- American Society of Plant Biologists
Abstract
Vascular cell axialization refers to the uniform alignment of vascular strands. In the Arabidopsis cotyledon vascular pattern1 (cvp1) mutant, vascular cells are not arranged in parallel files and are misshapen, suggesting that CVP1 has a role in promoting vascular cell polarity and alignment. Characterization of an allelic series of cvp1 mutations revealed additional functions of CVP1 in organ expansion and elongation. We identified CVP1 and found that it encodes STEROL METHYLTRANSFERASE2 (SMT2), an enzyme in the sterol biosynthetic pathway. SMT2 and the functionally redundant SMT3 act at a branch point in the pathway that mediates sterol and brassinosteroid levels. The SMT2 gene is expressed in a number of developing organs and is regulated by various hormones. As predicted from SMT2 enzymatic activity, the precursors to brassinosteroid are increased at the expense of sterols in cvp1 mutants, identifying a role for sterols in vascular cell polarization and axialization.
INTRODUCTION
The plant vascular system is formed progressively during embryogenesis and postembryonic development. In organ primordia, provascular cells appear among ground cells in hierarchical patterns that give rise to the patterns of veins (Nelson and Dengler, 1997). Provascular cells proliferate and elongate in axialized files that appear to depend on the intercellular polar transport of auxin through the developing tissue. Screens for pattern-defective mutants have identified several genes that regulate or influence this process. The monopteros mutation causes a discontinuous venation pattern associated with improper vascular differentiation (Przemeck et al., 1996) and corresponds to an auxin response transcription factor (Hardtke and Berleth, 1998).
In gnom/emb30 mutants, which correspond to a brefeldin A–sensitive ADP-ribosylation factor guanine-nucleotide exchange factor (ARF-GEF), veins are unaxialized and patterned irregularly (Mayer et al., 1993; Shevell et al., 1994; Steinmann et al., 1999; Grebe et al., 2000; Koizumi et al., 2000). Additional venation pattern mutants have been described, such as the van mutants, lop1/tornado1, and scarface, but their gene products have not been identified (Carland and McHale, 1996; Cnops et al., 2000; Deyholos et al., 2000; Koizumi et al., 2000).
We recently described a number of additional mutants with various effects on cotyledon vascular pattern (Carland et al., 1999). One of these, cotyledon vascular pattern1 (cvp1), resembles the monopteros phenotype in that a discontinuous, poorly axialized venation pattern is formed. Here, we show that cvp1 corresponds to a defect in sterol biosynthesis.
Derivatives of the sterol pathway in plants and animals include both membrane sterols and signaling steroids. Recently, the membrane sterol cholesterol was shown to influence signaling and vectorial processes through its key role in the organization of lipid rafts (Simons and Ikonen, 1997). Lipid rafts appear to provide scaffolding for protein localization and distribution, membrane trafficking, and membrane signaling (Simons and Toomre, 2000). Although such rafts have not yet been described in plants, the diversity of plant sterols makes the possibility intriguing, particularly in highly polarized cells such as those in procambial strands.
In contrast to animals and fungi, in which a single major sterol is accumulated, plants accumulate many sterols (Hartmann, 1998). The most abundant plant sterols differ mainly by the number of carbon additions at the C-24 position, which are catalyzed by sterol methyltransferases (SMTs) in an S-adenosyl Met–dependent manner (Benveniste, 1986). In Arabidopsis, there are three SMTs, each of which can complement the defects in yeast SMT-deficient erg6 mutants (Husselstein et al., 1996; Bouvier-Nave et al., 1997; Diener et al., 2000). SMT1, which is the most homologous with yeast ERG6, catalyzes the initial sterol methyltransfer reaction and serves as a branch point between cholesterol and the more abundant sterols and brassinosteroids (BRs) (Diener et al., 2000). SMT2 and SMT3 catalyze a second methyltransfer, which distinguishes the sterols from BR precursors (Bouvier-Nave et al., 1997). As a result of SMT activities, cholesterol, stigmasterol, campesterol, and sitosterol accumulate (in order of increasing abundance), as do a number of downstream derivatives.
The phenotypes of sterol biosynthetic mutants suggest that there are diverse sterol-dependent processes. The single SMT1 in yeast is defined by erg6 mutants, which were recovered in mutant screens for impaired Trp uptake, increased resistance to polyene antibiotics, hypersensitivity to various metals and ion salts, sensitivity to brefeldin A, suppression of vesicle trafficking, decreased mating efficiency, and modulation of steroid receptor signaling (Parks et al., 1999). The causality between sterol deficiency and phenotype may have a different basis in each case and is unlikely to be limited to alterations in membrane rigidity or permeability. Arabidopsis smt1 mutants, in which cholesterol is aberrantly abundant, have compact rosettes, slightly reduced plant statures with blunt siliques, and a conditional root growth defect (Diener et al., 2000). Tobacco plants that overexpress soybean SMT1 have decreased cholesterol and increased sitosterol and 24-methyl cholesterol without significant effects on plant growth and development (Sitbon and Jonsson, 2001).
Antisense SMT2 plants, which have lower sterol levels and higher campesterol levels, exhibit dwarfism accompanied by reduced apical dominance, floral organ elongation, and fertility (Schaeffer et al., 2001). Plants that overexpress SMT2 contain higher levels of sitosterol and lower levels of campesterol and exhibit reduced plant stature that can be rescued with exogenous application of BRs (Schaeffer et al., 2001). fackel mutants, which are defective in C-14 reductase, display altered embryonic and postembryonic patterning, including a vascular patterning defect, as a result of abnormal cell division and expansion. In fackel mutants, both BRs and sterols after the defective step are reduced and uncommon 8,14-diene sterols accumulate (Jang et al., 2000; Schrick et al., 2000).
We cloned the gene that corresponds to the cvp1 mutant described above and found that it encodes the Arabidopsis C-24 SMT2. As predicted from the role of SMT2 in the sterol biosynthetic pathway, sterol levels were altered significantly. Based on the phenotypes of an allelic series of cvp1 mutants, we propose that cvp1 mutants are defective in sterol signaling or an aspect of membrane organization that is essential for the polarization and axialization of vascular cells.
