The role of JAGGED in shaping lateral organs

The aerial part of the plant develops new lateral organs continuously through a reiterative process of organogenesis that takes place at the shoot apical meristem. During the life cycle, these lateral organs are modified to suit changing needs. In flowering plants, lateral organs will take the shape of leaves, sepals, petals, stamens and carpels. Leaves typically develop a wide blade adapted to gather light, whereas floral organs promote reproduction. Throughout evolution, lateral organs have also been modified to help plants adapt to almost every terrestrial habitat on earth. For example,the leaves of cactus plants have been modified to form spines to protect against herbivores.

While our understanding of the specification of organ identity is advanced,a mechanistic understanding of the processes that sculpt lateral organs is only now emerging. In Antirrhinum, two members of the TCP family of transcription factors, encoded by CINCINNATA (CIN) and CYCLOIDEA (CYC), have been implicated in the regulation of lateral organ development (Cubas et al.,1999). CIN controls growth of the leaf blade and is required to generate flat leaves with zero Gaussian curvature(Nath et al., 2003). cin mutants develop excessive growth at the periphery of the leaf blade, resulting in wavy leaves. Interestingly, CIN appears to prevent excessive growth by sensitizing peripheral cells to a cell-cycle arrest front that moves from the tip of the leaf to the base. CYC,which also appears to regulate growth via effects on the cell cycle,suppresses growth of dorsal floral structures, producing the asymmetric flowers typical of Antirrhinum(Gaudin et al., 2000; Luo et al., 1996). It is interesting to note that both genes appear to affect organ shape by promoting differentiation.

We have characterized JAGGED (JAG) (formerly BRACTS), a gene involved in the formation of lateral organs. JAG encodes a putative transcription factor with a single C₂H₂ zinc-finger domain and is expressed in the growing regions of lateral organs. Mutations in JAG most severely affect the distal regions of organs, resulting in a jagged edge. In petals, the entire distal half of the organ can be eliminated. In jag mutants,cell-cycle activity declines earlier that in wild type, suggesting that JAG promotes distal petal development by suppressing premature cell-cycle arrest. Misexpression studies with an activation-tagged allele reveal that JAG can promote the growth of many different tissues including the cryptic bract. We discuss a possible role for JAG in maintaining growth of lateral organs and in the diversification of plant form.

Terminologies to differentiate between the leaf types of a flowering plant are varied. In loose definitions, a bract is a flower- or inflorescence-associated leaf, without a distinction between inflorescence-associated leaves in general and ones that specifically subtend a flower (Harris and Harris,1994). We adopt the terminology of Irish and Sussex(Irish and Sussex, 1990) and Bowman and colleagues (Bowman et al.,1993), who describe a bract as being elongate, with a pointed tip,lacking a petiole (leaf stalk), having stipules and being delayed in senescence compared to sepals. We do not consider epidermal cell pattern in the classification of bracts, since there can be differences between bracts produced early and late in development. Bracts are similar to cauline leaves,which develop on the inflorescence. Cauline leaves that are sessile and subtend co-florescence shoots, can be considered bracts sensu lato. They are distinguished from rosette leaves, which develop during the vegetative phase of growth and have a defined petiole at their base(Bowman, 1993).

Plants were grown under a 3:1 ratio of Cool White and Gro-Lux (wide spectrum) fluorescent lights at 23°C in long day conditions (16 hour light, 8 hour dark).

