The sterol methyltransferases SMT1, SMT2, and SMT3 influence Arabidopsis development through nonbrassinosteroid products.

Plant sterols are structural components of cell membranes that provide rigidity, permeability, and regional identity to membranes. Sterols are also the precursors to the brassinosteroid signaling molecules. Evidence is accumulating that specific sterols have roles in pattern formation during development. COTYLEDON VASCULAR PATTERNING1 (CVP1) encodes C-24 STEROL METHYLTRANSFERASE2 (SMT2), one of three SMTs in Arabidopsis (Arabidopsis thaliana). SMT2 and SMT3, which also encodes a C-24 SMT, catalyze the reaction that distinguishes the synthesis of structural sterols from signaling brassinosteroid derivatives and are highly regulated. The deficiency of SMT2 in the cvp1 mutant results in moderate developmental defects, including aberrant cotyledon vein patterning, serrated floral organs, and reduced stature, but plants are viable, suggesting that SMT3 activity can substitute for the loss of SMT2. To test the distinct developmental roles of SMT2 and SMT3, we identified a transcript null smt3 mutant. Although smt3 single mutants appear wild type, cvp1 smt3 double mutants show enhanced defects relative to cvp1 mutants, such as discontinuous cotyledon vein pattern, and produce novel phenotypes, including defective root growth, loss of apical dominance, sterility, and homeotic floral transformations. These phenotypes are correlated with major alterations in the profiles of specific sterols but without significant alterations to brassinosteroid profiles. The alterations to sterol profiles in cvp1 mutants affect auxin response, demonstrated by weak auxin insensitivity, enhanced axr1 auxin resistance, ectopically expressed DR5:β-glucuronidase in developing embryos, and defective response to auxin-inhibited PIN2-green fluorescent protein endocytosis. We discuss the developmental roles of sterols implied by these results.

Sterols are integral components of the cell membrane and serve to provide stability to the phospholipid bilayer. In addition to this structural role, in higher plants, animals, and insects, sterols are precursors of the steroid hormones. There are over 100 types of sterols identified in plant species, but the most abundant are sitosterol, stigmasterol, and campesterol (Guo et al., 1995;Hartmann, 1998). The diversity in plants is in marked contrast to the single sterol accumulated in fungi and animals, ergosterol and cholesterol, respectively (Benveniste, 2004). The balance of phytosterols with varied properties has been postulated to provide membrane integrity under the broad range of temperatures to which plants are exposed (Beck et al., 2007).
Sterols are isoprenoid derivatives with a four-ring steroid nucleus (Edwards and Ericsson, 1999). A distinguishing feature of cholesterol and the major phytosterols is the number of carbons at the carbon 24 (C-24) position (Benveniste, 1986). Phytosterols have one to two additional carbons due to alkylation events catalyzed by S-adenosyl-L-Met-dependent C-24 STE-ROL METHYLTRANSFERASEs (SMTs; Fig. 1). SMT1 catalyzes the initial step in sterol biosynthesis by a single methyl addition at C-24 (C 1 addition). SMT2 and SMT3 can each add a second methyl group (C 2 addition), thus completing an ethyl side chain addition on C-24 (Husselstein et al., 1996;Bouvier-Navé et al., 1997;Schaller et al., 1998;Nes et al., 1999). In plants, cycloartenol is the primary starting compound in sterol biosynthesis and is the substrate of SMT1, although there is some promiscuity in substrate specificity among the SMTs (Diener et al., 2000). Arabidopsis (Arabidopsis thaliana) SMT2 and SMT3 can promote both C 1 and C 2 additions in a yeast ergosterol6 mutant blocked at SMT1 and SMT2, and SMT1 can perform both methyl additions in bacteria (Husselstein et al., 1996;Bouvier-Navé et al., 1997). Recently, a second sterol pathway was identified that utilizes lanosterol rather than cycloartenol in its initial step. This lanosterol pathway is a minor branch, estimated to contribute a small fraction (1.5%) of total sitosterol (Ohyama et al., 2009). Redundancy among enzymes and path-ways ensures a balanced sterol composition and underscores the importance of sterols in plant growth and development.
Evidence is growing that specific sterols have regulatory roles in plants, independent of their contribution to brassinosteroid (BR) biosynthesis (Schaller, 2004). First, the phenotypes of sterol biosynthetic mutants are distinct from those in the downstream BR biosynthetic pathway. Mutants including sterol methyltransferase1 (smt1; also called cephalopod [cph]), fackel (fk; allelic to hydra2 [hyd2]), hyd1, cyclopropylsterol isomerase (cpi), and cotyledon vascular pattern1 (cvp1) reveal roles for sterols in embryo, vein, shoot and root patterning, cell expansion, polarity and proliferation, fertility, cellulose level maintenance, gravitropism, and hormone signaling (Diener et al., 2000;Jang et al., 2000;Schrick et al., 2000Schrick et al., , 2002Schrick et al., , 2004aCarland et al., 2002;Peng et al., 2002;Souter et al., 2002Souter et al., , 2004Willemsen et al., 2003;Men et al., 2008). None of these are rescued by supplementation with BRs. In contrast, the dwarf phenotypes of BR biosynthetic mutants, including dwarf7/sterol1 (dwf7/ste1) and dwf1/dim, are rescued by exogenous brassinolide (BL) application (Choe et al., 1999a(Choe et al., , 1999b. Second, transcriptional profiling experiments suggest that different sets of downstream genes are affected in sterol and BR pathway mutants. Microarray studies indicate some overlap between BR-and sterolregulated cell expansion and division genes in the sterol mutant fk and the BR mutants de-etiolated2 (det2) and dwf4, but overall the profiles are highly distinct, suggesting that independent targets are activated (Goda et al., 2002;He et al., 2003). Sterol mutants share characteristics and pathways with other hormones, providing additional evidence for sterols as hormonal regulatory molecules. Cholesterol, the major animal sterol, has a well-documented role as a transcriptional regulator through its modification of the C-24, the site of SMT methyl additions, is indicated. Multiple steps are designated with dashed lines. Mutants are shown in italics with corresponding enzymes in uppercase letters. dwf1/dim has homology to FAD-linked oxidase and is predicted to have C-24 reductase activity.
Hedgehog (Hh) ligand for the cell surface receptor Patched (Ptc; Incardona and Eaton, 2000). The Hh signal transduction pathway activates target genes in cell expansion and proliferation during embryonic and adult plant development (Duman-Scheel et al., 2002).
Third, sterol mutants influence membrane structure and traffic. The sterol biosynthetic mutants smt1 orc , fk/hyd2, cpi, and cvp1 exhibit misdistribution of the polarly localized PIN protein, an efflux transporter of auxin, suggesting a sterol requirement at the level of PIN endocytosis (Simons and Ikonen, 1997;Souter et al., 2002;Willemsen et al., 2003;Men et al., 2008;Pan et al., 2009). In animals, cholesterol influences the polar trafficking of proteins through its ability to interact with sphingolipids in specialized membrane microdomains or lipid rafts (Simons and Ikonen, 1997). These membrane compartments serve to concentrate associated proteins for enhanced interaction and, thus, more efficient cellular processes. Likewise, sterol regulation in plants may not be solely at the transcriptional level.
Fourth, the sterol biosynthetic genes are expressed in regions of active cell division and expansion. Indeed, sitosterol, stigmasterol, and some abnormal fk sterols up-regulate characteristic cell expansion and proliferation genes (He et al., 2003). Sterol balance is affected in all mutants, but not always as predicted based on a simple linear pathway. For example, compromised SMT1 activity does not completely restrict further sterol transformations, as some downstream sterol levels remain unaffected (Diener et al., 2000). Consistent with this result, the smt1 mutant phenotype, particularly at the adult stage, is relatively mild compared with other sterol biosynthetic mutants, even though smt1 acts at the initial step. The downstream but more severe sterol mutants fk, hyd1, and cpi display a more drastic reduction in sterol levels, accumulate abnormal sterols, and show decreased BR precursor levels (Souter et al., 2002;Schrick et al., 2004a;Men et al., 2008). Furthermore, genetic data indicate a function for HYD1 and FK independent from SMT1 and suggest that the sterol biosynthetic pathway is more complex than previously believed (Schrick et al., 2002).
In this study, we provide further evidence that specific sterols influence numerous processes in plant development, independent of BR action. We previously reported on CVP1 as encoding SMT2, a branch point enzyme functioning to balance sterol and BR levels (Carland et al., 2002). Although SMT2 is expressed in regions of rapid cell division and cell expansion throughout development, the phenotypic abnormalities of cvp1 mutants are predominantly restricted to a cotyledon vein pattern defect and do not share the gross embryo defects of smt1, cpi, fk, and hyd1. The close relative, SMT3, has a similar expression profile. Since overexpression of SMT3 can complement cvp1 mutants, we reasoned that the mild cvp1 phenotype may be due to genetic redundancy with smt3. Here, we assign function in plant development to the third SMT and report causal effects of depleting C-24 ethylidene SMTs both at the phenotypic and sterol levels. Our results reveal some novel roles for sterols, particularly in flower development. A complete BR profile shows that BR levels are unaffected in the cvp1 mutant alleles, indicating that the phenotypic abnormalities are BR independent.

