Loss of function of 3-hydroxy-3-methylglutaryl coenzyme A reductase 1 (HMG1) in Arabidopsis leads to dwarfing, early senescence and male sterility, and reduced sterol levels


For correspondence (fax +81 45 503 9650; e-mail muranaka@postman.riken.jp).

Present address: Japan Women's University, Bunkyo-ku, Tokyo 112-8681, Japan.

Present address: Institute for Advanced Sciences, Keio University, Tsuruoka, Yamagata 997-0035, Japan.

These authors contributed equally to this work.


3-Hydroxy-3-methylglutaryl-CoA reductase (HMGR) catalyzes the first committed step in the cytosolic isoprenoid biosynthesis pathway in higher plants. To understand the contribution of HMGR to plant development, we isolated T-DNA insertion mutants for HMG1 and HMG2. The hmg1 and hmg2 mutants were both more sensitive than the wild type (WT) to lovastatin, an inhibitor of HMGR. The hmg2 mutant showed no visible phenotype under normal growth conditions. In contrast, the hmg1 mutant exhibited dwarfing, early senescence, and sterility. Expression of senescence-associated genes 12 (SAG12), a marker gene for senescence, was induced in the hmg1 mutant at an earlier stage than in the WT. Levels of trans-cytokinins – hormones known to inhibit senescence – were not lower in hmg1. The mutant did not have the typical appearance of brassinosteroid (BR)-deficient mutants, except for a dwarf phenotype, because of the suppression of cell elongation. The expression of several genes involved in cell elongation was suppressed in hmg1. WT plants treated exogenously with inhibitors of sterol biosynthesis had similar gene expression and sterility characteristics as the hmg1 mutants. Pleiotropic phenotypes were rescued by feeding with squalene, the precursor of sterols and triterpenoids. The sterol levels in hmg1 mutants were lower than in the WT. These findings suggest that HMG1 plays a critical role in triterpene biosynthesis, and that sterols and/or triterpenoids contribute to cell elongation, senescence, and fertility.


Plants produce a wide variety of isoprenoid compounds. All plant isoprenoid compounds are derived from a common precursor, isopentenyl diphosphate (IPP), which is biosynthesized from acetyl-CoA by the cytoplasmic mevalonate (MVA) pathway and from pyruvate and glyceraldehyde-3-phosphate by the plastid 2-C-methyl-d-erythritol-4-phosphate (MEP) pathway. Sesquiterpenes, triterpenes, sterols, and brassinosteroids (BRs) are biosynthesized via the MVA pathway (Newman and Chappell, 1999) (Figure 1), whereas gibberellins, abscisic acid, carotenoids, and chlorophyll side chains are biosynthesized via the MEP pathway (Lichtenthaler, 1999). In mammals, 3-hydroxy-3-methylglutaryl-CoA reductase (HMGR) is the rate-limiting enzyme in the MVA pathway, and inhibition of HMGR strongly reduces the biosynthesis of cholesterol. The level of HMGR in mammals is highly regulated at the transcriptional and post-translational levels (Goldstein and Brown, 1990).

Figure 1.

Overview of phytosterols biosynthetic pathways.

The MEP pathway, localized to plastids, is shown in a box. Genes mentioned in the text are in bold and italics, and inhibitors used in this work are underlined. Arrows with dashed lines indicate metabolic flow between the cytosol and plastids. MVA, mevalonic acid; IPP, isopentenyl diphosphate; DMAPP, dimethylallyl diphosphate; GPP, geranyl diphosphate; FPP, farnesyl diphosphate; GAP, glyceraldehyde 3-phosphate; DXP, 1-deoxy-d-xylulose 5-phosphate.

Several plant HMGRs have been characterized. Two genes encoding HMGR, HMG1 (Caelles et al., 1989; Learned and Fink, 1989) and HMG2 (Enjuto et al., 1994), exist in the Arabidopsis thaliana genome. Although the two encoded proteins (HMGR1 and HMGR2) have the same structural organization and intracellular location, expression profiles of these genes are different. The HMG1 mRNA can be detected in all Arabidopsis tissues (Enjuto et al., 1994), but HMG2 is expressed exclusively in meristematic and floral tissues (Enjuto et al., 1995). In addition, the expression of HMG1 is suppressed by light (Learned, 1996; Learned and Connolly, 1997). Although transgenic plants that overexpress HMG1 are less sensitive to lovastatin and have greater HMGR activity, these plants do not have an altered morphology or isoprenoid content (Re et al., 1995).

The physiological functions of plant HMGRs have been characterized using specific inhibitors, such as lovastatin or mevinolin. HMGR is essential for cytokinin biosynthesis in tobacco Bright Yellow-2 cultured cells, and inhibition of HMGR reduces cytokinin content and cell proliferation activity (Crowell and Salaz, 1992; Hemmerlin and Bach, 1998; Laureys et al., 1998; Miyazawa et al., 2002). Arabidopsis seedlings grown in the presence of lovastatin display inhibited cell elongation and reduced expression of many cell elongation-related genes (Suzuki et al., 2003), but the expression of genes involved in isoprenoid biosynthesis is not significantly affected in these plants (Laule et al., 2003; Suzuki et al., 2003). Moreover, the MVA and MEP pathways are not independent, as metabolic flow occurs between these pathways (Kasahara et al., 2002; Nagata et al., 2002).

