Cross-talk between the cytosolic mevalonate and the plastidial methylerythritol phosphate pathways in tobacco bright yellow-2 cells.

In plants, two pathways are utilized for the synthesis of isopentenyl diphosphate, the universal precursor for isoprenoid biosynthesis. The key enzyme of the cytoplasmic mevalonic acid (MVA) pathway is 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGR). Treatment of Tobacco Bright Yellow-2 (TBY-2) cells by the HMGR-specific inhibitor mevinolin led to growth reduction and induction of apparent HMGR activity, in parallel to an increase in protein representing two HMGR isozymes. Maximum induction was observed at 24 h. 1-Deoxy-d-xylulose (DX), the dephosphorylated first precursor of the plastidial 2-C-methyl-d-erythritol 4-phosphate (MEP) pathway, complemented growth inhibition by mevinolin in the low millimolar concentration range. Furthermore, DX partially re-established feedback repression of mevinolin-induced HMGR activity. Incorporation studies with [1,1,1,4-2H4]DX showed that sterols, normally derived from MVA, in the presence of mevinolin are synthesized via the MEP pathway. Fosmidomycin, an inhibitor of 1-deoxy-d-xylulose-5-phosphate reductoisomerase, the second enzyme of the MEP pathway, was utilized to study the reverse complementation. Growth inhibition by fosmidomycin of TBY-2 cells could be partially overcome by MVA. Chemical complementation was further substantiated by incorporation of [2-13C]MVA into plastoquinone, representative of plastidial isoprenoids. Best rates of incorporation of exogenous stably labeled precursors were observed in the presence of both inhibitors, thereby avoiding internal isotope dilution.

With over 30,000 entities being structurally identified, most of them of plant origin, isoprenoids represent the largest family of natural compounds (1,2). They function in respiration, signal transduction, cell division, membrane architecture, photosynthesis, and growth regulation (3). Furthermore, they also play an important role in the exchange of signals between plants and their environment (4) or in defense against pathogens. These structurally diverse compounds all originate from a branched five-carbon unit, the so-called active isoprene, iso-pentenyl diphosphate (IPP), 1 and its isomer dimethylallyl diphosphate (DMAPP).
In higher plants, it was revealed that two pathways are involved in the biosynthesis of the active isoprene unit. Over the course of evolution, plants have maintained the well known eukaryotic mevalonic acid (MVA) pathway (5) in the cytosol, also called the classical pathway, and acquired the more recently discovered prokaryotic 2-C-methyl-D-erythritol 4-phosphate (MEP) or alternative pathway (6 -8) from the endosymbiotic ancestor of plastids. Under normal physiological conditions, cytoplasmic isoprenoids, i.e. sterols, or the side chain of mitochondrial ubiquinone are synthesized from MVAderived IPP (9), whereas plastidial isoprenoids take their origin from both simultaneously formed DMAPP and IPP (10), synthesized via the MEP pathway (8,11,12). In a series of earlier experiments (see Refs. 3,13, and 14 for review of the literature), the key enzyme of the classical mevalonate pathway in plants was identified as 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase (HMGR, EC 1.1.1.34). It catalyzes the formation of MVA by two successive reductions of HMG-CoA, using two entities of NADPH as cofactor. In contrast to the situation in animals, HMGR in plants is encoded by gene families (13)(14)(15). These HMGR isoforms are differentially expressed, depending on physiological conditions (14, 15), and are thought to be linked to specific channels of the cytoplasmic pathway leading to different end products (16). In animal cells, HMGR is subject to multiple control mechanisms at different levels (17). Overexpression of HMGR cDNA in tobacco (Nicotiana tabacum L.) plants increased apparent enzyme activity and the amount of sterols and sterol pathway intermediates in the form of fatty acyl esters but did not affect the accumulation of carotenoids or chlorophylls (18 -19). This observation further confirmed the hypothesis (20) that in plants, at least for phytosterol biosynthesis, HMGR plays a regulatory role similar to that in animals. The metabolic importance of this enzyme is possibly underlined by the existence of natural inhibitors identified in an array of ascomycetes occurring in the rhizosphere. Among those, mevinolin (21), also referred to as lovastatin, has been revealed as a highly efficient plant growth inhibitor (22) and as a valuable tool for the study of isoprenoid biosynthesis in intact plants and in plant cells (23). Fosmidomycin (24) is a more recently identified inhibitor of the first committed en-zyme of the plastidial MEP pathway, the 1-deoxy-D-xylulose-5phosphate reductoisomerase, or MEP synthase (25). With such molecular probes at hand, it is possible to deregulate pathways artificially and to deplete cells gradually from essential metabolites.
In this study, we take advantage of using pale-yellow tobacco (N. tabacum L.) Bright Yellow 2 (TBY-2) cells, which represent a highly suitable system toward studying regulatory interactions between isoprenoid biosynthesis and fundamental processes like cell division and growth (26 -28). They also provide an excellent system for incorporation studies (9,10,29), due to the extremely high productivity of TBY-2 cell suspensions, based on a cell division cycle of only about 13 h (30). This guarantees high metabolic flux rates to end products of biosynthetic pathways, i.e. sterols, but also of putatively rate-limiting enzymes like HMGR, which are otherwise difficult to measure. In this contribution we demonstrate by various approaches the usefulness of TBY-2 cells and inhibitors in elucidating how and to what extent the cytoplasmic MVA and the plastidial MEP pathways communicate under various conditions.

