ABSTRACT
Flower meristems comprise several distinct cell layers. To understand the role of cell interactions between and within these layers, we have generated plants chimeric for a key floral homeotic gene, floricaula (flo). These chimeras arose in Antirrhinum by excision of a transposon, restoring flo gene function. Activity of flo in a subset of cell layers gives fertile flowers with an abnormal morphology. This shows that flo can act non-autonomously between layers, although some aspects of its function are impaired. In addition, we show that flo exhibits some cell-autonomy within layers.
INTRODUCTION
Plant and animal development depends on interactions between cells in different layers. In animal systems, interactions between ectoderm (outer), mesoderm (middle) and endoderm (inner) layers during embryo development have been intensively studied, mainly through transplantation experiments or by producing chimeras in which part of a layer is genetically altered. Whereas transplantation methods assay the effects of juxtaposing particular tissues that might differ in the expression of many genes, the use of chimeras can assess the effects of altering the activity of individual genes in different layers. In plants, studies on the interactions between L1 (epidermal), L2 (subepidermal) and L3 (core) layers have largely depended on chimeras in which the genotype of one or more layer is altered. Plants are particularly amenable to this type of approach because chimeras, in which entire layers have been genetically altered, can be vegetatively propagated. In spite of the advantages afforded by plant chimeras, they remain poorly understood because it has been impossible to analyse the gene activities underlying their phenotypes. However, by using a range of genetic and molecular techniques these problems can now be overcome. We have used such an inter-disciplinary approach, in this and the accompanying paper (Hantke et al., 1994), to try to understand the role of cell-cell interactions in the control of flower development.
The shoot meristems of many flowering plant species consist of three layers of dividing cells (Fig. 1). The cells descended from L1 mainly give rise to the epidermis, those from L2 produce subepidermal tissues and the germ cells of reproductive organs, and those from L3 give rise to core tissues (Tilney-Bassett, 1986; Satina and Blakeslee, 1941, 1943). Most of the divisions in the two outer layers of the meristem result in daughter cells remaining in the same layer (i.e. the new cell walls are perpendicular to the meristem surface). This ensures that the cell lineages of L1 and L2 are kept separate from each other and from L3 cells, which can divide in any plane (occasionally divisions can result in daughter cells invading a different layer). If a somatic mutation occurs early in meristem development, it may give rise to a branch in which only one layer of the meristem is mutant (Fig. 1). This is called a periclinal chimera. Vegetative propagation of such a branch enables the chimera to be maintained. In some cases, mutant cells may occupy only part of a meristem, giving rise to branches in which part of a layer is affected; these are termed mericlinal chimeras (Jorgensen and Crane, 1927). Whereas periclinal chimeras give information about interactions between the cell layers (interdermal effects), mericlinal chimeras can also reveal the extent of cell-cell interactions within the same layer (intradermal effects).
Origin of periclinal chimeras in meristems with three layers. An L1 cell carrying a mutation, shown black, divides anticlinally to produce a patch of cells that overlies the developing side branch, thus giving rise to a periclinal chimera.
The study of flower development provides one of the best systems for exploiting chimeras. Several floral homeotic genes involved in switching the fate of meristems and primordia have been described (Coen and Meyerowitz, 1991; Coen and Carpenter, 1993; Davies and Schwarz-Sommer, 1994). Furthermore, chimeras involving several flower developmental genes have revealed interactions between cells in different layers (Stewart, 1972; Szymkowiak and Sussex, 1992, 1993; Zimmerman and Hitchcock, 1951; DeLong et al., 1993). Unfortunately, however, none of these chimeras involve well-characterised genes controlling meristem and primordium identity.
Here we describe the genetic and morphological analysis of plants chimeric for the floricaula (flo) gene of Antirrhinum, an early acting gene that affects meristem identity. A wild-type Antirrhinum plant produces several inflorescences, each comprising about 20 flowers borne in the axils of bracts (small leaf-like organs) (Fig. 2). In contrast, a plant homozygous for a stable flo allele (such as flo-640; Coen et al., 1990) produces inflorescences with an indeterminate shoot growing in place of each flower (Fig. 2). The flo-613 allele carries Tam3 inserted within the second exon (Carpenter and Coen, 1990; Coen et al., 1990) and confers an unstable flo phenotype: flowering sectors are occasionally observed on otherwise mutant inflorescences (Figs 2 and 3). Seeds collected from self-pollinated flowering sectors sometimes produce only mutant plants, whereas in other cases the progeny segregate for wild type and mutant in a 3:1 ratio (Carpenter and Coen, 1990). These observations suggest that the flowering sectors may have arisen by somatic excision of Tam3 during early development, restoring gene function in only one layer (Coen, 1991). Transmission of the revertant allele would be observed in cases where excision has occurred in L2. Similar examples in which chimeras have been shown to arise by transposon excision have been described in morning glory, maize and petunia (Imai, 1934; Dawe and Freeling, 1990; DeLong et al., 1993; Doodeman and Bianchi, 1985).