In spite of these findings, the roles of HMGRs in plant development are only partially understood as no molecular genetic analyses of plant HMGRs have been performed. In addition to the two HMGR genes in Arabidopsis, tomato has four HMGR genes (Daraselia et al., 1996) and potato has three (Korth et al., 1997), and it is difficult to distinguish the functions of different HMGR isoenzymes. This is in sharp contrast with mammalian systems, in which the HMGR enzyme is encoded by a single gene. Molecular genetic analyses should be useful in elucidating the functions of multiple HMGR genes in plant development. We report here the isolation and characterization of Arabidopsis lines with T-DNA inserts in HMG1 and HMG2. The hmg1 mutant shows early senescence and sterility, as well as a dwarf phenotype. Surprisingly, these phenotypes are not related to cytokinins and BRs, which are phytohormones biosynthesized via the MVA pathway, but are instead derived from reduced sterol content. The roles of sterols and triterpenoids in plant development are discussed.


The hmg1 mutant shows dwarfing, early senescence, and sterility

Two mutant alleles of HMG1 and one mutant allele of HMG2 were identified using a PCR-based strategy with T-DNA-tagged Arabidopsis mutant pools generated at the University of Wisconsin (for hmg1-1 and hmg1-2) and at the Kazusa DNA Research Institute (for hmg2-1). T-DNA sequences were inserted in the first exon in hmg1-1, at the junction between the first exon and the first intron in hmg1-2, and in the third exon in hmg2-1 (Figure 2a). RNA gel blot analysis was used to investigate whether these T-DNA insertion mutants are null mutants. The HMG1 transcript was not detected in hmg1-1, but the HMG2 transcript was detected in hmg2-1 as a species longer by 1 kb than that in the wild type (WT) (Figure 2b). Therefore, hmg1-1 is a null mutant, but hmg2-1 is not. However, as the catalytic domain of HMGR2 is located in the C-terminal half of the protein (Enjuto et al., 1994), it is possible that the HMGR2 mutant polypeptide in hmg2-1 loses its HMGR activity.

Figure 2.

Isolation of hmg1 and hmg2 mutants.

(a) Structure of HMG1 and HMG2 T-DNA insertion alleles. Exons are indicated by black boxes. The T-DNA insertion site in each allele is indicated by a flag.

(b) RNA gel blot analyses of HMG1 and HMG2. Total RNA was extracted from WT and hmg1-1 seedlings for detection of HMG1 and from WT and hmg2-1 inflorescences for detection of HMG2. rRNAs of these samples, stained ethidium bromide, are shown.

The root length and the cotyledon size of 1-week-old hmg1-1 seedlings were smaller than those of the WT. These phenotypes were similar to those of the WT seedlings treated with low concentrations of the HMGR-specific inhibitor lovastatin (Figure 3a). hmg1-1 seedlings grown in the presence of lovastatin had a more severe phenotype than did the WT (Figure 3a).

Figure 3.

hmg1 displays dwarf, early senescence and sterility phenotypes.

(a) One-week-old seedlings of WS, hmg1-1, hmg2-1, and Col grown with or without lovastatin. Lovastatin concentrations in the medium are indicated to the left.

(b) Dark-grown 1-week-old seedlings of WS, hmg1-1, det2, and Col.

(c) Five-week-old plants of WT and hmg1-1 plants.

(d) Longitudinal sections of the first internode of WT (left) and hmg1-1 (right) inflorescence stem.

(e) Five-week-old hmg1-1 plants. The edges of rosette leaves of hmg1-1 were visibly yellow.

(f) An hmg1-1 inflorescence. Sterile siliques are indicated by arrows.

(g) Split immature siliques of hmg1-1 heterozugous (upper panel) and homozygous plants (lower panel).

(h) Micrographs of aniline-blue-stained pollen on pistils 18 h after pollination. The genotypes of male and female gametes are indicated.

(i) Histochemical analysis of GUS activity in an inflorescence of a transgenic Arabidopsis plants carrying the chimeric gene of the 2.4 kbp HMG1 promoter sequence and GUS gene.

Bars indicate 1 cm in (a) and (b), 5 cm in (c), 100 µm in (d), 500 µm in (g), 100 µm in (h), and 1 mm in (i).

In mammals, the expression of HMGR is increased in HMGR-inhibitor-treated cells (Brown et al., 1978). As hmg1-1 is a loss-of-function mutant of HMG1, feedback regulation of HMG2 or enzymes in the MEP pathway may exist. To investigate this possibility, the expression of HMG2 or 1-deoxy-d-xylulose-5-phosphate (DXP) synthase (DXS), DXP reductoisomerase (DXR), and CDP-methylerythritol synthase (CMS) in the MEP pathway (Rodríguez-Conceptión and Boronat, 2002) in hmg1-1 seedlings was examined using RT-PCR. As shown in Figure 4(a), however, the expression of these genes in hmg1-1 seedlings was not affected.

Figure 4.

Gene expression analyses of hmg1.

(a) RT-PCR comparison of the expression of HMG2, DXS, DXR, and CMS. EF-1α transcripts were amplified as a control.

(b) RT-PCR comparison of the expression of cell elongation-related genes, XTR9, and extensin-like protein and a cell cycle regulator gene AtSKP1. EF-1α transcripts were amplified as a control.

(c) SAG12, senescence marker gene, was examined by RT-PCR analysis. Ethanol (lanes 1 and 2), MVA (final concentration 3 mm in lane 3), c-ZA (1 µm in lane 4 and 10 µm in lane 5), and t-ZA (1 µm in lane 6 and 10 µm in lane 7) were added to the medium at the time of transfer to darkness. EF-1α transcripts were amplified as a control.