EXPERIMENTAL PROCEDURES
Materials- [2-13 C]Mevalonolactone (99% isotope abundance) was purchased from Aldrich. Non-labeled 1-deoxy-D-xylulose (DX) was synthesized from (ϩ)-2,3-O-benzylidene-D-threitol. 2 [1,1,1,4-2 H 4 ]DX was synthesized as described previously (10) for the synthesis of [4-2 H]DX. Deuterium-labeled methylmagnesium chloride (99% isotope abundance) was utilized in the place of the reagent of natural abundance. According to the integration of the 4-H and 3-H signals of the 1 H NMR spectrum, the expected main isotopomer (88%) with deuterium labeling at C-4 was accompanied by a minor isotopomer (12%) with deuterium labeling at C-3 derived from deuterium migration during the last step of the synthesis, the hydrogenolysis of the benzylidene and benzyl protecting groups (10 ); 4.12 (0.88H, d, J 5a,5b ϭ 11.5 Hz, 5-H a of A); 4.13 (0.12H, dd, J 4,5a ϭ 6.4 Hz, J 5a,5b ϭ 11.5 Hz, 5-H a of AЈ); 4.28 (0.88H, d, J 5a,5b ϭ 11.5 Hz, 5-H b of A); 4.29 (0.12H, dd, J 4,5b ϭ 5.7 Hz, J 5a,5b ϭ 11.5 Hz, 5-H b of AЈ); 5.23 (0.88H, s, 3-H of A); 5.57 (0.12H, dd, J 4,5a ϭ 6.4 Hz, J 4,5b ϭ 5.7 Hz, 4-H b of AЈ). For 13 (26). Inhibitor solutions were filter-sterilized before addition to the growth medium. Results were compared with cultures containing the amount of solvents, if any, but in the absence of compound (control cells). After suction filtration of aliquots, cell growth was evaluated by determination of the packed cell volume of cells in 1 ml of cell suspension. The corresponding fresh weight of this packed cell volume was estimated. For feeding experiments, cell cultures (80 ml in 250-ml Erlenmeyer flasks) were kept in the dark at 26°C and shaken at 174 rpm. In the preceding test series, growth was determined as increase in biomass (fresh weight), in order to optimize the composition of the culture medium and to obtain sufficient amounts of labeled isoprenoids for the analyses. Concentrations of inhibitors were chosen such as to block cell division; a minimum rate of cell growth was found optimum with 2% (w/v) sucrose, 5 M mevinolin, and 1 mM [1,1,1,4-2 H 4 ]DX for examining the cross-talk between plastids and the cytosol. For the study on MVA complementation of the plastidial isoprenoid pathway, optimal conditions required 2% (w/v) sucrose, 20 M fosmidomycin, and/or 5 M mevinolin and 2 mM [2-13 C]MVA. One Erlenmeyer flask was used for each incorporation experiment, including controls and inhibitors alone or in combination. After 8 days, cells were harvested by vacuum filtration on a sintered glass funnel for further chemical analysis.
Preparation of Microsomes and Measurement of HMGR Activity-Frozen TBY-2 cells were powdered in a mortar in the presence of liquid nitrogen. The powder was suspended in 12.5 ml g Ϫ1 fresh weight of a 4°C cold phosphate buffer system A (0.2 M K x PO 4 , pH 7.5, 0.35 M sorbitol, 10 mM Na 2 EDTA, and 5 mM MgCl 2 ), to which 20 mM DTT and 4 g of 100 ml Ϫ1 insoluble polyvinylpyrrolidone (Sigma) were freshly added. The homogenate was filtered through nylon gauze (50 m) and centrifuged at 1500 ϫ g for 5 min at 4°C (rotor JA-20, RC-5 superspeed centrifuge, Beckman Instruments). The pellet containing cell debris and polyvinylpyrrolidone particles was removed, and the supernatant was again centrifuged at 16,000 ϫ g for 40 min at 4°C. The following supernatant was centrifuged at 105,000 ϫ g (at 4°C for 1 h). The resulting pellet (P105,000), referred to as microsomal fraction, had to be washed free from inhibitors (e.g. mevinolin). Therefore, it was resuspended in 20 ml of the same buffer system and centrifuged again at 105,000 ϫ g as described above. Buffer-washed microsomes were finally re-dissolved in buffer system A and stored at Ϫ80°C. Protein content was quantified by a modified Lowry protocol (33) using bovine serum albumin as a standard. HMGR activity was determined as described by  Analysis of SDS-PAGE Separated Proteins and Western Blotting-Radiolabeled cells were homogenized in phosphate buffer system A (0.2 M K x PO 4 , pH 7.5, 0.35 M sorbitol, 10 mM Na 2 EDTA, and 5 mM MgCl 2 ) supplemented with 5 mM DTT, 1:1 g fresh weight per ml. The homogenate was centrifuged for 5 min at 8000 rpm and 4°C. Vertical SDS-PAGE was done according to Laemmli (34) using a Protean II slab cell (Bio-Rad) and 15% acrylamide gels containing 1% SDS. 40 g of protein (pellet or supernatant) were loaded per slot, and the protein bands were stained using Coomassie Brilliant Blue R-250 (Sigma). Dried gels were exposed to x-ray films (Eastman Kodak Co.). After a period of 4 -6 months, they were developed and scanned. Images were handled and processed for printing using Photoshop 5.0 (Adobe Systems, Mountains View, CA). For Western blot analysis of HMGR, 40 g of microsomal proteins from TBY-2 cells, treated with mevinolin or untreated, were separated in a 15% polyacrylamide gel and electroblotted onto a nitrocellulose membrane (Amersham Biosciences) following standard protocols and immunodetected using polyclonal antibodies raised in rabbit against the soluble domain of radish HMGR2, as described by Hemmerlin and Bach (27). Bands were quantified using a Bio-Rad GS-800 densitometer system. The antibodies had been found previously (35) to recognize apparently all isoforms of plant HMGR.

