A Highly Conserved Signal Controls Degradation of 3-Hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) Reductase in Eukaryotes*

Sterol synthesis by the mevalonate pathway is modulated, in part, through feedback-regulated degradation of 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGR). In both mammals and yeast, a non-sterol isoprenoid signal positively regulates the rate of HMGR degradation. To define more precisely the molecule that serves as the source of this signal, we have conducted both pharmacological and genetic manipulations of the mevalonate pathway in yeast. We now demonstrate that farnesyl diphosphate (FPP) is the source of the positive signal for Hmg2p degradation in yeast. This FPP-derived signal does not act by altering the endoplasmic reticulum degradation machinery in general. Rather, the FPP-derived signal specifically modulates Hmg2p stability. In mammalian cells, an FPP-derived molecule also serves as a positive signal for HMGR degradation. Thus, both yeast and mammalian cells employ the same strategy for regulation of HMGR degradation, perhaps by conserved molecular processes.

The mevalonate pathway is responsible for the biosynthesis of numerous essential molecules including prenyl groups, coenzyme Q, dolichol, and sterols such as cholesterol (1). 3-Hydroxy-3-methylglutaryl-coenzyme A reductase (HMGR) 1 is a key enzyme of the pathway and is rate-determining for cholesterol synthesis in mammals (2,3). The mevalonate pathway is modulated in large part by feedback control of the amount of HMGR protein (1), and a significant portion of HMGR feedback control occurs through regulation of HMGR degradation (4 -6). HMGR is an integral endoplasmic reticulum (ER) membrane protein, and its degradation occurs without exit from the ER (6 -8). The non-catalytic, N-terminal transmembrane anchor of HMGR is both necessary and sufficient for regulated degradation (6, 8 -10). When there is abundant synthesis of pathway products, HMGR degradation is fast, and steady-state levels of the protein tend to be low. Conversely, when synthesis of pathway products is reduced, for instance when a patient is given the HMGR inhibitor lovastatin, HMGR degradation is slowed, and steady-state levels of the protein tend to increase. However, neither the identity of the mevalonate-derived signal nor the mechanism by which this signal is coupled to HMGR degradation is known.
In order to discover and understand the mechanisms of HMGR-regulated degradation, we have been studying the process in the yeast Saccharomyces cerevisiae (6,11,12). Our earlier work has shown that the yeast HMGR isozyme Hmg2p is degraded in a regulated manner with many similarities to the analogous process in mammals. Through the use of genetic selections and screens, we have been able to identify genes required for the degradation of Hmg2p, called HRD genes (11), and genes required for normal regulation of Hmg2p degradation. 2 Concurrently with these screens, we have been studying the nature of the mevalonate-derived signals that control Hmg2p stability. Hmg2p degradation is regulated by an unknown signal from the mevalonate pathway (Fig. 1a). Inhibiting early pathway enzymes, such as HMGR itself or HMG-CoA synthase, decreases the rate of Hmg2p degradation (6). These early pathway blocks decrease the availability of a downstream signal for degradation. Conversely, inhibiting the enzyme squalene synthase, which is downstream of HMGR, stimulates degradation and ubiquitination of Hmg2p (12). Furthermore, the degradation-enhancing effect of squalene synthase inhibition is abolished by simultaneous inhibition of HMG-CoA synthase or HMGR.
These pharmacological studies imply that the signal for Hmg2p degradation is a pathway molecule between mevalonate and squalene (Fig. 1a). The most reasonable candidate for this signal is farnesyl diphosphate (FPP) or an off-pathway FPP derivative. The idea that FPP, or a derivative, is a positive signal for Hmg2p degradation is particularly interesting since there is accumulating evidence from in vitro and in vivo studies that farnesol, an FPP-derivative, is a signal for regulation of mammalian HMGR stability (13)(14)(15)(16)(17).
We have now tested the hypothesis that FPP provides a molecular signal for control of Hmg2p stability using unique genetic opportunities available in yeast. In conjunction with pharmacological and biochemical approaches, we have constructed yeast strains that allowed either overexpression or down-regulation of specific mevalonate pathway genes (Fig.  1b). Our results indicated that FPP was indeed a source of a signal for Hmg2p degradation in yeast, demonstrating that there is striking conservation for this mode of HMGR regulation among eukaryotes.