Mature hmg1-1 plants had short stems, bushy inflorescences, and small leaves (Figure 3c). The height of hmg1-1 plants was approximately one-third that of the WT. In order to understand the cellular basis of the dwarf phenotype of hmg1-1, we observed longitudinal sections of the first internodes of WT and hmg1-1 inflorescence stems. The cell size was smaller in hmg1-1 stems than in WT stems (Figure 3d), indicating that the dwarf phenotype of hmg1-1 is at least partly because of reduced cell elongation. We also examined whether the expression of genes that are potentially involved in cell division and cell elongation was affected in hmg1-1. The Xyloglucan endotransglycosylase-related 9 (XTR9) (Yokoyama and Nishitani, 2001) and extensin-like protein (GenBank Accession no. BT006378) genes, thought to be involved in cell elongation, are downregulated by lovastatin treatment (Suzuki et al., 2003). The expression levels of these genes were lower in hmg1-1 than in the WT (Figure 4b). However, the expression of AtSKP1, which is a homolog of a cell cycle regulator gene and is used as a marker for cell division (Porat et al., 1998), was not affected by the hmg1-1 mutation.

At approximately 5 weeks of culturing, yellowing appeared at the edges of hmg1-1 rosette leaves (Figure 3e). No yellowing was observed in rosette leaves of WT plants of the same age (data not shown). Despite this early senescence phenotype for hmg1-1, the mutant plants continued to produce flowers at 7 weeks of age, at a time when WT plants have generally ceased flowering, indicating a prolonged life span for this mutant. The SAG12 gene, which encodes a cysteine protease, is expressed late in natural senescence (Lohman et al., 1994; Weaver et al., 1998), which makes it useful as a senescence marker gene. To investigate whether the expression of SAG12 is induced earlier in hmg1-1 than in the WT, RT-PCR analysis was performed. As the expression of SAG12 is enhanced in the dark (Weaver et al., 1998), total RNA was extracted from 2-week-old seedlings after 2 days of dark treatment. Although no SAG12 expression was detected in WT plants, the gene was highly expressed in hmg1-1 plants of this age (Figure 4c, lane 2). The high SAG12 expression in hmg1-1 was suppressed by treatment with MVA, the product of HMGR (Figure 4c, lane 3).

In addition, hmg1-1 siliques were markedly shorter than those of the WT, and hmg1-1 plants were infertile (Figure 3f (arrows), g). As hmg1-1 homozygous seeds could be harvested from hmg1-1 heterozygous plants, but not homozygous plants, it appeared that the sterility of hmg1-1 was derived from a parental defect. Most of the seeds produced from heterozygous plants had a normal appearance and were viable (Figure 3g), unlike seeds from plants with typical embryonic lethal mutations. To investigate whether a defect in the male or the female gametes was responsible for the sterility of hmg1-1 homozygous plants, we crossed hmg1-1 homozygous and WT plants. When pollinated with WT pollen, hmg1-1 plants were able to produce seeds, but WT plants pollinated with hmg1-1 pollen were unable to produce seeds. Thus, the hmg1-1 mutant exhibits typical male sterility. Pollen tubes of these plants were stained with aniline blue 18 h after pollination and observed with a fluorescence microscope. When WT pollen was used, pollen tubes germinated and penetrated into the stigma, but in pollinations with hmg1-1 pollen, no pollen tubes germinated (Figure 3h). These results suggest that hmg1-1 pollen grains have an abnormality that is caused by a defect in the parental cells, most likely the tapetum. Histocemical GUS staining of transgenic plants carrying a chimeric HMG1 promoter::β-glucuronidase fusion gene showed that HMG1 is expressed in anthers (Figure 3i). These findings are consistent.

The overall morphology of hmg1-2 plants was similar to that of hmg1-1 plants (data not shown). No distinctive phenotypic differences from the WT were observed in hmg2-1 plants under normal growth conditions. However, hmg2-1 seedlings grown in the presence of lovastatin showed a more severe dwarf phenotype than did the WT grown with lovastatin (Figure 3a). This means that HMGR2 is also a functional HMGR and that the hmg2-1 mutant is a loss-of-function allele of HMG2. To examine whether feedback regulation of HMG1 exists in hmg2-1, RT-PCR analysis was performed. The expression of HMG1 was not affected in hmg2-1 seedlings (data not shown).

hmg1-1 did not show the typical appearance of BR-deficient mutants

Brassinosteroids, including brassinolide, are biosynthesized by the MVA pathway (Figure 1). As the hmg1-1 mutant shows a dwarf phenotype because of reduced cell elongation, it is postulated that hmg1-1, like deetiolated 2 (det2), is a BR-deficient mutant (Fujioka et al., 1997). In addition to the dwarf phenotype, the most typical phenotype found in BR-deficient mutants is photomorphogenesis in the dark (Clouse, 1997). As shown in Figure 3(b), det2 mutants grown in the dark have short hypocotyls and opened cotyledons. By contrast, when hmg1-1 mutants were grown under the same conditions, the length of the hypocotyl and root were comparable to those of the WT and the cotyledons were not opened (Figure 3b).

The levels of trans-cytokinins were not lower in the hmg1-1 mutant

Cytokinins are a class of hormones synthesized from IPP and dimethylallyl diphosphate (DMAPP), and are strong antagonists of senescence (Gan and Amasino, 1995; Ori et al., 1999). As a first approach to understanding the mechanisms of early senescence in hmg1-1, we examined the effect of exogenously supplied cis-zeatin (c-ZA) and trans-zeatin (t-ZA) on hmg1-1. Although 10 µmt-ZA, which is a bioactive cytokinin, repressed SAG12 expression in hmg1-1 (Figure 4c, lane 7), the same concentration of c-ZA, which is known to be inactive, did not suppress but slightly enhanced expression of this gene (Figure 4c, lane 5). A low t-ZA concentration did not affect SAG12 expression in hmg1-1 (Figure 4c, lane 6). Despite expectations that the t-ZA levels would be lower in hmg1-1 than in the WT, endogenous trans-cytokinin levels were higher in mutant plants than in the WT (Table 1). WT plants treated with lovastatin had a cytokinin profile similar to that of hmg1-1 (Kato et al., unpublished data). These results suggest that cytokinins are not involved in the early senescence of hmg1-1.