RESULTS
Identification of the smt3 Mutant SMT2 and SMT3 genes are highly homologous (83% identity) and encode highly similar sterol 24-carbon methyltransferases (Diener et al., 2000;Carland et al., 2002). The overexpression of SMT3 complements the cotyledon venation pattern defects of cvp1 (SMT2) mutants. In addition, antisense smt3 plants did not display a phenotype. These results suggest that SMT3 and SMT2 are functionally redundant. To investigate the effects of complete deficiency in C-24 ethylidene SMTs, we identified a T-DNA insertion in the coding region of SMT3 from the SALK SIGNAL collection ( Fig. 2A) and combined it with the cvp1 mutant allele of SMT2. The Arabidopsis SMTs are members of the methyltransferase superfamily, with the signature methyltransferase domain and an S-adenosyl-L-Met-binding region. S-Adenosyl-L-Met provides the methyl group for transfer to the SMT substrate, 24-methylenelophenol (24-ML). The T-DNA insertion resides at amino acid 55 and is predicted to eliminate expression of the 359-amino acid SMT3 protein. RNA gel-blot analyses showed the absence of an SMT3 transcript in smt3 mutants (Fig.  2B). As has been demonstrated previously with cvp1 and smt1 mutants, SMT1 and SMT2 transcript levels were reduced in the smt3 mutant, suggesting that SMT transcription may be regulated through a sterol feedback mechanism (Diener et al., 2000;Carland et al., 2002).