EXPERIMENTAL PROCEDURES
Materials and Reagents-All enzymes were obtained from New England Biolabs (Beverly, MA). Chemical reagents were obtained from Sigma. Lovastatin and zaragozic acid were generously donated by Merck. Terbinafine was commercially obtained as a 1% Lamisil® solution from Novartis (East Hanover, NJ). ECL TM chemiluminescence immunodetection reagents were from Amersham Pharmacia Biotech. The anti-Myc 9E10 antibody was used as a cell culture supernatant * This work was supported by National Institutes of Health Grant DK5199601 (to R. Y. H.) and a Searle scholarship (to R. Y. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
obtained by growing the 9E10 hybridoma (ATCC CRL 1729) in RPMI 1640 culture medium (Life Technologies, Inc.) with 10% fetal calf serum. The anti-HA antibody was an ascites fluid obtained from Babco (Berkeley, CA). The anti-HSV-Tag antibody was obtained from Novagen (Madison, WI). Affinity-purified goat anti-mouse horseradish peroxidase-conjugated antiserum was obtained from Sigma. Recombinant DNA and Molecular Cloning-PCR was performed as described previously (18). The genes encoding squalene synthase (ERG9), farnesyl-diphosphate synthase (ERG20), and squalene epoxidase (ERG1) were PCR-amplified from yeast strain RHY623 genomic DNA (18), isolated by the Winston method (19), using separate primers that contained PstI and BamHI sites in the upstream primers and NheI and SalI sites in the downstream primers. The amplified ERG9 and ERG20 genes were cloned between the PstI and SalI sites in pRH423 (12), thereby placed under control of the glyceraldehyde-3-phosphate dehydrogenase gene (GAPDH) promoter (P GAPDH ) (20). The squalene synthase-containing plasmid was named pRH440, and the farnesyldiphosphate synthase-containing plasmid was named pRH830.
Squalene synthase (ERG9) was tagged at the C terminus with the HSV epitope sequence (21). The plasmid containing an HSV-tagged squalene synthase expressed from the GAPDH promoter was made by annealing primers that encoded the HSV epitope sequence (QPELAPE-DPED) and cloning the resulting DNA fragment between the NheI and SalI sites in pRH440 to yield pRH442 (ERG9-HSV). The plasmid to tag ERG9 at its genomic locus with HSV was made by digesting pRH442 with MunI, and the 5.4-kb vector fragment was reclosed to yield pRH885. The remaining portion of erg9 included codons 208 -446 and the 3Ј HSV sequence.
Plasmids that allowed expression of the genomic copy of either ERG9, ERG20, or ERG1 from the MET3 promoter (P MET3 ) (22,23) were constructed as follows: pRH442 was digested with EcoRI, and pRH448 was digested with KpnI. Each vector was reclosed with the insert removed. The erg9 vector was named pRH948, and the erg20 vector was named pRH950. The MET3 promoter was cloned into pRH948 and pRH950 by digesting each plasmid with SspI and PstI and replacing the insert with the MET3 promoter containing SspI-PstI fragment from pHAM8 obtained from Dr. Harry Mountain (Staffordshire, UK). The P MET3 -erg9 plasmid was named pRH973 and the P MET3 -erg20 plasmid was named pRH975. A PCR product containing the ERG1 coding region was digested with BamHI and PvuII. The 840-base pair fragment was then cloned between the BamHI and HpaI sites in pRH973, and the resulting plasmid was named pRH1204.
Strains and Media-Escherichia coli DH5␣ strains were grown at 37°C in LB ϩ amp (100 g/ml). Yeast strains were grown at 30°C in minimal medium supplemented with glucose and the appropriate amino acids, as described (6). The lithium acetate method was used to transform yeast with plasmid DNA (26).
Methionine chase assays were performed as follows. Cells were grown to early log phase with an absorbance (A 600 ) of 0.01. Methionine was added to a final concentration of 2 mM, and the cells grown at 30°C for 15 h. Cells were then either used for the cycloheximide chase assay as described above or for the FACS analysis described below.
Hmg2p ubiquitination assays were performed as described previously. 3 Strains were transformed with pRH1100, 3 which expressed a triple HA epitope-tagged ubiquitin from the constitutive GAPDH promoter. Transformants were selected for Ade ϩ prototrophy. Hmg2p ubiquitination was assayed by immunoprecipitation of 1Myc-Hmg2p and then immunoblotting the precipitate for covalently linked HA-ubiquitin.
Flow Cytometry (FACS Analysis)-FACS analysis was performed as described previously (18). Living cells were analyzed by flow microfluorimetry using a FACScalibur TM (Becton Dickinson, Palo Alto, CA) flow microfluorimeter with settings for fluorescein-labeled antibody analysis. Histograms were produced from 10,000 individual cells and were plotted with log fluorescence (arbitrary units) on the horizontal axis and cell number on the vertical axis.