Table 1.  Quantification of endogenous cytokinins (ng g−1 FW)
  1. Two-week old seedlings grown in liquid culture were harvested and frozen in liquid nitrogen. The samples were then analyzed by LC–MS/MS (Miyazawa et al., 2002). The values show an average of three samples. Parenthesis shows standard error. n.d., not detected.

t-ZA0.121 (0.011)0.187 (0.012)
c-ZR0.951 (0.076)0.687 (0.063)
t-ZR1.38 (0.12)3.96 (0.13)

Exogenous application of inhibitors of enzymes in the MVA pathway mimics the hmg1 mutation

To investigate which metabolites are involved in the phenotypes of the hmg1 mutants, we performed gene expression analyses using inhibitors of enzymes in the MVA pathway, such as lovastatin; squalestatin, an inhibitor of squalene synthase (Hartmann et al., 2000); and brz2001, an inhibitor of steroid 22-alpha-hydroxylase (Asami et al., 2001; Sekimata et al., 2001; Figure 1). WT seedlings were germinated on agar plates containing 3% sucrose, grown in a 16-h light/8-h dark photoperiod for 8 days, transferred to liquid medium containing 1% sucrose, and cultured under the same photoperiod for 3 days. At that point, lovastatin, squalestatin, or brz2001 was added to the medium and the seedlings were cultured for an additional 3 days. The seedlings were then grown in darkness for 2 days, after which total RNA was extracted. SAG12 expression was detected in seedlings treated with lovastatin and squalestatin, but not in seedlings treated with brz2001 (Figure 5). Expression of XTR9 and extensin-like protein was decreased in seedlings treated with lovastatin or squalestatin, but was not affected in seedlings treated with brz2001 (Figure 5). Moreover, WT plants treated with squalestatin were infertile (Figure 6a), but WT plants treated with brz2001 remained fertile (data not shown). These results are consistent with the fact that det2 mutants are fertile.

Figure 5.

Gene expression analyses of WT plants treated with inhibitors of enzymes in the MVA pathway.

Influence of lovastatin, squalestatin, and brz2001 on the expression of SAG12, XTR9, and extensin-like protein. EF-1α transcripts were amplified as a control.

Figure 6.

Squalene rescues the hmg1 mutation.

(a) From left to right: inflorescences of WT, hmg1-1, squalene-treated hmg1-1, and squalestatin-treated WT plants. Fertilized siliques of squalene-treated hmg1-1 plants are indicated by yellow arrows. Sterilized siliques of squsalestatin-treated WT plants are indicated by red arrows.

(b) Squalane (left)- and squalene (right)-treated hmg1-1 plants. Bar indicates 5 cm.

Squalene, the precursor of sterols and triterpenoids, rescues the hmg1 mutation

The above studies that used inhibitors of enzymes in the MVA pathway suggested that the pleiotropic phenotypes of the hmg1 mutants are derived from a deficiency in sterols and/or triterpenoids. We therefore examined the inflorescences of hmg1-1 plants exogenously supplied with MVA pathway intermediates. Treatment with MVA (data not shown) and squalene (Figure 6a,b) rescued the pleiotropic phenotypes of hmg1-1, while treatment with brassinolide did not rescue these phenotypes (data not shown). It has been reported that the fertility of some mutants with reduced fertility can be restored in conditions of high humidity (Chen et al., 2003). To eliminate the possibility that the restored fertility was because of the high viscosity of the compounds applied, squalane, a chemically hydrogenated compound of squalene, was applied to hmg1-1. Squalane had no effect on the pleiotropic phenotypes of hmg1-1 (Figure 6b).

Decreased sterol content in the hmg1 mutant

To investigate the role of HMG1 in sterol biosynthesis, the sterol contents of hmg1-1 and the WT were compared. Total sterols were extracted from 2-week-old seedlings and inflorescences of hmg1-1 and the WT (Table 2). The mass of sterols in hmg1-1 seedlings was approximately 47% lower than in WT seedlings, and hmg1-1 inflorescences had approximately 25% less sterols by mass than WT inflorescences. Only the level of cholesterol in the mutant was not much affected.

Table 2.  Quantification of endogenous sterols (microgram per 100 mg DW)
Major sterolSeedlingInflorescence
  1. Two-week-old seedlings and inflorescences of 5-week-old plant were harvested and freeze-dried. The samples were then analyzed using GC–MS as described in Experimental procedures. The values show an average of three samples. Parenthesis shows standard error. Trace, detected but under quantifiable limitations.

CycloartenolTraceTrace3.82 (0.75)2.28 (0.45)
Cholesterol26.5 (4.3)21.2 (10)28.2 (2.2)26.8 (1.3)
Sitosterol200 (11)93.3 (5.6)250 (7.7)178 (3.7)
Stigmasterol11.6 (0.79)7.04 (0.41)3.04 (0.27)0.89 (0.28)
Campesterol49.3 (3.1)33.2 (1.7)69.2 (2.4)58.7 (1.4)
24-Methylenecholesterol4.73 (0.75)1.63 (0.12)3.92 (0.17)1.67 (0.08)
Campestanol1.30 (0.10)0.51 (0.05)1.14 (0.10)0.89 (0.25)
Sitostanol4.65 (0.28)1.89 (0.17)3.45 (0.21)2.72 (0.18)
Total sterol298 (18)159 (4.4)363 (14)272 (4.8)


hmg1-1 is a sterol-deficient mutant

To understand the contributions of HMGR to plant development, we isolated T-DNA insertion mutants for HMG1 and HMG2. hmg1-1 showed dwarfing, early senescence, and sterile phenotypes, but hmg2-1 had no obvious morphological differences under our growth conditions. Analysis of hmg1-1 revealed that these phenotypes were not derived from a decrease in cytokinins, but instead from a decrease in metabolites downstream of squalene, namely sterols and/or triterpenoids.

