Genetic Analysis of Organ Fusion in Arabidopsis thaliana
Susan J. Lolle, Wendy Hsu, Robert E. Pruitt


Postgenital organ fusion occurs most commonly during reproductive development and is important in many angiosperms during genesis of the carpel. Although a number of mutants have been described that manifest ectopic organ fusion, little is known about the genes involved in regulating this process. In this article we describe the characterization of a collection of 29 Arabidopsis mutants showing an organ fusion phenotype. Mapping and complementation analyses revealed that the mutant alleles define nine different loci distributed throughout the Arabidopsis genome. Multiple alleles were isolated for the four complementation groups showing the strongest organ fusion phenotype while the remaining five complementation groups, all of which show only weak floral organ fusion, have a single representative allele. In addition to fusion events between aerial parts of the shoot, some mutants also show abnormal ovule morphology with adjacent ovules joined together at maturity. Many of the fusion mutants isolated have detectable differences in the rate at which chlorophyll can be extracted; however, in one case no difference could be detected between mutant and wild-type plants. In three mutant lines pollen remained unresponsive to contact with the mutant epidermis, demonstrating that organ fusion and pollen growth responses can be genetically separated from one another.

IN plants the outermost layer of cells covering the shoot and root surfaces offers the first site of contact with biotic as well as abiotic factors present in the surrounding environment. The barrier presented by this epidermal layer is dynamic and selective, permitting some biological interactions while blocking others. During plant growth, for example, epidermal cell interactions play an important role in the elaboration of the shoot by regulating organ fusion, thus contributing to variation of the body plan, especially in flowers. During reproductive development epidermal derivatives provide a receptive surface for the hydration and germination of pollen, an important first step in the fertilization process. In some species the specialized epidermal cells that interact with pollen (the stigmatic papillar cells) provide the selection point where self-pollen is blocked during early development (Ockendon 1972; de Nettancourt 1977; Bell 1995), thereby selecting against self-fertilization and promoting outcrossing. Clearly, mediating the responsiveness of the shoot epidermis to external factors is critical to the viability, proper elaboration and growth of the plant, as well as the fertilization process itself.

In many plant species primordial structures are known to unite following initiation at the shoot apical meristem. This process, known as postgenital fusion, is achieved by a change in the responsiveness of the epidermis to physical contact with other epidermal cells and occurs most commonly during floral ontogenesis (Cusick 1966). Postgenital fusion represents a special case where the developmental potential of the epidermis is altered so that it no longer remains unresponsive to contact. In Catharanthus roseus, where this developmental response has been studied most extensively, it has been demonstrated that organ fusion involves the exchange of small, water-soluble morphogenetic factors (Siegel and Verbeke 1989). Although many of the epidermal cells along the fusion suture in C. roseus respond to contact by forming a tight cell wall association, a subset of approximately 400 cells dedifferentiate and redifferentiate into parenchyma cells (Walker 1975a,b,c; Verbeke and Walker 1985, 1986). Both cell wall adhesion and the cellular redifferentiation response are known to require recognition events because noncarpel epidermal derivatives will not undergo any of the changes associated with this contact-mediated response (Siegel and Verbeke 1989). The details of how this developmental pathway is regulated and information about the signals and signalling cascades involved remain elusive.

To date six mutations have been described which cause ectopic expression of an epidermal fusion response: adherent1 (Kempton 1920), adherent2 (Neufferet al. 1997) and crinkly4 (Becraftet al. 1996) in maize and fiddlehead (Lolleet al. 1992), wax1 and wax2 (Jenkset al. 1996) in Arabidopsis. In all six cases the mutations were identified by virtue of some impairment of normal expansive growth in mutant plants. In maize plants harboring the crinkly4 (cr4) mutation, for example, epidermal cells adhere to one another early in development, culminating in significantly stunted and misshapen adult plants (Becraftet al. 1996). Closer examination of cr4 mutants reveals that the epidermis itself is atypical and forms graft-like fusions. The CR4 gene has been isolated and encodes a putative receptor protein kinase (Becraftet al. 1996). In the Arabidopsis fdh-1 mutant, epidermal morphology is normal (with the exception of the fusion sutures) although plant growth is also severely affected (Lolleet al. 1992). In addition,analysis of the fdh-1 mutant revealed that the entire shoot surface was competent to promote pollen hydration, germination and tube growth (Lolle and Cheung 1993). The pollen response seen on the fdh-1 shoot surface reiterates a pollen-stigma type interaction in that the pollen growth response involves recognition events that limit the spectrum of pollen species that can grow on the fdh-1 surface to other members of the mustard family (Lolle and Cheung 1993). Furthermore, like the wild-type stigma, fdh-1 epidermis is unable to bypass the hydration block imposed by some of the cer mutations (Preusset al. 1993; Hülskampet al. 1995; Lolleet al. 1997). The wax1 and wax2 Arabidopsis mutants, on the other hand, show altered epicuticular wax as well as an organ fusion phenotype (Jenkset al. 1996). Whether or not these wax mutants interact with pollen has not been determined.

