Synthesis of Hydroxylated Sterols in Transgenic Arabidopsis Plants Alters Growth and Steroid Metabolism 1[C][W][OA]

To explore mechanisms in plant sterol homeostasis, we have here increased the turnover of sterols in Arabidopsis ( Arabidopsis thaliana ) and potato ( Solanum tuberosum ) plants by overexpressing four mouse cDNA encoding cholesterol hydroxylases (CHs), hydroxylating cholesterol at the C-7, C-24, C-25, or C-27 positions. Compared to the wild type, the four types of Arabidop- sis transformant showed varying degrees of phenotypic alteration, the strongest one being in CH25 lines, which were dark-green dwarfs resembling brassinosteroid-related mutants. Gas chromatography-mass spectrometry analysis of extracts from wild-type Arabidopsis plants revealed trace levels of a and b forms of 7-hydroxycholesterol, 7-hydroxycampesterol, and 7-hydroxysitosterol. The expected hydroxycholesterol metabolites in CH7-, CH24-, and CH25 transformants were identiﬁed and quantiﬁed using gas chromatography-mass spectrometry. Additional hydroxysterol forms were also observed, particularly in CH25 plants. In CH24 and CH25 lines, but not in CH7 ones, the presence of hydroxysterols was correlated with a con- siderable alteration of the sterol proﬁle and an increased sterol methyltransferase activity in microsomes. Moreover, CH25 lines contained clearly reduced levels of brassinosteroids, and displayed an enhanced drought tolerance. Equivalent transforma- tions of potato plants with the CH25 construct increased hydroxysterol levels, but without the concomitant alteration of growth and sterol proﬁles observed in Arabidopsis. The results suggest that an increased hydroxylation of cholesterol and/or other sterols in Arabidopsis triggers compensatory processes, acting to maintain sterols at adequate levels.

Sterols are important components of the membrane system in eukaryotes. In the membrane, sterols contribute to its physical properties and the activity of membrane-bound proteins. Sterols are also metabolic precursors to steroid hormones, and can function as signaling molecules by binding to regulatory proteins. While these functional aspects of sterols to a large extent are conserved within eukaryotes, the chemical structure of sterols varies considerably among different organisms (Hartmann, 1998). The main sterol in vertebrates is cholesterol, a sterol with eight carbon atoms in its side chain (a C8 sterol), while fungi contain ergosterol, a C9 sterol. Plants are characterized by a mixture of C8, C9, and C10 sterols, such as cholesterol, campesterol, and sitosterol, respectively (Fig. 1A). Insects are unable to synthesize sterols de novo, and depend on an intake of sterols or metabolic precursors from their diet.
Due to the complex cellular functions of sterols, a deficient sterol metabolism often leads to adverse developmental effects. To ensure an adequate sterol level, eukaryotes have evolved different regulatory mechanisms (Espenshade and Hughes, 2007). Cholesterol homeostasis in humans involves membranebound transcription factors, denoted sterol regulatory element-binding proteins (SREBPs), which directly activate expression of genes in the synthesis and uptake of cholesterol, fatty acids, and other lipids. Regulation of SREBP activity is mediated by its interaction with the SREBP cleavage-activating protein, a sterol sensor and escort protein, and the proteases S1P and S2P. The SREBP pathway is present in fruitfly (Drosophila melanogaster), nematode (Caenorhabditis elegans), and in fission yeast (Schizosaccharomyces pombe), but is absent in baker's yeast (Saccharomyces cerevisiae). This fungus depends on other transcription factors, Upc2p and Ecm22p, which bind to sterol response promoter elements and regulate expression of the main sterol synthesis genes (Vik and Rine, 2001).
Also part of the sterol regulatory network in humans are oxidized forms of cholesterol, oxysterols. Compared with cholesterol, oxysterols have a very short half-life, are present at 1,000-to 100,000-fold lower levels, and have a higher capacity to pass lipid membranes. Oxysterols are formed from cholesterol by action of separate cholesterol hydroxylases (CHs), but can also be formed nonenzymatically by reactive oxygen (Björkhem, 2002). Oxysterols function as signaling compounds in a variety of processes, including sterol homeostasis. Specific oxysterols bind to liver X receptors (LXRs), members of the nuclear hormone receptor family. LXR dimerizes with the retinoic X receptor, and binds to specific response elements of target genes in cholesterol homeostasis (Cummins and Mangelsdorf, 2006). An additional component of oxysterol signaling are the oxysterol-binding proteins (OSBPs), which bind oxysterols with high specificity and influence sterol homeostasis and lipid transport, but do not function as transcription factors (Olkkonen and Levine, 2004).
Compared to what is known about sterol homeostasis in mammals and yeast, knowledge about this process in plants is limited (Hartmann, 1998). Our analysis of the entire Arabidopsis (Arabidopsis thaliana) and rice (Oryza sativa) genomes has not revealed any obvious orthologs to the known key components in human sterol homeostasis such as SREBP, SREBP cleavage-activating protein, or LXR. However, one component in sterol homeostasis that seemingly is present in plants is OSBP-related proteins (ORPs). ORPs are conserved in most eukaryotes, and between four and 12 genes encoding this type of protein are present in yeast, nematode, fruitfly, mouse, and Arabidopsis (Ngo et al., 2010). An initial study of a plant ORP came from the search for interaction partners to a receptor-like kinase from petunia (Petunia hybrida; Skirpan et al., 2006). This identified PiORP1, an ORP localized to the plasma membrane of pollen tubes through a pleckstrin domain. PiORP1 was used to search the Arabidopsis genome, which revealed 12 transcribed genes falling into three distinct clusters. The precise function of ORPs in plants or outside humans is not clear, but disruption of all seven ORPs in yeast was lethal (Beh et al., 2001), suggesting an essential function. Binding of hydroxysterols such as 25-hydroxycholesterol has been shown to some of the mammalian ORPs, e.g. human ORP1 and ORP2 (Suchanek et al., 2007), although it is presently unknown to what extent plant ORPs also bind oxysterols, or the effects such a binding would evoke. In a recent study, Arabidopsis ORP3a was shown to possess binding affinity to sitosterol and to be localized to the endoplasmatic reticulum, where it might influence sterol transport processes (Saravanan et al., 2009). However, the ORP3a affinity for hydroxysterols was not investigated in that study.
While plant ORPs have been identified, the occurrence of oxysterols, their presumptive ligands, is less well documented. Oxysterols have tentatively been identified in potato (Solanum tuberosum) plants (Heftmann and Weaver, 1974), and were by gas chromatography-mass spectrometry (GC-MS) more convincingly identified in extracts from barley (Hordeum vulgare) roots (König and Seifert, 1998). The latter study identified nine oxysterol species, including the 7a-hydroxy and 7b-hydroxy forms of 24-methyl cholesterol, sitosterol, and stigmasterol, as well as the corresponding 7-oxo forms.
The homeostatic regulation of cholesterol in plants is of interest not only from the perspective of membrane function, but also from the specific role of cholesterol in certain plant species as a metabolic precursor of defense metabolites, such as saponins, phytoecdysones, and glycoalkaloids (Kreis and Mü ller-Uri, 2010). In most plant species, e.g. Arabidopsis and rice, cholesterol constitutes only a small fraction of total desmethylsterols. However, cholesterol can in some species be a major sterol, and the ratio of cholesterol in membrane sterols can be as high as 20% in potato plants (Arnqvist et al., 2003), possibly related to a use of cholesterol for synthesis defense metabolites. The regulatory mechanisms that underlie these metabolic differences between species are poorly understood. Mutation of the Arabidopsis STEROL METHYLTRANSFERASE1 (SMT1) gene has been cor- Figure 1. Chemical structure of end product desmethylsterols and of different forms of hydroxycholesterol. A, Chemical structure of the plant sterol precursor cycloartenol, and the sterol end products cholesterol, campesterol, and sitosterol, representing the C8, C9, and C10 side chain sterols, respectively. B, Cholesterol hydroxylated at the C-7, C-24, C-25, and C-27 positions. related with a lowered alkylation of sterol side chains and a drastically increased cholesterol level, indicating that cholesterol synthesis is influenced by this enzymatic step (Diener et al., 2000).
To gain more knowledge about the homeostasis of cholesterol and other sterols in plants, we have here increased the turnover of cholesterol in transgenic plants by overexpression of four mouse CHs hydroxylating cholesterol at different positions. In mouse, these enzymes hydroxylate cholesterol mainly at the C-7, C-24, C-25, and C-27 positions (Russell, 2000;Pikuleva, 2006;Fig. 1B); thus either in the sterol B-ring (CH7) or side chain (CH24, CH25, and CH27). Arabidopsis and potato were chosen as two plant species having a low, or high, relative level of cholesterol, respectively. The results demonstrate that increased hydroxysterol biosynthesis in Arabidopsis and potato plants causes different effects on sterol metabolism, growth, and development.

