MGOUN1 and MGOUN2: two genes required for primordium initiation at the shoot apical and floral meristems in Arabidopsis thaliana

The number and arrangement of the aerial organs of plants are entirely determined by the activity of small populations of mitotic cells, called shoot apical meristems (SAMs). These structures therefore determine an important part of plant architecture.

Based on analyses of ultrastructure and mitotic frequencies, it has been proposed that SAMs are organised into three distinct zones (reviewed by Steeves and Sussex, 1989 and more recently by Clark, 1997; Kerstetter and Hake, 1997): a central zone (CZ), a peripheral zone (PZ) and an underlying rib zone (RZ). The CZ contains slowly dividing cells, and is thought to ensure a constant renewal of stem cells. The progeny of these stem cells is subsequently recruited into the PZ, where organ primordia initiation takes place, or into the RZ which gives rise to the pith and central tissues of the shoot axis. According to this model, the number and arrangement of organs will depend on two parameters: (i) the number of cells produced by the CZ and the PZ that are available for organ initiation, and (ii) the way these cells are subsequently partitioned into the primordia.

Recent genetic and molecular studies have confirmed this general model and allowed a first insight in the regulation of organ initiation. In particular, in Arabidopsis, mutants with abnormal organ arrangements and numbers have been isolated. One class of mutants is mainly perturbed in meristem maintenance and cell proliferation within the meristem. This is thought to lead indirectly to the production of aberrant numbers of organs since the number of cells produced by the CZ and which become available for primordium formation is modified in these mutants. Examples of this class are shoot meristemless (stm) (Barton and Poethig, 1993; Endrizzi et al., 1996) and wuschel (wus) (Laux et al., 1996), which are unable to maintain a functional meristem, and clavata (clv 1,2,3) and fasciata (fas1,2), which have enlarged meristems (Leyser and Furner, 1992; Clark et al., 1993, 1995). So far only the STM and CLV1 genes have been isolated. STM belongs to the KNOTTED class of homeobox genes, and it is expressed in the central parts of the shoot apical and floral meristems but not in the primordia (Long et al., 1996). CLV1 encodes a putative receptor kinase, that is expressed in the inner parts of the meristem (Clark et al., 1997). It has been proposed that the CLV1 protein could function in cell-to-cell signalling. Another class of mutants appears to affect more directly the partitioning of cells into leaf, flower or floral organ primordia. The pin-formed and pinoid mutants of Arabidopsis, are unable to initiate flower primordia (Bennett et al., 1995; Okada et al., 1991), and in the perianthia mutant, the typical Arabidopsis crucifer-type flower is changed into a flower bearing 5 organs in the 3 outermost whorls (Running and Meyerowitz, 1996).

Although these results have provided a first insight in the molecular basis of organ initiation, it is obvious that we need to identify additional factors involved in this process. For this purpose, we have screened for mutants showing abnormal numbers of leaves and floral organs. Here we describe two mutants (mgoun1 and 2) which show such defects throughout development: abnormal leaf numbers and floral organ numbers are observed in combination with increased stem fasciation. The observed phenotypes and the genetic analyses suggest that the MGO genes are necessary for the proper initiation of primordia.

Plants were grown in vitro as described by Santoni et al. (1994). Two- to three-week-old plants were transferred to the greenhouse for further analyses.

The following strains were used: fas1-1, fas2-1, clv2-1 obtained from the ARBC, the Ohio State University, US. stm-5 was a generous gift from Dr T. Laux, Universität Tübingen, Germany. clv3-1 and stm- 2 were obtained from Dr S. Clark, University of Michigan, Ann Arbor, US. All strains were in the Landsberg erecta (Ler) background except fas1-1 which was in the Enkeim background.

Mutants were identified either in vitro at the seedling stage, or later in the greenhouse, by determining the number of leaves or floral organs. mgo1-3 was identified after EMS mutagenesis (Columbia ecotype (Col); Santoni et al., 1994). mgo1-1, mgo1-2, and mgo2 were identified in a T-DNA insertion mutagenesis screen (Wassilewskaya ecotype (WS); Bechtold et al., 1993; Bouchez et al., 1993) The mgo2 phenotype is linked to the T-DNA kanamycin resistance marker (all mutants tested are kanamycin resistant n=1 054).

