ASYMMETRIC LEAVES1 and auxin activities converge to repress BREVIPEDICELLUS expression and promote leaf development in Arabidopsis

Correct cell fate allocation in the Arabidopsis shoot depends upon mutual repression between the AS1 Myb protein, which promotes leaf fate, and class I KNOX transcription factors, which promote meristem activity(Long et al., 1996; Ori et al., 2000; Byrne et al., 2000; Byrne et al., 2002). These interactions result in the delimitation of AS1-expressing leaf founder cells in the meristem. Leaf initials are also defined by local auxin maxima generated by activity of the PINFORMED1 (PIN1) auxin efflux facilitator protein (Benkova et al., 2003; Reinhardt et al., 2003; Heisler et al., 2005). However, it is not known how the promotion of organ growth by auxin is integrated with the cell fate allocation pathway defined by AS1/KNOX proteins.

The disruption of mechanisms repressing KNOX expression in leaves is associated with perturbations in leaf development and disruption in auxin homeostasis (Tsiantis et al.,1999; Scanlon et al.,2002; Zgurski et al.,2005), indicating that KNOX misexpression may perturb an auxin-directed mechanism controlling leaf morphogenesis. However, whether regulated auxin gradients sculpt leaf shape and how KNOX activity in leaves may disrupt such auxin gradients is unclear.

Here, we investigate these questions by examining genetic interactions between components of the AS1/KNOX and auxin regulatory pathways in Arabidopsis. We show that auxin activity acts together with AS1 to repress expression of the KNOX gene BP and hence promote leaf development. We also provide evidence that local auxin maxima are required to initiate marginal serrations in the wild-type Arabidopsisleaf, indicating that auxin acts not only to define leaf inception at the meristem but also later in development to control leaf shape. We also show that ectopic BP expression in Arabidopsis leaves alters leaf shape, at least in part by perturbing these local auxin gradients that shape the leaf margin. Thus, the combined action of AS1 and auxin to repress BP expression in leaves plays a key role in safeguarding leaf fate and controlling leaf shape.

All mutant alleles and transgenic lines are listed in Table 1. as1;axr1: axr1 plants from the F2 generation of a cross between axr1-12 and as1-1 homozygotes were self-pollinated to generate F3 families that segregated as1;axr1 double mutants. bp;axr1: bp plants from the F2 generation of a cross between bp-9 and axr1-12 homozygotes were self-pollinated to generate F3 families that segregated bp;axr1 double mutants. No obvious difference in leaf phenotype was observed between axr1-12 and bp-9;axr1-12 leaves by visual inspection of 50 single and double mutants. bp;as1;axr1: as1;axr1 plants from the F2 generation of a cross between as1-1;axr1-12 and bp-9;axr1-12 plants were self-pollinated to generate F3 families that segregated bp;as1;axr1triple mutants. axr1;BP::GUS: GUS-positive axr1 plants from the F2 generation of a cross between axr1-3 and BP::GUShomozygotes were self-pollinated to generate axr1;BP::GUS F3 families. as1;axr1;BP::GUS: axr1 plants from the F2 generation of a cross between axr1-3;BP::GUS and as1-1;BP::GUS (Ori et al.,2000) homozygotes were self-pollinated to generate F3 families that segregated as1;axr1;BP::GUS double mutants. as1;axr1;BP::GUS double mutants were also constructed with the axr1-12 allele and showed an identical pattern of BP::GUSexpression to double mutants constructed with the axr1-3 allele. pin1-6 double mutant combinations were generated by crossing heterozygous plants with as1-1, bp-9 or blr-126 mutants. as1, bp or blr plants from the F2 generation were self-pollinated to generate respective F3 families that segregated as1;pin1-6, bp;pin1-6 or blr;pin1-6 double mutants. We excluded any effects of mixed backgrounds on the phenotypes by obtaining similar results using pin1-En134, which was generated in the Col ecotype, in crosses with as1-1 and bp-9. pid;bp: bp plants from the F2 generation of a cross between a heterozygous pid-3 and homozygous bp-9 plant were self-pollinated to generate F3 families that segregated bp;pid double mutants. All reporter lines were crossed into respective mutant or transgenic lines and expression analysis was performed in segregating F3 families.

Plants on soil were grown in a greenhouse with supplemental lighting (days:18 hours, 20°C; nights: 6 hours, 16°C) or in growth cabinets under the same conditions.