SMT3 Is Functionally Redundant with SMT2
Although SMT3 transcripts are absent and SMT1 and SMT2 RNA levels are reduced in smt3 mutants, smt3 mutant seedlings are indistinguishable from the wild type (Fig. 3A). This result is consistent with the absence of an antisense plant phenotype and with genetic redundancy. cvp1 seedlings have round, cupped epinastic cotyledons. cvp1 smt3 double mutants are smaller and show enhanced epinasty with rounded, curved cotyledons ( Fig. 3A; Table I). In comparison with wild-type cotyledon vein patterns, which consist of three to four continuous, closed loops, cvp1 mutant cotyledons have a fragmented and less reticulated vein pattern (Table I; Carland et al., 1999). cvp1 vascular cells are misshapen and exhibit abnormal planes of division that form knots of misaligned vascular cells (Fig. 3, B-D; Carland et al., 1999). Although smt3 cotyledon vein patterns are identical to that of the wild type, the patterns of cvp1 smt3 double mutants are less reticulated than either single mutant parent, with numerous misaligned vascular cells (Fig.  3, E-G; Table I). Defective SMT1 activity, which blocks sterol biosynthesis at an earlier step, results in similar vascular pattern defects ( Fig. 3H; Willemsen et al., 2003). When a construct containing an SMT3 genomic clone was introduced into cvp1 smt3 double mutants, it restored the less severe cvp1 phenotype, indicating that the enhanced phenotypes observed in the cvp1 smt3 double mutant were due to a dosage deficiency in SMT2 and SMT3 function (Fig. 3I). The rosette leaf vein patterns in the cvp1 mutant resembled those of the wild type and of smt3 mutants (Fig. 3J). However, the apical loop of the cvp1 smt3 rosette consisted of poorly axialized cells and often freely terminated to form an open areole (area bordered by veins). At later stages, slow-growing cvp1 smt3 double mutant plants displayed reduced stature and a loss of apical dominance, resulting in a bushy appearance (Fig. 3K), similar to cosuppressed SMT2 lines (Schaeffer et al., 2001). cvp1 smt3 double mutants have short siliques and poor seed yield, reflecting either gametophytic or early embryo lethality ( Fig. 3L; Supplemental Table S1). These results suggest that SMT3 is functionally redundant with SMT2 and that a complete deficiency in C-24 ethylidene SMT activity enhances the phenotypic abnormalities of cvp1 by causing more absolute changes in the composition of sterol products.

Mutants Deficient in Both SMT1 and SMT2 Are Embryo Lethal
To investigate the effect of combined SMT1 and SMT2 deficiency, affecting two distinct steps in the sterol pathway, we attempted to generate a cvp1 smt1 double mutant utilizing smt1-1-linked hygromycin resistance (HYG R ) as a selectable marker (Diener et al., 2000). Although we were able to construct a cvp1 homozygous, smt1 heterozygous (cvp1/cvp1 SMT1/smt1;HYG R ) plant, all self-fertilized HYG R progeny (n = 101) were confirmed as the same genotype, suggesting that cvp1 smt1 double mutants were embryo lethal (expect 19 cvp1 smt1 double mutants). cvp1/cvp1 SMT1/smt1 plants were identical to cvp1 plants, with the exception of reduced silique length and an increase in aborted embryos (Supplemental Table S1; data not shown). This result confirms the sensitivity of embryo development to the balance of sterol species. cvp1 smt3 Flowers Exhibit Morphological and Cellular Defects More Severe Than Those in Single Mutants cvp1 is allelic to frill1 (frl1), which was identified by its sepal and petal tip serrations (Hase et al., 2000(Hase et al., , 2005. In cvp1 mutants, serrations are restricted to the apical surface. cvp1 smt3 double mutants have enhanced sepal and petal margin projections with highly abnormal cell expansion and division patterns that extend beyond the tips to lateral positions of the organs (Fig. 4, A-C). Enlarged nuclei were observed by 6-diamidino-2-phenylindole (DAPI) staining of frl1 petal and sepal tip cells and were found to correlate with ectopic endoreduplication (Hase et al., 2005). Unlike wild-type petals and sepals, which had nuclei of uniform size, shape, and distribution, cvp1 smt3 petal and sepal lobes displayed nonuniform, large, and often elongated nuclei (Fig. 4,D and E). This result is consistent with the reported cell expansion defects in other sterol mutants and supports the hypothesis that sterol composition influences cell division and expansion.
Flowers of cvp1 smt3 double mutants display organspacing defects and homeotic organ transformations. Unlike wild-type, cvp1, and smt3 flowers, in which floral organs are uniformly spaced within four concentric whorls, cvp1 smt3 petals are improperly spaced, with overlapping petals and subsequent gaps (Supplemental Fig. S1). Late-forming flowers display unfused leaf-like carpels, increased carpel number, and homeotic transformations in which additional floral organs arise from fourth whorl gynoecia and deeply lobed leaf-like structures have stigma-like tips (Fig. 5). Floral defects are specific to cvp1 smt3 flowers and do not appear in smt1 (Supplemental Fig.  S1). These results suggest that the alteration in sterol biosynthesis caused by SMT2 and SMT3 deficiencies has a significant impact on the proper growth of flowers.
Root Defects in cvp1 smt3 Double Mutants Resemble Those in smt1 SMT2 is expressed in internal cells throughout the developing embryo and has significant overlap with SMT3 expression in both early and late stages of root growth. Although cvp1 and smt3 mutant roots are indistinguishable from those of the wild type, many sterol biosynthetic mutants, including smt1, have se- verely affected roots (Diener et al., 2000;Willemsen et al., 2003;Schrick et al., 2004a). Interestingly, cvp1 smt3 double mutant roots displayed the severely affected roots of smt1 mutants. cvp1 smt3 roots were much shorter than those of the wild type or either single parent but were similar to smt1 roots (Fig. 6A).
Scanning electron microscopy of smt1 and cvp1 smt3 mutant epidermal cells showed a complete loss of aligned longitudinal cell files and abnormal cell expansion and morphology (Fig. 6B). These results suggest that critical sterols for proper root development are downstream from SMT2 and SMT3.