hmg1-1 had a dwarf phenotype throughout its vegetative and reproductive phases. Microscopic analyses revealed that cell elongation was defective in hmg1-1. As hmg1-1 was a dwarf mutant that was defective in the MVA pathway and this was because of the inhibition of cell elongation, it was postulated that the dwarf phenotype was derived from a deficit in BRs. However, hmg1-1 seedlings that germinated in darkness did not show the photomorphogenic phenotype characteristic of BR-deficient mutants (Figure 3b). Moreover, the dwarf phenotype was scarcely rescued by exogenously applied BR. Furthermore, the gene expression profile in hmg1-1 is different from that of the BR-deficient mutant. We previously examined the gene expression in lovastatin-treated Arabidopsis plants using microarray analysis (Suzuki et al., 2003). Among the genes downregulated by lovastatin, there were several cell elongation-related genes, such as XTR9 (Yokoyama and Nishitani, 2001) and extensin-like protein. The expression of both XTR9 and extensin-like protein was downregulated in hmg1-1 (Figure 4b). These genes are not included in the list of BR-regulated genes (Goda et al., 2002). Experiments with inhibitors of enzymes in the MVA pathway showed that a decrease in isoprenoids upstream of campestanol and downstream of squalene regulated the expression of these genes (Figure 5). Although XTR6, a homolog of XTR9, is upregulated by BR (Goda et al., 2002), the expression of XTR6 in hmg1-1 was not different from that in the WT (data not shown). Moreover, sterol methyltransferase 1 (smt1) (Diener et al., 2000) and fackel (fk) (Jang et al., 2000; Schrick et al., 2000; Souter et al., 2002) mutants located downstream from hmg1 mutant in the MVA pathway are not reported to be BR-deficient mutants. Combining this evidence, we considered that hmg1-1 is not a BR-deficient mutant. He et al. (2003) reported that the expression of Touch 4 (TCH4), Meri5, and KORRIGAN (KOR) is regulated not only by BRs but also by sitosterol, stigmasterol, and the atypical fk sterols. This report and our data suggest that sterols other than BRs can regulate cell elongation. This is supported by the fact that hmg1-1 had lower levels of these sterols than did WT plants. It remains unknown whether sterols other than BRs regulate cell elongation through mechanisms similar to those of steroid hormones, or a reduction in the flexibility and permeability of plasma membranes suppresses cell elongation.

Cytokinins, a class of hormones synthesized from IPP and DMAPP, are strong antagonists of senescence (Gan and Amasino, 1995; Ori et al., 1999). Overexpression of farnesyl diphosphate synthase leads to a decrease in endogenous zeatin-type cytokinin levels and induces a cell death/senescence-like response (Masferrer et al., 2002). We postulated that the early senescence phenotype of hmg1-1 is because of the hmg1-1 mutation, causing a decrease in endogenous cytokinin levels. In fact, expression of SAG12, a senescence marker gene that is expressed earlier in hmg1-1 than in the WT, was suppressed by the exogenous bioactive cytokinin t-ZA, but slightly upregulated by exogenous c-ZA, an inactive cytokinin (Figure 4c). Quantification of endogenous cytokinins demonstrated that the hmg1-1 mutation reduces the levels of cis-cytokinins, but increases the levels of trans-cytokinins. Therefore, it was thought that the suppression of the early expression of SAG12 in hmg1-1 by exogenous t-ZA was not because of complementation of the hmg1-1 mutation, but because of the natural effects of t-ZA. The levels of both t-ZA and t-ZR, which is the ribosylated form of t-ZA and is a bioactive cytokinin, were increased in hmg1-1. Application of lovastatin to WT plants also caused a decrease in the levels of cis-cytokinins and an increase in trans-cytokinins (data not shown). Therefore, downregulation of the MVA pathway may lead to a decrease in cis-cytokinins and an increase in trans-cytokinins in Arabidopsis. This is in sharp contrast with our previous report (Miyazawa et al., 2002) in which we used tobacco Bright Yellow-2 cultured cells. These results imply that intricate pathways and regulation exist in cytokinin biosynthesis in both whole plants and cultured cells.

The expression of SAG12 was then examined using inhibitors of enzymes in the MVA pathway. Lovastatin and squalestatin induced expression of SAG12 in WT plants, but brz2001 treatment did not, which suggests that decreased levels of isoprenoids upstream of campestanol and downstream of squalene trigger early senescence in hmg1-1. The sterol levels of the mutant were lower than those of the WT, except for the cholesterol level, which was comparable in the hmg1-1 mutant and WT.

As pollen grains of hmg1-1 were unable to germinate, the sterility of hmg1-1 was male-derived. Fertility-rescue experiments and sterol quantification suggested that the pollen defect of hmg1-1 was derived from reduced levels of isoprenoids downstream of squalene, possibly sterols but not BRs. Microscopic analysis of DAPI-stained pollens of hmg1-1, however, showed that pollens of hmg1-1 have two sperm nuclei and one vegetative nucleus, which are comparable to those of WT (data not shown). Therefore, it was postulated that defect of pollens of hmg1-1 is derived from abnormalities of sterol distributions in anthers that do not affect meiosis or morphology of the pollen grains. Tapetal cells contain abundant sterol compounds and lipidic organelles, such as elaioplasts, which are plastidial lipid body structures enriched in sterol esters (Hernández-Pinzón et al., 1999; Piffanelli and Murphy, 1998). Various compounds are released from tapetal cells and are transferred to the developing pollen grains to provide nutrients for pollen development and to form the pollen exine and pollen coat. For example, it has been reported that lipids supplied from tapetum cells and that accumulate on pollen coats are essential for pollen tube elongation (Worters-Arts et al., 1998). Similarly, it is possible that sterols or sterol-derived products in tapetum cells are transferred to pollen grains or pollen coats and stimulate the germination of mature pollen grains. Brassica napus pollen coat lipids are also enriched in similar classes of sterol esters in tapetum elaioplasts (Hernández-Pinzón et al., 1999).

