Asymmetric leaf development and blade expansion in Arabidopsisare mediated by KANADI and YABBY activities

Most leaves are laminar structures with two distinct surfaces that are specialized for their distinct functions. Leaves are produced from the flanks of the apical meristem, and thus, there exists a fundamental positional relationship between leaves and the meristem from which they are produced. The adaxial side (ad - next to) of the leaf is derived from cells adjacent to the meristem, while the abaxial side (ab - away from) is derived from cells in the leaf primordium at a distance from the meristem. In many species, including Arabidopsis, the adaxial side differentiates to form the upper surface and is specialized for light capture, with the lower, or abaxial,specialized for respiration.

The formation of the leaf lamina requires the establishment of polarity and is proposed to be a result of interactions between juxtaposed abaxial and adaxial cells (Sussex, 1955; Waites and Hudson, 1995). The initial establishment of ad-abaxial polarity may result from a signal emanating from the apical meristem to induce or maintain adaxial identity, and in the absence of signal, abaxial identity may be the default. As communication between the abaxial and adaxial domains has been proposed to be responsible for lamina growth, factors promoting the identity of one or both domains are probably involved in directing the initial stages of lamina formation. In Arabidopsis, adaxial fates are promoted by the expression of members of the class III HD-ZIP family of genes, such as PHABULOSA (PHB), PHAVOLUTA (PHV) and REVOLUTA (REV)(McConnell and Barton, 1998; McConnell et al., 2001; Emery et al., 2003). Abaxial fates are promoted by members of two gene families, the KANADIs and the YABBYs(Sawa et al., 1999; Siegfried et al., 1999; Eshed et al., 1999; Kerstetter et al., 2001; Eshed et al., 2001). Gain-of-function alleles of some class III HD-ZIP genes (PHB, PHV)result in adaxialized radial leaves lacking lamina expansion(McConnell and Barton, 1998; McConnell et al., 2001), and loss-of-function alleles result in abaxialized radial cotyledons(Emery et al., 2003). Conversely, gain-of-function alleles of KANADI result in radial abaxialized organs (Eshed et al., 2001; Kerstetter et al., 2001). Thus, homogenization, either to all abaxial or all adaxial identities results in a loss of lamina development, consistent with the hypothesis of Waites and Hudson (Waites and Hudson,1995) that juxtaposition of these two domains is critical for lamina expansion.

Previously, members of the YABBY gene family have been proposed to promote abaxial cell fates. This hypothesis was based on the observations that gain-of-function alleles of several members, FILAMENTOUS FLOWER(FIL) and YABBY3 (YAB3) and CRABS CLAW(CRC), result in abaxial tissues differentiating in adaxial positions, namely in the cotyledon, leaf and petal epidermises(Sawa et al., 1999; Siegfried et al., 1999; Eshed et al., 1999). Conversely, loss-of-function alleles of CRC, when in combination with kan1 mutations result in adaxial tissues developing in abaxial positions in the gynoecium (Eshed et al.,1999). During leaf development FIL and YAB3 are expressed in the abaxial regions of Arabidopsis leaves and their expression patterns parallel that of the progress of leaf differentiation(Siegfried et al., 1999; Sawa et al., 1999; Kumaran et al., 2002). We present several lines of evidence that YABBY gene activity is associated with lamina expansion and propose that boundaries of YABBY gene expression marking the abaxial-adaxial boundary are intimately linked with the proposed communication between the abaxial and adaxial domains during leaf development.

All mutants were in the Landsberg erecta (Ler) background except kan3-1 and yab3-1 (both in Wassilewskija) which were backcrossed into Ler. Plants were grown under 18 hours cool white fluorescent light at 20°C. All mutant lines have been described previously: kan1-2 (Eshed et al.,1999); kan2-1, AS1>>KAN2(Eshed et al., 2001); kan3-1 (Emery et al.,2003); fil-5, yab3-1(Siegfried et al., 1999); fil-8, yab3-2 (Kumaran et al.,2002); phb-1d(McConnell and Barton, 1998); syd-2 (Wagner and Meyerowitz,2002). YJ158 was identified in an enhancer trap screen as previously described (Eshed et al.,1999).