Mutation in Both SMT2 and SMT3 Causes Extreme Alteration in Sterol Composition
SMT2 and SMT3 catalyze the second methyl addition onto the sterol compound 24-ML (Schaller et al., 1998;Benveniste, 2004). We compared the sterol profiles of cvp1 smt3 seedlings with those of the wild type and cvp1 and smt3 single mutants (Table II). Consistent with its wild-type appearance, sterol levels were unaffected in the smt3 mutant. Levels of sterol intermediates upstream of 24-ML were not affected in cvp1 or cvp1 smt3 mutants. However, as predicted, 24-ML was at increased amounts relative to that of the wild type and smt3. As a consequence of SMT2/3 deficiency, the quantity of the immediate product of SMT2 catalysis is reduced, particularly in the cvp1 smt3 double mutant, in which 24-ethylidenelophenol is barely detectable. Levels of downstream sitosterol and campesterol branch products are also decreased in the cvp1 single and cvp1 smt3 double mutants. The end products of the sitosterol branch pathway, sitosterol, stigmasterol, and sitostanol, are dramatically reduced in the cvp1 smt3 double mutant to only 1% to 3% of wild-type levels. Surprisingly, isofucosterol is relatively unchanged in cvp1 but is greatly reduced in cvp1 smt3. Cholesterol lacks methyl additions; thus, it bypasses SMT activities and is elevated in cvp1 single and cvp1 smt3 double mutants.
The cvp1 smt3 double mutant and the smt1 single mutant exhibit similar abnormalities in root growth, fertility, and seedling morphology (Supplemental Fig.  S2). Therefore, we compared the sterol composition in cvp1 smt3 and in smt1 to identify common features. SMT1 catalyzes the first methyl group addition using the substrate cycloartenol to initiate the sterol biosynthetic pathway. As reported previously for the cph allele of smt1, the cycloartenol content was strikingly increased in the smt1 mutant allele. Similar to cvp1 smt3 double mutants, sitosterol branch end products are in decreased quantities in smt1. The cholesterol branch of sterol biosynthesis is a minor branch, because intermediates feed into the dominant sitosterol and campesterol pathways. As previously reported with smt1-1 and other smt1 alleles, the amount of cholesterol is dramatically elevated, causing cholesterol to become the predominant sterol in smt1 mutants. Although there are many similar trends between altered sterol content in cvp1 smt3 and smt1 mutants, the most noteworthy is the decrease in the major sterol, stigmasterol, raising the possibility that this sterol is essential for normal root growth and fertility. However, we cannot rule out the possibility that elevated  cholesterol also influences the observed developmental abnormalities. To test the ability of these sterols to rescue or mimic smt defects, we performed sterol supplementation experiments. Plants treated with stigmasterol and sitosterol were not able to restore cvp1 cotyledon vein pattern or cvp1 smt3 root defects to a wild-type phenotype (Supplemental Fig. S3, A and B), and cholesterol-supplemented wild-type seedlings did not mimic smt defects (Supplemental Fig. S3, B and C), suggesting that exogenously applied end product sterols are not sufficient to reestablish sterol balance, possibly due to insufficient uptake or metabolism.

BR Levels Are Unaffected in cvp1 Alleles
The cvp1 mutation results in an aberrant accumulation of the SMT2 substrate, 24-ML. This sterol intermediate feeds into the BR pathway to yield BL, the only known active steroid signaling molecule. With elevated precursor levels, it is possible that the cvp1 mutant phenotype is due to alterations in the downstream BR levels. We previously showed that supplementation experiments of either wild-type plants with BL or cvp1 plants with the BR inhibitor brassinazole (Brz) did not mimic or rescue the cvp1 mutant phenotype, respectively, suggesting that alterations in the BR signaling molecules were not responsible for the phe-notypic abnormalities (Carland et al., 2002). To assess whether BL levels are affected in cvp1 mutant alleles, we performed BR profiling of cvp1 7-d-old seedlings. This allelic series has been described previously. Briefly, cvp1-1, cvp1-3, and cvp1-4 are weak, null, and intermediate alleles, respectively (Carland et al., 2002). Sterol precursors to the BR pathway are elevated in all alleles, ranging from approximately 3-to 14-fold over wild-type levels, and are consistent with the severity of the alleles (Table III). The Arabidopsis BR biosynthetic pathway has two parallel pathways, one with early C-6 oxidation (i.e. 6-oxocampestanol, cathasterone) and the second with late C6 oxidation (i.e. 6-deoxocathasterone; Clouse and Sasse, 1998). These branches converge to produce the end product, BL. Although there were minor deviations in BR levels between the wild type and cvp1 early in the pathway, there were no detectable differences in levels of biologically active BR. Low seed and tissue yield of cvp1 smt3 double mutants precluded BR profiling. Therefore, to assess the effects of elevated BR precursor levels in cvp1 smt3 double mutants, we performed supplementation experiments with the BL inhibitor, Brz (Sekimata et al., 2001). Exogenously applied Brz did not rescue cvp1 smt3 developmental abnormalities (Supplemental Fig. S4). These results are consistent with our previous studies and indicate that the cvp1 and cvp1 smt3 mutant phenotypes are not associated with changes in BR levels and suggest that other sterols may act as signaling molecules.
cvp1 Enhances the Auxin Resistance of axr1-3 A number of the phenotypic abnormalities observed in cvp1 and cvp1 smt3 mutants are reminiscent of those in auxin signaling mutants. To investigate whether some of the phenotypic defects of cvp1 are due to alterations in auxin signaling, we assayed the auxin sensitivity of cvp1. cvp1 mutant seedlings exhibited weak auxin insensitivity relative to wild-type seedlings in root inhibition studies (Fig. 7A). Although there was no evidence of auxin resistance at higher levels of 2,4-dichlorophenoxy acetic acid (2,4-D) for the cvp1 single mutant, the cvp1 auxin resistance1-3 (axr1-3) double mutants show enhanced auxin resistance (Fig. 7B). This result is in agreement with the ability of cvp1 to enhance auxin resistance of transport inhibitor resistant1 (Pan et al., 2009).
Numerous auxin signaling mutants have a simplified vascular pattern, including monopteros, axr1, axr6, bodenlos, and transport inhibitor resistant1, indicating an involvement of auxin signaling in establishing the vein pattern (Przemeck et al., 1996;Hobbie et al., 2000;Deyholos et al., 2003;Hardtke et al., 2004;Carland and Nelson, 2009). To investigate the relationship between sterols and auxin in contributing to the complexity of the vein pattern, we examined the cvp1 axr1-3 double mutant vein pattern. axr1 mutant cotyledons have a reduced reticulum that usually consists of two loops with regions of poorly differentiated vascular cells indicated by nonuniform tracheary element thickening (Fig. 7C). Although the cotyledon vein pattern of cvp1 axr1-3 was additive, sharing characteristics of both parents, the cvp1 axr1-3 seedlings produced an enhanced axr1-3 phenotype. Wild-type seedlings have an adaxially directed cotyledon cup. In cvp1 and cvp1 smt3 mutants, this cup is pronounced with raised cotyledon edges. In most cases (90%), axr1 cotyledons are similar to those of the wild type, while the remaining 10% invert the concavity of the cup, resulting in an abaxially directed cup. Eighty-five percent of cvp1 axr1-3 cotyledons have an abaxially oriented cup (Supplemental Fig. S5). Adult stages of plant growth are intermediate, indicating that the genetic relationship between cvp1 and axr1 is restricted to early stages of growth of specific organs (data not shown).

