Report

tasselseed1 Is a Lipoxygenase Affecting Jasmonic Acid Signaling in Sex Determination of Maize

See allHide authors and affiliations

Science  09 Jan 2009:
Vol. 323, Issue 5911, pp. 262-265
DOI: 10.1126/science.1164645

Abstract

Sex determination in maize is controlled by a developmental cascade leading to the formation of unisexual florets derived from an initially bisexual floral meristem. Abortion of pistil primordia in staminate florets is controlled by a tasselseed-mediated cell death process. We positionally cloned and characterized the function of the sex determination gene tasselseed1 (ts1). The TS1 protein encodes a plastid-targeted lipoxygenase with predicted 13-lipoxygenase specificity, which suggests that TS1 may be involved in the biosynthesis of the plant hormone jasmonic acid. In the absence of a functional ts1 gene, lipoxygenase activity was missing and endogenous jasmonic acid concentrations were reduced in developing inflorescences. Application of jasmonic acid to developing inflorescences rescued stamen development in mutant ts1 and ts2 inflorescences, revealing a role for jasmonic acid in male flower development in maize.

Most flowering plants produce perfect flowers containing both the male organs (stamens) and female organs (pistils). In maize, which has physically separated male and female inflorescences, floral meristems become unisexual through sex determination [reviewed in (1, 2)]. The basic unit of the maize inflorescence, called a spikelet, contains one upper and one lower flower (known as florets in grasses). Each floret initiates a series of floral organs including three stamen primordia and a central pistil primordium (3, 4). These initially bisexual florets become exclusively staminate in the tassel (by abortion of pistil primordia) and exclusively pistillate in the ear (by arrest of developing stamens) (3, 5). Each ear spikelet produces a solitary functional pistil in the upper floret due to abortion of the pistil in the lower floret (35).

Mutations altering the sexual fate of florets in maize indicate that sex determination is under genetic control. The nonhomeotic tasselseed (ts) mutations ts1 and ts2 result in the conversion of the tassel inflorescence from staminate to pistillate (6, 7). Both ts1 and ts2 are required to eliminate pistil primordia through cell death (8, 9). The ts2 gene encodes a short-chain dehydrogenase/reductase (10) with broad activity, which has complicated the discovery of its natural substrate (11). It is unknown how ts genes mediate pistil cell death, although it has been suggested that the dehydrogenase/reductase activity of ts2 may produce a proapoptotic signal or metabolize a substrate required for cell viability (8, 11). Even less is known about the ts1 gene. TS2 transcripts are low or undetectable in ts1 mutant tassels, which suggests that ts1 may act upstream of ts2 by regulating ts2 RNA levels and possibly other sex determination genes (8).

We positionally cloned and mapped ts1 (12) (fig. S1), which is located in a region with extensive synteny with rice (13). The ts1 syntenic interval within the sequenced genome of rice contains nine genes (12). We found no maize orthologs for four of these genes, and another three mapped to locations unlinked to the ts1 locus in maize. Maize homologs of the remaining two rice genes, one encoding a putative glutamate decarboxylase and the other encoding a putative lipoxygenase, were confirmed to be contained within the ts1 physical interval (fig. S1).

Sequencing showed that the gene encoding glutamate decarboxylase was monomorphic, whereas the gene encoding lipoxygenase in the ts1-ref line showed an 864–base pair (bp) insertion in the predicted first exon with complete linkage with the ts1 phenotype in mapping populations. To confirm that the lipoxygenase corresponded to the ts1 gene, we analyzed eight ts1 mutant alleles. Each contained an independent mutation in the gene encoding lipoxygenase (Fig. 1A and table S1). Complementary DNA sequence analysis showed that the ts1 gene contains seven exons with a coding sequence of 2757 bp (Fig. 1A). A closely related gene was also identified in the database of the TIGR AZM 4.0 assembly and by Southern blot analysis (fig. S2). This gene, named ts1b, has an identical exon-intron structure to that of ts1 and shares 93% nucleotide similarity. The ts1b gene is located on maize chromosome 10S, a segmental duplication of chromosome 2S (14). The TS1 protein displays 38 to 60% similarity to plant lipoxygenases and contains two conserved domains characteristic of this family: a beta-barrel (cd01751) and a catalytic helical bundle (pfam00305) (15) (Fig. 1B and fig. S3).

Fig. 1.

