Auxin biosynthesis by the YUCCA genes in rice.

Although indole-3-acetic acid (IAA), the predominant auxin in plants, plays a critical role in various plant growth and developmental processes, its biosynthesis and regulation have not been clearly elucidated. To investigate the molecular mechanisms of IAA synthesis in rice, we identified seven YUCCA - like genes (named OsYUCCA1-7 ) in the rice genome. Plants overexpressing OsYUCCA1 exhibited increased IAA levels and characteristic auxin-overproduction phenotypes, whereas plants expressing antisense OsYUCCA1 cDNA displayed defects that are similar to those of rice auxin-insensitive mutants. OsYUCCA1 was expressed in almost all of the organs tested, but its expression was restricted to discrete areas including the tips of leaves, roots, and vascular tissues, where it overlapped with expression of a GUS reporter gene controlled by the auxin-responsive DR5 promoter. These observations are consistent with an important role for the rice enzyme OsYUCCA1 in IAA biosynthesis via the Tryptophan-dependent pathway. we and characterized a YUCCA-like gene in rice, OsYUCCA1 . Plants overexpressing OsYUCCA1 exhibited increased IAA levels and auxin-overproduction phenotypes, whereas plants suppressing antisense OsYUCCA1 cDNA showed abnormal phenotypes similar to auxin-insensitive rice. Moreover, OsYUCCA1 is not expressed ubiquitiously, rather its expression is restricted to discrete places such as the tips of and roots, and parenchyma cells surrounding large vascular stem tissues. Based on these observations, we propose a molecular mechanism for IAA synthesis in rice.

INTRODUCTION enzyme, was identified by isolation of a dominant mutant with elevated levels of IAA (Zhao et al., 2001). Since YUCCA can convert tryptamine (TAM) to N-hydroxytryptamine (NHT) in vitro, Zhao et al. (2001) suggested that YUCCA catalyzes the N-oxygenation of TAM, a rate-limiting step in auxin biosynthesis in many plants. Involvement of YUCCA in IAA synthesis is also supported by molecular analysis of a petunia mutant, floozy (fzy) (Tobeña-Santamaria et al., 2002). The fzy mutant was first isolated as a flower mutant in which the formation of the outermost three floral whorl formations and one of the two bracts is blocked at an early stage. FZY encodes an FMO-like enzyme homologous to Arabidopsis YUCCA, and its overexpression results in increased IAA levels and in an auxin-overproduction phenotype (Tobeña-Santamaria et al., 2002). More recently, Cheng et al. (2006) reported that double, triple, and quadruple mutants of four Arabidopsis YUCCA genes display severe defects in floral patterning, vascular formation, and the other developmental processes. The defects in the as enzymes that catalyze conversion of Trp to IAOx in vitro (Hull et al., 2000;Mikkelsen et al., 2000), are critical enzymes in auxin biosynthesis in vivo. Arabidopsis plants overproducing CYP79B2 contained increased levels of free auxin and exhibited auxin-overproduction phenotypes, and double knock-out of the CYP79B2 and CYP79B3 genes caused a reduction in IAA levels and induced growth defects related to partial auxin deficiency. These findings concerning YUCCA and CYP79B2/CYP79B3 strongly suggest that a Trp-dependent pathway occurring via an IAOx intermediate is a major source of IAA in Arabidopsis.
Although recent molecular genetic studies have successfully identified a few genes involved in the Trp-dependent IAA biosynthetic pathway in Arabidopsis, auxin biosynthesis mechanisms in monocots have not been defined (Zhao et al., 2002). As a model monocot, rice has been extensively studied. The entire rice genome sequence is known, and full-length cDNAs, transformation systems, and many mutant collections are