cvp1 Mutants Display Elevated Auxin Response
Although the cotyledon vascular pattern is established during embryogenesis, veins continue to develop throughout cotyledon expansion, as monitored with the vascular specific reporter gene Athb8::GUS (Baima et al., 2001;Kang and Dengler, 2002), and are possibly amenable to perturbation by the auxin transport inhibitor 1-N-naphthylphthalamic acid (NPA; Fig.  7D). Indeed, wild-type seeds germinated on NPAsupplemented medium displayed the nonuniform thickening characteristic of cvp1, indicating that compromised auxin transport results in a loss of vascular cell axialization (Fig. 7E). Similar to reports of NPAtreated leaves (Sieburth, 1999;Mattsson et al., 2003), NPA supplementation shifted and expanded regions of auxin accumulation in cotyledons, monitored with the synthetic auxin response reporter DR5::GUS (Fig.  7, F and G; Ulmasov et al., 1997). To examine auxin distribution in cvp1 mutant developing embryos, we crossed plants containing the DR5::GUS reporter construct to cvp1. In developing wild-type embryos,  DR5::GUS was visible in developing procambial strands and the hydathode of walking stick-stage embryos (Fig. 7H). At later stages, DR5::GUS staining was detectable in apical portions of basal loops in bent cotyledon-stage embryos, indicating its transient expression during vein initiation (Fig. 7J). In cvp1 walking stick and bent cotyledon embryos, DR5::GUS was much stronger and broader, extending into areole portions beyond incipient vascular cells and the hydathode (Fig. 7, I and K). DR5::GUS root tip expression was indistinguishable between the wild type and cvp1 and served as a control. This broad DR5::GUS expression correlated with the poorly axialized vascular cells of cvp1 to reflect abnormalities in cotyledon vein patterns and is consistent with a cvp1 defect in localized auxin transport.

Sterol Dependence of Auxin-Inhibited PIN2-GFP Endocytosis
PIN proteins facilitate auxin transport through trafficking between the plasma membrane and the internally localized endosome (Geldner et al., 2001). In the presence of brefeldin toxin A (BFA), PIN accumulates in endosomal aggregates known as BFA bodies. However, in response to auxin, BFA body formation is disrupted, because auxin promotes its own efflux by preventing internalization of PINs (Paciorek et al., 2005). To monitor auxin transport defects in the root, we crossed a PIN2-GFP line into cvp1 and cvp1 smt3 mutant backgrounds. Similar to results in wild-type roots, PIN2-GFP in cvp1 and cvp1 smt3 mutant roots is plasma membrane localized and responds appropriately to BFA with the formation of BFA bodies (Fig. 7, L and M; Supplemental Fig. S6, A and B). However, in contrast to wild-type roots, BFA bodies are retained in cvp1 and cvp1 smt3 mutants supplemented with auxin, demonstrating that correct sterol composition is required for auxin-inhibited PIN2-GFP endocytosis (Fig. 7,N and Q;Supplemental Fig. S6C).

