Cell-Cell Signaling and Movement by the Floral Transcription Factors LEAFY and APETALA1

See allHide authors and affiliations

Science  04 Aug 2000:
Vol. 289, Issue 5480, pp. 779-781
DOI: 10.1126/science.289.5480.779


LEAFY (LFY) and APETALA1(AP1) encode unrelated transcription factors that activate overlapping sets of homeotic genes in Arabidopsisflowers. Sector analysis and targeted expression in transgenic plants were used to study whether LFY and AP1 can participate in cell-cell signaling between and within different layers of the floral meristem. LFY signaled equally well from all layers and had substantial long-range action within layers. Nonautonomous action of LFY was accompanied by movement of the protein to adjacent cells, where it directly activated homeotic target genes. In contrast, AP1 had only limited nonautonomous effects, apparently mediated by downstream genes because activation of early target genes by AP1 was cell-autonomous.

Shoots and flowers are derived from collections of stem cells called meristems, which are stratified into distinct cell layers. In many plants, restrictions in the plane of cell division in the two outer layers lead to the generation of three cell lineages—the epidermal layer (L1), the subepidermal layer (L2), and the internal layer (L3)—thus allowing the generation of genetically mosaic shoots and flowers. Mosaic studies have shown that some floral transcription factors can signal from layer to layer, although signaling within layers always appeared to be largely absent (1–6).

Although cell-cell communication initiated by transcription factors is not unusual, plant cells differ from animal cells in that they are connected by plasma membrane–lined channels called plasmodesmata, which provide cytoplasmic continuity between adjacent cells. On the basis of the precedence of intercellular trafficking of viral proteins, it has been proposed that cell-cell communication by trafficking of transcription factors is a widespread phenomenon in plants (7). Two transcription factors, KN1 in maize and DEF in Antirrhinum, have indeed been shown to move to cells in which their RNAs are not found (4, 8, 9). However, because direct target genes have not been conclusively identified for either factor, the biological activity of the exported proteins could not be assayed, although DEF movement at late stages of development correlated with some nonautonomous phenotypic effects during early stages (4).

To investigate transcription factor trafficking inArabidopsis flowers, we used two complementary approaches to compare the cellular autonomy of LFY and AP1, two unrelated transcription factors that activate overlapping sets of target genes (10, 11). In the first approach, we used FLP recombinase to create genetically mosaic plants with sectors marked by excision of a β-glucuronidase (GUS) gene. Activation of FLP under the control of a heat shock promoter (HSP::FLP) (12) resulted in 35S::AP1 + GUS sectors in a35S::AP1 GUS + ap1-1 background (13). In ap1-1 flowers, first-whorl sepals are replaced by bracts in the axils of which secondary flowers arise, whereas second-whorl petals are typically absent (14). Analysis of mosaic shoots from heat-shocked ap1-1 HSP::FLP FLP.AP1 plants revealed that the recombined allele had to be present in all layers for full rescue and that clones expressing 35S::AP1only in L3 were indistinguishable from ap1-1 mutants (13).

Clones expressing 35S::AP1 only in L1 produced first-whorl organs with L1 cells typical of wild-type sepals, but L2 and L3 cells more typical of ap1 bracts (Fig. 1D). Second-whorl organs were restored, and these had petal identity in L1 but not in the internal layers (Fig. 1E). Conversely, expression of 35S::AP1 in L2 and L3 produced first-whorl organs with sepal anatomy in the internal layers, but a bract-like L1 (Fig. 1F). In the second whorl, organs with petal shape were produced, but L1 typically lacked petal identity (13). None of the L1, L2, or L3 clones suppressed the formation of secondary flowers (13). Mericlinal sectors, in which 35S::AP1 + and35S::AP1 cells abutted in the same layer, showed complete autonomy of AP1 within layers (Fig. 1F). In summary, these genetic mosaics revealed that AP1acts largely cell-autonomously to control cellular identity, but nonautonomously to promote outgrowth of second-whorl organs.