Two distinct mechanisms for promoting organ fusion are suggested by the studies done on the cr4 and fdh-1 mutants. It seems plausible in light of the sequence homology shown by the CR4 gene that fusions in these maize mutants result from disruption of the signalling pathway directing epidermal differentiation such that cells either do not achieve or do not maintain a terminal and unresponsive developmental state typical of epidermal derivatives. In fdh-1 mutants, on the other hand, a change in the permeability barrier of the outer epidermal cell wall and cuticle is correlated with and may be responsible for organ fusion (Lolleet al. 1997). It is postulated that the factors promoting fusion become accessible and are exchanged across fdh-1 cell walls between contacting cells, initiating the cascade of developmental events leading to cell wall adhesion and organ fusion. In the former case a putative receptor in the signal transduction pathway has been identified while in the latter case early events in signalling have become constitutive.

In this article we describe the isolation and characterization of 29 independently derived mutations causing organ fusion in Arabidopsis. The mutations fall into nine complementation groups putatively identifying nine genes. In the majority of mutants recovered, a change in the permeability barrier of the epidermal cell wall cosegregates with the organ fusion phenotype but in one case it does not, suggesting that organ fusion can be achieved by at least two distinct mechanisms in Arabidopsis. In addition, pollen hydration does not occur on all mutants showing the organ fusion phenotype, supporting the notion that organ fusion and pollen growth promotion define distinct biological processes.


Plant growth conditions: Plants were maintained under a 24-hr light regimen and were illuminated with a mix of fluorescent and incandescent lights (100–175 μmol m−2 sec−1 at pot level). Plants were grown in CustomBlen Plus (Griffin Greenhouse Supplies, Tewksbury, MA) soil mix and watered as needed with distilled water. The environment in the Percival (I-37LLVL and I-60LLVL) growth chambers (Percival Manufacturing Company, Boone, IA) was maintained at 25° (±1°) and 50% relative humidity.

Mutagenesis and plant manipulations: All of the new mutations isolated in our laboratory as a part of this study were recovered from mutagenesis with ethyl methanesulfonate (EMS). Seeds were treated with EMS essentially as described in Robinson-Beers et al. (1992). M1 plants were grown on soil and a single silique was harvested from each plant. All of the seeds from a single M1 plant were grown as a family and screened for the segregation of organ fusion mutants. In some cases mutations could be recovered directly from selfing of homozygous mutant plants, while in others they were recovered through heterozygous siblings present in the family. Approximately 15,000 M1 families were screened and 21 new organ fusion mutations were recovered (see Table 1).

Lines polymorphic for molecular markers were generated by manually crossing homozygous or heterozygous mutant plants (in one of three ecotype backgrounds: Landsberg, Columbia or Wassilewskija) to an appropriate wild-type line. F1 plants were allowed to self pollinate and set seed. F2 plants homozygous for the mutation of interest were then collected and tissue stored at −80° until processed for DNA-based mapping analyses.

Complementation analyses were done by crossing homozygous or heterozygous mutant plants to one another and the resulting F1 plants scored for the presence or absence of the mutant phenotype. In some instances, lines polymorphic for molecular markers were used to construct trans-heterozygous F1 plants in order to facilitate identification of both F1 and F2 progeny using the molecular markers.

Determination of genetic map positions: DNA was prepared from individual F2 plant samples according to Edwards et al. (1991). Briefly, tissue was ground in microcentrifuge tubes using disposable plastic pestles. Following grinding, 400 μl of extraction buffer (200 mm Tris-HCl pH 7.5, 250 mm NaCl, 25 mm EDTA, 0.5% SDS) were added and samples vortexed. Samples were centrifuged for 1 min in a microcentrifuge, 300 μl of the supernatant transferred to a fresh tube and an equal volume of isopropanol added. Samples were vortexed and left for 2 min at room temperature. Following a 5-min centrifugation, supernatants were discarded and pellets air-dried for 5–10 min at room temperature. DNA preparations were then resuspended in 100 μl TE (10 mm Tris-HCl pH 8.0, 1 mm EDTA) and stored either at 4° or at −20°.

Amplification of simple sequence length polymorphisms (SSLPs; Bell and Ecker 1994) or codominant amplified polymorphic sequences (CAPS; Konieczny and Ausubel 1993) was performed using polymerase chain reaction (PCR). Oligonucleotide primers were obtained from Research Genetics Inc. (Huntsville, AL) and Taq DNA polymerase and reaction buffer from Perkin-Elmer (Norwalk, CT). Deoxynucleotide triphosphates (dNTPs) were purchased from GIBCO-BRL (Gaithersburg, MD). Each 20-μl reaction mixture contained 1 μl of DNA, buffer, 200 μmol dNTPs, 0.5 units of AmpliTaq DNA polymerase (Perkin-Elmer, Norwalk, CT) and 5 pmol each of the forward and reverse primers.