RESULTS
Generation and General Phenotype of Transgenic Arabidopsis Plants Expressing the Mouse CHs CH7, CH24, CH25, and CH27 To explore regulatory processes in plant sterol homeostasis, coding sequences from the mouse CH7, CH24, CH25, and CH27 cDNA were expressed from the cauliflower mosaic virus 35S promoter in transgenic Arabidopsis and potato plants. The expression of transgenes was verified in Arabidopsis T3 plants by reverse transcription (RT)-PCR (Supplemental Fig. S1), and for each type of transformant, two lines with a strong transgene expression were chosen for further characterizations.
The general effects on growth and development of Arabidopsis CH transformants were distinct for each type of transformation (Table I), and representative plants are depicted in Figure 2A. With regard to growth and development, the CH27 plants displayed a phenotype similar to control transformants and the wild type. The general phenotype of CH7 and CH24 plants was also rather similar to that of control lines, although several CH24 lines were somewhat smaller in size. The strongest effects of the transformation were observed in CH25 lines. Strong CH25 lines had dark-green leaves and were severe dwarfs. The dwarf phenotype was clearly stronger for homozygous plants than for heterozygous ones (Fig. 2, B-D). Analysis of the T3 generations revealed that the occurrence of fertile CH25 homozygous T3 lines was significantly underrepresented (Table I), and that the flower development was disturbed (Fig. 2D). The flower opened to a lesser degree than wild-type flowers, and styles were clearly shorter (Supplemental Fig. S2), likely leading to the reduced self fertility of CH25 plants in homozygous form. Particularly when grown under short-day conditions, several CH25 lines could be preserved only as heterozygotes, unless the homozygous transformants were pollinated by hand.