Phenotypic Abnormalities Are Due to Altered Sterol Levels Independent of BR Signaling
The changes in the relative abundance of specific sterols caused by biosynthetic blocks have been associated with distinct developmental abnormalities. The SMT2 and SMT3 enzymes act at the branch point between sterol and BR synthesis. We show that in the cvp1 and smt3 mutants (in SMT2 and SMT3, respectively), the sterol content is dramatically affected without disrupting BR levels. This suggests that the vascular patterning defects observed in cvp1 mutants are due to deficiency in particular sterols, rather than in BRs. Although other sterol mutants also have distinct patterning defects, suggesting that particular sterol imbalances have specific consequences, the extreme sterol block provided by the double cvp1 smt3 mutant provides the most extensive detail thus far on the developmental processes affected by specific sterols.
Several mutants with sterol pathway blocks and consequent phenotypes have been characterized with regard to the specific sterols that accumulate. In wildtype plants, the sterol biosynthetic pathway produces a balance of key compounds, including the predominant end product sterol, sitosterol, the precursor to the BR pathway, campesterol, and the minor sterol, cholesterol. Mutations in upstream enzymes FK and HYD1 result in meristem and root patterning defects and are seedling lethal. In these mutants, overall typical sterol levels, including campesterol, decrease to 2% or less, with atypical sterols accounting for the majority of sterols (Schrick et al., 2004a). smt mutants show a more restricted phenotype, with a loss of polarity both at the vascular cell level and the organ level, characterized by rounded cotyledons, short roots, and reduced plant stature, and are able to grow to maturity, allowing an examination of sterol roles in adult development. Unlike fk, hyd1, and cpi mutants, total sterol levels in the smt mutants are approximately equivalent to wild-type levels, and there is no accumulation of abnormal sterols. Atypical sterols are not detected in smt mutants, most likely because the enzymes act at bifurcation steps and the accumulated substrate feeds into a parallel branch and is further transformed into sterol/steroid derivatives. Lastly, smt mutants are not depleted in BR precursor levels. Collectively, this argues that the less severe phenotype in smt mutants is due to reduced sitosterol levels and not to accumulated abnormal sterols or reduced BRs. However, it is also possible that elevated cholesterol or 24-ML has an impact on smt phenotypic abnormalities. The sterol profiles and phenotypes of smt1 mutants are consistent with genetic evidence indicating distinct pathways between SMT1, FK, and HYD1 (Schrick et al., 2002).
Although the significant accumulation of cholesterol, which lacks methyl additions, in smt mutants is consistent with proposed SMT function, there were some inconsistencies with the relative amounts of downstream sterol products. Because SMT1 acts at the initial step in the pathway, it is predicted that all downstream products would be completely or significantly reduced, and although some sterol intermediates are reduced, some are at equivalent or increased levels relative to the wild type. For example, the downstream campesterol branch-specific sterol, 24-methylenecholesterol, is elevated in smt1. The presence of this methylated product may be due to promiscuity of SMT2 or SMT3 that is enhanced in SMT1-deficient plants. Likewise, even with complete deficiency of C-24 ethylidene SMT function, residual levels of downstream end product sitosterol are still detected in cvp1 smt3 double mutants, albeit only at 2.5% of wild-type levels. These results suggest that alternative, not normally active, pathways are upregulated in sterol mutant backgrounds. Recently, it was shown that the lanosterol pathway, the sterol biosynthetic pathway in yeast and animals, exists in plants (Ohyama et al., 2009). In wild-type plants, it is a minor pathway, contributing to only 1.5% of total sitosterol accumulation, but it could be exploited in smt mutant backgrounds and offers an explanation for the presence of downstream sterols.

Sterol Requirements for Cell Expansion and Cell Division
Sterols are abundantly synthesized during early stages of seed development, coincident with a period of intense cell division and membrane synthesis (Herchi et al., 2009). Some of the cell expansion and division defects in fk, smt1, hyd1, and cvp1 may be the result of membrane structural defects from deficiency in specific sterols. Cell morphological defects in fk, smt1, and hyd1 were shown to be associated with cell wall gaps and aberrant cell wall thickenings with ectopic deposits of lignin and callose (Schrick et al., 2004a). The sterol biosynthetic genes are expressed in regions of cell division and expansion, and SMT transcription is strongly induced by the cell division promoter, the cytokinins (Diener et al., 2000;Jang et al., 2000;Schrick et al., 2000;Carland et al., 2002;Souter et al., 2002). In addition, sterols are able to transcriptionally activate genes associated with cell expansion and proliferation (He et al., 2003). Lastly, specific sterol deficiencies are associated with defects in cellulose synthesis and deposition (Schrick et al., 2004a). The sterol mutants smt1/ cph, fk, and hyd1 exhibit a decrease specifically in bulk cellulose, without apparent effects on other cell wall components such as pectin and sugars. This result is consistent with the postulated role of the sterol conjugate, sitosterol-b-glucoside, as a primer of cellulose synthesis by providing Glc monomers to the developing cellodextrin chain (Peng et al., 2002). Although tight association of sitosterol-b-glucoside with cellulose was observed, recent evidence suggests that this interaction may not serve to prime biosynthesis. Mutations in the enzymes that catalyze the formation of sterol glucoside conjugates, UDP-Glc:sterol glucosyltransferase (i.e. ugt80A2 and ugt80B1 mutants), do not disrupt cellulose biosynthesis, nor does the single or double mutant combination display decreased cellulose levels relative to the wild type (DeBolt et al., 2009). Thus, if there is a sitosterol role in priming cellulose synthesis, there must be alternative enzymes, such as glucosylceramide synthase, which has the ability to produce glycosylated conjugates of sterols in vitro (Hillig et al., 2003). Alternatively, residual sterol glucosyltransferase in the ugt80A2 ugt80B1 double mutants may be enough to facilitate the priming reaction.
The severe sterol block in the cvp1 smt3 double mutant is associated with aberrant floral development. Although fk and hyd1 mutants are seedling lethal, precluding an examination of sterol-dependent floral development, null smt1 plants produce flowers and seeds. No floral defects are obvious in the smt1-1 allele used in this study; however, smt1 orc displays ruffled sepals and mild gynoecia fusion similar to cvp1 (Willemsen et al., 2003). smt1 orc is in the Utrecht background, and like other sterol mutants, it displays an ecotypic enhancement, suggesting that additional factors may be involved. Ectopic endoreduplication is associated with cvp1 petal and sepal tip cell enlargement, suggesting that sterol-mediated cell cycle regulation may be responsible (Hase et al., 2005). Cholesterol-dependent cell cycle progression through Hh signal transduction has been reported (Incardona and Eaton, 2000). In this case, cholesterol covalently binds to the ligand, Hh. Modified Hh subsequently interacts with its partner, Ptc, and activates transcription of the primary regulators of the cell cycle to promote cell growth. If depleted of cholesterol, the cells were arrested in the G2 phase of mitosis but resumed phase progression with the addition of cholesterol or a cell cycle protein. In insects, a rough eye phenotype, consisting of enlarged eye cells due to abnormal cell expansion and strikingly reminiscent of cvp1 smt3 petal cells, is a direct consequence of defective Hh signaling (Duman-Scheel et al., 2002). Therefore, it is conceivable that the cell expansion defects in the smt mutants may be due to a sterol ligand and its binding partner that acts in cell cycle regulation analogous to Hh/Ptc.
Mature ethylating SMT-depleted plants also induced misspecification of floral organ identity and closely resembled strong ap2 mutants, characterized by a loss of petals and transformation of sepals to carpels (Bowman et al., 1991). AP2 and AP2-like transcription factors are regulated by microRNA172 (MiR172; Aukerman and Sakai, 2003). Overexpression of MiR172 induced stigmatic papillae on cauline leaf margins, closely resembling cvp1 smt3 double mutant flowers. The means by which sterols influence floral patterning are unknown, but the observations suggest that sterols have roles in floral development and possibly share common pathways/targets with micro-RNA-regulated transcription factors. Another connection between sterols and microRNA-regulated gene expression is discussed below in relation to START domain proteins.