Conditions for amplification of SSLPs were as follows: 1 cycle of 2 min at 94°, 15 sec at 55°, 30 sec at 72°, followed by 39 cycles of 15 sec at 94°, 15 sec at 55°, 30 sec at 72°. Conditions for amplification of CAPS were as follows: 1 cycle of 2 min at 94°, 15 sec at 55°, 2 min at 72°, followed by 39 cycles of 30 sec at 94°, 15 sec at 55°, 2 min at 72°. Restriction digests of CAPS products were carried out using the restriction enzymes specified in Konieczny and Ausubel (1993). PCR products were size-separated by Tris-Borate-buffered agarose gel electrophoresis. Agarose (4%) gels were used to resolve SSLP products and 1.5–2% agarose gels to resolve CAPS products. Linkage to specific SSLP or CAPS markers was determined by scoring a sufficient number of F2 individuals to give an LOD score larger than 3.0 using Mapmaker/EXP version 3.0 (Landeret al. 1987). Mutations which mapped to similar map positions were tested for allelism by complementation analysis.

Phenotypic analyses of organ fusion, fertility, porosity and pollen hydration: Plants were visually inspected for evidence of organ fusion during seedling, juvenile, adult and reproductive developmental stages. Fusion was scored as positive if two organs adhered to one another and could not easily be separated by gentle physical manipulation. Ovules were analyzed by dissecting open the ovary walls of the gynoecium and viewing ovules under a dissecting microscope (Leica/Wild M3C, Heerbrugg, Switzerland). Samples analyzed by scanning electron microscopy were fixed in FAA (50% ethanol, 3.7% formaldehyde, 10% acetic acid) and dehydrated through a graded ethanol series to 100% ethanol. Samples were critical point dried using a Samdri-PVT-3B unit and sputter coated with gold. Samples were viewed in an AMR 1000 scanning electron microscope at 10 kV.

Mutant plants were tested for male fertility by outcrossing to a male sterile/female fertile line (TH154; R. E. Pruitt, unpublished results) and scoring for silique elongation and seed set. Female fertility was tested by pollinating mutant plants with wild-type pollen and scoring for silique elongation and seed set. In cases where floral organ fusion was severe mutant flowers had to be dissected open to reveal the gynoecium.

The rate at which chlorophyll could be extracted as a measure of porosity was determined as described in Lolle et al. (1997). Tissue samples were collected following elongation of the inflorescence bolt (approximately 4 wk after planting). In all cases where fusion did not prevent physical separation, the rosettes were removed from the inflorescence bolt and the two tissue samples immersed separately in 80% ethanol. Samples were agitated gently on a rotating platform and 100μl aliquots of the ethanol solution removed at 10 min, 20 min, 40 min, 60 min, 80 min and 24 hr following immersion. Total chlorophyll content in each sample was determined using absorption readings taken at 647 and 664 nm using a UV-1201 spectrophotometer (Shimadzu Corp., Tokyo). Pollen hydration assays were carried out as described previously (Lolleet al. 1997).


Mutant isolation and complementation analysis: The isolation of the original fdh-1 mutant demonstrated the possibility of identifying genes whose products are required to suppress interorgan fusion through the isolation of mutations that promote such fusions. In order to exploit this we undertook the isolation of a large number of such mutations in a single line genetic screen, which would allow the recovery of any fully male and female sterile mutant through heterozygous sibling plants. As shown in Table 1, 21 mutants were isolated in this screen, which displayed an organ fusion phenotype. During the course of our genetic screen we also obtained an additional eight mutants from the sources indicated in Table 1. All of the mutant phenotypes described segregated in a manner consistent with monogenic, recessive mutations. Each mutant line was outcrossed to an appropriate wild-type line to generate an F1 line bearing polymorphisms for molecular markers suitable for use in DNA-based mapping procedures. Analysis of F2 plants for each of these outcrossed mutant lines revealed linkage to nine different SSLP or CAPS DNA markers distributed throughout the genome. Complementation analysis between mutants with similar map positions confirmed that our collection of fusion mutants defines nine distinct genes. The genetic map positions of these complementation groups are illustrated in Figure 1 and their morphological phenotypes are described individually as follows:

airhead (ahd) complementation group: Only a single allele was isolated in this complementation group. Of all the fusion mutants we have characterized, this mutant showed one of the weakest organ fusion phenotypes, which is manifested only in fusion between floral organs (Table 2). Fusions are most frequently observed between sepals and petals, resulting in poor emergence of the latter. Although interorgan fusions within the flowers are common, no morphological aberrations have been observed in the ovules. Fertility is normal (see Table 2).

bulkhead (bud) complementation group: Only one allele belonging to this complementation group was identified. Like airhead, this allele causes only weak fusion where fusion events are essentially restricted to floral organs (Figure 2). A striking feature of this mutant, however, is seen in the severity of the ovule phenotype. As shown in Figure 3, C and D, ovule morphology is highly abnormal. Although the funiculus can be identified for each individual ovule the remaining structure is severely compromised and shows little morphological integrity. Not surprisingly, this mutant is a tight female sterile (Table 2).

conehead (cod) complementation group: Plants homozygous for the conehead-1 mutation display either a weak fusion phenotype similar to bulkhead-1 mutants or a severe fusion phenotype like that seen in fdh mutants with approximately equal frequency (Figure 2). We were unable to separate the weak and strong fusion phenotypes even after 3 backcrosses; when the two phenotypes were mapped independently they mapped to the same location. These two distinct phenotypes may represent modification of the cod mutant phenotype by a second segregating locus, but we have not yet attempted to map this modifier. These mutant plants show no ovule defects and are both male and female fertile (Table 2).

