A gain-of-function mutation in IAA18 alters Arabidopsisembryonic apical patterning

Plant embryos provide a simple context in which to study patterning and differentiation events that are reiterated at later stages. In Arabidopsis, cotyledon placement and outgrowth, flower meristem formation, and flower organ emergence all require the auxin-responsive transcription factor MP/ARF5 and the auxin efflux transporter PIN1, indicating that apical patterning in embryos and adults share mechanisms(Berleth and Jürgens,1993; Hardtke et al.,2004; Okada et al.,1991; Przemeck et al.,1996). Thus, studies of cotyledon formation in embryos may reveal how patterning occurs de novo and provide insight into how organs form postembryonically.

Auxin regulates gene expression by inducing turnover of Aux/IAA proteins. Aux/IAA proteins dimerize with Auxin response factor (ARF) transcription factors through shared C-terminal motifs, and inhibit gene induction through a conserved sequence in Aux/IAA proteins called motif I(Szemenyei et al., 2008; Tiwari et al., 2004; Tiwari et al., 2001). In the presence of auxin, F-box auxin receptors TIR1 and AFB1-AFB3 (and possibly also AFB4 and AFB5) bind to a conserved sequence called motif II in Aux/IAA proteins, leading to Aux/IAA protein ubiquitination by SCF complexes and subsequent turnover by the proteasome(Dharmasiri et al., 2005a; Dharmasiri et al., 2003; Dharmasiri et al., 2005b; Gray et al., 2001; Kepinski and Leyser, 2004; Kepinski and Leyser, 2005; Ramos et al., 2001; Tan et al., 2007; Tian et al., 2003; Walsh et al., 2006; Zenser et al., 2001). Auxin-induced Aux/IAA turnover frees ARF proteins to activate target genes(Guilfoyle and Hagen, 2007; Ulmasov et al., 1997a; Ulmasov et al., 1999a; Ulmasov et al., 1999b). Dominant or semi-dominant missense mutations that affect motif II in Arabidopsis IAA genes decrease interaction of the corresponding Aux/IAA proteins with auxin receptors and thereby stabilize them(Dharmasiri et al., 2005b; Gray et al., 2001; Kepinski and Leyser, 2004; Kepinski and Leyser, 2005; Reed, 2001; Tatematsu et al., 2004; Tian et al., 2003; Yang et al., 2004). In most cases, gain-of-function iaa mutations decrease auxin response,consistent with the model that Aux/IAA proteins inhibit gene activation by ARFs. However, gain-of-function axr3 mutations in AXR3/IAA17increase response to auxin in some assays(Leyser et al., 1996).

Both root formation and shoot patterning in Arabidopsis embryos require auxin-regulated gene expression responses. mp(monopteros) mutants deficient in MP/ARF5, gain-of-function mutations in BODENLOS(BDL)/IAA12 or IAA13, and embryos lacking multiple auxin receptors lack the hypophysis (the precursor of the root cap and quiescent center) and fail to form a primary root(Berleth and Jürgens,1993; Dharmasiri et al.,2005b; Hamann et al.,2002; Hamann et al.,1999; Weijers et al.,2005a). Hypophysis formation requires MP/ARF5 activity in overlying pro-embryo axis cells (Weijers et al., 2006). These studies have led to the model that BDL/IAA12 and IAA13 regulate MP/ARF5 activity in the axis. axr3 embryos form a root but have aberrant root cap cell morphology, indicating that auxin also regulates embryonic root cell differentiation(Sabatini et al., 1999).

In the embryonic shoot, mp mutations cause frequent cotyledon fusions and vascular patterning defects(Aida et al., 2002; Berleth and Jürgens, 1993; Hardtke and Berleth, 1998; Przemeck et al., 1996). nph4 (non-phototropic hypocotyl) single mutants defective in NPH4/ARF7 have normal embryo patterning, but mp nph4 double mutants lack both root and shoot organs(Hardtke et al., 2004). Mutation of the miR160 target site in ARF17, which may cause increased or ectopic expression of ARF17, causes defects in leaf shape, cotyledon outgrowth and flower morphology(Mallory et al., 2005). tir1 afb2 afb3 triple and tir1 afb1 afb2 afb3 quadruple auxin receptor mutants also often have aberrant cotyledon outgrowth(Dharmasiri et al., 2005b),although Aux/IAA proteins that act in the apical domain to regulate patterning have not been identified.