Wild type was either Columbia [Col-0, Col glabrous (Col(gl)] or Wassilewskija (Ws). jag-1 and jag-4 were identified by screening the University of Wisconsin Knockout collection(Krysan et al., 1999) using JAG-specific primers N-1365 and N-1366 (see Supplementary Table 1 for oligonucleotide sequences; Table S1). The jag-1 allele is disrupted by a T-DNA derived from pSKI015; jag-4 by a T-DNA derived from pD991, which contains an AP3::GUS fusion. As has been observed in other lines derived from this collection(http://www.biotech.wisc.edu/Arabidopsis/), jag-4 has defects in the development of petals and stamens that appear to be due to co-suppression by AP3 sequences in the T-DNA (see Fig. S1). The jag-5D mutation was obtained from a population of Col(gl) pop1^-/- plants(Preuss et al., 1993)transformed with pSKI015 (Weigel et al.,2000). jag-5D is linked to the pop1 mutation. The effect of jag-5D on lateral organ development was the same in Col-0 and Col(gl). ap1-15 is in Col-0(Ng and Yanofsky, 2001). Transgenic Col-0 plants were generated by floral-dip method(Clough and Bent, 1998).

Genotyping of jag-1 used primers oJD126 and oJD127, to amplify a 0.7 kb fragment from the wild-type locus. To detect the jag-1 allele,a T-DNA-specific oligonucleotide, JL202, and N-0681 were used to amplify a 1.1 kb fragment. Genotyping of jag-5D was done using N-1509, N-1510 and SKC12 in a single PCR reaction. PCR products were digested with EcoRV producing a band of 0.5 kb representing the wild-type allele and a band of 0.4 kb representing jag-5D. jag-5D homozygotes can also be distinguished by phenotypes associated with the pop1 mutation(Preuss et al., 1993). Genotyping of ap1-15 was as described previously(Ng and Yanofsky, 2001).

The third and fourth rosette leaves from Ws and jag-1 plants were scanned. Area and perimeter of leaf blade region were measured using the NIH Image program (developed at the US National Institutes of Health and available at http://rsb.info.nih.gov/nih-image/).

Plasmid rescue of jag-5D genomic DNA was performed as described previously (Weigel et al.,2000). pJDP2, which was used for jag-5D recapitulation,was plasmid rescued from PstI digested jag-5D genomic DNA. An initial JAG cDNA clone was obtained by RT-PCR (see below) using primers targeting exonic sequences predicted by Genscan(Burge and Karlin, 1997). 5′ and 3′ RACE (rapid amplification of cDNA ends) was performed using the First Choice RLM-RACE kit (Ambion, Inc.) according to the manufacturer's instructions. A full-length cDNA was obtained using N-0335 and N-0336. An initial JAGGED-LIKE (JGL) cDNA was amplified based on the At1g13400 gene model. A full-length JGL clone was obtained using oJD119 and oJD120 (pPY1).

The JAG recapitulation construct pJD71 was made by isolating a HindIII/ClaI fragment containing the genomic region and 35S enhancer elements from pJDP2 and subcloning it into pBJ36 (pJD63)(Gleave, 1992). The cassette was then shuttled into pART27 (pJD71). For AP1::JAG, the JAGcDNA was PCR amplified from first-strand cDNA with Pfu Turbo polymerase (Stratagene), using primers N-0680 and N-0681, to introduce HindIII and BamHI restriction sites. The product was cloned into pBluescript-SK⁺ and sequenced (pJD37). The JAGfragment was subcloned into pJD33 (Wu et al., 2003), replacing the AP1 cDNA (pJD41). For AP1::JAG:GFP, the JAG cDNA was PCR amplified from pJD37 with N-0680 and N-1413, adding a HindIII site to the 5′ UTR and replacing the stop codon with a PstI site (pJD62). GFPcoding sequence was amplified from pCAMBIA1302(Hajdukiewicz et al., 1994),replacing the start codon with a PstI site and adding an XbaI site to the 3′ UTR (pJD60). JAG was then translationally fused to GFP and shuttled into pJD51(Wu et al., 2003), replacing AP1:GFP with JAG:GFP (pJD105).

Samples for scanning electron microscopy were fixed for 4.5 hours in FAA(50% ethanol, 5% acetic acid, 3.7% formaldehyde) at room temperature,dehydrated through an ethanol series and critical point dried. Samples were sputter-coated using gold and palladium and viewed using a Quanta 600 microscope.