Zaragozic Acid Increased Hmg2p Degradation by Squalene
Synthase Inhibition-We previously discovered that addition of zaragozic acid (ZA), a potent inhibitor of squalene synthase (30) (Fig. 1a), to yeast cells stimulated degradation of Hmg2p (12). Squalene synthase inhibition would be expected to cause a build-up of farnesyl diphosphate (FPP), the substrate of the enzyme, implicating FPP as a source of the positive signal for Hmg2p degradation. However, it was possible that the effect of ZA was the result of some action of the drug. We addressed this by testing the effect of ZA on Hmg2p degradation in a strain that overexpressed squalene synthase. If ZA was inducing Hmg2p degradation through squalene synthase inhibition, then cells expressing increased levels of squalene synthase should require increased amounts of ZA to cause the same degree of degradation.
We placed the squalene synthase gene (ERG9) under control of the constitutive GAPDH promoter, P GAPDH (20), and transformed the P GAPDH -ERG9 construct into yeast cells. P GAPDH -ERG9 was present as a single integrated copy through targeted insertion of an integrating vector. The strain expressing a 3 Gardner, R. G., and Hampton, R. Y. (1999) EMBO J., in press. FIG. 1. Manipulation of the mevalonate pathway. A, representation of the mevalonate (Mev) pathway indicating key enzymes and their inhibitors. B, construction of a MET3 promoter-regulated allele by "promoter switching." A mevalonate pathway gene, ERGX, was placed under control of the MET3 promoter through targeted integration of a promoter-switch plasmid at the ERGX genomic locus. The plasmid consisted of a non-functional, 5Ј portion of the ERGX gene placed behind the MET3 promoter (P MET3 ). Integration resulted in a functional copy of the ERGX gene placed under control of the MET3 promoter and a non-functional, deleted copy behind the native promoter.
single, integrated copy of P GAPDH -ERG9 was 8-fold more resistant to the growth-slowing effect of ZA than a wild-type strain, consistent with approximately 8-fold higher squalene synthase levels in the P GAPDH -ERG9 strain than the wild-type strain, as measured by Western blot of HSV-tagged versions of squalene synthase. 4 We then tested if the Hmg2p degradation-enhancing effect of ZA required higher doses upon squalene synthase overexpression. The yeast strains used above also co-expressed two versions of Hmg2p, 1Myc-Hmg2p and the fluorescence reporter protein Hmg2p-GFP which have identical degradation behaviors as normal Hmg2p (12,18). Cells containing a single, integrated copy of P GAPDH -ERG9 required 8-fold more ZA to decrease 1Myc-Hmg2p and Hmg2p-GFP steady-state levels as was required for wild-type cells (Fig. 2, a and b, 40 g/ml for P GAPDH -ERG9 versus 5 g/ml for wt), consistent with the 8-fold overexpression of squalene synthase. These results suggested that the mechanism for ZA-induced degradation of Hmg2p was through squalene synthase inhibition.
Squalene Synthase Levels Determined the Degree of Hmg2p Degradation-The above results indicated that ZA altered Hmg2p degradation by decreasing squalene synthase activity. In that case, genetic down-regulation of the squalene synthase gene (ERG9) should also increase Hmg2p degradation. Because mevalonate pathway enzymes are essential for yeast viability, a null allele of ERG9 in yeast results in cell death (32,33). Therefore, we made a conditional allele of squalene synthase by placing the wild-type ERG9 gene under control of the MET3 promoter (22), which is repressed by the presence of high extracellular concentrations (Ͼ0.5 mM) of methionine (23). We constructed a "promoter-switch" plasmid that contained a truncated version of erg9 placed behind the MET3 promoter. Targeted integration of this plasmid into the ERG9 locus resulted in the creation of a single, functional copy of ERG9 under control of the regulated MET3 promoter (P MET3 ) (Fig. 1b). This plasmid was used to transform a methionine prototroph yeast strain to allow growth in any concentration of methionine. The strain also co-expressed 1Myc-Hmg2p and Hmg2p-GFP allowing a complete characterization of Hmg2p degradation. When grown in the absence of methionine, normal regulated Hmg2p degradation was observed in the yeast strain that expressed squalene synthase from the MET3 promoter (P MET3 -ERG9). 4 Genetic down-regulation of squalene synthase enhanced Hmg2p degradation in a manner identical to inhibition with ZA. After 15 h growth in 2 mM methionine, Hmg2p degradation in the P MET3 -ERG9 strain was increased. This was indicated by a lower steady-state level of Hmg2p-GFP and 1Myc-Hmg2p in the P MET3 -ERG9 strain compared with the wild-type strain (Fig. 3, a and b, P MET3 -ERG9 versus wt). The effect of downregulation was similar to that in the wild-type strain after 15 h incubation in the presence of ZA (P MET3 -ERG9 compared with wt,ϩZA). Hmg2p-GFP in the P MET3 -ERG9 was stabilized by the addition of lovastatin to a similar degree as the stabilization of Hmg2p-GFP by lovastatin addition in the wild-type strain preincubated with ZA (P MET3 -ERG9, ϩLov compared with wt,ϩZAϩLov), indicating that enhanced Hmg2p degradation caused by squalene synthase down-regulation or ZA addition was still regulated by the mevalonate pathway.