Plants mutant for kan1, kan2, fil and yab3 were selected as follows: 25 phenotypically wild-type plants from the F₂ of kan1 kan2 × fil yab3 were self-fertilized and tested in the F₃. Families in which approximately 3/16 plants looked like either kan1 kan2 or fil yab3 (so kan2 and yab3 are homozygous) were subjected to further analysis. The rare 1/16 phenotype was considered to be the quadruple mutant, while the 3/16 plant that had trichomes on both sides of their first leaf were considered as fil yab3 kan2 kan1/+.

Total mRNA was extracted from vegetative shoot tips using the RNeasy kit(Qiagen). First strand cDNA was generated using reverse transcriptase and an oligo(dT) primer. Two degenerate primers targeting the zinc finger domain(5′ GTIACIGTIMGITGYGGICAYTG 3′) and the YABBY domain (5′GCCCARTTYTTIGCIGC 3′) were used to amplify Solanum tuberosumpartial cDNA sequences. Products were cut from the gel and TA cloned using the TOPO TA cloning kit (Invitrogen). Phylogenetic analysis including a translation of the sequence obtained from S. tuberosum and the A. thaliana YABBY gene amino acid sequences indicate S. tuberosum YABBY1 (GenBank accession no. AY495968) is orthologous to FIL/YAB3.

SEM, tissue clearing, GUS staining and in situ hybridization were carried out according to the methods of Eshed et al.(Eshed et al., 1999) and Emery et al. (Emery et al., 2003). KAN1, KAN2, KAN3, FIL, PHV and PHB probes were generated by linearizing cDNA plasmids and synthesizing DIG-labeled antisense RNA using T7 RNA polymerase. Histological analyses were carried out as described by Eshed et al. (Eshed et al., 1999) for kan1 kan2 and Emery et al. (Emery et al., 2003) for kan1 kan2 kan3.

Four closely related members of the GARP gene family, which we refer to as the KANADI genes (KAN1-4), are present in the Arabidopsisgenome, all of which are capable of inducing abaxial cell fate upon uniform expression (Eshed et al., 2001; Kerstetter et al., 2001) (Y.E. and J.L.B., unpublished observations). KAN1 is expressed in the abaxial regions of all lateral organs(Kerstetter et al., 2001). We examined whether other KANADI genes are expressed in a similar manner as KAN1. At the heart stage of embryogenesis, KAN1, KAN2 and KAN3 displayed a similar expression pattern in the abaxial basal portion of emerging cotyledon primordia(Fig. 1A-C). Based on reporter gene constructs, KAN4 appears to have a more limited expression pattern in the ovules (data not shown). The use of promoter-reporter gene fusions suggested that KAN1-3 genes also share phloem-associated vascular bundle expression (Emery et al.,2003).

Since none of the kan1, kan2 or kan3 single mutants exhibited a dramatic loss of polarity, we constructed the multiple mutant combinations kan1 kan2 and kan1 kan2 kan3. All lateral organs had gross morphological defects in kan1 kan2 plants as described previously (Eshed et al.,2001); leaves were narrow, dark green and developed ectopic outgrowths on their abaxial side (Fig. 1D,G,H), a feature never found in wild-type leaves(Fig. 1E-F). In contrast, the local blade outgrowths characteristic of the abaxial side of kan1 kan2 leaves were greatly reduced in the triple mutant, and when outgrowths occurred in the triple mutant they developed much later temporally(Fig. 1I,L). While kan1 kan2 kan3 leaves were nearly radial at initiation, these leaves still displayed some polarity, as trichomes were not present on the abaxial sides of the first leaves (Fig. 1L). In kan1 kan2 kan3 leaves the mature blade expanded in various planes giving rise to long narrow leaves with a fan-like blade at their distal end(Fig. 1M). Ectopic meristems were commonly formed at the base of the abaxial side of the phb-1dleaves, a feature that has not been observed in kan1 kan2 plants. However, ectopic meristems could occasionally be found at the abaxial base of kan1 kan2 kan3 leaves (Fig. 1J,K). The flowers of kan1 kan2 kan3 plants were similar to kan1 kan2 plants (not shown)(Eshed et al., 2001).