In situ hybridization was carried out as described previously(Long et al., 1996; http://www.its.caltech.edu/~plantlab/protocols/insitu.htm)with the following modifications. Tissue samples were fixed in FAA for 2.5 hours at room temperature. RNase treatment of slides was left out. Substrate solution for alkaline phosphatase color reaction was prepared using 2% of a NBT/BCIP stock solution (Roche Diagnostic, Germany) in 100 mM Tris pH 9.5, 100 mM NaCl, 50 mM MgCl₂.

The JAG probe was transcribed using T3 RNA polymerase (Promega)from pJD37 linearized with HindIII. A HISTONE 4(H4) cDNA clone was obtained by RT-PCR amplification of ATHH4GA (GenBank accession number M17132) from inflorescence tissue using oJD133 and oJD134 and cloned into pBluescript-SK⁺ (pH4-T7). H4 probe was transcribed using T7 RNA polymerase (Promega) from pH4-T7 linearized with HindIII. FIL probe was transcribed using T7 RNA polymerase from pY1-Y (provided by J. Bowman) linearized with EcoRI.

RNA isolation and reverse transcription were performed as described previously (Kardailsky et al.,1999). PCR amplification was performed on 2 μl of reverse transcription reaction. A fragment of the JAG gene that spans the entire coding sequence was amplified using N-0680 and N-0681. A JGLfragment was amplified using oJD119 and oJD120. A TUBULIN fragment was amplified as a control using N-1136 and N-1137. The number of PCR cycles performed is indicated in Fig. 5. Annealing temperature was 58°C for all reactions.

We identified two T-DNA insertion alleles in the At1g68460 gene, which encodes a zinc-finger protein similar to SUPERMAN (SUP) (see below). Because of the defects in leaf and floral organ development common to both mutants, we named this gene JAGGED (JAG). jag-1 has an insertion in intron 2, and jag-4 in intron 4. Both alleles result in the loss of full-length JAG transcript as detected by RT-PCR (Fig. 5B). We focused our subsequent analysis on jag-1.

Inactivation of JAG affects the shape of most lateral organs including leaves and floral organs (Fig. 1). Rosette leaves in the parental Ws strains are paddle shaped and have a smooth edge (Fig. 1A). jag-1 leaves, however, develop with a serrated edge(Fig. 1B). To quantify these shape changes, the area and perimeter of rosette leaf blades were measured. While leaf blade area is not significantly different, jag-1 leaves have a larger perimeter/area ratio (P=0.0016), reflecting the irregular blade margins of these leaves (see Table S2). All four whorls of floral organs are also affected in jag-1, with the defects being most apparent in sepals, petals and stamens(Fig. 1C,D). These defects are more severe in flowers produced late in development.

jag-1 sepals and petals are narrower and shorter than the wild-type organs (Fig. 1C-H). Mutant sepals develop white jagged tissue at the tip(Fig. 1E,F). Petals can be jagged at the tip, with the distal region being absent or reduced in jag-1 (Fig. 1G,H). This region forms the part of the petal that appears white and contains cone shaped cells (Bowman, 1993). Scanning electron microscopy confirmed that the distal region of petals is absent or reduced and that only a few cells resemble wild-type cone cells(Fig. 1K,L,N,O). Often, cells at the tip of jag-1 petals display a cuticular wax pattern and overall shape intermediate between cone cells and those at the base of wild-type petals (Fig. 1K-P).

Wild-type stamens consist of two components: the anther and the filament(Bowman, 1993). jag-1filaments appear normal (Fig. 1C,D), but anthers are reduced with an altered shape and typically produce less pollen than wild-type ones(Fig. 1I,J). Late flowers are often male-sterile. Carpel development is relatively normal, but under some growth conditions, surface irregularities develop along the valve walls (data not shown). Thus, JAG is required for the proper formation of all lateral organs, with a particularly important role in the development of the distal regions of sepals, petals and stamens.