It was possible that enhanced Hmg2p degradation was the result of cell death inadvertently caused by squalene synthase down-regulation rather than the build-up of a positive signal for degradation. Cells containing the P MET3 -ERG9 allele ceased to grow after 15 h incubation in 2 mM methionine (6 doublings) (see below, Fig. 8b). However, when the P MET3 -ERG9 cells from this 15-h time point were transferred to media without methionine, they retained the same plating efficiency 4 and growth curve as identically treated wild-type ERG9 cells (see below, 4 R. Gardner and R. Hampton, unpublished observations. Cycloheximide chase assay of strains expressing squalene synthase from the wild-type promoter (wt) or from the MET3 promoter (P MET3 -ERG9) in the presence of a normal HRD1 gene or the hrd1⌬ allele (P MET3 -ERG9, hrd1⌬, and wt, hrd1⌬). Cells were grown for 15 h at 30°C in the presence of 2 mM methionine, and then cycloheximide was added to 50 g/ml final concentration. After addition of cycloheximide, lysates were prepared at the indicated times and immunoblotted to determine Hmg2p levels. Fig. 8b), indicating that these cells were still viable.
The enhanced Hmg2p degradation by ZA addition or squalene synthase down-regulation required the HRD gene-encoded proteins. Enhanced degradation by squalene synthase or ZA addition was completely eliminated by the presence of the hrd1⌬ allele (Fig. 3b, P MET3 -ERG9, hrd1⌬, and wt ϩZA, hrd1⌬), which normally stabilizes Hmg2p (11). This indicated that the degradation-enhancing effect of squalene synthase down-regulation or ZA addition required a functional HRD pathway and was not due to aberrant degradation by an alternate pathway. Furthermore, identical steady-state levels of 1Myc-Hmg2p in the wild-type, hrd1⌬ strain and the P MET3 -ERG9, hrd1⌬ strain indicated that the lower steady-state levels of Hmg2p in the P MET3 -ERG9 strain were due only to enhanced degradation and not reduced translation efficiency.
Additionally, we observed that overexpression of squalene synthase stabilized Hmg2p. In the strains previously described in Fig. 2, which overexpressed squalene synthase by the presence of the P GAPDH -ERG9 allele, both versions of Hmg2p were significantly stabilized. This was observed as both an increase in the steady-state level of 1Myc-Hmg2p (Fig. 4, P GAPDH -ERG9 versus wt) and Hmg2p-GFP, 4 and as a decrease in 1Myc-Hmg2p degradation when squalene synthase was overexpressed ( Fig. 4, P GAPDH -ERG9 versus wt). Thus, squalene synthase overexpression had the opposite effect on Hmg2p degradation as squalene synthase down-regulation.