RESULTS
Mevinolin Induces Apparent HMGR Activity in Tobacco BY-2 Cells-Previously it was shown that treatment of cells with statins, to which mevinolin belongs (21), induces apparent HMGR activity (17). We showed that this observation also holds true for mevinolin-treated TBY-2 cells and that antibodies, raised against a recombinant radish HMGR2, were able to quantify not only radish HMGR isozymes (35) but also the tobacco enzymes (26,27). We achieved a higher resolution and indeed could separate two isoforms of tobacco HMGR, by increasing the polyacrylamide concentration to 15% (w/v) and allowing proteins with molecular masses smaller than 30 kDa to elute from the gel (Fig. 1). The estimated molecular masses correspond to 64 kDa for isoform 1 and 63 kDa for isoform 2. By assuming that the antibodies recognized each isoform to the same degree, without treatment the expression of isoform 1 was higher (2-fold) than the expression of isoform 2 (Fig. 1A).
Mevinolin induced the accumulation of both isoforms, which can be due to increased translation or decreased degradation of the proteins. The quantification of bands demonstrated a 3-fold increase for each isoform (Fig. 1B). The addition of MVA 24 h after mevinolin treatment and incubation for another 24 h resulted in a decrease in band intensity for both isoforms, but not to the same extent, although that of isoform 1 dropped down to base level, isoform 2 remained stimulated about twice over the initial value (Fig. 1B).
To determine the highest value of apparent HMGR activity in mevinolin-treated TBY-2 cells, we realized a time course study (Fig. 2). Microsomes were buffer-washed in order to remove all traces of inhibitor, which would otherwise mask the apparent HMGR activity in vitro. In this way, we estimate the "real activity" of the enzyme (Fig. 2), but not that in vivo, where the inhibitor is present. In control cells, we found some slight stimulation of apparent HMGR activity within the first 24 h, thereby corroborating earlier observations (26). In contrast to this, within a very short time, mevinolin treatment led first to a slight decrease in the apparent total activity, which was very rapidly compensated for by a 13-fold increase over the initial value after 16 h. At 24 and 36 h, two peaks were observed, putatively corresponding to two different isoforms of HMGR with a distinct induction behavior. At 24 h, a maximum activity corresponding to 550 pmol min Ϫ1 mg Ϫ1 was measured, whereas the peak at 36 h was lower. The activity then slightly decreased to the base level determined at the beginning of the experiment.
1-Deoxy-D-xylulose Partially Reverses Induction by Mevinolin of HMGR Activity-It had been shown previously (26,36) that MVA could overcome the inhibition by mevinolin in TBY-2 cells. DX was tested for its efficiency to down-regulate the mevinolin-induced increase of apparent HMGR activity as the key enzyme of the MVA pathway. The same inhibitor concentrations were used for the follow-up experiment, and time points at 24 and at 36 h were selected for the study. At 24 h, the mevinolin-induced HMGR activity could be partially repressed by addition of DX to the treated cells ( Fig. 3A) compared with untreated cells (column C). Because in our initial experiments we had observed a second induction peak at 36 h, we tested also this time point for possible repression by DX, but there was no significant effect (Fig. 3A). Consequently, we performed a time . A, proteins in microsomes isolated from these cells were separated by SDS-PAGE on a 15% acrylamide gel, and HMGRs were visualized by Western blotting using an antibody raised against a recombinant HMGR2 from radish. B, bands were quantified using a Bio-Rad Imaging densitometer.

FIG. 2. Time course of apparent HMGR activity in mevinolintreated tobacco BY-2 cells.
TBY-2 cells in the stationary phase were diluted 5-fold into fresh culture medium and incubated with 5 M mevinolin or not (control). Aliquots were withdrawn at different time points and microsomal fractions isolated. Microsomes were bufferwashed; before apparent HMGR activity was measured for the generation a time profile without inhibition by residual mevinolin bound nonspecifically to the enzyme. Black dots, activity in microsomes from mevinolin-treated cells; black triangles, activity in microsomes from untreated control cells.
course study with mevinolin-treated cells in the presence of DX, compared with only mevinolin-treated cells (Fig. 3B). DX could repress by 1.6-fold the first mevinolin-induced HMGR activity peak, with the highest activity at 24 h. After 26 h, HMGR activity increased to the same level as in mevinolintreated cells, with a maximum value between 36 and 40 h.

1-Deoxy-D-xylulose Reverses Inhibition by Mevinolin of Cell Growth in Tobacco BY-2 Cells-Because the mevinolin-induced
increase of HMGR activity was partially reversed by addition of DX, we tested the capacity of this compound to complement the growth inhibition. Increasing concentrations of DX (0, 0.1, 0.5, 1, or 2 mM) were used for reversion of the mevinolin-induced growth arrest. DX proved to be as efficient as MVA for reversal of mevinolin inhibition (Fig. 4). At a concentration of 2 mM each, 75% of the fresh weight of non-treated cells (control) was measured. In order to determine the minimum concentration of mevinolin necessary to efficiently block cell division and the concentration of DX capable of reversing the inhibition, we set up the combinatorial experiments described in Fig. 4. The same kind of experiment was performed using methylerythritol, the dephosphorylated form of the first committed precursor of the alternative MEP pathway for the biosynthesis of IPP. Although in E. coli methylerythritol is well incorporated in isoprenoid end products (37,38), it was found to be toxic for TBY-2 cells (data not shown), most likely because of the inability of TBY-2 cells to phosphorylate this compound. Both MVA and DX are individually able to reverse the inhibition by mevinolin of cell growth of TBY-2 cells, but the simultaneous addition of both compounds had a synergistic effect (Fig. 5). Indeed, addition of a low concentration of DX to the culture medium, with almost no effect on cell growth, helped the cells to gain 30% more mass than a cell culture treated with MVA alone (Fig. 5).
