The RON1/FRY1/SAL1 Gene Is Required for Leaf Morphogenesis and Venation Patterning in Arabidopsis 1[W][OA]

To identify genes involved in vascular patterning in Arabidopsis ( Arabidopsis thaliana ), we screened for abnormal venation patterns in a large collection of leaf shape mutants isolated in our laboratory. The rotunda1-1 ( ron1-1 ) mutant, initially isolated because of its rounded leaves, exhibited an open venation pattern, which resulted from an increased number of free-ending veins. We positionally cloned the RON1 gene and found it to be identical to FRY1/SAL1 , which encodes an enzyme with inositol polyphosphate 1-phosphatase and 3 # (2 # ),5 # -bisphosphate nucleotidase activities and has not, to our knowledge, previously been related to venation patterning. The ron1-1 mutant and mutants affected in auxin homeostasis share perturbations in venation patterning, lateral root formation, root hair length, shoot branching, and apical dominance. These similarities prompted us to monitor the auxin response using a DR5-GUS auxin-responsive reporter transgene, the expression levels of which were increased in roots and reduced in leaves in the ron1-1 background. To gain insight into the function of RON1/FRY1/SAL1 during vascular development, we generated double mutants for genes involved in vein patterning and found that ron1 synergistically interacts with auxin resistant1 and hemivenata-1 but not with cotyledon vascular pattern1 ( cvp1 ) and cvp2 . These results suggest a role for inositol metabolism in the regulation of auxin responses. Microarray analysis of gene expression revealed that several hundred genes are misexpressed in ron1-1 , which may explain the pleiotropic phenotype of this mutant. Metabolomic proﬁling of the ron1-1 mutant revealed changes in the levels of 38 metabolites, including myoinositol and indole-3-acetonitrile, a precursor of auxin.

During the vegetative development of Arabidopsis (Arabidopsis thaliana), leaves are produced from the shoot apical meristem in an orchestrated program that involves patterning and cell division, expansion, and differentiation. The mature vegetative leaves of Arabidopsis are histologically simple and consist of the outer epidermis and internal mesophyll and vasculature (Tsukaya, 2005). Veins are crucial for normal leaf function, transporting water, minerals, and photosynthate and providing mechanical support to the lamina (Evert and Eichhorn, 2006). The leaves of many vascular plants, such as the angiosperms, exhibit a closed reticulate venation pattern (Roth-Nebelsick et al., 2001). In Arabidopsis, the leaf venation pattern is brochidodromous, with a single primary vein (midvein) and a series of loops formed by secondary veins that connect other secondary and higher order veins (Hickey, 1973;Candela et al., 1999).
Vein differentiation must be spatially and temporally regulated throughout leaf development. Many aspects of venation patterning in plant leaves can be explained by the auxin canalization model (Sachs, 1991;Rolland-Lagan and Prusinkiewicz, 2005), which is supported by considerable experimental evidence. The role of auxin in venation pattern formation is supported by the phenotypes of mutants possessing altered auxin biosynthesis or perception (Alonso-Peral et al., 2006;Cheng et al., 2006), experimental perturbation of auxin transport (Mattsson et al., 1999;Sieburth, 1999), and the expression pattern of auxinresponsive reporter transgenes (Mattsson et al., 2003;Scarpella et al., 2006). The phenotypes of mutants impaired in auxin transport, such as scarface (sfc; Deyholos et al., 2000;Sieburth et al., 2006) and pin-formed1 (pin1; Okada et al., 1991;Gälweiler et al., 1998), and perception, such as monopteros (mp; Hardtke and Berleth,1998), are pleiotropic and include defects in vein patterning or differentiation. The sfc mutant exhibits a disconnected venation pattern (Deyholos et al., 2000), and the lateral organs of strong mp mutants display a reduced venation pattern with no peripheral veins (Przemeck et al., 1996). In contrast, the leaf venation pattern of pin1 mutants resembles that of wild-type plants treated with auxin transport inhibitors, exhibiting extra primary and secondary veins and an accumulation of vascular elements along the leaf margin (Mattsson et al., 1999).
Unlike sfc, pin1, or mp, other leaf venation mutants are not primarily affected in auxin production, perception, or transport (Carland et al., 1999). Examples include cotyledon vascular pattern1 (cvp1), the cotyledons of which exhibit isolated patches of vascular tissue (Carland et al., 1999(Carland et al., , 2002, and cvp2, which exhibits increased numbers of free-ending veins in the cotyledons and leaves (Carland et al., 1999;Carland and Nelson, 2004). CVP1 encodes the STEROL METHYLTRANSFERASE2 (SMT2) protein, an enzyme that functions in the sterol biosynthetic pathway (Carland et al., 2002). CVP2 encodes an inositol polyphosphate 5#-phosphatase (5PTase; Carland and Nelson, 2004), which mediates the hydrolysis of inositol 1,4,5-trisphosphate (IP 3 ), a eukaryotic second messenger with a pivotal role in calcium signaling (Berridge, 2009). IP 3 controls cytosolic calcium levels by regulating calcium release from the vacuole and endoplasmic reticulum (Krinke et al., 2007). The disconnected, open venation pattern of cvp2 cotyledons and leaves suggested a role for intracellular IP 3 levels in vascular development (Carland and Nelson, 2004). Recently, CVP2 and another 5PTase, CVP2-LIKE1 (CVL1), have been shown to regulate vein patterning through the production of a specific phosphoinositide (PI) that acts as a ligand for SFC/VASCULAR NETWORK3 (VAN3), which in turn controls the traffic of vesicles that accounts for the polar subcellular localization of PIN1 proteins (Carland and Nelson, 2009;Naramoto et al., 2009). Another inositol 5PTase, At5PTase13, has been shown to play a role in auxinmediated vein development in cotyledons (Lin et al., 2005). Furthermore, the open vein networks present in the leaves of forked and tornado mutants (Steynen and Schultz, 2003;Cnops et al., 2006) may be due to altered auxin perception or distribution.