Because inactivation of JAG has such severe effects on petal shape, we further explored the role JAG plays in petal development. To test whether the loss of distal petal tissue could reflect reduced cell division, we analyzed the expression of the HISTONE 4 (H4)gene, which has previously been used to assay cell-cycle activity in developing tissues (Gaudin et al.,2000; Krizek,1999; Nath et al.,2003). H4 expression correlates with the expression of other cell-cycle markers such as Cyclin D3b in Antirrhinum(Fobert et al., 1994). As in Antirrhinum, at any given time only a fraction of cells show strong H4 expression (Fig. 2B,C). In wild-type petals, the number of cells expressing H4 increases during floral stages 7 and 8(Fig. 2A,B)(Smyth et al., 1990). The number of H4-positive cells plateaus during stages 9, 10 and 11, and it begins to fall at stage 12. This peak in the number of H4-positive cells correlates with the rapid growth of petals during stages 9-11(Smyth et al., 1990). H4-expressing cells are also much more frequent in the distal region of petals, especially during stages 8-10.

In jag-1 petals, the number of cells expressing H4 is normal during stages 7 and 8 (Fig. 2A-C). After stage 8, however, the number of H4-positive cells soon declines, and continues to decrease through stage 12. Also, the distal regions of jag-1 petals have fewer H4-expressing cells than in wild type. These data, along with the localized defects of jag-1 petals, suggest that JAG promotes the development of the distal region of petals by maintaining tissues in an actively dividing state.

We also characterized a JAG gain-of-function allele, jag-5D, which was obtained using enhancer activation-tagging, in which a T-DNA with a multimerized, constitutively active enhancer is randomly inserted in the genome (Walden et al.,1994; Weigel et al.,2000). jag-5D mutants have a striking phenotype, with bract-like organs forming ectopically on the inflorescence, subtending most flowers (Fig. 3A-C). These leaf-like organs have stipules, lack a petiole, are pointed at their tip, and exhibit delayed senescence, confirming their identity as bracts(Fig. 3D,E, and data not shown). Using scanning electron microscopy, we found that the bracts emerge from the inflorescence meristem, with floral meristems subsequently developing in their axils (Fig. 3K,L, Fig. 6I). Thus, the bract in jag-5D is a product of the shoot and not the floral meristem. The degree of bract development varies from flower to flower, with alternating waves of nodes with and without bracts(Fig. 3B). The size and shape of bracts also vary, such that the bracts can be large and laminar or small and filamentous (Fig. 3D). Bracts tend to be slightly jagged (Fig. 3D), and carpelloidy is sometimes seen in late developing bracts,which are tipped with stigmatic papillae (data not shown). The jag-5Dmutant is different from other mutants with ectopic bracts like leafy(lfy), unusual floral organs (ufo) or filamentous flower (fil) in that jag-5D flowers are fairly normal, although organ number is occasionally increased in the first three whorls of flowers subtended by bracts(Fig. 3A-C, and data not shown)(Levin and Meyerowitz, 1995; Sawa et al., 1999; Schultz and Haughn, 1991; Weigel et al., 1992; Wilkinson and Haughn,1995).

The selfed progeny of the original jag-5D mutant fall into three phenotypic classes with a 1:2:1 ratio of plants that appear wild type, plants that develop bracts, and plants with additional replacement of some flowers by leafy shoots (Fig. 3A-C,F; Table 1). Genotyping and progeny testing determined that the latter are homozygous for jag-5D. Bracts are more prominent and develop more often in these plants(Fig. 3B,C). In addition, the number of coflorescences (side shoots with flowers) that form on the main inflorescence is increased, and coflorescences are occasionally replaced by shoots containing only bracts (Fig. 3F, Table 1). These leafy shoots most frequently develop on the basal nodes of the inflorescence;occasionally, they interrupt a series of flowers. Replacement of flowers with coflorescences is less frequent in jag-5D heterozygotes(Fig. 3G). These phenotypes suggest that JAG can suppress floral meristem identity, which leads to a copy number-dependent conversion of flowers into coflorescences or leafy shoots.