RESULTS
Additional Alleles of cvp1 Exhibit Defects Consistent with Aberrant Cell Expansion and Cell Elongation
As reported previously, four alleles of cvp1 (cvp1-1, cvp1-2, cvp1-3, and cvp1-4) were identified from a screen of >34,000 ethyl methanesulfonate–mutagenized M2 seedlings derived from 10 M1 pools (Carland et al., 1999). cvp1-1 and cvp1-2 showed identical phenotypes and were found to have the same mutation (described below). Therefore, we compared in detail only the phenotypes of cvp1-1, cvp1-4, and cvp1-3, which correspond to weak, intermediate, and strong alleles, respectively.
All cvp1 alleles exhibited a discontinuous cotyledon venation pattern that consisted of isolated patches (Figures 1A to 1D) of vascular tissue that lacked cellular axialization. There was a reduction of vein loops in cotyledons of cvp1-4 and cvp1-3, and the veins were thicker. The reduced cotyledon expansion of cvp1-4 and cvp1-3 in the proximal-distal axis resulted in a “soup spoon”–shaped cotyledon. cvp1-3 cotyledons were smaller than those of the wild type. No vascular aberrations were apparent in the rosette leaves of the cvp1 alleles (Figures 1E to 1H). Flowers in all cvp1 alleles were normal in gross morphology. However, petals appeared scalloped, with highly serrated margins (Figures 1I and 1J), and sepals were highly serrated, with ectopic protrusions extending from the apical margins (Figures 1K and 1L). These affected foliar organs often exhibited vascular alterations such as vein regions with nonuniform thickening.
Phenotypes of cvp1 Alleles.
(A) to (H) Venation patterns of cleared cotyledons (A) to (D) and first rosette leaves (E) to (H) viewed under dark-field illumination. Arrows denote vascular discontinuities.
(A) and (E) Wild type (Columbia ecotype).
(B) and (F) cvp1-1.
(C) and (G) cvp1-4.
(D) and (H) cvp1-3.
(I) to (L) Margins of cleared petals (I) and (J) and sepals (K) and (L) viewed under dark-field illumination. Arrows indicate smooth margins of the wild type in contrast to the marginal protrusions of cvp1-1.
(I) and (K) Wild type.
(J) and (L) cvp1-1.
(M) Seven-day-old seedlings of the wild type (WT) and cvp1-3. The growth of cvp1-3 is retarded, and seedlings are smaller than those of the wild type.
(N) Vegetative stage of wild-type, cvp1-3, and cvp1-4 plants. cvp1-4 is indistinguishable from the wild type, but cvp1-3 has a compact rosette.
(O) Siliques of the wild type and cvp1. cvp1 mutant siliques are shorter and wider than wild-type siliques.
(P) Inflorescence stem internodal elongation defect in cvp1-3 compared with that of the wild type. Nodes between siliques fail to elongate, resulting in a cluster of siliques.
Bars in (A) to (D) and (I) to (L) = 250 μm; bars in (E) to (H) = 500 μm; bar in (M) = 1 mm; bar in (N) = 1.5 cm; bar in (O) = 3 mm; bar in (P) = 1 cm.
cvp1-1 and cvp1-4 were similar to wild-type plants in general morphology and growth rate (Figure 1N), whereas cvp1-3 plants had retarded growth and were smaller at both the seedling and rosette stages (Figures 1M and 1N). The compact cvp1-3 rosette was composed of small vegetative leaves that failed to expand to wild-type size. This strong cvp1-3 allele also exhibited delayed flowering time and senescence and reduced plant height and apical dominance.
cvp1 siliques were shorter (wild type, 13.1 ± 1.1 mm; cvp1, 10.1 ± 0.5 mm [n = 61 siliques from 12 plants]) and wider than wild-type siliques (Figure 1O). However, a reduction in fertility occurred only in the strong cvp1-3 allele. As described previously, cvp1-1 had a decrease in internodal stem elongation in the apex of primary and secondary inflorescences in ∼10% of the inflorescence stems. Transverse sections of stem tissue within this mutant region revealed an overproliferation of lignified tissue (Carland et al., 1999). This phenotype occurred at a higher frequency in cvp1-3, affecting 55% of plants (77 of 139) (Figure 1P). In summary, in addition to the loss of vascular cell axialization, strong cvp1 mutants exhibited defects in elongation at the cellular level (vascular cells) and/or the organ level (stem and siliques) and an expansion defect similar to that seen in leaves.
CVP1 Encodes SMT2
We previously mapped CVP1 to chromosome 1 near MONOPTEROS between m59 and m235 (Hardtke and Berleth, 1998). To clone CVP1, a large mapping population was generated and scored for recombinants, as diagrammed in Figure 2A . Polymorphic markers were designed based on limited sequence data in the region and used to further localize CVP1 to two overlapping BACs. PCR products were amplified from the unordered BAC sequences and used as probes to screen a transformation-competent artificial chromosome (TAC) library (Liu et al., 1999), resulting in an ordered contig of TAC clones spanning the CVP1 region.
Positional Cloning of CVP1.
(A) Line drawing of CVP1 cloning strategy. cvp1 in the Columbia background was crossed to the polymorphic ecotype Landsberg erecta to generate a mapping population of 480 meiotic events. Using breakpoint analysis, CVP1 was positioned on chromosome 1 between cleaved amplified polymorphic markers m59 and g2395. The numbers of recombinants between CVP1 and polymorphic flanking markers are shown below the line with double arrows. The BACs from which markers were designed are noted at top. CVP1 was localized to the two overlapping BACs, T20H2 and F14O10. Probes designated by small thick lines were generated from these BACs and used to identify overlapping TAC clones as labeled. Only the TAC clones that yielded transgenic plants are shown.
(B) Complementation of cvp1. One TAC clone complemented the cvp1-1 mutant vascular patterning defect (K3J15) and contained six unique annotated genes relative to a noncomplementing TAC clone (K3L10). A genomic clone of SMT2 was generated and shown to complement cvp1. Bars = 250 μm.
(C) Gene structure with mutations. Motifs proposed to be functionally important, as listed from amino acid 1, are as follows: the hatched box represents the signal anchor; the box with diagonal and horizontal lines indicates the sterol binding site (Nes et al., 1999); the black box denotes a conserved region among SMTs of unknown function; S-adenosyl Met (SAM) sites are labeled boxes. Mutations within the cvp1 alleles are denoted. Specifically, cvp1-3 has a G-to-A nucleotide substitution in the initiation codon, changing Met to Ile. cvp1-1 and cvp1-2 are independent alleles that have a G-to-A nucleotide substitution, changing Asn to Asp. cvp1-4 has a C-to-T nucleotide substitution, resulting in an early termination codon.