Auxin movement also contributes to correct patterning. PIN1 and other related proteins are required for auxin efflux, and their polar localization in plasma membranes have suggested routes of auxin movement in embryos,meristems and other tissues (Benkova et al., 2003; Friml et al.,2002; Friml et al.,2003; Galweiler et al.,1998; Heisler et al.,2005; Reinhardt et al.,2000; Reinhardt et al.,2003). pin1 mutants frequently have fused cotyledons,fail to form flowers on inflorescence meristems, and have fewer organs in rare flowers that do form (Aida et al.,2002; Furutani et al.,2004; Okada et al.,1991; Vernoux et al.,2000). PIN2, PIN3, PIN4 and PIN7 also contribute to embryo patterning, and pinoid mutants with defects in PIN localization also have phyllotactic defects (Benjamins et al., 2001; Bennett et al.,1995; Blilou et al.,2005; Friml et al.,2002; Friml et al.,2004; Vieten et al.,2005; Weijers et al.,2005b). PIN genes and PIN proteins are subject to positive and negative feedback. Auxin regulates PIN and PIDgenes, and individual pin mutations cause other PIN genes to be expressed in expanded domains (Benjamins et al., 2001; Blilou et al.,2005; Vieten et al.,2005). Auxin also inhibits PIN protein endocytosis and it can affect PIN polar localization (Paciorek et al., 2005; Sauer et al.,2006a). mp pin1 mutants and mp mutants treated with an auxin transport inhibitor have no leaves, indicating that auxin transport and MP/ARF5 have partly independent effects in shoot apical meristems (Schuetz et al.,2008).

Here, we describe the iaa18-1 gain-of-function mutation in IAA18, which stabilizes an Aux/IAA protein that is expressed in the apical domain of embryos. The mutation affects PIN1 expression in the apical domain, and causes aberrant cotyledon positioning and outgrowth.

The iaa18-1 mutant was backcrossed six times to the Landsberg erecta (Ler) ecotype prior to detailed characterization. mp-CSH1 and bdl-1 mutations are in the Ler ecotype,and axr1-13, axr6-1 and nph4-1 mutations are in Columbia(Col) ecotype (Hamann et al.,1999; Hardtke and Berleth,1998; Hobbie et al.,2000; Liscum and Briggs,1996) (M.-R. Cha and M. Estelle, personal communication). Plants were genotyped by PCR using CAPS, dCAPS or T-DNA border markers. PIN1:GFP fusion plants in both Ler(Heisler et al., 2005) and Col(Benkova et al., 2003)backgrounds were used.

N-terminal and full-length IAA18 genomic fragments were amplified from wild-type or iaa18-1 plants by PCR using a 5′ primer covering the PstI site 2576 bp upstream of the IAA18 start codon and 3′ primers terminating at either C135 or R267 of the coding sequence or 334 bp downstream of the stop codon. Genomic fragments were subcloned, sequenced and subsequently inserted into the PstI and BamHI sites of the pCAMBIA 1391Xa vector (CAMBIA, Canberra,Australia) to generate translational IAA18:GUS fusions, or into the pPZP211 vector (Hajdukiewicz et al.,1994) to generate the iaa18-1 plasmid. Introduction of a BamHI site into the 3′ primers generated a C135W mutation in the IAA18NT:GUS and iaa18-1NT:GUS plasmids, and an additional R at the C terminus of the IAA18:GUS and iaa18-1:GUS plasmids. Constructs were introduced into wild-type Ler plants by the floral dip method. T1 transformed seedlings were selected on plates containing hygromycin, transferred to soil and allowed to self-fertilize. For IAA18NT:GUS, iaa18-1NT:GUS and IAA18:GUSconstructs, GUS activity was assayed among progeny of at least 16 primary transformants. Transformed lines carrying the same construct exhibited qualitatively similar staining patterns with some variation in intensity. Transformants with single loci of the constructs were selected for subsequent analyses and crossing into mutant backgrounds. Reporter gene activities in Col/Ler mixed backgrounds were consistent among F3 and F4 progeny of at least three independent F2 lines. iaa18-1 or iaa18-1:GUSconstructs were silenced in all surviving T1 plants and their progeny.

For P_35S:ARF6 and P_35S:ARF8constructs, ARF6 or ARF8 cDNA sequences were amplified with primers flanking the coding region, cloned into pENTR/D-TOPO (Invitrogen) and subcloned into pB2GW7 using LR clonase(Karimi et al., 2002). Primers used for RT-PCR were: MP/ARF5,5′-GGAGATGATCCATGGGAAGAGT-3′ and 5′-GTTAATGCCTGCGCTGTTCA-3′; ARF6,5′-CACCTTTGTGAAGGTGTACAAGTC-3′ and 5′-ACGTCGTTCTCTCGGTCAAC-3′; NPH4/ARF7,5′-GCGATGATCCATGGGAAGA-3′ and 5′-GCGATGATCCATGGGAAGA-3′; ARF8,5′-CACGAGCTGCGAGAAGAGTTAG-3′ and 5′-CAAACGTTATTCACAAATGACTCC-3′; and UBQ10,5′-AACTATCACTTTGGAGGTGGAGA-3′ and 5′-TGTGGACTCCTTCTGAATGTTG-3′.

