INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plants and animals deploy equivalent four-phase processes for the detoxification of toxins that they produce themselves or to which they are exposed (Ishikawa, 1992; Sandermann, 1992; Coleman et al., 1997; Ishikawa et al., 1997; Rea et al., 1998). Phase I (activation) is the introduction or exposure of functional groups of the appropriate reactivity for phase II enzymes. Cytochrome P450-dependent monooxygenases and mixed function oxidases are examples of phase I enzymes. Phase II (conjugation) is covalent attachment of the activated compound to a bulky hydrophilic molecule, a process considered to increase water solubility and promote recognition by phase III transporters. Phase III (elimination) is transport of the conjugates out of the cytosol into intracellular compartments and/or the extracellular space. In animals, the conjugates are usually excreted into the urine or bile. In plants, organisms that otherwise lack bona fide excretory organs, the conjugates are often sequestered in the vacuole, a large acidic intracellular compartment that constitutes 40% to 90% of total cell volume. Phase IV (transformation) is the processes responsible for further substitution, degradation, and/or partial salvage of the conjugates. Whether the compound concerned is a secondary metabolite or xenobiotic, the underlying principle is that activation and conjugation increase its hydrophilicity, thus diminishing its inherent lipid solubility and capacity for cell-to-cell diffusion, while introducing a specific chemical tag for carrier-mediated transport out of the cytosol.

Because of their pharmacological importance, the best characterized phase II reactions are those catalyzed by mammalian UDP-glucuronyltransferases that attach GlcUA to a wide range of acceptor molecules (Meech and Mackenzie, 1997). Although closely related homologs exist in plants, as indicated by the presence of some 100 open reading frames in Arabidopsis encoding polypeptides containing a C-terminal consensus sequence common to all members of the UDP-glycosyltransferase superfamily (Li et al., 2001; Lim et al., 2001), much less is known about these enzymes than their mammalian counterparts. The majority of the plant enzymes are presumed to use UDP-Glc as the sugar donor and an increasing number have been purified and enzymologically characterized in the last several years (Ford et al., 1998; Fraissinet-Tachet et al., 1998; Lee and Raskin, 1999; Vogt et al., 1999; Jackson et al., 2001), but their natural substrates and physiological roles largely remain to be determined. Notwithstanding our ignorance of these enzymes, however, it is suspected that one of the roles of plant UDP-glucosyltransferases is to target endogenous and exogenous toxins to the vacuole.

An impressive array of naturally occurring plant secondary metabolites are glycosylated (Harborne, 1993) and known to accumulate in the vacuole (Wink, 1997; Martinoia et al., 2001), including coumaryl glucosides, flavonoids, anthocyanins, cardenolides, saponins, cyanogenic glucosides, glucosinolates, and betalains. On this basis and because UDP-glucosyltransferases are considered to have a cytosolic localization, glucosylation has been invoked as a prerequisite for the vacuolar uptake of many compounds. In agreement with this concept, the results of in vitro experiments demonstrate that isolated vacuoles and/or vacuolar membrane vesicles are capable of accumulating certain Glc conjugates (Wink, 1997; Martinoia et al., 2001). However, what is not clear is the identity of the carriers responsible, their mode(s) of energization, and the minimum structural requirements that must be fulfilled by a glucoside for its net uptake by vacuoles. Although uptake usually requires an energy source, for instance ATP, and occurs against a concentration gradient, the elucidation of general principles has been confounded by the paucity of systematic investigations and the seeming dependence of the energetics and kinetics of uptake not only on the identity of the Glc conjugate but also on the plant species from which the vacuoles were isolated. Another complicating factor is that there are examples of Glc conjugates that cannot enter the vacuole unless they are further modified, for instance by acylation (Matern et al., 1986; Hopp and Seitz, 1987; Wink, 1997). This is the case for apigenin 7-O-glucoside (the major pigment in parsley), which is conjugated to malonic acid before vacuolar transport (Matern et al., 1986). Although malonic acid hemiesters of glucosylated herbicides have also been found in plants (Onisko et al., 1994; Wink, 1997), it is not known if these auxiliary modifications are required for or contribute to the vacuolar sequestration of xenobiotics of this type.

Three principal mechanisms for vacuolar glucoside accumulation have been proposed: H⁺-antiport, conformational trapping, and direct energization by ATP. Of these, the first two depend on the H⁺-electrochemical potential difference across the vacuolar membrane, established by the vacuolar H⁺-ATPase (V-ATPase) and vacuolar H⁺-pyrophosphatase (Rea and Sanders, 1987; Zhen et al., 1997; Sze et al., 1999). Both pumps catalyze electrogenic H⁺-translocation from the cytosol into the vacuole to establish an inside-acid pH gradient (pH) and an inside-positive electrical potential difference ().

Depending on the glucoside and/or the source of vacuolar membranes, the H⁺ gradient can drive uptake in different ways. For example, in barley (Hordeum vulgare) vacuoles, the carrier that mediates the uptake of isovitexin, an endogenous flavone C-glucoside derived from apigenin, is thought to be coupled to the H⁺ gradient by an H⁺-antiporter (Klein et al., 1996). Thus, uptake of isovitexin is stimulated by ATP and partially inhibited by agents that collapse or prevent the formation of an H⁺ gradient. Similar findings have been reported for other endogenous glucosides, including those of oleanolic acid (Szakiel and Janiszowska, 1993) and esculetin (Werner and Matile, 1985), suggesting that the coupled efflux of intravacuolar protons provides the motive force for glucoside influx, without the need for modification or structural rearrangement of the transport substrate.