View this table:

Summary of organ fusion mutations

deadhead (ded) complementation group: Four alleles were identified which fell into this complementation group. All of the mutants in this group show a marked surface luster consistent with a waxless or eceriferum phenotype. As summarized in Table 2, for the ded-1 and ded-2 alleles, fusion can occur upon emergence of the first true leaves. Only a small fraction of these mutant plants grows to produce an inflorescence as fusion often limits normal vegetative growth. Early fusion events are also typical for the ded-4 allele but not the ded-3 allele where fusion events are commonly limited to the inflorescence and flowers. Ovules show a strongly aberrant phenotype in ded-1 whereas ovule defects seen in ded-2 mutant plants are less severe. No ovule phenotype was observed in ded-4 mutant plants (ded-3 ovaries were not surveyed). ded-1 and ded-2 are female semi-sterile (Table 2), whereas ded-3 and ded-4 mutants show normal fertility.

Complementation analyses revealed unusual allelic interactions at this locus. As a consequence all possible pair-wise allele combinations were constructed and F1 plants scored for their wax and fusion phenotype. As summarized in Figure 4, some transheterozygotes showed a wild-type phenotype while others showed only a waxless phenotype or both a waxless and organ fusion phenotype. Because of the wax phenotype observed in ded mutants, all of these mutations were also complementation tested against cer5-1, the only cer mutant known to map in the same vicinity. All of these crosses produced only wild-type F1 progeny. Mapping data confirmed that cer5-1 mapped to a nearby but distinct map position relative to the DED gene.

eceriferum10 (cer10) complementation group: The cer10 locus has been described previously by Koornneef et al. (1989). These mutants have a pleiotropic phenotype including reduced plant height (Koornneefet al. 1989). However, in these descriptions no mention was made of the organ fusion phenotype. As indicated in Table 2, cer10 mutants manifest a fusion phenotype similar to the bulkhead-1 mutant. Although fusion is restricted to floral organs, the ovules are not affected (Table 2). These mutants are male semisterile, as previously noted by Koornneef et al. (1989) but are female fertile (Table 2).

fiddlehead (fdh) complementation group: Five new alleles of the FDH locus were identified in this study and one additional allele was obtained from a colleague (see Table 1). Like the original fdh-1 mutant, described previously by Lolle et al. (1992), all of these new alleles show a strong organ fusion phenotype (Figure 2). Although organ fusion is variable from one individual plant to the next, for every allele it is clear that all organs are competent to fuse, including the first true leaves. Both fdh-5 and fdh-6 manifest a defective ovule phenotype (Table 2) and all alleles tested (fdh-1, fdh-3, fdh-5 and fdh-6) are female semisterile (Table 2).

Figure 1.

Schematic representation of genetic map positions for the nine genetic loci. The distance of each locus from a linked molecular marker is indicated, together with the direction from that marker, if known. Three of the loci map to chromosome 1 and two each to chromosomes 2, 3 and 5, respectively.

hothead (hth) complementation group: Eleven alleles define the HTH locus. The fusion phenotype of these mutants is intermediate in strength with organ fusion generally being limited to flowers (Table 2). The floral organ fusion phenotype is stronger than that seen in either bud-1 or ahd-1 but not strong enough to completely block petal emergence (see Figure 2) or self-fertilization. Although the majority of gynoecia show no ovule defects, approximately 10–20% of the gynoecia surveyed in the hth-8 and hth-10 mutants showed evidence of ovule abnormalities while no defects were seen in any of the ovaries sampled from hth-4 mutants.

View this table:

Summary of mutant phenotypesa

Figure 2.

Light micrographs showing the fusion phenotypes associated with various mutant alleles. As shown in A, wild-type Landsberg er inflorescences show a typical spiral phyllotaxy where contact between neighboring buds decreases as development progresses. As individual flowers mature petals elongate and emerge unimpeded by the surrounding sepals. In bulkhead (B), hothead (G) and pothead (H) mutant petals are restricted in their growth but do emerge from some flowers (arrow in G). In mutants showing stronger fusion phenotypes, like conehead (C), deadhead (D), thunderhead (E), and fiddlehead (F) mutants, organ fusion alters the configuration normally seen in the inflorescence and blocks petal and anther emergence. The pistil, however, usually protrudes out from the individual floral buds. In thunderhead-1 mutants the first 3–4 rosette leaves (as shown in E) do not adhere and have a normal phyllotactic relationship. Upon emergence of subsequent leaves, however, fusion commences resulting in a central trumpet-like clustering of leaves (arrow in E). The inflorescence is completely sequestered within these leaves but often will emerge as the plant continues to grow. Magnification of panel A, B, D, F and H is the same. Magnification of panels C and G is the same. Bars, 1 cm.