For mapping, mgo1-1 and mgo2 (WS background) were crossed with the wild-types Ler and Col. Forty-four seedlings homozygous for mgo1-1 and mgo2 were selected in the F₂ population. We determined the chromosome location of mgo1-1 and mgo2 mutations by searching for linked WS-type alleles of cleaved amplified polymorphic sequence (CAPS) markers (Konieczny and Ausubel, 1993).

In all cases, double mutants were identified, in the F₂ generation, as plants with a new phenotype. Frequencies of these plants are compatible with a segregation ratio of 1:16 confirming that they correspond to the double mutants. In addition, the different genetic backgrounds did not have any major effect on the phenotypes of the single mutants which could be clearly recognised in the segregating F₂ population.

mgo1-1 mgo2, mgo1-1 fas1, mgo1-1 fas2, mgo2 fas2 and mgo1- 1 stm-2 double mutants could be clearly distinguished 2 weeks after germination. When quantifying the floral organs and primordia of stm and double mutants (Table 2 and 3), we used stm mutants (stm- 2 and stm-5) which segregated in F₂ populations, deriving from crosses with WS wild-type, as controls. The double mutants clv3-1 mgo1-1 and clv3-1 mgo2 were identified in the greenhouse as plants with an extremely fasciated stem. Since they were fertile, further analysis was performed on the progeny of these plants. No mgo2 fas1 and mgo2 stm double mutants were found due to the tight linkage of these loci: stm, mgo2 and fas1 map respectively at position 75, 82 and 88 of the classical genetic map (see Results and http://mutant.lse.okstate.edu/genepage/classical_map.html).

Histological sections were prepared as described by Traas et al. (1995) except that the sections were stained in toluidine blue (0.01% in water). No in situ data are available for the expression pattern of the KNAT2 gene but experiments using a KNAT2 promoter GUS fusion suggest that this gene is likely to be a L3 marker (Dockx, 1995). GUS was detected using standard procedures. Sections (unstained) were made as described above and viewed in a Nikon FXA microscope using dark-field illumination.

Plants were fixed in 4% formaldehyde (fresh from paraformaldehyde) in PBS under vacuum for 2× 20 minutes, and left in fixative overnight. After fixation, plants were washed, dehydrated, and embedded in paraffin, essentially as described by Jackson (1991). Paraffin sections (8-10 μm thick) were cut with a disposable metal knife and attached to precoated glass slides (Fischer Scientific, US). Sense and antisense probes of a full length cDNA of STM (a generous gift from Dr K. Barton, University of Wisconsin-Madison, US). were synthesised using digoxigenin (DIG-UTP; Boehringer Mannheim) according to the manufacturer’s instructions. In situ hybridisation was carried out as described by Jackson (1991). Immunodetection of the DIG-labelled probes was performed using an anti-DIG antibody with phosphatase as described by the manufacturer (Boehringer Mannheim). Sections were stained with calcofluor white (0.01% in water).

Plants were fixed, and then stained with propidium iodide to visualise DNA, as described by Clark et al. (1995). Optical sections were made using a Leica TCSNT confocal microscope. Meristem structure was studied using low-temperature scanning electron microscopy as described by Traas et al. (1995).

In our screen we identified four mutants with a similar phenotype. While these four mutants are all the same size as the wild type, at the adult stage they all produce fewer leaves and have enlarged and flattened (fasciated) stems (Fig. 1A-I). The mutants have flowers with abnormal numbers of organs that are of variable sizes (Fig. 1J-L).

Genetic tests showed that we have identified four recessive mutations falling into two complementation groups : mgoun1 (3 alleles) and mgoun2 (1 allele). Mgoun is a ridge-shaped mountain in the Atlas range which reminded us of the fasciated meristem of these mutants (see later). The three mgo1 alleles are very similar and most of the results presented here were obtained with mgo1-1.

mgo1 was mapped to the bottom of chromosome 5: 11.8 cM distal to LFY3, and 6.8 cM and 19.2 cM proximal to g2368 and m555, respectively. mgo2 was mapped to chromosome 1 between GAPB (16.4 cM away) and ADH (22.3 cM away), which corresponds to position 82 on the genetic map. mgo2 shows complementation with clv2, stm and fas1 mutants.

