A genetic framework for fruit patterning in Arabidopsis thaliana

Remarkably, while multicellularity arose independently during the evolution of plants and animals, the processes that pattern organs in these two lineages are similar. In both lineages, organs such as limbs in animals and lateral organs in plants are patterned by two parallel systems: one system defines the type of organ that will develop, while the second system defines the coordinates of the organ (Engstrom et al.,2004; Logan,2003). While the combination of factors that confer organ identity tend to be unique for individual organ types, the patterning system that defines the coordinates of the organ tends to be the same for all organs.

The development of the fruit in the model plant, Arabidopsis,provides an excellent system for dissecting the mechanisms that pattern an organ in plants because of the presence of distinctive morphological landmarks and the availability of reporter lines that mark specialized tissues(Dinneny and Yanofsky, 2005). The region of the fruit that encloses the seeds (ovary) is divided into three tissue zones: valve, replum and valve margin(Fig. 1A-D). The valves, or seed pod walls, protect the seeds during their development and detach after maturation to promote seed dispersal in a process known as dehiscence. The two valves are separated by a central ridge of replum tissue. At the valve/replum junction, a specialized stripe of tissue, termed the valve margin, forms that facilitates the detachment of the valves from the replum through the action of two different cell types. On the replum side of the valve margin, the separation layer forms, which permits the detachment of the valve from the replum through cell-cell separation (Fig. 1E) (Jenkins, 1999; Peterson, 1996). On the valve side of the margin, a layer of rigid lignified cells forms(Fig. 1E,F). The lignified layer of the valve margin is continuous with an adjacent layer of lignified cells present in the valves, termed the endocarp layer b(enb) (Fig. 1E-G). Together, these tissues are thought to provide spring-like tension that mechanically drives valve detachment (Spence, 1996).

Recent work has identified several genetic components that are important for specifying valve margin development and defining its borders(Fig. 7). The redundant MADS-box genes, SHATTERPROOF 1/2 (SHP), are at the top of a cascade of transcription factor genes that specify valve margin identity(Liljegren et al., 2000). When SHP activity is eliminated, fruits are indehiscent and lack lignified layer and separation layer development. Acting both downstream and in parallel to SHP, the INDEHISCENT (IND) and ALCATRAZ(ALC) basic helix-loop-helix-type transcription factor genes also specify valve margin identity with IND playing important roles in lignified and separation layer development, and ALC regulating separation layer development (Liljegren et al., 2004; Rajani and Sundaresan, 2001).

The expression of the valve margin identity genes is limited to the valve margin through negative regulation by FRUITFULL (FUL) in the valves and REPLUMLESS (RPL) in the replum. FULencodes a MADS-domain transcription factor that is expressed in the valves(Gu et al., 1998). When FUL is mutated, the valve margin identity genes become ectopically expressed in the valves, imparting valve margin-like development, including the ectopic formation of lignified and separation layer-like cell types(Ferrándiz et al.,2000b). Similar to the role that FUL plays, RPLnegatively regulates the expression of the valve margin identity genes in the replum (Roeder et al., 2003). RPL encodes a BELL-family homeodomain transcription factor and is expressed in the replum. When RPL is mutated, ectopic SHPexpression in the replum causes the valve margins to coalesce into a single stripe of tissue in place of the replum, rendering rpl fruit partially indehiscent. SHP expression is positively regulated by AGAMOUS (AG), a MADS-domain transcription factor gene that controls carpel identity (Savidge et al.,1995). However, AG probably acts redundantly with other factors, since SHP activity is still present in ag apetala2 mutants (Lee et al., 2005; Pinyopich et al., 2003).