Mapping data from the HTH locus showed that a subset of hth alleles recombined with a flanking SSLP marker at a higher frequency than the remaining group of hth alleles, although crosses between these two groups of alleles failed to show complementation in the F1 generation. To determine the frequency of recombination which took place between these different classes of hot-head alleles, hth-4, hth-8 and hth-10 mutants polymorphic for Columbia and Landsberg molecular markers were used to generate F1 plants. The identity of these F1 plants as heterozygotes was confirmed by demonstrating that they were heterozygous for SSLP markers which were homozygous in each parental line but polymorphic between the parental lines. F2 plants derived from individual F1 plants were then scored for segregation of wild-type individuals. The identity of all putative wild-type F2 plants was confirmed by demonstrating that they were homozygous for at least three SSLP markers that were homozygous in the F1 plants and represented both Landsberg and Columbia alleles. All of these plants were also progeny tested and shown to segregate 3:1 for the fusion phenotype after selfing. As shown in Table 3 only F2 progeny derived from hth-8/hth-10 transheterozygotes segregated wild-type plants. No wild-type progeny were found among F2 plants derived from the hth-4/hth-10 transheterozygotes.

pothead (phd) complementation group: Only one allele for this locus was isolated. Organ fusion in the phd-1 mutant plants is limited to the flowers and is even weaker than that seen in ahd-1 (see Figure 2). As in ahd-1, fusion is most often seen between sepals and petals and interferes with the proper deployment of the petals. Ovules are normal in appearance and mutant plants are self-fertile (Table 2).

thunderhead (thd) complementation group: Organ fusions in plants harboring mutations at this locus initiate late in vegetative development (see Table 2). Of the two alleles isolated, plants homozygous for the thd-1 allele show a sharper transition to fusion competence. In thd-1 plants, all leaves formed subsequent to the expansion of the first four to five leaves fuse together, as do the inflorescence and flowers (see Figure 2). thd-2 mutants, on the other hand, show a more gradual onset of late vegetative fusion but display similarly severe floral organ fusions. In plants homozygous for either mutant allele, ovules are joined to one another at maturity but the severity of the phenotype is greater in the thd-1 mutant ovaries (Figure 3). Although many ovules are affected adversely in the mutants, some ovules in either case appear morphologically normal and fully differentiated. Mutant plants are female semisterile (Table 2).

Figure 3.

Scanning electron micrographs showing normal wild-type ovules and ovule phenotypes of three representative mutants. In all cases, ovary walls were removed to reveal the ovules. The boxed areas in A, C, E and G are shown at higher magnification in the adjacent and corresponding panels (B, D, F, and H). In the wild-type ovary (A and B) each locule contains two parallel rows of ovules (o) on either side of the septum. Pollen grains land on the stigmatic papillar cells (p), germinate and grow tubes which penetrate into the ovary and fertilize the ovules. As shown in B, ovules are tightly packed within the ovary but manifest no adhesion. Each ovule is anchored to the ovary only by the funiculus (f). In bulkhead-1 mutants (C and D) ovules (o) show little morphological integrity. The only identifiable structure is the funiculus (f). In thunderhead-1 mutants (E and F), although the majority of ovules are physically joined to one another at maturity, some ovules develop relatively normally. As shown at higher magnification in F, one ovule has developed without attachments to neighboring structures and not only has a funiculus (f) but also a micropyle (arrow). In thunderhead-2 mutants (G and H), on the other hand, a larger proportion of the ovules remain unattached to their neighbors and show normal morphology. Bars, magnification in micrometers.

Pollen hydration: One goal in characterizing a larger collection of fusion mutants was to determine whether the pollen hydration response always cosegregated with the organ fusion phenotype. Table 2 summarizes the results of the pollen hydration assays. Some mutants clearly support a rapid hydration response time similar to that seen on fdh-1 plants (cod-1 and thd-2, data not shown). In other instances hydration takes place but is attenuated (ahd, cer10, ded, hth, data not shown). However, three mutants show no hydration response with wild-type pollen: bud-1, phd-1 and thd-1. Two of these, bud-1 and phd-1, manifest a weak fusion phenotype while the thd-1 mutant has a very strong organ fusion phenotype. Although the thd-1 mutant does not promote hydration, pollen grains hydrate within 10 min when applied to the thd-2 mutant surface. No hydration of the cer1-147 pollen was observed on any of the mutant plants tested, indicating that in all cases the pollen hydration observed requires specificity on the pollen side of the interaction similar to that seen on the Arabidopsis stigma (Hülskampet al. 1995; Preusset al. 1993).