To identify genes required for venation patterning, we screened for naturally occurring variations in the venation pattern of Arabidopsis vegetative leaves (Candela et al., 1999). In this way, we discovered the spontaneously occurring hemivenata-1 (hve-1) mutation, which causes a venation pattern that is significantly simpler than those of other wild types, such as Landsberg erecta (Ler) and Columbia-0 (Col-0). We positionally cloned the HVE gene, which encodes a CAND1 protein involved in ubiquitin-mediated auxin signaling (Alonso-Peral et al., 2006). To identify additional loci necessary for vascular patterning, we screened for venation pattern defects in a collection of leaf shape mutants isolated in our laboratory after ethyl methanesulfonate (EMS) mutagenesis (Berná et al., 1999) and found that the rotunda1-1 (ron1-1) mutant, named after the round laminae of its vegetative leaves, displays disconnected leaf veins. Here, we describe the phenotypic characterization of the ron1-1 mutant and the map-based cloning of RON1, which encodes an inositol polyphosphate 1-phosphatase that plays a role in venation patterning, as determined by morphological, reporter gene, and double mutant analyses. Our results suggest an interplay between inositol and auxin signaling in a number of developmental pathways, including those responsible for leaf venation pattern formation.

Effects of the ron1-1 Mutation on Leaf Development
To gain insight into the developmental basis of the phenotype caused by the ron1-1 recessive mutation ( Fig. 1), we studied cotyledons and first-and third- Figure 1. The visible vegetative phenotype of ron1-1. A, Basal rosettes of wild-type Ler and ron1-1 mutant plants. B, Cotyledons and leaves excised from Ler and ron1-1 rosettes. The two cotyledons are shown on the left, followed by the vegetative leaves in the order that they initiated. A number of leaves are not fully expanded, given that photographs were taken at 21 DAS. Bars = 5 mm. node vegetative leaves 22 d after stratification (DAS), which were found to be shorter and wider in the mutant than in the wild type (Table I). These changes translated into length-width ratios close to 1 for all organs examined in ron1-1, ranging from 0.95 6 0.05 in third leaves to 1.11 6 0.16 in cotyledons. In the Ler wild type, the first leaves were the most rounded, with a length-width ratio of 1.17 6 0.07. We measured the area of individual palisade cells in the first leaves, which were 42.05% larger in ron1-1 (1,617.10 6 698.81 mm 2 , n = 337) than in Ler (1,138.41 6 617.95 mm 2 , n = 473; P = 0), suggesting that the RON1 gene restricts cell expansion.
We next characterized the venation pattern of cotyledons and leaves (Fig. 2) using two parameters that quantitatively describe venation pattern complexity: venation density, expressed as the ratio of vein length to leaf area (mm mm 22 ), and the number of branching points per unit leaf area (Candela et al., 1999). Both parameters demonstrated significantly reduced values in ron1-1 first leaves but not in cotyledons and third leaves ( Fig. 2; Table II). In addition, leaves exhibited more free-ending veins per unit venation length in ron1-1 than in Ler (Table II), indicating altered vein connectivity. Vein disconnections are obvious at the leaf margins and in the apical zone of cleared ron1-1 leaves. In wild-type leaves, the midvein and two secondary veins usually merged at the leaf apex to form two closed loops (areoles) and an apical hydathode. In contrast, in ron1-1 leaves, these loops were often incompletely closed (Fig. 3, A and B).
To trace the defects of ron1-1 mutant leaves to earlier stages, we used an ATHB-8-GUS transgene as a reporter of the procambial cells that will differentiate into veins (Baima et al., 1995). The ATHB-8-GUS reporter revealed a closed reticulate expression pattern that began during early stages of wild-type leaf development (Fig. 3C). In ron1-1 leaves, ATHB-8-GUS revealed an open network with supernumerary freeending veins, suggesting that fewer cells had been recruited to a vascular fate, particularly near the leaf margins (Fig. 3D). This defect was observed in developing first leaves, which had the simplest venation patterns, as well as in leaves produced later in development (Fig. 3, G and H). The closed ATHB-8-GUS expression patterns observed early in wild-type leaf primordia correlated with a closed venation pattern in cleared mature leaves (Fig. 3E), and the aberrant ATHB-8-GUS expression patterns observed in young ron1-1 leaf primordia correlated with partially open venation patterns and with the disconnected secondary veins near the leaf margins in mature leaves (Fig. 3F).
The expression pattern of RON1/FRY1 has been previously determined by northern-blot and promoter-GUS fusion analyses (Xiong et al., 2001). RON1/FRY1 is expressed in the primary root and root hairs as well as the leaves, stems, and flowers. The highest expression signal detected in the leaves occurs in the veins. We analyzed RON1 expression in silico using Genevestigator (Zimmermann et al., 2005) and found the gene to be expressed almost ubiquitously, except in the stigmata and in pollen, and throughout development, with the highest levels observed in young seedlings  (1-4 DAS) and during the flowering stages (21-25 DAS). Using Mutant Surveyor in Genevestigator, we found that RON1 is overexpressed (1.3-to 2.7-fold change) in the continuous vascular ring1 mutant, which is characterized by increased vascular tissue development in the stem (Parker et al., 2003). The G-to-A substitution in ron1-1 affects the last nucleotide of the fourth intron in the At5g63980 gene, which damages the splice acceptor site. This suggests that intron splicing is perturbed. We performed reverse transcription (RT)-PCR using primers that anneal to the first and fifth exons of At5g63980 and detected a single band for Ler but two bands for ron1-1, one of which was slightly smaller and the other larger than that of the wild type (data not shown). Nine nucleotides were absent from the smaller of these mutant transcripts as a consequence of the presence of a cryptic splice site located nine nucleotides downstream of the boundary between the fourth and fifth exons. The absence of these nine nucleotides is predicted to delete three conserved amino acid residues (FLR) present in an a-helix adjacent to the WVLDPIDGT motif (York et al., 1995), which is highly conserved in inositol polyphosphate 1-phosphatases. Intron 4 (119 bp) remained unspliced in the large tran-script, which encodes a truncated protein in which the active and the substrate-binding sites of the conserved inositol monophosphatase 3#-phosphoadenosine 5#-phosphate (PAP) phosphatase domain are affected. Because normal transcripts were not found, we assumed that the ron1-1 allele is either hypomorphic or null.