The jag-5D mutation also affects the growth of rosette leaves(Fig. 3H,I). In wild type, the petiole is most defined in the first rosette leaves and gradually becomes reduced leading up to the cauline leaves. In both homozygous and heterozygous jag-5D mutants, the petiole is either absent or develops blade tissue along its edge, creating a teardrop shape. Homozygous mutants have slightly more blade tissue on the petiole than heterozygotes, although it is difficult to distinguish them on this basis (data not shown).

We also found ectopic blade tissue on the stems of coflorescences that arise in the axils of rosette leaves (Fig. 3J). This blade tissue appears to be continuous with that of the first leaves of the coflorescence. The development of blade tissue on rosette leaves and stems shows that jag-5D can activate ectopic growth of tissue and suggests that this activity of JAG may be the cause of ectopic bract development, as well.

The JAG gene was initially identified starting with the jag-5D mutant, which contains a single activation-tagging T-DNA inserted downstream of the annotated gene, At1g68480. JAG encodes a putative single zinc-finger of the C₂H₂ type, with high sequence similarity (35% identity at amino acid level) to its closest characterized Arabidopsis homolog, SUP(Sakai et al., 1995). To confirm the identity of the JAG gene, a genomic fragment including the At1g68480 coding sequence and 35S enhancer sequences from the activation-tagging T-DNA was transformed into wild-type plants(Fig. 4A). 53 of 150 primary transformants showed the jag-5D phenotype to various degrees,including plants that had a stronger phenotype than the original jag-5D mutant (Fig. 4Band data not shown). We were also able to complement jag-1 mutants with a similar genomic fragment that lacked the 35S enhancer sequences (data not shown).

JAG has a close homolog in Arabidopsis, At1g13400, which we named JAGGED-LIKE (JGL) (53% identity). A JAG-related sequence was also found in rice, represented by an EST,which we named OsJAG (GenBank accession numbers C26936, C50482). JAG and JGL form a well-supported monophyletic group, with OsJAG as sister group (data not shown). Alignment of JAG, JGL, OsJAG and SUP proteins indicates very similar sequences in the C₂H₂ zinc-finger, including the plant-specific QALGH motif (Fig. 4D)(Takatsuji, 1998). Some of the residues that provide DNA binding sequence specificity in the C₂H₂ zinc-finger region differ between SUP and JAG, JGL and OsJAG, which all share identical residues at these positions, suggesting that the JAG-like proteins bind to other sequences than SUP(Takatsuji, 1996)(Fig. 4D). A proline-rich region C-terminal to the zinc-finger may function as a protein-protein interaction domain (Falquet et al.,2002; Kay et al.,2000). While the EAR motif [ERF-associated amphiphilic repression motif (Hiratsu et al., 2002)]of SUP is only weakly conserved in JGL and OsJAG, there is a similar motif in JAG (Fig. 4D). At the N terminus, the JAG-like proteins contain a motif that distinguishes them from SUP. In addition, JAG contains a putative monopartite nuclear localization signal (NLS) near the N terminus (Fig. 4D) (Kalderon et al.,1984). We confirmed the predicted subcellular localization of JAG with a GFP fusion, which localizes to the nucleus(Fig. 4C).

By RT-PCR, JAG RNA is detected during vegetative and reproductive development in the shoot apex and open flowers, but not in roots, leaves or inflorescence stems (Fig. 5A). As expected, JAG mRNA levels are increased in jag-5D mutants(Fig. 5B). JGL mRNA is found in the same tissues where JAG RNA accumulates(Fig. 5A). The overlap in RNA accumulation along with the high degree of sequence similarity between JAG and JGL suggests that the two genes are partially redundant.