A crucial aspect of fruit development that is not well understood is how the patterns of gene activities that control valve margin formation are initially established. Here, we show that the FILAMENTOUS FLOWER(FIL) and YABBY3 (YAB3) genes, which regulate the polarity of tissues in lateral organs(Eshed et al., 2004; Sawa et al., 1999; Siegfried et al., 1999), are required to promote the expression of FUL and SHP in the valves and valve margin, respectively. The unrelated gene, JAGGED(JAG), which promotes the growth of tissues in lateral organs(Dinneny et al., 2004; Ohno et al., 2004), acts redundantly with FIL and YAB3 to promote the expression of FUL and SHP, with jag fil yab3 triple mutants lacking FUL and SHP expression in the valves or valve margins. We also provide further insight into the mechanism by which RPL regulates valve margin development by showing that RPLpromotes the formation of two stripes of valve margin by dividing JAG/FIL activities into two separate domains. Our work describes the first functional link between patterning systems in plants that define organ polarity and growth and those that control tissue identity, and provides a genetic framework for the patterning of the three tissue types composing the fruit.

Landsberg erecta [Ler] or a wild-type segregant that is mutant for erecta when tissue is stained for GUS activity was the wild type used. fil-8, yab3-2 (Kumaran et al., 2002), jag-1, jag-5D(Dinneny et al., 2004),rpl-3 (Roeder et al.,2003), shp1, shp2(Liljegren et al., 2000),ful-1 (Gu et al., 1998), 35S::FUL (Ferrándiz et al., 2000b) and SHP2::GUS (Savidge, 1995) have been described before.

Genotyping of jag-1 has been previously described(Dinneny et al., 2004). To genotype fil-8 we used primers oJD174 and oJD175 to detect the wild-type FIL allele and oJD175 and oJD176 to detect the fil-8 allele. To genotype yab3-2, we used primers oJD172 and oJD173 to detect the wild-type YAB3 allele and oJD173 and oJD177 to detect the yab3-2 allele. Genotyping of ful-1(Gu et al., 1998) and rpl-3 (Roeder et al.,2003) has already been described. The jag-5D^+/–shp1,2 triple mutant was identified phenotypically.

The sequences of the oligonucleotide primers used were as follows:

• oJD172, 5′-GAC TCC ATG TCG AGC ATG TCC ATG-3′;

• oJD173, 5′-CTA GTA ATT CAT GTG AAA CCC TAG C-3′;

• oJD174, 5′-GAC AAT GAA AGT AGT TTT AAT GAG-3′;

• oJD175, 5′-CAG CTT TGA TAC GTT GGA TCT CCT CC-3′;

• oJD176, 5′-ACC GTT ACG ACC GTT TTC ATC C-3′; and

• oJD177, 5′-ACG GTC GGG AAA CTA GCT CTA C-3′.

In situ hybridization was performed as previously described(Dinneny et al., 2004). Tissue was prepared for scanning electron microscopy (SEM) as previously described(Dinneny et al., 2004). Late stage 17 fruit were fixed, sectioned (8 μm) and stained using Phloroglucinol (Liljegren et al.,2000) or Saffranin O and Alcian Blue(Roeder et al., 2003) as previously described.

To monitor FUL expression, the ful-1 enhancer trap line was crossed to fil-8 yab3-2 or jag-1 fil-8 yab3-2^+/– mutants. F₂ segregants were genotyped for jag-1, fil-8, yab3-2 and ful-1 alleles and stained for GUS activity if plants were heterozygous for ful-1. The yab3-2 allele is also a GUS enhancer trap line, however, no overlapping GUS staining was detected with that of ful-1 in fil yab3 and jag fil yab3 mutant backgrounds during the stages of development described (data not shown). To monitor SHP2 expression,the SHP2::GUS reporter line was crossed to fil-8 yab3-2, jag-1 fil-8 yab3-2 and jag-5D mutant backgrounds. F₂segregants were genotyped for jag-1, fil-8, yab3-2 alleles and stained for GUS activity. Plants heterozygous for jag-5D were identified phenotypically. To monitor AG expression, the KB9reporter line was crossed to fil-8 yab3-2 mutants(Busch et al., 1999). GUS staining was performed as previously described(Blázquez et al.,1997).