For the UAS:iaa18-1 construct, a HindIII-BamHI fragment with the GAL4 UAS from pSDM7023(Weijers et al., 2005b) was subcloned into pB7WG2 (Karimi et al.,2002) to obtain pB7WG2-UAS, in which the UAS fragment replaced the P_35S promoter. iaa18-1 cDNA was reverse transcribed from 10-day-old seedlings and PCR amplified using primers 5′-CACCACTAGTATGGAGGGTTATTCAAGAAA-3′ and 5′-CCGAGCTCTCATCTTCTCATTTTCTCTT-3′. The PCR product was cloned into pENTR/D-TOPO and subcloned into pB7WG2-UAS using LR clonase(Invitrogen).

For β-glucuronidase staining, seedlings and ovules were fixed in cold 90% acetone for 20-30 minutes, washed three to four times for 5 minutes in cold 50 mM PO₄ buffer and stained at 37°C for 1 to 16 hours in 50 mM PO₄ buffer (pH 7.2), 0.5 mM potassium ferro/ferricyanide and 1 μg/ml 5-Bromo-4-chloro-3-indolyl-β-D glucuronic acid (X-Gluc). Seedlings were cleared in a 70, 80 and 95% ethanol series, mounted in chlorohydrate:water (8:3) and photographed either with a Wild stereomicroscope or a Nikon E800 photomicroscope equipped with a SPOT cooled color digital camera using differential interference contrast (DIC) optics.

X-Gluc staining was very similar in embryos extruded from both acetone fixed and unfixed ovules. Both protocols were employed. Early stage embryos were released from dissected ovules by forcing tissue submerged in staining buffer through fine steel mesh. Late stage embryos were hand dissected from ovules. Embryos were stained at 37°C for 4 to 16 hours in watch glasses sealed in humidified chambers, mounted directly in 5% glycerol and photographed under an oil immersion 100× objective using DIC optics.

Propidium iodide-stained roots were imaged using a Leica TCS NT/SP confocal microscope with excitation at 488 nm(Truernit et al., 2006). For imaging PIN1:GFP in embryos (Fig. 4), PIN1:GFP (Heisler et al., 2005) gynoecia were fertilized with either wild-type or iaa18-1 homozygous pollen and ovules were harvested 3 to 7 days later, fixed in 4% paraformaldehyde/1×PBS overnight and stained with DAPI (1 μg/ml in 1x PBS) (Sauer et al.,2006b). F1 embryos were extruded from ovules into 1×PBS, 5%glycerol, 0.01% Tween-20) and mounted on slides. Stacks of 1 μm optical sections were acquired on a Zeiss 510 LSM Meta confocal microscope using an oil immersion 40× objective. To image GFP and DAPI together, we used multitracking in line-scan mode. For GFP we used a 488 nm laser line attenuated to 10% and a 505-530 nm band pass filter. For DAPI we used a 364 laser line attenuated to 5% and a 385-470 nm band pass filter. Wild-type and iaa18-1/+ embryos were photographed under identical settings. Images in Fig. S3 in the supplementary material were taken on a Zeiss DUO confocal microscope.

An IAA18 fragment was amplified from first-strand cDNA as described above, and was cloned into pGEM-T vector (Promega). The plasmid was linearized by SpeI digestion and the antisense probe was synthesized by in vitro transcription with SP6 RNA polymerase using a DIG RNA labeling kit(Roche). In situ hybridization was performed as described previously(Long and Barton, 1998).

Among progeny of EMS-mutagenized Landsberg erecta seed, we found the iaa18-1 mutant as a dwarfed plant with leaves that curled up(Fig. 1A-C). Homozygous mutant plants were much smaller than heterozygous plants(Fig. 1A), and segregation ratios among progeny of the mutant indicated that a single semi-dominant mutation caused both leaf curling and dwarfism (data not shown). We mapped the mutation to an interval including IAA18 (At1g51950), and found a guanine to adenine transition in IAA18 that changes the glycine codon at amino acid position 99 to glutamate in the mutant. Glycine 99 is a highly conserved residue in motif II, VGWPPV, and is important for instability of Aux/IAA proteins (Ramos et al.,2001). The shy2-3 mutation changes the corresponding glycine of SHY2/IAA3 to glutamate, and also causes significant phenotypes including upward leaf curling(Tian and Reed, 1999). Recently, Uehara et al. independently isolated iaa18 mutations in the Columbia ecotype that affect the same codon as iaa18-1 and cause very similar phenotypes (Uehara et al.,2008).