Individual TAC clones were introduced into cvp1-1 mutants and scored for complementation of the cvp1 mutant phenotype. A complementing TAC contained six unique genes relative to overlapping noncomplementing TACs, based on gene annotation that had become available by that date. These six genes were sequenced in the cvp1-1 background, and a mutation was identified in a previously identified gene encoding SMT2 (Bouvier-Nave et al., 1997). The additional cvp1 alleles contained mutations in this gene. cvp1 was complemented by transformation with the SMT2 genomic region, confirming the identity of CVP1 (Figure 2B).
Although cvp1-1 and cvp1-2 were identified from different M1 pools and therefore should be independent alleles, they contain the same mutation in an invariant residue within the first S-adenosyl Met binding site. cvp1-4 was found to have a nonsense mutation that is predicted to produce a truncated protein. The cvp1-3 allele contained a missense mutation in the initiation codon (Figure 2C) that should result in a null allele, consistent with its more severe phenotype. The next in-frame Met residue follows the sterol binding site.
Features of the SMT2 Protein
SMT2 is one of three SMT enzymes in Arabidopsis, all of which share regions of homology. SMT1 has substrate affinities that differ from those of SMT2 and SMT3 and acts earlier in the sterol biosynthetic pathway (Figure 3) . SMT2 and SMT3 are 83% identical in amino acid sequence and act on the same substrate at the pathway bifurcation that distinguishes sterols from BRs. The predicted protein structure of SMT2 includes a hydrophobic region at the N terminus (Figure 2C). This region meets the criteria for a signal anchor (as predicted with SignalP [http://www.cbs.dtu.dk/services/SignalP]) and may serve to direct SMT2 to the endoplasmic reticulum, where sterol biosynthesis is postulated to occur. The sequence predicts three sites for binding of S-adenosyl Met, the methyl donor for SMT reactions, a sterol binding site (Nes et al., 1999), and a region of unknown function conserved among all SMTs.
The Sterol Biosynthetic Pathway.
A simplified version of the sterol biosynthetic pathway is shown. Sterol intermediates are labeled. Genes that are enzymes and that have been identified by mutation, with the exception of SMT3, are shown in uppercase letters (Klahre et al., 1998; Choe et al., 1999a, 1999b, 2000; Diener et al., 2000; Jang et al., 2000; Schrick et al., 2000). The site of SMT methyl addition, C-24, is indicated. SMT2 is shown in boldface and underlined. Brz is a BR biosynthetic inhibitor (Asami and Yoshida, 1999). Dashed arrows indicate multiple biosynthetic steps.
SMT2 Is Expressed in Several Developing Organs
We examined the effects of the cvp1 molecular lesions on SMT2 transcript levels by RNA gel blot analysis. RNA isolated from 6-day-old seedlings revealed decreases in SMT2 transcript levels in all alleles (Figure 4A) in proportion to their phenotypic severity, even though the mutations were predicted to alter translational products rather than transcription. There were undetectable levels of SMT2 transcripts in the severe cvp1-3 allele, confirming the designation of cvp1-3 as a null allele.
RNA Gel Blot Analysis of SMT.
(A) SMT transcript levels in different alleles (SMT2, SMT3, and SMT1) were hybridized sequentially to filters of the cvp1 alleles subsequent to probe removal. RNA was isolated from 6-day-old seedlings. WT, wild type.
(B) SMT2 and SMT3 exhibit similar expression profiles. RNA was isolated from the indicated organ. Roots were excised from 1-week-old seedlings. Seedlings were 1 week old. Young leaves and expanding leaves were 2 to 3 mm and 4 to 5 mm in length, respectively. Siliques were from all developmental stages. Elongating regions of the stem were selected for RNA analysis.
(C) Tissue was incubated with the indicated hormone (see Methods). Mock, no hormone, BA, benzyladenine; ACC, 1-aminocyclopropane-1-carboxylic acid; GA, gibberellic acid; IAA, indoleacetic acid, BL, epibrassinolide. Ten micrograms of total RNA was loaded onto lanes. Methylene blue staining of rRNA served as the loading control. Only the top rRNA band is shown.
The cvp1 mutations in SMT2 also affected the accumulation of transcripts for SMT1 and SMT3. The transcript levels of the highly homologous SMT3 and of SMT1, determined with gene-specific probes, were increased in cvp1-4 but were unaffected in cvp1-1. Remarkably, neither transcript was detected in cvp1-3, suggesting that the null allele is deficient in all SMT activity. We reproduced these measurements several times, because the results were unexpected. Some effects on SMT2 and SMT3 transcript levels were reported previously for smt1 mutants, but these were not as great as the effects we observed in cvp1 alleles (Diener et al., 2000). The increased SMT1 and SMT3 transcript levels in cvp1-4 and the decreases in cvp1-3 may reflect a feedback mechanism between sterol production and activation of the SMT genes. Alleles that differ in severity may disrupt domains in SMT2 involved in distinct aspects of this regulation.
The SMT2 gene was expressed in most developing tissues in a pattern that differed slightly from that of SMT3 and SMT1. RNA gel blot analysis showed that SMT2 mRNA was present in all developing tissues surveyed and revealed low-level expression in roots and siliques, moderate levels in seedling and rosette leaves, and abundant levels in stems and inflorescences (Figure 4A). SMT3 showed a similar expression profile, except that mRNA was at undetectable levels in siliques (Figure 4B). Likewise, the pattern of SMT1 mRNA (data not shown) resembled that of SMT2, in agreement with a previous report (Diener et al., 2000).
The expression of SMT2 was influenced by several hormones. We measured SMT2 and SMT3 mRNA after treatment with ethylene, cytokinin, indoleacetic acid, epibrassinolide, or gibberellic acid, as described in Methods. Hormone induction studies indicated that SMT2 was induced strongly and moderately by ethylene and cytokinin, respectively. SMT3 was induced strongly by cytokinin. SMT2 and SMT3 were induced weakly and moderately, respectively, by indoleacetic acid. The BR epibrassinolide and gibberellic acid did not appear to have any effect on SMT transcript levels. This finding suggests that SMT2 and SMT3 are induced differentially by at least three hormones that promote cell division and/or elongation.