Anatomical differences along the abaxial/adaxial (ab/ad) axis of Arabidopsis leaves are evident shortly after leaf initiation. Wild-type leaves have distinct dense columnar palisade mesophyll cells on the adaxial side (Fig. 2A-C). The abaxial spongy mesophyll cells appear more cubic and intercelluar air spaces gradually became more prevalent. The distinct layers of mesophyll are maintained by predominantly anticlinal cell divisions(Fig. 2C). In contrast, leaf primordia of kan1 kan2 and kan1 kan2 kan3 seedlings were nearly radial (Fig. 2E,H), and only later displayed signs of polar anatomical features(Fig. 2F,I). Cells on the abaxial side of kan1 kan2 leaf primordia maintained a densely cytoplasmic appearance for a prolonged time, and periclinal divisions were common (Fig. 2F). As a consequence, kan1 kan2 leaves were 10-20 cells across, compared to only six cell layers in wild type. The excess cell divisions were not distributed equally, resulting in localized outgrowths on the abaxial side(Fig. 2F). Unlike those of the double mutant, leaves of the triple mutant remained radial until much later in development and did not exhibit the ectopic abaxial periclinal divisions characteristic of kan1 kan2 leaves(Fig. 2I).

To further characterize the polar nature of kan1 kan2 kan3 leaves,adaxial- and abaxial-specific gene expression in the mutant background was compared to that of the wild type. To assess abaxial gene expression, we assayed YAB3 expression in kan1 kan2 kan3 leaves using the yab3-2 enhancer trap line, which faithfully reproduces YAB3expression in leaves (Kumaran et al.,2002). β-glucuronidase (GUS) activity was undetectable in the kan1-2 kan2-1 kan3-1 yab3-2 background(Fig. 3B), while in an otherwise wild-type background, GUS activity was detected in the abaxial regions of developing yab3-2 leaves(Fig. 3A). Thus, loss of lamina expansion is correlated with loss of YABBY gene expression in the triple mutant.

Various members of the class III HD-zip transcription factors are expressed in regions of the apical meristem and in the adaxial regions of lateral organs(McConnell et al., 2001; Otsuga et al., 2001; Eshed et al., 2001; Emery et al., 2003). For example, in wild type, PHB mRNA was localized to the SAM and was restricted to the adaxial domain as developing primordia separated from the meristem (Fig. 3F)(McConnell et al., 2001). Later expression was confined to the provascular and vascular tissues of leaves and stems. In contrast, PHB was expressed throughout kan1 kan2 kan3 leaf primordia, although there was still a gradient with expression highest towards the adaxial side(Fig. 3C,D). In summary, loss of KANADI activity results in an adaxialization of leaves at the morphological, anatomical and molecular levels.

When KANADI was expressed uniformly throughout leaf primordia, such as in AS1>>KAN2 plants, the leaves that developed were radialized and abaxialized (Fig. 3E)(Eshed et al., 2001). To further test the relationships between KANADI and YABBY or PHB, we examined their expression patterns in these plants. While PHB was localized to the adaxial regions of wild-type leaves(Fig. 3D), no PHBexpression was detected in AS1>>KAN2 leaves, and a low level of PHB expression was still present in the apical meristem(Fig. 3G). In contrast, FIL expression was detected throughout young AS1>>KAN2leaf primordia (Fig. 3I). However, unlike wild-type leaves in which FIL expression is maintained in the marginal abaxial regions(Fig. 3H), FILexpression in AS1>>KAN2 was transient, only being detected in one leaf per apex (Fig. 3I). Thus, even though AS1>>KAN2 leaves were abaxialized, FIL expression was ephemeral. We conclude that KANADI can induce uniform FIL expression in organ primordia, but that lack of maintenance of FIL expression correlates with the limited extent of lamina formation.

Unlike kan1 kan2 floral organs, which can largely be viewed as adaxialized, the blade outgrowths on the abaxial side of leaves represent a novel phenotype. To address their nature, anatomical, morphological and molecular characterizations were carried out. The outgrowths appeared first in a row on the lower third of the leaf, with additional outgrowths continuing to initiate and expand for as long as the leaf differentiates(Fig. 1G-H, Fig. 4A). The epidermis of these outgrowths in mature leaves had a high density of stomata interspersed amongst long rectangular cells similar to those found typically at leaf margins. In wild-type seedlings, the unique leaf marginal cells exhibited specific GUS staining in the enhancer trap marker line YJ158(Fig. 4B). The outgrowths of kan1 kan2 young leaves showed GUS expression of this marker throughout their circumference (Fig. 4C), suggesting acquisition of radial blade identity.