As homozygous iaa18-1 plants were almost sterile (see below), for phenotypic analyses we used progeny of self-pollinated heterozygous plants. Three to 8% of iaa18-1/IAA18 progeny had aberrant cotyledon outgrowth, with higher frequency in a Landsberg erecta (Ler)background than in a mixed Ler/Columbia background(Fig. 1, Table 1). Some seedlings had a single cotyledon that wrapped around the meristem to encompass more than half of the circumference of the seedling apex, or had a single cotyledon of normal appearance (monocots, Fig. 1F). Enlarged cotyledons in monocots had more veins than did cotyledons of wild-type seedlings, suggesting that they might have arisen from incomplete separation of two cotyledon primordia (data not shown). Other seedlings had two distinguishable cotyledons that were fused along one edge (fused dicots, Fig. 1E), had two cotyledons with asymmetric placement (asymmetric dicots, Fig. 1D), or had three cotyledons (tricots, not shown). These aberrant cotyledon outgrowth phenotypes occurred at higher frequency among iaa18-1 homozygotes than among heterozygotes (data not shown). Mutant seedlings that had two cotyledons in normal locations unfolded their cotyledons more slowly than did wild-type seedlings (Fig. 1H). All seedlings with aberrant cotyledon placement or folded cotyledons later developed curled leaves, suggesting that the iaa18-1 mutation affected cotyledon outgrowth with incomplete penetrance.

In addition to cotyledon patterning defects, mutant seedlings also had long hypocotyls when grown in the light, reduced primary root growth,differentiated cells closer to the root meristem than normal, and fewer lateral roots than wild-type seedlings (see Fig. S1 and Table S1 in the supplementary material) (Uehara et al.,2008). Mutant seedlings were visibly purple, similar to stressed wild-type seedlings. Auxin inhibited mutant root growth to a similar degree as in wild-type seedlings (see Fig. S1 in the supplementary material). A small proportion (0.1-0.3%) of progeny of self-fertilized iaa18-1heterozygotes lacked a root (Fig. 1G, Table 1). iaa18-1 also increased the frequency of rootless seedlings in bdl, axr1-13 and tir1-1 backgrounds(Table 1). Flowers of homozygous plants had fewer petals and stamens than did flowers of wild-type plants (see Table S2 in the supplementary material), and they had short stamen filaments. Most ovules in homozygous mutant siliques aborted without forming a seed, even if pollinated with wild-type pollen.

Transgenic plants carrying either an iaa18-1 genomic construct(iaa18-1) or a fusion of the iaa18-1 promoter and full-length open reading frame to the GUS gene (iaa18-1:GUS)also had curled leaves, closed cotyledons and/or cotyledon phyllotaxy defects(see Materials and methods; Fig. S2 and Table S3 in the supplementary material), indicating that the iaa18-1 mutation can cause the phenotypes we observed. Consistent with the semidominance of iaa18-1,different transformants had a range of phenotypes that were generally stronger than those of the original mutant, suggesting that these phenotypes are sensitive to iaa18-1 gene dosage or expression level. Most T1 seedlings, including all of those with strong phenotypes, failed to survive to adulthood or to set seed, and a significant frequency also lacked a primary root (see Fig. S2, Table S3 in the supplementary material). iaa18-1and iaa18-1:GUS transgenes had similar effects, indicating that the IAA18-1:GUS fusion protein retains function.

We also generated plants carrying IAA18:GUS, with a full-length wild-type gene fused to GUS, and plants carrying IAA18NT:GUSor iaa18-1NT:GUS constructs with the N-terminal region of IAA18,including motifs I and II but lacking the C-terminal dimerization domain(Fig. 2A). These plants had normal morphology, in contrast to plants with the full-length mutant iaa18-1:GUS construct. Apparently, the truncated fusion proteins do not interact with ARF proteins. We used these plants as reporters for IAA18 expression, and to explore the effect of the iaa18-1mutation on IAA18 protein.

Plants carrying IAA18NT:GUS or IAA18:GUS constructs had X-Gluc staining in the stele of roots and vascular tissues of hypocotyl,cotyledons and leaves (Fig. 2D,E, data not shown). The staining in the root was strongest in the stele in the elongation zone just above the meristem, weaker in the meristem and in older parts of the root, and excluded from the root cap and meristem initials (Fig. 2E). Adaxial domains of developing leaf primordia also had staining(Fig. 2G), which became restricted to the vasculature as the leaves expanded, similarly to the staining in cotyledons (data not shown, Fig. 2D). Staining also appeared in chalazal pole cells of mature ovules(Fig. 3A).

