The role of SEUSS in auxin response and floral organ patterning

Among plant hormones, auxins are well known for their widespread role in growth and development. The most biologically active of these simple molecules is indole-acetic acid (IAA), a significant proportion of which is synthesized in young actively dividing cells in the aerial part of the plant. IAA is transported in a polar fashion throughout the plant to affect a variety of morphological processes (Davies,1995). At the cellular level, auxin regulates cell division and elongation, which explains its essential role in developing tissues(Sachs, 1991). Auxin contributes to such fundamental processes as embryonic axis establishment,organization of the root meristem and subsequent root elongation, and lateral organ outgrowth in both the root and the shoot(Friml et al., 2003; Jiang and Feldman, 2002; Casimiro et al., 2003; Reinhardt et al., 2000). Auxin continues to influence development throughout the life of the plant. During flowering, it promotes organ outgrowth and development of the gynoecium, the female reproductive organ of angiosperms(Vernoux et al., 2000; Nemhauser, 2000; Okada et al., 1991; Bennett et al., 1995).

The Arabidopsis flower is particularly well suited to study the effects of auxin on plant development because the flower arises in a very stereotypical pattern, resulting in four types of organs that are morphologically distinguishable from one another. Each organ type develops in a concentric whorl: four evenly spaced sepals envelop and protect the developing internal organs; four petals arise just inside the sepal margins;six stamens are patterned as two lateral and two pairs of medial organs; and two fused carpels collectively comprise the central gynoecium. The conceptual framework for flower development has been the elegantly simple ABC model, in which three classes of genes are expressed in concentric overlapping domains that combinatorially specify organ identity in each of the four floral whorls(Bowman et al., 1991). Although identification of genes involved in meristem function and organ identity has increased exponentially (Carles and Fletcher, 2003; Franks and Liu, 2001), relatively little is known about how organs are patterned both within and between whorls. Transcription factors such as SUPERMAN and CUP-SHAPED COTYLEDON2 function in whorl and/or organ partitioning(Aida et al., 1997; Sakai et al., 1995), while UNUSUAL FLORAL ORGANS and PERIANTHIA affect initiation and spacing of floral organs (Levin and Meyerowitz,1995; Running and Meyerowitz,1996). Yet how floral organs are initiated in such specific positions and in such a reproducible pattern is still not well understood.

The ettin (ett) mutant has pleiotropic effects on flower development, including increases in perianth organ number and aberrations in regional differentiation of reproductive organs. In ett, development of the gynoecium has been the best characterized. The gynoecium in the Brassicaceae, including Arabidopsis, develops from two congenitally fused carpels that arise from the center of the floral meristem. The two outer ovary walls, or valves, are separated from each other by a longitudinal furrow, or replum. In ett, the organization of apical style and stigma, central ovary and basal internode (termed stipe), is disrupted. Strong alleles of ett are characterized by a shift in boundaries between these tissues, including a reduction in valve length, an increase in basal stipe, and an overproliferation of apical stigma and style(Sessions and Zambryski,1995). There is also a concomitant eversion or abaxialization of internal tissues, the severity of which increases towards the gynoecium apex. The reduction in the proportion of ovary in ett is reminiscent of the phenotypes of the auxin regulatory mutants pinoid (pid) and monopteros (mp), and therefore similarly may be due to a disruption in auxin signaling (Bennett et al., 1995; Przemeck et al.,1996). In fact, transient application of naphthylphthalamic acid(NPA), a polar auxin transport inhibitor, to the inflorescence apex of wild-type plants results in a reduction of ovary relative to apical style and basal stipe, effectively phenocopying ett and weak alleles of pid. Because NPA presumably causes auxin to pool at its sites of synthesis instead of being transported in a normal polar fashion, it has been hypothesized that ETT responds to a gradient of auxin to establish ovary boundaries within the gynoecium (Nemhauser et al., 2000).