Permeability to chlorophylls: Representative alleles from each of the loci identified in this study were assayed for changes in the rate at which chlorophyll could be extracted. The results are summarized in Figure 5. As is evident from the graphed data, by this criterion there exists a wide variation in porosity among the mutants tested. Clearly, the most permeable samples are found in the fdh and ded complementation groups (Figure 5, B and D). Only one mutant showed a relatively intact chlorophyll permeability barrier (that is, similar to wild type): phd-1 (Figure 5G). The remaining loci show a range of permeabilities, but all show some enhancement of permeability to chlorophyll relative to Landsberg. In cases where the rosette could be assayed independently from the inflorescence, the inflorescence usually showed a greater permeability (see Figure 5, A–D, F and H). thunderhead plant tissues were further subdivided by separating the juvenile from adult leaves (or adult leaves plus inflorescence in the case of thd-1). As shown in Figure 5H, the adult leaves of thd-2 show a permeability profile similar to the juvenile leaves of the thd-1 allele. However, both show approximately equivalent permeability changes in their inflorescence tissues.


Examples of postgenital organ fusion can be found in many angiosperm species; this fusion is thought important in both increasing developmental flexibility during floral ontogeny and facilitating the mechanics of the pollination process. The most extensive analysis of the cell biology of postgenital organ fusion has been done in C. roseus (Walker 1975a,b,c; Verbeke and Walker 1985, 1986). Based on the results of these studies it is known that organ fusion is an epidermis-specific interaction, involves reciprocal recognition events and is mediated by small diffusible water-soluble molecules (Siegel and Verbeke 1989). Another type of cell-cell interaction showing many similarities to epidermally-mediated organ fusion is the pollen-stigma interaction. In this case a specific recognition reaction also takes place as well as an exchange of factors between the participating cells. Not only do pollen grains adhere to the papillar cell (an epidermal derivative) in a manner akin to an organ fusion-like event, but this interaction triggers the further development of the male gametophyte. Interestingly, lipids have recently been highlighted as playing regulatory roles in both processes (Hülskampet al. 1995; Lolleet al. 1997; Preusset al. 1993).

Figure 4.

A summary of the complementation phenotypes from all pair-wise crosses of the ded alleles. As shown, one allele combination shows full interallelic complementation (ded-2 allele crossed to ded-4) while two combinations show partial complementation restoring only the fusion phenotype to normal (ded-3 crossed to either ded-2 or ded-4). cer5 complements all ded alleles.

In this article we describe the isolation of 21 new mutations and the further characterization of an additional eight mutants which manifest an organ fusion phenotype. In addition to the original fdh locus described previously (Lolleet al. 1992), eight additional loci were identified which are involved in mediating organ fusion. Multiple alleles were found for three loci (ded, fdh and hth), two alleles for the thd locus but only single alleles at the remaining five loci (ahd, bud, cer10, cod, phd). With the exception of the conehead mutant, the four genes with only one representative allele are represented by mutants that all manifest weak organ fusion. Due to the subtle nature of the fusions other mutations that fall into these complementation groups may have been overlooked in the course of our genetic screen. It may be possible in the future, however, to manipulate environmental conditions or to use other sensitized genetic backgrounds such that the frequency or severity of organ fusion is enhanced, simplifying identification of other alleles at these loci. The relative map positions as determined using DNA-based mapping procedures reveal that the genes identified in this article are not clustered but map throughout the Arabidopsis genome.

As summarized in the results section, the fusion phenotype varies among members in this mutant collection from a severe phenotype like that of the original fdh-1-mutant (Lolleet al. 1992), to a very subtle fusion phenotype such as is the case for the ahd-1 and phd-1 mutants. Plants with a severe fusion phenotype can suffer serious growth impairment while plants with a weak phenotype are relatively normal in appearance. In addition to fusion of organs comprising the external shoot, in some mutants ovules with abnormal morphology were also found. Although these abnormal structures (in which adjacent ovules are joined together at maturity) are reminiscent of fused structures seen in other parts of the plant, determination of the mechanism which leads to these structures will require a more detailed examination of ovule development in these mutants. Surprisingly, ovule defects were detected not only in mutants showing strong fusion between organs but also in mutants like bud-1, which manifest a relatively weak fusion phenotype. Although the bud-1 mutant is a tight female sterile, other mutants where ovule abnormalities were detected (ded-1 and thd-1) are not, indicating that ovaries containing large abnormal structures derived from many ovules need not block the fertilization process in the few remaining normal ovules. The detectable presence of a funicular structure with relatively normal morphology in all mutants that manifest ovule defects suggests that deviation from normal ovule development occurs after the ovule primordia have elongated and the integuments have initiated (stage 2-II; Schneitzet al. 1995). Further analysis of ovule development should help illuminate the ontogeny of these aberrant structures and what, if any, cellular consequences these abnormalities have for the embryo sac as well as the sporophytic tissues of the ovule.

View this table:

Summary of hothead intragenic recombination frequencies

The deadhead locus which we have described has an interesting pattern of interallelic complementation. Plants homozygous for any of the individual ded mutant alleles have a strong eceriferum phenotype as well as manifesting relatively severe organ fusion. Heteroallelic combinations have widely varying phenotypes, however, ranging from completely wild-type plants to plants with as severe a phenotype as the individual alleles. The present complementation analysis fails to define any obvious subgroupings of alleles. One allele (ded-1) fails to complement all other alleles, but each of the other three alleles results in both the cer and organ fusion phenotypes only when homozygous or in combination with ded-1. The pattern is further complicated by the fact that plants bearing some combinations of alleles (ded-3/ded-4; ded-2/ded-3) have an obvious cer phenotype but fail to undergo any organ fusion whatsoever. More mutant alleles of this locus are needed to allow a better definition of this complex complementation pattern. Characterization of the mutants presently available as well as molecular analysis of the DED gene may allow us to determine the nature of the interallelic complementation seen at this locus.