Cloning of rice YUCCA homologs and analysis of their expression patterns
To identify rice homologs of the Arabidopsis YUCCA genes, we performed a tBLASTn search against all available rice DNA databases, using the known Arabidopsis YUCCA1 protein as a query sequence (Zhao et al., 2001). This search yielded seven rice homologs (OsYUCCA1-7), each of which encodes a protein containing the conserved binding motifs for FAD and NADPH (Supplemental figure 1). We also analyzed the phylogenic relationships among these homologs based on alignments of their full-length amino acid sequences (Fig. 2). As previously reported (Zhao et al., 2001;Woodward et al., 2005, Cheng et al., 2006, the rice and Arabidopsis YUCCA proteins cluster into three groups. One group includes YUCCA1, YUCCA4, YUCCA10, YUCCA11, and one of the OsYUCCA proteins, which we named OsYUCCA1. The second group contains YUCCA2, YUCCA6, and four of the OsYUCCA proteins, which we named OsYUCCA2-5. The third group contains YUCCA3, YUCCA5, YUCCA7, YUCCA8, YUCCA9 and two rice homologues, which we named OsYUCCA6 and OsYUCCA7. To determine whether the OsYUCCA genes are actually expressed in rice organs, we used semi-quantitative reverse transcription-PCR (qRT-PCR) analysis with primers that specifically amplified the 3'-noncoding regions of OsYUCCA1-7. This approach was necessary because the levels of OsYUCCA mRNA were too low to detect using RNA gel blot analysis (data not shown). Before performing the qRT-PCR analysis, we tested the efficiencies of the primer sets by using them in PCR reactions with the rice genomic DNA as a template. The amounts of the resulting PCR products increased similarly as a function of cycle number (Fig. 3A), confirming that the primer sets were of similar efficiencies.
Using the same experimental conditions as for the genomic template, we performed qRT-PCR with template cDNA produced from the total RNA isolated from various organs. At 35 cycles, strong bands corresponding to OsYUCCA1 were amplified from cDNAs from leaf blade, vegetative shoot apex, reproductive shoot apex, and flower, bands of intermediate intensity were amplified from leaf sheath cDNA, and faint bands were amplified from root cDNA (Fig. 3B,35 cycles). In contrast, only intermediate or faint intensity bands of ambiguous specificity were observed for OsYUCCA6 or 7, and sheath and vegetative shoot apex, whereas the products corresponding to OsYUCCA2 and 5 were preferentially produced from leaf sheath, vegetative shoot apex, and root. When we used DNA samples skipped the reverse transcriptase step, no bands were produced.
These results indicate that all of the OsYUCCA genes are expressed in rice, but OsYUCCA1 expression predominates in almost all organs.
To help elucidate which the OsYUCCA genes are associated with IAA biosynthesis, we examined their spatial expression patterns in coleoptiles because of the localization of auxin in the tip of grass coleoptiles, (Fig. 3C). If the OsYUCCA gene products are involved in IAA production, the OsYUCCA mRNAs may be preferentially expressed at the coleoptile tip. As we expected, OsYUCCAs 1, 5, and 6 were preferentially expressed in the top of rice coleoptiles in comparison to the middle or bottom, whereas the other OsYUCCAs did not exhibit a preferential pattern (Fig. 3D). Taken together these observations, we expected that OsYUCCA1 may dominantly play a role in IAA production in rice, and therefore focused on this gene in further study. rice ACT1 gene promoter (McElroy et al., 1990) was used to induce constitutive, high-level expression of OsYUCCA1 in rice calli (Fig. 4A). The growth rate and regeneration frequency of calli transformed with pAct-OsYUCCA1 were greatly decreased, as compared to those of calli transformed with a control vector, probably because increased auxin production is inhibitory. Because addition of 1.5× kinetin to the growth and regeneration media partially improved the frequency of overexpressing calli (data not shown), we used 1.5× kinetin in subsequent transformations of several hundred rice calli with pAct-OsYUCCA1. Twenty-one independent reentrants were obtained.