Evidence for Cross-talk between the Plastidial and the Cytosolic Pathway-In TBY-2 cells, as in other higher plants, phytosterols are synthesized via the MVA pathway (9). Some contribution of the MEP pathway is expected, however, due to crosstalk between the cytoplasmic and plastidial pathways. This contribution should be especially pronounced when DX, an easily incorporated precursor in the MEP pathway, is added to the culture medium. Our previous results (10) strongly suggest that DX was efficiently absorbed by TBY-2 cells, but also that it had to be subsequently phosphorylated, and that it did enter the MEP pathway. To verify this, we now used the stably labeled precursor [1,1,1,4-2 H 4 ]DX and carried out in vivo incor- poration experiments in the presence and in the absence of mevinolin. According to the current knowledge of the fate of the hydrogen atoms in the MEP pathway (38,39) and especially in TBY-2 cells (10), the three hydrogen atoms of the C-1 methyl group of DX are preserved in the isoprene units. However, some loss has been reported in plant systems, due to the lack of selectivity in the reaction catalyzed by the IPP isomerase (40). In addition, in TBY-2 cells, some deuterium retention occurs from [4-2 H]DX in all isoprene units from the prenyl side chain of plastoquinone as well as from phytoene, which are both essentially synthesized via the MEP pathway. This is due to a significant contribution of the DMAPP branch of the MEP pathway. This branch is characterized by the deuterium retention from [4-2 H]DX in the isoprene units derived from DMAPP, as well as from IPP resulting from its isomerization (10).
When [1,1,1,4-2 H 4 ]DX is incorporated into sterols of TBY-2 cells, six methyl groups can be labeled with three deuterium atoms in epoxysqualene, the precursor of all sterols (Fig. 6). Two labeled methyl groups are lost at C-4 and C-14 by decarboxylation during the conversion of cycloartenol into sterols. The C-19 methyl group bears only two deuterium atoms due to the formation of the cyclopropane ring in cycloartenol and its opening by protonation. Only three positions (C-17, C-20, and C-25) may be labeled from deuterium at C-4 in DX in the sterol skeleton (see dots in the structures in Fig. 6). The other three deuterium atoms, which were still present in cycloartenol, are eliminated through the loss of the two C-4 methyl groups by decarboxylation, by elimination in the conversion of cycloeucalenol into obtusifoliol, and by the introduction of the ⌬ 5 double bond in the sterol nucleus (Fig. 6).
Campesterol, stigmasterol, and sitosterol, along with isofucosterol, represent the major sterols in TBY-2 cells (41). When [1,1,1,4-2 H 4 ]DX (0.5 mM) was added to the culture medium, significant incorporation was observed into the sterols, which were analyzed as acetates by gas chromatography coupled to mass spectrometry (Table I). Analysis of the isotopomer distribution can be made on the fragment corresponding to the loss of acetic acid, which is the major ion in the mass spectrum of the acetates of most ⌬ 5 -sterols. Next to the naturally occurring isotopomer, which represented by far the most abundant one, an isotopomer bearing three deuterium atoms was found as the major labeled one, accompanied by smaller amounts of isotopomers with six and nine deuterium atoms, respectively, corresponding to sterols with 0 -3 CD 3 groups. Due to the limited incorporation, it is justified to consider only these isotopomers in a first approximation. For campesterol for instance, it can be inferred from the relative intensities of the signals at m/z 382 and 385 that about half of the sterol molecules possess one labeled CD 3 methyl group, corresponding to a value of about 10% incorporation. Isotope abundance was slightly higher in sitosterol and in stigmasterol but still of the same order of magnitude. These results were corroborated by the analysis of the signals resulting from the McLafferty rearrangement and the loss of acetic acid in the mass spectrum of isofucosteryl acetate. A maximum of three labeled methyl groups was expected. The fragment of natural abundance (m/z 296) corresponding to the isotopomer of natural abundance was accompanied by the signals from minor isotopomer with one CD 3 (m/z 299), one CD 3 and one CD 2 H (m/z 301), and finally two CD 3 and one CD 2 H (m/z 304) (Table I).