SMT2 appeared to be expressed predominantly in regions with active cell division. We visualized SMT2 mRNA accumulation by in situ RNA hybridization (Figure 5) . As reported previously (Diener et al., 2000), SMT2 was expressed throughout the developing embryo (Figure 5A). In agreement with the RNA gel blot studies, SMT2 was expressed in a number of nascent organs. These included the primordia of leaves, sepals, stamens, gynoecia (Figures 5B to 5E), and petals (data not shown). SMT2 also was expressed in the inflorescence meristem (Figure 5C) and in developing ovules (Figure 5F). SMT2 expression at these sites was limited in each case to stages that included actively dividing cells.
In Situ Localization of SMT2 mRNA.
(A) SMT2 is expressed throughout the developing embryo. Seed coat staining is background.
(B) Longitudinal section of a 7-day-old seedling to show SMT2 expression in leaf primordia. The shoot apical meristem is not visible in this section.
(C) Longitudinal section of an inflorescence shows SMT2 expression in the inflorescence meristem and sepals.
(D) Slightly oblique section of a stage 3 flower showing intense SMT2 expression in the sepals.
(E) Longitudinal section of a stage 5 to 6 flower reveals SMT2 expression in stamens and in the gynoecium before fusion.
(F) Ovule expression of SMT2.
In (C) and (E) images were photographed with a blue filter. cpt, cotyledon petiole; gy, gynoecium; IM, inflorescence meristem; lp, leaf primordia; ov, ovules; se, sepals; st, stamens. Bars = 40 μm.
β-Glucuronidase (GUS) reporter analysis revealed some additional features of the SMT2 and SMT3 expression patterns. Both were expressed specifically in the elongation zone of the roots (Figures 6A and 6B) . In rosettes, GUS histochemical staining was apparent in cotyledons, expanding leaves, and the apical tips of young leaves (Figures 6C and 6D). These are regions in the leaf that are undergoing cell expansion and reflect the basipetal maturation of leaves. In SMT2::GUS and SMT3::GUS flowers, staining was localized to the apical region of the sepal, particularly in the vascular bundles (Figures 6E and 6F). In SMT2::GUS siliques, GUS staining also was detected in ovules (Figure 6E). These data, in conjunction with the RNA gel blot analysis of organs and hormone application, indicate that SMT2 and SMT3 have similar but not identical expression patterns.
GUS Expression Studies of SMT2 and SMT3.
(A), (C), and (E) SMT2::GUS reporter expression.
(B), (D), and (F) SMT3::GUS reporter expression.
(A) and (B) Two-day-old germinated seedlings. Note the intense GUS staining in the elongation zone of the root.
(C) and (D) Plants with rosette leaves. In young leaves, GUS activity is restricted to the apical zone (arrows).
(E) and (F) Detached flowers show GUS enzyme activity in sepals, veins, and ovules ([E] only). For the inset in (E), outer floral organs have been removed to expose the carpel.
ez, elongation zone; ov, ovules. Bars in (A) and (B) = 250 μm; bars in (C) to (F) = 2 mm.
SMT2 and SMT3 Are Functionally Redundant
SMT2 and SMT3 are highly homologous, act on the same substrate, and exhibit similar but distinguishable expression patterns (Figures 4 and 6), suggesting that the proteins are functionally redundant. Because SMT3 expression is restricted to the outer cell layers of the developing embryo, and thus is excluded from the majority of procambial cells of developing cotyledons (Diener et al., 2000), it appears that the function is provided solely by SMT2 in these locations. This probably explains the aberrant venation of cvp1 mutants and provided an experimental system to test for functional redundancy in vivo.
We expressed either SMT3 or SMT2 in the cvp1-1 background, under the control of the 35S promoter of Cauliflower mosaic virus, and observed full restoration of the vascular patterning, petal and sepal margin, and silique elongation defects in cvp1-1 (Figures 7A to 7F and data not shown). However, the stem elongation defect was not rescued by SMT3 (Figure 7G). This finding appears to reveal a role for SMT2 in the stem that is not shared by SMT3, indicating that the functional overlap is extensive but not complete.
Functional Redundancy of SMT2 and SMT3.
(A) to (F) Cleared images of cotyledons ([A] to [D]) and petals ([E] and [F]) viewed under dark-field illumination. Bars = 250 μm.
(A) Wild type.
(B) cvp1-1.
(C) and (D) cvp1-1 transformed with 35S/SMT2 (C) and 35S/SMT3 (D) to show complementation of the vascular patterning defect.
(E) and (F) cvp1-1 petal with a scalloped apical region (E) and cvp1-1 transformed with 35S/SMT3 (F) to show complementation of the petal defect.
(G) The stem internodal elongation defect is not rescued by SMT3 overexpression. A normal cvp1-1 inflorescence stem is shown at left; an affected inflorescence stem of cvp1-1::35S/SMT3 is shown at right. Three of 36 plants had affected inflorescences. Bar = 5 mm.
(H) and (I) RNA gel blot analysis of SMT2 (H) or SMT3 (I) overexpressors (OE). The first lane contains RNA from wild-type (WT) seedlings. The transcripts are weak as a result of a short exposure time. The next three lanes contain RNA isolated from three different SMT2 or SMT3 overexpressors. Each lane contained 5 μg of total RNA isolated from 9-day-old seedlings.
SMT2 and SMT3 have overlapping functions but are not completely functionally redundant. The overexpression of SMT2 and SMT3 yielded high transcript levels in the cvp1-1 background but did not produce any phenotypic abnormalities (Figures 7H and 7I). In addition, SMT2 overexpression did not have a detectable effect on SMT3 transcript levels, and likewise, SMT3 overexpression did not have an effect on SMT2 transcript levels, indicating that the cvp1 complementation is attributable specifically to SMT3 ectopic expression. The overexpression of SMT3 in the sense and antisense orientations in the wild-type background in ∼60 transgenic plants also did not produce a phenotype (data not shown).