Phenotypes of plants overexpressing
In many cases, the transformed calli did not produce plantlets; instead, they produced extensive hairy roots that developed from adventitious roots directly formed from calli (Fig. 4B). Sometimes, small leaflets sporadically protruded from calli enveloped with hairy roots (Fig. 4C). The leaflet morphology was aberrant in that the vascular tissues often stopped at the leaf margin ( Fig. 4D). Several regenerated plantlets produced several or more leaves with a more normal structure (Fig. 4E), but these plantlets with the mildest phenotypes still exhibited dwarfism (Fig. 4E, right).
In addition to inhibiting leaf growth, overexpression of OsYUCCA1 also severely inhibited root elongation in the mild phenotype plants, whereas formation of crown root (adventitious root, the major component of the rice root system) was promoted (Fig. 4E).
Interestingly, some plantlets with a mild phenotype initiated internode elongation at an early stage ( Fig. 4F and G); initiation of internode elongation does not occur in WT plants under normal conditions until the vegetative-to-reproductive phase changes. Active crown-root formation still occurred at the upper node after internode elongation (Fig. 4F).
The increased formation of crown root corresponded well to the high frequency of adventitious root formation from calli, demonstrating that increased IAA promotes adventitious root formation from both callus and stem. When the crown roots of regenerated plantlets carrying the control vector reached the bottom of the plant box, almost all of the roots coiled at the bottom (Fig. 4H), whereas the roots of plants overexpressing OsYUCCA1 did not coil, and their root diameters were greater than those of WT plants (Fig. 4I). The lack of root coiling in the overexpressing plants may be related to their abnormal gravitropism, as discussed below.
All of the above observations indicate that overexpression of OsYUCCA1 causes abnormal leaf, root, and stem development. These phenotypes are so severe, however, that the possibility that indirect effects of OsYUCCA1 overexpression had contributed to the observed phenotypes could not be discounted. Therefore, to more directly observe the effects of increased OsYUCCA1 function, we used a steroid-hormone-inducible system for OsYUCCA1 expression (Aoyama and Chua, 1997). We also constructed a plasmid containing the GUS reporter gene under the control of the DR5 promoter (Fig. 5A). The DR5 promoter, which is often used as an artificial auxin-responsive promoter in Arabidopsis, was used to create a marker for visualization of the in vivo distribution of auxin in rice (Ulmasov et al., 1997;Sabatini et al., 1999;Scarpella et al., 2003).
Ten independent transgenic plants carrying the DR5-GUS transgene were obtained.
From these plants, we selected three lines that expressed GUS at their root tips and did not express detectable OsYUCCA1 mRNA in the absence of the inducer dexamethasone (DEX) (data not shown). In the absence of DEX treatment, these transgenic plants exhibited normal phenotypes, as did plants carrying the control vector ( Fig. 5B, left). In contrast, when the plants were treated with DEX for one week, the newly developed crown roots produced abundant hairy roots, as was also observed for the constitutively overexpressing plants (compare Fig. 4C and 5B, right). In DEX-containing agar medium, the roots of the transgenic plants did not respond normally to gravity, but their gravity response was normal in a medium without DEX ( showed no abnormal phenotype (Fig. 6B), although its expression was almost completely suppressed in these antisense plants (Fig. 6C). These observations strongly support that OsYUCCA1 is important for growth and development of shoot and root in rice.