Diagrammatic representation of inflorescences from wild type, a stable flo mutant and an unstable flo-613 mutant showing a revertant sector with the first three flowers numbered.
A revertant sector with near wild-type flowers on a flo-613 mutant plant.
Here we show that the flowering sectors on flo-613 plants behave as chimeras and can be vegetatively propagated to give whole chimeric plants with various phenotypes. This suggests that flo gene activity in a subset of cell layers can act interdermally to promote the floral programme. In addition, the analysis of mericlinal chimeras suggests that intradermal spread of flo activity is limited.
MATERIALS AND METHODS
Antirrhinum stocks and SEM
The flo-613 mutant and its progenitor have been described previously (Carpenter and Coen, 1990). Cuttings were taken from inflorescence tips, then excess bracts were removed and the base of the stem dipped in hormone rooting powder. Cuttings were planted in a compost consisting of half peat/sand mix and half perlite and placed in a mist unit until they had rooted. Scanning electron microscopy was carried out on replicas of flowers and developing inflorescences as described previously (Williams and Green, 1988).
RESULTS
Genetic analysis of flowering sectors
If the revertant sectors on homozygous flo-613 plants were due to excision of Tam3 from the flo locus during early development, each flowering sector should represent an independent periclinal chimera. Flowers from the same sector should therefore show a greater similarity in genetic behaviour than those from different sectors. To test this, 37 different flowering sectors were individually drawn and labelled as illustrated in Fig. 2. In some cases, revertant sectors from different inflorescences on the same plant were selected, and labelled as A, B and C. The flowers from the sectors were self-pollinated and the seeds of 76 of the resultant seed capsules were sown to give families of up to 48 plants each, grown outside. The phenotypes observed in 23 representative families are described in Table 1.
Most of the families contained only flo mutant plants, indicating that reversion had not occurred in the germ line. However, about 7% of the families, segregated wild type to mutant in a 3:1 ratio, suggesting that in these cases the germ line (i.e L2) was heterozygous for flo and Flo+. Germinal transmission reflected the particular sector from which seed was collected. For example, amongst the progeny derived from plant D13917, only the seed derived from sector A showed germinal transmission (Table 1). A further 7% of the families consisted of 47 or 46 mutant plants with 1 or 2 wild types (e.g. D1326 sector A, D1328, sector A, Table 1). In these cases L2 was presumably homozygous for flo, and the wild-type plants were due to late reversion events in the germ line or to occasional invasions of a few revertant cells into L2.
Many of the flo mutants in the 76 families showed somatic instability, similar to that observed in the previous generation. In most cases, families contained both stable and unstable mutant plants. The stable plants might have arisen by imprecise excision of Tam3 or they might have contained Tam3 but produced no observable excision events. About 4% of the families, contained only unstable plants with a particularly high level of instability, reflecting the sector of origin (e.g. D13257, sector A). A further 4% of the families consisted entirely of stable flo plants, no flowering sectors being observed throughout the season. This behaviour depended on the sector of origin (e.g. plant D1328, sector B), suggesting that it could be the result of an imprecise excision, giving rise to a stable flo mutant allele in L2. However, in one case, seed from an individual flower (D1328, sector C, flower 4) gave only stable mutant progeny even though other flowers from the same sector gave unstable mutant plants. One possibility is that imprecise excision, during the ontogeny of this particular flower, may have resulted in its L2 being homozygous for a stable allele.
Genetic analysis of vegetatively propagated flowering sectors
The similar genetic behaviour of flowers from the same sector confirmed that the flowering sectors might be the result of independent Tam3 excision events giving rise to periclinal chimeras. According to this view, it should be possible to propagate the sectors to produce whole plants that are chimeric. To test this, cuttings were taken from the tips of the flowering sectors and rooted. This gave rise to 12 flowering plants.