Based on the SMT3 complementation of cvp1 and the observation that SMT3 is not expressed independently from SMT2, we propose that the lack of a phenotype in SMT3 antisense plants is caused by the functional redundancy of SMT2 and SMT3. This redundancy also may provide an explanation for the wild-type gross morphology of cvp1-4. cvp1-4 has only 35% of SMT2 wild-type RNA levels and is predicted to have a truncated SMT2 protein. However, increased levels of SMT3 mRNA accumulate in this allele. We propose that the wild-type morphology of cvp1-4 is maintained by the increase in SMT3 activity, which compensates for the decrease in SMT2 activity. The overlap in SMT2 and SMT3 expression patterns does not include the majority of cotyledon procambial cells, so a vascular patterning defect remains (Diener et al., 2000).
cvp1 Mutants Alter the Balance between Sterols and BRs
Because SMT2 catalyzes the methylation that distinguishes sterols from BRs, we determined whether the levels of sterol and BR intermediates were affected in the cvp1 alleles. The levels of 22 sterol compounds were assayed in 1-week-old cvp1-1, cvp1-3, and cvp1-4 seedlings (Table 1). Because whole seedlings were extracted for measurements, it is possible that individual cells and tissues exhibited significantly different values. In all tested alleles, the levels of intermediates before 24-methylenelophenol, the substrate of SMT2, were not altered significantly. By contrast, 24-methylenelophenol was increased in the cvp1 mutants, and products of the sterol-specific pathway, such as 24-ethylidenelophenol, sitosterol, stigmasterol, and sitostanol, were decreased. 24-Methylenelophenol also was channeled into the BR pathway.
Sterol Profile of cvp1 Alleles
These precursors to the BRs were increased in cvp1 relative to the wild type. For example, the null allele cvp1-3 showed more than a 25-fold increase in 24-methylenecholesterol. Cholesterol lacks methyl additions and is formed by bypassing the methyl addition reactions. As predicted, cholesterol was higher in cvp1 mutant alleles. Although there were gross changes in individual sterol levels, the total sterol content of cvp1 was not markedly different from that of the wild type. Therefore, sterol levels were altered significantly in cvp1, most notably in the null allele cvp1-3, by shifting the balance of intermediates in favor of the BR pathway at the expense of the sterol pathway.
Because SMT2 regulates the biosynthetic flux that controls the amount of sterol and campesterol, the alteration in the flux often is expressed as a ratio of campesterol to sitosterol. In the wild type, this ratio was equal to 0.2, whereas in cvp1, the ratio was 1.9 or greater. In general, the decrease in sterol levels of the three alleles was proportional to the severity of their phenotypes. Although the cvp1-3 null allele exhibited a reduction in transcript levels in all SMTs, it was able to produce alkylated sterols, albeit at reduced levels. The presence of sterols was reported in a null smt1 allele (Diener et al., 2000). It is possible that lateral sterol pathways not normally used in wild-type sterol production are used in smt1 and cvp1 mutants.
The BR-Specific Inhibitor Brassinazole Does Not Rescue the cvp1 Mutant Phenotype
All cvp1 mutant alleles exhibited a decrease in sterols and a concomitant increase in the precursors to the BRs. Although BR levels were not assayed because of the large amount of tissue required, the cvp1 sterol profile suggested that there was an increase in BR levels. Therefore, we determined whether the inhibition of the BR biosynthetic pathway could rescue any aspects of the cvp1 phenotype. The cvp1 vascular patterning defect occurs during embryogenesis. Thus, we applied the BR biosynthetic inhibitor brassinazole (Brz; Figure 3) (Asami and Yoshida, 1999) to soil-grown plants before bolting to test its effects on the stem internodal elongation defect.
As shown in Figure 8A , Brz application did not rescue this aspect of the mutant phenotype. Subsequent to treatment, seeds were isolated and the cotyledon vascular patterning defect was assessed in seedlings. Brz application during embryogenesis did not rescue the cvp1 vascular patterning defect, nor did it affect the wild-type venation pattern (Figures 8B and 8C), suggesting that the phenotype is not the result of altered BR levels. The cvp1 alleles exhibited a significant reduction in sitosterol and stigmasterol levels. We were unsuccessful in altering cvp1 by supplementation with these compounds, although supplementation is made difficult by the limited solubility of the compounds.
The BR Inhibitor Brz Does Not Rescue the cvp1 Mutant Phenotype.
(A) Failure of Brz to rescue the stem elongation defect of cvp1-3. Note that there is no increase in plant height in Brz-treated cvp1 plants relative to untreated cvp1 plants.
(B) and (C). Cleared cotyledons of wild-type (B) or cvp1-3 (C) seedlings whose embryos were treated with Brz, viewed under dark-field illumination.
Plants were watered with 1 μM BRZ2001 (Sekimata et al., 2001) before bolting and until dessication. For (B) and (C) seed was collected from Brz-treated plants and germinated on Murashige and Skoog (1962) medium without Brz. Bar in (A) = 2.5 cm; bars in (B) and (C) = 100 μm.
DISCUSSION
The first recognizable event associated with the differentiation of provascular tissue is cell polarization and the formation of axialized procambial files. Our analysis of mutations that disrupt the SMT2 gene demonstrates that this process is dependent on a functional sterol pathway. Arabidopsis has two very similar proteins (SMT2 and SMT3) acting at the same step in sterol biosynthesis, but only SMT2 is expressed in cotyledon vascular cells. This was important for two reasons. First, it allowed the critical observation of aberrant venation in the cvp1 mutants, establishing a vascular role for the pathway. Second, it provided an experimental opportunity to introduce ectopic SMT3, demonstrating functional redundancy.
SMT2 acts at the bifurcation of the sterol and BR pathways, at which the substrate 24-methylenelophenol is methylated to enter the sterol pathway or is not further methylated and enters the BR pathway. Thus, SMT2 regulates the ratio of campesterol, precursor of the BRs, to sitosterol, the major sterol in plants. In cvp1 mutants, we showed that this ratio, which normally favors the sterol pathway, was inverted to favor the BR pathway, possibly most dramatically in cotyledons. The phenotypic consequences could be the result of the enrichment of one or more BRs or of the depletion of one or more sterols, or a combination of both effects.