Figure 1

AP1 and LFY sectors. The first six panels are cross sections of floral organs; adaxial side is up. (A) Wild-type sepal, with adaxial curvature, no internal layers at the margin (arrows), large cells in the abaxial epidermis (arrowhead), and isodiametric internal cells. (B) Wild-type petal with distinct adaxial and abaxial L1 of conical cells. (C) ap1-1 FLP.AP1 first-whorl bract, stained for GUS activity, and showing blunt margin, uniform epidermis of small cells, and subepidermal columnar cells (arrow). The next three panels show sectioned organs from mosaic ap1-1 HSP::FLP FLP.AP1 plants. (D) First-whorl organ with a35S::AP1 L1. The epidermis has large cells typical of the wild type (arrowhead). Internal layers are ap1-like (arrow). (E) Dark-field image of second-whorl organ with 35S::AP1 L1 consisting of petal-typical conical cells. The internal cells, which areap1-1 mutant (GUS-positive tissue appears orange), differ from those of wild-type petals, although their exact identity is unclear. (F) First-whorl organ with a mericlinal L2-L3 clone of 35S::AP1 cells. Sepal-like cells abut bract-like cells at the clone border between arrows. (G) Dark-field image of cross section through a GUS-stained flower from alfy-12 HSP::FLP FLP.LFY plant. Two35S::LFY clones—one in the stamens (st) on the top, and one in the sepal (s), petals (pe), and stamens below—are indicated (g, gynoecium). Layer composition of organs is indicated (m, mericlinal sector; p, periclinal sector). Scale bars, 68 μm (A to F), 250 μm (G).

A strategy similar to that for AP1 was used to generate35S::LFY sectors in a lfy-12 mutant background (13). Mosaic plants were obvious because they produced flowers with petals and stamens on lfy shoots (13), which normally produce only shoot-like flowers that lack petals and stamens (15). Some mosaic plants showed conversion of lateral branches into solitary rosette flowers as well as terminal flowers on the primary shoot, similar to35S::LFY plants (13, 16). Flowers were mosaic in many different ways, but often contained phenotypically wild-type organs that were genotypically completelylfy mutant. Sectors that occupied less than half of a meristem could reorganize it into a normal flower, demonstrating nonautonomous behavior of LFY both across and within layers (Fig. 1G) (13).

Genetic chimeras cannot be sexually propagated, and because of their sporadic nature, it is often impossible to analyze a particular chimeric arrangement at different developmental stages. To overcome these limitations, we generated molecular mosaics by expressingAP1 and LFY under the control of the L1-specificAtML1 promoter (13) in ap1 andlfy mutants (Figs. 2B and3B). Although most ap1 ML1::AP1 lines had phenotypes similar to those of L1 genetic mosaics, a minority showed more extensive rescue of the mutant phenotype, suggesting that higher levels of AP1 in L1 had limited nonautonomous effects on the cellular identity of internal layers. These lines also had gain-of-function phenotypes that included bract-like organs on the abaxial base of pedicels (13). In contrast to ap1, lfy mutants were fully rescued by ML1::LFY. Most lfy ML1::LFY lines had phenotypically wild-type flowers; about one-quarter of these lines also had gain-of-function phenotypes similar to those of 35S::LFY plants (13).

Figure 2

RNA expression in ML1::AP1plants. Top row, ap1-1 (left panel) or wild type (two right panels); bottom row, ap1-1 ML1::AP1. (Aand B) In ap1-1 ML1::AP1 plants,AP1 is expressed in addition to its endogenous domain throughout L1, including the inflorescence meristem (im), pedicel organs (asterisks), and stem. (C and D)AG is ectopically expressed in L1 of ap1-1 ML1::AP1. (E and F) AP3is ectopically expressed in L1 of ap1-1 ML1::AP1. The inset shows L1-specific AP3 expression in a rescued petal (pe) primordium of a stage 8 flower, which contrasts withAP3 expression throughout all layers of the adjacent stamen (st) primordium, whose formation is AP1-independent. Scale bar, 100 μm, except 50 μm in (C) and 70 μm for inset in (F). Numbers indicate floral stages (21).

Figure 3

RNA expression in ML1::LFYplants. Top row, wild type (left panel) or lfy-12 (two right panels); bottom row, lfy ML1::LFY. (A) Inset shows absence of LFY RNA in the lfy-30deletion allele. (B) L1-specific expression ofLFY in lfy-30 ML1::LFY. (Cand D) AG expression is restored in all layers oflfy-12 ML1::LFY. (E and F)AP3 expression is restored in all layers of lfy-12 ML1::LFY, but is initially shallower than in the wild type (compare to Fig. 2E). Scale bar, 50 μm, except 30 μm for inset in (A).