Sites of OsYUCCA1 expression in rice plants
From the above observations, we concluded that OsYUCCA1 is an important enzyme for auxin biosynthesis in rice. Reasoning that the specific localization of OsYUCCA1 expression might indicate the site of auxin biosynthesis, we next examined OsYUCCA1 expression by comparing the expression levels of GUS reporter genes under control of either the OsYUCCA1 promoter or the DR5 promoter (Fig. 7). In shoots, GUS staining controlled by the OsYUCCA1 promoter was observed at the tip of young leaves OsYUCCA1-GUS expression was broadly seen in young leaf primordia around the vegetative shoot apical meristem (SAM) but was virtually absent in the SAM itself ( Fig.   7B). Almost no GUS staining was observed in the DR5-GUS plants, and it was absent from the leaf primordia (Fig. 7C). Thus, IAA may undergo rapid transport after synthesis.
OsYUCCA1-GUS expression was also seen in parenchymal cells surrounding large vascular bundles of stem (Fig. 7B, arrows), and a similar staining pattern was observed in transgenic plants carrying DR5-GUS with more faint activity (Fig. 7C, arrows). Stem cross-sections of OsYUCCA1-GUS plants clearly showed localized GUS staining in cells surrounding large vascular bundles (Fig. 7D), as well as in a broader area of the peripheral vascular cylinder in DR5-GUS plants (Fig. 7F).
In root, OsYUCCA1-GUS expression was faint and was localized at the tip (Fig. 7G), whereas that of DR5-GUS was stronger and also occurred in the root apical meristem (Fig.   7K). OsYUCCA1-GUS expression was detected at the root tip, and not in other root portions (Fig. 7H), whereas DR5-GUS expression was also observed strongly in stele, especially at the base of lateral root (Fig. 7L). OsYUCCA1-GUS expression was not detected in the floral SAM itself, but it was observed in cells surrounding the vascular bundle ( Fig. 7I), whereas no DR5-GUS expression was detected in the same region (Fig.   7M). In developing flower, GUS expression was observed at the flower tip and vascular tissues for both promoters ( Fig. 7J and N) 2000;Zhao et al., 2001Zhao et al., , 2002.
In contrast to CYP79B2/CYP79B3 whose counterparts in rice have not been identified, rice has at least seven genes encoding proteins similar to Arabidopsis YUCCA proteins. Expression analysis of these YUCCA-like genes indicated that one of them,

OsYUCCA1, functionally predominates in various organs of rice, although other
OsYUCCAs were also expressed in various organs and may confer redundant functions in these organs (Fig. 3B)