Sterols as Signaling Molecules
The phytosterol biosynthetic pathway leads to the production of the BRs, the only known steroid signaling molecules in plants (Belkhadir and Chory, 2006). Sterols themselves may act as signaling molecules in plants in a manner analogous to the action of cholesterol in mammalian systems. In these systems, the candidate target-binding partners for cholesterol are the steroidogenic acute regulatory protein-related lipid transfer (START) domain-containing proteins (Ponting and Aravind, 1999). These classic START domain proteins were reassigned as members of the helix-grip fold superfamily due to structural similarity in the ligand-binding pocket (Iyer et al., 2001). Mem-bers of this superfamily have been demonstrated to bind a diverse range of molecules, including polyketide antibiotics, antigens, and RNA for both catalytic and noncatalytic activities. Plants contain numerous START domain proteins. One class of plant START protein receptors belonging to the helix-grip superfamily has been shown to bind the phytohormone abscisic acid via the START domain Ma et al., 2009;Melcher et al., 2009;Miyazono et al., 2009;Park et al., 2009;Santiago et al., 2009;Sheard and Zheng, 2009;Yin et al., 2009;Szostkiewicz et al., 2010). Other plant START domain proteins are homeodomain-Leu zipper (HD ZIP) transcription factors that serve to regulate plant morphogenesis, including meristem and vascular development, and thus are likely targets of sterol ligands (Schrick et al., 2004b). GLA-BROUS2 (GL2), a regulator of trichome and root hair development, acts to negatively regulate phospholipid signaling, thereby possibly integrating GL2 START domain ligand binding with downstream lipid signaling (Ohashi et al., 2003).
Of the approximately 90 HD transcription factor family members, 21 include START domains in an arrangement that only occurs in plants (Schrick et al., 2004b). This tight linkage may suggest a unique START domain transcriptional activation role in plants that relies on sterol ligands. Given the vein patterning defects in smt mutants, the vascularly expressed class III START domain HD ZIP subfamily (HD ZIP III) represents candidate SMT-mediated sterol targets . However, these HD ZIP proteins, including PHABULOSA (PHB), PHAVOLUTA (PHV), REVOLUTA, Athb15/CORONA (CRN), and Athb8, are microRNA regulated within an evolutionally conserved site in the START domain, precluding an examination of sterol/START binding by mutational analysis (McConnell et al., 2001;Otsuga et al., 2001;Green et al., 2005;Kim et al., 2005;Melcher et al., 2009). Multiple loss-of-function mutant combinations reveal redundant, distinct, and antagonistic functions among HD ZIP III family members . Interestingly, phb phv cna triple mutant flowers display reproductive organs within the fifth whorl, similar to cvp1 smt3 mutant flowers. In addition, loss-of-function mutations in Athb-8 in a background of reduced MONOPTEROS function revealed a role for Athb-8 in vein patterning (Donner et al., 2009). Plant START domains are also coupled with other modules, including the phosphoinositide-binding pleckstrin homology domain, indicating a function for lipid control of intracellular signaling pathways (Schrick et al., 2004b). A specific function for phosphoinositide signaling in vein patterning involving CVP2 and its phosphoinositide target, the pleckstrin homology domain-containing VASCULAR NET-WORK3/SCARFACE, was recently reported (Carland and Nelson, 2009). Collectively, these results support a possible role for sterol regulation of START domain proteins in influencing patterning events during plant morphogenesis.