These plants were self-pollinated and the seed sown to analyse their progeny. In most cases, several capsules from the same plant gave only mutant progeny (plants 1, 2, 5-11; plant 12 gave no viable seed). Seed from the second generation cuttings from some of these plants were also sown and gave rise to only mutant progeny. One plant consistently gave progeny segregating Flo+ and flo plants in a 3:1 ratio (plant 4, five capsules tested). Therefore, in this case L2 had been affected by the original reversion event and was heterozygous. Plant 3 gave Flo+ and flo progeny in a 3:1 ratio (8 capsules tested), indicating that L2 was heterozygous. However, when the second generation cuttings from plant 3 were progenytested, one plant gave only flo mutants whereas two others gave Flo+ and flo progeny in a 3:1 ratio. This indicated that cuttings taken from plant 3 could be of two types: those in which L2 was homozygous for flo and those in which L2 was heterozygous.
Phenotypes of flowering sectors and chimeric plants
In addition to the flowers from the same sector or the same chimeric plant being similar with respect to their genetic behaviour, they also showed similarities in their flower phenotype. The phenotypes of flowers were divided into three categories: near wild type, intermediate and extreme (Fig. 4). Flowers on the same sector usually had a similar appearance and those on chimeric plants usually resembled their parental sectors, although there was some variablity. Furthermore, the genetic behaviour of the chimeric plants correlated with their phenotype: only plants of the intermediate type consistently gave revertant progeny. These results suggested that the phenotype reflected the particular type of Tam3 excision event. The phenotypes of the chimeric plants were described in detail and compared to wild type.
Wild type
Flowers are borne on stems in the axils of bracts (Fig. 2). Each flower comprises four whorls of floral organs. The outermost whorl comprises five sepals, with the lower two being alternate to the bract. The second whorl consists of five petals, united for part of their length to form a corolla tube bearing five separate lobes, the two upper lobes differing in shape from the lower three. The third whorl contains four stamens, alternate with the petals and the fourth whorl comprises two united carpels.
Near wild type (plants 1,2,3)
Flowers were similar to wild type except that sometimes they had six sepals and occasionally the corolla tube was not completely united, causing twisting and bulging.
Intermediate (plant 4)
Both sepals and petals were slightly narrower and the sepals more pointed than wild type. The corolla was divided into an upper and lower half. The upper half consisted of two petals united for part of their length. The lower half comprised petals that were only sometimes united at one or both of the petal junctions. Occasionally the lower half of the corolla comprised four separate sections. The lower sections of the corolla tube bulged outwards and away from the upper half of the corolla and were sometimes twisted, thus exposing the stamens and carpels. The lower side petals often resembled the lowest centre petal of wild type. The stamens and carpels were wild type.
Extreme (plants 5-12)
The sepals were narrower than wild type and were sometimes petaloid. The corolla comprised four to six separate sections which sometimes had sectors of green tissue, particularly in chimeric plants 9 and 10. A central upper section was usually present, consisting of a narrow strip of corolla tube ending with two very small distorted lobes. The lower petals were often much reduced in size and comprised narrow strips of corolla tube terminating in small lobes that usually resembled the central lower lobe of wild type. This resulted in the lower sections bulging outwards and sometimes twisting. The stamens varied in number from three to five and were sometimes petaloid. Both upper and lower stamens were represented and stood proud of the petals. The carpels were as wild type.
In addition to examining fully developed flowers, we studied the development of floral meristems in some of the chimeras in order to establish how early divergence from wild type occurred. Floral meristems were initiated in a spiral on the periphery of the inflorescence apex. Each inflorescence therefore consisted of floral meristems at a range of developmental stages, the youngest (node 0) being nearest the top. Scanning electron microscopy showed that differences between chimeras and wild type were sometimes discernable by about node 12 but became clearly visible by about node 14, when the five sepal primoridia were visible surrounding a central dome of cells (Fig. 5). At this stage, near wild-type meristems often showed six sepal primordia rather than five and the central dome was flatter than wild type, suggesting that primordium initiation within the dome was less advanced. Comparable meristems from the extreme class had a smaller diameter and the sepal primordia were less well synchronised in development than wild type, the two lower primordia being further advanced than the others. By node 16, broad tongueshaped sepal primordia curved over the central dome and petal primordia were clearly visible in wild-type meristems. In the near wild-type class, the lower sepal primordia appeared narrower and more finger-like and only some of the petal primordia were clearly visible. In node 16 meristems of the extreme class, the development of sepal primordia did not seem to be as synchronised as in wild type but they had a spiral pattern of development and were narrow. Only some of the petal primordia could be seen and appeared to be more separated from each other. These observations correlated with the phenotype of mature flowers which had narrow sepals and separate narrow petals (Fig. 4).
