INTRODUCTION

TOP INTRODUCTION LEAF INITIATION LEAF SHAPE QTL MAPPING AS A... CELL BIOLOGY IN THE... THE FORMATION OF LEAF... THE EVOLUTION OF LEAF... LITERATURE CITED

Most leaves appear simple at first sight and they consist of only a few cell types. Despite that, many developmental processes are involved in leaf ontogeny, including positioning and initiation of leaf primordia, specification of leaf identity, establishment of dorsiventrality, the control of cell division and expansion, and pattern formation. This report highlights and reviews some of what we know about the genetic circuitry that underlies leaf form, as discussed at a workshop on "Leaf Development" held February 11 through 13, 2002, at the Instituto Juan March (Madrid).

	LEAF INITIATION

TOP INTRODUCTION LEAF INITIATION LEAF SHAPE QTL MAPPING AS A... CELL BIOLOGY IN THE... THE FORMATION OF LEAF... THE EVOLUTION OF LEAF... LITERATURE CITED

Shoot meristems produce leaves on their flanks in regular patterns called phyllotaxy. As leaves are initiated, meristem cells divide and replace the cells that have just been committed to initiating a leaf primordium. Thus, the meristem balances self-renewal with organ initiation. The regular pattern of leaf initiation allows one to predict where the next leaf will appear. Leaves that have just appeared as bumps from the meristem are in Plastochron 1 and are referred to as P₁ leaves. The cells in the meristem that will become the next leaf are designated P₀. The P₀ cells, although still part of the meristem, soon become radically different from adjacent cells. They divide at higher rates, their growth axis changes from isodiametric to axial, they loose their indeterminate nature, and they gain leaf cell identity. Even more amazing, the boundary that establishes the P₀ cells from the meristem is continually being remade and reinterpreted with every initiating leaf.

The identification of genes that are expressed in the P₀ cells or are specifically excluded from these cells has been useful in understanding meristem function. In many plant species such as maize (Zea mays) and Arabidopsis, knox (knotted1-like homeobox) genes are expressed in shoot meristems but not in P₀ cells. Recessive, loss-of-function mutations in these genes impair the generation of organs from the shoot apical meristem (SAM), whereas their overexpression leads to the production of meristems on leaves (Lincoln et al., 1994; Long et al., 1996).

Miltos Tsiantis (Oxford) has explored the relationship of the gibberellin pathway and KNOX genes such as SHOOTMERISTEMLESS (STM), which is required for meristem maintenance (Long et al., 1996). Gibberellic Acid (GA) is a diterpenoid plant growth hormone that is important for germination, stem elongation, and flowering time. Previous studies have shown that KNOX gain-of-function phenotypes in tobacco (Nicotiana tabacum) could be suppressed by the addition of GA. They also demonstrated a direct link between NTH15, the tobacco KN1 ortholog, and the promoter of Ntc12, a gene encoding a GA 20-oxidase required for GA biosynthesis (Sakamoto et al., 2001). Tsiantis and colleagues demonstrated the negative regulation of the GA-oxidase gene, GA5, by KNOX genes in Arabidopsis. He showed that the leaves of the asymmetric leaves1-1 (as1-1) Arabidopsis mutant, which are lobed and misexpress KNOX genes, become more severe and gain meristematic activity when grown on media supplemented with paclobutrazol, an inhibitor of GA biosynthesis. In addition, the phenotype of stm-2 is enhanced by the spindly-5 (spy-5) mutation, which increases GA signaling (Jacobsen and Olszewski, 1993). Furthermore, the expression of a GA5::-glucuronidase transgene is excluded from the wild-type SAM, but is found in the region where a meristem would have been in stm mutants. Taken together, these results indicate that one of the roles of the STM gene is to exclude GA5 expression. They also lead to the interesting hypothesis that KNOX genes negatively regulate GA in the meristem, thus keeping cells from expanding and differentiating. Cells that do not express KNOX genes, as found in P₀ and later leaf stages, have increased GA biosynthesis, resulting in increased cell expansion and differentiation.

It has been proposed that differential regulation of KNOX genes may be involved in the generation of dissected leaf morphology (Hareven et al., 1996; Janssen et al., 1998). Tsiantis also reported that the dominant Mouse ear mutation of tomato (Lycopersicon esculentum) increases leaf dissection, a phenotype that is suppressed by the constitutive GA response procera mutation, suggesting again that GA acts antagonistically to KNOX genes in simple (Arabidopsis) and compound (tomato) leaves.