We favor the second of these possibilities. We believe that the potential increase of BR levels is unlikely to be the cause of the cvp1 phenotype, for several reasons. First, we were unable to phenocopy the cotyledon vascular defect by supplementation of wild-type plants with brassinolide. Second, we were unable to cure the cvp1 mutant phenotype by the application of Brz, an inhibitor that acts at a point in the BR pathway downstream from campesterol. Third, the BR pathway appears to be highly regulated, such that increases of early intermediates do not necessarily result in increased levels of the downstream active BR, brassinolide (Clouse and Sasse, 1998). Although we cannot exclude the possibility that BR alterations cause some of the cvp1 phenotype, we believe it more likely that sterol depletions or modifications are responsible.
Sterols as membrane-ordering components are likely to have roles in organizing the many asymmetric structures and processes that are essential for the development of axialized files of polarized cells, such as the provascular cells and procambial strands that are affected in cvp1 mutants. In addition, the sterol branch of the pathway may produce compounds distinct from BRs, with roles in intercellular signaling for the alignment of cell files and elongation axes or in conferring cell identity distinct from neighbor cells. We discuss each of these processes below in relation to the cvp1 phenotype.
Role of SMT2 in Signaling
The cvp1 venation defect may be the result of the absence or alteration of a sterol not identified previously as an intercellular or intracellular signal. Such a signal might serve to align provascular cells along an axis or to prevent neighboring cells from joining in the provascular cell fate. Vascular patterning is altered in several biosynthetic mutants that affect both sterols and BRs, including fackel, which is blocked upstream of the sterol/BR bifurcation, and dwf7/ste1 (Choe et al., 1999b), which is blocked downstream. In addition, bri1 mutants, with a defective BR receptor, also exhibit aberrant venation (Li and Chory, 1997; Wang et al., 2001). Unlike cvp1, these mutants generally show an overproliferation of phloem at the expense of xylem (Choe et al., 1999a). However, weak fackel alleles exhibit discontinuous venation patterns in cotyledons, similar to cvp1 (Jang et al., 2000).
Although we did not measure brassinolide, the active signal in Arabidopsis, directly, it is unlikely that the cvp1 defect is caused by excessive BR levels, for the reasons described above. Instead, another sterol-derived signal may be affected in cvp1 and fackel mutants. There is indirect evidence for such signaling sterols in the spatial organization of various organs. Sterol/lipid binding domains (START) have been identified in a subtype of homeodomain/Leu zipper–containing proteins, which includes Athb-8, Athb-9, Athb-14, and REVOLUTA/INTERFASCICULAR FIBERLESS (Sessa et al., 1998; Zhong and Ye, 1999; Ratcliffe et al., 2000). In the case of the radial patterning genes Phabulosa (phab) and Phavoluta (phav), which encode the proteins ATHB-14 and ATHB-9, respectively, an unidentified sterol or lipid with spatial distribution is postulated to bind to the START domain of the receptors PHAB and PHAV to activate the genes involved in radial patterning (McConnell et al., 2001).
SMT2 and the Synthesis of Cellulose Microfibrils
The cvp1 sitosterol deficiency might cause a vein pattern defect by interfering with the pattern of cellulose synthesis in the walls of provascular cells. Recently, the major plant sterol sitosterol was suggested as a primer for cellulose synthesis, which is initiated with the conjugation of Glc to sitosterol, forming sitosterol-β-glucoside (Peng et al., 2002). Further addition of Glc residues forms sitosterol cellodextrin, which is cleaved, allowing the transfer of the cellodextrin chain to another cellulase. After the attachment of additional Glc residues, the chain then assembles into cellulose microfibrils, which are the primary scaffolding component of plant cell walls and are postulated to orient the direction of cell expansion.
Because sitosterol is reduced greatly in cvp1, both sitosterol-β-glucoside and cellulose microfibrils might be reduced as a consequence, causing aberrations in the pattern of cell wall extensibility that normally guide the elongation and morphology of provascular and vascular cells. The aberrant cell expansion patterns observed in cvp1 foliar organs and in fackel may be attributable to the same effects on cellulose priming. As discussed below, cobra mutants have reduced levels of cellulose and a concomitant alteration in cellulose microfibrils and cell expansion (Schindelman et al., 2001).
Lipid Rafts and the Compartmentalization of Proteins
In other systems, sterols have been shown to influence protein activity through their central role in organizing lipid rafts. Lipid rafts are membrane microenvironments that consist of dynamic compositions of sterols and lipids; they were identified initially in polarized epithelial and neuronal cells, in which they accumulate at a polar locality in the plasma membrane (Simons and Ikonen, 1997; Simons and Toomre, 2000). Through the inclusion or exclusion of specific proteins on the basis of affinity, rafts lead to the clustering of membrane-bound and membrane-associated proteins and serve as compartments for protein signaling complexes, cytoskeletal associations, channels, and other functional features of membranes. Numerous proteins, such as glucosylphosphatidylinositol (GPI)-linked proteins and transmembrane proteins, have been demonstrated to reside in such rafts. In many cases, sterol depletion results in the dissociation of membrane-bound proteins from lipid rafts, leading to mislocalization of the protein and/or inactivation of the signaling cascade (Simons and Toomre, 2000).
Rafts are assembled initially with endoplasmic reticulum–derived sterols in the Golgi, the site of lipid synthesis, and then are transported to the plasma membrane carrying proteins destined for polarized delivery. After protein delivery, rafts are endocytosed from the cell surface and then recycled either directly back to the plasma membrane or indirectly by endosomes. In plants, there are low sterol levels in the endoplasmic reticulum, where sterols are synthesized as corroborated by the presence of signal anchors on SMT2 and SMT3, and high levels in the plasma membrane, providing support for this mode of transport.
Although lipid rafts have not been observed in plants, their small size may have precluded observation by conventional microscopy. In other systems, raft components have been resolved through advanced microscopy techniques coupled with biochemical methods. Recently, a GPI-anchored protein was found to be the product of the Arabidopsis COBRA gene, whose mutant phenotype includes a loss of polarized, longitudinal expansion (Schindelman et al., 2001). The COBRA protein is localized asymmetrically in polarized cells, much like GPI-anchored proteins in other systems, and is postulated to play a role in oriented cell expansion, possibly through the recruitment of enzyme components involved in cellulose deposition. Although not demonstrated directly, this localization has the properties expected of a membrane raft-localized protein.