We identified three lines (SALK_020882, SALK_ 079259, and SALK_005741) carrying T-DNA insertions in At5g63980 (http://atidb.org/cgi-perl/gbrowse/ atibrowse/), but phenotypically mutant plants were found only among the T4 progeny of SALK_020882. Mutants also appeared in the F1 progeny of a complementation cross between SALK_020882 (in the Col-0 background) and ron1-1 (Ler), confirming that they carry allelic mutations. We PCR amplified and sequenced the genomic region adjacent to the left T-DNA border in SALK_020882 and determined that the T-DNA is inserted immediately upstream of the last nucleotide of exon 3 (Fig. 5). This allele, which had been designated ron1-2 by us and fry1-6 by previous authors (Gy et al., 2007), is predicted to yield a truncated protein with 135 amino acids, the first 125 of which are shared with the wild-type RON1 protein. The absence of most of the PAP phosphatase domain in this protein suggests that ron1-2 is a null allele. The ron1-2 mutant showed rounded leaves (Supplemental Fig. S1) and venation pattern defects even stronger than those of ron1-1, at least in the first leaf (Table III;  We examined 16 T-DNA insertion mutants corresponding to four other Arabidopsis genes that encode  inositol polyphosphate 1-phosphatases (see "Materials and Methods"): SAL2 (At5g64000), SAL3 (At5g63990), SAL4 (At5g09290), and AHL (At5g54390). SAL2 and SAL3 are arranged in tandem to FRY1/RON1 (Xiong et al., 2001). No visible phenotypes were detected in the T3 and T4 generations of these insertion lines, possibly due to functional redundancy between members of the same gene family. In contrast, the unique phenotypes observed in ron1 mutants demonstrate that RON1 performs at least some discrete, nonoverlapping functions.

Effects of ron1 Mutations on Auxin Homeostasis
Venation patterning, lateral root formation, root hair length, shoot branching, and apical dominance, all of which are affected in ron1 mutants, are processes known to be regulated by auxin (Pitts et al., 1998;Alonso-Peral et al., 2006;De Smet and Jü rgens, 2007;Ongaro and Leyser, 2007). To ascertain whether the ron1-1 mutation affects auxin perception, we monitored the expression pattern of the synthetic auxin reporter DR5-GUS (Ulmasov et al., 1997) in different tissues. Consistent with previous results (Sabatini et al., 1999), GUS signal was detected in the columella initial cells, the quiescent center, and the columella root cap of wild-type roots (Fig. 6A). In ron1-1, the GUS-staining area was broader in the root tip and often extended to the root vasculature ( Fig. 6G), indicating a disrupted pattern of auxin perception or accumulation in roots. In wild-type first leaf primordia, strong DR5-GUS expression was detected at the position of the apical hydathode and in the differentiating vascular cells, which were visible as a network ( Fig. 6B; Mattsson et al., 2003). In contrast, in ron1-1 leaf primordia, the signal was consistently reduced compared with the wild type and confined to the apical region (Fig. 6H). As the leaf developed and tracheary elements differentiated distally in wild-type plants, DR5-GUS expression persisted in the apex and appeared as a network only in the proximal portion of the leaf (Fig. 6, C and D). At this stage, however, DR5-GUS expression was weak and fuzzy at the basal portion of ron1-1 leaves and persisted strongly only in the leaf apex (Fig. 6, I and J). Later, close to the end of first leaf expansion, DR5-GUS expression appeared along the leaf margin, with maximum levels observed at the leaf apex and the apical and lateral hydathodes in wild-type leaves and, to a much lesser extent, in ron1-1 leaves (Fig. 6, E and K). A similar decrease in DR5-GUS expression levels was observed in leaves from other nodes (Fig. 6, F and L).
To gain insight into the role of RON1 in auxin homeostasis, we grew mutant and wild-type plants in vertically oriented petri dishes in the presence of different concentrations of the synthetic auxin 2,4dichlorophenoxyacetic acid (2,4-D) or the auxin efflux inhibitors N-1-naphthylphthalamic acid (NPA) and 2,3,5-triiodobenzoic acid (TIBA). The effects of 2,4-D, NPA, and TIBA were similar in ron1-1 and Ler. Growth of ron1-1 in solid medium supplemented with 0.1 mM 2,4-D failed to rescue the aberrations in the leaf venation pattern (data not shown). We also tested if auxin responses were reduced in ron1-1 by applying exogenous auxin (1 mM indole-3-acetic acid [IAA] or 1 mM 2,4-D). ron1-1 DR5-GUS plants were grown in nonsupplemented solid medium for 4 weeks and then incubated for 6 h in liquid medium supplemented with the hormone. We observed a reduced response of the DR5-GUS reporter in the leaves of ron1-1 DR5-GUS plants compared with the response of a nonmutant control (Supplemental Fig. S2).
Phenotypes of the ron1 cvp1 and ron1 cvp2 Double Mutants CVP2 encodes an inositol polyphosphate 5#-phosphatase, and its loss-of-function mutations determine Table III. Morphometric analysis of the venation pattern of first-node leaves in double mutants All values are means 6 SD. In italics, values significantly different from those of Col-0 (P # 0.05); in boldface, values significantly different from those of ron1-2 (P # 0.05); marked by asterisk, values of the double mutants significantly different from those of Col-0 and both single mutants (P # 0.05). an open venation pattern in leaves and cotyledons due to premature vein termination, which affects higher order veins more severely than lower order veins (Carland and Nelson, 2004). Similar vein phenotypes are caused by mutations in CVP1, a gene that encodes the sterol methyltransferase SMT2, although only in cotyledons, sepals, and petals (Carland et al., 2002). Whole organ morphology is not visibly altered in the leaves of cvp1 and cvp2 mutants, which nevertheless exhibit an increased number of free-ending veins and reduced venation complexity, as observed for ron1-1 (Fig. 7).
