A Role for Flavin Monooxygenase-Like Enzymes in Auxin Biosynthesis

See allHide authors and affiliations

Science  12 Jan 2001:
Vol. 291, Issue 5502, pp. 306-309
DOI: 10.1126/science.291.5502.306


Although auxin is known to regulate many processes in plant development and has been studied for over a century, the mechanisms whereby plants produce it have remained elusive. Here we report the characterization of a dominant Arabidopsis mutant,yucca, which contains elevated levels of free auxin.YUCCA encodes a flavin monooxygenase–like enzyme and belongs to a family that includes at least nine other homologousArabidopsis genes, a subset of which appears to have redundant functions. Results from tryptophan analog feeding experiments and biochemical assays indicate that YUCCA catalyzes hydroxylation of the amino group of tryptamine, a rate-limiting step in tryptophan-dependent auxin biosynthesis.

Auxin is an essential plant hormone that influences many aspects of plant growth and development, including cell division and elongation, differentiation, tropisms, apical dominance, senescence, abscission, and flowering (1). Although auxin has been studied for over 100 years, the mechanisms of its biosynthesis remain elusive. Multiple pathways have been proposed for the biosynthesis of indole-3-acetic acid (IAA) (the main auxin), including several tryptophan-dependent pathways and a tryptophan-independent one (2). Auxin biosynthesis can be catalyzed not only by endogenous plant enzymes, but also by enzymes encoded on plasmids of plant pathogens such as Agrobacterium(2). Whereas the bacterial pathways are well documented, knowledge of the endogenous plant pathways is scant. Rate-limiting steps have not been identified. Overexpression of the only enzyme conclusively shown to be involved in one of the tryptophan-dependent pathways does not cause any auxin-related phenotypes, although loss-of-function mutants are resistant to the effects of the auxin precursor indole-3-acetonitrile (3–6). Part of the difficulty in studying auxin biosynthesis has been the inability to generate auxin-deficient mutants, although genetic screens have identified several recessive Arabidopsis mutants with elevated levels of free auxin (7–11). However, it is not known whether elevated auxin levels in these mutants are caused by indirect effects on auxin biosynthesis or by defects in auxin conjugation. In addition, the mutants are either sterile or have intrinsic heterogeneity in phenotypes, which limits their utility for dissection of auxin biosynthetic pathways. Here we report a dominant and fertile mutant, yucca, which has an elevated level of endogenous auxin. We present evidence that YUCCA, a flavin monooxygenase (FMO)–like enzyme, catalyzes a key step in Arabidopsistryptophan-dependent auxin biosynthesis.

Two independent yucca mutants were identified with long hypocotyls by activation tagging (12). The genomic DNA adjacent to the right border of the T-DNA insertions in the two alleles was cloned by plasmid rescue (Fig. 1A). Comparison of the rescued DNA insertions showed that the cauliflower mosaic virus 35S (CaMV 35S) enhancer arrays in the two mutants had inserted within a 3-kilobase (3-kb) region downstream of the coding region of the same gene (GenBank accession CAB79971). Northern blot analysis of total RNA from both alleles showed that the expression level of this gene was higher inyucca than in wild type. This indicated that the two mutants carried dominant alleles at the same locus. Furthermore, transforming wild-type Arabidopsis either with genomic DNA of this gene placed next to a tetramer of 35S enhancers or with a cDNA of this gene under the control of the complete 35S promoter recapitulated the yucca phenotype. Together, these experiments identified this gene as theYUCCA gene. Because the original yucca alleles were poorly fertile, recapitulation lines with weaker phenotypes were used for further studies. For simplicity, we refer to the recapitulation lines as yucca mutants as well.

Figure 1

Identification of YUCCA. (A) Rescued 10-kb Eco RI plasmid from yucca-2D mutant, which was used for recapitulation of phenotype. ampr, bacterial ampicillin resistance; RB, right T-DNA border. (B) Wild-type (left) andyucca seedlings grown on 0.5X MS medium in white light. (C) Roots of wild-type (left) and yucca grown on a vertical plate in white light. (D) Dark-grown wild-type (left) and yucca seedlings. (E) Matureyucca (top) and wild type (bottom) grown in soil in the greenhouse. (F) Delayed development of yucca(top). (G) Increased apical dominance of yucca(right).

In addition to long hypocotyls, yucca plants had phenotypes characteristic of elevated auxin levels during all stages of development. In the aerial part, yucca seedlings had epinastic cotyledons and elongated petioles when grown in white light (Fig. 1B) similar to known auxin overproduction mutants such assur1 (7, 9), sur2(8), and transgenic plants that overexpress theiaaM auxin synthetic gene from Agrobacterium(13). Roots of yucca were shorter, but the root hairs were much longer and more plentiful than those of wild-type seedlings (Fig. 1C). Dark-grown yucca seedlings had short hypocotyls and lacked an apical hook (Fig. 1D). Mature leaves ofyucca were narrower than those of wild type and were epinastic with long blades and petioles (Fig. 1, E and F). Adultyucca plants had increased apical dominance (Figs. 1G and2G). All these traits are expected from elevated free auxin levels (13, 14). In addition, the matureyucca leaves curled downward and had a semi-erect growth habit, resembling the commonly known yucca plant (Agave sp.) (Fig. 1, E and F). This gives rise to the mutant's name.