In other cases, it has been difficult to distinguish an H⁺-antiport from other modes of translocation that merely exploit the acidity of the vacuole lumen for conformational trapping. For example, conformational trapping has been invoked for the vacuolar accumulation of 7-O-(6-O-malonylglucoside) (Matern et al., 1986). The kinetics of uptake for this compound are consistent with a carrier-mediated process but one in which directionality is imposed by structural isomerization of the glucoside consequent on its delivery into the acidic vacuole lumen. A similar mechanism has been proposed for trans-O-coumaric acid glucoside, which undergoes an acid-induced isomerization to its transport-incompetent cis form (Rataboul et al., 1985). In both cases, the backbone structure is taken up against a concentration gradient, not by a concentrative active transport process per se, but instead through a mass action effect driven by structural modifications of the substrate after uptake.

The one known mechanism for the accumulation of secondary metabolites and xenobiotics that is not obligatorily dependent on vacuolar acidity is ATP-energized transport by ATP-binding cassette (ABC) transporters (Rea et al., 1998; Rea, 1999; Martinoia et al., 2001). First defined in the context of glutathione conjugate transport (Martinoia et al., 1993; Li et al., 1995), it is now clear that many other xenobiotics and secondary metabolites are amenable to accumulation by vacuolar ABC transporters (Rea et al., 1998; Rea, 1999; Martinoia et al., 2001). The salient properties of ABC transporter-catalyzed vacuolar uptake are: (a) direct energization by ATP hydrolysisother nucleoside triphosphates can substitute, but non-hydrolyzable analogs and inorganic pyrophosphate cannot; (b) exquisite sensitivity to vanadatemicromolar concentrations of vanadate, a phosphoryl transition-state analog, strongly inhibit uptake (Li et al., 1995); (c) little or no reliance on the transmembrane H⁺-electrochemical gradientabolition of the H⁺ gradient can partially inhibit transport but seldom, if ever, to the extent that vanadate does (Li et al., 1995; Lu et al., 1998; Klein et al., 2000).

Despite the abundance and ubiquity of ABC transporters in plants, as exemplified by the identification of 103 open reading frames encoding membrane proteins of this type in Arabidopsis (Sánchez-Fernández et al., 2001), and the established roles of at least two of them, both members of the multidrug resistance-associated protein (MRP) subclass, in the vacuolar sequestration of glutathione conjugates (Lu et al., 1997, 1998), we are aware of only two in vitro studies in which they have been directly implicated in the vacuolar uptake of glucosides. Moreover, in both studies, the same compound, hydroxyprimisulfuron glucoside (HPSG), the major detoxification product of the sulfonylurea herbicide primisulfuron, was employed. Based on the finding that HPSG uptake by barley vacuoles is relatively insensitive to agents that interfere with the H⁺ gradient but strongly inhibited by vanadate (Klein et al., 1996), the susceptibility of the AtMRP3 gene (EST 107J19T7) of Arabidopsis to induction by primisulfuron (Tommasini et al., 1997), and the capacity of herbicide safeners to increase HPSG transport in barley vacuoles (Gaillard et al., 1994), it has been suggested that the carrier responsible for ATP-energized uptake of this glucoside is an ortholog of AtMRP3 (Rea, 1999). Although highly speculative and derived from the results of only a few experiments, the idea has arisen that naturally occurring glucosylated plant secondary metabolites enter the vacuole by H⁺-antiport, whereas glucosylated xenobiotics enter via ABC transporters, possibly MRP-subclass glutathione conjugate pumps (Martinoia et al., 2001). An extension of this idea, supported by the results of very recent comparative analyses of flavone glucoside uptake into barley mesophyll and Arabidopsis cell culture vacuoles, is that although endogenous flavone glucosides are subject to H⁺-antiport, their nominally "xenobiotic" equivalents, exogenous flavone glucosides, are transported by ABC transporters (Frangne et al., 2002).

There have been two major impediments to the rigorous analysis of vacuolar glucoside transport. The first is the scarcity and expense of radiolabeled glucosides for vacuolar transport assays. The second is that no laboratory has examined vacuolar uptake of different conjugates of the same or closely related compounds. Here, we attempt to address both issuesthe first through the development of a liquid chromatography (LC)-mass spectrometry (MS)-based protocol for quantitative analysis of the transport of unlabeled compounds by vacuolar membrane vesicles, and the second by direct comparisons of the energetics and kinetics of uptake of closely related compounds conjugated to different ligands. Using vacuolar membrane vesicles isolated from the red beet (Beta vulgaris) storage root as a model system (Rea and Turner, 1990; Li et al., 1995), we have examined the transport properties of two structurally related sulfonylurea herbicides, chlorsulfuron (CS) and chlorimuron-ethyl (CE), before and after conjugation with Glc and glutathione, respectively. The results demonstrate that the glucosylated herbicide is transported by a carrier that is energetically coupled to the transmembrane H⁺ gradient, whereas the glutathionated herbicide is transported by an ABC transporter-like carrier that is directly energized by ATP. We further show that attachment of Glc to the aromatic OH group, but not the carboxyl group, of p-hydroxycinnamic acid (pHCA) and p-hydroxybenzoic acid (pHBA), enables vacuolar accumulation and that the transport properties of these two naturally occurring plant secondary metabolites are remarkably similar to those of the glucosylated herbicide.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Glucosylation of 5-Hydroxychlorsulfuron Is a Prerequisite for Transport