Plants with the iaa18-1NT:GUS construct had strong X-Gluc staining throughout the hypocotyl, cotyledons and leaves, rather than just in vascular tissues as for the IAA18:GUS and IAA18NT:GUS fusions(Fig. 2D). Staining in leaf primordia appeared somewhat stronger in the adaxial domain than in the abaxial domain (Fig. 2H). In roots, iaa18-1NT:GUS plants had staining in the stele just behind the root meristem (Fig. 2F) and also in older parts of the root (Fig. 2D). In ovules, staining was present in cells at the chalazal pole, in the endothelium and integuments that later form the seed coat, and,to a lesser extent, in the endosperm (Fig. 3B). The IAA18, iaa18-1 and fusion protein transcripts were present at comparable levels (Fig. 2C), and were not induced by auxin(Fig. 2B)(Okushima et al., 2005; Tian et al., 2002). These results indicate that the iaa18-1 mutation increased the amount of fusion protein in mature root stele and in shoot organs, most probably by stabilizing the IAA18 protein as motif II mutations in other IAAgenes do. IAA18-1NT:GUS protein may be present in domains where IAA18:GUS protein is absent because the wild-type reporter protein is turned over quickly in most tissues.

To ascertain in which cells IAA18-1 acts to affect cotyledon formation, we assessed IAA18 expression pattern in embryos. In iaa18-1NT:GUS embryos, X-gluc staining appeared initially in the apical domain sometimes at the 16-cell stage, and with complete penetrance at the 32-cell stage (Fig. 3G). At late globular stage, staining became restricted to a strip of cells encompassing the nascent shoot apical meristem and extending through the periphery of the embryonic apex (Fig. 3H,I,A′). This apical expression pattern persisted through subsequent stages so that it became apparent that staining on the apical periphery had been restricted to cells between the cotyledons. Staining appeared on the adaxial sides of the cotyledons by mid-heart stage and this persisted through heart and early torpedo stages(Fig. 3J,K). At late torpedo and walking stick stages, staining appeared throughout the cotyledons,especially in developing vasculature, and in axis vasculature(Fig. 3L,M). Expression patterns from globular to heart stages seen by in situ hybridization with an IAA18 antisense probe mirrored these iaa18-1NT:GUS staining results almost exactly (Fig. 3N-R; J. Long, personal communication), revealing that the iaa18-1NT:GUS reporter expression reflects the IAA18transcript pattern. This expression pattern, and the ability of the iaa18-1:GUS transgene to recapitulate all iaa18-1 embryo phenotypes, suggest that the stabilized IAA18-1 protein acts in apical cells in globular embryos, and in adaxial and nascent shoot apical meristem domains of heart stage embryos.

In mp-CSH1, iaa18-1, axr6-1 and axr1-13 embryos, iaa18-1NT:GUS staining was present in the same apical domain as in wild-type embryos (Fig. 3U-Y). In mp-CSH1 nph4-1 double mutant embryos, which do not form cotyledons, staining was present in the presumptive shoot apical meristem but not on the flanks (Fig. 3Z,B′). Together with the absence of auxin regulation of IAA18 transcript level (Fig. 2B), these results suggest that ARF proteins do not regulate IAA18 expression directly.

In contrast to iaa18-1NT:GUS embryos, IAA18:GUS and IAA18NT:GUS embryos expressing wild-type fusions lacked X-Gluc staining (Fig. 3A, data not shown). Some axr1-13 and axr6-1 mutant embryos with defects in SCF ubiquitin ligase function had IAA18:GUS staining in the apical domain of embryos (Fig. 3S,T). SCF complexes may normally mediate efficient IAA18 turnover in the apical domain of embryos, so that the wild-type IAA18:GUS and IAA18NT:GUS fusion proteins do not accumulate to detectable levels.

In siliques of self-pollinated iaa18-1/IAA18 heterozygous plants,we observed heart-stage embryos with extra cells between margins of adjacent cotyledon primordia, as well as monocot embryos(Fig. 3C-F, Fig. 4K). To detect patterning phenotypes before cotyledon outgrowth, we used embryos heterozygous for iaa18-1 and carrying a PIN1:GFP protein fusion construct(Heisler et al., 2005). In wild-type early globular stage embryos, PIN1:GFP was expressed throughout the apical half and in nascent provascular cells(Fig. 4A). At mid- and late-globular and transition stages, expression was highest at two foci on the flanks of the apex, and was also present in the cell tier beneath the apical half of the pro-embryo and in the provascular cells of the incipient hypocotyl(Fig. 4D,G). In heart stage embryos, expression persisted in the L1 cells of the nascent shoot apical meristem; in tips, abaxial L1 and distal adaxial L1 cells of cotyledons; and in provascular cells of the cotyledons and axis(Fig. 4J). As heart stage embryos elaborated L2 and L3 layers of the nascent shoot apical meristem,expression was lost in these cell layers.