Other candidates for lipid raft localization in plants are the PIN family of auxin efflux carriers (Galweiler et al., 1998). Auxin flow is postulated to direct vascular cell patterning through the vectorial transport of auxin by basally localized efflux carriers. The mislocalization of a PIN efflux carrier in cotyledons of cvp1 mutants might lead to the failure to form axialized files of provascular cells. PIN is cycled between the plasma membrane and an undefined endosomal compartment requiring ARF-GEF (Geldner et al., 2001) encoded by EMB30/GNOM (Shevell et al., 1994; Busch et al., 1996; Steinmann et al., 1999; Geldner et al., 2001). Although PIN-containing vesicles are not derived directly from the Golgi, such cycling is seen with other lipid raft-associated proteins.
It is tempting to speculate that PIN asymmetric delivery acts in association with lipid rafts. Alternatively, PIN may reside in plasma membrane microdomains rich in sterols and lipids. Although pin mutants have not been reported to exhibit any cotyledon vascular patterning discontinuities, emb30/gnom mutants exhibit a loss of vascular cell axialization similar to that observed in cvp1 mutants, albeit much more severely (Mayer et al., 1993; Koizumi et al., 2000). There are many members of the PIN family, so functional redundancy probably exists. In emb30/gnom, vesicle transport–dependent polar localization of all PIN members would be affected; thus, a much more severe phenotype would occur.
Based on the known role of sterols in mammalian and yeast cells, it is likely that sterols act in more than one process. We propose that the cvp1 mutant phenotype is not caused by a disruption in membrane permeability but rather reflects either (1) the mistargeting of proteins involved in polarizing and aligning vascular cells or (2) the synthesis of novel sterol signaling molecules. These two possibilities are not mutually exclusive. Sterol signaling may involve a receptor that subsequently initiates a signaling cascade that resides in a plasma membrane microdomain. The further analysis of cvp1 and other sterol-defective mutants should reveal the roles of sterols in vascular pattern formation and other developmental processes.
METHODS
Plant Growth Conditions
All cvp1 alleles were recovered from an ethyl methanesulfonate–mutagenized population of Arabidopsis thaliana ecotype Columbia in the screen described previously (Carland et al., 1999) and backcrossed at least three times. Seeds were surface-sterilized, plated on Murashige and Skoog (1962) (MS) medium (Sigma) with (for transgenic plants) or without antibiotics, and grown under continuous light. Approximately 1-week-old seedlings were either collected for RNA gel blot analysis, β-glucuronidase (GUS) staining, or sterol analysis or transplanted to soil and grown to maturity for additional manipulations. Wild-type and mutant plants were at the same developmental stage for comparative morphological studies. Plants in soil (Metromix; Scotts, Hope, AR) were grown at 22°C with a 16-h day and a light intensity of 200 μmol·m−2·s−1. The phenotype of cvp1-3 plants grown at a higher light intensity (290 μmol·m−2·s−1) was not as severe.
Histology and GUS Histochemical Staining
To score cotyledon defects, one cotyledon was detached and cleared of chlorophyll by immersing cotyledon in fixative (ethanol:acetic acid [3:1]) and then rinsed in 70% and incubated in 100% ethanol at 4°C overnight. Cotyledons were viewed with a dissecting microscope. For photography, after the 100% ethanol dehydration step described above, specimens were further cleared by 1 h of incubation in 10% NaOH at 42°C and mounted in 50% glycerol. Vascular patterns were viewed under dark-field illumination using a Zeiss Axiophot microscope (Jena, Germany).
Plant organs were placed in 5-bromo-4-chloro-3-indolyl-β-glucuronic acid (X-Gluc) staining solution, infiltrated under vacuum for 10 min, wrapped in foil, and placed at 37°C. GUS staining was monitored for 2 to 12 h. X-Gluc staining solution of aerial organs contained 1 mM X-Gluc, 1 mM Na-phosphate buffer, pH 7.0, 1.0 mM EDTA, pH 8.0, 0.1% Triton X-100, 1 mM β-mercaptoethanol, 100 μg/mL chloramphenicol, 0.5 mM K3Fe(CN)6, and 0.5 mM K4Fe(CN)6. Improved specificity of X-Gluc staining in roots was obtained by following the method of Malamy and Benfey (1997). Samples were rinsed in 70% ethanol two times and mounted in 50% glycerol for photography using a Zeiss Stemi 2000-C dissecting microscope.
Positional Cloning of CVP1
cvp1-1 (ecotype Columbia) was crossed to Landsberg erecta. The resulting hybrid was self-pollinated to generate a large mapping population. Medium-grown F2 seedlings were scored for the mutant cotyledon vascular patterning phenotype. cvp1 mutants were transferred subsequently to soil and allowed to mature. DNA was isolated from leaf tissue and scored for recombinants with the flanking cleaved amplified polymorphic markers g2395 and m59 (Hardtke and Berleth, 1998). Seeds were harvested from the cvp1/g2395 and cvp1/m59 recombinants. At this time, we noticed that cvp1-1 in the Landsberg erecta background had a pleiotropic phenotype that included dwarfism and a severely stunted silique growth defect accompanied by a drastic reduction in fertility. This phenotype did not segregate with the Landsberg erecta mutation.
CVP1 was localized further by searching for polymorphic markers. To obtain new polymorphic markers, as BAC sequences became available in this region, 0.8-kb regions of DNA were amplified by PCR from Landsberg erecta, sequenced, and compared with the Columbia sequence. We designed several novel markers with this method and used them to further define the position of CVP1 to an ∼150-kb region spanning two BACs. These markers are available upon request. Transformation-competent artificial chromosome (TAC) filters, obtained from the ABRC (Columbus, OH), were probed with 1-kb probes amplified from the unordered working-draft BAC sequences and hybridized using a formamide-based buffer, according to the protocol provided by Bio-Rad for Zetaprobe membranes. The positive TAC clones then were ordered into a TAC contig based on their hybridization signals.
Although eight TAC clones were transformed into cvp1-1, seeds were obtained from only five TAC clones because of low transformation frequency. Restriction analysis and gene annotation indicated the presence of six unique genes in a complementing TAC relative to an overlapping noncomplementing TAC. Upon sequencing cvp1-1 alleles of these genes, a mutation was identified only in the SMT2 gene. The remaining cvp1 alleles were amplified by PCR, cloned into pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA), and sequenced to identify base-pair changes within SMT2.