	LEAF SHAPE

TOP INTRODUCTION LEAF INITIATION LEAF SHAPE QTL MAPPING AS A... CELL BIOLOGY IN THE... THE FORMATION OF LEAF... THE EVOLUTION OF LEAF... LITERATURE CITED

Maize leaves are characterized by clear proximal/distal domains as seen in the morphological differences of the sheath and blade. Not as obvious to the eye are the lateral domains. Mike Scanlon (University of Georgia, Athens) discussed establishment of lateral domains in the maize leaf and the role of narrow sheath genes, ns1 and ns2, in the process (Scanlon et al., 1996). Deletion of a lateral domain characterizes the narrow sheath phenotype. A single copy of the wild-type allele of either ns gene is sufficient for normal leaf development (Scanlon et al., 1996; Scanlon, 2000). Clonal analysis has demonstrated that NS1 is required at two discrete foci in the meristem to act in a noncell autonomous way directing recruitment of marginal, leaf founder cells (Scanlon, 2000). NS1 is not required for development of the central domain of the leaf. With the aim of identifying the NS recruitment signal, ns2 was cloned by transposon tagging and was found to encode a protein with similarity to nitrilases, enzymes that are known to participate in the biosynthesis of auxin in plant shoots. This finding, together with the narrow sheath-like phenotype of tobacco and petunia (Petunia hybrida) transgenic plants overexpressing iodoacetamide biosynthetic genes, suggests a role for auxin synthesis and/or transport in the regulation of lateral growth (Berleth and Sachs, 2001; Reinhardt et al., 2000). In the model proposed by Scanlon (2000), iodoacetamide initiates founder cell recruitment of the central domain, and the NS signal completes recruitment of founder cells in the NS domain.

One of the properties of a leaf is that it is a bifacial organ that has dorsiventral asymmetry as soon as it emerges from the flanks of the SAM. This asymmetry allows plants to capitalize on the distribution of specialized cell types such as trichomes, photosynthetic cells, and stomata. Beginning with the analysis of the phantastica (phan) mutant in Antirrhinum majus by Waites and Hudson (1995), where the leaves are partially abaxialized, a number of mutations affecting other genes have been found that produce leaves with different degrees of perturbation in dorsiventrality.

The PHAN gene of A. majus encodes a MYB transcription factor, the mRNA of which is expressed throughout lateral organ anlagen and primordia (Waites et al., 1998). Its activity is partially redundant with that of Handlebars (Hb), and the phan hb double mutant displays abaxialized, radialized cotyledons and leaves (Waites and Hudson, 2001). In Arabidopsis, semidominant phabulosa (phb) and phavoluta (phv) mutations adaxialize lateral organs, which are filamentous and radially symmetrical in phb homozygous plants (McConnell and Barton, 1998; McConnell et al., 2001). On the other hand, revoluta (rev) mutants lack axillary meristems, a trait that can be considered as a partial loss of adaxial identity (Talbert et al., 1995). The PHB, REV, and PHV genes encode members of the homeodomain/Leu zipper (HD-ZIP) family that are expressed adaxially in lateral organs. PHB is first expressed throughout lateral organ anlagen and is later restricted to adaxial regions as the lateral organ primordia emerge (McConnell et al., 2001; Ratcliffe et al., 2000).

John Bowman (University of California, Davis) emphasized that the adaxial and abaxial sides of lateral organ primordia are, respectively, adjacent to and at a distance from the apical meristem from which they derive. Such a positional relationship was proposed by Wardlaw as early as 1949 (Wardlaw, 1949) to underlie the asymmetry of leaves, and was supported by surgical experiments performed by Sussex (1955) and Snow and Snow (1959). These classical analyses, together with recent molecular studies, indicate that the SAM produces an adaxial-promoting signal (Sussex, 1955), the perception of which is likely to be mediated through the HD-ZIP proteins PHB, PHV, and REV (McConnell and Barton, 1998; McConnell et al., 2001).