Hydroxysterol and Sterol Levels in Wild-Type Arabidopsis and CH Transformants
To monitor functional expression of the introduced CH genes, hydroxysterols were enriched using solidphase extraction (SPE) and analyzed by GC-MS in wild-type Arabidopsis plants and in strong CHexpressing lines. Analysis of wild-type plants revealed very low levels of hydroxysterols on a fresh weight (f. w.) basis, and extraction of as much as 20 g leaf tissue was necessary to obtain measurable amounts. GC-MS analysis revealed two minor peaks having the same retention time and MS fragmentation pattern as authentic 7a-hydroxycholesterol and 7b-hydroxycholesterol standards, respectively (Fig. 3, A-C). Levels were estimated to 0.008 mg kg 21 f.w. 7a-hydroxycholesterol and 0.005 mg kg 21 f.w. 7b-hydroxycholesterol (Supplemental Table S1). In addition, other metabolites with a molecular ion at mass-to-charge ratio (m/z) 560 and 574 were based on their mass spectra tentatively identified as the a and b forms of 7-hydroxycampesterol (not shown) and 7-hydroxysitosterol ( Fig. 3D), respectively, although the position of the hydroxyl group needs to be confirmed. None of the three other hydroxylated cholesterol species that were expected to occur in the CH transformants (i.e. 24-hydroxycholesterol, 25hydroxycholesterol, and 27-hydroxycholesterol) could be detected in the wild type. Table I. Growth characteristics of CH transformants CH7, CH24, CH25, and CH27 CH7, CH24, CH25, and CH27 transformants, and an empty-vector control line, were selected for kanamycin resistance before being transferred to soil (day 0), and further growth in a climate room. The frequency of T3 homozygotes was analyzed in the offspring from 10 to 12 self-fertilized T2 plants with 3:1 segregation. Rosette diameter and flowering frequency refer to the lines depicted in Figure 2, but are representative for other lines from the same transformation. In contrast, 7-, 24-, and 25-hydroxycholesterol species were readily resolved and identified in the CH7, CH24, and CH25 lines, respectively (Table II; Fig. 4). The hydroxysterol fraction in leaves from the transgenic CH7 lines number 4 and number 8 contained 0.09 mg kg 21 f.w. and 0.12 mg kg 21 f.w. 7-hydroxycholesterol, respectively, thus more than 10-fold higher than the trace levels in wild-type plants. Based on the GC retention time and MS fragmentation pattern, the hydroxyl group was determined as being in the a orientation, in agreement with the known catalytic function of the CH7 enzyme (CYP7A1) in rat (Ogishima et al., 1987). The hydroxysterol fraction of CH24 line number 3 contained 0.22 6 0.02 mg kg 21 f.w. 24-hydroxycholesterol, whereas CH25 line number 5 contained as much as 11.1 6 3.7 mg kg 21 f.w. 25-hydroxycholesterol (Table II). However, despite a strong expression of the CH27 transcript, no 27-hydroxycholesterol could be detected in CH27 lines, indicating that the CH27 construct was not functional.
In addition to the expected presence of 24-hydroxycholesterol and 25-hydroxycholesterol in CH24 and CH25 lines, respectively, these lines also contained other hydroxysterols (Table II). In CH24 lines, this included 24-hydroxy-24-methyl cholesterol, but the C-24 R/S orientation could not be deduced from the GC-MS analysis. In CH25 lines, 25-hydroxylated forms of campesterol, 24-methylene cholesterol, and stigmas-terol were identified, together with eight other hydroxysterols that were either tentatively identified, or unidentified. Presence of an abundant fragment with m/z 131, characteristic of 25-hydroxylated desmethylsterols, in 10 out of the 11 hydroxysterols suggests that the different hydroxylations in CH lines specifically occurred at C-25. The amounts of the different hydroxysterol forms in CH25 lines was about 3-to 5-times higher in homozygotes than in heterozygotes (Table  II), thus suggesting a causal relationship between the level of 25-hydroxylated sterols and the stronger effects on growth and development in homozygotes. The invariant position of the hydroxyl group at C-24 or C-25, respectively, for these additional hydroxysterol species is consistent with the known catalytic function of the transgene-encoded protein, and suggests that the mouse enzymes are able to use other sterols than cholesterol as substrates in a plant sterol background. In contrast, there were no increases of hydroxysterol species other than 7a-hydroxycholesterol in CH7 lines, indicating a higher substrate specificity of the CH7 enzyme compared with CH24 and CH25, or that any additional product was below detection.
The demonstration of new hydroxysterols in CH transformants prompted an analysis of the level of cholesterol and other sterols. GC and GC-MS sterol analyses did not reveal any major alterations in the composition of desmethylsterols in CH7 and CH27 lines as compared to the wild type (Supplemental Fig. S3,A and B). In contrast, both CH24 and CH25 lines displayed a clearly altered sterol composition. Compared to the wild type, the ratio of C9 to C10 sterols in CH24 lines increased from 9% to 18% of total desmethylsterols, while the total sum of desmethylsterols was not altered (Fig. 5A). Moreover, for both the C9 and C10 branch of the sterol biosynthesis pathway, the proportion of sterols that are substrates for the D24(28) reduction step increased, and the corresponding products decreased, leading to a higher proportion of 24methylene cholesterol to campesterol (C9 branch), and of isofucosterol to sitosterol (C10 branch). Thus, whereas 24-methylene cholesterol was not detectable in the wild type, this sterol was in CH24 lines present at levels even higher than those of campesterol. In the CH25 lines, the desmethylsterol profile showed a pattern similar to that in CH24 lines, but was even more pronounced (Fig. 5A), and was stronger in homozygotes than in heterozygotes (Supplemental Fig. S3C). The C9:C10 sterol ratio increased strongly, and C9 sterols contributed to 77% of total desmethylsterols. The ratios of 24-methylene cholesterol:campesterol, and of isofucosterol:sitosterol, was even higher than those in CH24 lines. Furthermore, the level of cholesterol was raised to 11% of total desmethylsterols. The increased proportion of C9 sterols was mainly at the expense of C10 ones, which accounted for 89% in the wild type, but only 12% in CH25 (Fig. 5A). The ratio of hydroxysterols to desmethylsterols varied widely between the transformants; in CH7 and CH24 the hydroxysterols were less than 1%, but in the CH25 homozygotic lines they amounted to about 20% of the total of desmethylsterols and hydroxylated desmethylsterols (Table II; Fig. 5).
In the CH25 lines, there was an increase not only in the total amounts of 4-desmethylsterols but also in their precursors; 4-monomethylsterols and 4,4#dimethylsterols (Fig. 5). The increase in precursors appeared from visual examination of the berberinestained thin-layer chromatography (TLC) plates used to separate the sterols (Supplemental Fig. S3D) and was verified by GC-MS analyses. In the 4-monomethylsterol fraction there were increased levels of cycloeucalenol, obtusifoliol, and 24-methylene lophenol, whereas the levels of the product of the SMT2 reaction, 24-ethylidene lophenol, was low ( Fig. 5B), in accordance with the altered profile of desmethylsterols. It should be noted that the low amounts of 4-monomethylsterols make comparisons within this sterol class less precise. In the 4,4#-dimethylsterol fraction from CH25 lines, levels of both cycloartenol and 24-methylene cycloartanol were clearly increased as compared to the wild type ( Fig. 5C), and there was a tendency for an increased ratio of cycloartenol to 24-methylene cycloartanol. Also CH24 lines displayed somewhat elevated 4,4#-dimethyl and 4-monomethyl sterol levels, as compared to wildtype plants (Fig. 5, B and C).

Generation and Sterol Analysis of Potato CH Transformants
The changes in sterol composition in the CH25 Arabidopsis transformants suggested that the channeling of sterols along the C9 and C10 branch was severely inhibited, possibly already at the level of Figure 3. GC-MS analysis of the hydroxysterol profile in wild-type Arabidopsis (Col.) plants. Hydroxysterols were extracted from rosette leaves, enriched using the SPE technique, and TMS-ether derivatized prior to GC-MS analysis using 19-hydroxycholesterol as an added internal standard. A, GC-MS chromatogram of the hydroxysterol fraction. Peaks were identified as 7a-hydroxycholesterol (I), 7a-hydroxycampesterol (II), 7b-hydroxycholesterol (III), 7a-hydroxysitosterol (IV), 7b-hydroxycampesterol (V), and 7b-hydroxysitosterol (VI). Additional peaks could not be identified with certainty as hydroxysterols from the GC-MS analysis. B and C, Mass spectrum of the peaks identified as 7a-hydroxycholesterol (B), and 7b-hydroxycholesterol (C) from comparisons of the retention time and fragmentation pattern of authentic standards, or identified as 7a-hydroxysitosterol (D), based on published spectra. cycloartenol methylation to 24-methylene cycloartanol. To further investigate these effects on sterol profiles, we performed equivalent CH transformations of potato, a plant species that has a higher relative level of cholesterol than Arabidopsis. We hypothesized that the higher amounts of cholesterol available as substrate for CH25, coupled to a higher flux of cycloartenol toward cholesterol in this species, might lead to less severe effects of CH25 hydroxylation.
The transformations yielded plants that in the heterozygotic T1 generation were quite similar to the wild type ('Desirée'). For instance, none of the 21 independent CH25 transformants that were regenerated from two separate transformations displayed a dwarf phenotype. Two CH25 transformants with a strong transgene expression were chosen for further analyses of hydroxysterol and sterol levels (Supplemental Fig. S4). The level of 25-hydroxycholesterol was for the amount of tissue analyzed below detection in wild-type 'Desirée' plants, but was in two strong CH25 overexpressors clearly detectable at 6.6 6 1.3 mg kg 21 f.w. and 8.9 6 2.0 mg kg 21 f.w. (n = 3, mean 6 SD), demonstrating that the CH25 enzyme was functional also in potato.
The levels of 25-hydroxycholesterol were about 5-fold higher than in equivalent Arabidopsis heterozygotes, which is in accordance with the endogenous higher cholesterol levels in potato. Additional hydroxysterol species were present also in the CH25 potato lines (not shown), although their identity was not further investigated. However, in contrast to the CH25 Arabidopsis lines, the potato CH25 lines exhibited normal growth and no major changes in sterol profiles, although cholesterol levels were reduced somewhat (Supplemental Fig. S4). Hence, an alteration of the proportion of C8, C9, and C10 sterols was not significant in potato.