The ron1-2 Mutation Synergistically Interacts with hve and axr1 The pleiotropic phenotype of ron1-1 and ron1-2 and the reduced DR5-GUS expression pattern observed in ron1-1 leaves seemed compatible with a mild auxin perception defect. These observations prompted us to analyze the interactions between ron1 and other mutants with impaired auxin perception and abnormal venation patterns, such as auxin resistant1 (axr1; Deyholos et al., 2003;Alonso-Peral et al., 2006) and hve (Candela et al., 1999;Alonso-Peral et al., 2006;Fig. 7). We obtained double mutant combinations of the ron1-2, axr1-12, and hve-2 mutations in the Col-0 background. The ron1-2 hve-2 and ron1-2 axr1-12 rosettes were small, particularly in ron1-2 axr1-12 (Supplemental Fig. S3). The venation pattern defects were also greatly enhanced in the leaves of these double mutants (Fig. 7), especially in ron1-2 axr1-12. In this strain, all the morphometric descriptors but one were significantly different from both single mutants and Col-0, suggesting a possible synergistic interaction (Pérez-Pérez et al., 2009). In these double mutants, the venation of first-node leaves was often completely open ( Fig. 7) and reduced to a single primary vein with a few secondary veins. In agreement with this reduction in venation pattern complexity, the values obtained for venation density and branching points per mm 2 in the first-node leaves of these double mutants were similar to, or even lower than, those of Ler or ron1-1 cotyledons (Tables I and III). No closed areoles were observed in 31 of 32 first-node leaves of ron1-2 axr1-12 plants and five of 20 first-node leaves of ron1-2 hve-2 plants. The veins of ron1-2 axr1-12 leaves included parallel vascular strands that often failed to contact each other (data not shown). In addition, the leaves of both double mutants were significantly smaller (Fig. 7) and deeply lobed later in development (Supplemental Fig. S4). The enhanced vein defects and the pronounced lobing of double mutant leaves suggest an interaction between RON1, HVE, and AXR1.
Global Expression Profiling of the ron1-1 Mutant A transcriptomic analysis of ron1-1 shoot apices was carried out using ATH1 microarrays, which demon-strated 663 genes that were differentially expressed at P , 0.01, 392 and 271 of which were up-regulated and down-regulated, respectively (Supplemental Table S3). To validate these results, quantitative (q) RT-PCR analysis of 14 genes selected from among the up-regulated ones was performed. Twelve of these 14 genes were validated by qRT-PCR, confirming the microarray results (Supplemental Tables S4 and S5).
BiNGO (Biological Networks Gene Ontology; Maere et al., 2005) analysis showed that some up-regulated genes participate in developmental processes, such as the determination of symmetry, adaxial/abaxial patterning, primary shoot apical meristem specification, meristem initiation, and root cap development ( Fig.  8A; Supplemental Table S6). Other up-regulated genes participate in physiological processes such as amino acid and derivative metabolism, flavonoid synthesis, nitrate assimilation, and basic amino acid transport. This analysis also showed that genes involved in responding to some stimuli (ethylene and metal ion) and to microRNA (miRNA)-mediated gene silencing are overexpressed in ron1-1. Indeed, changes in the expression of 24 genes that are known or predicted miRNA targets were detected (Supplemental Table  S7). All of them were up-regulated in ron1-1, suggesting a defect in miRNA-mediated gene silencing that might be related to the proposed role of RON1/FRY1 as an endogenous RNA-silencing suppressor (Gy et al., 2007). On the other hand, genes related to salinity and cold responses, organic acid biosynthesis, starch catabolism, fatty acid biosynthesis, phagocytosis, and chromatin assembly or disassembly were down-regulated in ron1-1 ( Fig. 8B; Supplemental Table S6).
To ascertain whether hormone-activated developmental pathways are affected in ron1-1, we compared our ron1-1 microarray results with those of Nemhauser et al. (2006;Supplemental Table S8). These authors searched for transcriptional targets of seven plant hormones: abscisic acid (ABA), methyl jasmonate (MJ), auxin, brassinolide (BL), the ethylene precursor 1-aminocyclopropane-1-carboxylic acid, the cytokinin zeatin, and the gibberellin GA 3 . The set of genes differentially expressed in the ron1-1 microarray made up an overrepresented sample of genes that are upregulated in the presence of ABA, which is in line with the role of FRY1/RON1 in the regulation of ABA responses (Xiong et al., 2001). In addition, a significant number of genes whose expression is deregulated by MJ were up-regulated in ron1-1. Indeed, RON1 itself was found to be specifically regulated by MJ and is considered a marker for this hormone (Nemhauser et al., 2006). To determine if ron1-1 is affected in the MJ response, we assayed the effects of MJ on the growth of wild-type and ron1-1 roots. Root growth was slightly more inhibited by 5, 10, and 50 mM MJ in ron1-1 than in wild-type plants (data not shown). The ron1-1 microarray data set was also enriched in BL target genes but exhibited expression changes of opposite sign: genes up-regulated by BL were down-regulated in the ron1-1 mutant, and genes down-regulated by BL were up-regulated in the ron1-1 mutant. We studied the effect of BL (25, 50, 100, 300, and 500 mM 24-epibrassinolide) on the growth of ron1-1 roots, but no effects were detected (data not shown).
The 266 nuclear genes down-regulated in ron1-1 were analyzed using Athena (O'Connor et al., 2005), and the nine-nucleotide AAAATATCT "evening element" (Harmer et al., 2000) was found to be significantly overrepresented in 39 of them (P , 10 29 ; the P value threshold for significant enrichment is ,10 24 ; Supplemental Table S9). The set of genes downregulated in ron1-1 was also significantly enriched in nuclear genes encoding chloroplast-targeted proteins (96 of 266 genes; P , 10 210 ; the P value threshold for significant enrichment in Gene Ontology terms is ,10 26 ; Supplemental Table S10), 15 of which contained the evening element in their promoters. It is known that the evening element is essential for circadian rhythmicity and that nuclear genes encoding plastid proteins are coregulated by the circadian clock, which helps plants become more efficient in using solar energy (Harmer et al., 2000).

Metabolic Profiling of the ron1-1 Mutant
To characterize the metabolic changes that are caused by the ron1-1 mutation, the metabolomes of ron1-1 and Ler were compared using gas chromatography-mass spectrometry (GC-MS). Over 80 metabolites were detected, but only those showing significant changes (i.e. false discovery rate , 0.05 and absolute fold change higher than 1.2) were used for subsequent comparison. The levels of 21 metabolites were significantly increased in ron1-1, whereas those of 17 metabolites were decreased when compared with Ler (Supplemental Table S11; Supplemental Fig. S5). These included sugars, amino acids, non-tricarboxylic acid cycle organic acids, and other unknown compounds. The levels of myoinositol, an intermediate of inositol metabolism, were significantly changed in ron1-1 (1.3-fold), which may be related to altered inositol metabolism. Although the biosynthesis of auxin can be dependent on Trp, the levels of this amino acid did not change in ron1-1. However, appreciable levels of the related compound indole-3-acetonitrile, a precursor of IAA in one of the Trp-dependent auxin biosynthesis pathways, were up-regulated in the mutant (2.6-fold). The induction of myoinositol and indole-3-acetonitrile may suggest that RON1 is an interplayer between inositol and auxin signaling. Variation in the levels of several amino acids was detected in the ron1-1 mutant.