Abaxial cell fate, which is likely to be the default state in the absence of the adaxial signal produced by the SAM (Bowman et al., 2002), is promoted by the YABBY and KANADI (Kerstetter et al., 2001) genes. YABBY gene expression is restricted to abaxial organ domains by the activity of PHB and related genes. Based on the epistasis displayed by double mutants, KANADI genes are assumed to act upstream of the YABBY genes, but their precise relationships remain to be determined. Andrew Hudson (University of Edinburgh) highlighted the fact that in A. majus, the activity of PHAN is required in adaxial domains for the expression of PHB-like genes and the repression of YABBY-like genes. These observations suggest a role for PHAN in the promotion of organ asymmetry via PHB-like functions.

Several KNOX homeobox genes are known to be targets of PHAN regulation in A. majus as well as of its ortholog, AS1, in Arabidopsis (Byrne et al., 2000). AS1 represses several genes, including KNAT1 and KNAT2, although their loss-of-function mutations do not suppress the phenotype of as1 mutants. This observation suggested to Hudson that AS1 has other targets than KNAT1 and KNAT2. Thus his group screened for suppressors of the as1 mutant phenotype, and identified the SYMMETRICA (SYM) gene, which may function between the AS1 and KNOX genes, as indicated by epistasis experiments.

Mary Byrne (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY) spoke on the partial redundancy of KNAT1 and STM in regulating stem cell function (Byrne et al., 2002). By a series of genetic studies, she was able to place AS1 and AS2 between different KNOX genes. Her results suggest that AS1 and AS2 are negative regulators of KNAT1 and KNAT2, but are negatively regulated by STM. A suppressor screen performed to isolate genes redundant with STM allowed her to identify the QUASIMODO (QUASI) gene, which encodes a BELL1-like homeodomain protein. Triple mutant analysis suggests that QUASI is required for meristem function in the absence of STM and AS1.

Hirokazu Tsukaya (National Institute for Basic Biology, Okazaki, Japan) described his analysis of AS2 and BLADE-LIKE PETIOLE (BLP) two genes that are, in his view, involved in the positioning of the central and proximodistal leaf axes in Arabidopsis. Whereas abnormal vein patterning is displayed by leaves of as2 mutants, which also develop leaflet-like structures (Semiarti et al., 2001), the blp mutant develops leaf lamina-like structures on the petioles. AS2 and BLP are required in the leaf for repression of KNAT genes. The recent cloning of AS2 revealed that it encodes a novel protein with Cys repeats and a Leu-zipper-like sequence. This nuclear-localized protein (Iwakawa et al., 2002) belongs to a novel family that has been named LOB, after the expression of gene family members at lateral organ boundaries (Shuai et al., 2002).

Tsukaya (1994) studied two mutations, angustifolia (an) and rotundifolia3 (rot3), which, respectively, narrow or widen the leaves of Arabidopsis (Tsuge et al., 1996). His group identified the ROT3 gene product as a CYP90 cytochrome of the P450 family, which may be involved in steroid biosynthesis (Kim et al., 1998; 1999). However, they also discovered that the rot3-1 mutation only affects the size of leaf cells, whereas their number and size are reduced in mutants impaired in brassinosteroid biosynthesis (Nakaya et al., 2002). This group has recently cloned the AN gene, which encodes the first known plant member of the CtBP family (Kim et al., 2002). The AN gene is likely to regulate the polarity of cell growth by modulating the arrangement of cortical microtubules, which is abnormal in leaf cells of the an mutant. Positional cloning of AN has also been carried out by Martin Hülskamp's group (University of Köln). They demonstrated a role for AN in the regulation of microtubule organization (Folkers et al., 2002), together with its genetic and physical interaction with ZWICHEL (ZWI), which encodes a kinesin motor molecule involved in trichome branching.

Dorsiventral patterning of the maize leaf was discussed by Marja Timmermans (Cold Spring Harbor Laboratory) on the basis of her studies on two mutants: leafbladeless1 (lbl1) and Rolled1-O (Rld1-O). Whereas severe lbl1 mutants are embryo lethal and shoot meristemless, ectopic laminae arise at the boundaries of abaxial sectors on the adaxial leaf surface of weak lbl1 mutants, suggesting a role for LBL1 in the specification of adaxial leaf identity. Although the number of founder cells recruited from the meristem and incorporated into the leaves of lbl1 mutants is strongly reduced, the semidominant Rld1-O mutation inverts the dorsiventral polarity of leaves but has no apparent effect on founder cell recruitment. Timmermans proposed that these two genes negatively regulate each other or act in opposite ways on the same pathway, given that their phenotypes are mutually suppressed in double mutants. To characterize the phenotypes of lbl1 and Rld1-O, Timmerman's group isolated several maize homologs of the Arabidopsis YABBY genes; their protein products are highly conserved in the Zn-finger and YABBY domains. As already shown in Arabidopsis, the yabby genes of maize are expressed throughout the incipient leaf primordium. It is interesting that later in development, their expression becomes restricted to the adaxial domain of the leaf, contrary to what happens in Arabidopsis. These results, taken together with the demonstration that yabby genes are downstream of lbl1, indicate that the lbl1, rld1, and yabby genes are involved in dorsiventral patterning of the monocotyledonous maize leaf, whose polarity may be inverted relative to that of the dicotyledonous Arabidopsis leaf.