Altered Activity of Sterol-Biosynthetic Enzymes in CH24 and CH25 Transformants
The SMT1 and SMT2 enzymes act at the branch points of the C8, C9, and C10 sterol pathways. Cycloartenol is the preferred substrate for SMT1, and 24methylene lophenol for SMT2. To investigate if the altered sterol composition in Arabidopsis CH24 and CH25 lines could be attributed to an altered activity of SMT1 or SMT2, these enzyme activities were analyzed in microsome preparations from leaf materials. Table II. Hydroxylated sterols in transgenic CH7, CH24, and CH25 Arabidopsis lines Hydroxysterols were extracted from rosette leaves and the TMS ether derivatives were quantified by GC-MS using 19-hydroxycholesterol as an added internal standard. Samples were from homozygous CH7 and CH24 plants, and heterozygous ( a ) or homozygous ( b ) CH25 plants. Major fragmentation ions are in bold and the predominant peaks are underlined. Identification of hydroxysterols is based on relative retention times, mass spectrum, and reported occurrence of the corresponding nonhydroxylated sterol in the plant. Asterisks indicate an additional identification by comparison to authentic standards. Mean value 6 range for two independent leaf samples. n.d., Not deduced. Compared to the wild type, the enzymatic assays revealed in CH24 plants a 1.7-fold and 3.2-fold higher activity of SMT1 and SMT2, respectively, and in CH25 plants a 3.4-fold and 6.3-fold higher activity (Table III). Parallel analyses of SMT1 and SMT2 gene expression using quantitative RT-PCR did not reveal any major increases in SMT1 and SMT2 gene expression in CH24 and CH25 lines, as compared to the wild type (Supplemental Fig. S5). Rather, the expression of both SMT1 and SMT2 was in CH25 lines somewhat reduced.

Brassinosteroid Levels and Dose Responses in CH25 Transformants
The dark-green color and dwarf phenotype of CH25 plants resembled Arabidopsis mutants having a defect metabolism of, or sensitivity to, the plant growth hormone brassinosteroid (BR). To investigate the BR metabolism in the transformants, the BR level was determined using GC-MS in two separate CH25 lines (lines number 3 and number 5), and compared to wild-type plants. This revealed a general reduction of BRs in the CH25 lines, being more pronounced in homozygous CH25 plants than in heterozygous ones (Table  IV). For instance, the level of 6-deoxocastasterone was reduced about 15% in heterozygotes of both lines compared to the wild type, whereas it in homozygous lines was reduced 60% in line number 5, and 80% in line number 3. Also the level of castasterone (CS), the immediate precursor of brassinolide (BL), was reduced by approximately 70%. However, the level of BL was below detection for all genotypes when grown and analyzed under our conditions. The degree of BR reduction correlated with the shorter stature of CH25 homozygotes compared with heterozygotes, and a somewhat stronger phenotype of the line number 3 compared with line number 5 (not shown).
The dwarf phenotype of CH25 lines could be caused by an inhibition of the perception/transduction of the BR signal, or alternatively, by lower levels of active BRs. As a measure of the BR responsiveness, the dosedependent stimulation of hypocotyl elongation and inhibition of root growth by CS and BL was followed in vitro. For comparison, the Arabidopsis dwarf mutant det2-1 was used as an example of a plant with reduced BR synthesis and a dwarf phenotype. This mutant has reduced activity of the enzyme steroid 5areductase, resulting in lower synthesis of the early BL precursor campestanol (Fujioka et al., 1997). The results showed that heterozygous CH25 lines displayed a dose-dependent stimulation of hypocotyl elongation (Fig. 6), similar to that shown for det2-1 (Fujioka et al., 1997). Heterozygotes were fully restored to the growth pattern of empty-vector control transformants, by treatments with 100 to 1,000 nM BL or 1,000 nM CS, whereas the stimulatory effect in homozygotes did not reach the same extent as in controls. The capacity to react to BRs with an inhibition of seedling root growth was also clearly contained in the CH25 transformants, which showed progressively reduced root lengths at treatments with increasing concentration of BL up to 1,000 nM and CS 100 nM (Supplemental Fig. S6). In addition, when grown for a longer period of time on CS-and BL-supplemented media, an increased petiole length, a larger leaf blade, and an epinastic leaf growth were evident in both CH25 and det2-1 mutants (not shown). Together these results demonstrate that CH25 lines are responsive to BRs such as CS and BL in the root, hypocotyl, and shoot. Hence, a reduced endogenous level of BRs, rather than a reduced BR responsiveness, may be considered as the prime cause of the CH25 dwarf phenotype.

CH25 Expression Increases Drought Tolerance
To investigate whether the altered sterol and steroid metabolism in CH24 and CH25 transformants influenced the physiology of the plants, their responses to drought, cold, and heat stresses were monitored and compared to empty-vector transformants and the BR det2-1 mutant. Due to the severe dwarf phenotype of CH25 lines in homozygote form, CH25 heterozygotes were selected to obtain a test material that was more comparable in size to the other genotypes. After 14 d without watering, severe drought stress symptoms such as growth reduction, leaf wilting, and drying of the leaf blade, were observed in almost all control transformants and in CH24 plants, whereas heterozygote CH25 plants and det2-1 mutants showed clearly milder symptoms (Fig. 7), and in a significantly lesser proportion of the plants analyzed (Table V). In a separate test, plants were rewatered after a continuous drought stress for 21 d, and the proportion of surviving plants was scored. Only 5% of the control transformants and the CH24 plants recovered after the treatment, whereas 52% of CH25 heterozygote plants and 40% of det2-1 survived. However, no significant differences compared to the wild type were observed among the transformant genotypes with regard to their capacity to withstand heat or cold stresses (not shown).