(A) Structure of the ts1 gene and the ts1 mutant alleles. Hollow boxes at left and right are 5′ and 3′ untranslated regions (UTRs), respectively; black boxes are exons and angled lines are introns. Mutations in eight ts1 mutant alleles are positioned above the corresponding exons. Insertions are represented by inverted triangles and a single deletion by a triangle. (B) TS1 protein features include a predicted chloroplast transit peptide (cTP, green), the PLAT/LH2 beta-barrel (pink), and the lipoxygenase domain (blue) as well as five conserved residues (H, His; I, Ile; N, Asn) necessary for iron binding (red) and the phenylalanine (F) residue predicting 13-LOX regiospecificity (black). (C) Bayesian and maximum parsimony consensus tree of predicted type 2 13-lipoxygenases in angiosperms. The red arrowhead indicates the position of the maize ts1-encoded lipoxygenase. Posterior probabilities from Bayesian inference and bootstrap support from maximum parsimony analysis less than 100% are displayed below internal nodes to the left and right of a slash, respectively. This subclade is part of a more extensive phylogenetic analysis shown in fig. S4.

Lipoxygenases are nonheme iron–containing fatty acid dioxygenases that catalyze the peroxidation of polyunsaturated fatty acids such as linoleic acid, α-linolenic acid, and arachidonic acid. They are classified according to the positional specificity of linoleic acid oxygenation, which occurs at carbon 9 of the hydrocarbon backbone for the 9-LOX types and at carbon 13 for the 13-LOX types; a further subdivision (classes 1 and 2) has been recognized for 13-lipoxygenases without or with a putative chloroplast transit peptide (cTP), respectively (16). According to ChloroP, a neural network–based method for predicting cTPs, the N-terminal 48 amino acids of the TS1 protein contain a cTP (17) (Fig. 1B and fig. S3). Additionally, TS1 contains a conserved phenylalanine (Phe636) previously identified as a determinant of 13-LOX regiospecificity (18, 19). Therefore, the primary structure of TS1 suggests that it is a member of the class 2 plastid-localized 13-lipoxygenases. This prediction was supported by Bayesian and maximum parsimony phylogenetic analyses of plant lipoxygenases, which placed TS1 and TS1b in a clade including characterized and predicted class 2 13-lipoxygenase (Fig. 1C and fig. S4).

The tissue-specific expression of both ts1 and ts1b was established by quantitative reverse transcription polymerase chain reaction (RT-PCR) analysis of root, stem, leaf, tassel, and ear transcripts. The ts1 RNA was detected in all maize tissues examined, whereas ts1b RNA was detected at very low levels (less than that of ts1 by a factor of 90 to 500) (Fig. 2A). The low expression of ts1b mayexplainwhy it does not also appear to be a component of sex determination in ts1 mutant plants. The broad expression of ts1 was unexpected because its mutant phenotype suggests a sex-specific function. Although no alterations in other tissues have been reported, it is possible that additional phenotypes for the ts1 mutation may be uncovered by more careful analyses. The ts2 gene was expressed almost as broadly as ts1, except in stem tissue, where expression was less than that of ts1 by a factor of ∼35.

Fig. 2.

(A) Expression profile of ts1, ts1b, and ts2 in different maize tissues by quantitative RT-PCR on three biological replicates for each tissue. Results are plotted as the ratio to the lowest detected level (ts1b in root) ± SE. Note that the y axis is in logarithmic scale. (B to E) RNA in situ hybridization targeting the 3′UTR of ts1 (dark purple) in developing inflorescences. Scale bars, 200 μm. [(B) and (C)] Wild-type heterozygote male inflorescences (tassels) of 1.6 and 1.5 cm, respectively. (D) Wild-type female inflorescence (ear) of 1.5 cm. (E) Homozygous ts1-Mu01 deletion mutant tassel showing no hybridization signal. (F to I) Colocalization of TS1:mCherry and RbcSnt:GFP fusion proteins in plastids of transfected onion epidermal cells. Scale bars, 50 μm. (F) TS1:mCherry red fluorescence. (G) RbcSnt:GFP green fluorescence. (H) Merge of Ts1:mCherry and RbcSnt:GFP plus two additional channels: 4′,6′-diamidino-2-phenylindole (blue fluorescence) for distinguishing nuclei, and differential interference contrast (DIC) for displaying cellular morphology. (I) Scatterplot of pixel gray value frequencies for RbcSnt:GFP (x axis) and Ts1: mCherry (y axis) channels. Frequencies are displayed using a rainbow lookup table (bottom, units between 0 and 255). Region 3 (upper right) contains pixels with signal above background in both channels, and a linear correlation in this region is a qualitative indicator of colocalization (12).

In situ hybridization showed that TS1 transcripts form stripes following the borders of the central inflorescence axis and projecting toward the spikelet attachment points (Fig. 2, B and D). In spikelet adaxial views, TS1 expression domains surround the spikelets, delineating their base (Fig. 2C). None of these expression patterns were observed in a homozygous ts1-Mu01 deletion mutant line containing a functional ts1b gene (Fig. 2E). These observations indicate that TS1 transcripts subtend maize spikelets at their junction with the central inflorescence axis (rachis). This expression domain suggests a function for TS1, as metabolites synthesized through the lipoxygenase encoded by ts1 could act as diffusible signals affecting floral development in a non–cell-autonomous fashion.