Figure 2

Evidence for elevated levels of endogenous auxin inyucca. (A) GC-MS analysis of free IAA levels (15). (B) Explants of wild-type (left) andyucca cotyledons grown on MS medium for 4 weeks. (C) Explants of wild-type (left) and yuccacotyledons grown on MS containing 6-(γ,γ-dimethylallylamino)-purine riboside (IPAR) (2 mg/l) for 4 weeks. (D) Explants ofyucca cotyledons grown on MS containing IPAR (2 mg/l) for 7 weeks. (E) Histochemical staining of dark-grown DR5-GUS (left) and yucca/DR5-GUS seedlings. (F) Histochemical staining of light-grown DR5-GUS (left) andyucca/DR5-GUS leaves. (G) Suppression of theyucca adult phenotype by overexpression of theiaaL gene: yucca (left), yucca/iaaL(middle), and iaaL overexpression line (right).

We carried out several experiments to show directly thatyucca plants contain elevated auxin levels. First, we used gas chromatography–mass spectrometry (GC-MS) to measure endogenous levels of free IAA (15). To obtain sufficient quantity of tissue for IAA measurements, we used a rather weak recapitulation line. This line contained about 50% more free IAA than wild-type plants, suggesting that strong yucca alleles contain even higher levels of free IAA (Fig. 2A).

Next, we used physiological and genetic experiments to test whether this modest increase in free IAA is physiologically important. It has been established previously that Arabidopsis explants cannot proliferate without addition of auxin to the medium. It has also been established that the ratio of auxin to cytokinin determines the relative growth of roots, shoots, and callus (16). In general, a high ratio of auxin to cytokinin leads to root growth, but a lower ratio induces callus and shoot growth (16). Extensive root growth was observed whenyucca cotyledon explants were grown in Murashige-Skoog (MS) medium, whereas the wild-type explants did not proliferate at all (Fig. 2B). When cytokinin was added to the medium, the length and the abundance of the yucca explants' roots were both reduced, whereas the wild-type explants were left unchanged (Fig. 2C). As expected, an increase in cytokinin concentration led to the formation of callus, which produced shoots from which flowers were generated eventually (Fig. 2, C and D). The ability of yucca mutant tissue to proliferate and differentiate in an auxin-free medium indicates that yucca produces more auxin than the wild type. This conclusion is also supported by the observation that an auxin-reporter gene was much more active in yucca than in wild type (17) (Fig. 2, E and F).

If the yucca phenotype is caused by elevated levels of free auxin, then it should be possible to suppress the phenotype by reducing levels of free auxin. To test this hypothesis, we made use of the bacterial iaaL gene, which encodes an enzyme that conjugates free IAA to lysine. When iaaL is overexpressed inArabidopsis plants, the pool of free IAA is reduced, which leads to decreased apical dominance (18). We found that overexpression of iaaL did indeed mask the yuccaphenotype (Fig. 2G), consistent with the interpretation that elevated auxin levels were the cause of yucca's phenotype.

The predicted YUCCA protein with 414 amino acid residues has similarity to FMOs from mammals (19, 20) and contains conserved motifs for binding of flavin-adenine dinucleotide (FAD) and NADPH (the reduced form of nicotinamide adenine dinucleotide phosphate) (Fig. 3A). BLAST searches indicated that there are two families of FMO-like proteins encoded in theArabidopsis genome (Fig. 3B). The YUCCA family contains nine other proteins, which share 44 to 64% amino acid sequence identity with YUCCA. To test whether the YUCCA homologs have similar functions in vivo, we overexpressed several of them under the control of CaMV 35S enhancers. Overexpression of YUCCA2 (CAB41936), which has 53% amino acid sequence identity to YUCCA, causedyucca-like phenotypes (21). We also isolated two other yucca-like mutants with activation tagging. Both mutants have elevated expression of YUCCA3 (AAB80641), which has 51% amino acid sequence identity to YUCCA (22). The similar phenotypes of plants overexpressing YUCCA,YUCCA2, and YUCCA3 demonstrate that these genes are functional paralogs and that they likely have redundant functions. Functional redundancy may explain why neither single- nor double-knockout alleles of YUCCA and YUCCA2 have obvious phenotypes (23). However, not allYUCCA-like genes have identical effects. BAS3(CAA22980), which has 50% and 63% amino acid sequence identity with YUCCA and YUCCA3, respectively, was identified in an activation-tagging screen for phyB4 suppressors. Whereas dominantyucca mutants have long hypocotyls in the light, overexpression of BAS3 shortened the long hypocotyls ofphyB mutants (24). Thus, plant FMO-like enzymes have distinct in vivo functions, which cannot be predicted a priori from sequence analysis alone.