A more detailed picture of JAG gene activity was obtained by in situ localization of JAG mRNA. During the vegetative phase, JAG mRNA is detected in emerging leaf primordia, but is absent from the shoot apical meristem (Fig. 6A). JAG mRNA is localized in the distal region of leaves(Fig. 6A). Serial sections through leaf primordia revealed that JAG is predominantly expressed in the blade regions (Fig. 6B,C). No JAG expression was detected in the petioles(Fig. 6B). During the reproductive phase, JAG mRNA is excluded from the inflorescence and floral meristem, up to early stage 2 of floral development(Fig. 6E).

The accumulation of JAG mRNA in leaf primordia, but its absence from flowers until mid-stage 2, contrasts with expression of FIL,another lateral organ marker that is detected not only in emerging leaves, but also in cells on the abaxial side of stage-2 floral primordia(Fig. 4D)(Siegfried et al., 1999). These cells, from which JAG mRNA is notably absent, constitute the cryptic bract (Long and Barton,2000).

Like FIL, JAG mRNA is expressed in initiating sepal, petal, stamen and carpel primordia (Fig. 6E,F). Accumulation of JAG mRNA in sepals and stamens declines soon after stage 6, while expression in the distal region of the petal is maintained through stage 11 (Fig. 6H). This expression domain overlaps with the region of the petal that has the highest proportion of H4-positive cells(Fig. 2B). In the carpel, JAG mRNA initially accumulates throughout the valves during stages 7 and 8 (Fig. 6G). By stage 9, JAG mRNA accumulates primarily near the valve margins and gradually disappears by stage 12 (Fig. 6G). The JAG signal seems somewhat patchy in most tissues, with some cells apparently accumulating more JAG transcript than others.

That JAG is expressed in leaf primordia, but excluded from cryptic bracts suggests that ectopic expression of JAG in the cryptic bract of jag-5D mutants leads to ectopic bract development. Consistent with this hypothesis, JAG RNA accumulates in the initiating bracts of jag-5D heterozygotes (Fig. 6I), where it is maintained long after emergence of the bracts. JAG RNA is also detected in the mutant in other tissues in which it is normally not expressed, such as the inflorescence meristem and ovules(Fig. 6I,J). Finally, JAG mRNA also accumulates in the ectopic bracts that replace first-whorl sepals in apetala1 (ap1) mutants(Fig. 6K)(Bowman et al., 1993; Irish and Sussex, 1990).

The appearance of bracts in jag-5D mutants is of special interest,since Arabidopsis normally does not produce such organs, although, as mentioned above, ectopic bracts are produced in flowers of strong ap1mutants. To determine whether JAG is not only sufficient, but also necessary for bract formation, we introduced the jag-1loss-of-function allele into the ap1-15 background. Most jag-1 ap1-15 flowers do not develop any organs in the first whorl(Fig. 7A-D, Table 2). Bracts are sometimes replaced by rudimentary organs that are similar to those found in late-arising ap1-15 flowers (data not shown). Scanning electron microscopy showed that early bract development is similar in ap1-15 and jag-1 ap1-15 flowers (Fig. 7E,F). In ap1-15 flowers, medial bracts continue to grow,while lateral bract primordia typically abort. In jag-1 ap1-15flowers, both medial and lateral bracts abort soon after initiation. Importantly, other aspects of the ap1-15 mutant phenotype are unaffected by jag-1 (Fig. 7C,D, Table 2). Thus, JAG is required for the outgrowth of bracts in ap1mutants, but dispensable for their initiation. This is consistent with the finding that cryptic bracts in wild-type plants also initiate independently of JAG.