PIN1:GFP expression in iaa18-1/IAA18 embryos deviated from the wild-type expression pattern starting at early globular stage. Whereas in wild-type embryos expression was uniform throughout the apical half, in iaa18-1 embryos expression was often asymmetric with stronger fluorescence on one side than the other(Fig. 4B,C,E,F,H,I). We observed such asymmetric expression in about 19% of early globular stage embryos examined, and about 40% of mid- to late-globular and transition stage embryos (see Table S4 in the supplementary material). (The overall frequency of asymmetric PIN1:GFP in wild-type embryos was about 1%.) Some iaa18-1 transition stage embryos also had discontinuities in PIN1:GFP fluorescence in the nascent vasculature(Fig. 4H).

In addition to asymmetries in PIN1:GFP expression, iaa18-1globular embryos occasionally had aberrant or ectopic cell divisions in the L1 layer within a focus of strong PIN1:GFP expression. In one case, periclinal divisions of adjacent apical L1 cells were not aligned with each other,leading to disordered cell layers (Fig. 4C). In another case, an ectopic periclinal cell division occurred(Fig. 4F).

iaa18-1 heart-stage embryos often lacked PIN1:GFP fluorescence in cells in a single tier just beneath the embryonic shoot apical meristem(Fig. 4K, see Fig. S3 in the supplementary material). As seen in transition stage embryos, some iaa18-1 heart stage embryos also had discontinuities in PIN1:GFP fluorescence in developing cotyledon vasculature (see Fig. S3 in the supplementary material). Last, at heart and torpedo stages, the axis vascular column visualized by PIN1:GFP fluorescence was narrower in iaa18-1embryos than in wild-type embryos (Fig. 4J,K; see Fig. S3 in the supplementary material).

In contrast to the effects on PIN1:GFP, we detected no effect of iaa18-1 on expression of the synthetic auxin-responsive reporter genes P_DR5:GUS and P_DR5:GFP(Benkova et al., 2003; Ulmasov et al., 1997b) at globular or transition stages (data not shown). At heart or torpedo stages after morphological abnormalities appeared, some mutant embryos lacked normal P_DR5 expression in the root pole or had a broader expression domain along the margin of fused cotyledons (see Fig. S3 in the supplementary material).

mp mutants often have fused cotyledons and also have decreased PIN1:GFP expression in leaf primordia(Berleth and Jürgens,1993; Przemeck et al.,1996; Wenzel et al.,2007), suggesting that IAA18-1 might cause embryonic phenotypes by inhibiting MP/ARF5. We found that a P_35S:MP/ARF5 construct(Hardtke et al., 2004) could suppress iaa18-1 vegetative phenotypes(Fig. 1, see Table S5 in the supplementary material). About one-third of iaa18-1/P_35S:MP/ARF5 T1 plants had flat rather than curled leaves (Fig. 1O,P,S,T). Some of these had leaves as large as those of IAA18 P_35S:MP/ARF5plants and were nearly as tall (Fig. 1M,O). Two iaa18-1/iaa18-1 P_35S:MP/ARF5 plants had flat leaves and produced significant yields of seeds. Thus, overexpressing MP/ARF5 could rescue leaf curling, stem elongation and fertility defects of iaa18-1. Conversely, the iaa18-1 mutation appeared to suppress the P_35S:MP/ARF5 terminal flower phenotype (Fig. 1N,Q,R, see Table S5 in the supplementary material). These data indicate that IAA18-1 and MP/ARF5 can antagonize each other in plants.

To test whether IAA18-1 protein can inhibit endogenous MP/ARF5 activity in embryos, we expressed iaa18-1 in the central domain of the embryonic axis by transforming a UAS:iaa18-1 construct into the GAL4:VP16-expressing driver line Q0990(Haseloff et al., 1999; Weijers et al., 2006). Of 18 Q0990 UAS:iaa18-1 T1 plants we obtained, 16 lacked a root. Of these,four had two cotyledons, five had fused cotyledons and seven were monocots(see Fig. S4 in the supplementary material). Thus, iaa18-1 can produce a mp-like embryo phenotype when expressed in the domain crucial for MP/ARF5 function.

However, iaa18-1 and mp-CSH1 mutations enhanced each other, indicating that IAA18-1 must also affect targets other than MP/ARF5. Whereas iaa18-1 and mp-CSH1 single mutants always had at least one cotyledon, mp-CSH1 iaa18-1/ seedlings (either homozygous or heterozygous for iaa18-1) often lacked cotyledons and sometimes also leaves (Fig. 1I,J,K; Table 1). After several weeks,some mp-CSH1 iaa18-1/ double mutant seedlings developed radialized finger-like organs from the apex resembling those of mp pin1seedlings (Fig. 1L)(Schuetz et al., 2008). Occasionally, these had pistil-like tissue at the tip (data not shown).