Endogenous Hydroxysterol Production in Wild-Type and Transgenic Plants
We have here attempted to modify the sterol composition and turnover in Arabidopsis and potato plants by channeling cholesterol to hydroxylated forms via expression of mouse CH enzymes in planta. Little is presently known about the occurrence, metabolism, or biological activity of hydroxysterols in plants. Technical difficulties in analyzing the minute levels of hydroxysterols that are present in plants has probably been one factor hampering studies in this area. Among few previous reports is one from barley, where 7ahydroxy, 7b-hydroxy, and 7-keto forms of campesterol, sitosterol, and stigmasterol were identified in roots, but not in leaves (König and Seifert, 1998). No hydroxylated forms of cholesterol were reported, in line with the low levels of cholesterol (below 0.5% of total) in the investigated cultivar.
The SPE protocol used in our study offers a convenient and reproducible method for hydroxysterol extraction from plant materials (Dutta and Appelqvist, 1997), and should simplify further studies in this field. Using this protocol we were able to demonstrate low amounts of hydroxysterols in wild-type Arabidopsis plants, mainly 7a-hydroxycholesterol, 7b-hydroxycholesterol, and four metabolites tentatively identified as the a and b forms of 7-hydroxycampesterol and 7-hydroxysitosterol (Fig. 3). At present we do not know if these hydroxysterols are formed in planta by enzymatic activity or spontaneous oxidation, or alternatively, formed in vitro during the extraction process. The later explanation is considered as less likely, since the SPE protocol does not involve TLC separation, and thus minimizes sample exposure to atmospheric oxygen during purification. In line with this suggestion is the fact that further oxidized hydroxysterols (e.g. ketosterols) were not detected. Together with the previous study in barley, our results suggest that hydroxysterols are endogenous sterol metabolites in plants, and that their composition varies between species, as does the Table III. Enzymatic activity of SMT in wild-type Arabidopsis plants and transgenic CH24 and CH25 lines Enzymatic activity of SMT1 and SMT2 was measured in microsome preparations using the native substrates cycloartenol and 24-methylene lophenol, respectively. Measurements were corrected for the activity with endogenous substrate present in microsome preparations. Mean value 6 range of two independent microsome preparations. composition of main sterols. Time-course feeding studies using labeled sterol precursors may now be performed to clarify the process of hydroxysterol synthesis. The identification of the main hydroxysterol forms in Arabidopsis presented here, should simplify such attempts. Hydroxysterol levels were increased in transgenic CH7, CH24, and CH25 lines, but not in CH27 ones. The absence of 27-hydroxysterols in CH27 lines could be due to several factors, such as mislocalization of the protein from its normal targeting to mitochondria (Pikuleva, 2006). It may also be noted that the hepatic 27-hydroxylase and 7a-hydroxylase are inhibited by sitosterol (Nguyen et al., 1998), a major sterol species in Arabidopsis. Nevertheless, the normal phenotype and absence of detectable alterations in the sterol/ hydroxysterol profile of CH27 lines strengthens a causal relation between presence of hydroxysterols in the CH24 and CH25 lines, and the observed effects on their growth and sterol metabolism (Figs. 2 and 5).
CH7 transformants contained higher levels of 7ahydroxycholesterol, but not of other 7-hydroxysterols (Table II). This indicates that the CH7 enzyme only accepts cholesterol as a substrate, even in a plant sterol background. The introduced CH24 and CH25 CHs were, in contrast, not fully specific for cholesterol, and hydroxylated forms of other sterols also occurred, most notably in CH25 lines (Table II). This wider substrate acceptance of the CH25 hydroxylase presumably reflects different enzymatic properties of CH25, a nonheme monooxygenase, compared to CH7 and CH24, which both are cytochrome P450 proteins.

Hydroxysterol Synthesis Alters Sterol Metabolism
For both CH24 and CH25 lines, the synthesis of hydroxysterols was correlated with alterations in the composition of common (nonhydroxylated) sterols.
There was a striking decrease in C10 side chain sterols, and a corresponding relative increase in C8 and C9 ones (Fig. 5). By contrast, sterol profiles were apparently normal in CH7 lines. This may be due to the low amounts of 7-hydroxycholesterol formed, but also indicates that hydroxylations in the sterol B-ring (CH7) have a weaker biological effect than hydroxylations in the sterol side chain (CH24, CH25). The altered sterol profiles in the CH24 and CH25 transformants indicated that mainly steps in the sterol side chain metabolism were affected. For instance, in CH25 transformants, the ratio of cycloartenol to 24-methylene cycloartanol increased, as did the ratio of C9 to C10 sterol species, and the ratio of substrates for the D24 (28) reduction step relative to its products. There were no indications of altered ratios of sterol intermediates representing metabolism of other parts of the sterol skeleton. This suggests that the enzymatic steps affecting side chain modifications [SMT1, SMT2, and the DWF1 D24(28) reductase] are inhibited in vivo by the side chain hydroxylated sterols. Reports from various mutants, transformants, and inhibitor-treated plants with modified expression or activity of these enzymes support this suggestion. The increased sterol channeling in CH25 lines toward C8 and C9 side chain sterols can thus be thought to resemble that observed for C8 side chain sterols in Arabidopsis smt1 mutants (Diener et al., 2000), and for C9 side chain sterols in transformants with cosuppression of the SMT2;1 gene (Schaeffer et al., 2001) as well as in smt2 mutants (Carland et al., 2002(Carland et al., , 2010. Likewise, the altered proportions in CH24 and CH25 lines between 24methylene cholesterol and campesterol, as well as between isofucosterol and sitosterol, parallels the situation in the dwf1 mutant, where the sterol D24(28) reduction steps are inhibited (Klahre et al., 1998). The changes in CH25 sterol patterns are also rather similar to those reported in suspension cultures of bramble cells treated with 25-azacycloartanol (Schmitt et al., Table IV. BR levels in aboveground tissue from wild-type Arabidopsis plants and transgenic CH25 lines Transgenic CH25 lines were selected for kanamycin resistance and visually screened for the characteristic phenotypes of heterozygotes and homozygotes. BRs were analyzed by GC-MS using deuterium-labeled internal standards. Mean value 6 range for two independent replicates (wild type), or one single analysis (CH25, line numbers 3 and 5). imp., Metabolite not detected due to an interfering substance; n.d., not detectable.