To identify factors required for the establishment of fruit patterning we examined genes that are expressed early during gynoecium development. One candidate was the YABBY family transcription factor gene, FIL, which is expressed strongly during floral stages 6-11 in the gynoecium, in the valves and in cells that will probably contribute to the formation of the valve margin (Fig. 2A) (Sawa et al.,1999; Siegfried et al.,1999). YAB3, the next closest homologue of FILin Arabidopsis, is expressed in a similar manner(Siegfried et al., 1999). Previous studies have identified FIL and YAB3 as downstream facilitators of the pathways that establish organ polarity(Eshed et al., 2004; Sawa et al., 1999; Siegfried et al., 1999). Lateral organs in plants develop bilateral symmetry with the half of the organ facing the shoot meristem, termed the adaxial side, while the half facing away from the shoot meristem is termed the abaxial side. All lateral organs develop distinct cell types that distinguish the adaxial and abaxial sides. In leaves, FIL expression is limited to the abaxial tissue layers, while in carpels, expression extends one cell layer adaxially to include the enb layer (Fig. 2A)(Siegfried et al., 1999). To determine whether FIL or YAB3 genes play a role in patterning the tissues of the fruit, we examined mature fruit of filand yab3 single mutants but did not observe any major defects in dehiscence (data not shown). Since other work has shown that FIL and YAB3 play redundant roles in leaf development, we examined fruit from fil yab3 double mutants and found that they were largely indehiscent(Fig. S1A,B in supplementary material). The lack of dehiscence is most apparent in the apical regions of fruit, with some dehiscence occurring in the basal region (Fig. S1C in supplementary material). A closer examination of the fil yab3 fruit, by scanning electron microscopy (SEM), revealed that,whereas the apical region of the fruit appeared to lack valve margins, the basal region appeared to develop regions with ectopic valve margin-like tissues (Fig. 2B, see below).

Consistent with the reduction in dehiscence of fil yab3 fruits,the apical region lacks valve margin development. Whereas wild-type fruit develop clear creases on either side of the replum, where the smaller valve margin cells develop (Fig. 1B),these creases are absent in fil yab3 fruit(Fig. 2B,C). We took a closer look at the valve margin region of fil yab3 fruits in cross section and stained for the presence lignified cells. In wild-type fruit, stripes of lignified cells develop on either side of the replum(Fig. 1F), however, in the apical region of fil yab3 fruit these lignified cells are absent(Fig. 2D). Similar to our results for the lignified layer, we found that the apical region of fil yab3 fruit lacked a separation layer at the valve margin region(Fig. 1E and S1D). Thus, the two cell types that compose the valve margin are absent in the apical region of fil yab3 fruit.

The loss of valve margin development in fil yab3 fruit suggests that FIL and YAB3 have important roles in controlling the expression of the valve margin identity genes. To determine this relationship,we examined the activity of a SHP2::GUS reporter. Consistent with the loss of valve margin development in the apical region, reporter activity was absent from the valve margins of fil yab3 mutants but was unaffected in regions where FIL and YAB3 are not expressed(Fig. 3A-D). Thus, FILand YAB3 play important roles in promoting valve margin development,in part by promoting SHP expression at the valve margin.

Paradoxically, fil yab3 mutants develop ectopic valve margin-associated cell types in the basal region of the fruit. By SEM, ectopic valve margin can be seen as patches, or stripes, of small cells(Fig. 2B,F). In cross section these patches stain similarly to cells of the separation and lignified layers(Fig. 2G,H and Fig. S1E in supplementary material). Ectopic valve margin in fil yab3 mutants can develop both as an expansion of the valve margin normally present on either side of replum and in stripes in the middle of the valves(Fig. 2F,G,H and Fig. S1E in supplementary material).

The development of the ectopic valve margin suggests that SHPexpression might have expanded into the valves at the base of fil yab3 fruit. We examined SHP2::GUS activity in fil yab3mutants and found that, in contrast to the apical region, reporter activity in the basal region was increased in the valves(Fig. 3A-E). The stark contrast in reporter activity between the apical and basal regions of fil yab3fruit is most dramatic in whole mounts(Fig. 3A,B).