In order to determine the effects of over-expressing JAGspecifically in developing flowers, we used the AP1 promoter, which is active in flower primordia from stage 1 on(Hempel et al., 1997). AP1::JAG plants produce a range of phenotypes. In the weakest lines,organs in the first three whorls are fused(Fig. 8D). Typically, sepals are connately fused at their base. Both petals and stamens fuse to form abnormal organs that consist predominantly of petal tissue. In intermediate lines, the severity of sepal fusion increases(Fig. 8A,E). In the most extreme cases, a tube of sepal tissue sheathes the inner whorls of the flower(Fig. 8E). The development of the tissue interior to the sepals is often reduced, although carpels can occasionally be seen developing in the center of the flower(Fig. 8E). Another peculiarity of these flowers is that sepals are not clearly distinct from the pedicel, and that sepal tissue extends down the length of the pedicel to the inflorescence stem (Fig. 8A,E-G).

In the strongest AP1::JAG lines, flowers are replaced by snake-like stumps that do not develop any organs(Fig. 8B,C). First-whorl organs seem to initiate, but they fail to grow. Organs of the inner whorl either do not initiate or, if there is any growth, grow amorphously and do not resemble any particular organ type (Fig. 8H). We confirmed by RT-PCR that the phenotypic severity of our AP1::JAG lines correlated with increased levels of JAGtranscript (see Fig. S2).

To determine the identity of the structures that develop in the place of flowers in strong AP1::JAG lines, we examined the expression of FIL, which, in stage 3 flower primordia, marks discrete populations of cells that develop into sepals(Siegfried et al., 1999). In strong AP1::JAG lines, FIL expression extends down the length of the snake-like outgrowths, suggesting that these structures have at least partial lateral organ identity (Fig. 8I,J).

In aggregate, the phenotypes caused by the AP1::JAG transgene suggest that JAG can override the arrest or delay in the growth of several tissues. In weak and intermediate lines, growth is activated between organs leading to fusion of sepals, petals and stamens. In intermediate lines,precocious growth of sepals leads to an elongated sepal tube. In the strongest lines, growth may be initiated before floral organs are properly established leading to the outgrowth of stumps that lack any discrete organ development.

We have described the function and expression of the JAG gene, an important regulator of growth in lateral organs of Arabidopsis. JAGcan activate growth of many different tissue types, including bracts, which are suppressed in wild-type plants. The absence of JAG expression from the cryptic bracts of wild-type plants contrasts with expression in most other lateral organs, suggesting a possible cause for bractless flowers in Arabidopsis and its relatives.

The plant shoot apex is a dynamic structure in which a meristem maintains a population of undifferentiated stem cells that later undergoes a process of differentiation to form lateral organs. This process involves several steps,beginning with the loss of meristem identity and the acquisition of lateral organ identity, followed by organ outgrowth and morphogenesis, then cell-cycle arrest and finally histogenesis. Progress through differentiation, however,must be regulated so that organs can be properly shaped. If tissues exit the cell cycle prematurely, morphogenesis is incomplete, whereas if cell-cycle arrest does not occur, lateral organs would continue to grow ad infinitum.

In jag loss-of-function mutants, lateral organs do not develop completely, with the strongest defects in the distal regions of organs. In petals, we have shown that this is probably due to a reduction in cell-cycle activity. In eudicots, lateral organs typically differentiate in a basipetal fashion with cells at the tip exiting the cell cycle before those at the base(Donnelly et al., 1999; Nath et al., 2003; Poethig and Sussex, 1985). We propose that JAG slows the cessation of cell division in the distal region of organs until proper morphogenesis has occurred, which is particularly apparent in floral organs, since these form different structures along the proximal-distal axis. JAG is also expressed in the blade region of leaves, and jag loss-of-function mutants develop serrated leaves. While we did not detect a change in total leaf area, the serrated leaves may be caused by a reduction in growth in the inter-hydathode regions of the blade. In this regard it is interesting to note that plants that overexpress ICK1 or KRP2, negative regulators of the cell cycle, also have jagged leaf blades (De Veylder et al., 2001; Wang et al., 2000). This interpretation is also consistent with jag-5D gain-of-function phenotypes in which rosette leaves develop ectopic blade tissue along the petiole. It is an intriguing concept that the shape of organs and the architecture of shoots in plants may be, in part, the result of controlled differentiation of tissues. Thus, the plant may be sculpted by the opposing inputs from genes like CIN that promote cell-cycle arrest, and genes like JAG that suppress cell-cycle arrest.