Of all of the organ fusion mutants we have recovered, only mutations at the DEADHEAD locus show a wax defect. The deadhead eceriferum phenotype suggests a defect in lipid biosynthesis (Aartset al. 1995; Hannoufaet al. 1996; Negruket al. 1996; Xiaet al. 1996), and, as indicated by biochemical analyses of fdh-1 mutants, changes in lipid composition of the outer epidermal cell wall and cuticle may play some role in promoting organ fusion. Two other Arabidopsis mutants which also show organ fusion and have a wax defect have been described (Jenkset al. 1996) and may be allelic to our ded mutants, although complementation tests have not yet been performed. However, deadhead does not appear to be allelic to cer5 which maps to a similar location in the genome. A detailed biochemical analysis of these mutants may reveal the loss or modulation of a lipid component that may be important for maintaining epidermal developmental integrity.

A second locus which has interesting phenotypic behavior is CONEHEAD. In this case, plants which are homozygous for the mutation fall in equal numbers into two discrete phenotypic classes: those that exhibit strong fusion and those that exhibit weak fusion. Both of these phenotypes were present in the original single line family in which the mutation was isolated and both phenotypes have persisted through three backcrosses to the parental wild-type line, Landsberg erecta. Although it is possible that this variation in phenotype is due to a genetic modifier which is segregating in this line, it is hard to account for the fact that it has been carried along through the backcrosses (implying close linkage to the cod mutation) and yet is readily separable from cod to produce the two different phenotypes. Alternatively, it is possible that the mutation in COD is itself responsible for both phenotypes with variable expressivity of the gene somehow producing two discrete classes of plants. The isolation of additional alleles of this locus and COD gene isolation may help clarify how these mixed phenotypes are achieved.

Figure 5.

Graphs showing the percent of total chlorophyll extracted from various fusion mutant tissues as a function of time. Rosette leaves (R) were assayed separately from the inflorescence (I) tissues. Rosettes were further subdivided for thunderhead mutants into juvenile (J) leaves and adult (A) rosette leaves. Results from tests using wild-type tissue of the relevant ecotype (Col: Columbia or Ler: Landsberg erecta) are included on all graphs. Data from assays on fiddlehead (fdh) mutants are shown in B. Those cases where the percentage of chlorophyll extracted is significantly different from wild type after 80 min are marked in the legends to the graphs (*P < 0.05; **P < 0.01).

The disproportionately large number of mutations recovered in the HOTHEAD complementation group suggest that this locus may represent an unusually large gene (or at least a very large target for the mutagen EMS). One corroborating piece of evidence comes from our analysis of recombination frequencies between different hth alleles. Based on estimates derived from these recombination data the HTH locus may span as much as 1.6 cM. This is dramatically larger than similar recombination estimates for other Arabidopsis genes (0.07 cM for GA1; Koornneefet al. 1983; 0.01 cM for CSR1; Mouradet al. 1994). A number of possibilities may explain this discrepancy. First, it is possible that the apparent recombinants actually represent reversion of one of the two mutations (presumably hth-8, since hth-10 was present in both crosses). While we have not attempted to measure this directly, we consider this an unlikely explanation since it would indicate a spontaneous reversion rate of 1.6 × 10−2. Second, it is possible that the HTH locus contains a particularly recombinogenic region within its borders. In this case the HTH locus may be a gene of quite ordinary size despite the measured recombination frequencies between alleles. Third, the HTH locus may be split by a large intervening sequence as is the case in a number of other genes, such as some of the homeotic genes of Drosophila (Gehring and Hiromi 1986). Fourth, it remains a formal possibility that this locus is made up of two distinct genes that interact in some manner such that they fail to complement one another in a simple cis-trans test. Although further analysis of recombination frequencies may reveal how the alleles group on the genetic map, not until the gene is isolated and characterized will its structure be revealed.

At the THUNDERHEAD locus, one of the two mutant alleles (thd-1) shows a sharp transition from normal growth to fusion competence, which appears to coincide well with the juvenile to adult phase transition. In these mutants fusion cannot be detected until emergence of the fourth or fifth leaf. The mutant plants when mature show a central trumpet-like clustering of leaves surrounded by a relatively normal rosette. The thd-2 allele does not show a similarly sharp transition, but the onset of organ fusion is still confined to late adult development. Plants are known to manifest a variety of morphological and biochemical changes during different phases in their development (Lawson and Poethig 1995) and it may be that at this locus the juvenile to adult phase transition plays some role in regulating THD gene expression. Interestingly, the adherent2 locus in maize shows a similar restriction of organ fusion to the adult phase of the plant life cycle (Neufferet al. 1997).