The development of ectopic patches of valve margin tissue at the base of fil yab3 fruits is reminiscent of the ful mutant(Ferrándiz et al.,2000b). We used the ful-1 enhancer trap line to examine FUL expression in fil yab3 mutants and found that reporter activity was missing from valves in both the apical and basal regions(Fig. 3F-I). Occasionally a small patch of cells near the base of the valves showed reporter activity(data not shown). Reporter activity is present but elevated in the vasculature of the style, replum and gynophore where FUL is normally expressed,but FIL and YAB3 are not. These data show that FILand YAB3 are important for promoting FUL expression in the valves and suggest that the ectopic expression of SHP2 at the base of fil yab3 fruits is an indirect effect, caused by the loss of FUL expression. This hypothesis is supported by the finding that constitutive expression of FUL can suppress most of the ectopic lignification that occurs at the base of fil yab3 fruit(Fig. 2H, Fig. 3J,K).

We examined the expression of SHP2 and FUL during early stages of gynoecium development when FIL and YAB3 are most strongly expressed. In wild type, the SHP2::GUS reporter is initially expressed in a broad domain during early gynoecium development in cells that will develop into the replum, septum and valve margins (Fig. S3 in supplementary material). Expression in the replum eventually fades, leaving SHP2 expression at the valve margins and septum. In fil yab3mutants, SHP2::GUS expression initiates as in wild type, however,expression is quickly lost in the replum and presumptive valve margins (Fig. S2 supplementary material).

FUL expression, in wild-type gynoecia, initiates in the adaxial cell layers of the valves, and then expands into all cell layers by stage 10. In fil yab3 mutants, however, expression is never detected in the valves. These data show that FIL and YAB3 are required for expression of SHP2 and FUL during early gynoecium development in addition to later stages in the fruit. The observation that SHP2 and FUL expression is affected in tissues where FIL and YAB3 are not expressed (i.e. replum for SHP2 and adaxial epidermis for FUL) suggests that they can act cell non-autonomously.

In fil yab3 mutants, enb lignification is reduced in the apical region of the fruit (Fig. 1F,G and Fig. 2D,E). The development of the enb layer is redundantly specified by both FUL and the valve margin identity genes(Liljegren et al., 2004). Expression of at least one of these factors is sufficient for enblignification: consistent with this scenario, a 35S::FUL transgene restored enb lignification in the apical region of fil yab3mutants (Fig. 2D,E and Fig. 3J,L). Since a complete loss of enb lignification is otherwise only observed in the ful shp1 shp2 ind alc quintuple mutant, we conclude that besides FULand SHP, the activity of IND and ALC is also compromised in fil yab3 mutants. To test this possibility further, we examined the effect that loss of FIL and YAB3 function had on the ectopic valve margin development seen in ful mutants(Fig. 3M). The apical phenotype of ful fil yab3 fruit is very similar to fil yab3 double mutants, with a clear absence of valve margin development and enblayer lignification, indicating that the fil yab3 mutations are epistatic to ful (Fig. 2E and Fig. 3M,N)and necessary for the activity of the genes that confer ectopic valve margin identity to ful valves. Importantly, the base of ful fil yab3 fruit still develop ectopic valve margin, however, the ectopic tissue appears to be composed only of cells with separation layer-like characteristics (not shown).

Our data show that FIL and YAB3 promote the expression of both SHP and FUL. However, while FIL and YAB3 are essential for FUL expression in the valves, SHP expression is only partially affected. One possible reason for this incomplete effect may be genetic redundancy of FIL and YAB3 with another factor that also promotes SHP expression. We have tested whether FIL and YAB3 act redundantly with other non-YABBY family members and in other work have found that FIL and YAB3 function redundantly with an unrelated C₂H₂ zinc-finger transcription factor gene, JAG, to promote leaf blade growth (Fig. S3A,B in the supplementary material). To determine if this redundancy extends into the fruit we examined the effect of various combinations of fil, yab3 and jagmutations.

Like FIL and YAB3, JAG is expressed during the early stages of gynoecium development in cells of the valves and presumptive valve margin (Fig. 4A)(Dinneny et al., 2004; Ohno et al., 2004). Unlike FIL and YAB3, however, JAG is expressed in all tissue layers of the gynoecium, between stage 6 and stage 8(Fig. 4A), at which point JAG expression diminishes in the valves(Dinneny et al., 2004; Ohno et al., 2004). We have previously reported that jag mutants develop slightly bumpy fruit(Dinneny et al., 2004; Ohno et al., 2004). Upon closer inspection we noticed that the shoulders of the valves in the apical region of the fruit are somewhat sloped down(Fig. 4B). This phenotype is particularly apparent in the presence of the erecta mutation.