1-N-naphthylphthalamic acid (NPA; Duchefa) was dissolved in dimethyl sulfoxide (DMSO) to a stock concentration of 500 mM and added to Murishuge Skoog (MS) medium to a final concentration of 5 or 10 μM. 2,3,5-Triiodobenzoic acid (TIBA; Sigma) was also dissolved in DMSO to a stock concentration of 500 mM and added to MS medium to a final concentration of 20μM. In all experiments, MS plates with the same concentration of DMSO were used as controls.

SEM and confocal microscopy were carried out as previously described(Bowman et al., 1991; Running and Meyerowitz, 1995). Seedlings for confocal microscopy were mounted and observed in water without fixation using the 458 nm argon laser of a Zeiss LSM510 microscope.

Total RNA (1 μg) extracted from mature leaf tissue was DNaseI treated and used for cDNA synthesis with an oligo(dT) primer and Superscript reverse transcriptase (Invitrogen). cDNA was amplified on the ABI PRISM 7300 Sequence Detection System (Applied Biosystems). Amplification reactions were prepared with the SYBR-Green PCR Master Kit (Applied Biosystems), according to manufacturer's specifications, with 0.4 μM of primers and 10 μl of cDNA per reaction. Each reaction was made in triplicate, and each experiment was repeated three times. The efficiency of each set of primers and calculation of the level of induction was determined according to Pfaffl(Pfaffl, 2001). Error bars represent the standard error calculated on experiment repetitions. Expression levels were normalized with values obtained for the ORNITHINE TRANSCARBAMILASE (OTC) gene, which was used as an internal reference gene as described by Cnops et al.(Cnops et al., 2004). Primers are listed in Table 2.

GUS activity was detected as described(Scarpella et al., 2004). Tissue was fixed in 90% acetone at -20°C for one hour, washed briefly with 100 mM phosphate buffer, and stained overnight in freshly prepared 100 mM sodium phosphate buffer with 10 mM sodium EDTA, 1 mM 5-bromo-4-chloro-3-indolyl-β-D-glucuronic acid (Molecular Probes) and ferrocyanide and ferricyanide salts (1 mM PIN1::GUS, 2 mM other GUS lines). Reactions were terminated with 95% ethanol, and leaves were dissected, mounted in 50% glycerol and viewed with dark-field microscopy.

Leaves were flattened onto clear adhesive, adhered to white paper and digitally scanned.

To investigate whether AS1 and auxin act in the same or distinct pathways to control leaf development, we examined genetic interactions between as1 and axr1, which confers a primary defect in auxin signalling (Lincoln et al.,1990). AXR1 encodes a subunit of the RUB1 activating enzyme that regulates the protein degradation activity of Skp1-Cullin-F box complexes, primarily, but not exclusively, affecting auxin responses(Leyser et al., 1993; Pozo et al., 1998; Gray et al., 2001; Schwechheimer et al., 2002; Xu et al., 2002). Leaf phenotypes of the as1;axr1 double mutant were enhanced with respect to either single mutant, with deeply lobed margins and ectopic stipules present in the sinus of each lobe (arrowheads, Fig. 1A-L). These novel phenotypes observed in as1;axr1 double mutants, but not in either single mutant, suggest that AS1 and auxin may act in overlapping pathways to direct leaf development.

To investigate the basis of this genetic interaction, we examined whether auxin signalling is required to repress KNOX expression in a manner similar to AS1. We observed inappropriate expression of the KNOX gene BP but not SHOOTMERISTEMLESS (STM) in axr1mutant leaves, although the level of BP expression was substantially higher in as1 than in axr1 leaves(Fig. 1M). This ectopic BP expression in axr1 leaves was not accompanied by a reduction in AS1 transcripts (Fig. 1M), indicating that AXR1 is unlikely to repress BP by promoting AS1 transcription. We observed a similar expression profile of ectopic BP but not STM in leaves of a dominant Aux/IAA17 mutant, axr3-1 (data not shown), suggesting that regulation of BP by AXR1 probably reflects SCF activity related to auxin, rather than other signalling pathways. Notably, ectopic expression of BP in the axr1 leaf is not responsible for the mild leaf phenotypes that distinguish axr1 from wild type(Lincoln et al., 1990), as bp;axr1 double mutant leaves appeared identical to axr1single mutants (Fig. 1C,J,N,P).