In Arabidopsis leaves, ab/ad polarity is also evident by the arrangement of the different tissues in the vascular bundles. Xylem is located adaxially, while phloem is positioned abaxially(Fig. 4D,G). In a transverse section of kan1 kan2 leaf blades, the numerous bundles present displayed varied xylem/phloem arrangements. In most bundles, xylem elements were closer to the adaxial leaf surface as in wild type, yet, skewed orientations, in which the phloem/xylem axis is perpendicular to the ab/ad leaf axis, were quite common (not shown), possibly the result of the prolonged period of cell division in these leaves. While mesophyll cell anatomy was ambiguous, the vascular organization provided a clue to the identity of the blade outgrowths. Several vascular strands connected each outgrowth to the main leaf bundles (Fig. 4E). These strands formed enlarged, centrally located bundles composed of central xylem tissue surrounded by multiple clusters of phloem tissue(Fig. 4F,H). On the periphery of the older outgrowths, secondary vascular bundles developed with the xylem usually closer to the external surface and phloem cells closer to the phloem of the central bundle (not shown).

Similar amphicribral vascular arrangement has been found in the abaxialized radial leaves of Antirrhinum phantastica mutants(Waites and Hudson, 1995) and the abaxialized cotyledons of the phb phv rev triple mutant(Emery et al., 2003). An opposite, amphivasal arrangement was found in adaxialized phb-1d/+leaves, where a cluster of phloem tissue is surrounded by a ring of xylem(McConnell and Barton, 1998). Combined with the formation of leaf-margin-specific cell types around the blade outgrowths, the outgrowths can be viewed as ectopic filamentous leaves,consisting primarily of marginal circumference and abaxial internal tissues. These observations are consistent with the ectopic boundaries of FILexpression detected in these outgrowths as they temporally persist beyond the normal abaxial FIL expression(Fig. 4J) and are associated with prolonged cell divisions (Fig. 2F). Earlier in leaf development, FIL expression appeared to be at highest levels in the region in which outgrowths will soon develop,and at lower levels throughout the abaxial regions of the majority of the leaf(Fig. 4K).

Since the blade-like outgrowths were associated with strong and localized FIL expression, we tested whether FIL is required for their formation. While kan1 kan2 fil leaves still developed outgrowths,their formation was reduced and delayed (not shown). However, when the activity of YAB3, which is molecularly and functionally redundant with FIL (Siegfried et al.,1999; Kumaran et al.,2002) is also compromised, leaves of quadruple mutants, kan1 kan2 fil yab3, exhibited almost no traces of focal outgrowths. The two different quadruple mutant genotypes, kan1-2 kan2-1 fil-8 yab3-2 or kan1-2 kan2-1 fil-5 yab3-1, exhibited similar phenotypes; the first two leaves appeared nearly radial, with trichomes, which are normally found only adaxially on the first few leaves, present around their entire circumference (Fig. 5A). The quadruple mutant plants were greatly reduced in overall size, and organ expansion was severely decreased. Subsequently produced leaves lacked a clear plane of blade symmetry and expanded at random orientations, resulting in short and thick structures (Fig. 5B,I). As in phb-1d homozygotes, the quadruple mutant plants lacked stipules entirely, a feature seen in fil yab3 mutants as well (Fig. 5A,B). Unlike the abaxial epidermal surface of kan1 kan2 cotyledons, which has a clear abaxial appearance, the abaxial epidermis of the quadruply mutant cotyledons and leaves resembled those of the wild-type adaxial epidermis(Fig. 5C-E). In the quadruple mutant all floral organs except carpels were completely radialized, reduced in number and lacked defining cell types (Fig. 5F). Overall, these flowers closely resembled those of severely adaxialized plants homozygous for phb-1d(Fig. 5K).

Anatomical analyses of the kan1-2 kan2-1 fil-5 yab3-1 leaves revealed a dramatic loss of tissue asymmetry(Fig. 5G-I). Subepidermal cells on both the abaxial and adaxial sides resemble each other, and exhibited an overall similarity to adaxial cell types. In the first two leaves, vascular bundles were greatly reduced with normal phloem/xylem orientation. In the later produced leaves, there were 10-12 cell layers and vasculature was often lacking. As FIL is normally expressed in fil-5 mutants (data not shown), it could serve as an abaxial marker in the quadruple mutant background. Overall, FIL expression was greatly reduced, and found only at the margins of the leaves where expansion occurs(Fig. 5J).