The combination of jag with fil results in the apparent expansion of the basal gynophore and apical style tissues into the ovary region (Fig. 4C). Furthermore, jag fil mutants develop a stripe of ectopic tissue that runs through the center of the valves (Fig. 4D). The cells in this region are narrow and can occasionally be seen separating from each other. Examination of cross sections of this stripe shows that it is composed of cells with separation and lignified layer characteristics, indicating that jag fil mutants develop an ectopic stripe of valve margin in the middle of the valves(Fig. 4E). Interestingly, this stripe of valve margin always develops overlying the main vascular bundle of the valves. The ectopic development of valve margin tissues also correlated with the expansion of SHP2::GUS activity into the valves of jag fil double mutants (Fig. 5F,J). Some ectopic reporter activity can be seen in the valves of jag and fil single mutants as well(Fig. 5F,G,H).

The ectopic development of valve margin cell types in the valves suggests that FUL activity might be compromised in jag fil mutants. Consistent with this hypothesis, we found that jag fil mutants develop a stripe of tissue in the valves in which FUL expression is not detected (Fig. S4A,C in supplementary material). We also examined the expression of FUL in jag single mutants and could also see a small cleft of FUL-negative tissue near the apical shoulder of the valves (Fig. S4A,B in supplementary material), suggesting that the downward sloping of the valves in the mature fruit may be caused by a loss of FUL expression.

The effect of the yab3 mutation was next examined in the context of the jag and fil mutations. Loss of a single copy of YAB3 resulted in a strong increase in the expansion of gynophore and style regions into the ovary of jag fil yab3^+/–fruit (Fig. 4F). The increase in the size of the style was particularly apparent when the fruits were not mutant for ERECTA (Fig. 4G). The fruit of jag fil yab3^+/– also developed a dramatic expansion of valve margin tissues, with large patches forming in and around the valves (Fig. 4I). The replum of jag fil yab3^+/– fruit was also greatly expanded and twisted along the length of the fruit(Fig. 4H). These phenotypes are reminiscent of those observed in ful mutant fruit, which develop enlarged and twisted repla (Fig. 4J) along with ectopic valve margin tissue in the valves(Fig. 4K), although to a greater extent than jag fil yab3^+/–. Consistent with these phenotypes, the expression of the SHP2::GUS reporter further expands into the valve regions in jag fil yab3^+/–fruit (Fig. 5K,L), with FUL expression diminishing to small islands(Fig. 5A,B). Together, these data indicate that JAG acts redundantly with FIL and YAB3 to promote FUL expression in the valves. Importantly, fil yab3^+/– mutants, which can develop stripes of ectopic valve margin in the valves, do not develop ectopic valve margin or ectopic SHP2::GUS activity to the extent of jag fil yab3^+/– mutant fruit(Fig. 5I and data not shown).

Surprisingly, the loss of the second copy of YAB3 in jag fil yab3 triple mutants reverses the expansion of valve margin development seen in jag fil and jag fil yab^+/– mutants. Fruits of jag fil yab3 mutants do not develop valve margin-like cell types in either the apical or basal portions(Fig. 4M-P and data not shown). In the valve regions, the epidermis is composed of elongated cells with interspersed stomata, similar to wild-type valve cells(Fig. 4N). Some valve cells have an irregular shape. The expression of FUL is completely absent from the valve regions of jag fil yab3 mutant gynoecia(Fig. 5C,D), similar to fil yab3 mutants. Contrary to what was observed in jag fil yab3^+/– fruit, however, SHP2::GUS activity is strongly reduced in both the valves and valve margins(Fig. 5M,N). A low level of SHP2 expression occurs only in the vicinity of the vascular bundles(Fig. 5M,N). These data indicate that JAG not only acts redundantly with FIL and YAB3 to promote FUL expression, but also to promote SHP expression. Furthermore, the strong reduction of SHPexpression in jag fil yab3 mutants demonstrates that genetic redundancy with JAG is a cause of the remaining SHPexpression seen in fil yab3 mutants. The other floral organs of jag fil yab3 mutants are also severely affected and develop little,if any floral characteristics, suggesting that JAG, FIL and YAB3 promote the patterning of other floral organs as well(Fig. 4L).