To determine whether the convergence of AS1 and AXR1 activities on BP regulation might account for the novel phenotypes observed in as1;axr1 double mutants, we examined the pattern of BPexpression in axr1, as1 and as1;axr1 leaves. Although BP::GUS was absent from wild-type leaves(Fig. 1R), expression was observed in the serration tips of axr1 leaves(Fig. 1S), and in the petiole,midvein and serration tips of as1 leaves(Fig. 1T). However, the pattern of BP::GUS expression in as1;axr1 double mutant leaves was different than that of either single mutant, being sharply localised to margin cells in the sinus of every lobe from an early stage in leaf development(arrows, Fig. 1U,V),correlating with the ectopic initiation of stipules. These results indicate that both AS1 and auxin signalling are required to exclude BPexpression from leaves, and that, in their absence, BP is misexpressed at the leaf margin. To test whether this novel pattern of BP expression observed in as1;axr1 leaves is responsible for the ectopic initiation of stipules, we analysed as1;axr1;bp triple mutants. Ectopic stipules observed in cauline leaves of as1;axr1double mutants, but not in as1 or axr1 single mutants, were not found in as1;axr1;bp triple mutants(Fig. 1W-Y, n=20 triple mutant leaves), indicating that their initiation is likely to depend on BP activity.

It is of note that as1 mutants do not initiate ectopic stipules,which are a hallmark of BP overexpression in Arabidopsis(Chuck et al., 1996), nor does loss of BP function suppress the as1 mutant phenotype(Byrne et al., 2002). By contrast, in as1;axr1 double mutants, ectopic stipules initiate at the sites of ectopic BP expression, and loss of BP function suppresses this phenotype. These observations suggest that, in as1mutant leaves, the inability of ectopic BP expression to elicit ectopic stipule initiation reflects the additional repression of BPby AXR1. As previously suggested, the ectopic initiation of stipules may reflect the formation of ectopic meristem-leaf boundaries within the leaf,because normal stipule formation occurs at the boundary between initiating leaves and the shoot apical meristem (SAM)(Ori et al., 2000). Notably, as1;axr1;bp triple mutant leaves do not revert to wild type, but rather to an as1-like morphology(Fig. 1O,Q), possibly because of the effects of additional KNOX genes misexpressed in as1or because AS1 also regulates KNOX-independent pathways to promote leaf development.

To test whether polar transport of auxin at the shoot apex acts in concert with AS1 to promote leaf development, we analysed pin1;as1 double mutants. The auxin efflux facilitator PIN1 transports auxin in the epidermis towards leaf initial cells that then act as auxin sinks, and this local accumulation of auxin triggers organ initiation(Benkova et al., 2003; Reinhardt et al., 2003; Heisler et al., 2005). Strong pin1 mutants with impaired auxin transport, therefore, initiate a reduced number of leaves and no flowers(Okada et al., 1991). However, pin1;as1 double mutants initiate significantly fewer leaves than do pin1 single mutants (Fig. 2A-E, Student's t-test, P=2.4×10^-3), demonstrating that AS1 and PIN1 function redundantly to promote leaf development.

To test whether PIN1 promotes lateral organ development by regulating BP, we analysed pin1;bp double mutants, predicting that those aspects of the pin1 phenotype that are BP dependent will be suppressed. We observed that the failure of pin1 mutants to initiate both leaves and flowers is partially rescued in these double mutants(Fig. 2F-H,J,K, Student's t-test, P=0.025; see also Fig. S1 in the supplementary material, Student's t-test, P=0.0025), indicating that PIN1 activity in lateral organ formation involves the repression of BP activity. BP can act as a dimer with a related homeobox protein, BELLRINGER (BLR)/PENNYWISE(Byrne et al., 2003; Smith and Hake, 2003), and the flower initiation defects of pin1 were also suppressed in blr;pin1 double mutants (Fig. 2G,I,J,L), indicating that PIN1-mediated auxin action to promote lateral organ initiation is antagonised by both BLR and BP activities. Antagonistic actions of BP towards auxin-mediated organogenesis were also observed in double mutants between bp and pinoid(pid) (Bennett et al.,1995; Benjamins et al.,2001), in which organ initiation defects of pid mutants were partially suppressed (Fig. 2M, Student's t-test, P=2.1442×10^-6).

To test whether regulated auxin transport is required to exclude BP expression from leaves, we analysed BP expression in pin1 mutants and plants where auxin transport was perturbed by treatment with auxin transport inhibitors. We observed misexpression of BP transcripts in pin1 leaves, relative to wild type,whereas AS1 transcript levels were unaltered(Fig. 2N), which indicates that PIN1 activity is required to repress BP expression, but not via promoting AS1 transcription. Similarly, we observed ectopic BP::GUS expression in the leaves of plants treated with the auxin transport inhibitors 1-N-naphthylphthalamic acid (NPA) or 2,3,5-triiodobenzoic acid (TIBA) (Fig. 2O,P and data not shown), whereas STM::GUS expression was unaltered by these treatments (Fig. 2Q,R and data not shown). This suggests that the correct regulation of auxin transport is required to repress BP expression independently of other meristem-expressed genes.