	QTL MAPPING AS A TOOL FOR ANALYZING LEAF DEVELOPMENT

TOP INTRODUCTION LEAF INITIATION LEAF SHAPE QTL MAPPING AS A... CELL BIOLOGY IN THE... THE FORMATION OF LEAF... THE EVOLUTION OF LEAF... LITERATURE CITED

José Luis Micol and colleagues (Universidad Miguel Hernández, Elche, Spain) analyzed variations in the architecture of vegetative leaves in a large sample of Arabidopsis accessions. Crosses between accessions that displayed extreme and opposite variations in leaf architectural traits revealed that the F₂ progeny could not be placed into discrete phenotypic classes. Because these results suggested that intraspecific variability in Arabidopsis leaf morphology arises from an accumulation of mutations at quantitative trait loci (QTL), a mapping population of recombinant inbred lines (RILs; Lister and Dean, 1993) was studied. One hundred RILs were grown, and the third and seventh leaves of several plants from each RIL were analyzed. More than 20 QTL were identified, harboring naturally occurring alleles that contribute to natural variations in the architecture of leaves (Pérez-Pérez et al., 2002).

A similar approach has been followed by Andrew Hudson and colleagues using interfertile hybrids in the genus Antirrhinum majus. The 20 species under study display marked variations in morphology such as large leaves in A. majus and small leaves in A. molle. Analysis of hybrids between these species suggests that overall leaf size or length and width are independently controlled by several genes. Their identification is under way by following a QTL mapping approach.

	CELL BIOLOGY IN THE LEAF

TOP INTRODUCTION LEAF INITIATION LEAF SHAPE QTL MAPPING AS A... CELL BIOLOGY IN THE... THE FORMATION OF LEAF... THE EVOLUTION OF LEAF... LITERATURE CITED

The relationship between cell cycle and leaf development has been explored with different techniques. Gerrit Beemster (University of Gent) has quantified cell division, cell expansion, and endoreduplication as well as their variation with time, in the abaxial surface of the first vegetative leaves of Arabidopsis. Cell division rates remain relatively constant throughout 9 d after initiation, and decline over a 4-d period, in which the onset of endoreduplication becomes apparent. With the aim of determining the effect of altered cell cycle regulation on the interplay between the division and expansion of leaf cells, Beemster obtained transgenic lines overexpressing the KRP2 gene; its product is a Kip-related protein with cyclin-dependent kinase binding specificity (De Veylder et al., 2001). Overexpression of KRP2 inhibited cell cycle progression in leaf primordia cells without affecting the temporal pattern of cell division and differentiation and resulted in markedly serrated leaves, which consisted of enlarged cells. Based on the comparison of the effects of the overexpression of KRP2 and a number of other cell cycle genes, which differentially affect individual growth parameters, Beemster concluded that they are independently controlled.

The switch from the mitotic cell cycle to endoreduplication is well known in the differentiation of trichomes and guard cells in the developing leaf epidermis. Yuki Mizukami (University of California, Berkeley) described experiments aimed at determining whether or not the cell cycle can be uncoupled from cell differentiation in the epidermis of Arabidopsis and tobacco leaves. She overexpressed cell cycle regulator genes such as D-type cyclins and found that cell division was accelerated in tobacco Bright Yellow 2 cells. In transgenic Arabidopsis plants, the mitotic cycle continued, leading to the development of multicellular trichomes and multicellular guard cells with essentially normal morphology, although stomatal patterning was affected.