Decreased levels were consistently detected in ron1-1 for several sugars, including Fru, Glc, sorbose, and altrose. Similar changes in sugar metabolism had been recently observed in the fry1-1 mutant, suggesting a role for RON1 in transient starch accumulation and osmotic stress protection (Wilson et al., 2009). We detected in addition higher levels of trehalose and precursors of polyamine biosynthesis in ron1-1 leaves.

DISCUSSION
In a screen for EMS-induced leaf shape mutants in Arabidopsis (Berná et al., 1999), we identified the recessive ron1-1 mutation, which causes vegetative leaves to be rounded and increases leaf palisade mesophyll cell size. Mutants with fewer but enlarged cells, such as ron1-1, might provide insight into the intriguing phenomenon named compensation (Tsukaya, 2005). In addition, the ron1-1 mutant exhibited an open venation pattern, which resulted from an increased number of free-ending veins. The pleiotropic phenotypes of ron1-1 included late flowering, long root hairs, a reduced number of lateral roots, reduced inflorescence size, and increased inflorescence branching. Positional cloning demonstrated that RON1 is identical to SAL1 (Quintero et al., 1996), FRY1 (Xiong et al., 2001), HOS2 , and ALX8 (Wilson et al., 2009), which encodes an inositol polyphosphate 1-phosphatase. FRY1 has been shown to rescue sulfur assimilation mutants in yeast (Quintero et al., 1996) and to negatively regulate ABA and stress signaling in Arabidopsis (Xiong et al., 2001. Mutations in FRY1 are known to enhance the induction of stress genes in response to cold, ABA, salt, and dehydration and have been correlated with elevated levels of the IP 3 second messenger (Xiong et al., 2001). Additional roles for FRY1 in the repression of drought tolerance and attenuation of light responses have recently been proposed (Kim and Von Arnim, 2009;Wilson et al., 2009). An analysis of the stromal proteome of Arabidopsis has shown that FRY1 is localized to the chloroplast (Peltier et al., 2006). However, transient expression analysis of a FRY1-yellow fluorescent protein fusion in onion (Allium cepa) epidermal cells showed that FRY1 is present in the nucleus and the cytoplasm (Kim and Von Arnim, 2009).
The insertional allele of RON1/FRY1/SAL1, ron1-2, is likely to be null because it is predicted to produce a protein lacking 282 amino acids. The similar leaf shape and vein-patterning phenotypes of ron1-1 suggest that it is also null. fry1-1 has also been proposed to be a null allele because it results in the absence of the a-5 helix in the mutant protein (Xiong et al., 2001). The leaves of fry1-1 and fry1-2 are more serrated than those of their corresponding wild type (C24; Xiong et al., 2001), a trait that is also observed for the ron1-2 mutant (in the Col-0 background) but not for ron1-1 (Ler), suggesting that this trait is suppressed in the Ler background.

A Role for RON1 in Vein Patterning
The role of inositol in vascular development was first revealed by cloning of CVP2, which encodes an inositol 5PTase (Carland and Nelson, 2004), and subsequently confirmed by cloning of At5PTase13, which encodes a different 5PTase enzyme (Lin et al., 2005). The similar vascular phenotypes of cvp2, at5ptase13, and ron1 mutants suggest that inositol signaling is critical for vascular patterning. The distinct effects of cvp2 and ron1 mutations on leaf development suggest that different inositol metabolites participate in distinct developmental processes or, alternatively, that some phenotypic traits that are visible in ron1 mutants, such as the rounded leaves, are masked in cvp2 mutants by the presence of a redundant 5PTase activity. Indeed, as many as 15 5PTases are encoded in the Arabidopsis genome (Carland and Nelson, 2004), and one of them, CVL1, has been shown to act with CVP2 to determine vein continuity. While cvp2 mutants only show free-ending veins restricted to leaf and cotyledon margins, and cvl1 mutants do not show any vascular defect, cvp2 cvl1 double mutants also have vein islands, which are not seen in either single mutant (Carland and Nelson, 2009;Naramoto et al., 2009). Our results suggest that cvp2-1 is epistatic on ron1-2 as regards the formation of free-ending veins, since this trait was similar to cvp2-1 in ron1-2 cvp2-1 double mutants. However, ron1-2 cvp2-1 double mutants also showed a slight, but statistically significant, reduction in venation density compared with the single mutants, which resemble the wild type in this parameter, suggesting that the inositol polyphosphate 1-phosphatase activity of RON1 can compensate for the reduction of 5PTase activity in cvp2-1 mutants to some extent regarding this trait.
An interplay between inositol metabolism and auxin perception is indicated by the altered expression patterns of the DR5-GUS reporter in cvp2 and at5p-tase13 mutants as well as by the altered expression levels of auxin-responsive genes in at5ptase13 (Carland and Nelson, 2004;Lin et al., 2005). Although DR5-GUS is often presented as being a reporter of the auxin concentration, it is more accurately described as a reporter of auxin perception. DR5 is an artificial promoter that includes tandemly repeated auxin response elements, known binding sites for auxin response factors (ARFs; Ulmasov et al., 1997). The auxin receptor was identified as the F-box protein TIR1 (Dharmasiri et al., 2005;Kepinski and Leyser, 2005), a component of the SCF TIR1 ubiquitin ligase complex. Binding of auxin to TIR1 leads to the ubiquitination and proteolytic degradation of members of a family of shortlived Aux/IAA proteins via the proteasome. In the absence of auxin, Aux/IAA proteins titrate ARF transcriptional repressors, preventing their binding to auxin response element-containing promoters and thus allowing transcription. Therefore, reduced expression levels of DR5-GUS may reflect low auxin concentrations, the inability to perceive auxin, or a failure to degrade Aux/IAA proteins. Conversely, elevated expression levels of DR5-GUS may translate into high auxin concentrations, active auxin perception, or the degradation of Aux/IAA proteins.