In order to improve the DX incorporation into free sterols, the MVA pathway providing the precursors of sterol biosynthesis was tentatively blocked with mevinolin. Addition of mevinolin (0.25 M) in the presence of [1,1,1,4-2 H 4 ]DX (0.5 mM) did not really affect the cells (data not shown). Sterol patterns and isotopomer distribution were nearly identical to those obtained in the absence of mevinolin. Thus, a higher but sublethal concentration of mevinolin (5 M) was used, which could be rescued by the addition of [1,1,1,4-2 H 4 ]DX (2 mM) and which allowed sufficient cell growth (20% of the control) for mass spectrometry analysis of isoprenoids (Table I). Under those growth conditions, the sterol pattern was similar to that in the absence of mevinolin. However, the concentration of isofucosterol significantly increased, and some sterols that were not previously detected appeared as minor compounds (not shown). The analysis of the signals of the fragments corresponding to the loss of acetic acid indicated that the isotopomers of natural abundance were absent for campesterol, sitosterol, and stigmasterol. An isotopomer corresponding to the maximum incorporation of deuterium at the level of the methyl groups (i.e. 11 deuterium atoms corresponding to three CD 3 and one CD 2 H), indicated that the MEP pathway was now the only IPP and DMAPP source and, consequently, that the MVA pathway was completely shut down by 5 M mevinolin (Table I). Isotopomers with fewer or more than 11 deuterium atoms accompanied the major one. However, some deuterium loss at the level of the methyl groups may have occurred by scrambling of the methyl groups in the reaction catalyzed by IPP isomerase and might be due to the resulting possible elimination of deuterium in the place of a proton (40). On the other hand, some deuterium retention from C-4 of [1,1,1,4-2 H 4 ]DX was expected and is responsible for the presence of minor isotopomers with additional deuterium atoms. Analysis of the fragment resulting from the McLafferty rearrangement and the loss of acetic acid in the mass spectrum of isofucosterol also showed the presence of isotopomers with all isoprene units labeled and again with a major isotopomer with 8 deuterium (m/z 304, corresponding to two CD 3 and one CD 2 H).   (Fig. 7). Predominantly isotopomers with 9 labeled methyl groups were detected. In the case of incorporation of [1,1,1,4-2 H 4 ]DX in the absence of mevinolin, the major isotopomer corresponding to 9 CD 3 groups as shown by the fragment cluster culminated at m/z 775, corresponding to the molecular ion of plastoquinone and to the incorporation of 27 deuterium atoms. The same incorporation pattern was observed when [1,1,1,4-2 H 4 ]DX was fed in the presence of the highest mevinolin concentration. In this case, two prominent signals were observed at m/z 775 and 777. They corresponded to the molecular ions of plastoquinone and plastoquinol, both containing again 9 CD 3 groups. Indeed, analysis of prenylated quinones is often hampered by the reduction of the quinone inside the source of the mass spectrometer, a process that is unavoidable and hardly reproducible (42). The high level of incorporation was confirmed in the mass spectra from all three feeding experiments performed with [1,1,1,4-2 H 4 ]DX. The signal m/z 189 in the spectrum of natural abundance plastoquinone, corresponding to the quinone ring and four carbon atoms of the isoprene unit linked to this ring (Fig. 7), was absent in the spectra obtained after feeding with [1,1,1,4-2 H 4 ]DX. The appearance of the m/z 192 fragment indicated that the methyl group of the isoprene unit was labeled with 3 deuterium atoms. This was confirmed by the shift of the m/z 69 signal found in the natural abundance spectrum to m/z 72, indicating that the terminal isoprene unit contained a trideuterated methyl group (Fig.  7). Some deuterium loss due to the IPP isomerase-catalyzed reaction is also likely as signals with m/z 191 and m/z 71, corresponding to the incorporation of two deuterium atoms, were observed for the two fragments discussed above. From the relative intensities of the m/z 189, 191, and 192 and m/z 69, 71, and 72 signals, it is estimated that at least 85% of the isoprene units had a labeled methyl group. The fragment corresponding to the quinone ring plus the adjacent carbon of the first isoprene unit in the side chain served as control. Its signal (m/z 151) was not shifted and did not display any deuterium labeling.
These three incorporation experiments showed that the pre-nyl chain of plastoquinone was essentially synthesized from exogenous DX added to the culture medium. If the contribution of the non-labeled material introduced with the inoculum is deducted (ϳ3%), no significant, or only a little, dilution occurred from non-labeled DX synthesized de novo from sucrose, the main non-labeled carbon source found in the culture medium.
To determine whether DX is also incorporated into the isoprenyl moiety of isoprenylated proteins, [2-14 C]DX was enzymatically synthesized and incorporated in the presence or absence of 5 M mevinolin. After separation by SDS-PAGE and autoradiography of dried gels, labeled proteins in the low mass range (ϳ20 kDa) appeared in the presence of [2-14 C]DX (Fig. 8). Interestingly, the [2-14 C]DX was also incorporated into nonmevinolin-treated membrane proteins (Fig. 8, lane 3).
Mevalonate Partially Reverses Inhibition by Fosmidomycin of Cell Growth in Tobacco BY-2 Cells-Because DX complements the mevinolin-induced growth inhibition in TBY-2 cells, the opposite experiment was performed, in which we blocked the alternative MEP pathway, and we tried to reverse the inhibition with a product of the classical MVA pathway. TBY-2 cells were treated with fosmidomycin, which blocked TBY-2 cell proliferation at concentrations as low as 20 M (Fig. 9A). The simultaneous addition of fosmidomycin and MVA led to a better cell growth than observed after fosmidomycin treatment alone, although control levels could not be reached again. The effect of fosmidomycin on HMGR activity in TBY-2 cells was also tested. The activity remained unaffected during these treatments (Fig. 9B).