ETT is a member of the auxin response factor (ARF) family of transcription factors (Sessions et al.,1997). This family is central to auxin response, as almost immediately after auxin entry into a cell, ARFs activate transcription of early auxin response genes (Ulmasov et al., 1999a). The ARF family is difficult to study genetically,owing to a high level of sequence homology and probable functional redundancy. Only three out of 22 ARFs from Arabidopsis have demonstrated loss-of-function phenotypes (Guilfoyle and Hagen, 2001). ETT is unique among ARFs because it lacks two C-terminal domains responsible for heterodimerization with Aux/IAAs, a family of transcriptional repressor proteins. Instead, ETT has a unique C-terminal half, containing a region rich in serine residues. Thus, the prevailing model for auxin action – in which Aux/IAA repressors are rapidly targeted for degradation when auxin is present, liberating ARFs to activate transcription of target genes – cannot hold for ETT. Yet ETT is clearly involved in auxin response in the gynoecium, as NPA treatment of wild type gynoecia phenocopies ett. In addition, auxin may have an as yet undescribed role in outer whorl development, because the auxin response mutant ett, the auxin signaling mutant pid, and the auxin transport mutant pin-formed1 (pin1) all alter floral organ numbers(Bennett et al., 1995; Sessions, 1997).

To understand more about how ETT responds to auxin to influence initiation and patterning of floral organs, modifiers of the ett phenotype were identified. One such modifier was found to carry a mutation in the SEUSS (SEU) gene, which encodes a putative transcriptional co-regulator of the floral homeotic gene AGAMOUS (AG)(Franks et al., 2002). We describe the effects of seu on ett and extend genetic and morphological studies using the novel seu-3 allele. We demonstrate that seu confers auxin resistance to the root and disrupts flower development in combination with the auxin response mutants ett and pid. We propose a model in which SEU acts in concert with ETT to promote floral organ development by transcriptionally regulating auxin-responsive genes.

Seeds of seu-1/+ ag-1/+ are courtesy of Bob Franks. The following seeds have been deposited at the ABRC: seu-3 in Col-0; seu-3in Ler-0; ett-7 in Col-0; ett-11 in Ler-0. seu-3was originally identified by Allen Sessions and Judy Roe as s27 from an EMS-mutagenized population of Col seeds. seu-4 is a T-DNA insertion line from the Salk Institute Genomic Analysis Laboratory; the ABRC stock number is SALK_069303. seu-3 CAPS genotype markers can be generated by PCR with forward primer GAAAATGTTCCGCCTTCGAT (+40 bp past the start of translation in SEU) and reverse primer GAATTTGCTGCGGTTCCAACT(beginning at SEU +619). BslI or PflMI cleaves wild-type DNA to yield 235 and 345 bp fragments, whereas in seu-3 the uncleaved product is 580 bp. Alternatively, BsrI can be used to generate wild-type fragments of 241 and 339 bp. The seu-4 allele can be identified with forward primer TGGACCGCTTGCTGCAACT (235 bp inside the T-DNA left border of pROK2, the vector used to generate the SALK lines) and reverse primer CCGGCATTTGAAGAAATTGT (at SEU +742), which generate a 264 bp PCR product. T-DNA left border (110 bp) is missing in the seu-4allele, and there is 8 bp of homology at the insertion point between the left border and SEU.

Plants grown on plates were in a growth chamber with 24 hours light at 22°C. Culture media consisted of half-strength MS salts, 1% sucrose and 0.8% bacto-agar at pH 5.8. For naphthaleneacetic acid (NAA) experiments, NAA was dissolved in 70% ethanol and added to the medium before autoclaving, or in 1 M NaOH and added after autoclaving. Plants grown in soil were sown on Metromix in 7.6 cm² pots and grown at a 1-4 per pot density under long-day greenhouse conditions.

seu and wild-type seeds were grown on vertical plates. At 1-2 days after planting (dap), seedlings were examined for presence of an emerged radicle, and these seedlings were transplanted at 3 dap to treatment plates(100 mm²). seu and wild-type seedlings were grown on the same treatment plate. For lateral root counts, roots were counted at 10 dap. For measurements of primary root length, digital photos were taken at 8 dap and roots measured in ImageJ (public domain software available from NIH at http://rsb.info.nih.gov/ij/). The segmented line tool was used to trace roots, and the measure function was used to determine root length. For root reorientation assays, plates were rotated 90° with respect to gravity and photos were taken 24 hours later. Root angles were calculated in Adobe Photoshop using the measure tool.

β-Glucuronidase staining was performed using 2 mM X-gluc substrate from Rose Scientific, 2 mM potassium ferrocyanide, and 2 mM potassium ferricyanide, at room temperature from 3 hours to overnight(Weigel and Glazebrook,2002).

Yeast two hybrid protocols are from Clontech. SEU prey is in pGAD424, ETT bait is in pBD, ARF1 bait is in pGBT9 and UFO bait is in pAS1. pBD and pGAD424 were used as empty vector controls. All assays were performed in yeast strain YD116.