The ChloroP-based prediction that TS1 localizes in plastids was confirmed with a fluorescent-tagged TS1 protein (TS1:mCherry) and a plastid-localized RbcSnt:GFP protein (20). Both fluorescent fusion proteins colocalized in a punctuated pattern throughout the cytoplasm of transfected cells (Fig. 2, F to I), indicating that TS1:mCherry protein is targeted to the same subcellular compartment as RbcSnt:GFP. Therefore, we conclude that TS1 is targeted to plant plastids. Biochemical analysis of protein extracts from developing tassels for activity on the lipoxygenase substrate linoleic acid suggests that TS1 is capable of lipid peroxidation. Protein extracts from wild-type tassels catalyzed hydroperoxidation of linoleic acid, whereas no such activity was detected in mutant ts1-ref/ts-1ref developing tassels (Fig. 3A). Mass spectrometry (MS) and high-performance liquid chromatography (HPLC) analyses showed that the products of this lipoxygenase activity are a mixture of 9- and 13-hydroperoxides in a 50:50 ratio (Fig. 3A), a somewhat unexpected result given that TS1 was predicted to be a lipoxygenase with 13-regiospecificity. Therefore, it is possible that TS1 possesses dual 9- and 13-regiospecificity—which has not previously been described for a plastid-localized lipoxygenase—or that TS1 function promotes the action of a separate 9-lipoxygenase.

Fig. 3.

(A) Partial gas chromatography–MS chromatograms displaying linoleic acid oxidation products generated by crude extracts of wild-type W22 tassels (blue line) but not ts1-ref tassels (red line). HPLC analysis of oxidation products (inset) indicated that the lipid hydroperoxide (HOD) peak was a mixture of 9-hydroxy-10,12-octadecadienoic acid (9-HOD) and 13-hydroxy-9,11-octadecadienoic acid (13-HOD). (B) Box plots summarizing the distribution of jasmonic acid in three tassel sets. Gray circles represent individual measurements. Blue diamonds show the 95% confidence interval of the mean (horizontal blue line). +/+ corresponds to inbred line W22. (C) Blank-treated mutant ts1 tassel. (D) JA-treated ts1 tassel. (E) JA-treated ts2 tassel.

Class 2 13-lipoxygenases participate in the biosynthesis of the plant hormone jasmonic acid (JA) (21) (Fig. 4). We tested whether TS1 is involved in JA biosynthesis by measuring endogenous JA levels in developing wild-type and ts1-ref/ts1-ref mutant tassels. The average concentration of JA in wild-type and ts1-ref/+ heterozygotes was 44.2 ± 13.9 ng per gram of fresh weight (ng/g FW) and 40.3 ± 20.2 ng/g FW, respectively (Fig. 3B). Homozygous ts1-ref/ts1-ref tassels showed an average JA concentration of 4.3 ± 2.1 ng/g FW (Fig. 3B), significantly below that of the wild type in a Kruskal-Wallis test and pairwise comparisons with a Bonferroni correction (P < 0.0001). Therefore, we estimate that the ts1 mutation reduces JA levels by a factor of ∼10, indicating a role for the hormone in the pistil cell death process. Note that JA levels of wild-type and mutant ts1 tassels are similar to those of wounded and nonwounded maize seedlings, respectively (22), which supports the notion that JA is actively synthesized during normal tassel development.

Fig. 4.

Biosynthesis of jasmonic acid through the octadecanoid pathway [adapted from (21)]. The first dedicated step in jasmonate biosynthesis is the peroxidation of α-linolenic acid (18:3) by 13-lipoxygenase to form (13S)-hydroperoxyoctadecatrienoic acid (13-HPOT). This is the putative function of TS1. 13-HPOT is transformed into the specific stereoisomer cis-(+)-12-oxophytodienoic acid (OPDA) through the sequential action of allene oxide synthase [yielding (13S)-12,13-epoxy-octadecatrienoic acid (12,13-EOT)] and allene oxide cyclase. These steps in JA biosynthesis occur in plant plastids (green box), where the corresponding enzymes are localized. Subsequent reactions occur in the peroxisomes (red box). First, the cyclopentenone ring of OPDA is reduced to 12-oxophytoenoic acid (OPC-8) by OPDA reductase. Next, three β-oxidation cycles are proposed to shorten the carboxylic side chain of OPC-8 to produce the 12-carbon JA (21, 24). A β-oxidation cycle is a set of four enzymatic reactions: oxidation, hydration, oxidation, and thiolysis. Not all enzymes acting on β-oxidation during JA biosynthesis have been identified. Because the oxidation in the third step is normally performed by a dehydrogenase activity, it is possible that TS2 may participate in this step of JA biosynthesis.