The data presented thus far show that JAG, FIL and YAB3act redundantly to promote the expression of FUL in the valves and SHP in the valve margins. Presumably, the domains of FUL and SHP expression are limited to the valves and valve margins,respectively, because the expression domains of JAG, FIL and YAB3 are limited to these tissues. To test this hypothesis, we wanted to determine if ectopic expression of JAG, FIL or YAB3 could promote the ectopic development of valve or valve margin-like tissues. We used the dominant activation-tagged allele, jag-5D, in which JAGexpression is expanded into the replum (Fig. S5A in supplementary material)(Dinneny et al., 2004). When we examined the fruits of jag-5D mutants we found that the outer replum,which normally develops in between two stripes of valve margin, was apparently missing (Fig. 6A,B). Cross sections of the fruit showed that the replum had been replaced with tissues characteristic of the separation and lignified layers of the valve margin. In strongly affected regions of jag-5D fruit, the normally separate lignified layers fused, forming a `lignified bridge' that joined the two valves (Fig. 6B). Consistent with this observation, fruit of jag-5D mutants are partially indehiscent (data not shown).

The ectopic development of valve margin tissues in jag-5D mutants suggested that the presence of JAG activity in the replum might be able to promote the expansion of SHP expression into this region. We therefore examined SHP2::GUS activity in jag-5D mutants, and found that reporter activity had expanded into the replum(Fig. 3C and Fig. 6C). To test whether expanded SHP expression was the cause of the ectopic valve margin development, we removed SHP activity by constructing a jag-5D shp1 shp2 triple mutant and found that replum development was rescued(Fig. 6A,B,D,E). Thus, JAG is sufficient to drive the ectopic expression of SHP in the replum, converting this region into valve margin.

The conversion of the replum into valve margin in jag-5D is very similar to the effect that the rpl mutation has on replum development(Fig. 6F)(Roeder et al., 2003). As in jag-5D, rpl mutants develop ectopic separation and lignified layer tissues in the replum, which is caused by the expansion of SHPexpression into the replum. Likewise, loss of SHP function in a rpl mutant background rescues replum development. These similarities suggest the basis of the rpl mutant phenotype may be ectopic JAG expression in the replum. To test this, we examined JAGexpression in rpl mutants by in situ hybridization, however, we could not detect JAG transcript in the replum(Fig. 6G). Nevertheless, we examined the effect of removing JAG activity from a rplmutant background by constructing jag rpl double mutants and found that replum development was partially rescued(Fig. 6F,H,I). The rescue of replum development is somewhat variable, however (data not shown). Thus, while expression of JAG can not be detected in the replum of rplmutants, genetic evidence suggests some JAG activity may be present.

Since JAG and FIL play similar roles in patterning the fruit, we wanted to determine if FIL was also regulated by RPL in the replum. We examined FIL expression by in situ hybridization in rpl mutants and found a clear expansion of FIL expression into the replum region(Fig. 2A, Fig. 6J). This expansion was particularly apparent during floral stages 7-9. To determine if this ectopic FIL expression was a cause of the rpl phenotype, we constructed fil rpl double mutants. The loss of FIL function had a strong effect in rescuing the rpl phenotype with fil rpl fruit consistently developing large repla(Fig. 6F,K,L).

We also examined the expression of FIL in jag-5D mutants to see if the replumless phenotype could be caused by ectopic FILexpression in the replum. No effect on FIL expression was apparent in jag-5D gynoecia, however, indicating that JAG is not sufficient to activate FIL expression in these tissues (Fig. S5B in supplementary material). Together, our data suggest that the effect of RPL on SHP is indirect, and mediated by repression of FIL and JAG, two activators of SHP expression. This hypothesis is consistent with in situ hybridization experiments that show RPL expression in the replum during the same stages that FILexpression expands into the replum region of rpl mutants. Our data,however, do not exclude the possibility that RPL may repress SHP expression through a parallel pathway as well.