The seu-3 mutant characterized in this study was originally identified as s27 in a screen for floral mutants (R. A. Sessions, PhD Thesis, University of California, Berkeley, 1996). Because of its synergistic double mutant phenotype with ett (see below), map-based cloning was begun. s27 mapped close to GAPB on chromosome I, near the SEU gene (Franks et al.,2002). Based on the comparable gynoecium defects in both mutants,we tested allelism, and all descendants from the selfed progeny of a complementation cross had a seu-like phenotype. Thus, s27was renamed seu-3. The SEU gene in seu-3 was sequenced and a C-to-T transition was found at amino acid 127 in the first exon, causing a mutation from glutamine to a stop codon. A fourth allele, seu-4, was later found in the SALK lines, with a T-DNA inserted into codon 199 (Alonso et al.,2003). seu-3 and seu-4 appear to be strong alleles, while seu-1 and seu-2(Franks et al., 2002) are weaker in phenotype.

seu plants have decreased apical dominance and appear shorter overall (Franks et al., 2002). Compared with wild-type flowers, seu-3 flowers are reduced in size because of smaller petals and stamens (Fig. 1A,B). The most noticeable aspect of seu-3 flower morphology is the apical cleft in the gynoecium. The stigma and style are split so that two balls of stigmatic tissue are usually seen. Often, the distal valve tips splay out where the style divides in two, resembling small horns. Occasionally these horns have stigmatic papillae at their tips(Fig. 1C). The lack of fusion rarely extends far into the ovary. seu-3 flowers are semi-fertile and produce few and irregular numbers of seeds when selfed. This is probably due to both male and female fertility defects, as reciprocal outcrosses produce low numbers of seeds (data not shown).

There are significant differences between seu alleles in different backgrounds. seu-3 and seu-4 are in the Columbia (Col)background, while seu-1 and seu-2 are in the Landsberg erecta (Ler) background. Unlike seu-1 and seu-2, seu-3 does not show homeotic transformations of the outer whorl from sepals to petals, stamens or carpels. Yet when seu-3 is introgressed into Ler, it displays more frequent homeotic transformations than does seu-1 (not shown). In addition, seu- 3 in Ler has reduced numbers of petals and stamens(like seu-1), while seu-3 in Col has close to wild-type outer whorl organ numbers (Table 1). Finally, fertility of strong seu alleles is higher in Ler than in Col. This may result from greater pollen production and a less substantial split in the gynoecium in Ler, as measured by the frequency of horn-like valves.

ett flowers have severe gynoecium defects. Apical and basal valve boundaries are displaced towards each other so that overall ovary size is reduced, and sometimes ovaries are completely absent(Fig. 1H). Internal tissue is everted at the apex, revealing internal style and transmitting tract tissue and occasionally exposing ovules.

Flowers of the ett-7 seu-3 double mutant are markedly different from flowers of either single mutant. Mature flowers are half the size of single mutant flowers (Fig. 1E,I,J). The gynoecium is most often reduced to a stalk-like,ovary-less gynophore topped with stigmatic tissue, although some gynoecia retain a seu-like irregular tip(Fig. 1I). There is no eversion of internal transmitting tract or ectopic stigma, as in ett single mutants. ett-7 seu-3 intermittently develops a large ovary(Fig. 1J), which is often split open to reveal malformed ovules. In those few double mutants that develop ovaries, valve length is greater than half the length of the gynoecium 99% of the time, whereas in ett-7 single mutants, only 16% of the carpels have valves greater than half the length of the gynoecium.

ett affects outer whorl organs such that sepal and petal numbers are increased from four in wild type to an average of five in ett-7(Fig. 1D, Table 1). The outer whorls are also affected in ett-7 seu-3 compared with each single mutant, and this phenotype becomes more severe acropetally. Remaining petals are variably narrowed with most being reduced to mere filaments(Fig. 1E,I). Some sepals are also filamentous, and others are wrinkled and reduced in size (arrowhead in Fig. 1I). Stamens are generally stunted and/or withered. In addition, ∼15% of flowers have a stamen fused to the length of the gynoecium (Fig. 1J). Organ numbers are more variable in whorls one and two, and reduced in whorl 3 (Table 1)compared with organ numbers in ett-7 or seu-3 single mutants. Phyllotaxy in the outer two whorls is disrupted and organ spacing becomes increasingly disorganized in later arising flowers, so that it becomes difficult to distinguish whorl 1 from whorl 2. A few sepals and petals are recognizable, but most organs have aberrant morphologies. These organs sometimes appear to develop from the same (merged) outer organ whorl.