1981)
, where a large decrease in the proportion of C10 side chain sterols was observed in parallel with a decrease of sterols with a saturated side chain. It should be noted that the positive charge at position 25 was an important aspect of the inhibitory effect. Similar effects have been described also for other types of charged nitrogen-containing compounds (Song and Nes, 2007). Of relevance for this study is that compounds such as 24,25-epiminolanosterol and triparanol inhibit both SMT1 and DWF1 activity, suggesting mechanistic similarities between the two types of reaction (Song and Nes, 2007). This raises the possibility that sterol side chain hydroxylations have a similar inhibitory effect as charged nitrogen-containing compounds. In line with this are unpublished data of ours showing that addition of 24-hydroxycholesterol or 25-hydroxycholesterol to microsome preparations can reduce SMT1 activity by more than 50% (T. Shirazi and F. Sitbon, unpublished data). However, more measurements are needed to determine the specificity of this effect, as well as the subcellular distribution of hydroxysterols in wild-type plants and CH transformants. The increases in SMT1 and SMT2 enzyme activity determined in vitro in microsomes from CH24 and CH25 Arabidopsis transformants (Table III) do not support the substrate:product ratios measured in vivo (Fig. 5). At present we have no explanation for this apparent contradiction, but suggest that the enzyme measurements in vitro do not give an adequate picture of the situation in vivo. This is to some extent supported by the fact that there were no indications of a stronger expression of the SMT1 and SMT2 genes (Supplemental Fig. S5), as has been reported for SMT transformants with elevated SMT1 and SMT2 activities in vitro, leading to an increased sitosterol production (Sitbon and Jonsson, 2001;Arnqvist et al., 2003). Another suggestion is that the increased SMT enzyme activity may be a compensatory effect to overcome a metabolic bottleneck caused by substrate accumulation. This would be in some agreement with the increased SMT1 activity found in potato discs, which accumulated cycloartenol after wounding (Bergenstråhle et al., 1993). The fact that the potato CH25 transformants did not change their proportion of sterols (Supplemental Fig. S4), supports to some extent a SMT1-catalyzed conversion of cycloartenol to 24methylene cycloartanol as one of the steps that are inhibited in CH25 Arabidopsis lines. In potato, the relative amount of cholesterol is much higher than in Arabidopsis, and the flux of sterol precursors is to a significant extent already going along the C8 branch to cholesterol. This might minimize the effect of sterol 25hydroxylations in potato due to a lower strain on the SMT1-catalyzed reaction and following steps in the C9 and C10 side chain branches.
Interestingly, transgenic tobacco (Nicotiana tabacum) plants with an increased conversion of sterols to stanols from expression of an Actinomyces cholesterol oxidase gene, CHO, also displayed a dwarf phenotype, but did not contain an alteration of the sterol composition similar to that in our CH Arabidopsis transformants (Corbin et al., 2001;Heyer et al., 2004). This indicates that the hydroxylation process, rather than  the increased turnover of sterols, causes the altered sterol profiles in our Arabidopsis CH lines. However, since also tobacco is a plant species with a relatively high proportion of cholesterol (Sitbon and Jonsson 2001), an alternative interpretation is that the regulation of sterol metabolism in CHO tobacco may be similar to that in our CH25 potato plants, where no significant effects on desmethylsterol profiles were observed. Transformation of Arabidopsis with the CHO construct would be one way of investigating these observations further.

CH25 Expression Alters Metabolism of BL Precursors
The level of several BL precursors, e.g. CS and typhasterol, was lower in CH25 plants than in the wild type (Table IV). The level of BL was below detection in both genotypes, although the complementation of the CH25 dwarf phenotype with exogenous CS and BL (Fig. 6), suggests that also the BL level was reduced in CH25 lines compared to the wild type. One explanation for these results may be that CH25 hydroxylates one or several BL precursors, consistent with the relaxed sterol substrate specificity of CH25 in transgenic plants. Since the identified hydroxylated sterols in CH25 plants were hydroxylated at the C-25 position, we speculate that such a BR inactivation would be due to a steroid C-25 hydroxylation, reminiscent of what has been shown for Arabidopsis transformants overexpressing the BAS1 gene (Neff et al., 1999). BAS1 (CYP734A1, formerly denoted CYP72B1) encodes an enzyme down-regulating BL activity by a C-26 hydroxylation of BL and its precursor CS, and leads to a dwarf phenotype if the gene overexpressed. Interestingly, also the BAS1-related protein CYP72C1 confers a dominant dwarf phenotype when overexpressed (Nakamura et al., 2005;Takahashi et al., 2005;Turk et al., 2005). Analysis of BR levels in CYP72C1 overexpressors revealed lower levels of 6-deoxoCS, CS, and BL, suggesting that also this enzyme is able to catabolize BL precursors. The mechanism of this inactivation is presently unknown, but considering the similarity at the protein level of CYP72C1 to BAS1, a hydroxylation of BL precursors seems possible. Thus, different types of reaction seem to be able to inactivate BL precursors, and expression of CH25 might have a similar effect. The dwarf growth of CH25 lines in our study may thus be interpreted as a CYP72-like effect on BR metabolism, due to the more relaxed substrate specificity of CH25 compared to CH7 and CH24.

Overexpression of CH25 Increases Drought Tolerance in Arabidopsis
In addition to a reduced growth rate, the CH25 plants also displayed an enhanced resistance to a drought stress, comparable to that in the BR-deficient control det2 mutant (Fig. 7). However, an increased 24hydroxysterol synthesis in CH24 lines was without effect in this aspect. Previous observations have shown that the det2 mutant has an enhanced tolerance to oxidative stress (Cao et al., 2005), as well as to drought (Divi and Krishna, 2009). It has been suggested that the enhanced tolerance could be due to a BR deficiency causing a constant cellular stress, leading to activation of stress tolerance genes and enhanced activities of relevant enzymes (Divi and Krishna, 2009). The similar effects on drought tolerance in det2 and CH25 lines, indicate that a decreased level of BRs is an important factor for tolerance in both genotypes, although it is presently unclear how low BR levels is translated into drought tolerance. Recent studies suggest a cross talk between BRs and other plant hormones (Divi et al., 2010;Depuydt and Hardtke, 2011) including abscisic acid (Zhang et al., 2009a(Zhang et al., , 2009b. Reduced BR levels in CH25 plants might increase abscisic acid levels or action, which in turn would contribute to reduced water losses. Other possible explanations are that the enhanced drought tolerance is mediated by the altered membrane sterol composition, or by developmental effects on stomata numbers. Possibly, decreasing endogenous BRs by transgenic modification or breeding could be a future means of creating crops with enhanced water stress tolerance. Taken together, we have in transgenic Arabidopsis plants with an increased hydroxysterol synthesis demonstrated effects on plant growth and development, as well as a perturbed metabolism of sterols and BRs. The results suggest that hydroxysterols constitute endogenous sterol metabolites in plants, and more information is now needed to demonstrate the precise origin and identity of plant hydroxysterols, as well as their role in sterol metabolism, transport, and other lipidrelated cellular process. Coupled to the recent identi- Table V. Drought stress response of transgenic empty-vector control Arabidopsis, CH24 and CH25 lines, and the det2-1 mutant Transgenic lines were selected for kanamycin resistance and transferred to soil for further growth. CH25 lines were also selected for the characteristic appearance of heterozygotes. Two-week-old soil-grown plants were subjected to a water stress and the relative proportion of plants with drought stress symptoms was scored after 2 weeks. Numbers in parentheses denote the number of plants showing symptoms. A statistical difference from empty-vector control transformants was calculated by the x 2 test. NS, Difference not significant.  (10) 13.3% (4) P , 0.001 fication of plant proteins with strong similarity to OSBPs, our results suggest that the metabolism and action of oxysterols may have been an overlooked aspect in plant sterol research. Further analyses in this area might be rewarding.