Other candidate IAA18-1 targets closely related to MP/ARF5 include NPH4/ARF7, ARF6 and ARF8. The nph4-1 mutation did not enhance the frequency of apical patterning defects of iaa18-1 seedlings(Table 1), suggesting that IAA18-1 inhibits NPH4/ARF7 in embryos. However, a P_35S:NPH4/ARF7 construct(Hardtke et al., 2004), as well as P_35S:ARF6 and P_35S:ARF8constructs, each failed to suppress iaa18-1 vegetative phenotypes(see Table S5 in the supplementary material).

Quantitative real-time RT-PCR experiments revealed that MP/ARF5transcript was present at 15-75 times the wild-type level in shoots of 2-week-old T2 plants from four different iaa18-1/P_35S:MP/ARF5 lines with suppressed phenotypes (see Fig. S5 in the supplementary material). Similar analyses of four to five lines each carrying the other overexpression constructs revealed only up to sixfold increases over control transcript levels in P_35S:ARF6 or P_35S:ARF8 lines, and up to 15-fold increases in P_35S:NPH4/ARF7 lines (see Fig. S5 in the supplementary material).

The bdl mutation interacted similarly to iaa18-1 with ARF overexpression constructs. In particular, P_35S:MP/ARF5 suppressed the curled leaf and dwarfed stature of bdl mutant plants, but P_35S:NPH4,P_35S:ARF6, and P_35S:ARF8 did not(Hardtke et al., 2004) (Table S5 in the supplementary material). bdl/BDL iaa18-1/IAA18 embryos retained cotyledons (Table 1),suggesting that neither BDL nor IAA18-1 (nor both together) completely inhibited ARF activity in domains relevant for cotyledon outgrowth. Plants with loss-of-function mutations in IAA12/BDL (SALK138684) or IAA18 (S.E.P., J. M. Alonso, J. R. Ecker and J.W.R., unpublished),and double loss-of-function iaa12 iaa18 mutant plants developed normally (data not shown).

IAA18-1 is present throughout the apical domain of globular stage embryos,and its turnover is required for correct PIN1:GFP expression and for proper cotyledon outgrowth. As iaa18-1 patterning defects resemble those of pin1 and pinoid mutants(Aida et al., 2002; Bennett et al., 1995; Okada et al., 1991), it is likely that altered expression of PIN1 (and possibly other PIN genes) in iaa18-1 embryos contributes to subsequent cotyledon placement defects. Thus, some cotyledon anlagen cells may acquire insufficient auxin and fail to grow, whereas others may accumulate too much auxin and therefore overproliferate. Intercellular positive feedback might reinforce initial asymmetries in PIN expression; for example, to cause a single initiated cotyledon to occupy an enlarged domain that includes cells that normally form cotyledon margins. This model can explain why inhibiting auxin response symmetrically throughout the apical domain causes asymmetry in the resulting morphology. As auxin does not induce IAA18transcription, the iaa18-1 mutant may be partially insulated from negative feedback that might normally promote robust patterning.

We observed iaa18-1 globular embryos with asymmetric PIN1:GFP expression more frequently than seedlings with aberrant cotyledon placement,suggesting that many mutant embryos recover normal patterning, despite early PIN1 misexpression. Similarly, pin1, pin7 and higher-order pin mutant seedlings often appear normal, despite having reduced embryonic PIN expression and exhibiting early cell division patterning defects (Blilou et al.,2005; Friml et al.,2003; Okada et al.,1991; Vieten et al.,2005). Although we could not follow individual embryos as they developed, a subset of embryos with asymmetric PIN1 expression, perhaps those with ectopic cell divisions or discontinuous vasculature, may later develop aberrant cotyledon outgrowth. Subsequently, at heart stage, absence of PIN1:GFP expression in sub-meristem cells and vascular discontinuities may arise from persistent action of IAA18-1, or as indirect consequences of earlier perturbed auxin transport.

Ectopic expression of axr2/iaa7 or slr/iaa14gain-of-function mutant genes in transition and heart stage cotyledon primordia eliminated growth of one or both cotyledons, but did not affect cotyledon patterning as iaa18-1 does(Muto et al., 2007). These results suggest that ARF function is necessary for cotyledon outgrowth, even after patterning has been established. It is therefore possible that, in addition to affecting PIN expression, IAA18-1 might affect expression of ARF target genes whose products drive cell expansion or cell division. IAA18-1 might decrease expression of such genes in cotyledon primordia and/or increase their expression in cotyledon margin zones. Although simple models of auxin response imply that gain-of-function iaa mutations should decrease ARF target gene expression, increases in expression might arise from decreased intracellular negative feedback or from decreased auxin efflux. Gain-of-function axr3 mutants have increased auxin response in some assays, and also have accelerated hypocotyl growth as do iaa18-1seedlings (Leyser et al.,1996).

iaa18-1 and mp mutant embryos each have similar apical phenotypes, and the ability of overexpressed MP/ARF5 to suppress iaa18-1 phenotypes indicates that IAA18-1 can interact with MP/ARF5. MP/ARF5 is expressed and present throughout the embryo except in the L1 layer(Hamann et al., 2002; Hardtke and Berleth, 1998; Hardtke et al., 2004; Weijers et al., 2006), so IAA18-1 could inhibit MP/ARF5 in most apical domain cells.