Disparities in ett-7 seu-3 floral organ position, shape and size can be seen at primordia inception (Fig. 2). In wild type, sepals originate from the floral meristem as four distinct bulges, which overlie the meristem by stage 6(Fig. 2A). seu-3sepals develop similarly, with initiation of four discrete sepal primordia occurring at stage 3 (Fig. 2B). The sepals in the lateral plane arise slightly later than the sepals in the medial plane and are therefore smaller (arrows versus arrowheads in Fig. 2A,B). Beginning at stage 4, a modest difference can be seen between seu-3 and wild-type flower development. seu-3 sepals appear smaller and often do not completely enclose the bud, leaving some of the developing meristem exposed (compare asterisks in Fig. 2A with 2B). In ett-7, although there are five sepal primordia, they are clearly distinct from each other as with wild-type sepals(Fig. 2C).

Sepal initiation in the ett-7 seu-3 double mutant is noticeably different than in either single mutant. Sepals arise as one uneven ring of tissue, with no distinct boundaries between individual primordia(Fig. 2D,E). Observation of sepal primordia at stages 5-6 often provide an inflated impression of sepal number, because some of these primordia cease to develop and others appear congenitally fused in their basal halves. The ring of sepal tissue eventually grows out erratically to produce some mature organs. Petal and stamen primordia are difficult to identify because of their aberrant positioning. Some stamen primordia appear incompletely separated from the meristem proper(not visible in Fig. 2), which may portend the high frequency of mature stamens fused with the gynoecium. Together, these observations demonstrate that seu-3 and ett-7 together affect morphology and positioning of floral organs,including sepal partitioning, petal shape and stamen number.

seu causes ectopic and precocious AG expression in whorls one and two, leading to partial homeotic transformations of sepals into mosaic carpelloid, stamenoid or petalloid organs(Franks et al., 2002). AG is a MADS-box transcription factor responsible for stamen and carpel identity, as well as for determinacy of the floral meristem(Bowman et al., 1989; Yanofsky et al., 1990). This latter growth suppression role is supported not only by the indeterminate flower in the ag mutant, but also by evidence that ectopic expression of AG in whorl 2 under the AP3 promoter inhibits growth of whorl 2 petals (Jack et al.,1997). If AG expression suppresses petal growth and affects floral organ identity, the ectopic AG present in seucould be responsible for the filamentous organ phenotype in the ett seu double mutant.

The null ag-1 allele produces flowers with petals and sepals in place of stamens and carpels (Fig. 3A). In addition, the indeterminate floral meristem produces an indefinite number of additional organ whorls as (sepal, petal,petal)ⁿ. The seu-1 mutant in Ler occasionally has a mosaic sepal, owing to the ectopic expression of AG(Fig. 3B). When ag-1is crossed with seu-1, AG protein activity is removed and sepals no longer exhibit homeotic transformations(Franks et al., 2002). An ag-1 seu-1 double mutant retains the organ identity and indeterminacy of ag-1 single mutant flowers, causing reiterating whorls of sepals,petals, petals (Fig. 3C). An ag-1 ett double mutant also resembles ag-1 in its floral organ identity and floral meristem indeterminacy(Sessions et al., 1997). Like the double mutants, an ag-1 ett-11 seu-1 triple mutant produces a repeating pattern of sepals, petals, petals(Fig. 3D). However, both sepals and petals are variably narrowed, reminiscent of the ett seu double mutant phenotype. Each flower contains recognizable sepals and petals in its initial whorls, but as the floral meristem develops acropetally, the organs become increasingly filamentous. Floral meristems occasionally continue to elongate without producing lateral organs(Fig. 3D, arrowhead). Determination of organ numbers in these ag ett seu mutants is difficult because organs are not evenly spaced around the floral meristem and individual whorls cannot be distinguished from each other. Nevertheless, it is clear that removing ectopic and precocious AG from ett seu by generating ag ett seu does not rescue the filamentous, mispositioned organ phenotype present in the ett seu double mutant.