Figure 3

YUCCA is a member of the FMO-like family. (A) Amino acid sequence of YUCCA. The putative FAD and NADPH binding motifs near the NH2 terminus and in the middle of the protein, respectively, are indicated in bold. (B) Phylogenetic tree of YUCCA and related Arabidopsis genes along with human FMOs (Hs FMO1/2/4). GenBank accession numbers are shown. Proteins analyzed here are indicated in bold. (C) Tryptamine is a substrate for YUCCA in vitro. Reaction mixture with tryptamine as the substrate was separated on a TLC plate (29). The arrow points to the enzymatic product; 1 is the control and 2 is the reaction (29). The product was characterized by mass spectrometry (29). Assignments of the pseudo-molecular ion (MH+) and major fragmentation ions are indicated.

To test whether the elevated level of auxin in yuccais produced via a tryptophan-dependent auxin biosynthetic pathway or a tryptophan-independent pathway, we grew plants in media containing tryptophan analogs. Analogs such as 5-methyl-tryptophan (5-mT) are toxic to plants because they inhibit tryptophan biosynthesis and disrupt the functions of proteins in which they have been incorporated (25). Although 5-mT is ultimately toxic to both wild type and yucca, yucca plants are much more resistant to its effects (Fig. 4A). Although wild type cannot grow in MS medium containing 100 μM 5-mT,yucca not only survived but also gained additional phenotypes, including the growth of adventitious roots from the hypocotyl (Fig. 4A). The same root phenotype could also be produced by the addition of 5-methyl IAA to the medium. This suggests that 5-mT is rendered nontoxic by conversion of 5-mT to 5-methyl IAA and that the latter is an active auxin that induces adventitious root growth. These results indicate that auxin is produced in yucca via a tryptophan-dependent pathway.

Figure 4

(Left) YUCCA is involved in tryptophan-dependent auxin biosynthesis, and the YUCCA pathway is functional in other plants. (A) yucca is less sensitive to toxic tryptophan analogs. Wild-type (left) and yucca seedlings were grown on 0.5X MS medium containing 100-μM 5-mT for 10 days. (B) Comparison of wild-type (left) and transgenic tobacco plants overexpressing YUCCA.

To test whether the YUCCA pathway is generally used by other plants to synthesize auxin, YUCCA was overexpressed in tobacco (26) (Fig. 4B). The transgenic tobacco plants had a dramatic change in phenotype, including long, narrow, epinastic leaves, similar to the phenotypes observed inArabidopsis.

There are several proposed auxin biosynthetic pathways that use tryptophan as the precursor (Fig. 5). Based on the well-characterized oxidation of heteroatom-containing compounds by mammalian FMOs (19, 20, 27,28), we identified the proposed auxin biosynthetic intermediate tryptamine as the most likely candidate as a YUCCA substrate. Indeed, when tryptamine was used in vitro as a substrate for recombinant YUCCA, a product was formed that had a mass-spectral fragmentation pattern consistent with that of N-hydroxyl tryptamine (Fig. 3C) (29).

Figure 5

(Right) YUCCA catalyzes a key step in auxin biosynthesis. Putative tryptophan-dependent auxin biosynthesis pathways and intermediates are shown (2). The indole-3-acetaldoxime intermediate was proposed recently (25).

These results lead us to propose that YUCCA catalyzes the N-oxygenation of tryptamine, and that this transformation is a rate-limiting step in auxin biosynthesis in many plants (Fig. 5). We propose furthermore that the limiting factor is the level of YUCCA and its paralogs, consistent with the observation that indole-3-acetaldoxime but not tryptamine has auxin activity when added to growth medium (30).

Although mammalian FMOs have been well characterized by biochemical means (27, 28), their physiological functions remain unknown (19, 20). That anArabidopsis FMO-like enzyme is involved in tryptophan metabolism suggests that some mammalian FMOs may also have important physiological functions in tryptophan metabolism. Further genetic analysis of Arabidopsis FMO-like genes may yield additional clues that can be used to elucidate the physiological roles of their mammalian counterparts.

  • * Present address: Department of Molecular, Cell, and Developmental Biology, University of California, Los Angeles, 2124 Life Sciences Box 951606, Los Angeles, CA 90095–1606, USA.

  • Present address: Department of Molecular Biology, 30 quai Ernest Ansermet, 1211 Geneve 4, Switzerland.

  • To whom correspondence should be addressed. E-mail: chory{at}


View Abstract

Navigate This Article