Detoxification of the sulfonylurea herbicide CS in some plant species is initiated by one or more members of the cytochrome P450 superfamily (Lau and O'Keefe, 1996), which generates the 5-hydroxy derivative, 5-hydroxychlorsulfuron (CSOH). The latter is then rapidly glucosylated to yield the Glc conjugate of CSOH (CSG; Fig. 1A). However, although it is established that the glucoside of the structurally related herbicide hydroxyprimisulfuron (Fig. 1, structure C) is transported into isolated vacuoles in an ATP-dependent manner (Klein et al., 1996), nothing is known about CSG transport. Moreover, although it is assumed that the sugar ligand enables vacuolar uptake of sulfonylurea herbicides, nowhere has the transport competency of the parent aglycone been tested directly. One of our initial objectives, therefore, was to determine if glucosylation is an essential prerequisite for energy-dependent uptake of CS using vacuolar membrane vesicles purified from red beet storage root.

View larger version (20K): [in this window] [in a new window]   Figure 1.   Chemical structures of three herbicide conjugates that undergo vacuolar uptake. A, CSG, a Glc conjugate of 5-hydroxychlorsulfuron. B, CE-GS, a glutathione conjugate of chlorimuron-ethyl. C, HPSG, a Glc conjugate of hydroxyprimisulfuron.

The results shown in Figure 2A and Table I demonstrate that conjugate formation is required for net uptake. Incubation of vacuolar membrane vesicles in uptake medium containing 50 µM CS, CSOH, or CSG, plus or minus ATP, and quantitation of net uptake by LC-MS, clearly establishes that only the Glc conjugate undergoes ATP-dependent transport (Table I). CSG accumulation is linear for approximately 20 min under energized conditions, whereas little or no transport is observed when ATP is omitted (Fig. 2A). In contrast, the rates of uptake of CS and CSOH are 18- to 116-fold lower than those of CSG, and the same with or without ATP (Table I).

View larger version (21K): [in this window] [in a new window]   Figure 2.   ATP-dependent uptake of two structurally related herbicides by vacuolar membrane vesicles. Red beet vacuolar membrane vesicles were incubated at 25°C in the standard uptake medium containing 50 µM CSG (A) or 50 µM CE-GS (B). ATP was added or omitted as indicated. Uptake was terminated by rapid filtration and the amount of compound taken up by the vesicles was determined by LC-MS. The results of a representative experiment are shown. Each data point is the mean of duplicate measurements, all of which were within 15% of each other. The insets show the original LC-MS traces from the 60-min time points for vesicles incubated in the presence (top) or absence (bottom) of ATP. The traces are slightly offset from each other to emphasize the high signal-to-noise ratio for energized uptake.

ATP-dependent vesicular CSG uptake occurs against a steep concentration gradient. Estimates of the internal volume of vacuolar membrane vesicles prepared in the same manner as those employed in the present study yield a value of 10 µL mg¹ membrane protein (Li et al., 1995, and refs. therein). Assuming a 1:1 mixture of right-side-out and inside-out vesicles, the accumulation of 15 nmol mg¹ protein after 60 min from a medium containing 50 µM CSG (Fig. 2A) amounts to an accumulation ratio of 60. This value is similar to that computed for S-(2,4-dinitrophenyl) glutathione (DNP-GS), a model MRP-type ABC transporter substrate, for the same membrane preparation (Li et al., 1995).

The ion detected after the uptake and LC-MS analysis of CSG has precisely the m/z value predicted for the (M + H)⁺ ion of CSG, demonstrating that this compound is not modified before, during, or after uptake by this membrane vesicle preparation. Because the signal-to-noise ratio for vesicular CSG uptake is large, the amount taken up is readily estimated using appropriate calibration standards (Fig. 2A). Moreover, direct, side-by-side comparisons of net [¹⁴C]CSG uptake, estimated by liquid scintillation counting, versus net uptake of unlabeled CSG, estimated by LC-MS, confirm that measurement by either technique yields similar results. Thus, a 30-min incubation of an equivalent but different vesicle preparation from that used for the experiment in Figure 2A with 50 µM [¹⁴C]CSG or unlabeled CSG in the presence of ATP yields estimates of net uptake of 16.4 ± 0.9 and 13.7 ± 1.4 nmol mg¹ protein by the radioactive assay and LC-MS, respectively. Because of its sensitivity and accuracy, all subsequent measurements of transport were performed by LC-MS.

When measured in the 12.5 to 400 µM concentration range, the initial rates of ATP-dependent CSG uptake approximate Michaelis-Menten kinetics to yield K_m and V_max values of 46.3 ± 6.1 µM and 0.9 ± 0.1 nmol mg min¹, respectively, as determined from three independent experiments similar to the one shown (Fig. 3A).