Down-regulation of Farnesyl-diphosphate Synthase Stabilized Hmg2p-The above studies implicated the substrate of squalene synthase, FPP, as a central molecule in the regulation of Hmg2p stability. Manipulation of squalene synthase predicted to increase FPP levels hastened Hmg2p degradation, whereas manipulation of squalene synthase predicted to decrease FPP levels slowed Hmg2p degradation. We wanted to test further the hypothesis that FPP was the source of the positive signal for Hmg2p degradation, by eliminating FPP production. This could be accomplished by inhibition of farnesyl-diphosphate synthase (FPP synthase), which generates FPP as a product. Unfortunately, no drugs are currently available that inhibit yeast FPP synthase in vivo. Therefore, we again used a genetic approach to lower FPP synthase production. As with the other enzymes of the mevalonate pathway, yeast cells that contain a null allele of FPP synthase are not viable (34), so we generated a conditional allele of the FPP synthase coding region (ERG20) that resulted in the wild-type ERG20 gene placed under control of the MET3 promoter, similar to ERG9 described above. When grown in the absence of methionine, normal regulated Hmg2p degradation was observed in the yeast strain that expressed FPP synthase from the MET3 promoter (P MET3 -ERG20). 4 In contrast to enhanced Hmg2p degradation caused by squalene synthase down-regulation, FPP synthase down-regulation resulted in stabilization of Hmg2p. When the P MET3 -ERG20 strain was grown 15 h in 2 mM methionine, Hmg2p was exceedingly stable, as indicated by both a higher steady-state level and decreased degradation of 1Myc-Hmg2p in the P MET3 -ERG20 strain compared with the wild-type strain (Fig. 5a, P MET3 -ERG20 versus wt). Similarly, Hmg2p-GFP steady-state levels were increased dramatically in the P MET3 -ERG20 strain (Fig. 5b, P MET3 -ERG20 versus wt), and this effect was mimicked by growth of the wild-type strain in the presence of lovastatin (P MET3 -ERG20 compared with wt,ϩLov), which acts to slow Hmg2p degradation (6). The effect of FPP synthase down-regulation was reversed by the presence of ZA in the degradation assay (Fig. 5a, P MET3 -ERG20, 4ZA lane), and in Hmg2p-GFP steady-state fluorescence (Fig. 5b, P MET3 -ERG20,ϩZA compared with wt,ϩZA), indicating that regulated degradation of Hmg2p was still operative. Although the methionine treatment halted the growth of the P MET3 -ERG20 strain after 15 h (see below, Fig. 8b), if these cells were then transferred to media that did not contain methionine, they retained a similar plating efficiency 4 and growth curve as the wild-type ERG20 control cells (see below, Fig. 8b), indicating that Hmg2p stabilization was not a result of inadvertent cell death.
Thus, Hmg2p stabilization was caused by FPP synthase down-regulation, further strengthening the model that FPP was the source of a positive signal for Hmg2p degradation.
FPP Synthase Down-regulation Also Blocked Hmg2p Ubiquitination-The covalent attachment of ubiquitin is a critical and regulated step in Hmg2p degradation (12). Because FPP synthase down-regulation stabilized Hmg2p, we also deter- FIG. 4. Squalene synthase overexpression decreased Hmg2p degradation. Effect of squalene synthase overexpression on 1Myc-Hmg2p degradation. Otherwise identical strains expressing squalene synthase from the wild-type promoter (wt) or from a single, integrated allele with the GAPDH promoter (P GAPDH -ERG9) were subjected to a cycloheximide chase assay. Lysates for each indicated time point after addition of cycloheximide were made and immunoblotted to determine Hmg2p levels.
FIG. 5. FPP synthase down-regulation decreased Hmg2p degradation. a, effect of FPP synthase down-regulation on 1Myc-Hmg2p degradation. Otherwise identical strains expressing FPP synthase from the wild-type promoter (wt) or from the MET3 promoter (P MET3 -ERG20) were grown for 15 h at 30°C in the presence of 2 mM methionine and then subjected to a cycloheximide chase assay. Lysates for each indicated time point were made and immunoblotted to determine Hmg2p levels. ZA (10 g/ml final concentration) was added to the indicated sample (4ZA) at the same time as addition of cycloheximide. Lovastatin (50 g/ml final concentration) was added to the indicated wild-type strain (wt,ϩLov) at the same time as addition of methionine. b, effect of FPP synthase down-regulation on Hmg2p-GFP steady-state levels. Cells expressing FPP synthase from either the wild-type promoter (wt) or the MET3 promoter (P MET3 -ERG20) were grown for 15 h at 30°C in the presence of 2 mM methionine. ZA (ϩZA, 10 g/ml final concentration) or lovastatin (ϩLov, 25 g/ml final concentration) was added to the appropriate cultures, and all cultures were grown an additional 4 h at 30°C. Hmg2p-GFP fluorescence was analyzed by flow cytometry. mined its effect on Hmg2p ubiquitination. To assay Hmg2p ubiquitination, strains containing the appropriate wild-type or regulated alleles of the FPP synthase gene were transformed with plasmids that expressed HA epitope-tagged ubiquitin. 3 Hmg2p ubiquitination was assayed by immunoprecipitation of 1Myc-Hmg2p, followed by anti-HA immunoblotting to detect covalently attached HA-ubiquitin. 3 Regulation of Hmg2p ubiquitination was assessed by performing each assay in the presence of lovastatin (Lov), which decreases ubiquitination, or zaragozic acid (ZA), which increases ubiquitination (12) (Fig. 6,  wt).