Comparison with mutants deficient in sterol biosynthesis

We have shown that the dwarf, early senescence, and sterile phenotypes of hmg1-1 are derived from decreases in compounds downstream of squalene. One possibility is that a decrease in sterol levels causes these phenotypes. Mutants defective in the early steps of the sterol biosynthesis pathway have been reported, such as smt1 (Diener et al., 2000), fk/hydra2 (Jang et al., 2000; Schrick et al., 2000; Souter et al., 2002), hydra1 (Souter et al., 2002), and cotyledon vascular pattern 1 (cvp1)/smt2 (Carland et al., 2002; Schaeffer et al., 2001). Most of these mutants have a dwarf phenotype and low fertility. Although most display abnormal embryogenesis, embryogenesis progresses normally in hmg1-1. One explanation for this is that hmg1-1 does not accumulate atypical sterols, but has reduced total sterol levels, whereas smt1 and fk accumulate atypical sterols. These atypical sterols may cause abnormal embryogenesis. Another explanation is that the distribution of auxin is altered in smt1, fk, and hydra1. Altered auxin distribution causes conspicuous cell polarity defects, possibly affecting cell shape, cell patterning, and embryogenesis in these mutants (Souter et al., 2002; Willemsen et al., 2003). It has been proposed that the distribution of auxin is altered in these mutants because altered membrane sterol levels lead to an alteration in the distribution of membrane proteins (Willemsen et al., 2003). As hmg1-1 displays reduced apical dominance, but no abnormal embryogenesis, the distribution of auxin in hmg1-1 plants may be altered. 35S::SMT2-1 co-suppressed plants show dwarf, reduced apical dominance and low-fertility phenotypes, as does hmg1-1. The campesterol to sitosterol ratio of 35S::SMT2-1 co-suppressed plants was increased by approximately twofold compared with the WT (Schaeffer et al., 2001). By contrast, the ratio in hmg1-1 was increased by only 44% (for seedlings) and 19% (for inflorescences) compared with the WT. Therefore, the causes of the phenotypes of the hmg1-1 mutant and 35S::SMT2-1 co-suppressed plants may be different.

Another possibility is that a decrease in triterpenoids in hmg1-1 causes the pleiotropic phenotypes. 2,3-Epoxysqualene is cyclized to not only cycloartenol, the precursor of phytosterols, but also β-amyrin or lupeol, precursor of triterprenoids. Several oxidosqualene cyclase genes have been reported (Husselstein-Muller et al., 2001; Kushiro et al., 1998; Segura et al., 2000; Shibuya et al., 1999). However, other than their roles as phytoalexins, the physiological roles of triterpenoids have not yet been well characterized. There are no reports of early-senescence phenotypes in mutants that are defective in a sterol biosynthesis pathway. Considering these findings, it is possible that a decrease in triterpenoids rather than a decrease in sterols is responsible for the early senescence of hmg1-1. Our research may provide new insights into the roles of triterpenoids.

The transcriptional and post-translational regulation of HMG1

HMG1 encodes two different HMGR1 isoforms: HMGR1S and HMGR1L (Lumbreras et al., 1995). HMGR1L has 50 more amino acid residues at its N-terminal end than does HMGR1S. Although both HMGR1 isoforms have the same intracellular localization (Lumbreras et al., 1995), it is possible that the stabilities of HMGR1S and HMGR1L differ. Moreover, HMGR is regulated by reversible phosphorylation in mammals (Clarke and Hardie, 1990). It has been reported that plant HMGR is also phosphorylated and inactivated in vitro (Sugden et al., 1999). However, the functional difference between the two HMGR1 isoforms and the mechanism of post-translational regulation of HMGR in vivo remain unsolved. Introducing the HMGR1 isoforms and modified HMGR1 isoforms with mutated putative phosphorylation sites into the hmg1-1 mutant should resolve this problem.

Contribution of HMGR2 and enzymes in the MEP pathway in the hmg1-1 mutant

Although the sterol levels in hmg1-1 were lower than those in the WT, the total sterols were reduced to 53% of those of the WT in seedlings and 75% of the WT in inflorescences. The expression of HMG2 was much lower than that of HMG1 (Figure 2b; Enjuto et al., 1994) and the expression of HMG2 in hmg1-1 was not affected. Whereas HMG1 mRNA is detected in all tissues, HMG2 is expressed only in meristematic and floral tissues (Enjuto et al., 1994). Therefore, HMG2 may complement the lost function of HMG1 in hmg1-1 slightly. IPP is biosynthesized via not only the MVA pathway, but also the MEP pathway (Figure 1). Although the genes in the MEP pathway are not affected in hmg1-1 (Figure 4a), the metabolic flow from the MEP pathway may partially compensate sterol and triterpenoid biosynthesis in hmg1-1 because metabolic cross-talk between the MVA and MEP pathways has been reported (Kasahara et al., 2002; Nagata et al., 2002). The chloroplastos alterados 1-1 (cla1-1) mutant, identified as an albino mutant (Mandel et al., 1996), is a loss-of-function allele of DXS in the MEP pathway (Estévez et al., 2000). To understand the contribution of HMGR2 and enzymes in the MEP pathway to sterol and triterpenoid biosynthesis in hmg1-1, we are now generating hmg1 hmg2 and hmg1-1 cla1-1 double mutants. Examining these double mutants should clarify the function of HMG2 and the influx from the MEP pathway.