Mizukami also overexpressed the Arabidopsis homologs of the FZY/FZR (Fizzy/Fizzy-related) proteins, which are known to target Cyclin B for degradation, and found that they induced endoreduplication in tobacco Bright Yellow 2 cells. Tobacco plants expressing a 35S::AtFZR transgene developed unicellular trichomes instead of the wild-type multicellular trichomes. However, the branching pattern of such hair cells remained undisturbed. Taken together, these results led Mizukami to conclude that the entry into endoreduplication is not essential for trichome cell morphogenesis and that patterned cell cycle transition can be uncoupled from determination and progression of cell differentiation in leaf epidermis.

Hülskamp's group has studied the TRIPTYCHON (TRY) gene, which encodes a MYB-related transcription factor that negatively regulates trichome patterning (Schnittger et al., 1999) and is a homolog of the root-hair patterning gene CAPRICE (CPC). Given that these two genes act redundantly in trichome development and in root-hair cell fate choice, Hülskamp proposed a genetic model for trichome patterning based on a reaction-diffusion model (Meinhardt and Gierer, 1974) that couples a short range autocatalytic activator process with a long range inhibitory process. According to Hülskamp, TRY and CPC fit as inhibitors in this model.

Aiming to ascertain the nature of the controls in the switch from mitosis to endoreduplication, Hülskamp and colleagues have generated transgenic lines that express known cell cycle genes in a trichome-specific manner. In agreement with the results of Mizukami, they demonstrated that the ectopic expression of specific B-type and D-type cyclins suppresses endoreduplication (Schnittger et al., 2002a; 2002b), thereby producing multicellular trichomes but not substantially modifying trichome shape.

Laurie Smith (University of California, San Diego) has isolated mutations in three brick (brk) genes that are required for formation of the lobes that characterize the margins of epidermal pavement cells in maize (Frank and Smith, 2002). The lobed shape is assumed to be determined by the arrangement of cortical microtubules. Expanding pavement cells of brk mutants display apparently normal microtubule organization, but lack the patches of cortical F-actin that colocalize with the tips of elongating lobes and the sites of lobe initiation, suggesting a role for brk genes in multiple, actin-dependent cell polarization in the developing maize leaf epidermis. Evidence from clonal analyses and genetic interactions indicate that the brk genes play different roles but act in a common process. The brk1 gene has been cloned and found to encode a small protein of 8 kD belonging to a family highly conserved throughout plants and animals. Smith speculated that the BRK1-like proteins function in actin-dependent aspects of cell polarization, and that, at least in maize, BRK1 regulates localized actin polymerization.

	THE FORMATION OF LEAF PATTERNS

TOP INTRODUCTION LEAF INITIATION LEAF SHAPE QTL MAPPING AS A... CELL BIOLOGY IN THE... THE FORMATION OF LEAF... THE EVOLUTION OF LEAF... LITERATURE CITED

A striking aspect of leaves is their venation pattern. Veins differentiate from procambial cells that form de novo from meristematic precursors within the leaf primordium (Nelson and Dengler, 1997; Dengler and Kang, 2001). Procambial strands arise from ground cells, which elongate, proliferate, and finally differentiate into veins. Aiming to ascertain how positional signals translate into procambium, Nancy Dengler (University of Toronto) has used a -glucuronidase reporter construct for the HD-ZIPIII homeobox gene ATHB-8 (Baima et al., 2001), a member of the above mentioned family, including PHB, PHV, and REV, as a molecular marker of early procambial development. An almost exact correlation in spatial and temporal patterns of cell cycling and ATHB-8 expression was found, suggesting that this gene might be involved in the maintenance of procambium-specific patterns of cell cycling during leaf development.

Tim Nelson's group (Yale, New Haven, CT) is studying venation patterning by identifying mutants, one of which, cotyledon vascular pattern1 (cvp1; Carland et al., 1999), displays discontinuous cotyledon veins. The CVP1 gene encodes a sterol methyltransferase involved in sterol biosynthesis (SMT2; Schaeffer et al., 2001), and its lack of function is rescued by the overexpression of a related gene, SMT3. Another of the mutants found by Nelson is cvp2, which encodes a synaptojanin-like phosphatidylinositol (4, 5)-bisphosphate phosphatase (McPherson et al., 1996). They also identified genes expressed specifically or predominantly in provascular cells. One of these is VASCULAR HIGHWAY1 (VH1), which encodes a protein product that resembles CLAVATA1 and BRASSINOSTEROID INSENSITIVE1, in having an extracellular domain with a Leu-rich repeat and an intracellular Ser/Thr kinase domain. A possible role for this putative receptor is to receive signals for differentiation versus proliferation in provascular cells.