Plant Materials
Wild-type Arabidopsis (Arabidopsis thaliana) ecotype Columbia (Col.) and derived transformants were germinated on sterile Murashige and Skoog medium (Duchefa Biochemie Bv.), solidified with 0.25% Gelrite, and complemented with kanamycin as a selection marker for transformants. After 12 to 14 d, plants were further grown in sterilized soil in climate rooms equipped with metal halogen lamps giving a photon flux density of 100 mmol m 22 s 21 , and under an 8 or 16 h photoperiod for sterol or plant growth analyses, respectively. In addition, the BR-related dwarf mutant det2-1 (Col. background, Nottingham Arabidopsis Stock Centre, cat. no. N6159) was used for certain comparisons. Leaf tissues were harvested, frozen in liquid nitrogen, and stored at 270°C until analysis.

Verification of Transgene Expression
Total RNA was extracted from Arabidopsis leafs using a Qiagen RNeasy plant mini kit (Qiagen Ltd.) as recommended by the manufacturer. To monitor transgene expression, cDNA was synthesized from 2 mg of total RNA and used as the template in a RT-PCR using the same CH primers as described above, and with ACTIN2 as the reference gene. The primers for ACTIN2 were 5#-ACCAGCTCTTCCATCGAGAA-3# and 5#-GAACCACCGATCCAGACACT-3#, which yielded an amplified product of 340 bp. RNA extraction and gel blot analysis of potato transformants were as described (Arnqvist et al., 2003).

Quantitative Real-Time PCR Analysis of SMT1 and SMT2 Expression
Total RNA extracted from rosette leaves was DNAse treated and quantified with a spectrophotometer. All preparations were as a control visually inspected for a distinct appearance of rRNA after agarose gel electrophoresis. cDNA was synthesized using SuperScript III reverse transcriptase (Invitrogen Corporation) according to the manufacturer. QPCR reactions were performed with ABsolute QPCR SYBR green fluorescein mix (Thermo Fisher Scientifics Inc) and analyzed on the iQ5 cycler real-time PCR detection system (Bio-Rad Laboratories Inc) following the manufacturer's directions. PCR conditions were 95°C for 15 min, followed by 40 cycles of 95°C for 10 s, and annealing/ extension at 60°C for 30 s. A melt curve analysis was performed by gradually increasing the temperature from 60°C to 98°C. Threshold values were set manually within the exponential phase of the logarithmic scale amplification plot, and relative expression values were calculated using the 2 2DDC T method (Livak and Schmittgen, 2001). Primer pairs were: 5#-GTATGCTCTTCCTCAT-GCTATCCTT and 5#-TTCCCGTTCTGCGGTAGTG (for ACTIN); 5#-TGCA-CAAAGATGGAAAGGAG and 5#-CGGTAACAACTGAATTGCTG (for SMT1); 5#-TGGGTAGGCTTGCTTATTGG and 5#-ATTCCGGTTTCACCTCCTCT (for SMT2). Specificity of SMT1 and SMT2 primer pairs for the respective gene was verified by PCR cross analysis of cDNA clones.

Hydroxysterol Analysis
Lipids were extracted from crushed leaf samples (2 g f.w.) with chloroform: methanol (2:1; v/v) at room temperature for 90 min, and with between 2 and 10 mg of 19-hydroxycholesterol added as an internal standard. For isolation of unsaponifiables, lipid extracts were dried, resolved in 1 mL dichloromethane, and saponified for 20 h at room temperature by adding 3.3 mL 2 M ethanolic KOH. After addition of 2.3 mL of dichloromethane and 3.3 mL water the samples were thoroughly mixed, left for 30 min, and centrifuged for 1 min 2,500 rpm. The water phase was removed and the washing step was repeated twice to obtain a clear organic phase. Samples were dried under nitrogen and dissolved in hexane:diethyl ether (75:25, v/v). For wild-type analyses 20 g f.w. were extracted, and volumes were increased accordingly. Hydroxylated sterols were then separated from other unsaponifiables by 2-fold SPE (Dutta and Appelqvist, 1997). Trimethylsilyl (TMS) ether derivatives of hydroxysterols were prepared according to Dutta and Normén (1998).
Hydroxysterols were quantified by capillary GC using a GC model 6890N (Agilent Technologies) equipped with an autosampler GC PAL G6502-CTC (CTC Analytics AG). A combination of two capillary columns was used to separate the hydroxysterols. The columns, DB35-MS (10 m 3 0.2 mm, 0.33 mm) and DB5-MS (15 m 3 0.18 mm, 0.18 mm; J & W Scientific), were connected to the injector side and detector side, respectively. The peak areas were computed with Agilent ChemStation Rev. B.02.01 (Agilent Technologies), and quantified relative to the 19-hydroxycholesterol internal standard (Dutta and Appelqvist, 1997). The temperature program was 60°C for 1 min, and then raised 50°C min 21 up to 290°C, held for 10 min, and then raised 0.8°C min 21 up to 300°C and held for 17 min. The TMS ether derivatives dissolved in hexane was injected in splitless mode and injector temperature was at 260°C. The carrier gas was helium and nitrogen was used as make-up gas at a flow rate of 30 mL min 21 . A flame ionization detector with a temperature of 320°C was used for detection.
Hydroxylated sterols were characterized and determined by GC-MS performed on a GC8000 Top Series gas chromatograph (CE Instruments) coupled to a Voyager mass spectrometer (Finnigan) operated with Xcalibur version 1.2 (ThermoQuest). The same column combination was used as in the GC analysis. Injector and detector temperatures were 260°C and 310°C, respectively. Oven conditions were 60°C for 1 min and raised to 290°C at rate of 50°C min 21 and held for 10 min, and then raised to 300°C at rate of 1°C min 21 and held for 15 min. Helium was used as carrier gas and nitrogen as the make-up gas at a flow rate of 25 mL/min. The samples were injected in splitless mode. The full scan mass spectra were recorded at 70 eV, ion source temperature was 200°C, and extracted manually. GC-MS reference standards 7a-hydroxycholesterol, 7b-hydroxycholesterol, 24-hydroxycholesterol, 25hydroxycholesterol, and the internal standard 19-hydroxycholesterol, were obtained from Research Plus Inc. (Barnegat) or Avanti Polar Lipids Inc. (Alabaster). Other hydroxysterol metabolites reported were identified based on the MS fragmentation pattern as compared to published spectra (Dutta 2002(Dutta , 2004, or tentatively identified by manual interpretation. A summary of identified hydroxysterols with references is given in Supplemental Table S2.