The YABBY genes were previously suggested to promote abaxial cell fate on the basis of their expression pattern and gain-of-function alleles. Yet, the loss-of-function phenotypes in either fil or fil yab3 mutant backgrounds do not provide clear support for this assumption, as replacement of abaxial cell types by adaxial ones is limited. For example, trichomes were not found on the abaxial surfaces of fil yab3 leaves until the 5th or 6th leaf as in wild type (Fig. 5L inset). However, when KANADI activity is partially compromised the role of the YABBYs as promoters of abaxial identity is revealed. Although kan1-2/+ kan2-1 plants resembled wild type, fil-5 yab3-1 kan1-2/+ kan2-1 plants had short and narrow leaves, with trichomes on both sides of the first produced leaves(Fig. 5L).

Previous analyses of mutations of genes involved in establishing the ab/ad axis suggested an association between this axis and proper development along the proximodistal axis (Waites and Hudson,1995). While no clear morphological markers define the Arabidopsis proximodistal axis beyond petiole and blade tissues,neither the kan1 kan2 nor the fil yab3 plants have leaves that grow to the normal length. Leaves of the quadruple mutants only grew to 20% of the normal length of wild-type leaves, similar in size to homozygous phb1-d leaves (Fig. 5M), corroborating the concept that proper establishment of adaxial-abaxial and proximodistal axes are related.

SPLAYED (SYD) encodes a SWI/SNF homolog required for proper B class gene expression during flower development(Wagner and Meyerowitz, 2002). In a screen for genetic enhancers of gym kan1 mutations, we identified two alleles of syd that in the gym kan1 sydtriple mutants resulted in ovules developing from the abaxial regions of the carpels (Eshed et al., 2001). To examine whether syd enhances other aspects of the kanadimutant phenotype, we constructed the kan1-2 kan2-1 syd-2 triple mutant. The leaves of kan1 kan2 syd were radially symmetric and lacked outgrowths (Fig. 6A). Consistent with their adaxialized appearance, PHB was expressed throughout the developing leaves (Fig. 6B). FIL expression was greatly reduced in this background (Fig. 6C-D). Thus,the suppression of abaxial outgrowths by syd mutations in a kan1 kan2 background is associated with a loss of YABBY gene expression in this background.

Previously, YABBY expression was associated with lamina expansion of phb-1d/+ plants, and this association was maintained even when YABBY activity was detected on the upper side of the trumpet shaped leaves(Siegfried et al., 1999). To examine whether YABBY gene expression is tightly correlated with regions of active cell division and blade expansion in developing leaves of other species as well, we isolated YABBY gene family members from Solanum tuberosum. The leaves of S. tuberosum facilitated examination of gene expression patterns in greater detail than the small and rapidly differentiating leaves of Arabidopsis(Fig. 6E). StYABBY1 of S. tuberosum is orthologous to FIL and YAB3 of Arabidopsis and NbYABBY of Nicotiana benthamiana(Foster et al., 2002). StYABBY1 was found to be expressed abaxially in young developing potato leaf primordia. As leaves differentiate, it continues to be expressed abaxially, but only in regions that consist of cytoplasmically dense cells that presumably retain the capacity to divide(Fig. 6F). These cells are located primarily towards the margins, and along the ab-ad boundary of the leaf. In conclusion, expression of StYABBY1 appears abaxial at early stages of leaf development and parallels regions associated with extensive future lamina expansion.

Although four KANADI genes are present in the Arabidopsis genome,simultaneous inactivation of three of them is enough to transform most abaxial features of leaves into adaxial features. This transformation includes adaxialization of vasculature and mesophyll and the generation of nearly radial leaf primordia accompanied by occasional production of abaxially placed axillary buds. Compromising most of the activity of both KANADI and YABBY genes, as in the kan1-2kan2-1 fil-5 yab3-1 quadruple mutants, further reduces the polar nature of lateral organs, as evidenced by formation of trichomes on both sides of the first leaves. However, the leaves of the quadruple mutant do not display all the adaxial characteristics seen in homozygous phb1-d plants(McConnell and Barton, 1998). They are not filamentous but rather bulbous and have many more cell layers than phb-1d leaves. Expression of the other YABBY and KANADI gene family members may still be promoting abaxial identity and limited random lamina expansion in this background, or alternatively, other abaxial identity genetic programs could be invoked. Other suppressors of PHB-like activity such as the recently identified microRNAs could also account for residual abaxial identity (Reinhart et al.,2002; Tang et al.,2002). Without further understanding of the `polarity ground state' it will be difficult to predict whether residual abaxial identity stems from failure to activate adaxial factors or from activity of additional abaxial factors.