Based on our phenotypic observations of ron1 mutants and on the synergistic phenotypes of ron1 axr1 and ron1 hve double mutants, we hypothesize that inositol compounds function to modulate auxin perception. The mechanism for this modulation is unknown at present, but the presence of a myoinositol 1,2,3,4,5,6-hexakisphosphate (also known as IP 6 and phytate) molecule in the structure of the SCF TIR1 ubiquitin ligase complex bound to auxin (Tan et al., 2007) indicates that the interaction might be direct. Inositol hexakisphosphate has been proposed to be a structural cofactor in the SCF TIR1 complex. Negative charges provided by the phosphates in inositol hexakisphosphate are important in the interaction with positively charged residues in TIR1 (Tan et al., 2007).
Our results provide genetic evidence that loss-offunction mutations in RON1, which affect inositol phosphorylation, dramatically enhance the phenotypes of double mutants with axr1 or hve. It has been suggested that the dynamics of highly phosphorylated inositols, such as IP 6 or IP 5 , may be altered in ron1 mutants (Xiong et al., 2001). AXR1 is required for the neddylation of CULLIN1 (CUL1), one of the core components of the SCF TIR1 complex. HVE encodes CAND1, a protein that physically interacts with unneddylated CUL1 to regulate the formation and/or the function of SCF complexes (Chuang et al., 2004;Feng et al., 2004;Alonso-Peral et al., 2006). Loss-of-function alleles of AXR1 and HVE lead to similarly reduced venation patterns in which most tertiary and higher order veins are absent, highlighting the importance of auxin perception for vascular patterning (Alonso-Peral et al., 2006;Candela et al., 2007). These phenotypes closely resemble those caused by loss-of-function alleles of YUCCA genes in triple or quadruple mutant combinations in Arabidopsis (Cheng et al., 2006). YUCCA proteins are flavin monooxygenases, key enzymes in Trp-dependent auxin biosynthesis (Zhao et al., 2001;Cheng et al., 2006). In our study, the enhanced axr1 and hve mutant phenotypes caused by ron1 alleles led to extremely simple, completely open vascular networks and to deeply lobed leaf laminae. Indeed, some pleiotropic traits in ron1-1 are reminiscent of a mild defect in auxin perception. The increased branching of ron1-1 shoots, although clearly distinct from the severe effects of axr1 and hve mutations on inflorescence branching, may also result from a decrease in apical dominance due to defective auxin perception. In line with this idea, increased shoot branching has recently been described for plants that overexpress an inositol polyphosphate 6-/3-kinase gene (Zhang et al., 2007). Although our assays did not allow us to detect changes in sensitivity to 2,4-D, those authors found that these overexpressors are less sensitive to IAA than the wild type (Zhang et al., 2007).
Alternatively, it is also possible that the effect of ron1 mutations on auxin responses results from a perturbation of auxin transport. This idea is not necessarily in disagreement with changes in auxin perception. Recently, a role of CVP2 in PIN1 trafficking has been described (Carland and Nelson, 2009;Naramoto et al., 2009). In addition to its activity on soluble forms of IP, CVP2 can generate phosphatidylinositol 4-monophosphate (Carland and Nelson, 2009), a specific PI ligand of SFC/VAN3 (Koizumi et al., 2005), an ARF-GAP protein involved in the cell polar localization of PIN1 through vesicle transport (Koizumi et al., 2005;Sieburth et al., 2006). Leaves of the cvp2 cvl1 double mutant show open vein patterns similar to the ones observed in the leaves of sfc (Carland and Nelson, 2009;Naramoto et al., 2009). The function of at least one of these 5-phosphatases is necessary for the correct subcellular localization of VAN3 (Naramoto et al., 2009), which in turn affects vesicle transport and therefore PIN1 polar localization. To our knowledge, no inositol polyphosphate 1-phosphatase activity against PIs has been described so far. However, we cannot rule out that changes in inositol metabolism due to ron1 mutations affect PI metabolism and alter vesicle trafficking. In fact, our double mutant analysis suggests that cvp2-1 is epistatic to ron1-2 for some aspects of vein patterning, suggesting that they act in the same developmental pathway. Future analyses involving genetic interactions between ron1 and van3/sfc mutations and PIN1 subcellular localization will shed light on this issue.
Our ron1-1 transcriptomic analysis did not reveal changes in genes previously described to participate in vein patterning, potentially suggesting that the re-quirement for RON1 activity in vein development functions downstream of these genes. However, one of the most down-regulated genes in the ron1-1 microarray was GH3.3 (At2g23170; 0.36-fold change), which encodes an enzyme that conjugates IAA in vitro to different amino acids, such as Asp (Staswick et al., 2005). Plants may regulate the effects of auxin through the production of different IAA-amido conjugates. In fact, because Asp is one of the amino acids that can be conjugated to auxin with high efficiency in vitro, and because IAA-Asp is a target for auxin degradation (Ljung et al., 2002), the formation of IAA-Asp may result in decreased active IAA levels. Moreover, GH3.3 is highly up-regulated after IAA treatment (Nemhauser et al., 2006), which may reflect the need for the plant to remove excess auxin. The down-regulated expression of GH3.3 in ron1-1 would have the opposite effect: the reduced auxin response in some parts of the mutant may be counteracted by a reduction in the removal of this hormone in an attempt to maintain the correct developmental program controlled by free auxin. Besides, IAA-amido conjugates and IAA-ester conjugates, such as IAA-myoinositol and IAAmyoinositol glucosides, have also been described in plants and may contribute to regulate IAA levels (Ljung et al., 2002), since these ester forms of IAA constitute a storage form of auxin. Changes in the metabolism of inositol may cause an imbalance in the levels of different inositol compounds, which in turn may affect the levels of different inositol conjugates, such as IAA-inositol esters.