Although the mutants described in this article were all selected on the basis of having an organ fusion phenotype, epidermal cells in many of them also interact with pollen, as was the case for fdh-1 (Lolle and Cheung 1993). This fact clearly indicates that these two seemingly distinct processes are in some way genetically related, since mutations in several different genes bring about similar alterations in both processes. On the other hand, three of the new organ fusion mutants which we have identified, representing three different complementation groups, fail to support pollen hydration on their nonreproductive epidermal surfaces. Taken together, these results indicate that while these two processes may be related, they are also clearly separable genetically, and it will be very interesting to learn the molecular identities of the gene products which are in common as well as those which are unique to each of the two pathways.

In the original fdh-1 mutant a striking increase in the permeability of the epidermal cell wall and cuticle was the only alteration detected that distinguished mutant plants from wild type (Lolleet al. 1997). Although it seemed plausible that this permeability change might play a role in the acquisition of fusion competence by the epidermal cells, no corroborating evidence supported this hypothesis. The fact that many of the new organ fusion mutations isolated, which fall into a number of different complementation groups, also show changes in permeability to chlorophyll, however, is consistent with this hypothesis. Furthermore, it is clear that thunderhead mutant plants switch to organ fusion competence at a discrete time in development and that this switch coincides precisely with a change in permeability. As such, these data strongly support the notion that changes in the permeability of the epidermal cell wall and cuticle to small molecules represent one mechanism which can lead to ectopic organ fusion during Arabidopsis development.

One of the mutants recovered in this study fails to show any change in permeability that can be detected with our present assay. This mutant may, of course, have permeability changes below the limit of our detection or it may have changes in permeability to the relevant signalling molecules that do not affect the permeability of our test molecule, chlorophyll. Alternatively, this mutant may represent a class of mutations which allow organ fusion to take place by some other mechanism, perhaps one related to the cr4 mutant of maize (Becraftet al. 1996). Interestingly, an Arabidopsis mutant (blasig) having an epidermal phenotype similar to cr4 but manifested only in the ovules has previously been characterized in this lab (Schneitzet al. 1997). A more detailed examination of the epidermal morphology of the mutant showing organ fusion without an accompanying change in permeability as well as the blasig mutant will be required to see if any of the same features characterizing the cr4 epidermis are also manifested in these mutants.

We undertook a genetic approach to studying organ fusion with the goal of identifying as many genes as possible which are involved in this developmental process. Ultimately, we hope to identify a variety of molecular players in this process either directly by screening for plants showing organ fusion or secondarily by looking for enhancers and suppressors of the primary mutations. At the outset we expected to find mutations altering the barrier to diffusion of signalling factors (such as described for fdh-1; Lolleet al. 1997), possibly mutations identifying the factors involved in the responses (the signalling molecules themselves) or mutations affecting members of the downstream signal transduction pathway. The collection of mutants that we present here clearly contains mutations in a number of different genes which alter the permeability barrier as well as one example of a mutation which probably leads to organ fusion by a different type of primary defect. It may be that this class of mutations will include the Arabidopsis homolog of the putative CR4 receptor protein kinase (Becraftet al. 1996). Although the evidence is only circumstantial, it may be that both processes are part of the same developmental pathway where changes in wall permeability permit exchange of factors that modulate CR4 receptor kinase activity. In cr4 mutant plants epidermal cells may not fully differentiate or only become partially dedifferentiated and hence be competent to form graft-like fusions, while in fdh-1 mutant plants the epidermis may become primed to exchange factors that will initiate the cascade of biological events that promote cell wall adhesion. Molecular characterization of the genes identified in this article may help to resolve questions about how these two types of organ fusion mutations are related mechanistically.


We are indebted to Ceri Batchelder, Martin Hülskamp, Steve Kopczak and Kay Schneitz for their willingness to screen for organ fusion mutations while pursuing other mutants of their own. We also thank Angeline Chong, Phyllis Itoka, Katherine Krolikowski, Phyllis Maffa, Evelyn Pizzi and Janet Sherwood for invaluable assistance with various aspects of this project. We thank Edward Seling of OEB and MCZ at Harvard for assistance with sample preparation and use of the scanning electron microscope. We also thank Allen Sessions, David Smyth, Lawrence Hobbie, Chris Somerville and the Arabidopsis Stock Center for making seeds available from mutant lines showing a fusion phenotype. We are grateful to Ueli Grossniklaus and members of his laboratory as well as Graeme Berlyn for their continued interest in this project and for many helpful discussions. This work was supported in part by a grant from the Clark Fund at Harvard University, Harvard College Research Foundation Fellowships and by National Science Foundation (NSF) Grant IBN-9405391 awarded to R.E.P. The main body of the work described in this article was supported by NSF Grant IBN-9596044 awarded to S.J.L. and NSF Grant IBN-9723563 awarded to R.E.P. and S.J.L.


  • Communicating editor: D. Preuss

  • Received December 31, 1997.
  • Accepted March 3, 1998.


View Abstract