DEX-dependent manner in transgenic rice plants carrying a DEX-inducible OsYUCCA1
construct (Fig. 5E). Second, when GUS was expressed under control of the DR5 promoter, DEX treatment led to enhanced GUS staining in the root tips of the transgenic rice plants carrying a DEX-inducible OsYUCCA1 construct (Fig. 5D). Third, OsYUCCA1 antisense plants exhibited abnormal phenotypes associated with auxin, such as shoot dwarfism and inhibited development and elongation of root (Fig.  6B). Fourth, OsYUCCA1-overexpressing rice plants exhibited phenotypes typical of IAA overproduction including root hair production ( Fig. 4B and C, and 5B) and agravitropism ( Fig. 4I and 5C). Finally, OsYUCCA1 was found to preferentially express in the top coleoptiles region (Fig. 3D), which is considered to be one of the most active sites of IAA production (Fig. 3C, Ribnicky et al., 1998;Philippar et al., 1999;Koshiba and Matsuyama, 1993).
As is often pointed out, phenotypic results from transgenic plants overexpressing a gene of interest must be interpreted with caution. Conclusions as to the biological function(s) of the overexpressed gene based on the transgene phenotype may be erroneous in the overexpression context. As an artifact of overexpression of the transgene, the gene may appear to have functions other than or in addition to its normal function(s).
However, our results demonstrating preferential expression of OsYUCCA1 in the area  at least seven YUCCA-like genes including OsYUCCA1 (Fig. 2), and therefore, it is very likely that these OsYUCCA genes are also redundantly involved in IAA biosynthesis in rice. The fact that antisense plants of OsYUCCA1 showed abnormal phenotypes associated with auxin deficiency indicates that genetic redundancy of the YUCCA genes may be less complex (Fig. 6B). OsYUCCA1 is a sole rice gene classified into the group of Arabidopsis YUCCA1 and 4; inactivation of both yuc1 and yuc4 affects flower development in Arabidopsis (Cheng et al., 2006).
Although the mechanism by which the product of YUCCA catalysis, NHT, is converted to IAA has not been conclusively demonstrated, a possible pathway leading from NHT to IAA in Arabidopsis involves conversion of NHT to IAOx, followed by conversion of IAOx to IAA via indole-3-acetonitrile (IAN) or indole-3-acetaldehyde (IAAld) ( Fig. 1; Zhao et al., 2002). If rice synthesizes IAA by the same pathway, then rice should contain the enzymes catalyzing these steps. Recently, Park et al. (2003) reported that the maize nitrilase ZmNIT2 hydrolyzes IAN to IAA with efficiency at least 7-to 20-fold higher than that of Arabidopsis NIT enzymes. This result, and the fact that ZmNIT2 is expressed in kernels, where high concentrations of IAA are synthesized by the Trp-dependent pathway, led them to suggest that ZmNIT2 is involved in IAA synthesis in maize. In our study, we found two genes similar to ZmNIT2 in the rice genome, and one of these genes encodes a protein 90% identical to ZmNIT2 at the amino acid sequence level (data not shown). Thus, this rice gene may be an orthologue of ZmNIT2 that functions in IAA synthesis in rice.
The alternative pathway leading from IAOx to IAA occurs via an IAAld intermediate. Seo et al. (1998) reported that aldehyde oxidase (AAO1), which catalyzes the conversion of IAAld to IAA, may be involved in IAA synthesis in Arabidopsis seedlings. We found three homologues of Arabidopsis AAO1 in the rice genome (data not shown). Although distinguishing between IAA-and ABA-synthesis-related aldehyde oxidases by their primary structures alone is difficult, owing to their highly similar sequences, one or more of these apparent AAO1 homologues may be involved in conversion of IAAld to IAA in rice.
The above results are consistent with the proposed biosynthetic pathway for IAA in rice shown in Figure 1. In the first two steps of this proposed pathway, which occur in the cytoplasm, Trp is converted to TAM by a decarboxylase reaction, and TAM is then oxygenized by OsYUCCA1 to produce NHT. The direct conversion of Trp to IAOx by CYP79B2/CYP79B3, an alternative pathway occurring in chloroplasts, appears not to contribute to IAA production in rice. In the third and fourth steps, the product of the YUCCA-catalyzed reaction, NHT, is converted to IAOx and then to IAN. In the final two steps, nitrilase converts IAN to IAAld, and aldehyde oxidase converts IAAld to IAA, as in Arabidopsis.
al. (1994). Transformed cells and plants were screened by selection for hygromycin (Wako Pure Chemical Industries) and were maintained in sterile culture. Regenerated plants were then grown to maturity in pots in a greenhouse. The primary transformants were self-pollinated, and the resulting seeds (T1) were collected.

RNA isolation and RT-PCR analysis
To estimate mRNA expression levels for OsYUCCA genes 1-7 in rice organs, semi-quantitative RT-PCR analysis was performed. Rice plants in vegetative phase were grown for 2 weeks in MS medium containing 3% sucrose and 0.3% Gelite, unless otherwise indicated, and plants in reproductive phase were grown for 2 months in soil-filled pots in a greenhouse. Total RNA was extracted with TRIzol regent (Invitrogen, Carlsbad, CA, USA), and the first-strand cDNA was synthesized from 2 µg of total RNA using an Omniscript reverse transcription kit (Qiagen, Hilden, Germany). The PCR parameters for the detection of the OsYUCCA genes were 94°C for 5 min followed by 25, OsIAA3 gene was analyzed using the forward primer 5'-gggttctccaagacatgcaatc-3' and the reverse primer 5'-ggatggaaatcaaagcattgaagc-3'. The PCR parameters for the detection of OsIAA3 were 94°C for 5 min, followed by 30 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min.

Phenotypic analysis
Plants were photographed using a digital camera (Nikon, Tokyo, Japan). To examine the response of plants to gravity, seedlings were grown vertically for 10 days under continuous light and dark conditions, rotated 90° to the horizontal plane, and then grown for 2 more days.

Quantification of endogenous IAA
The preparation of samples and procedure of IAA quantification were according to Mori   Genes or mutants that have been identified with particular enzymatic steps are specified boxes. The steps indicated by dotted arrows are shared with the glucosinolate pathway, which is restricted to a few plant families but not in rice.