To understand the molecular mechanisms underlying the phenotypes of ap1 ML1::AP1 and lfy ML1::LFY, we analyzed the RNA expression patterns of the target genes AP3 and AG. Early expression ofAP3 and AG is not markedly changed inap1 mutants, because AP1 is only a redundant activator of these genes (10, 11). However, in strongap1 ML1::AP1 lines, AG was ectopically expressed in L1 of the shoot meristem and pedicel organ primordia and occasionally prematurely in incipient flowers (Fig. 2D). AP3was also ectopically expressed in L1 of stems and later-arising flowers of strong ap1-1 ML1::AP1 lines. AP3 RNA was found only in L1 of restored petals, indicating cell-autonomous activation of AP3 within the normal expression domain of AP1 (Fig. 2F). Thus, AP1activates AG and AP3 cell-autonomously.

AP3 and AG expression, although much reduced in lfy flowers (10), was restored throughout all layers in lfy ML1::LFY plants (Fig. 3, D and F), indicating nonautonomous activation of AP3 andAG by LFY. Because it has been proposed that such nonautonomous effects might be mediated by protein trafficking, we analyzed the expression of LFY protein in these plants. In contrast toLFY RNA (Fig. 3B), we detected LFY protein in all layers oflfy ML1::LFY plants (Fig. 4C). We also studied the expression of the KB18 AG::GUS reporter in lfy ML1::LFY, because KB18 activity, in contrast to that of endogenous AG, is completely LFY-dependent (17). The activity of KB18, which contains two essential LFY binding sites, was restored in all layers of lfy ML1::LFY flowers; this result confirmed that LFY protein that had moved to adjacent cells was active as a DNA-binding transcription factor.

Figure 4

Expression of LFY protein and a LFY-dependent reporter in ML1::LFY. (A) In the wild type, LFY is detected in the nuclei of all cells of young floral buds. (B) No LFY protein is detected in lfy-12 mutants. (C) In this lfy-12 ML1::LFYinflorescence, a gradient of LFY protein is observed, with the highest levels in L1 and L2 and lower levels in internal layers. This gradient is also apparent in the stem shown in the inset. (D andE) KB18 GUS reporter is active in the AG domain in the wild type, but inactive in lfy-12. (F) KB18 GUS activity is restored in lfy-12 ML1::LFY, often in a gradient, with the highest levels in the outer cell layers (inset). Scale bars, 60 μm (A to C), 85 μm (D to F), and 42 μm [inset in (F)].

Because our experiments used a reporter whose expression requires that it be directly bound by LFY, we were able to show not only that LFY antigen was found in cells where its RNA was not detected, but also that LFY protein was active in these cells. We have also documented substantial long-range action of an exported transcription factor, because expression of LFY in less than half of a meristem was sufficient to reorganize the entire meristem. Additionally, our studies have highlighted the importance of discriminating between nonautonomous effects at the level of mature phenotypes versus the level of early target genes. We found that L1-restricted AP1 could restore petal formation even though it activated the petal identity gene AP3, which is likely to be a direct AP1 target (18), only in L1 of rescued petals.

The finding of LFY movement raises the question of what its role in normal development is. Given that the patterns of LFY RNA and LFY protein in the wild type are similar, it is possible that movement of LFY protein provides only a redundant mechanism to ensure complete conversion of a meristem into a flower. Indeed, shoot-flower chimeras are rare in the wild type but are frequently observed inlfy mutants (19). On the other hand, nonautonomous effects of LFY and its ortholog FLOhave been reported. For example, FLO is required for activation of CEN in the shoot meristem (20), whereas LFY is required to prevent ectopic activation ofAG in the stem (10), although it is not known whether these effects are direct. To determine the requirement of LFY movement in wild-type plants, it will be necessary to examine the effects of disabling LFY movement during normal flower development.

  • * Present address: Novartis Agricultural Discovery Institute, 3115 Merryfield Row, San Diego, CA 92121, USA.

  • To whom correspondence should be addressed. E-mail: marty{at}, weigel{at}


View Abstract

Stay Connected to Science

Navigate This Article