The developmental timing of the pistil abortion process occurs in tassel inflorescences when they are 1.0 to 3.0 cm in length (23), which was also the stage at which ts1 expression occurred in the subtending glumes (Fig. 2, B to D). We applied 0.005% ethanol with or without 1 mM JA to tassels of ∼1 cm in wild-type (ts1-ref/+) or ts1 mutant (ts1-ref/ts1-ref) sibling plants (12). In ts1 mutant plants, JA application reversed feminization, as evidenced by the presence of staminate spikelets mostly in the mid- to apical regions (Fig. 3D). Wild-type rescue in ts1 mutants was observed in the appearance of subtending floral bracts (glumes) about 3 to 4 weeks after treatment. JA-treated glumes in ts1 mutants were elongated, were covered with numerous trichomes, and had a ring of anthocyanin deposited at the base (Fig. 3D), all three characteristics of wild-type staminate spikelets. Later in floral development, stamens emerged from JA-treated ts1 mutant spikelets (fig. S5B).

Staminate florets from rescued JA-treated ts1/ts1 plants produced viable pollen, which was used for both self-pollination and test crosses to untreated ts1/ts1 mutant sibs. All test cross progeny (n > 100) were homozygous for the ts1-ref allele and displayed a complete ts1 mutant phenotype. The JA-rescued phenotype of the tassel inflorescence was incomplete in that some spikelets were bisexual (fig. S5A), containing both pistils and stamens, and others (mainly those located at the base of the inflorescence) were pistillate. These effects, however, may have been due to the timing of JA treatments, because the stage of floral maturation differs in a positionally dependent fashion throughout the inflorescence. Rescued staminate spikelets were never observed in blank-treated ts1-ref/ts1-ref plants (Fig. 3C, fig. S5D, and table S2), nor did JA treatment affect heterozygous ts1-ref/+ sibs (table S2). The similar phenotype of ts1 and ts2 mutations indicates that both genes may act in the same metabolic pathway. Therefore, we also applied JA to mutant ts2-ref/ts2-ref and ts2-ref/+ plants, which responded in the same manner as the JA-treated ts1 mutants (Fig. 3E, fig. S5C, and table S2). These results indicate that JA can restore the wild-type phenotype in both ts1 and ts2 mutant plants. Moreover, this suggests an unexpected role for TS2 in JA biosynthesis, perhaps as one of the yet-unidentified enzymes catalyzing a series of β-oxidations in this metabolic pathway (Fig. 4) (24).

Genes regulating meristem determinacy early in maize inflorescence development are expressed at the boundary of the meristem and the inflorescence axis rather than within the meristem itself. These genes, such as ramosa1, ramosa3, and barren stalk1 (2527), probably act non–cell-autonomously by producing a diffusible signal at the base of the meristem (27). Analogously, ts1 expression at the boundary of developing spikelet initials and inflorescence axis produce the hormone JA, which may diffuse within the spikelet to regulate sexual development. This situation parallels JA-mediated anther dehiscence in Arabidopsis, where JA biosynthetic genes are highly expressed in the anther filament where it signals development both in the filament and within the anther (28, 29). We previously reported that the expression of ts2 is reduced in ts1 mutants (8). The finding that ts2 may be involved in the same biosynthetic pathway as ts1 is not necessarily at odds with our previous observation, as most genes encoding enzymes of the JA biosynthetic pathway are transcriptionally up-regulated by JA in a characteristic positive feedback loop (21).

JA signals plant responses to biotic and abiotic stresses (21) and regulates plant developmental processes such as root growth (30) and mechanotransduction (31). In Arabidopsis, JA is required for male fertility because pollen maturation and anther dehiscence are blocked in mutations that impair JA biosynthesis (28, 29). JA may promote anther dehiscence by signaling degeneration of the stomium, a group of specialized cells that run along the length of the anther and are necessary for dehiscence (32, 33).

The plant hormone ethylene has been observed to promote feminization in cucumber [reviewed in (34)]. Recent genetic and biochemical evidence has confirmed the role of ethylene in sex determination of melon, a related species (35). Conversely, gibberellin has masculinizing effects in cucumber but promotes feminization in maize (36), and auxin also has opposing effects in cucumber and Mercurialis annua (34). The present results imply a role for JA in maize sex determination, wherein JA is necessary for signaling the tasselseed-mediated pistil abortion and the acquisition of the male characteristics of staminate spikelets. The diverse mechanisms of hormonal control in plant sex determination support the notion that the systems have evolved independently multiple times (37).

Supporting Online Material

www.sciencemag.org/cgi/content/full/323/5911/262/DC1

Materials and Methods

SOM Text

Figs. S1 to S5

Tables S1 to S5

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article