Owing to the synergistic phenotype between seu-3 and ett-7, we reasoned that SEU, like ETT, may have a role in auxin response. In support, seu plants are shorter with increased shoot branching, and after 5 weeks seu-3 develops an average of six lateral inflorescences from the rosette, instead of the three lateral inflorescences typically found in wild type(Fig. 4A). Such decreased apical dominance is commonly found in auxin resistant mutants such as axr1 (Lincoln et al.,1990). Because many auxin signaling mutants also have root defects, Arabidopsis roots have become a well-characterized system with which to study auxin signaling. To determine whether seu-3affects auxin response, we analyzed seu-3 roots for three archetypal auxin responses: lateral root initiation, resistance to exogenous auxin and gravitropic response.

seu-3 possesses fewer lateral roots than does wild type(Fig. 4B). Some seu-3seedlings have no lateral roots, while most seu-3 seedlings have half of the number found in wild-type roots(Fig. 4B). Close examination of these seu-3 roots indicates fewer points of lateral root initiation from the pericycle, resulting in fewer mature lateral roots. When exogenous auxin in the form of naphthaleneacetic acid (NAA) is added to growth medium,primary root elongation is inhibited in wild-type plants. Cotyledons also grow downward (epinasty) owing to differential elongation of adaxial and abaxial surfaces. In the presence of NAA, seu-3 roots are inhibited less and therefore elongate more than wild-type roots(Fig. 4C,D). Moreover, seu-3 cotyledons are not epinastic like those of wild type(Fig. 4E,F). Therefore, seu-3 is less sensitive to the inhibitory effects of NAA. To determine the effect of the seu-3 mutation on root gravitropsim, root tip angles were measured 24 hours after rotating the vertically grown seedlings 90°. While wild-type roots are consistently aligned to the gravity vector, seu-3 roots are less likely to orient properly(Fig. 4G). These phenotypes– decreased apical dominance, fewer lateral roots, reduced sensitivity to exogenous auxin in both the root and the shoot, and more variable growth vector in the root – suggest that seu has a decreased response to auxin.

To further test the hypothesis that seu displays diminished auxin response, we examined expression of the synthetic auxin response reporter DR5::GUS (Ulmasov et al.,1997). seu-3 shows a clear decrease in DR5 expression in both the root and the shoot. In wild type, DR5 expression in the root meristem extends into the root cap and stele, whereas seu-3 roots have decreased DR5 expression in the root cap and no expression in the developing vascular tissue (Fig. 5A,B). The attenuation of DR5 expression in seu-3 is even more apparent in the shoot. DR5 expression in 5-day-old wild-type seedlings is most intense at the distal leaf tip, where auxin is produced, and also is present throughout the leaf in incipient secondary and tertiary veins(Fig. 5C). DR5 expression in seu-3 seedlings, however, occurs only occasionally at the leaf tip and in hydathodes (Fig. 5D). In 7-day-old wild-type seedlings, DR5 expression occurs in the basal half of the leaf, paralleling the basipetal differentiation of vascular tissue(Fig. 5E). seu-3seedlings never show DR5 expression in these procambial sites of future vein formation (Fig. 5F). That sites of maximum auxin response in both the root and the leaves are diminished in seu-3 supports a role for SEU in auxin signaling.

The serine threonine kinase PINOID (PID) has been postulated to affect polar auxin transport as well as auxin signaling(Benjamins et al., 2001; Christensen et al., 2000). Because the pleiotropic effects of pid on flower development are presumed to be auxin related, we examined pid seu double mutants for effects of seu on the pid floral phenotype. pidinflorescences produce only a few flowers before terminating in a pin-like structure reminiscent of the pin1 auxin transport mutant(Bennett et al., 1995; Okada et al., 1991). pid flowers have larger, supernumerary petals but fewer sepals and stamens. Organs within each whorl are often fused together, frequently resulting in heart-shaped petals (Fig. 6A). The central whorl of pid flowers phenocopies ett gynoecia in that valve length is variably reduced, decreasing the size of the ovaries (Fig. 6B).

pid-1 seu-3 double mutants exhibit a strikingly synergistic phenotype. Very few outer whorl organs develop in pid-1 seu-3flowers, resulting in clusters of naked gynoecia(Fig. 6C,D). While developing gynoecia are normally enveloped by sepals in wild-type flowers, pid-1 seu-3 flowers rarely contain a wrinkled sepal, a filamentous petal or stunted stamens (Fig. 6C-F). More often the remnants of the floral meristem appear as a swollen ring at the base of the gynoecium, without evidence of distinct primordia(Fig. 6C,D). Although half-formed sepals occasionally develop from this region, there is no discernible boundary between the swollen area and the pedicel(Fig. 6D). Lack of pedicel elongation may contribute to that region becoming overly bulbous. In fact,double mutant plants are shorter and internode length is reduced compared with those features in either single mutant (not shown).