In wild type, the first two opposite leaves emerge 5 days after germination under our in vitro conditions and 5 to 6 leaves will develop before bolting (Table 1). In mgo1-1 and mgo2, the first leaves emerge 2 days later than in the wild type and the number of rosette leaves is variable, with a mean value of about 2 (Table 1). Mutant leaves are lanceolate or asymmetrical. In mgo1-1, one of the two first leaves is replaced by a finger-like structure in 23% of the plantlets (Fig. 1G-I). Root growth is unaffected in mgo1-1, whereas mgo2 has a shorter root than the wild type. The position of the cotyledons is abnormal in 45% of mgo1-1 plantlets.

The number of floral organs is constant in the wild-type: all flowers have 4 sepals, 4 petals, 2 carpels. The average stamen number is 5.2±0.1 due to the absence in some flowers of 1 or 2 of the expected 6 stamens (Table 2 and Fig. 1J).

In mgo1-1 and mgo2, the number of floral organs in each whorl is highly variable (Table 2 and Fig. 1K, L). For instance, 3–6 sepals are found in mgo1-1, and 1–7 stamens are found in mgo2 flowers. Although the average number of sepals is close to WT (4), the average number of petals and stamens is reduced in both mutants (Table 2). There are slightly more carpels in mgo1-1 and mgo2. The average total number of floral organs is reduced in both mutants compared to the wild type (Table 2). Both mutants are fertile, but mgo1-1 shows reduced fertility.

In addition, organ width is variable, mainly in the first whorl, although overall organ shape is normal (Fig. 1K-L). Filamentous structures in the second or third whorl are visible in some flowers.

We subsequently investigated the stage at which organ development is perturbed in both mutants. We chose to study sepal initiation as an indicator of organ development because these organs are easily accessible for microscopy. In wild-type plants, the abaxial sepal primordium arises first, followed by the adaxial, and finally, the two lateral primordia (Fig. 2A,B; see also Smyth et al., 1990).

Both mgo1-1 and mgo2 show similar defects in sepal formation (Fig. 2C-E). Floral meristem partitioning into whorls is not perturbed by these mutations (Fig. 2C,D). Cells are allocated to the outer whorl where sepal initiation will take place, but the further partitioning into sepals is altered. Primordium size and position are irregular and abnormal in both mutants. Some sepal primordia appear to be partially fused (see arrows Fig. 2D). The number of sepal primordia (arrow Fig. 2E) is variable. In 12 developing floral buds of mgo2, we observed from 3 to 7 sepal primordia. We frequently observed that some sectors of the first whorl failed to initiate sepal primordia even at a stage where the petal and stamen primordia were already visible. This suggests that in some sectors the partitioning into sepals does not take place at all. Together these observations suggest that the MGO genes are required for the proper initiation of floral organs.

In order to obtain more insight into the role of the MGO genes, we analysed the genetic interactions between mgo1-1, mgo2 and other mutants which have reduced organ numbers, namely fas1, fas2 and stm.

The double mutant mgo1-1 mgo2 has the same characteristics as the mutants. However both the effects on organ numbers (Table 1 and 2) and stem fasciation (see below) are greatly enhanced in the double mutant.

The STM gene is necessary for shoot meristem formation during embryogenesis and meristem maintenance during postembryonic development (Barton and Poethig, 1993; Clark et al., 1996; Endrizzi et al., 1996). In stm-2, a weak allele, meristematic-like cells are formed during embryogenesis, and are entirely incorporated into the first primordia (Clark et al., 1996; Endrizzi et al., 1996 and Table 3). The mutant is therefore able to initiate primordia, but incapable of maintaining a SAM. stm-5 has a stronger phenotype as most seedlings do not form any meristematic cells at all, and only a few seedlings form primordia (Endrizzi et al., 1996 and Table 3).

When mgo1-1 is introduced into a stm-2 background, most of the double mutant embryos do not form any leaf primordia or clusters of meristematic-like cells (Table 3). About half of the double mutants do not develop a visible shoot after 19 days (Table 3 and Fig. 3G,K). Very few double mutants develop flowers, but when present they show reduced floral organ numbers when compared to mgo1-1 or stm-2 mutants (Table 2). Together, these data indicate that mgo1 reduces the capacity of the plant to initiate organs of stm-2 throughout development.