View larger version (23K): [in this window] [in a new window]   Figure 3.   Kinetics of ATP-energized uptake of CSG and CE-GS by vacuolar membrane vesicles. Uptake of CSG (A) or CE-GS (B) was measured at 25°C in standard uptake medium in the presence or absence of ATP. Uptake was allowed to proceed for 20 min (CSG) or 15 min (CE-GS) before rapid filtration and determination of the amount of the compound taken up by LC-MS. The values shown are the means of duplicate measurements minus the amount of compound that was taken up in the absence of ATP. Insets, Eadie-Hofstee (v/[S] versus v) plots, where v = velocity of uptake and [S] = substrate concentration.

Glutathionation of CE Is a Prerequisite for Transport

Because it is known that several herbicides, chloroacetanilides primarily, are targeted to the vacuole by glutathionation (Hopp and Seitz, 1987; Martinoia et al., 2001), it was of interest to know if the same would hold true for CS and, if so, whether the corresponding glutathione conjugate is transported by the same carrier as CSG. To examine this directly, we attempted to conjugate glutathione to CSOH but were unsuccessful. It was for this reason that we focused our attention on another sulfonylurea herbicide, CE, the structure of which is closely related to CS (Fig. 1). CE, unlike CSOH, is amenable to electrophilic displacement and substitution with glutathione at the 4-chloro position to yield CE-GS.

By analogy with CS and its glucoside, vacuolar uptake of CE is ATP dependent, occurs against a steep concentration gradient, and is contingent on conjugation. The omission of ATP from the uptake medium abolishes CE-GS accumulation (Fig. 2B), whereas unconjugated CE is not accumulated regardless of the conditions of energization (Table I). Working from the same assumptions as those described for CSG, it can be estimated that the intravesicular concentration of CE-GS exceeds the extravesicular concentration by a factor of 900 after 60 min (Fig. 2B).

The maximal rate of vesicular CE-GS accumulation is very high compared with CSG. Although the apparent K_m values for these conjugates differ by only a factor of two, the maximum rate of energized uptake of CE-GS (6.4 nmol mg min¹) exceeds that of CSG by at least 7-fold (Fig. 3B).

Different Carriers Mediate CSG and CE-GS Uptake

Superficially, the transport characteristics of CSG and CE-GS are remarkably similar. The key question, therefore, is whether these compounds are transported by the same or different carriers because one is a Glc conjugate and the other a glutathione conjugate. Given that previous investigations have shown that several glutathionated xenobiotics (Rea et al., 1998; Martinoia et al., 2001) and a few endogenous plant secondary metabolites (Li et al., 1997; Walczak and Dean, 2000) are transported into the vacuole by MRP-type ABC transporters that are directly energized by ATP and are not energetically coupled to the H⁺ gradient, it was anticipated that CE-GS uptake would be mediated by a similar carrier. One of the hallmarks of these so-called "GS-X pumps" is that they are potently inhibited by vanadate.

As shown in Figure 4, vacuolar uptake of CE-GS has all the properties expected of a process catalyzed by an MRP transporter. Net uptake of CE-GS, as is the case for the model MRP substrate DNP-GS, is completely abolished by the addition of vanadate, whereas the macrolide antibiotic bafilomycin A₁, which inhibits the V-ATPase on this membrane and abrogates intravesicular acidification, inhibits uptake only moderately. Likewise, CE-GS and DNP-GS uptake are only partially inhibited by an uncoupling mixture of the K⁺-specific ionophores nigericin and valinomycin (Fig. 4). A 5 µM concentration of the monovalent cation-selective ionophore gramicidin-D has a similar effect (data not shown). Nigericin dissipates pH with little effect on , valinomycin has the opposite effect, and gramicidin-D dissipates both components. Evidently, both of these glutathione conjugates interact with a transporter whose activity is modulated by but not obligatorily dependent on the transmembrane H⁺ gradient, a property that this transporter shares with heterologously expressed AtMRP2 (Lu et al., 1998) and the MRP-like functionality responsible for the uptake of glucuronides into isolated rye (Secale cereale) and barley vacuoles (Klein et al., 2000). However, in the system described here, as is the case for heterologously expressed AtMRPs 1 and 2 (Lu et al., 1997, 1998), vanadate totally abolishes uptake even when the H⁺ gradient is maximal.

View larger version (27K): [in this window] [in a new window]   Figure 4.   Effects of inhibitors and ionophores on the uptake of CE-GS, DNP-GS, and CSG. The conditions for ATP-dependent uptake were as in Figure 2 except uptake was terminated after 30 min. All of the conjugates were present at a concentration of 50 µM. Where indicated, bafilomycin A1, nigericin, valinomycin, and vanadate were added at concentrations of 0.4, 2, 2, and 200 µM, respectively. The bafilomycin A1 stock solution was made up in dimethyl sulfoxide; the valinomycin and nigericin stock solutions were made up in ethanol. Control experiments confirmed that neither dimethyl sulfoxide nor ethanol at the concentrations at which they were added to the uptake medium (0.004% [v/v] and 0.04% [v/v], respectively) affected uptake. Uptake is expressed as a percentage of the uptake by controls to which no inhibitors or ionophores had been added. The control activities were 52.5, 7.7, and 13.4 nmol mg1 protein for CE-GS, DNP-GS, and CSG, respectively. The values shown are the means of duplicate measurements from a representative experiment.