Sterol Analysis
The levels of free desmethylsterols, 4-monomethylsterols, and 4,4#-dimethylsterols were analyzed in rosette leaf samples (400 mg f.w.) or for dwarfed plants, samples of aboveground tissue (300-800 mg f.w.). Total lipids were extracted from crushed tissue with chloroform:methanol (2:1, v/v) for 1 h at 70°C. The 4-des-, 4-mono-, and 4,4#-dimethylsterols were separated by two runs of TLC with dichloromethane as developing solvent, and extracted from the plate. Different amounts (1-5 mg) of 5a-cholestane (Sigma-Aldrich Inc.) was added to the sterol fractions as an internal standard, and sterols were quantified relative to the standard by GC as described (Arnqvist et al., 2003), and identified by GC-MS as described above. Correction for sample losses was performed by parallel analyses of a desmethylsterol mixture. Standards for sterol identification by retention time and mass spectra were obtained for cycloartenol (Steraloids), 24-methylene cholesterol (Research Plus Inc.), brassicasterol (Steraloids Inc.), 24-methylene lophenol (a gift from Dr. Bouvier-Navé, Institut de Botanique, Strasbourg, France), cholesterol, sitosterol, and stigmasterol (Sigma-Aldrich Inc.). Other sterol metabolites reported were identified based on the MS fragmentation pattern as compared to published spectra, or tentatively identified by manual interpretation of spectra. A summary of identified sterols with references is given in Supplemental  Table S2.

BR Measurement and Dose-Response Assays
Aerial parts of wild-type plants and two CH25 lines (10 g f.w.) were harvested and lyophilized. The tissues were extracted twice with 250 mL of MeOH. Deuterium-labeled internal standards were added to the extracts. Purification and quantification of BRs were carried out according to Fujioka et al. (2002).
For analysis of dose-dependent BR effects on seedling growth, seeds of wild-type Arabidopsis (Col.), one CH24 homozygote, two CH25 lines in heterozygous form, and the det2 mutant, were sterilized and sown on petri dishes containing Murashige and Skoog medium supplemented with 2% Suc and solidified with 0.25% Gelrite. The petri dishes were incubated at 20°C under a 16/8 h light cycle using fluorescent tubes giving a photon flux density of 100 mmol m 22 s 21 . After 2 weeks, seedlings were transferred to fresh petri dishes containing a gradient from 0 to 1,000 nM of either BL or CS (Chemiclones Inc.). Plates were incubated vertically for 5 d, after which the root elongation and hypocotyl length was measured to the nearest 0.1 mm. For analysis of dose-dependent effects on general development, plates were incubated horizontally for 2 weeks, and documented by photography.

Enzymatic Assay of SMT1 and SMT2 Activity
Microsomal fractions were isolated from rosette leaves (5 g f.w.) as described (Sitbon and Jonsson, 2001). Proteins were quantified by the Bradford assay using bovine serum albumin as the reference. The assay volume of 0.1 mL contained 60 mL of microsomal suspension (approximately 0.1 mg of protein) in sample buffer (pH 7.5), 1 mg mL 21 Tween 80, and 20 mM S-adenosyl-L-(methyl-14 C)-Met (1.86 TBq mol 21 ; Perkin Elmer Inc.). Cycloartenol and 24-methylene lophenol were used as the substrate at a concentration of 20 mM. Incubations with or without substrate were carried out at 30°C for 120 min, and reactions were terminated by the addition of 100 mL of 2 M ethanolic KOH. A mixture of sterol carriers (lanosterol, lophenol, and cholesterol) was added before extracting with n-hexane. Sterols were separated by two TLC runs with dichloromethane as the developing solvent, visualized by berberine staining, and fractions of 4#-4#-dimethylsterols (SMT1 activity) and 4-monomethylsterols (SMT2 activity) were scraped directly into a liquid scintillation cocktail. Radioactivity was determined by liquid scintillation, and values were corrected for activity with the endogenous substrates present in microsome preparations.

Stress Tolerance Tests
Seeds of empty-vector Arabidopsis (Col.) control transformants, CH24 and CH25 transformants, and the det2 mutant were surface sterilized and sown on petri dishes containing half-strength Murashige and Skoog medium supplemented with 2% Suc, with or without kanamycin, and solidified with 0.25% Gelrite. Seeds were grown under the same condition as mentioned previously. To obtain a more uniform size distribution in the test material, only CH25 heterozygotes were used in the test. After 14 d, seedlings were transferred to sterilized soil in separate pots (0.1 L), and further grown under short-day conditions (8 h photoperiod) in a controlled growth cabinet (120 mmol photons m 22 s 21 , 66% relative humidity, 20°C) for 21 d with constant watering. All pots received an equal amount of water when watered. Prior to the drought stress, all pots received an equal amount of water. Drought stress was initiated by withholding water for 14 d, whereas control plants were watered normally. Changes in response to the drought stress, such as leaf wilting, growth reduction, and leaf senescence were scored. In a separate test, plants were rewatered after 3 weeks of constant drought, and the number of recovered plants was scored as mentioned above, and also including a visual inspection of flower induction.
For cold tolerance tests, plants were grown in soil for 3 weeks, and then incubated for 2 weeks at 4°C under short-day conditions in a growth chamber, and scored as above. For heat tolerance tests, plants were grown during shortday conditions for 6 weeks. The temperature was then increased to 45°C during the day, and 22°C at night. Plants received equal amounts of water, and were scored after 1 week.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S3. Sterol levels in Arabidopsis CH lines.
Supplemental Figure S6. BR effects on root growth.