Plants produce IAA through different pathways, including Trp-dependent and Trp-independent pathways. Our metabolomic analysis revealed an increase in the levels of indole-3-acetonitrile, a precursor of IAA biosynthesis (Bartling et al., 1994). In line with this result, our transcriptomic analysis showed the downregulation of a gene encoding a putative nitrilase (At4g08790), an enzyme that might catalyze to some extent the conversion of indole-3-acetonitrile into IAA (Piotrowski, 2008). Although we did not detect any change in the levels of Trp, those of shikimate, one of its precursors, were increased. Thus, we cannot rule out that the poor auxin response in the ron1-1 mutant may be due to some extent to an imbalance in auxin levels.

Pleiotropy of ron1 Mutants
It is likely that loss of function of FRY1 restores posttranscriptional gene silencing in the argonaute1-27 (ago1-27) and ago1-33 backgrounds by corepressing the XRN2, XRN3, and XRN4 exoribonucleases, which in turn increases RNA-silencing triggers (Gy et al., 2007). The same authors found similar levels of the fulllength mRNAs of the miR160 target genes ARF10, ARF16, and ARF17 in fry1 mutant and control plants. However, increased levels of the 3# products of the miRNA-guided cleavage products of these genes were found in the fry1 mutant background, probably as a result of inhibition of the cytoplasmic exoribonuclease XRN4 by PAP. In contrast to these results, our microarray analysis revealed increased levels of ARF10, ARF16, and ARF17 mRNAs in ron1-1. It is possible that we detected the accumulation of 3# cleavage products from miRNA target genes in ron1-1 because the probe sets in the ATH1 array were designed using the last 600 bp of the coding sequence (Redman et al., 2004). This premise could explain the apparently contradictory results and should make us cautious about interpreting microarray results involving miRNA target genes. In fact, although we found expression changes in 24 miRNA target genes, it is still possible that the transcription of most of these genes remains unchanged in the ron1-1 background, since, for 17 of them, the probes in the ATH1 microarray were designed downstream of their miRNA cleavage sites.
In an ATH1 microarray analysis performed on leaf RNA from the fry1-1 and alx8 mutants, which carry alleles of RON1, a total of 2,447 and 1,844 were found to be deregulated (Wilson et al., 2009), in contrast with the 663 genes that we found in ron1-1. These differences may be due to the different experimental setup; while Wilson et al. (2009) used 5-week old green leaves from plants grown on soil, we used shoot apices (comprising shoot apical meristem and first and second leaf primordia at the petioleless stage, between 7 and 9 DAS) of seedlings grown on Murashige and Skoog (MS) medium containing Suc. We compared the three microarray analyses and found in common 142 up-regulated and 35 down-regulated genes. We did not detect changes in the expression of genes known to participate in vein patterning in the microarray data set of Wilson et al. (2009). However, two auxin transporters are down-regulated in these mutants: PIN7 in alx8 and fry1-1 and PIN4 in alx8 (Feraru and Friml, 2008).
It is possible that perturbations in the homeostasis of a main signaling molecule such as IP 3 affect a number of developmental, response, and metabolic processes. For example, one of the most conspicuous developmental phenotypic traits of ron1 mutants is late flowering. In the ron1-1 microarray, overexpression of CCA and LHY, two genes encoding MYB-related transcription factors that are part of the Arabidopsis core oscillator, was detected, and constitutive overexpression of CCA1 and gain-of-function alleles of LHY have been shown to cause late flowering (Schaffer et al., 1998;Wang and Tobin, 1998). Thus, it is plausible that the delayed flowering observed for ron1 mutants is caused, to some extent, by the up-regulated expression of these two genes. Moreover, overexpression of CCA1 disrupts circadian gene expression, which may be responsible for the ron1-1 microarray enrichment for down-regulated genes including in their promoters the evening element, which is present in genes with circadian rhythmic expression (Harmer et al., 2000). Since IP 3 triggers Ca 2+ release from intracellular stores (Tang et al., 2007) and since Ca 2+ levels oscillate daily, probably encoding circadian clock signaling infor-mation (Johnson et al., 1995;Love et al., 2004), it is possible that the altered IP 3 homeostasis in the ron1-1 mutant causes changes in the oscillator, resulting in abrogated rhythmic gene expression and changes in light and metabolic pathways that are typically synchronized with day/night cycles. The involvement of RON1 in the attenuation of light responses through its 3#(2#),5#-bisphosphate nucleotidase activity has been recently described (Kim and Von Arnim, 2009). In this study, late flowering under long-day conditions (16 h of light, 8 h of dark) has been related to the photoperiodic control of flowering, since low levels of FLOW-ERING LOCUS T (FT) mRNA are observed in the fry1-6 mutant. Our transcriptomic analysis did not reveal any change in FT transcription in ron1-1, probably due to the different experimental conditions.
The characterization of ron1 suggests that inositol compounds participate in a number of plant developmental processes. Further experiments will help us better understand the interplay between inositol metabolism and auxin responses and the regulatory roles of inositol in plant biology.
For morphometric, metabolomic, and physiological assays, plants were grown on agar medium at 20°C 6 1°C and 60% to 70% relative humidity under 5,000 lux continuous fluorescent light (Ponce et al., 1998). Crosses and allelism tests were carried out as described previously (Berná et al., 1999). For microarray analysis, the medium contained half-strength MS salts (microelements and macroelements), 1 g L 21 Suc, 0.5 g L 21 MES, pH 6.0, and 6 g L 21 plant tissue culture agar. Sixty seeds were sown per 150-3 25-mm round petri dish, and the dishes were sealed with Urgopore tape and incubated in the dark at 4°C for 3 d. The growth chamber conditions were 16/8 h (day/night) with white light (neon tubes, cool-white), 100 mE m 22 s 21 photosynthetically active radiation, and 20°C.

Morphometric and Histological Analyses
Morphometric analyses of leaf venation patterns were performed at 22 DAS for cotyledons and leaves of ron1-1 and Ler and 21 DAS for the comparisons of wild-type, single mutant, and double mutant strains in the Col-0 background. Leaf clearing, venation pattern visualization, and morphometry were performed as described by Alonso-Peral et al. (2006). The number of free-ending veins was determined manually based on drawings of cotyledon and leaf venation patterns.