Experimental procedures

Plant growth conditions

Arabidopsis thaliana (L.) Heynh. ecotypes Wassilewskija (WS) and Columbia (Col), hmg1-1 (WS background), hmg1-2 (WS background), and hmg2-1 (Col background) plants were germinated on agar plates of 1× MS medium (Invitrogen, Carlsbad, CA, USA) containing 3% sucrose. Seedlings were transplanted onto rock–wool bricks (Nitto-bo Co., Ltd., Japan) and supplemented with vermiculite after 2 weeks of culture on agar plates. Plants were grown at 23°C in a photoperiodic cycle of 16 h light/8 h dark.

Isolation of hmg1 and hmg2

To isolate hmg1 and hmg2 mutant alleles, large populations of T-DNA-tagged Arabidopsis plants, generated at the University of Wisconsin Knockout Arabidopsis facility and the Kazusa DNA Research Institute, were screened using PCR. The gene-specific primers, 5′-TGGAGAGATTATTCATTCCCTCCAATGGA-3′ and 5′-GGATGATGATTCAGATTCAGATCATGTTG-3′ for HMG1, and 5′-ACAACCTTGACGTTGATTTCGATGTCACA-3′ and 5′-AAGCCATAAAGATGGAAGGACAAAAGACT-3′ for HMG2, were used. T-DNA-specific primers were synthesized using sequences recommended by the University of Wisconsin Knockout Arabidopsis facility and the Kazusa DNA Research Institute.

RNA gel blot analysis

For RNA gel blot analysis, 15 µg of total RNA was separated on agarose–formaldehyde gels and blotted to nylon membrane as described by Chomczynski (1992). Blots were hybridized with digoxigenin (DIG)-labeled antisense RNA probes specific for each HMGR gene. Hybridization was performed in DIG Easy Hyb solution (Roche, Indianapolis, IN, USA) for 16 h at 68°C. The membrane was then washed two times for 10 min at 68°C with 2× SSC (1× SSC = 0.15 m NaCl + 0.015 m sodium citrate) containing 0.1% SDS, and for 20 min at 68°C with 0.1× SSC containing 0.1% SDS. Hybridized mRNAs were detected using anti-DIG antibody conjugated to alkaline phosphatase and the CDP-Star™ chemiluminescent substrate (Roche). To obtain gene-specific probes, cDNA fragments containing the almost complete first exons of HMG1 or HMG2 were amplified using RT-PCR and cloned into the pCR II-TOPO cloning vector (Invitrogen, Carlsbad, CA, USA). The resulting plasmids were digested with the appropriate enzyme to yield template for antisense in vitro transcription. Antisense RNA probes were synthesized using an RNA transcription kit (Stratagene, La Jolla, CA, USA) and labeled with DIG-UTP using DIG RNA Labeling Mix (Roche). The gene-specific primers for HMG1 were 5′-GGCCTCCTAAACCACCGGTTACC-3′ and 5′-AGACTACTGCCAAAGTATCAAAG-3′ and those for HMG2 were 5′-AGATTTCCGACTAAAAAGAACGG-3′, and 5′-CTACACGCAAACCTCGGATACAG-3′.

RT-PCR analysis

To compare the expression levels of HMG2, DXS, DXR, CMS, XTR9, extensin-like protein, AtSKP1, and SAG12 in WT and hmg1-1 plants, RT-PCR was performed. Total RNA was extracted from seedlings grown in liquid 1× MS medium, containing 1% sucrose, in darkness for 2 days after 2 weeks of culture on 1× MS agar plates, containing 3% sucrose in a photoperiod of 16 h light/8 h dark. MVA, c-ZA, or t-ZA was added to the liquid cultures when the seedlings were transferred to the liquid medium. To investigate the cause of the altered expression of SAG12, XTR9, and extensin-like protein in hmg1-1, we used RT-PCR to examine their expression in WT plants treated with inhibitors of enzymes in the MVA pathway. WT plants were germinated and grown for 8 days on agar plates containing 1× MS medium and 3% sucrose transferred into liquid 1× MS medium containing 1% sucrose, and grown for 3 days under the photoperiod described above; 1 µm lovastatin, 10 µm squalestatin, or 3 µm brz2001 was added to the medium and the seedlings were cultured for three more days. The seedlings were then grown in darkness for 2 days, after which total RNA was extracted from them. The primers specific for HMG2, DXS, DXP, CMS, XTR9, extensin-like protein, AtSKP1, SAG12, and EF-1α were: HMG2-fwd 5′-AATGCCTTGCATTGAGGTTGGTA-3′, HMG2-rev 5′-CATCACCTGTTGACTTGAGACGAAG-3′; DXS-fwd 5′-GAAGTCGCAAAGGGTATGACAAAGCA-3′, DXS-rev 5′-CTGGATCAAATTTCACAACACCATGGTAT-3′; DXR-fwd 5′-TTCTGCCATATTTCAGTGTATTCAAGGTT-3′, DXR-rev 5′-GCACAGATGAATCCTGTGTTTCAATC-3′; CMS-fwd 5′-CAGTGGACTTCAGGAAATCGATGTG-3′, CMS-rev 5′-TGGTTTGATCACCTGTGGTGTCTG-3′; XTR9-fwd 5′-CTCCTTGCCATTGGCTTCTTTGTGGT-3′, XTR9-rev 5′-GCTTGATGTCCTGCTGCATGAGCTTTG-3′; extensin-like protein-fwd 5′-ATGGCTTTCTCACGCCTCTCATTTGC-3′, extensin-like protein-rev 5′-GTAGTAAGTAGGTTCAGACGTGTAGAAG-3′; AtSKP1-fwd 5′-GGTGGCTCTCGAGTCACAAACCATAGCGCA-3′, AtSKP1-rev 5′-GCAAGTCATCAACAACACTTGAAAACGGCA-3′; SAG12-fwd 5′-AGAATCAAGGCAGCTGCGGAT-3′, SAG12-rev 5′-ACACGGTTTTGAATTCATATAGTTGGGTA-3′; and EF-1α-fwd 5′-ACACATTCTTCCTCCGCATCATCCT-3′, EF-1α-rev 5′-TGGCATCCATCTTGTTACAACAGCAG-3′.