In a 19-day-old F₂ population of a cross between the heterozygotes of the mgo1-1 and stm-5 mutations no new phenotype was visible and we found 219 wild type, 69 mgo1-1 seedlings and 93 stm-5-like seedlings (segregation 9/3/4, χ²=0.24). This indicates that stm-5 is epistatic to mgo1. No mgo2 stm double mutants were found (see Materials and Methods).

fasciata 1 and 2 mutants described by Leyser and Furner, (1992) have an enlarged shoot apical meristem, disturbed leaf production (Fig. 3E,F) and fewer floral organs. After 19 days, SAMs of fas (1 or 2) mgo (1 or 2) double mutants have formed a round dome (Fig. 3H-J,L-M). No leaves are produced by mgo2 fas2 double mutants (Fig. 3J,M), whereas finger-like structures develop on the meristem flank of mgo1-1 fas2 and mgo1-1 fas1 double mutants (Fig. 3H,I,L). In addition, mgo1- 1 fas1 double mutants have a short root and narrow cotyledons (Fig. 3H). No mgo2 fas1 double mutants were found (see Material and Methods).

In conclusion, leaf production is extremely inhibited when fas1 or fas2 are combined with mgo. This suggest that MGO genes are absolutely required for leaf production in a fas background.

Histological analyses of wild-type and mutant meristems The morphology of the apical meristem was analysed on longitudinal sections of wild-type and mutant apices. In the wild- type mature embryo, the SAM is made up of an average of 25-35 small dense cells organised into 3–4 layers (Fig. 4A); the two outermost cell layers (L1 and L2) are well defined. At 8 days, the wild-type SAM is flat and 4–5 cells deep, and by 12 days, it has formed a convex structure (Fig. 4D,G). Cells in the two outer layers divide preferentially in the anticlinal plane; periclinal divisions occur at the periphery of the second layer. Cells in the underlying L3 layer do not show any preferential division plane.

The mgo1-1 mutation affects meristem organisation in the embryo: cell shape is irregular and the layered structure is less evident (Fig. 4B). SAMs of mgo2 embryos are normal (Fig. 4C) and the first difference from the wild type is visible after 6 days (data not shown). In both mgo mutants, the organisations of 8- and 12-day-old SAMs are perturbed in the same way (Fig. 4E,F,H,I). Cell layers are not clearly visible and cell division planes are not well oriented. The meristems enlarge progressively between 8 and 12 days until they reach a diameter which is 2 to 3 times wider than in the wild type. On the flank of mgo2 meristems, cells become bigger and vacuoles form (Fig. 4I).

In conclusion, mgo1-1 and mgo2 affect meristem structure and lead to disorganisation of the cell division plane alignment and to progressive overgrowth. mgo1-1 affects the SAM during embryogenesis, whereas mgo2 only has a role in postembryonic SAM development.

Since the organisation of mgo1-1 and mgo2 meristems is perturbed, we looked at the identity of cells within the meristem by analysing the expression patterns of meristem markers. In this way, we tested two members of the KNOTTED gene family, STM and KNAT2.

The activity of the KNAT2 promoter controlling the uidA gene was studied in 10-day-old plants. In the wild type, GUS expression is restricted to the SAM, the base of developing young leaves and the vascular strands near the apex (Fig. 5A). Within the SAM, the GUS signal is only visible in the lower part corresponding to the L3 layer and is more concentrated to the periphery. The expression pattern of pKNAT2::uidA in mutant seedlings is similar to the wild-type pattern (Fig. 5B,C).

The STM expression pattern was studied by in situ hybridisation on sections of 10-day-old seedlings. As described by Long et al. (1996), STM is expressed in the wild type in all shoot apical and floral meristems but it is absent from the incipient primordia (Fig. 6A,D). No major changes in the overall pattern of STM expression were found in mutant seedlings; STM expression is visible in the central part of the SAM and absent from domains at the periphery of the meristem which could be the developing primordia (Fig. 6B,C,E,F). However, the domain expressing STM is more extensive than in the wild-type, confirming that the SAM in mgo is larger. When the mutant apex is composed of several domes, STM is expressed in several distinct domains (Fig. 6G). In some apices (Fig. 6H) STM is expressed in two distinct domains although the apex still forms a single morphological unit.