To determine palisade mesophyll cell size, six first-node leaves of Ler and ron1-1 plants grown in vitro were harvested at 23 DAS. Leaves were cleared with chloral hydrate, and the palisade mesophyll was visualized using differential interference contrast optics and photographed at the widest part of the lamina. Cell size was scored using NIS Elements (Nikon Imaging).

Statistical Analyses
Unless otherwise stated, we tested the normality of our data sets using nonparametric Kolmogorov-Smirnov tests. For data sets comprising two independent samples with 10 or more normally distributed values, the statistical significance of the mean differences of both samples was analyzed using Student's t test. For data sets comprising fewer than 10 values in one or both samples, we employed the Mann-Whitney U test. The analyses were carried out using SPSS 10.0.5 software (SPSS Inc.).

Map-Based Cloning
DNA extraction, amplified fragment length polymorphism, insertion/ deletion, and single-nucleotide polymorphism analyses were performed according to Peters et al. (2004) and Cnops et al. (2004). Fine-mapping of the RON1 locus was conducted using the markers described in Supplemental  Table S1. The seven candidate genes were amplified from genomic DNA and sequenced to identify the nucleotide substitution in ron1-1 (Supplemental Table S2).

Physiological Assays
To study the effects of the auxin polar transport inhibitors NPA and TIBA and the synthetic auxin 2,4-D on root elongation, seedlings were grown on vertically oriented MS agar plates supplemented with different concentrations of NPA (1, 5, and 10 mM), TIBA (1 and 10 mM), or 2,4-D (10, 50, 100, 200, 500, and 1,000 mM). NPA and 2,4-D were added from stock solutions dissolved in dimethyl sulfoxide and NaOH, respectively. To determine root growth inhibition in the presence of NPA, the primary roots of over 50 seedlings were measured at 7 DAS. For the 2,4-D treatments, seedlings grown on vertically oriented MS agar plates were transferred at 5 DAS to plates containing 2,4-D, and the primary roots of 15 or more seedlings were measured 5 d later.
To study the effects of 2,4-D and IAA on the expression of the DR5-GUS transgene, plants grown on agar medium were incubated for 6 h in scintillation vials containing liquid medium supplemented with 1 mM 2,4-D or IAA and subsequently stained for GUS activity as described previously (Alonso-Peral et al., 2006).

Microarray Preparation, Hybridization, and Data Analysis
The microarray experiment, including RNA extraction, experimental design, and data analysis, was processed as described by Fleury et al. (2007) using ATH1 Affymetrix chips of Arabidopsis, Bioconductor Project Release 1.4 (Gentleman et al., 2004) and BiNGO version 2 (Maere et al., 2005) software. The false discovery rate method for multiple testing was used to correct the P value of the Bayesian t test. The differentially expressed genes between the wild type and the mutant were selected at P , 0.01. Since some microarray elements of ATH1 match more than one annotated gene (Arabidopsis Genome Initiative code), we used the list of all annotated genes in the Gene Ontology analysis (BiNGO). The data are available on ArrayExpress (E-TAMB-566).
Aliquots of the RNA samples used for the microarray analysis were employed as templates for qRT-PCR validation. Two micrograms of mRNA was heated for 5 min at 65°C and immediately transferred to ice, followed by addition of 500 ng of 10Xp(dN) 6 primers (Boehringer Mannheim) and incubation of the mixture at room temperature for 10 min. cDNA synthesis was performed in a 40-mL reaction mixture containing 0.5 mM of each deoxyribonucleotide triphosphate, 10 mM dithiothreitol, 200 units of SuperScript II enzyme (Gibco BRL), and 40 units of RNaseOUT. After cDNA synthesis, 80 mL of RNase-free water was added to the reaction mixture. One microliter of the resulting cDNA solution was used for qRT-PCR amplifications, which were carried out in an ABI PRISM 7000 sequence detection system (Applied Biosystems) as described by Pérez-Pérez et al. (2004). Primer pairs (Supplemental Table S4) were designed to yield amplification products of approximately 100 bp. When possible, at least one primer of each primer pair contained sequences complementary to the ends of two contiguous exons, so that genomic DNA could not be amplified. Each 25-mL reaction mix contained 12.5 mL of the SYBR-Green PCR Master Mix (Applied Biosystems), 0.4 mM of primers, and 1 mL of cDNA solution. Relative quantification of the gene expression data was performed using the 2 -DDCT or comparative C T method (Livak and Schmittgen, 2001). Each reaction was performed in triplicate, and expression levels were normalized using the C T values obtained for the OTC housekeeping gene (Quesada et al., 1999).
To compare the ron1-1 microarray data set with that of Nemhauser et al. (2006), a binomial sampling distribution was used (P , 0.05). To analyze gene promoters, we used the Athena Web-based application, which includes a database containing 30,067 predicted promoter sequences from Arabidopsis together with consensus sequences for 105 characterized transcription factor binding sites. The significance of enrichment (P value) for specific sequences was calculated using a hypergeometric probability distribution with Bonferroni correction (http://www.bioinformatics2.wsu.edu/cgi-bin/ Athena/cgi/home.pl; O'Connor et al., 2005).

Metabolite Data
Extraction and derivatization of metabolites from third and fourth leaves of plants collected at 21 DAS using GC-MS were performed as outlined by Lisec et al. (2006). GC-MS data were obtained using an Agilent 7683 series autosampler (Agilent Technologies) coupled to an Agilent 6890 gas chromatograph coupled to a Leco Pegasus 2 time-of-flight mass spectrometer (LECO). Identical chromatogram acquisition parameters were used as those described previously (Weckwerth et al., 2004). Chromatograms were exported from Leco ChromaTOF software (version 3.25) to R software. An in-house R script was employed to perform peak detection, retention time alignment, and library matching. Metabolite data were normalized by dividing each raw value by the median of all measurements of the experiment for one metabolite. To calculate the significant changes in the profiles of ron1-1 and the wild type, t tests with false discovery rate correction were performed using R software.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S2. Expression of DR5-GUS in ron1-1 and wild-type plants incubated with auxin.
Supplemental Figure S3. Rosette phenotype of double mutants.
Supplemental Table S1. Molecular markers used for the fine-mapping of RON1.
Supplemental Table S2. Candidate genes within the 30-kb interval defined by fine-mapping of RON1.
Supplemental Table S4. Primers used for qRT-PCR validation of microarray results.