Our results suggest that auxin and AS1 activities promote leaf fate, in part by excluding meristem-expressed BP transcripts from leaves. Previous work has shown that aberrant leaf development resulting from inappropriate KNOX expression is associated with reduced polar auxin transport and altered auxin distribution(Tsiantis et al., 1999; Scanlon et al., 2002; Zgurski et al., 2005). Thus,regulated auxin transport may be an important determinant of leaf shape. To investigate whether PIN1-directed auxin flux controls leaf shape, we compared the margin configuration of pin1 and wild-type leaves. Whereas wild-type (Col ecotype) rosette leaves have a serrated margin(Fig. 3A), pin1-En134mutants (Col ecotype) have a smooth margin(Fig. 3B), indicating that PIN1 activity promotes the development of leaf marginal serrations, hence determining leaf shape. This effect of loss of PIN1 function on the leaf margin is independent of the ectopic BP expression observed in pin1 leaves, as the margins of bp;pin double mutant leaves are indistinguishable from those of pin1 single mutants (data not shown).

To investigate whether serrations are elaborated from the leaf margin through the generation of PIN1-directed auxin maxima, similar to those driving organogenesis at the tips of emerging organs(Benkova et al., 2003), we assayed DR5rev::GFP and PIN1:GFP expression in wild-type leaf margins. DR5rev::GFP was localised in the tips of initiating serrations (Fig. 3C), and polar expression of PIN1:GFP in the epidermis indicated that these auxin maxima may be generated by PIN1-directed auxin efflux (arrowheads Fig. 3D). This observation suggests that aspects of the mechanism whereby PIN1-dependent auxin activity gradients trigger leaf initiation at the SAM may be recapitulated within the Arabidopsis leaf to determine the shape of the leaf margin. These local auxin maxima correlated with epidermal convergence points of PIN1 polarity in the leaf margin have recently been shown to control vein positioning (Scarpella et al.,2006), suggesting that a common auxin-mediated mechanism may underlie patterning of leaf venation and elaboration of leaf shape.

To further test whether local auxin activity gradients active in the developing leaf margin are required to initiate serrations, we perturbed these gradients by growing wild-type plants (Col ecotype) on NPA. Compared with the serrated leaf margin of plants grown on MS medium(Fig. 3E), a smoother leaf margin developed when these plants were grown on MS medium supplemented with NPA (Fig. 3F). Local auxin activity gradients and polar localisation of PIN1:GFP in margin cells of plants grown on MS medium (arrowheads, Fig. 3G) were abolished in NPA-grown plants and pin1 mutants(Fig. 3H,I, see also Fig. S2 in the supplementary material), demonstrating that these local gradients of auxin activity, generated by PIN1 polarity, are required for the development of a serrated wild-type leaf margin.

To test whether auxin activity in the developing leaf margin responds to ectopic KNOX expression, we examined PIN1::GUS and DR5::GUS expression in wild-type and 35S::BP leaves. In comparison with wild-type leaves (Fig. 4A,C), PIN1::GUS and DR5::GUS expression was repressed in the distal lamina and concentrated in developing lobes of 35S::BP leaves (Fig. 4B,D). Therefore, KNOX exclusion from leaves is required to establish the wild-type pattern of auxin activity gradients and PIN1expression in leaves. To further examine whether PIN1 localisation in the leaf margin is altered in response to ectopic KNOX activity, we assayed PIN1:GFP expression during the development of wild-type and dissected leaves that result from ectopically expressing BP under the control of the FILAMENTOUS FLOWER (FIL) promoter(Hay and Tsiantis, 2006). PIN1:GFP expression maxima were observed at sites of developing serrations along the wild-type leaf margin (arrow Fig. 4E), and were shifted basipetally as new serrations were initiated(arrow Fig. 4F). PIN1:GFP expression in initiating leaflets of FIL≫BP leaves was indistinguishable from that of wild type; however, expression persisted as leaflets developed in FIL≫BP leaves (arrows Fig. 4G), suggesting that KNOX activity in the leaf prevents the normal basipetal displacement of PIN1:GFP expression maxima, correlating with prolonged localised growth and leaflet formation. BP-induced alterations in PIN1:GFP expression were mirrored by similar alterations in expression of the AINTEGUMENTA (ANT)gene (arrows, Fig. 4H-J), which promotes growth in Arabidopsis lateral organs(Krizek, 1999; Mizukami and Fischer, 2000; Grandjean et al., 2004). These observations suggest that auxin-mediated reorganisation of growth at the leaf margins underpins leaflet formation in FIL≫BP leaves.