Further Evidence for Cross-talk between the Cytosolic and the Plastidial Pathway, Incorporation of [2-13 C]MVA into Phytosterols and Plastoquinone-Sterols are essentially synthesized via the MVA pathway in higher plants. Thus, incorporation of [2-13 C]MVA is an efficient way to evaluate the turnover of this metabolic route. Addition of [2-13 C]MVA alone to the culture medium resulted in a strong labeling of all sterols (campesterol, sitosterol, stigmasterol, and isofucosterol) with similar isotope enrichment (Table II) according to the pattern presented in Fig.  10. In a first approximation, only the isotopomer with 5 labeled isoprene units (with a very modest contribution of the isotopomer with 4 labeled isoprene units), i.e. with the maximum labeling expected from the MVA pathway, was observed. It was accom-  [1,1,1,4-2 H 4

]DX into the sterols of tobacco BY-2 cells
The TBY-2 cell suspensions were incubated for 7 days as described under "Experimental Procedures." Reference cultures contained nothing but growth medium. Other treatments are as indicated. MV, mevinolin.  (5)  Isofucosteryl acetate 296 (0) 297 (0) 298 (0) 299 (0) 300 (0) 301 (15) 302 (26) 303 (33) 304 (100) 305 (48) 306 (10) panied by a small cluster of less labeled isotopomers with 0 -3 labeled isoprene units, which seemed to have been synthesized separately from the major 13 C 5 -isotopomer with a significant contribution of non-labeled MVA derived from sucrose, the carbon source of the culture medium. An explanation for the nonstatistical distribution of side peaks around the dominant mass peaks in the recorded spectra (not shown) can be given by assuming the presence of two pools of sterols that incorporate MVA: one dominant, indicative of a high degree of labeling, and one small, with a slightly reduced mass distribution and lower percentage of incorporation. The latter phenomenon can be interpreted to mean that with depletion of exogenous MVA during the first half of the cultivation period, the feedback regulation of HMGR by downstream products is gradually lost. Consequently, endogenous MVA is again synthesized and utilized for incorporation into newly formed end products of the sterol pathway. This interpretation is also supported by comparison of differences in labeling patterns between control and treated samples, using one or more inhibitors. In the presence of inhibitors of both the MVA and the MEP pathways, there was essentially no further dilution by endogenous substrate (Table II) 1 and 2) and membrane (lanes 3 and 4) proteins were extracted and separated on a 12% SDS gel. companied by small amounts of the [ 13 C 4 ]isotopomer, was still nearly the only one, but the above-mentioned pool of sterols with low isotope abundance completely disappeared in the presence of mevinolin. Simultaneous treatment with mevinolin and fosmidomycin in the presence of [2-13 C]MVA inhibited both isoprenoid biosynthetic pathways. Consequently, all sterols were solely represented by their 13 C 5 -isotopomer, indicating that the TBY-2 cells had to completely rely on the exogenous MVA source for ensuring the sterol supply.
Normally, plastoquinone, like all chloroplast isoprenoids, is synthesized via the MEP pathway (Fig. 10). This had also been verified in TBY-2 cells (10). After feeding of labeled MVA, only partial labeling of plastoquinone is expected due to the possibility of exchanges of intermediates between the cytoplasm and the chloroplasts. Incorporation of [2-13 C]MVA into plastoquinone of TBY-2 cells was followed by fast atom bombardment ionization mass spectrometry, which showed essentially the molecular ion of the plastoquinol (m/z 750) formed by reduction of the quinone in the source of the mass spectrometer, accompanied by a minor signal corresponding to the quinone (m/z 748) (Table III).
After feeding with [2-13 C]MVA alone or in the presence of mevinolin, a similar labeling pattern was obtained. The predominant natural abundance isotopomer was accompanied by isotopomers with 1-3 labeled isoprene units. Addition of fosmidomycin (20 M) increased the [2-13 C]MVA incorporation into plastoquinone, the isotopomer with three labeled isoprene units becoming preponderant. In the presence of fosmidomycin, the failure of the MEP pathway to produce the precursors for the plastoquinone side chain was significantly rescued by exogenous mevalonate. In the presence of fosmidomycin (20 M) and mevinolin (5 M), both MEP and MVA pathways were blocked. This resulted in a more efficient incorporation of [2-13 C]MVA, which was now the only source for intermediates of the isoprenoid biosynthetic pathway for the cells. The isotopomer with 5 labeled isoprene units was the major one (m/z 753) with a significant contribution of the isotopomers with 6 and 7 labeled isoprene units (Table III). DISCUSSION The discovery of the alternative pathway for the biosynthesis of plastidial isoprenoids raised considerable interest in the development of potential herbicidal compounds, which would thus interfere with carotenoid and chlorophyll biosynthesis, besides affecting that of at least one important class of phytohormones, gibberellic acids. Such herbicidal compounds would only be fully active if both the cytosolic MVA and the plastidial MEP pathways would not exchange downstream intermediates such as IPP, geranyl diphosphate, and/or geranylgeranyl diphosphate. The purpose of our study was to demonstrate more directly biosynthetic and regulatory links between the classical MVA pathway in the cytosol and the alternative MEP pathway in plastids. Some of our previous work (26) already gave some insights into this direction. For instance, mevinolin-treated cells needed more than 36 h to be completely blocked within the cell cycle. Another surprising observation was the time course of cell death induction by mevinolin. Whereas in animal cell cultures under those conditions the number of dead cells steadily increases to 100%, TBY-2 cells reached a dead cell population of 20% after 48 h, with this number only increasing to 30% over 1 week of culture (26). This phenomenon suggested that there might exist some sort of slowly inducible, intracellular complementation by IPP synthesized through the MEP pathway, thereby allowing the survival of these plant cells. The existence of some rather limited minor spillover of plastidial IPP units into cytoplasmic sterols had been observed earlier (11,40) using [5,5-2 H 2 ]DX as precursor. Thus, our goal was to examine the extent to which such low import and export processes could be stimulated. The use of inhibitors to control a metabolic flux circumvented the need to apply conditions where defense reactions come into play (43,44), and the use of cell suspension cultures prevented the possibility that absorption and transport by various tissues (45,46) might affect or even mask biosynthetic capacities. In cell cultures, exogenous compounds are integrated into cellular products only if they are efficiently absorbed, which is the case in TBY-2 cells. However, incorporation can be "diluted" by endogenously produced precursors. These can even be channeled into specific branches of a pathway through formation of metabolic units (metabolons), with no access to intermediates. Another possible problem with incorporation studies is that overloading of systems with a precursor molecule might challenge the cells to induce rescue pathways for rapid metabolization, because the pile up of potentially toxic intermediates must be avoided. In our system, we could observe some subtle changes in the ratio of certain sterols and the appearance of some minor sterols (data not shown), depending on the treatments. Sterol accumulation in TBY-2 cells follows a biphasic pattern (41), which coincides with a cell division phase, followed by cell elongation. As those processes are affected by mevinolin and, apparently, by fosmidomycin, the slight differences in the sterol pattern observed might be due to some adaptation of the developmental program to the inhibition, possibly to maintain the rigidity of the cell membrane despite limited integration of de novo synthesized sterols (47).
