI site of the yeast expression vector pvt101U (Vernet et al., 1987). pvt101 containing MDR1 and pvt101 alone were each independently transformed into Saccharomyces cerevisiae INVSC1 (MATαhis3-Δ1 leu2 trp1-289 ura3-52), YMR4 (MATαhis3-11,15 leu2-3 112ura3Δ5 can Res pho5,3::ura3Δ1), YPH499 (ura3-52 leu2-Δ1 his3-Δ200 ade2-101och lys2-801a trp1-Δ63), and YKKA-7 (ura3-52 leu2-Δ1 his3-Δ200 ade2-101och lys2-801a Δpdr5::TRP1) by a polyethylene glycol–lithium acetate procedure (Elble, 1992) and selected on uracil dropout medium.

Yeast Growth

Yeast were grown at 30°C under conditions of constant selection for uracil auxotrophy. Supplemented YNB (Bio101, Vista, CA) and 2% glucose were used to grow strains having pvt101 constructs. All media were at pH 5.5. The nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) assay for cell surface phosphatase activity was performed as described by Thomas et al. (1999). Cycloheximide (Sigma) was added to liquid medium or spread on solid medium to a final concentration of 500 ng/mL, unless otherwise stated. Nigericin (Sigma) was added to liquid medium or spread on solid medium to a final concentration of 0.025 mg/mL. For selection assays on plates, single colonies were streaked; for selection in liquid medium, 0.01 mL of saturated culture was added to 5 mL of fresh medium containing the xenobiotic. Plated colonies were grown for 3 to 5 days before being photographed. Yeast selection assays in liquid medium were quantified by turbidity as measured by absorbance at OD₆₆₀. All selection assays were repeated at least five times.

For those experiments testing the salvage hypothesis, all cell lines were grown in cycloheximide until , at which point 10-μL samples of each culture were removed and placed in each of the following seven media: medium with nothing added; medium plus 3 mM potassium phosphate; medium plus 3 mM adenosine; medium plus 9 mM potassium phosphate and 3 mM adenosine (for controls); medium plus potassium phosphate and cycloheximide; medium plus adenosine and cycloheximide; and medium plus adenosine, cycloheximide, and potassium phosphate. Cell cultures were grown for another 72 hr, and turbidity was measured. In the experiments using sodium vanadate and ATP in assays of MDR reversal, the vanadate concentration was 0.125 mM and the ATP concentration was 12 mM. Growth was assayed by turbidity at 96 hr and was normalized for the same cell line grown in cycloheximide to determine inhibition.

Transgenic Plant Construction

psNTP9 (Pisum sativum apyrase; GenBank accession number Z32743) was subcloned as a SalI-XbaI fragment into pKYLX71 (Schardl et al., 1987). This plasmid was transformed into Agrobacterium GV3101 (pMP90) (Koncz and Schell, 1986), which was used to infect root calli from Wassilewskija ecotype of Arabidopsis thaliana under kanamycin selection (Valvekens et al., 1992). The methods for construction of overexpressing and antisense-suppressed AtPGP-1 (A. thaliana MDR1 gene; accession number X61370) in Arabidopsis are described by Sidler et al. (1998). Homozygous plants for the overexpressed MDR1 and cDNAs encoding pea apyrase were used in all experiments.

Plant Growth

Arabidopsis seeds were sown in a solid germination medium containing Murashige and Skoog salt (Sigma), 2% sucrose, 0.8% agar, and vitamins (Valvekens et al., 1992). For selection assays, cycloheximide was spread on the medium to a final concentration of 250 ng/mL. For cotyledon expansion assays, the cytokinin N⁶-(2-isopentenyl)adenine (2IP) was applied to plates to a final concentration of 1 mM. For experiments using methyleneadenosine 5′-diphosphate, the cycloheximide concentration was 500 ng/mL and the ATPase inhibitor concentration was 1 mM. Plant growth was measured by the percentage of plants germinating after 6 days in experiments using cycloheximide alone (Figure 4) and after 10 to 20 days in experiments using methyleneadenosine 5′-diphosphate or 2IP.

ATP Collection and Luminometry

Yeast cells used in the luciferase assays were grown for 2 days and then transferred to 5 mL of fresh medium at the time of the assay. From this time forward, the cells were kept at room temperature on a rotator. Every hour, a 1-mL aliquot was taken and the cells were counted with a hemocytometer, checked for viability with methylene blue, and centrifuged; the supernatant was stored in liquid nitrogen until all the aliquots were collected. If viability was <95%, the experiment was ended. For luciferase assays involving plants, Arabidopsis seedlings were grown in sterile culture at 22°C under 150 to 200 μmol m⁻² sec⁻¹ of continuous light for at least 15 days. Foliar ATP was collected by placing a single 30-μL drop of luciferase buffer (Analytical Luminescence Laboratory, Cockeysville, MD) on a leaf and, without making direct physical contact with the plant, immediately collecting the droplet and snap-freezing it. For each leaf, the area was approximated as an integrated area of a two-dimensional image of the leaf by using National Institutes of Health 1.52 software. An average of at least 10 independent measurements is presented for data points in plots of yeast and plant ATP. For luminometry, samples were thawed and reconstituted in Firelight buffer (Analytical Luminescence Laboratory) to a final volume of 100 μL. After buffer was added, all samples were kept on ice. ATP standards (Sigma) were reconstituted in 100 μL of Firelight buffer, and the standards and samples were loaded onto a 96-well plate, where the absorbance of each well was read with an automated luminometer (model mLX; Dynex Technologies, Chantilly, VA). Samples were processed by addition of 50 μL of Firelight enzyme, followed by a reading delay of 1.0 sec and an integration time of 10 sec. Output was taken as an average for the integration time and then as averages for multiple samples. The sample handling time was <2 hr.

Pulse–Chase Experiments

Yeast were grown to saturation in liquid medium, as described previously, centrifuged, and resuspended in fresh medium containing 1 μCi/mL ³H-adenosine (Amersham, Arlington Heights, IL). The cells were rotated at room temperature for 20 min to allow adenosine uptake. After 20 min, the cells were pelleted by centrifugation. The pellet was washed twice in ice-cold medium, resuspended in culture medium at room temperature, divided equally among five tubes (five per cell line), and placed on a rotator. Every 10 min, a separate sample-containing tube from each cell line was centrifuged, and the pellet and the supernatant were placed in separate scintillation vials. The efflux activity was expressed as the ratio of counts in the supernatant to counts in the pellet.