Down-regulation of FPP synthase caused a drastic decrease in the level of Hmg2p ubiquitination (Fig. 6, P MET3 -ERG20 versus wt, no drug lanes). This effect was similar to the addition of lovastatin to the wild-type strain during the ubiquitination assay (wt, Lov lane). Furthermore, the addition of ZA for 10 min during the ubiquitination assay, which normally increases Hmg2p ubiquitination, had no effect on Hmg2p ubiquitination in the FPP synthase down-regulated strain (wt ZA lane versus P MET3 -ERG20, ZA10 lane). This effect of FPP synthase down-regulation on ZA action was identical to addition of lovastatin to the wild-type cells (P MET3 -ERG20, ZA10 lane compared with wt, ZAϩLov lane). However, addition of ZA for 1 h did increase Hmg2p ubiquitination in the FPP synthase down-regulated strain similar to that in wild-type strain (P MET3 -ERG20, ZA60 lane compared with wt ZA lane). Thus, pharmacological and genetic manipulations that altered the cellular levels of FPP correspondingly altered Hmg2p ubiquitination in a manner consistent with its role as a positive degradation signal.
Genetic Manipulations Did Not Alter General ER Degradation-The above genetic experiments were consistent with FPP being the source of a positive signal for Hmg2p degradation. However, the results could also be explained by effects on the machinery for ER degradation, rather than on the signal con-trolling Hmg2p degradation. To test this, we examined if these genetic manipulations altered the degradation of 6Myc-Hmg2p, a previously described unregulated mutant of Hmg2p (11). Degradation of 6Myc-Hmg2p requires the same degradation pathway as normal Hmg2p (11), but 6Myc-Hmg2p degradation is not regulated by signals from the mevalonate pathway. Thus, if down-regulation of mevalonate pathway enzymes altered only the regulatory signal for Hmg2p degradation, then 6Myc-Hmg2p degradation should be unaffected.
The effect of squalene synthase or FPP synthase down-regulation on 6Myc-Hmg2p degradation was determined in a strain that was isogenic to the wild-type strain described above, except that unregulated 6Myc-Hmg2p and 6Myc-Hmg2p-GFP were co-expressed rather than 1Myc-Hmg2p and Hmg2p-GFP. As shown in the top left panel of Fig. 7, 6Myc-Hmg2p-GFP steady-state levels were unaffected by addition of drugs, such as lovastatin or ZA, which normally altered the steady-state levels of Hmg2p-GFP (top right panel). Similarly, down-regulation of squalene synthase or FPP synthase had no effect on the steady-state levels of the unregulated 6Myc-Hmg2p-GFP (bottom left panel), whereas these perturbations appropriately altered the steady-state levels of the regulated Hmg2p-GFP (bottom right panel). The co-expressed 6Myc-Hmg2p was similarly unresponsive to the same pharmacological or genetic manipulations that affected 1Myc-Hmg2p or the optical Hmg2p-GFP reporter. 4 These results indicated that squalene synthase and FPP synthase down-regulation affected the regulation of Hmg2p stability and not the process of ER degradation.
Squalene Epoxidase Down-regulation Did Not Alter Hmg2p Degradation-The previous experiments provided strong evidence that FPP levels within the cell controlled the rate of Hmg2p degradation. However, it was not clear if this feature was unique to FPP or if other downstream products could also affect Hmg2p degradation. Squalene epoxidase is the next enzyme after squalene synthase in the pathway and is responsible for the formation of squalene epoxide from squalene (Fig.  1a). We wondered if altering squalene epoxidase levels would have any effect on Hmg2p degradation. Therefore, we generated a MET3-regulated allele of the squalene epoxidase gene (ERG1), similar to the ERG9 and ERG20 alleles described above. When grown in the absence of methionine, normal regulated Hmg2p degradation was observed in the yeast strain that expressed squalene epoxidase from the MET3 promoter (P MET3 -ERG1). 4 Unlike the other manipulations above, squalene epoxidase down-regulation had no effect on Hmg2p degradation. When grown in methionine, the P MET3 -ERG1 strain and the wild-type strain had identical Hmg2p-GFP steady-state levels (Fig. 8a, P MET3 -ERG1 compared with wt), and 1Myc-Hmg2p degradation. 4 This was in contrast to squalene synthase down-regulation (see above and Fig. 8a, P MET3 -ERG1 versus P MET3 -ERG9), which had the expected effects in the same experiment.