The capacity of higher plant cells to use two independent pathways for the synthesis of isoprene units is also found in some prokaryotic organisms like strains of Streptomyces. In these prokaryotes, however, both the MEP and the MVA pathways are sequentially utilized for the synthesis of isoprenoid compounds (48), whereas in plants they are compartmentalized and operate in parallel (46). The question arises as to why some organisms like higher plants acquired and maintained two independent pathways, despite numerous examples in others, including animals (17) or sterol and carotenoid-producing algae (49), which easily cope with the "disadvantage" of having only one of the pathways operating. Does this double choice exert a vital role specifically for higher plants or at least bear some advantage? Ascomycete species in the rhizosphere, like Aspergillus terreus (21), ubiquitously produce HMGR inhibitors such as mevinolin and its functional congeners (50). Plants, which are unavoidably exposed to the presence of such inhibitors without the possibility to escape, might therefore have maintained the possibility of internal complementation.
The potential of mevinolin to inhibit MVA production was used already over decades to study isoprenoid biosynthesis. For instance, root growth of radish seedlings was significantly diminished at concentrations between 0.125 and 1.25 M mevinolin and could be restored by the addition of 2 mM of exogenous MVA (22), which is similar to the values needed to restore growth in the TBY-2 cell system used in this study (Fig. 5). At Ͼ20 M, mevinolin caused some unspecific, generally toxic effects in etiolated wheat seedlings, which were prevented from greening after exposure to light or even bleached (22). This could have suggested a direct involvement of the MVA pathway in plastidial chlorophyll biosynthesis. However, at non-toxic concentrations of mevinolin (below 10 M), in light-grown radish seedlings, the amount of phytosterols and ubiquinone de-  (4)  Isofucosteryl acetate 296 (25) 297 (34) 298 (28) 299 (24) 300 (100) 301 (20) 302 (2) MVA (2 mM creased (the latter less dramatically), whereas the synthesis of carotenoids or chlorophylls in cotyledons remained practically unchanged (51). In a later study (52), a sharp decrease in tomato HMGR activity before ripening and carotenoid accumulation can nowadays be explained through a practically exclusive contribution of the MEP pathway (53,54) to carotenoid synthesis during ripening, with DXS apparently exerting flux rate control. Dependence on MVA biosynthesis during the early stages of tomato fruit development, shown by a correlation between growth and HMGR activity (52), once again indicates the importance of this precursor for cell division and elongation, comparable with the situation in TBY-2 cells. Growth inhibition by mevinolin of cultured TBY-2 cells has been shown to be reversed upon addition of MVA (26,36), as well as being partially reversed by cytokinins at low mevinolin concentrations (36), and the cell cycle arrest in G 1 could be overcome by alkalinization of the cytosolic pH (28). The present study shows that a further compound (DX, the dephosphorylated analog of the first intermediate in the MEP pathway) can be used for the re-initialization of cell division, strongly suggesting that the exogenous DX can stimulate the synthesis or is incorporated into normally MVA-derived compounds. To confirm that the biosynthesis of major non-inducible cytoplasmic isoprenoids from DX, viz. phytosterols, occurs in parallel to that of a major plastidial isoprenoid such as plastoquinone, we fed [1,1,1,4-2 H 4 ]DX to TBY-2 cells, after having established the best conditions (sucrose and inhibitor concentrations). Obtaining enough material for chemical analyses required incubating for 7 days (9).
The measurement of the activity of a rate-limiting and highly regulated enzyme like HMGR is an excellent indicator to understand how a biosynthetic pathway might be affected during a physiological situation. We have demonstrated, in agreement with previous observations (33), that HMGR protein is induced by mevinolin. When potato cells were transformed for expression of His-tagged HMGR, a similar dependence was shown in Western blot analyses using anti-His 6 antibodies (55). Our observations clearly indicate that this induction is counteracted at least at the level of one isozyme by feeding exogenous DX (Fig. 3), apparently through reconstitution of feedback regulation. Growth inhibition can be overcome as well, albeit not completely when inhibitor concentrations are much increased (Fig. 4). Two HMGR isozymes are induced, one more strongly than the other, and the kinetics of induction have a biphasic behavior, with overlapping peak characteristics (Fig. 2). We therefore postulate that under our cultivation conditions, we have two major forms of the enzyme active in TBY-2 cells: a housekeeping one that is slowly reacting, and a stress-induced one. Our results support the hypothesis that plant HMGR, at least the isozyme possibly linked to sterol biosynthesis, is feedback-regulated (13).