As iaa18-1 and mp mutations enhanced each other, IAA18-1 probably also targets other ARF proteins. ARF6, NPH4/ARF7, ARF8 and ARF19 are the most closely related ARF proteins to MP/ARF5(Remington et al., 2004). nph4 mutations enhanced mp mutations(Hardtke et al., 2004) but did not enhance iaa18-1, consistent with NPH4/ARF7 and IAA18-1 acting in a common pathway. Moreover, IAA18 can interact with NPH4/ARF7 and ARF19 in yeast two-hybrid assays, and the reduced frequency of lateral roots in iaa18 plants also suggests that IAA18-1 may inhibit NPH4/ARF7 and ARF19 in roots (see Table S1 in the supplementary material)(Okushima et al., 2005; Uehara et al., 2008; Wilmoth et al., 2005). Similarly, iaa18-1, arf6 and arf8 mutants each have long hypocotyls and short stamen filaments(Nagpal et al., 2005; Tian et al., 2004), suggesting that IAA18-1 might inhibit ARF6 or ARF8. However, overexpression of ARF6,NPH4/ARF7 or ARF8 did not suppress iaa18-1. In addition. in a wild-type background, only MP/ARF5 overexpression causes strong phenotypes (Hardtke et al.,2004; Wu et al.,2006). Regulation by the microRNA miR167 apparently limits the effectiveness of ARF6 and ARF8 overexpression constructs (Wu et al., 2006). A higher degree of overexpression might be needed to accumulate enough ARF6,NPH4/ARF7 or ARF8 protein to suppress iaa18-1 effects; for example,if translation of these ARF proteins is inefficient. Alternatively, these ARF proteins may differ from MP/ARF5 in some functional attribute. Thus, in embryos, IAA18-1 may inhibit one or more of these ARF proteins, or other ARFs that we have not tested.

Similarly to iaa18-1, the bdl mutation decreased PIN1 expression, and inhibited MP/ARF5 function when driven by the Q0990GAL4-expressing line (Weijers et al.,2006). Moreover, mp bdl embryos lacked cotyledons,indicating that BDL/IAA12 also has targets in addition to MP/ARF5(Hamann et al., 2002; Hamann et al., 1999). Together, IAA18-1 and BDL are expressed in all cells of globular and heart stage embryos, except the L1 layer on the abaxial flanks(Hamann et al., 2002; Weijers et al., 2006), so the presence of cotyledons in iaa18-1/IAA18 bdl/BDL embryos suggests either that the concentration of BDL or IAA18-1 was below a threshold required to inhibit the relevant ARF proteins fully, or that IAA18-1 and BDL proteins may attenuate the activity of each other, for example through protein-protein interactions. IAA12/BDL and IAA18 may normally act partially redundantly. However, bdl iaa18 double loss-of-function mutants developed normally. Higher-order loss-of-function mutants may reveal whether Aux/IAA proteins are in fact necessary for correct embryo patterning.

Root formation depends on auxin response in the axis(Weijers et al., 2006), where IAA18 is not expressed. Decreased root pole formation in iaa18-1/bdl/, iaa18-1/axr1-13 and iaa18-1/tir1-1 double mutants, and the narrower domain of PIN1:GFP expression in the axis of iaa18-1 heart and torpedo stage embryos, suggest that iaa18-1 acts non-autonomously on axis cells. IAA18-1 might reduce apical to basal auxin flux, either by affecting auxin efflux from apical cells, or as an indirect consequence of altered apical patterning or cotyledon outgrowth. Consistent with these models, WEI8/TAA1, YUCCA1(YUC1), YUC4, YUC10 and YUC11 genes involved in tryptophan-dependent auxin biosynthesis are expressed in the apical domain of globular stage embryos, and embryonic root formation requires these YUCCA genes, as well as polar transport of auxin from apical to basal domains starting at the globular stage(Cheng et al., 2007; Friml et al., 2003; Steinmann et al., 1999; Stepanova et al., 2008).

We thank R. P. Elumalai for identifying the iaa18-1 mutant, M. Ben-Davies for help with mapping, J. Dangl and S. Grant for use of their compound microscope, J. Haseloff for use of his confocal microscope, and L. Hobbie, E. Liscum, T. Berleth, J. Friml, M. Cha, M. Estelle, E. Meyerowitz and G. Jürgens for seeds of various genotypes.