The initiation of lateral organ development at the shoot apex sets off a series of patterning events that result in the formation of a mature organ with a unique set of tissues. In plants, one of the first events to take place in organ development is the establishment of polarity, which helps to determine the arrangement of tissues in the organ. In this work, we have determined that FIL and YAB3 play an important role translating this polarity information into the regulation of genes necessary for fruit patterning. FIL and YAB3 are expressed early during the initial stages of gynoecium development in cells of the valve and presumptive valve margin and promote the expression of FUL and SHP in these tissues, respectively. FIL and YAB3regulate FUL and SHP expression in cooperation with another unrelated gene, JAG. The involvement of JAG, which was originally identified by its role in promoting growth during organ development, suggests that multiple developmental pathways may converge to regulate FUL and SHP expression. We have also shown that JAG and FIL activity is negatively regulated by RPLin the replum, dividing JAG/FIL activity into two separate domains corresponding to the two valves of the fruit. Together, these data provide a mechanistic basis for the patterning of the valves, valve margin and replum and the activation of FUL and SHP, which control the identity of tissues in these regions.

Plant organogenesis is regulated by a genetic system that divides organs into abaxial and adaxial halves. Members of a miRNA-regulated clade of HD-ZIP family transcription factors, including PHABULOSA, PHAVOLUTA and REVOLUTA, control the identity of the adaxial domain, the half of the organ that is proximal to the shoot meristem(Emery et al., 2003; McConnell and Barton, 1998; McConnell et al., 2001). On the other side, the GARP-type transcription factor genes, KANADI 1and 2 control the identity of the abaxial domain(Eshed et al., 1999; Eshed et al., 2001; Eshed et al., 2004; Kerstetter et al., 2001). Not only do these factors impose adaxial or abaxial development on their respective sides, but they also restrict the spread of the opposing identity. FIL and YAB3 also function with KAN genes to control abaxial identity, however, they do not appear to be involved in the initial establishment of organ polarity. Instead it has been proposed that FIL and YAB3 function downstream of KAN to facilitate abaxial identity and to promote the growth of the leaf blade(Eshed et al., 2004). Thus, FIL and YAB3 may function downstream of polarity establishment to begin carrying out an abaxial-specific developmental program.

Although it is known that the establishment of organ polarity controls the distribution of cell types in an organ, it is not yet known what the functional relationship is between the specification of adaxial/abaxial identity and the development of specialized cell types that constitute these regions. By showing that FIL and YAB3 promote the expression of the FUL and SHP genes, our work provides the first functional link between the establishment of organ polarity and the regulation of genes that control tissue identity. Because the FIL and YAB3 genes are expressed in all lateral organs, it is likely that they control tissue identity in the fruit by interacting with other factors,such as those encoded by the floral homeotic genes. Thus, FIL and YAB3 would provide positional information that is common to all organs, with the floral homeotic genes contributing identity information specific to individual organs. Importantly, we have found that the expression of the floral homeotic gene, AG, which regulates the identity of the carpels, is unaffected in fil yab3 mutants (Fig. S6 in supplementary material). This indicates that FIL/YAB and AG represent independent pathways regulating FUL and SHP expression.

The fil yab3 double mutants and jag fil yab3 triple mutants also have defects in patterning all other floral organs(Siegfried et al., 1999). The defects seen in these other organs do not resemble the patterning defects of mutants, such as phb-1D or kan1 kan2, that have been classified as having ectopic adaxial development(Eshed et al., 2001; McConnell and Barton, 1998). It will be interesting to examine these floral organs more carefully to determine the exact nature of the patterning defects and to see if specific floral organ cell types are missing, as they are in the fruit.