The patterning of stomata, the cellular structures of the epidermis that mediate gas exchange, is under genetic and environmental control (Serna and Fenoll, 2000; van Groll and Altmann, 2001). Thomas Altmann's group (University of Potsdam, Golm, Germany) has isolated ethyl methanesulphonate-induced mutants with altered stomatal traits to unravel the mechanisms regulating stomatal distribution. Several of the mutants found display clustered stomata compared with stomata on wild-type leaves, which are surrounded by a stomata-free area (Berger and Altmann, 2000). The stomatal density and distribution1-1 (sdd1-1) mutation causes a 2- to 4-fold increase in the density of stomata, many of which are not separated by intervening pavement cells. The SDD1 gene encodes a subtilisin-like Ser protease that accumulates near the extracellular surface associated to the plasma membrane of meristemoids and guard mother cells, two successive stages of the stomatal developmental pathway. Based on the previously known activities of eukaryotic subtilases, Altmann proposed that SDD1 is a processing protease involved in the creation of a signal emanating from meristemoids or guard mother cells, which regulates the development of cell lineages forming stomatal complexes. SDD1 overexpression results in reduced stomatal density and premature arrest of divisions in stomatal precursor cells. SDD1 overexpression has no effect on too many mouths (tmm; Yang and Sack, 1995), a mutant that displays clustered stomata. The epistasis of tmm on the overexpression of SDD1 suggests that the genes act in the same pathway.

Carmen Fenoll (University of Castilla-La Mancha, Toledo, Spain) presented results on the influence of light in stomata formation, which they studied by means of cell lineage analyses in constitutive photomorphogenesis mutants, affected in the pleiotropic CONSTITUTIVE PHOTOMORPHOGENIC/DE-ETIOLATED/FUSCA (COP/DET/FUS; for review, see Hardtke and Deng, 2000) loci. They found that these mutants display perturbations in the development and positioning of stomata that are remarkably similar to those described for mutants specifically altered in stomatal patterning. These results strongly suggest that the genes involved in the repression of photomorphogenesis play a crucial role in stomatal pattern formation in a light-independent manner.

	THE EVOLUTION OF LEAF SHAPE

TOP INTRODUCTION LEAF INITIATION LEAF SHAPE QTL MAPPING AS A... CELL BIOLOGY IN THE... THE FORMATION OF LEAF... THE EVOLUTION OF LEAF... LITERATURE CITED

Leaf shape varies from simple, such as found in the Arabidopsis leaf, to dissected as in tomato. Two different groups presented work on genetic mechanisms that lead to dissected leaves. In each case, gene expression, normally excluded from simple leaves, was found expressed in species with dissected leaves and correlated with the dissected leaf shape.

Neelima Sinha (University of California, Davis) described how most species with dissected leaves express KNOX genes and species that have simple leaves do not express KNOX genes (Bharathan and Sinha, 2001). What appeared to be exceptions to the rule, i.e. species with a simple leaf that express KNOX genes, were shown by scanning electron microscopy to actually be dissected at leaf initiation. KNOX gene expression is found along the leaf margins of young leaves, leading to the elaboration of the lobes and leaflets. As found in plants with simple leaves, KNOX expression disappears in the P₀ cells of the meristem, consistent with the hypothesis that KNOX genes are needed to be repressed to initiate a leaf.

Julie Hofer (John Innes Centre, Norwich, UK) examined the dissected leaf of pea (Pisum sativum). She showed that the unifoliata mutant of pea, which no longer has a dissected leaf, is a mutation in a LEAFY ortholog (Hofer et al., 1997). LEAFY was first discovered as a gene that regulates the initiation of flowers from the inflorescence meristem in Arabidopsis (Weigel et al., 1992). Her analysis shows that pea leaves do not express KNOX genes, also discovered by Sinha, but instead express LEAFY. She has expanded this research to the related species of Medicago. A Medicago mutant was described that has unifoliate leaves and again the mutation is in a LEAFY ortholog. From the combined work of Sinha and Hofer, it appears that the phylogenetic clade that includes pea and Medicago gained dissected leaves, not by expressing KNOX genes, but expressing genes normally restricted to the floral meristems.

Received September 26, 2002; returned for revision September 26, 2002; accepted October 29, 2002.