When CSG is the transport substrate an entirely different pattern of inhibition is observed. Vanadate has no effect on the net uptake of CSG but agents, such as bafilomycin A₁, nigericin plus valinomycin, or gramicidin-D (data not shown) decrease net uptake by more than 95% (Fig. 4). The implication is that although uptake of CE-GS is catalyzed by an MRP-type ABC transporter, CSG accumulation is catalyzed by an H⁺-coupled carrier, possibly an H⁺-antiporter, whose activity is obligatorily dependent on a transmembrane H⁺ gradient.

Uptake of Glucosylated p-Hydroxycinnamic Acid

It has been reported that red beet vacuolar membrane vesicles, purified by the same procedure (Li et al., 1995), accumulate a glutathione conjugate of pHCA, a key intermediate in the phenylpropanoid pathway, and that conjugation is necessary for uptake (Walczak and Dean, 2000). In light of these findings, the likely involvement of an MRP-type transporter in uptake of the glutathione conjugate, and the fact that plants primarily glucosylate pHCA, we synthesized the phenolic glucoside of pHCA (pHCAG) to see if it also undergoes transport and, if so, whether it is subject to MRP-mediated or H⁺-coupled transport.

The results are unequivocal: pHCAG undergoes H⁺-coupled transport. Thus, pHCAG, but not its unconjugated precursor (Table I), is transported at high rates in the presence of ATP and net uptake is almost totally abolished by bafilomycin A₁ but little effected by vanadate (Fig. 5). The apparent K_m and V_max values for ATP-dependent, vanadate-insensitive pHCAG uptake are 62.0 µM and 0.9 nmol mg min¹, respectively (Fig. 5, inset), values similar to those obtained with the glucosylated herbicide, CSG (Fig. 3A). Virtually identical results were obtained with the phenolic glucoside of pHBA, which is another naturally occurring plant phenylpropanoid (data not shown).

View larger version (23K): [in this window] [in a new window]   Figure 5.   Uptake of the phenolic glucoside of pHCA, from media containing () or lacking ATP () plus no inhibitors (), 200 µM vanadate (), or 0.4 µM bafilomycin A1 (). The conditions for ATP-dependent uptake were as in Figure 2, but the transport substrate was 50 µM pHCAG. Inset, Eadie-Hofstee (v versus v/[S]) plot of the concentration dependence of ATP-energized pHCAG uptake. Uptake was measured at 25°C in standard uptake medium containing 50 µM pHCAG for the times indicated or for 20 min in uptake medium containing 12.5 to 400 µM pHCAG. pHCAG uptake was determined by LC-MS.

In no case in which energy-dependent accumulation was demonstrable was there any evidence of covalent modification of the compound concerned either before, during, or after transport. The chromatographic and mass spectral properties of all of the Glc and glutathione conjugates examined were the same before and after vesicular uptake. Enzymes capable of degrading or modifying these conjugates do not appear to be operative in this in vitro system.

The results of these studies indicate that it is the identity of the ligand, Glc, or glutathione to which the parent compound is attached that primarily determines which carrier the conjugate interacts with. A corollary of this inference is that xenobiotics and endogenous plant secondary metabolites can share common carriers if they are conjugated to the same ligand, even if they bear little structural resemblance to each other.

The finding that glucosides compete with each other for transport but not with glutathione conjugates, and vice versa, is consistent with this idea. CSG uptake is strongly inhibited by pHCAG (Fig. 6A) and the pHBA phenolic glucoside (data not shown), but is relatively insensitive to inhibition by CE-GS and DNP-GS (Fig. 6A). The converse is also true. Thus, DNP-GS, the only glutathione conjugate tested, was the only compound that significantly inhibited uptake of CE-GS (Fig. 6B). Although not shown, whenever an inhibitory effect of a compound on the accumulation of another compound was observed the effect was reciprocal. It is noteworthy that diminished CSG accumulation was also observed in the presence of p-nitrophenyl (pNP) -D-glucopyranoside [pNP()G]. In contrast, no inhibition of CSG uptake was observed with the corresponding -anomer pNP -D-glucopyranoside, pNP GlcUA, or the unconjugated parent compound, pNP (Fig. 6A).

View larger version (47K): [in this window] [in a new window]   Figure 6.   Effects of glutathione conjugates on Glc conjugate uptake and vice versa by vacuolar membrane vesicles. ATP-dependent uptake of 25 µM CSG (A) or CE-GS (B) was measured in standard uptake medium containing the other compounds indicated at concentrations of 400 µM. Reactions were terminated after 60 min and the amount of CSG or CE-GS taken up by membrane vesicles was determined by LC-MS. Uptake is expressed as a percentage of ATP-energized controls to which no other compounds were added. The compounds added were pHCAG, the phenolic glucoside of pHCA (-anomer); pNP()G, pNP -D-glucopyranoside; and pNP()GA, pNP -D-GlcUA. Values shown are means ± SE for triplicate measurements.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

These investigations demonstrate that vacuolar accumulation of sulfonylurea herbicides and phenylpropanoids is contingent on phase II conjugation and that the particular transport pathway taken by these compounds is determined by the ligand to which they are conjugated. Glc conjugates of the model sulfonylurea herbicide CS and of the phenylpropanoid pHCA are subject to H⁺-coupled transport by a vanadate-insensitive transporter, whereas the corresponding glutathione conjugates of CE and pHCA are taken up by an ATP-energized, vanadate-inhibitable transporter.