It was possible that squalene epoxidase down-regulation had no effect on Hmg2p degradation because the conditional allele was not correctly functioning. However, cells carrying the ineffective P MET3 -ERG1 allele showed identical methionine-dependent growth retardation as the other alleles that affected degradation (Fig. 8b). Thus, the conditional allele of squalene epoxidase was correctly integrated and down-regulated. To confirm further that a block at squalene epoxidase did not affect Hmg2p degradation, we treated the wild-type strain with terbinafine, a specific inhibitor of fungal squalene epoxidase (35)(36)(37)(38). No concentration of terbinafine had any observable effect on Hmg2p-GFP steady-state levels (Fig. 8c, ϩTb compared with no drug) or 1Myc-Hmg2p degradation, 4 despite examining the effect with lethal doses of the drug. Together these results indicated that FPP was the primary, if not the only, source of the positive signal for Hmg2p degradation. DISCUSSION HMGR degradation is regulated by signals generated downstream in the mevalonate pathway. The identity of these signals and the mechanism by which they control HMGR degradation is currently not known. To understand better the nature of this regulatory mechanism, we have examined the regulated degradation of the yeast HMGR isozyme Hmg2p through genetic and pharmacological manipulation of the mevalonate pathway.
The results detailed in this work strongly implicated the mevalonate pathway product FPP as the source of the positive signal for Hmg2p degradation. ZA addition or squalene synthase down-regulation, both predicted to increase FPP levels in the cell, increased Hmg2p degradation. Conversely, addition of lovastatin, FPP synthase down-regulation, or squalene synthase overexpression, all predicted to decrease FPP levels in the cell, stabilized Hmg2p. Identical Hmg2p steady-state levels in degradation-deficient strains containing either normal or down-regulated squalene synthase levels indicated that FPP levels affected only Hmg2p degradation.
Utility of the MET3 Promoter-In order to examine the molecular signals for Hmg2p degradation, we required a way to manipulate the mevalonate pathway at any enzymatic step. Drugs that inhibit some of the pathway enzymes are available, such as lovastatin and ZA. However, there are no currently available drugs that inhibit enzymes between HMGR and squalene synthase. Furthermore, because the products of the mevalonate pathway are essential for yeast viability, null alleles of mevalonate pathway genes result in cell death. Therefore, we required a way to alter the level or activity of mevalonate pathway enzymes in a controllable, yet viable, manner.
To accomplish this, we manipulated the levels of key mevalonate pathway enzymes by placing them under the control of a regulated promoter. We chose the MET3 promoter (22), which can be repressed by incubation of cells in high concentrations of methionine (Ͼ0.5 mM) (23). In all cases, the conditional alleles were similar to null alleles in that cells carrying the conditional alleles, when continuously incubated in methionine, were unable to grow after a few doublings. However, unlike cells with null alleles, cells carrying the conditional alleles could be induced to grow by transferring them to media containing no methionine. This type of genetic manipulation provided a facile way to alter the expression of target genes and could be used, in theory, to create regulated alleles of any yeast gene.
FPP as the Source of the Signal for Hmg2p Degradation-Manipulation of FPP levels by both pharmacological and genetic means resulted in the expected changes to Hmg2p degradation consistent with FPP, or a derivative, as a positive signal for Hmg2p degradation. The idea that intracellular levels of FPP, or a derivative, serve to modulate Hmg2p degradation in yeast paralleled similar observations in mammalian cells. It has been proposed that degradation of mammalian HMGR is regulated by the intracellular levels of farnesol, a derivative of FPP (13)(14)(15)(16)(17). It may also be the case that farnesol is the positive signal for Hmg2p degradation in yeast, and we are currently in the process of examining whether FPP or farnesol regulates Hmg2p degradation.
One way to distinguish between FPP and farnesol as the FIG. 8. Squalene epoxidase down-regulation or inhibition had no effect on Hmg2p-GFP steady-state levels. a, effect of squalene epoxidase down-regulation on Hmg2p steady-state levels. Cells expressing squalene epoxidase from either the wild-type promoter (wt) or the MET3 promoter (P MET3 -ERG1) or squalene synthase from the MET3 promoter (P MET3 -ERG9) were grown for 15 h at 30°C in the presence of 2 mM methionine. Hmg2p-GFP fluorescence was analyzed by flow cytometry. b, left panel, cells containing either the conditional alleles of squalene synthase (P MET3 -ERG9), FPP synthase (P MET3 -ERG20), or squalene epoxidase (P MET3 -ERG1) were grown to mid-log phase (A 600 approximately 0.5) in media that did not contain methionine. New cultures containing 2 mM methionine were inoculated from these previously grown cultures to an initial A 600 of 0.01 and then incubated for 32 h at 30°C. The new A 600 for these cultures was measured at the indicated time points and plotted versus time. Growth of each strain was compared with the wild-type strain (wt). Right panel, new cultures, which contained no methionine, were inoculated from the previous 15-h methionine-grown cultures to an initial A 600 of 0.01 and then incubated for 32 h at 30°C. The new A 600 for these cultures was measured at the indicated time points and plotted versus time. c, effect of terbinafine (Tb), a squalene epoxidase inhibitor, on Hmg2p-GFP steady-state levels. Cells were treated with either no drug, 40 g/ml terbinafine (Tb), or 10 g/ml ZA (ZA) and grown for 4 h at 30°C. Hmg2p-GFP steady-state fluorescence was analyzed by flow cytometry. regulatory signal would be by elimination of the pyrophosphatase activity required to convert FPP to farnesol (17). Recently, two pyrophosphatases, LPP1 and DPP1, that appear to convert FPP to farnesol in vitro have been described (39). Null alleles of the genes encoding these enzymes have been made in yeast, and the cells are viable. We examined Hmg2p degradation in these strains and found that the absence of Lpp1p and Dpp1p had no effect on Hmg2p regulated degradation. 4 This could mean that either FPP, not farnesol, is the signal for HMGR degradation in yeast or that these enzymes are not solely responsible for the conversion of FPP to farnesol in vivo. In either case, it is apparent that FPP serves as the source of a signal for HMGR degradation in both mammals and yeast.