Sterols extraction and quantification

Freeze-dried plant tissues (seedlings and inflorescences, 10 mg) were extracted three times with 1 ml of CHCl3–MeOH (1 : 1), and [25,26,26,26,27,27,27-2H7]cholesterol (98% D; Cambridge Isotope Laboratories, Inc., Andover, MA, USA) was added to the extract as an internal standard. The extract was dried in a rotary evaporator and chromatographed on a silica gel column with hexane–EtOAc (2 : 1) and CHCl3–MeOH (1 : 1). The hexane–EtOAc eluent was dried in a rotary evaporator and saponified with 1 ml each of MeOH and 20% KOH aq. for 1 h at 80°C. The CHCl3–MeOH eluent was dried in a rotary evaporator. The residue and the debris from extraction were combined and hydrolyzed with 1 ml each of MeOH and 4N HCl for 1 h at 90°C. These reaction mixtures were extracted three times with 2 ml of hexane, and the combined hexane layer was evaporated to dryness. The residue was trimethylsilylated and analyzed by GC–MS. The endogenous levels of sterols were determined as the peak area ratios of molecular ions for the endogenous one and for the internal standard.

GC–MS analyses were carried out under the following conditions. A mass spectrometer (JMS-AM SUN200, JEOL, Tokyo) was connected to a gas chromatograph (6890 A, Agilent Technologies, Wilmington, DE, USA), EI (70 eV), source temperature 250°C, DB-1 column (15 m × 0.25 mm, 0.25-µm film thickness; J&W Scientific, Folsom, CA, USA), injection temperature 250°C, column temperature program: 80°C for 1 min, then raised to 280°C at a rate of 20°C min−1, and held at this temperature for 8 min; interface temperature 300°C, carrier gas He, flow rate 1 ml min−1, splitless injection. Samples were trimethylsilylated with N-methyl-N-trimethylsilyltrifluoroacetamide at 80°C for 30 min.

Cytokinin analysis

Extraction, purification, and quantification of cytokinins were performed using LC–MS/MS as previously described by Miyazawa et al. (2002).


To compare the cell sizes of WT and hmg1-1 plants, the first internode of the inflorescence stem was fixed with 4% glutaraldehyde, dehydrated in a graded ethanol series, and embedded in Technovit 7100 resin (Heraeus Kulzer GmbH, Wehrheim, Germany). Sections of 5 µm were mounted on slides, stained in a toluidine blue solution, and examined under a light microscope.

To examine whether hmg1-1 pollen was able to germinate and form pollen tubes, we stained the callose of the pollen tubes on pistils. Eighteen hours after pollination, the pistils were placed for 10 min in a drop of 0.1% aniline blue in 0.1% K3PO4 buffer (pH approximately 12). The samples were then pressed briefly with a cover slip and observed under UV illumination using an Olympus IX70 fluorescence microscope (Olympus, Tokyo, Japan).

Construction of the chimeric HMG1 promoter::β-glucuronidase fusion gene and transformation of Arabidopsis

A genomic DNA fragment that included the 2459-bp HMG1 promoter sequence and part of the coding region was amplified by PCR from DNA of Arabidopsis ecotype Col using the primers 5′-TCATCCTCGAGATCGACATAGCCCGGTTTAGCC-3′ and 5′-GTTTGGATCCCCTCCGACGGAGATCCATTGGAG-3′, which contain XhoI and BamHI restriction sites. The PCR-amplified fragment was digested with these enzymes and inserted between the SalI and BamHI sites of the binary vector pBI101.1 (Clontech, Palo Alto, CA, USA) to obtain a translational fusion with the GUS gene. The resulting construct was transferred into Agrobacterium tumefaciens strain GV3101 followed by introduction into Arabidopsis ecotype Col using the floral dip method (Clough and Bent, 1998). T1 seeds were collected from dipped plants and selected by plating on medium containing kanamycin (50 µg ml−1). Five-week-old transgenic plants were subjected to histochemical staining for GUS activity as described by Jefferson (1987).

Feeding experiments

To examine the effect of the inhibitor of squalene synthase on WT plants, plant inflorescences were sprayed with 10 µm of squalestatin. The chemical complementation of hmg1-1 was performed basically as reported by Choe et al. (1998). Briefly, 2 µl each of squalane, squalene, or 10−5, 10−6, or 10−7 m brassinolide was applied directly to the tip of each apical shoot of hmg1-1. hmg1-1 inflorescences were sprayed with 100 mm MVA (3–5 ml per plant).


We thank Yasuyuki Hayashi (Plantech Research Institute) for his contribution to the sterol analysis, Dr Shozo Fujioka (Plant Functions Laboratory, RIKEN) for his advice on GC–MS analysis, Reiko Kiuchi (Plant Science Center, RIKEN) for technical assistance with the microscopic analysis, and Dr Joanne Chory (SALK Institute) for providing the det2 mutant. We also thank Dr Yutaka Miyazawa (Plant Functions Laboratory, RIKEN) for helpful discussions. Squalestatin was kindly provided from Glaxo Research and Development Limited. Brz2001 was kindly provided by Drs Tadao Asami (Plant Functions Laboratory, RIKEN) and Katsuhiko Sekimata (Plant Functions Laboratory, RIKEN). This work was supported in part by a Grant-in-Aid (No. 15510189) for Scientific Research (C) to T.M. and a Grant-in-Aid (No. 15770045) for Young Scientists (B) to N.N. from the Japan Society for the Promotion of Science.