The carriers for these glucosides and glutathione conjugates have the properties of H⁺-antiporters and MRP-type ABC transporters, respectively. Net uptake of CSG and pHCAG is abolished by agents that prevent or dissipate V-ATPase-catalyzed intravesicular acidification but is insensitive to vanadate. In contrast, net uptake of CE-GS, as shown here, and of glutathionated pHCA, as shown by Walczak and Dean (2000), is abolished by vanadate but only moderately sensitive to protonophores and V-ATPase inhibitors. Similar properties have been reported for the uptake of DNP-GS by red beet vacuolar membrane vesicles (Li et al., 1995), and the glutathione conjugate of the chloroacetanilide herbicide, metolachlor, by vacuolar membrane-enriched vesicles purified from yeast (Saccharomyces cerevisiae) heterologously expressing the Arabidopsis GS-X pumps, AtMRPs 1 or 2 (Lu et al., 1997, 1998). The implication is clear. Contrary to earlier speculations by us and others (Martinoia et al., 2001), it is evident from this study that not only glucosides of endogenous metabolites but also those of xenobiotics are amenable to vacuolar uptake by H⁺-antiport. It is also clear that the carrier principally responsible for CSG accumulation in red beet vacuolar membrane vesicles has a much higher substrate affinity and transport capacity than the carrier that mediates HPSG uptake into isolated barley vacuoles (Klein et al., 1996). As the authors of the latter study point out, the rate of uptake of HPSG into barley vacuoles is too low to demonstrate accumulation against a concentration gradient and the carrier(s) responsible show no indication of substrate saturation even at concentrations as high as 250 µM (Klein et al., 1996). Assignment of the transport pathway of a particular glucoside on the basis of whether the substrate is of endogenous or exogenous origin is not warranted even though a pattern of this type is found for some subsets of compounds (Frangne et al., 2002).

In the context of glutathione conjugate transport, it is interesting to note that although the detoxification of sulfonylurea herbicides has generally been assumed to depend on glucosylation, this is not always the case. Soybeans (Glycine max) are naturally resistant to CE because they metabolize it rapidlythis is why this agent has been used for selective weed control in this crop (Brown and Neighbors, 1987). However, the principal metabolite in soybean is an adduct formed by displacement of the pyrimidinyl chloride substituent by the sulfhydryl group of homoglutathione, an analog of glutathione found in legumes, in which the C-terminal Gly is substituted by -Ala (Brown and Neighbors, 1987).

Further studies are required to determine if the Glc conjugates and glutathione conjugates examined are transported by the same or different transporters within their respective categories, but it is apparent that the two categories interact only weakly if at all with the other's substrates. All of the Glc conjugates, with one exception, compete with each other but not with glutathione conjugates for uptake, and vice versa. The sole exception is the one -D-glucoside tested, pNP -D-glucopyranoside, which unlike its -anomeric equivalent, pNP()G, does not compete with CSG (Fig. 6A) or the pHCA or pHBA phenolic glucosides (data not shown) for H⁺-coupled uptake. The full significance of this finding remains to be determined, but it is consistent with the notion that H⁺-coupled vacuolar carriers for Glc conjugates can only transport 1-O--D-glucosides, which are the expected products of most plant UDP-glucosyltransferases. Our results further suggest that the Glc moiety must be attached to an aromatic OH group, not a carboxyl group. Only the pHBA phenolic glucoside (Fig. 7, structure I) accumulates in vacuolar membrane vesicles under energized conditions, whereas the corresponding Glc ester does not (Fig. 7, structure II), even though both compounds contain a 1-O--D linkage. Similar results were obtained with the two pHCA Glc conjugates: Only the phenolic glucoside was taken up by membrane vesicles (Fig. 5). More importantly, in competition experiments similar to those shown in Figure 6, the pHCA and pHBA Glc esters had no effect on accumulation of CSG or uptake of the pHBA and pHCA phenolic glucosides. These observations are somewhat surprising because many naturally occurring aromatic Glc esters are known to accumulate in plant vacuoles (Wink, 1997; Martinoia et al., 2001). The reason why the uptake of these compounds by isolated red beet vacuolar membrane vesicles cannot be measured under these conditions, even though appropriate carriers must exist, remains to be determined.

View larger version (11K): [in this window] [in a new window]   Figure 7.   Four other glucosides whose energy-dependent accumulation by vacuolar membrane vesicles was examined. I, pHBA phenolic glucoside. II, pHBA Glc ester. III, pNP glucoside. IV, Methylparaben glucoside. All four compounds are 1-O--D-glucopyranosides.

Another condition for the net uptake of phenolic glucosides that likely applies is that the conjugate must carry a negative charge. Although not apparent from its chemical structure, even CSG (Fig. 1) is negatively charged at the pH of the in vitro transport assay. The pKa of the acidic proton on the sulfonamide nitrogen is 3.6. To further test this hypothesis, methylparaben glucoside (Fig. 7, structure IV) and pNP()G (Fig. 7, structure III), uncharged analogs of phenolic glucosides, were synthesized and tested for vacuolar uptake. As would be expected if negative charge were a determinant of transportability, neither 1-O--D glucoside was taken up by isolated membrane vesicles in the present of ATP (data not shown), although both compounds partially inhibited the ATP-dependent accumulation of CSG and pHCAG, suggesting that they are able to some extent to interact with the same Glc conjugate carrier.