In mammalian cells, there is an additional sterol signal that acts to modulate HMGR degradation. The addition of sterols to mammalian cells results in increased degradation of HMGR (4,5). However, this downstream signal from sterols does not accelerate HMGR degradation in the absence of the upstream FPP-derived signal (4,17), indicating that sterols provide an additional positive signal for HMGR degradation that works only in conjunction with the FPP-derived signal. In this study, our results demonstrated that no other pathway product downstream of FPP was required for Hmg2p degradation in yeast. That is inhibition of squalene epoxidase activity or down-regulation of squalene epoxidase had no stabilizing effect on Hmg2p degradation. However, yeast may contain abundant stores of sterols, and so the existence of other signals cannot be ruled out by these negative results. We are further exploring, by both genetic and pharmacological means, whether an additional signal downstream of squalene can act in conjunction with FPP to hasten Hmg2p degradation.
Mechanisms by Which FPP Regulates Hmg2p Degradation-The identification of FPP, or a derivative, as the positive signal for HMGR degradation in yeast leads to several models for regulation of Hmg2p degradation. The simplest model for the regulatory mechanism is that the FPP-derived signal acts as an allosteric regulator and physically binds to the transmembrane domain of Hmg2p, altering its susceptibility to degradation. Alternatively, the FPP-derived molecule could modify or interact with a separate effector protein that alters Hmg2p stability. The effector could act to promote or prevent Hmg2p degradation, as either of these possibilities have precedent in cellular degradation (40,41). It is also possible that the FPP-derived signal directly affects the structure of the ER membrane. By this model, the Hmg2p transmembrane domain would respond to the altered membrane by becoming increasingly susceptible to degradation. In this regard, it is interesting that the distance from the ER surface of key residues in the Hmg2p transmembrane is critical in regulation of Hmg2p stability. 3 We are currently exploring the mechanism by which FPP acts to regulate Hmg2p degradation. It is clear that both yeast and mammalian cells measure FPP production as a means to control HMGR degradation. This conservation in signaling strategy, together with the similar location and machinery required for HMGR degradation, indicates that the mechanism for regulation of HMGR degradation may also be conserved among eukaryotes.
A final and important note should be made regarding a recent study on a fascinating side effect of the HMGR inhibitor lovastatin. It has been demonstrated that the closed ␤-lactone ring form of this particular statin can directly inhibit the proteasome (29). At first thought, it might be reasonable to conclude that this action is responsible for the stabilizing effects of lovastatin on Hmg2p, which is degraded by the proteasome (11). However, several observations show quite clearly that all actions of lovastatin we have observed were due to altered signal production and not proteasome inhibition. Degradation of 6Myc-Hmg2p, which is proteasome-dependent, is unaffected by doses of lovastatin that slow Hmg2p degradation (Ref. 11 and Fig. 7). Furthermore, the pathway inhibitor L659,699, which inhibits HMG-CoA synthase but contains no lactone moiety, is equally effective at stabilizing Hmg2p as lovastatin (12). Third, genetic manipulations of HMG-CoA synthase (6), or FPP synthase (this work), which slow production of the degradation signals but do not involve drugs, have identical actions to lovastatin on Hmg2p degradation but not 6Myc-Hmg2p degradation. Finally, lovastatin decreases ubiquitination of Hmg2p (Ref. 12 and this study), an effect opposite of that observed when the proteasome is compromised or inhibited (11). Although proteasome inhibition by the closed ring form of lovastatin is interesting and perhaps important clinically, it plays no obvious or important role in the actions of this drug on regulating Hmg2p stability in yeast and is most likely not involved in mammalian HMGR degradation, for similar reasons.