The pid-1 seu-3 double mutant gynoecium displays characteristics of both pid-1 and seu-3 mutants. Like pid-1, the central whorl of pid-1 seu-3 most often develops as a long gynophore topped with stigmatic tissue (Fig. 6E). The gynoecium is frequently missing major tissues, such as valves, and often is split into multiple stalks, as if carpel fusion did not occur normally (Fig. 6F).

Evaluation of organ numbers confirms the strong interaction between pid-1 and seu-3 (Fig. 7). Wild-type Ler flowers typically produce four sepals,four petals, six stamens and two carpels. When seu-3 is in the Ler background, petal and stamen numbers are decreased slightly, with a mean of 3.1 petals and 5.0 stamens. Sepal and carpel numbers in seu-3 are similar to wild type. In pid-1, variability in organ number increases, with a median of three sepals, six petals, three stamens and no valves. The double mutant produces a mean of 0.5 sepals, 0.5 petals, 0.5 stamens and 0.1 valves, with median numbers of zero for each organ type. Those few flowers that generate organs tend to be the first few in an inflorescence, with the lack of organ development increasing in later-formed flowers. This drastic reduction in organ numbers in pid-1 seu-3suggests a synergistic interaction between these two mutants, especially in whorl two, where the increased number of petals in pid-1 is reduced to almost none in pid-1 seu-3.

SEU is postulated to be a transcriptional co-regulator based on (1) its repression of AG in the outer whorls of the floral meristem, and (2)its sequence similarity with transcriptional co-factors such as Ldb1 in mouse and Chip in Drosophila (Franks et al., 2002). Given the identity of SEU as a putative transcriptional co-regulator and ETT as an ARF transcription factor, we tested whether SEU and ETT directly interact. Results from a yeast two hybrid assay suggest that SEU and ETT physically interact(Fig. 8). When ETT is expressed in a bait vector (ETT-BD) and SEU is expressed in a prey vector (SEU-AD),yeast grow robustly with uracil selection for the interaction. Yeast do not grow with ETT-BD and an empty prey vector (AD), or with SEU-AD and an empty bait vector (BD). SEU-AD does not interact non-specifically with all prey, as shown by lack of growth of yeast containing SEU-AD and UFO-BD.

To determine whether the interaction between SEU and ETT is specific to ETT/ARF3 among ARFs, growth of yeast expressing SEU-AD and ARF1 in a bait vector (ARF1-BD) was also examined. Yeast expressing ARF1-BD and SEU-AD show no growth at day 3, similar to ARF1-BD and AD. Thus, SEU interacts with ETT,but not with ARF1 in a yeast two hybrid system.

This work describes new alleles of the seu mutant along with morphological, physiological and genetic evidence of a role for SEUin auxin-regulated growth and development. seu displays several classic auxin response defects, such as reduced apical dominance, decreased lateral root initiation, resistance to exogenous auxin and diminished expression of the auxin response reporter DR5. This collectively suggests that SEU functions in the auxin signaling pathway. seu interacts synergistically with the auxin response pathway mutants ett and pid. In addition, SEU interacts molecularly with the auxin response factor ETT/ARF3, which suggests that SEU may regulate transcription of auxin response genes in concert with ETT.

seu-3 was originally identified as a modifier of the ettfloral mutant. ETT affects multiple patterning events during flower development, including number and spacing of perianth organs and regional differentiation of reproductive organs(Sessions et al., 1997). ETT function has been best characterized in the gynoecium because ett ovaries are severely reduced in size, with the missing ovary tissue replaced by abaxialized style and transmitting tract tissues. In the ett seu double mutant, the few ovaries that are produced are longer,making ett appear suppressed in the central whorl. SEUpromotes development of marginal tissue between the valves that gives rise to style, transmitting tract and ovules (R. G. Franks, personal communication). The partial suppression of ett may be explained by the lack of development of these tissues in ett-7 seu-3, which normally overproliferate in ett.