It is anticipated that LC-MS will find wide application in other areas of in vitro transport measurement. The exquisite sensitivity of this technique for certain organic compounds obviates the need for synthesis and/or purification of expensive radioactive transport substrates. Moreover, the fact that only 1% of the material extracted from a single membrane filter was typically analyzed, the equivalent of 70 to 180 ng of membrane vesicles, shows that total reaction volumes of only a few microliters may be sufficient for certain types of transport assays. This would increase the number of experiments that could be performed with a single preparation of membrane vesicles. Finally, the application of LC-MS, a mass-based technique, to in vitro transport circumvents a number of potential artifacts that might otherwise go undetected in assays that only rely on the recovery of radioactivity, including intra- and extravesicular modification of the transport substrate.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Materials

[¹⁴C]UDP-Glc (286 mCi mmol¹) was purchased from American Radiolabeled Chemicals, Inc. (St. Louis). Equine liver glutathione S-transferase, ATP, UDP-Glc, methylparaben, the - and -anomers of pNP()G, and pNP -D-glucuronide were purchased from Sigma (St. Louis). CS, CSOH, and CE (>95% [w/v] purity) were synthesized by DuPont Crop Protection Products (Wilmington, DE).

Preparation of Vacuolar Membrane Vesicles

Vacuolar membrane vesicles were purified from fresh red beet (Beta vulgaris) storage roots by a combination of differential and density gradient ultracentrifugation as described (Rea and Turner, 1990; Li et al., 1995). The final membrane pellets were resuspended in 5 mM Tris-MES (pH 7.5), 1 mM Tris-EDTA, 0.5 mM dithiothreitol, 100 µg mL¹ butylated hydroxytoluene, and 10% (v/v) glycerol at a protein concentration of 0.5 to 2.0 mg mL¹, frozen in liquid nitrogen, and stored at 80°C. Protein was determined by a modification of the method of Lowry et al. (Peterson, 1977).

Cloning, Expression, and Purification of IS5a

IS5a, a broad-specificity UDP-glucosyltransferase from tobacco (Nicotiana tabacum; Horvath and Chua, 1996; Fraissinet-Tachet et al., 1998), was employed for the synthesis of most of the glucosylated compounds used in these studies. The coding region of IS5a (GenBank accession no. NTU32644) was amplified from a tobacco cDNA library (catalogue no. 936002, Stratagene, La Jolla, CA) using two PCR primers. One primer (5'-CTA CTC ATT Tca tat gGG TCA GCT CCA TAT TTT C-3') introduced an NdeI site at the initiation codon, whereas the other (5'-CAT CTT ACT gga tcc TAA TGA CCA GTG GAA CTA TAT G-3') introduced a BamHI site just after the stop codon. The PCR fragment was cut with NdeI and BamHI, and ligated into double-digested pET-24a(+) vector (Novagen, Madison, WI). Escherichia coli BL21(DE3) was transformed with the ligation mixture and selected on Luria-Bertani medium containing kanamycin (50 µg mL¹).

For protein production, the recombinant strain was grown at 17°C in Luria-Bertani medium containing kanamycin (50 µg mL¹) to an A_{600 nm} of approximately 0.6 units before induction with isopropyl-1-thio--D-galactopyranoside (0.4 mM). After growth for a further 24 h, the cells were harvested by centrifugation. Cell-free extracts were prepared as described by Dickson et al. (1994) and stored at 80°C. Recombinant protein was purified to homogeneity by a combination of anion-exchange, hydroxylapatite, and gel filtration chromatography (a detailed protocol is available upon request from Paul V. Viitanen). Purified recombinant IS5a was stored at 80°C in 50 mM Tris-HCl (pH 7.7), 0.3 M NaCl, 1 mM dithiothreitol, and 5% (v/v) glycerol at 8 to 10 mg protein mL¹.

Synthesis and Purification of Glutathione Conjugates

CE-GS was prepared by displacement of the 4-chloro substituent of CE and characterized as described (Brown and Neighbors, 1987). Solid CE (2 mg) was added to a 1-mL solution of 0.25 M glutathione in 0.5 M Tris-HCl buffer (pH 8.5), and 5 mg of equine liver glutathione S-transferase was added. The mixture was stirred at 25°C for 20 h, after which time more than 93% of the CE has been converted to CE-GS as determined by HPLC and UV detection at 280 nm. The product was purified by HPLC using a Zorbax ODS column (6.2 mm × 8 cm, 3 µ, Mac-Mod Analytical, Chadds Ford, PA) and its identity was confirmed by LC-MS. The Zorbax column was eluted at a flow rate of 1.0 mL min¹ with a linear gradient (10 min) of 5% to 90% (v/v) acetonitrile in 0.1% (v/v) formic acid. Peak fractions corresponding to CE-GS were pooled from multiple runs, quantitated spectrophotometrically, and dried to a powder by lyophilization. DNP-GS was synthesized and purified as described (Li et al., 1995).