These results indicate that, although the SAM in the mutants is perturbed, the general organisation of the SAM into functional domains is maintained.

After bolting, wild-type plants form an inflorescence stem which bears cauline leaves and lateral inflorescence shoots. Flowers are produced along the stem in a regular spiral phylotaxy (Fig. 1A). The inflorescence stems of mgo mutants are fasciated: they are wider and flatter than in the wild-type (Fig. 1B,C,E,F). Some of the young emerging stems have a wild-type morphology, while in others, fasciation occurs immediately after bolting. The stem enlarges progressively and then bifurcates, restoring a wild-type structure which will enlarge again (Fig. 1F). This process, which is repeated several times, leads to a bushy phenotype for older plants. Note that this process is different from wild-type branching where an axillary meristem subtended by a cauline leaf develops and forms an accessory branch (Fig. 1D). In both mgo mutants, flower position along the stem is irregular, in that more than one flower stalk may start from the same stem point.

Using scanning electron microscopy, we observed that mgo apices do not appear as one continuous structure but as juxtapositions of several meristematic domes (Fig. 7A). Some of these domes are clearly separated, which could represent early stages of bifurcation. In mgo1-1 mgo2 double mutants, the fasciation of the meristem is enhanced (Fig. 7B). The domes do not separate when they become morphologically distinct, as this would not lead to the observed stem fasciation but to an extreme increase in the number of stems, which is not the phenotype of the mutants.

It has been proposed that in clavata1 and 3 meristem overgrowth and fasciation is due to the increased size of the CZ (Clark et al., 1995, 1996). The fasciated meristem of clavata, however, is different from that observed in mgo, as it forms a continuous structure with no individual subunits visible (Fig. 7C). In addition, organ production is increased in clv, whereas it is reduced in mgo.

In order to study the interaction between mgo and clv, mgo1- 1 clv3-1 and mgo2 clv3-1 double mutants were constructed. Up to 2 weeks after germination, double mutants are almost identical to single mgo mutants. Later on, double mutants show additive defects. Organ numbers are intermediary between those of single mutants (Table 1 and 2). In double mutant flowers, organ width and number are variable just as in the mgo mutants and, a fifth inner carpel whorl develops as in clv3-1 flowers (data not shown). Stems are extremely fasciated and can be up to 1 cm wide. SAMs are fasciated and, as in mgo, composed of juxtaposed domes (Fig. 7D,E). These results indicate that there is no epistatic relationship between MGO1, MGO2 and CLV3 and that the MGO and CLV3 genes are still active in a clv3 and mgo mutant background respectively.

In vitro, mgo1 and mgo2 meristems produce just two leaves before bolting, in contrast to the five leaves which are initiated by wild-type plants. mgo1 and mgo2 flowers show an overall reduction in the number of organs, with variable numbers in all four whorls. These results indicate that MGO1 and MGO2 are necessary for determining the number of organs produced during vegetative and floral development. This phenotype may be due to defects in the SAM at two levels: the CZ and the PZ. Abnormal functioning of the central parts of the meristem could result in decreased or increased cell numbers flowing to the PZ, which in turn would modify the number of cells available for primordium initiation and thus influence organ arrangements. However, as mentioned earlier, the mutants may also be perturbed in the process of primordium initiation at the meristem periphery.

Several observations indicate that the MGO genes are involved in the process of organ primordia initiation itself. First, the vegetative meristem produces fewer leaves, despite the fact that there are enough cells. In fact, the mgo meristem is larger than the wild-type meristem. Secondly, SEM observations of developing flowers clearly show that the initiation of sepals at the periphery of the meristem is perturbed. Cells are apparently allocated to the outer whorl, but then fail to be partitioned correctly. Floral organ number is abnormal in all four whorls showing that the MGO genes are necessary throughout flower development whenever organ initiation takes place. mgo mutants are therefore radically different from mutants such as clv, stm or wus, that are supposedly perturbed in the CZ, and where the number of organs is positively correlated with the number of cells produced by the meristem. It was proposed that in clavata, the enlarged meristem is due to an increase in size of the CZ (Clark et al., 1993, 1995, 1996). In mgo mutants, meristem overgrowth could be a secondary effect of the perturbed primordia formation; cells, instead of being ‘used’ by the primordia, remain in the meristem.