One of the most dramatic consequences of JAG over-expression in AP1::JAG plants is the inhibition of floral organ boundaries and the precocious growth of sepals creating tube- and snake-like structures. These phenotypes suggest that increased JAG levels can override factors that inhibit the growth of specific organ regions. That JAGover-expression has such dramatic consequences even though JAG is normally expressed in young organs indicates that expression levels need to be finely tuned for proper function of JAG.

One of the characteristic features of members of the Brassicaceae family is that most flowers lack bracts, which is in stark contrast to the presence of these organs in most eudicots. However, bracts are not completely absent in Brassicaceae as basal flowers in many species are subtended by bracts(Arber, 1931). Although bracts are normally absent in Arabidopsis, they can develop in certain backgrounds, such as lfy, ufo or fil mutants. Less pronounced activation of cryptic bracts is seen when flowers are ablated through expression of a toxin (Nilsson et al., 1998). Even in wild-type plants, molecular markers indicate the presence of a cryptic bract whose outgrowth is normally suppressed(Long and Barton, 2000). Thus,while the bract appears to be dispensable for the final architecture of the Arabidopsis inflorescence, the bract is still patterned.

In Arabidopsis, several genes, FIL, AINTEGUMENTA, ASYMMETRIC LEAVES1 and STYLISH1, are expressed in incipient lateral organs(Byrne et al., 2000; Kuusk et al., 2002; Siegfried et al., 1999). At least some of these genes are required for the development of lateral organs,but none is sufficient to activate bract development, even though they are expressed in the cryptic bract. Thus, their activity is either specifically suppressed in the cryptic bract, or bract outgrowth requires another factor. The latter scenario is supported by the finding that JAG is absent from the cryptic bract, but can activate its outgrowth when ectopically expressed in jag-5D mutants. That this function does not merely represent a neomorphic activity is demonstrated by our observation that JAG is required for outgrowth of ectopic bracts that form in ap1-15. It is particularly interesting that loss of JAGfunction in an ap1-15 background primarily affects bracts, even though it is also expressed in other organs. This may indicate that JAG is more important for the growth of bracts, and that redundancy with a gene such as JGL may mask essential functions in other organs. Alternatively, ap1 bracts may be particularly sensitive to defects in organogenesis. This could be the case if ap1 bracts are under similar suppressive influences as bracts of the main inflorescences. In this case,loss of JAG would sensitize the bract to suppression of outgrowth. This interpretation is further supported by the observation that loss-of-function in another gene important for lateral organ development, AINTEGUMENTA, also has a negative effect on bract development in ap1 mutants (B. Krizek, personal communication).

Since our data suggest that JAG is both necessary and sufficient for bract formation, and since JAG expression is normally excluded from the cryptic bract, one can speculate that bract-specific suppression of JAG expression is the cause of bractless flowers in the Brassicaceae. As there are also other examples of bract suppression in plants outside of Brassicaceae, such as in the maize inflorescence(Ritter et al., 2002), it will be interesting to determine the possible role that JAG plays in the evolution of this intriguing plant characteristic in other species, as well.

We thank Dennis Fernando for technical assistance; Evelyn York for training and advice with scanning electron microscopy, which was performed at the Scripps Institution for Oceanography; Beth Krizek for sharing unpublished results; John Bowman for providing the FIL probe; members of the Yanofsky and Weigel laboratories and Jennifer Nemhauser for comments and suggestions on the manuscript and Adrienne Roeder, Clint Whipple and Xuelin Wu for thoughtful discussion. This work was supported by a Howard Hughes Medical Institute Predoctoral Fellowship to J.R.D., grants from the National Science Foundation to R.L.F. (IBN-9817992) and M.F.Y., and the National Institutes of Health (GM62932) to D.W., who is a Director of the Max Planck Institute.