In the outer whorl organs, the interaction of seu and ettappears synergistic. In seu, sepals and petals are narrower and stamens produce little pollen. In ett, there is a slight increase in number of sepals and petals, affecting the spacing between organs. The ett seu double mutant exhibits a more severe phenotype, with substantial decrease in stamen number and filamentous mispositioned perianth organs. The spacing between organs is disrupted not only within whorls but between whorls,so that stamens are occasionally fused with the gynoecium. This suggests that SEU and ETT participate in convergent pathways in the floral meristem for positioning and growth of floral organs. It is not known whether SEU and ETT expression patterns overlap. By northern analysis, SEU is expressed at low levels in all tissues analyzed(Franks et al., 2002). ETT is expressed in incipient floral meristems, but later expression is restricted to vascular bundles of petals and stamens and the abaxial side of the gynoecium where valves and replum differentiate(Sessions et al., 1997).

Double mutant analysis of ett with homeotic mutants involved in the ABC model of floral organ development established that ETT does not play a role in floral organ identity(Sessions et al., 1997). By contrast, SEU causes ectopic and precocious expression of the MADS box transcription factor AGAMOUS, responsible for stamen and carpel identity (Franks et al.,2002). Removing this misexpressed AG from the outer floral whorls by generation of an ag-1 ett-11 seu-1 triple mutant does not affect the filamentous, mispositioned organ phenotype caused by ett and seu. Thus, the effects of SEU on ETT are independent of the repression of AG by SEU. These experiments suggest that SEU and ETT act in a distinct pathway affecting floral organ positioning and development.

Several facets of the seu phenotype resemble hallmark defects of auxin response. Auxin is required in the root for organization of the meristem, gravitropic response, primary root elongation and initiation of lateral roots (Sabatini et al.,1999; Moore, 2002; Casimiro et al., 2003). The meristem in seu-3 roots appears to function normally, but after reorientation of vertically grown seedlings, the angle of root growth in seu-3 is more variable than in wild type. seu-3 produces half as many lateral roots as wild type, suggesting that either not enough auxin is present in the root zone where lateral roots initiate, or that the auxin present is not perceived as efficiently as in wild type.

To distinguish between these hypotheses and to confirm that seuaffects auxin response, seu-3 seedlings were grown in the presence of exogenous NAA, a biologically active auxin. seu-3 roots are longer relative to wild-type roots when grown on NAA, indicating that seu-3root growth is less inhibited by exogenous auxin than wild-type roots. That seu-3 is less responsive to NAA treatment than wild type indicates seu is defective in auxin perception or response.

Further insight into the role of SEU in auxin response was gained by examining expression of the auxin response reporter DR5. If seu is unable to transport auxin normally, auxin would pool at sites of synthesis and be depleted in other tissues. This perturbed auxin flux would be reflected in an increase in DR5 expression at the distal leaf tip and a decrease in expression in outlying areas of vascular development. However, in seu, DR5 expression is reduced but not displaced in both developing leaves and the root tip, supporting the hypothesis that seu is defective in perception of/response to auxin within a cell and not in transport of auxin between cells.

The striking abolishment of outer whorl organs in the pid-1 seu-3double mutant demonstrates a strong synergism between seu-3 and pid-1. Because pid is thought to destabilize auxin levels in the inflorescence meristem, the seu-3 pid-1 interaction further implicates SEU in floral auxin response. Based on ectopic expression data, the PID kinase has been postulated to both positively regulate auxin transport and negatively regulate auxin signaling(Benjamins et al., 2001; Christensen et al., 2000). PID probably affects regional auxin distribution in the inflorescence meristem,because the auxin efflux carrier PIN1 is reduced and mislocalized in developing pid meristems(Reinhardt et al., 2003). This disruption in auxin flux, together with the decreased auxin response of seu, may account for the lack of organ primordia development in pid-1 seu-3 double mutant flowers. Those few outer whorl organs that develop in pid-1 seu-3 are stunted and/or filamentous, similar to organs in ett-7 seu-3. The parallel organ phenotypes in these double mutants supports the hypothesis that SEU functions in auxin response,with pid-1 compromising seu floral meristems even more severely than ett, owing to reduced auxin levels as well as reduced auxin response. Transient application of the polar auxin transport inhibitor NPA to wild-type inflorescences also reduces the number of outer whorl floral organs that develop in affected flowers(Nemhauser et al., 2000). By contrast, micro-application of the natural auxin indole-3-acetic acid (IAA)promotes lateral outgrowth of both leaves and flowers(Reinhardt et al., 2000). The rings of meristem tissue that develop in both ett-7 seu-3 and pid-1 seu-3 bear similarity to the rings of flowers and floral organs that develop in the auxin transport mutant pin1 after treatment with IAA at the top of the pin1 inflorescence meristem(Reinhardt et al., 2000). Taken together, the above findings suggest that lack of organ development in pid-1 seu-3 is due to insufficient levels of auxin and deficient response to auxin within organ primordia.