Sterols, Membrane Domains, and PIN Proteins
Sterols are integral membrane components that serve not only a structural role but also a regulatory role by forming small membrane domains enriched in specific lipids that have an affinity for selected proteins (Incardona and Eaton, 2000). This confinement of selected proteins to specific small microdomains or rafts establishes a polarity to the cell. Although there is no direct evidence of the existence of microdomains in plants, due to technical difficulties in monitoring sterol distribution, there is circumstantial evidence accumulating with the identification of proteins from detergent-resistant membrane fractions (Boutte and Grebe, 2009). The PIN auxin efflux carrier family members have been proposed as likely candidates for sterolenriched microdomain-localized proteins (Carland et al., 2002;Betts and Moore, 2003). Consistent with a function in directional transport of auxin, mutations that disrupt PIN activity or PIN asymmetric localization cause embryo, root, and vein polarity defects (Robert and Friml, 2009). Indeed, mislocalization of specific PIN family members has been shown in smt orc and hyd1 roots (Souter et al., 2002;Willemsen et al., 2003). In this report, we provide evidence that steroldependent endocytosis is required for efficient auxin transport. Both auxin signaling and correct sterol composition are required for auxin-inhibited endocytosis, as BFA bodies are retained in auxin signaling and sterol mutants, including smt mutants, in the presence of auxin (Pan et al., 2009).
The role of sterols in establishing cell polarity via PIN-mediated processes has been examined more closely with the characterization of CPI, which encodes the sterol biosynthetic enzyme CYCLOPRO-PYLESTEROL ISOMERASE1. cpi mutants are sterile, dwarf plants with small, compact rosettes and short roots and thus share characteristics with upstream smt1 mutants and downstream br mutants (Men et al., 2008). CPI localized to the cell plate-associated endoplasmic reticulum and supports endoplasmic reticulum as a site of sterol synthesis, as biochemical studies have suggested. Newly synthesized sterols are subsequently transported to the plasma membrane via the Golgi apparatus . Upon BFA treatment, plasma membrane sterols are internalized to early endosomes, colocalizing with PIN2. Similar to other sterol mutants, PIN2 was mislocalized in cpi mutants (Men et al., 2008). A detailed analysis indicated that PIN2 failed to redistribute asymmetrically at the end of cytokinesis in the endocytosis-defective cpi mutants, indicating a sterol dependence for establishing PIN asymmetry. Thus, the observed vein polarity defects in smt mutants may be a direct consequence of a disruption in sterol-mediated PIN endocytosis. Failure to transport auxin within veins is consistent with the ectopic DR5::GUS expression observed in cvp1 and smt orc mutants and with the cvp1 enhanced auxin resistance of axr1 and thus is likely responsible for the smt vein patterning defects. It will be interesting to examine the sterol dependence of additional PIN family members, particularly PIN5, which has been shown to reside in the endoplasmic reticulum, the site of sterol biosynthesis (Mravec et al., 2009). In summary, studies with cvp1 and additional sterol mutants indicate that sterols have multiple functions independent of BR signaling molecules. Future work will be to further elaborate on the mechanisms of SMT-derived sterols.

Microscopy
Vein patterns were imaged as described previously (Carland et al., 2002). For quantitation of cotyledon size and vein patterns, cleared, mounted specimens from 7-d-old seedlings were captured using Axiovision software on a Zeiss Axiophot microscope and measured with ImageJ software. Plant images were captured on a Zeiss dissecting microscope using Axiovision software. Samples were prepared for scanning electron microscopy by fixation in ethanol:acetic acid (3:1) overnight at 4°C and dehydrated in an ethanol series to 100% ethanol. The remaining preparation and microscopy were performed as reported (Carland and McHale, 1996). For DAPI staining, flowers of similar developmental stages were cleared in 70% ethanol for 2 h at room temperature. Samples were subsequently incubated in 1 mM DAPI in 70% ethanol for 1 h, rinsed three times with 0.53 MS, mounted in 0.53 MS/ 10% glycerol, and viewed under fluorescence on a light compound microscope (Zeiss Axiophot). GUS staining of embryos has been described in a previous report (Carland and Nelson, 2004). For PIN2-GFP studies, roots of 9-d-old seedlings were treated with 50 mM BFA (Sigma) for 90 min or a 30-min pretreatment of 10 mM naphthalene-1-acetic acid (NAA; Sigma) followed by incubation with 10 mM NAA and 50 mM BFA for 90 min (Pan et al., 2009). PIN2-GFP imaging was conducted on a Zeiss LSM 510 Meta NLO confocal microscope (Carl Zeiss International).

Constructs and Plant Transformation
The SMT3 construct for complementation of the cvp1 smt3 double mutant was generated by PCR amplification of the transcriptional unit (approximately 2.8 kb, containing a 1.5-kb promoter, an approximately 1-kb coding region, and a 0.3-kb 3# untranslated region) of SMT3 from Columbia using the primers SMT3-2 (5#-CGTGTGAGCAAATAGATCACG-3#) and SMT3-9 (5#-CTAATTTCCAGTGATCCATC-3#). The amplified product was cloned into pCR2.1 TOPO vector (Invitrogen), sequenced to ensure that there were no base pair changes, and subsequently subcloned into pCAMBIA 2300 using compatible restriction sites. The construct was electroporated into Agrobacterium tumefaciens strain GV3101 and transformed into smt3 plants using the floral dip method (Clough and Bent, 1998). Several transgenic plants were selected on medium with 50 mg mL 21 kanamycin and crossed to cvp1-3 mutants. Kanamycin-resistant F1 progeny from independent transformants were self-pollinated, and the resulting kanamycin-resistant F2 progeny were scored for cotyledon vein pattern, root, and petal defects. There was no evidence of the cvp1 smt3 double mutant phenotype among 401 kanamycin-resistant progeny. A portion of kanamycin-resistant cvp1 plants were genotyped as cvp1 smt3.

Measurement of BR and Sterol Levels
One gram and 20 g of 7-d-old lyophilized seedling tissue were used for sterol and BR profiling, respectively. Purification and quantification of sterols and BRs were carried out as described previously He et al., 2003).
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers NM_101884 (SMT2) and NM_106258 (SMT3).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S2. cvp1 smt3 double mutants share characteristics with smt1 mutants in fertility and seedling morphology.
Supplemental Figure S3. Sterol supplementation does not rescue or mimic smt defects.
Supplemental Figure S4. The BL inhibitor Brz does not rescue cvp1 smt3.