Scanning electron micrographs of floral meristems at nodes 12, 14 and 16 from wild type, a chimera with near wild-type flowers (plant 1) and a chimera with extreme flowers (plant 9). Each floral meristem is orientated with the bract at the base. Scale bar is 100 μm and is only shown for the top left meristem.
Scanning electron micrographs of floral meristems at nodes 12, 14 and 16 from wild type, a chimera with near wild-type flowers (plant 1) and a chimera with extreme flowers (plant 9). Each floral meristem is orientated with the bract at the base. Scale bar is 100 μm and is only shown for the top left meristem.
Stability of flowering phenotype during vegetative propagation
The phenotypes of the chimeric plants could often be transmitted through further generations of cuttings, but some of the plants with an extreme phenotype gave a proportion of flo mutant plants when propagated (plants 5,6,7,9,11,12). Plant 4 had an intermediate phenotype, although some cuttings taken from it exhibited a near wild-type phenotype. The variability sometimes observed following vegetative propagation might be explained if cells from one layer can occasionally be displaced by those from another layer.
Mericlinal chimeras
In addition to periclinal chimeras, we examined several mericlinal chimeras to study the interactions between cells residing within the same layer. Such chimeras were detected as revertant sectors that consisted of several partial or complete flowers on otherwise mutant inflorescences. Individual examples of these sectors were analysed in detail by recording the position and phenotype of flowers along the inflorescence.
Many of the flowering sectors appeared to form strips of revertant tissue that ran vertically along the inflorescence (Fig. 6). Narrow, green, thread-like organs were sometimes observed in the outer whorls or on the pedicels of the flowers, and the flowers were frequently deformed and twisted. This appeared to reflect the formation of organs that were part petal and part sepal or bract, resulting in uneven growth. Scanning electron micrographs of the junctions between petal and sepal-like tissue showed that in some cases they were separated by a sharp boundary of one or two intermediate cells (see flower 5, Fig. 6). In other cases, the boundary was less sharp. As shown in the revertant sector of Fig. 6, the distorted side of the flowers and the location of thread-like organs appeared to correspond to the borders of the sector, indicating that they arose where mutant and revertant tissue were next to each other. The flower at position 2 in this sector had a near wild-type phenotype, indicating that it was contained within the revertant sector. These observations suggested that this sector was a chimera that only occupied about 1/3 of the circumference of the stem and thus comprised a mericlinal chimera. The small flowering inflorescence at position 16 might have been due to the revertant layer occupying the central region of the meristem, so forming a complete periclinal chimera. It was not possible to determine the precise boundaries of the mericlinal chimeras solely on the basis of their phenotype.
Revertant sector comprising part of an inflorescence. The centre panel represents a section of the stem, shown flattened out with positions of bracts (black crescents), numbered sequentially from the base of the inflorescence. Positions of the bracts were determined by labelling and then removing each flower, placing the remaining inflorescence in a drinking straw and marking the position of each node. The straw was then cut longitudinally, removed from the inflorescence and flattened out. Structures in the axils of bracts are shown as circles, red for flowers, green for flo mutant shoots. In some cases no axillary structures were observed. Circles containing both red and green indicate deformed flowers, with the relative proportion and position of mutant tissue indicated diagrammatically. A flowering inflorescence at position 16 is indicated by a red spot within a green circle. Drawings of five individual flowers are shown numbered according to their position on the section of stem. A scanning electron micrograph of a region of flower 5, indicated by an arrow, is shown in the bottom right panel. This shows the junction between mutant (left) and petaloid (right) tissue. Scale bar is 50 μm.
Revertant sector comprising part of an inflorescence. The centre panel represents a section of the stem, shown flattened out with positions of bracts (black crescents), numbered sequentially from the base of the inflorescence. Positions of the bracts were determined by labelling and then removing each flower, placing the remaining inflorescence in a drinking straw and marking the position of each node. The straw was then cut longitudinally, removed from the inflorescence and flattened out. Structures in the axils of bracts are shown as circles, red for flowers, green for flo mutant shoots. In some cases no axillary structures were observed. Circles containing both red and green indicate deformed flowers, with the relative proportion and position of mutant tissue indicated diagrammatically. A flowering inflorescence at position 16 is indicated by a red spot within a green circle. Drawings of five individual flowers are shown numbered according to their position on the section of stem. A scanning electron micrograph of a region of flower 5, indicated by an arrow, is shown in the bottom right panel. This shows the junction between mutant (left) and petaloid (right) tissue. Scale bar is 50 μm.