The severity of the pid-1 seu-3 phenotype suggests several possible roles for the PID kinase in the ETT-SEU pathway. Because PID has an AuxRE in its promoter and is auxin-inducible(Benjamins et al., 2001), SEU and ETT may function together to directly affect PID transcription. A second possibility is that PID indirectly affects the SEUpathway by its effect on auxin transport and regional auxin levels(Benjamins et al., 2001; Reinhardt et al., 2003).

The collective auxin response phenotypes of seu may reflect a novel and fundamental role for SEU in auxin-mediated signal transduction. Based on the genetic and physical interactions of SEUwith ETT (Figs 1, 2, 8), we propose that SEU functions with ETT to regulate transcription of auxin response genes involved in floral meristem patterning (Fig. 9).

How might a SEU-ETT transcriptional complex regulate gene transcription?SEU contains two remarkably glutamine-rich domains that may function as transcriptional activation patches(Mitchell and Tjian, 1989; Petcherski and Kimble, 2000). However, SEU represses AG(Franks et al., 2002),suggesting that SEU may mediate activation and/or repression of target genes,depending on the cellular context (Freiman and Tjian, 2002; Mannervik et al., 1999). ETT is thought to function as a repressor based on protoplast transfection assays (Tiwari et al., 2003; Ulmasov et al.,1999b) and ectopic expression of the SPATULA(SPT) transcription factor in whorl 4 primordia of the ettmutant (Heisler et al., 2001). Moreover, ETT lacks the highly conserved ARF domains III and IV that are responsible for dimerization of ARFs and Aux/IAAs, which regulates ARF activity (Dharmasiri et al., 2002). The absence of domains III and IV in ETT,and the distinct C-terminal region of ETT, suggest that ETT acts differently than other ARFs. SEU may contribute to and/or regulate this unique function of ETT.

SEU is homologous in its central region with mouse Ldb1, which homodimerizes and binds LIM-homeodomain transcription factors to form a tetrameric complex (Jurata et al.,1998). SEU may similarly act as a bridging factor between ETT and other regulatory molecules (Fig. 9). These regulatory molecules could include ARFs and/or Aux/IAAs,which would allow regulation of the ETT transcriptional complex by the Aux/IAA family. However, SEU interacts with ETT/ARF3 but not with ARF1, another putative ARF repressor, in a yeast two-hybrid assay(Fig. 8). These results suggest a testable hypothesis that SEU interacts specifically with ETT among ARFs. In this case, ETT and SEU may function together in regions outside the flower, as seu mutants have global auxin response defects, and both SEUand ETT are expressed in a variety of plant tissues(Franks et al., 2002; Sessions et al., 1997; Ulmasov et al., 1999b). The present research demonstrates a role for SEU in auxin-mediated floral organ development. Future studies will clarify the role of SEU in transcriptional regulation of auxin response genes, and will undoubtedly reveal additional novel players in this complex developmental program.

We thank Bob Franks, Paula McSteen, Sheila McCormick and Mimi Tanimoto for their critical reviews of the manuscript. J.P. is immensely grateful to Lew Feldman for support, encouragement and advice; to Bob Franks for stimulating discussions and the generous exchange of seed and DNA stocks; and to Allen Sessions for the initial identification of seu-3. J.P. also thanks undergraduate researchers Lan Quang and Deborah Yang for their careful work. We are grateful to Zhongchi Liu, Anandkumar Surendrarao, Celine Forzani and Sakis Theologis for graciously sharing two-hybrid constructs. We appreciate SIGNAL T-DNA Express for creating seu-4, the ABRC for seu-4and pid-1 seeds, and Tom Guilfoyle for the DR5 reporter. For microscopy support, we thank the Bob Fischer laboratory; Denise Schichnes,Steven Ruzin and the Biological Imaging Facility; and the Electron Microscope Laboratory. This research was supported by grants from the USDA (#35304-09986)and the Torrey Mesa Research Institute at Syngenta.