Plasmid Construction and Plant Transformation
All genes used in this study were PCR-amplified products from Arabidopsis ecotype Columbia that were introduced into pCR2.1-TOPO cloning vector, sequenced to ensure that there were no base-pair changes, and subcloned into the appropriate vector using compatible restriction sites within the polylinker. For complementation of cvp1, a genomic clone containing the 1.5-kb upstream region, coding region, and 3′ untranslated region using primers MT10 (5′-CGAGTGAGTCAGTCATAT-3′) and MT9 (5′-GTCAGTAGTGTTACTCACACAGGC-3′) was cloned into pCAMBIA 2300. For GUS constructs, 1.5-kb upstream regions of SMT2 (MT10 and MT7 [5′-CGTTAAGAGTGAGGAAGACC-3′]) or SMT3 (SMT3-1 [5′-CACAGGGAGAAAGAGAGAAGC-3′] and SMT3-2 [5′-CGTGTGAGCAAATAGATCACG-3′]) were cloned into both pBI101 (Clontech, Palo Alto, CA) and pCAMBIA 1381.
For overexpression of SMT2 or SMT3, because there are no introns, a genomic region representing the cDNA of SMT2 (MT1 [5′-GGTCTTCCTCACTCTTAACG-3′] and MT9) or SMT3 (SMT3-6 [5′-CAGAGTCGTGAACTTAACG-3′] and SMT3-7 [5′-CCAATAGAATTTCCCGGC-3′]) was amplified by PCR and eventually cloned into pZP35 (Hajdukiewicz et al., 1994) that had been engineered with a nopaline synthase 3′ terminator sequence. SMT3 cDNA also was introduced in the antisense orientation in pZP35nos. Additional information on the design of these constructs will be provided upon request. For each construct, 25 to 60 transgenic plants were analyzed for GUS histochemistry or phenotype and/or RNA gel blot analysis.
Plant transformations were conducted (Bechtold and Pelletier, 1998) with modifications (Clough and Bent, 1998).
Hormone Induction Studies and RNA Gel Blot Analysis
For hormone induction experiments, a modification of a method described previously (Gil et al., 1994) was used. One-week-old light-grown seedlings were cut into 0.5-cm pieces and incubated in 0.5 × MS salts, 1% Suc, and 50 μM cycloheximide. After 4 h of incubation at room temperature, the buffer was replaced with fresh buffer containing the appropriate hormone and incubated for an additional 2 h. The tissue then was blotted dry with tissue paper, frozen in liquid N2, and stored at −70°C. Hormone levels were as follows: 20 μM indoleacetic acid, 20 μM benzyladenine, 50 μM gibberellic acid, 20 μM 1-aminocyclopropane-1-carboxylic acid, and 2 μM 24-epibrassinolide. Indoleacetic acid, gibberellic acid, 1-aminocyclopropane-1-carboxylic acid, and benzyladenine were obtained from Sigma. 24-Epibrassinolide was obtained from CIDtech Research (Cambridge, Ontario, Canada). The experiment was conducted two times and yielded the same results.
For RNA gel blot analysis, RNA was isolated using the Trizol method (Gibco BRL). Unless noted otherwise, 10 μg of RNA was loaded onto formaldehyde gels, electrophoresed under denaturing conditions, and transferred to a nylon membrane (Zetaprobe; Bio-Rad). To ensure equal loading of RNA, filters were stained in 0.02% methylene blue and 0.3 M NaOAc, pH 5.5, according to a published method (Herrin, 1988). After cross-linking RNA to filter, the filters were hybridized to gene-specific probes (SMT2, MT1 and MT9; SMT3, SMT3-6 and SMT3-7; SMT1, SMT1-1 [5′-CTCCGATTCATCTTTATCCTC-3′] and SMT1-2 [5′-GGGCATGTGCACATGATTCAG-3′]) using a formamide-based hybridization buffer according to the manufacturer's protocol (Bio-Rad). After hybridization, filters were washed in 0.5 × wash buffer (0.075 M NaCl, 0.0075 M Na-citrate, 2.5% SDS, and 0.05% NaPPi) at 55°C. Under these conditions, there was no cross-hybridization between probes. SMT1 was amplified from 1-week- old seedling RNA using the 5′ rapid amplification of cDNA ends system (Gibco BRL) for first-strand synthesis. Density measurements were made using Image Gauge version 3.3 (Fuji Photo Film Co., Tokyo, Japan).
RNA in Situ Hybridization
A full-length SMT2 probe and the SMT2 3′ untranslated region (MT11 [5′-GGTAGAAAGGAAACATCACCGG-3′] and MT9) were cloned into pBluescript II SK+ (Stratagene) in both orientations and digested with the appropriate enzyme (BamHI) to yield templates for sense and antisense in vitro transcription according to the manufacturer's suggestions (Roche, Indianapolis, IN). In situ hybridization experiments were conducted (Long et al., 1996) with a hybridization temperature of 42°C. The SMT2 3′ untranslated region probe yielded the same signal as the full-length probe and did not hybridize with SMT3. The signal was apparent after 2 days. Sense controls showed no signal. Photography was conducted using a Zeiss Axiophot microscope.
Measurement of Sterol Levels
Levels of 22 sterol intermediates were assayed from 1-week-old seedlings. Plant tissue was frozen and lyophilized before extraction, purification, and gas chromatography–mass spectrometry analysis (Noguchi et al., 1999).
Upon request, all novel materials described in this article will be made available in a timely manner for noncommercial research purposes. No restrictions or conditions will be placed on the use of any materials described in this article that would limit their use for noncommercial research purposes.
Acknowledgments
We are grateful to Brian Keith (University of Pennsylvania) for providing the cvp1 alleles and to Asami Tadao (Institute of Physical and Chemical Research) for providing Brz. We thank Lien Lai (Ohio State University) for advice on positional cloning. We gratefully acknowledge Neil McHale (Connecticut Agricultural Experiment Station) for a critical review of the manuscript. We acknowledge the ABRC for providing the TAC filter and BAC and TAC clones used in the cloning of CVP1. This research was supported by Grants IBN-9808295 and IBN-0110730 from the National Science Foundation.
Footnotes
Article, publication date, and citation information can be found at www.plantcell.org/cgi/doi/10.1105/tpc.003939.
- Received April 18, 2002.
- Accepted May 27, 2002.
- Published September 1, 2002.
References