While fil yab3 double mutants do not exhibit a conspicuous loss of polarity with respect to differentiation of leaf tissues(Siegfried et al., 1999; Kumaran et al., 2002), we show that FIL and YAB3 contribute to polar differentiation of tissues in the leaf when in the context of a kan1/+ kan2 background. Trichomes develop on the abaxial sides of the first leaves in fil yab3 kan1/+ kan2 plants, but not in kan1/+ kan2 plants, indicating that FIL/YAB3 contribute to the differentiation of cell types on the abaxial sides of these leaves. Thus, these loss-of-function phenotypes are consistent with the hypothesis, based on gain-of-function alelles, that YABBY gene activity promotes abaxial cell fates(Sawa et al., 1999; Siegfried et al., 1999).

The establishment of polarity is required for lamina expansion, a critical process in the development of nearly all lateral organs. Based on elegant genetic and classic dissection experiments it has been proposed that a juxtaposition of adaxial and abaxial domains is required for lamina development (Sussex, 1955; Waites and Hudson, 1995). Several lines of evidence lead us to propose that a major component of the genetic program directing lamina expansion in leaves is the activity of members of the YABBY gene family. As outlined below, YABBY gene expression is correlated with lamina expansion in several different contexts, ectopic YABBY gene expression leads to ectopic growth, and loss of YABBY gene activity results in a loss of lamina expansion.

YABBY gene expression patterns mirror distributions of cytoplasmically dense cells competent to undergo cell division. In Arabidopsis, FILand YAB3 are initially expressed throughout the abaxial regions of leaves, and later, expression becomes restricted to the laminar marginal domains as the central abaxial domain differentiates into vacuolated cells. Expression ceases in a basipetal manner, paralleling the differentiation of leaf cells and preceding the progression in cell division distributions(Donnelly et al., 1999; Sawa et al., 1999; Siegfried et al., 1999; Kumaran et al., 2002). In Solanum tuberosum, the correlation with densely cytoplasmic cells is readily observed, as YABBY expression becomes restricted to the marginal domains of the abaxial-adaxial boundary. Given that the YABBY-expressing cells represent a two dimensional sheet of cells, originally throughout the leaf,but then becoming restricted to two marginal regions of the differentiating leaf, it would follow that lamina expansion along the proximodistal axis would also be coupled to establishment of the abaxial-adaxial axis. Thus, in each of these cases YABBY gene expression patterns are consistent with activity being linked to lamina expansion.

The reduction in lamina growth, most markedly in the floral organs of fil yab3 plants is also consistent with YABBY gene activity contributing to lamina expansion(Siegfried et al., 1999). Some lamina expansion does occur in fil yab3 leaves, but this limited lamina development may be due to two other members of the YABBY gene family, YAB2 and YAB5, which are expressed in leaves(Siegfried et al., 1999). Characterization of plants lacking all YABBY activity will be required to address whether YABBY activity is absolutely required for leaf lamina development. The outer integument is a lateral organ in which expression of only a single YABBY gene family member, INNER NO OUTER (INO)has been detected (Villanueva et al.,1999). In this case, INO is expressed in the abaxial cell layer of the outer integument(Balasubramanian and Schneitz,2002; Meister et al.,2002) and loss of INO activity results in a complete loss of lamina expansion of the outer integument. Thus, loss-of-function alleles of YABBY gene family members are also consistent with a role of YABBY gene expression promoting lamina expansion.

The model of leaf blade development by Waites and Hudson(Waites and Hudson, 1995)proposed that juxtaposition of abaxial and adaxial domains is required for lamina outgrowth. Leaves of kan1 kan2 plants are mosaic, having both ectopic adaxial and abaxial characteristics. Young leaf primordia in the double mutant are nearly radial, with spatially expanded expression domains of adaxial promoting genes (Eshed et al.,2001). Laminar expansion in these leaves occurs largely as a result of the low level of properly localized YABBY genes since this expansion is lost in the kan1 kan2 fil yab3 quadruple mutant. However, as kan1 kan2 leaves differentiate, reactivation of YABBY genes occurs within specific abaxial foci by a presently unexplained mechanism. These foci develop as blade-like outgrowths with abaxial fates based on anatomical characters and vascular organization. YABBY activity is required for lamina expansion associated with the focal outgrowths since this growth disappears in the kan1 kan2 fil yab3 background, providing positive evidence that YABBY activity can induce lamina expansion.