To test whether perturbation of such auxin activity gradients contributes to KNOX-dependent alterations in leaf shape, we analysed leaflet formation in FIL≫BP plants grown on MS medium(Fig. 4K) or MS medium supplemented with NPA. Strikingly, we observed that NPA treatment completely blocked leaflet initiation (Fig. 4L; a similar suppression of lobe initiation by NPA treatment was observed in 35S::BP plants, data not shown), and prevented the generation of PIN1-directed auxin maxima in the leaf margin(Fig. 4M and data not shown). The smooth margin formed in both wild-type and FIL≫BP plants as a result of NPA treatment indicates that local auxin maxima generated by polar auxin transport may stimulate the localised growth required for development of both a wild-type serrated margin, when BP is absent from the leaf,and a dissected leaf margin, when BP is ectopically expressed in the leaf. Thus, although auxin activity gradients acting in the leaf margin to control leaf shape are sensitive to ectopic BP activity, their effects on leaf shape are mediated by factors that remain unknown.

By contrast, BP antagonises PIN1 activity in leaf initiation, and the repression of BP expression in leaves requires both PIN1 and AXR1 activities (Fig. 4N),suggesting that auxin-dependent repression of BP may be a component of leaf initiation processes. Alternatively, the suppression of pin1defects observed in bp;pin1 double mutants may reflect post-transcriptional or post-translational regulation of BP, or the repression of BP-dependent processes by auxin. High-resolution studies of BP expression in the apices of wild type and pin1 mutants(e.g. Heisler et al., 2005),and the characterization of downstream targets of BP, will help to distinguish between these two possibilities that are not mutually exclusive. The partial recovery of organ initiation defects of pin1 and pid mutants in bp;pin1 and bp;pid double mutants is reminiscent of the suppression of cotyledon boundary formation and the growth defects of pin1;pid double mutants in stm;pin;pid triple mutants(Furutani et al., 2004). These genetic interactions may indicate that the antagonism between KNOX and auxin activities operates in multiple contexts throughout Arabidopsisdevelopment to promote organ initiation and the associated elaboration of organ boundaries. Taken together, these observations highlight the modular nature of auxin action, and emphasize the significance of identifying factors that contextualise auxin action in distinct tissues at different developmental stages and possibly at various auxin concentrations(Ljung et al., 2001; Xu et al., 2006). For example,the auxin response factors (ARF) ETTIN/ARF3 and ARF4 act with KANADI proteins to facilitate axial patterning of lateral organs(Pekker et al., 2005);therefore, it will be interesting to determine whether the same or different transcriptional components mediate the repression of BP by auxin.

Our results establish two novel points about developmental patterning in plants. First, we show that auxin activity, directed by PIN1-dependent fluxes,is required together with AS1 to repress BP expression and promote leaf development. Secondly, we show that PIN1 activity is required later in leaf development to control leaf shape by regulating the initiation of marginal serrations (Fig. 4N). Ectopic KNOX expression in leaves perturbs these PIN1-dependent local gradients of auxin activity, resulting in lobe or leaflet outgrowth. Both KNOX activity in leaves and auxin signalling are involved in the development of dissected leaf forms in nature (Bharathan et al., 2002; Wang et al.,2005; Hay and Tsiantis,2006); therefore, it is possible that the differential regulation of auxin activity gradients by KNOX proteins mediates natural variation in leaf form.

We thank O. Leyser, A. Hudson, Y. Eshed and S. Hake for critical reading of the manuscript. We thank J. Friml for PIN1:GFP and DR5rev::GFP seeds,O. Leyser for axr1-12 seeds, Y. Mizukami for ANT::GFP seeds,B. Scheres and I. Blilou for pin1En134 seeds, J. Traas for pin1-6 seeds, W. Werr for STM::GUS seeds, S. Hake and N. Ori for bp-9 and 35S::BP seeds, J. Craft for FIL≫BPlines, and Y. Eshed and J. Bowman for FIL::LhG4 seeds. We also thank J. Baker for photography, I. Moore for assistance with confocal microscopy and the Arabidopsis Biological Resource Center for seeds. M.T. receives support from the BBSRC and the Gatsby Charitable foundation. A.H. is the recipient of a University of Oxford Glasstone Research Fellowship and a Balliol College Junior Research Fellowship. M.B. is the recipient of a Bodossakis Foundation Award. This work was also funded by an EU MECHPLANT project.