Genetic analysis also supports the hypothesis that MGO1 and MGO2 are involved in primordium production. Like mgo, the fas1 and fas2 mutants show variable organ numbers. Interestingly, double mutants between fasciata and mgoun develop vegetative meristems that are almost incapable of forming leaves. Since we do not know at this stage whether all the alleles are null alleles, it is impossible to say whether fas and mgo act in the same pathway, or in parallel pathways, which affect the same process. The results do show, however, that mgo genes become an absolute requirement for primordium initiation in the absence of fas, and vice versa. Analysis of stm mgo1 double mutants could be interpreted in a similar way. STM is thought to maintain undifferentiated source cells in the meristem and not to be required for primordium initiation per se (Long et al., 1996). It is this ability to initiate primordia which is severely reduced when mgo1 is introduced into a stm background.

Although the overall organisation of the meristem into distinct zones, as shown by pKNAT2::uidA and STM expression patterns, is apparently not perturbed, both mutations affect the fine structure of the vegetative SAM. The L1, L2 and L3 layers are less well-defined and division planes are abnormal. It could be that this is an indirect effect of the aberrant partitioning in the periphery, but we cannot rule out that MGO1 and MGO2 are also directly involved in setting up the architecture of the entire meristem. Indeed, some of the genetic evidence could also be interpreted in this way. For instance, if both fas and mgo affect the structure of the meristem itself, it is conceivable, that the accumulated effect of the mutations could lead to such a perturbation of meristem function, that primordium initiation would be completely abolished.

One of the most obvious characteristics of the mgo mutants is their fasciation. Meristem fasciation is not specific to the inflorescence meristem as it can also occur during vegetative development, when mutants are grown in SD (data not shown). As in clavata, this phenotype is affected by growth conditions, but usually the stems are 3-4 times as wide as in the wild type. Clear differences from clavata become apparent when the meristems are viewed in the electron microscope. As described earlier (Clark et al., 1993, 1995), clavata mutants show one central meristem which can be up to 1 cm wide. Fasciated apices in mgo, however, have a large dome, which consists of a variable number of juxtapositioned meristems. This is particularly clear in the double mutant mgo1-1 mgo2 shown in Fig. 8B. The observed fasciation of mgo can be explained in several ways: it could be the result of the continuous fragmentation of the existing SAM, or, the ectopic formation of new meristems. We have repeatedly observed mutant apices forming a unique dome expressing STM in two separate cell clusters. This observation suggests that fragmentation of the STM-expressing domain is an early event of meristem bifurcation. We never observed the small domains of STM expression that would be expected in the case of early steps of ectopic meristem formation. Although we cannot rule out either of these hypotheses, these observations suggest that fragmentation occurs within the meristem. Mutants homozygous for both mgo and clv3 still have an apex consisting of several juxtapositioned domes, each dome being significantly larger than in the mgo single mutants. This indicates that the full activity of MGO1 and MGO2 is not required to maintain the integrity of the large central meristem of the clv mutant.

The two complementation groups described here have comparable effects on reducing organ formation and altering meristem structure, while the mgo1-1 mgo2 double mutant has an enhanced phenotype. Both mutations have the same phenotypic effect in clv and fas backgrounds. These results suggest that these two genes have comparable roles during postembryonic plant development. MGO1 has an additional role during embryonic development. We propose that these genes are necessary for correct primordia development in all meristems. They may stimulate cells to leave the meristem and may also be necessary for the partitioning into primordia (Fig. 8).

We thank S. Clark and T. Laux for providing seeds, and K. Barton for the STM probe. We thank M. Caboche, R. Cowling, S. O’Neill and F. Nogué for critical reading of the manuscript. This work was supported by a grant from Le Ministère de l’Enseignement Superieur et de la Recherche to P. L. and an INRA postdoc fellowship to J. D.