DISCUSSION
We show that flowering sectors on flo-613 mutant plants behave as periclinal chimeras caused by excision of Tam3 early in meristem development, restoring expression of flo in a subset of cell layers. This conclusion is based on several types of evidence. (1) The genetic properties of flowers from the same sector are usually similar to each other, showing that each sector typically results from a single reversion event. (2) The phenotypes of flowers from the same sector are also usually similar to each other and can be classified as near wild type, intermediate or extreme. (3) Flowering sectors can be vegetatively propagated to give whole plants that show similar genetic and phenotypic properties to individual sectors. The flowering phenotype can be maintained through several generations of cuttings although in some cases flo mutant plants arise, presumably as a result of cells in the revertant layer being displaced by mutant cells (this is particularly common for L3 chimeras). (4) Germinal transmission of the revertant phenotype is only consistently observed for propagated plants with an intermediate phenotype, suggesting that they represent chimeras in which flo activity has been restored in L2 (ie L2 chimeras) whereas the other plants represent L1 or L3 chimeras.
These results indicate that restoration of flo in a subset of meristematic cell layers can activate the floral genetic programme in other layers. Flowers with sepals, pigmented petal-like organs, functional stamens and carpels are seen irrespective of the layer expressing flo. This shows that all chimeras can activate the genes needed for gametogenesis in L2. Similarly, all chimeras can activate genes needed for petal pigmentation in L1. The action of flo therefore involves interdermal cell communication. It is possible that this is mediated by a cell-transmissable signal. Alternatively, the FLO protein may be able to move directly between cells, as proposed for the KNOTTED protein of maize (Jackson et al., 1994).
Although flo expression in a subset of layers can confer some aspects of flower development, other aspects of flo function are hampered because the flower phenotypes of the chimeras are not normal in all respects. Both the timing of primordium development and the type of organs produced in whorls 1 and 2, are altered. The hindered action of flo in the chimeras may reflect reduced overall levels of flo activity in the meristems as a whole. Alternatively, flo may exhibit partial interdermal autonomy with respect to some aspects of its function.
The flo chimeras show some similarities to a Camellia chimera in which L1 genotype is wild type but both L2 and L3 derive from a variety in which reproductive organs are replaced by a proliferation of petals. The chimera has intermediate characteristics, indicating that both L1 and the internal layers contribute to the phenotype (Stewart et al., 1972). In contrast, in chimeras for the fasciated gene of tomato, only the genotype of L3 determines carpel number and flower meristem size (Szymkowiak and Sussex, 1992). For tasselseed2, a gene controlling sexual dimorphism in maize, the genotype of only L2 appears to matter, correlating with the expression of this gene in only the L2 cells of wild type (DeLong et al., 1993). The properties of each chimera therefore vary according to the mode of gene action and the normal site of gene expression.
The role of intradermal and interdermal communications in flo activity can be compared by analysing mericlinal chimeras. Many revertant sectors consist of several flowers running along an otherwise mutant inflorescence. Morphological analysis indicates that these events are clonal and represent mericlinal chimeras in which only a proportion of a meristematic layer comprises revertant cells. The production of such chimeras shows that restoration of flo to one area of the meristem is relatively contained and flo activity does not spread intradermally to other regions. This could suggest that intradermal communication is less effective than interdermal communication. Alternatively, the difference might reflect the number of cells that separate revertant and mutant cells. For a mericlinal chimera in which 1/3 of the inflorescence is revertant, flo activity would have to spread throughout the remaining 2/3 of the circumference of the inflorescence meristem (i.e. about 50 cells); whereas for a periclinal chimera, communication need only extend across 2 cells. The extent of intradermal autonomy can be further investigated by analysing deformed flowers from mericlinal chimeras that arise at the borders between mutant and revertant tissue. The junction between phenotypically mutant and revertant areas is sometimes but not always distinct, suggesting that intradermal spread of flo activity may be limited even within floral meristems.
The above results show that transposons can be used to generate chimeras for key developmental genes, allowing their action in different cell layers to be studied. Molecular analysis of these chimeras can give further insights into gene interactions, as described in the accompanying paper (Hantke et al., 1994).
ACKNOWLEDGEMENTS
We are grateful to Peter Walker for taking and maintaining the cuttings, to Lucy Copsey and Coral Vincent for scanning electron microscopy and to Sarah James for drawing the flowers. For critical reading of the manuscript and constructive comments, we thank Desmond Bradley, Pilar Cubas and David Hopwood. We are also grateful to the AFRC stem cell initiative and HFSPO for financial support.