While FIL, YAB3 and JAG are expressed in both the valves and presumptive valve margin, FUL and SHP are expressed in mutually exclusive domains in these tissues. How is it then that FIL,YAB3 and JAG pattern the expression of both sets of genes? Our loss-of-function genetic results show that FUL expression is most severely affected by reductions in FIL/YAB3/JAG activity whereas SHP expression is more robust. For example, FUL expression is strongly affected in fil yab3 mutants, whereas SHPexpression is lost in only part of the fruit. These data suggest that the activation of FUL and SHP expression may require different levels of FIL, YAB3 and JAG activity. In this light, it is intriguing that SHP is expressed at the periphery of the FIL/YAB3/JAG expression domain where FIL/YAB3/JAG activity may be weakest; possibly pointing to their proteins having properties of a morphogen that controls different downstream targets in a concentration-dependent manner. It will be necessary to develop tools to quantitatively detect FIL, YAB3 and JAG proteins, in order to enable a close comparison of their distribution with FUL and SHP expression during fruit development.

We have gained further insight into how fruit patterning is established by showing that JAG/FIL activity is negatively regulated by RPLin the replum. In rpl mutants, the two valve margins coalesce into a single stripe of tissue with valve margin characteristics. This expanded valve margin development correlates with the presence of ectopic FILexpression in these medial tissues. Removal of either JAG or FIL activity from a rpl mutant rescues replum development,demonstrating that ectopic JAG/FIL activity in the replum is the cause of the rpl phenotype. Thus, RPL effectively divides JAG/FIL activity into two domains, creating two separate valve margins that enable each valve to separate independently from the fruit.

The spatial regulation of JAG/FIL activity in the fruit identifies an interesting parallel to lateral organ and meristem patterning at the shoot apex. In the shoot, antagonism between the meristem and lateral organ primordia ensures that the meristem is not consumed during the process of organogenesis and prevents organ primordia from fusing together. In the fruit, RPL [a.k.a. BELLRINGER(Byrne et al., 2003), PENNYWISE (Smith and Hake,2003) and VAAMANA(Bhatt et al., 2004)]determines the limits of valve margin development by inhibiting JAG/FIL activity. BELLRINGER also plays important roles in meristem development and acts redundantly with SHOOTMERISTEMLESS to prevent the fusion of the cotyledons (seed leaves) at their base(Byrne et al., 2003). Thus, the mechanism that prevents the margins of the valves from fusing may be derived from the process by which organ fusion is suppressed in the seedling shoot. It will therefore be interesting to determine whether ectopic FIL or JAG activity is responsible for any of the shoot defects seen in rpl mutants.

In 1790, Goethe proposed that floral organs represented modified vegetative leaves. It would take nearly 200 years for plant biologists to develop the technologies to test this intriguing hypothesis. The formation of the ABC model of floral organ development led to an understanding of the molecular basis of floral organ identity (Coen and Meyerowitz, 1991). These discoveries demonstrated that the expression of a handful of genes was responsible for transforming simple leaves arising on flowers into sepals, petals, stamens and carpels. These principles came full circle when it was shown that the constitutive expression of floral homeotic genes and their co-factors could convert vegetative leaves outside of flowers into individual floral organs(Honma and Goto, 2001; Pelaz et al., 2001). The observation that these ectopic floral organs formed all of the appropriate tissue types in the correct spatial arrangement, despite the constitutive expression of the homeotic genes, suggests that these genes do not directly control the arrangement of tissues. Furthermore, since leaves can be converted into floral organs, and vice versa, the positional information that controls the arrangement of tissues must be the same for every organ. Our work showing that FIL, YAB3 and JAG, which are expressed in all organs,are important for the development of fruit-specific tissues and for the development of all floral organs, suggests that they are important for this positional information. Thus, not only do floral organs share a common ancestry with leaves, but the mechanisms that pattern the arrangement of tissues are also common. It will be interesting to determine how the FIL/YAB3 and JAG pathways were co-opted to pattern the wide array of organ types found in plants.

We would like to thank John Bowman for providing the FIL probe(pY1-Y) and fil-8 yab3-2 double mutant seeds. We thank Evelyn York for technical help with SEM work, which was done at the Scripps Institution of Oceanography. We would like to thank B. Crawford, G. Ditta, K. Gremski, S. Jeong, A. Roeder for careful reviewing of the manuscript. This work is funded by a National Science Foundation grant to M.F.Y. and a Howard Hughes Medical Institute Predoctoral Fellowship to J.R.D. D.W. is supported by NIH (GM062932) and the Max Planck Society.