During the course of our experiments, we noted that fosmidomycin was much less efficient than mevinolin as a growth inhibitor of TBY-2 and green tobacco cells (data not shown). In contrast, almost complete inhibition was seen in wild-type E. coli cells treated with less than 1 M. 3 But why should inhibition of a plastidial pathway in cells without photosynthetic activity and grown heterotrophically affect cell division? We hypothesized that in a system like TBY-2 cells, fosmidomycin might act somewhat nonspecifically and might also affect the enzyme acetolactate reductoisomerase, which is involved in the biosynthesis of branched-chain amino acids, due to mechanistic similarities with the MEP synthase (1-deoxy-D-xylulose-5-phosphate reductoisomerase) reaction. Indeed, we could observe a certain reversal of fosmidomycin-induced growth inhibition by feeding mixtures of valine, leucine, and isoleucine. 3 Intact integration of the carbon skeleton of leucine into isoprenic units as described for the trypanosomatid Leishmania mexicana (56) seems very unlikely in plants (see Ref. 3). At higher concentrations those amino acids were toxic to TBY-2 cells. An alternative possibility remains speculative, e.g. a catabolism of leucine to acetyl-CoA (57,58), which could then re-enter the MVA pathway. In such a case, it would be MVAderived isoprene units that complement the fosmidomycin-induced deficiency in plastids. This study has demonstrated that feeding MVA to TBY-2 cells can overcome growth inhibition by fosmidomycin (Fig. 9) and that MVA incorporation into plastoquinone, a product of the MEP pathway, could be enforced when both pathways were blocked, as demonstrated by the mass spectral data in Table III. These results confirm and strengthen those of earlier studies, in which chloroplast pigments, also synthesized via the MEP pathway, were found to be labeled by MVA even in the absence of inhibitor, although with a low efficiency of incorporation (59 -61). The earlier results could not be considered definitive because in most cases the radiochemical purity of products was not proven by repeated chromatography and recrystallization, which in the case of colored pigments like carotenoids and chlorophylls is difficult due to the instability of these compounds when subjected to such treatments.
Based on a series of experiments (62, 63) using purified plastids with no appreciable activities of enzymes in the MVA pathway up to IPP, it was hypothesized that an IPP transporter should be localized to the plastid envelope membrane. This was evidenced by Soler et al. (64), although the measured K m IPP values for plastids isolated from Vitis vinifera L. cell cultures exceeded the range of 0.5 mM, which seems extremely high or is simply reflecting a nonspecific phosphate or sugar phosphate transporter that accepts IPP. To our knowledge, no experimental evidence for being of a transport from the plastids to the cytosol has been described. We demonstrated that the transport is possible in both directions. Although we can only hypothesize that we have an exchange of C 5 units between the cytosolic and plastidial compartments, it is not impossible that longer chains may be transported across the envelope. Labeling by [1-13 C]DX of chamomile sesquiterpenoids after elicitation (65) suggested that a 3 A. Hemmerlin, unpublished observations.  [2-13 C]MVA into plastoquinone of tobacco BY-2 cells The TBY-2 cells suspensions were incubated for 7 days as described under "Experimental Procedures." Reference cultures contained nothing but growth medium. Other treatments are as indicated. MV, mevinolin; I, fosmiolomycin.

Treatment
Molecular ion m/z % of relative intensity cytosolic farnesyl diphosphate synthase utilized an allylic C 10 unit (geranyl diphosphate) synthesized in the plastidial compartment for the condensation with a terminal C 5 IPP unit to form farnesyl diphosphate. This molecule, a product of both the MVA and MEP isoprenoid pathways, would then be used for the subsequent cyclization reaction to form typical chamomile cytosolic sesquiterpenoids. A similar, but less conclusive situation, seems to exist with respect to the biosynthesis of the sesquiterpene germacrene D in Solidago canadensis (66).
In conclusion, although it is not yet possible to decide which product of the MEP pathway (IPP, DMAPP, geranyl diphosphate, etc.) is exported to the cytosol and used for the biosynthesis of farnesyl diphosphate-derived entities in TBY-2 cells, or which cytosolic intermediate enters the plastidial compartment, we have presented evidence that significant exchange of metabolites of the cytosolic and the plastidial isoprenoid pathways across the plastid envelope is possible. In order to address such questions more precisely, we are currently developing a system with transformed TBY-2 cells that will permit the direct determination of biosynthetic flux rates, without the need to apply inhibitors. Although the export of precursor units from the plastids to the cytosol seems to operate more readily, under specific conditions, intracellular complementation by the cytoplasmic MVA pathway of plastidial isoprenoid synthesis seems possible, but under rather restrictive conditions. These observations are essentially similar to other ones, of which we just recently became aware, based on wild-type and CLA1 mutant Arabidopsis seedlings (67) treated with mevinolin (68), or showing that in Croton sublyratus, isoprene units used for the synthesis of phytosterols have a double origin, suggesting an active exchange between the compartments (69). Taken together, these results have to be considered before stepping into the development of herbicides interfering with early steps in the biosynthesis of plastidial isoprenoids. Under stress conditions affecting vital processes, the higher plant cell could react by stimulating mechanisms involved in the exchange of intermediates in the biosynthesis of isoprenoids, which could render any treatment inefficient in the long run.