In kan1 kan2 leaves, three levels of YABBY gene expression are evident: the adaxial domain has no expression, the abaxial domain has low levels, and the abaxial foci have high levels. Lamina expansion is correlated with the boundaries between each of these expression domains, suggesting that relative levels, rather than absolute levels, of YABBY activity could be responsible for the blade growth in these leaves. The abnormal thickness of kan1 kan2 leaves can also be seen in a context of prolonged boundaries of YABBY gene expression leading to continued periclinal cell divisions in the abaxial regions of developing leaves. Such boundaries still exist even in kan1 kan2 fil yab3 quadruple mutants, as reflected by FIL expression, and therefore thickness of these leaves could also be attributable to residual YABBY activity, possibly mediated by YAB2and YAB5. However, when no boundaries (as assayed by FIL)are present, such as in phb-1d, and kan1 kan2 syd leaves,abnormal leaf thickness is not present.

In a phb-1d background, radial adaxialized leaves are produced and FIL is not expressed at detectable levels in the leaves(Siegfried et al., 1999). Thus, loss of YABBY activity is associated with loss of lamina expansion in this background. In contrast to the adaxialized leaves of the above genotypes, AS1>>KAN2 plants produce radial abaxialized leaves. If YABBY activity was solely associated with abaxial cell type specification, one might expect YABBY gene expression to be prominent. However, in AS1>>KAN2 leaves FIL expression is transient,consistent with the lack of lamina expansion in this genotype. YABBY gene family expression does not necessarily have to be localized to the abaxial domain to effect lamina expansion. For example, in phb-1dheterozygotes, peltate leaves, in which cells with abaxial identity are inside the cup, commonly develop (McConnell and Barton, 1998). In this case, YABBY gene expression is associated with expansion of the cup and is located at the tip of the developing leaf rather than the abaxial domain (Siegfried et al., 1999). Perhaps more remarkably, FIL is expressed adaxially in petal loss pistallata second whorl floral organs that are morphologically inverted (Siegfried et al., 1999; Griffith et al.,1999). As these organs display normal blade expansion, YABBY gene expression can still exert its role in lamina expansion as long as expression remains polar.

Thus, in each case examined, YABBY gene activity is associated with regions of lamina expansion, and is necessary in at least some contexts to promote lamina development. In these cases it appears that it is a boundary of YABBY gene activity that is associated with lamina formation. Unlike many genes whose products will promote growth per se, via cell division or cell expansion, we suggest that boundaries of YABBY activity act to promote cell division indirectly in regions spanning both sides of the boundary. This would imply that YABBY activity is signaling between domains, and consistent with this hypothesis ectopic expression of FIL or YAB3 results in dramatic non-autonomous phenotypic effects (Y.E. and J.L.B., unpublished observations). Determination of whether YABBY gene activity is sufficient to induce lamina expansion will require the analysis of genetic chimeric plants in which sectors of YABBY gene expression are produced in fields of cells lacking YABBY activity.

The proposal that boundaries of YABBY gene expression act as mediators of zones of lamina expansion has elements in common with the paradigm of boundaries acting as organizing centers in animal development(Lawrence and Struhl, 1996; Basler, 2000; Irvine and Rauskolb, 2001). In such a developmental paradigm an initial polarity leads to asymmetric short-range signaling creating a third type of cell at or near the boundary that produces long-range organizing signals that pattern the differentiation of surrounding tissues. Such systems provide a powerful mechanism to generate complex patterns from an initial simple asymmetry. While no signaling molecules that pattern differentiation within developing leaves have yet been identified, boundaries of YABBY gene activity appear to promote organ growth in adjacent cells and thus acting as a focus for lamina formation.

We thank John Alvarez, John Emery and Nathaniel Hawker for valuable discussions and Rick Harris for technical help. Supported byNSF grants IBN 9986054 and IBN 0077984 (J.L.B.); Research Grant award No. 3328-02 from BARD(Y.E. and J.L.B.) and Postdoctoral Award FI-314-2001 from BARD (A.I.).