Synthesis and Purification of Glc Conjugates

[¹⁴C]CSG was synthesized enzymatically from unlabeled CSOH using [¹⁴C]UDP-Glc and purified recombinant IS5a. The 50-µL reactions contained 50 mM Tris-HCl (pH 7.5), 1 mM MgCl₂, 5 mM dithiothreitol, 2 mM CSOH, 1.75 mM [¹⁴C]UDP-Glc (286 mCi mmol¹), and 4 µM IS5a protomer. After incubation for 18 h in the dark at 25°C, [¹⁴C]CSG was purified by reverse-phase HPLC on a C₁₈ column (VYDAC 218TP54, The Nest Group, Inc., Southborough, MA). The column was eluted at a flow rate of 1.0 mL min¹ with a linear gradient (20 min) of 10% to 50% (v/v) methanol in 0.1% (v/v) formic acid. Absorbance was monitored at 254 nm and the peak corresponding to CSG was collected. After adjusting its pH to 8 with ammonium bicarbonate, the preparation was stored at 80°C. For the synthesis of unlabeled CSG, the 5-mL reactions contained 0.1 M Tris-HCl (pH 8.5), 2 mM MgCl₂, 1 mM dithiothreitol, 5 mM CSOH, 50 mM UDP-Glc, and 5 µM IS5a. Approximately 60% of the CSOH was converted to CSG after incubation for 48 h at 25°C. Unlabeled CSG was purified, lyophilized, and stored as described above, after confirming its identity by LC-MS. For radioactive transport assays, an aliquot of [¹⁴C]CSG was dried under vacuum and mixed with an aqueous solution of unlabeled CSG to yield the desired specific activity (5-10 mCi mmol¹).

Methylparaben glucoside was synthesized in a similar manner using the components described above for unlabeled CSG, except methylparaben was substituted for CSOH and 5 mM UDP-Glc was used. After incubation for 18 h, the enzyme was removed by ultrafiltration using a Centriprep-10 (Amicon, Beverly, MA), and the volume of the filtrate was decreased under vacuum at 25°C. The product was purified by HPLC, lyophilized, and stored as described for CSG, after confirming its identity by electrospray LC-MS.

The phenolic glucosides and Glc esters of pHCA and pHBA used in these studies were synthesized, purified, and structurally characterized by Frank Conway and Andy Travis (DuPont Crop Protection Products).

Vacuolar Membrane Vesicle Transport Assays

The measurements of uptake by vacuolar membrane vesicles were performed individually in glass test tubes (12 × 75 mm) at 25°C as previously described (Rea and Turner, 1990; Li et al., 1995). The 200-µL reaction mixtures contained 25 mM Tris-MES (pH 8.0), 0.4 M sorbitol, 50 mM KCl, 3 mM MgSO₄, 0.1% (w/v) bovine serum albumin, 10 mM creatine phosphate, 3 mM ATP, 16 units mL¹ creatine kinase, and the indicated concentrations of labeled or unlabeled transport substrate. Both ATP and creatine kinase were omitted from the nonenergized controls. Assays were initiated by the addition of membrane vesicles (7-18 µg of protein) and brief agitation on a vortex mixer. At the times indicated, reactions were terminated with 1.0 mL of ice-cold 25 mM Tris-MES (pH 8.0) and 0.4 M sorbitol ("wash solution") and subjected to vacuum filtration through prewetted GV Durapore polyvinylidene difluoride membrane filters (0.2-µm pore diameter, Millipore, Bedford, MA). For radioactive transport assays, the filters were washed twice with 1.0 mL of wash solution and radioactivity was determined by liquid scintillation counting in 4 mL of Formula-989 (NEN Life Science Products, Boston). For transport assays with unlabeled compounds, the procedure was the same except that the washed filters were transferred to 20-mL glass vials containing 1.0 mL of 50% (v/v) acetonitrile (Li et al., 1995). The vials were capped and the filters were extracted for 1 h at room temperature in an orbital shaker. Typically, 5-µL aliquots of the eluate were analyzed by LC-MS.

LC-MS Determinations of Substrate Uptake

The acetonitrile extracts from the membrane filters that were used for transport assays were analyzed on an LC-MS system consisting of a model 1050 HPLC (Hewlett-Packard, Palo Alto, CA) equipped with a photodiode array detector and a Micromass Platform II single quadrupole mass spectrometer (Micromass, Inc., Beverly, MA). The compounds were separated on a Zorbax Rx-C8 (2.1 × 150 mm, 5 µ) or MacMod Hydrobond PS-C8 (3 × 150 mm, 5 µ) reverse-phase column. The columns were eluted with a linear gradient (7 min) of 5% to 95% (v/v) acetonitrile in 0.1% (v/v) formic acid followed by a 2-min holding period, and absorbance was monitored between 200 and 400 nm with a diode array detector. Mass spectrometric detection was by selected ion monitoring of the intensity of the mass ion for each of the compounds described, using either electrospray or atmospheric pressure chemical ionization. The following cone voltages were used for electrospray ionization: 20 V for CSG and p-nitrophenylglucoside, +30 V for CS, CSOH, and DNP-GS, and +40 V for CE and CEGS. Atmospheric pressure chemical ionization at a cone voltage of 10 V was used to detect and quantitate pHCA and its phenolic glucoside, and 30 V was used for methylparaben. The compounds were quantitated by peak height using standards of known concentration. The values reported for vacuolar transport have been corrected for nonspecific absorption to the membrane filter, normalized to protein, and are expressed as nanomoles of compound taken up per milligram of membrane vesicle protein.