Defective cell proliferation in the floral meristem of alloplasmic plants of Nicotiana tabacum leads to abnormal floral organ development and male sterility


  • Isabelle Farbos,

    Corresponding author
    1. Department of Plant Biology, Uppsala Genetic Center, Swedish University of Agricultural Sciences, Box 7080, S-75007 Uppsala, Sweden, and
      For correspondence (fax +46 18 673238; e-mail
    Search for more papers by this author
  • Armand Mouras,

    1. Unité de Biologie Cellulaire et Biotechnologies Végétales, IBVM, CR-INRA Bordeaux, BP81, 33883 Villenave d'Ornon Cedex, France
    Search for more papers by this author
  • Agnes Bereterbide,

    1. Unité de Biologie Cellulaire et Biotechnologies Végétales, IBVM, CR-INRA Bordeaux, BP81, 33883 Villenave d'Ornon Cedex, France
    Search for more papers by this author
  • Kristina Glimelius

    1. Department of Plant Biology, Uppsala Genetic Center, Swedish University of Agricultural Sciences, Box 7080, S-75007 Uppsala, Sweden, and
    Search for more papers by this author

For correspondence (fax +46 18 673238; e-mail


Flowers of an alloplasmic male-sterile tobacco line, comprised of the nuclear genome of Nicotiana tabacum and the cytoplasm of Nicotiana repanda, develop short, poorly-pigmented petals and abnormal sterile stamens that often are fused with the carpel wall. The development of flower organ primordia and establishment of boundaries between the different zones in the floral meristem were investigated by performing expression analysis of the tobacco orthologs of the organ identity genes GLO, AG and DEF. These studies support the conclusion that boundary formation was impaired between the organs produced in whorls 3 and 4 resulting in partial fusions between anthers and carpels. According to the investigations cell divisions and floral meristem size in the alloplasmic line were drastically reduced in comparison with the male-fertile tobacco line. The reduction in cell divisions leads to a discrepancy between cell number and cell determination at the stage when petal and stamen primordia should be initiated. At the same stage expression of the homeotic genes was delayed in comparison with the male-fertile line. However, the abnormal organ development was not due to a failure in the spatial expression of the organ identity genes. Instead the aberrant development in the floral organs of whorls 2, 3 and 4 appears to be caused by deficient floral meristem development at an earlier stage. Furthermore, defects in cell proliferation in the floral meristem of the alloplasmic male-sterile line correlates with presence of morphologically modified mitochondria. The putative causes of reduced cell number in the floral meristem and the consequences for floral development are discussed.


During the evolution of higher plants, the genomes of the nucleus, plastids and mitochondria have coevolved resulting in coordinated gene expression. The crucial importance of this coordination is obvious from and investigation of plant development in alloplasmic lines. Novel nuclear-cytoplasmic combinations in which the cytoplasm from one species has been combined with the nucleus from another species often result in aberrant development. The most obvious example of abnormality is in the development of floral organs, such as petals and stamens, which leads to inhibited pollen production (Bonnett et al., 1991; Hanson and Conde, 1985; Kaul, 1988; Vedel et al., 1994). The floral abnormalities in the alloplasmic lines have proven to be associated with the mitochondria (Aviv and Galun, 1980; Belliard et al., 1978; Gleba, 1978; Glimelius et al., 1981; Kofer et al., 1991). Floral morphology and male fertility can be restored to normal by introgressing nuclear genes (Bonnett et al., 1991; Burns and Gerstel, 1981). Clearly cooperative genetic interactions between nuclei and mitochondria are involved in the developmental processes of flowers, although the nature of these interactions is unknown.

Flower development has been studied extensively during the last decade. Detailed models regarding function and regulation of nuclear genes involved in floral initiation and organ formation have been proposed (Coen and Meyerowitz, 1991; Irish, 1999; Ma, 1998; Meyerowitz, 1997). Several nuclear genes have been shown to be involved in control of size, determinancy and identity of the floral meristem. A model explaining the genetic interactions was developed in which the floral organ identity genes were classified as A, B and C function genes (Coen and Meyerowitz, 1991; Meyerowitz et al., 1991; Weigel and Meyerowitz, 1994). Organ identity genes are transcribed from early stages in specific domains of the floral meristem, demarking the four whorls, and expression is subsequently maintained in floral organs as they develop. The domain of expression of AG, the C function genes, comprises whorls 3 and 4 (Drews et al., 1991). AP3 and PI, the B function genes, have overlapping expression patterns in whorls 2 and 3 (Jack et al., 1992). PI is expressed in whorls 2, 3 and, to a lesser extent, in whorl 4 (Goto and Meyerowitz, 1994). Later during development, AP3 and PI expression becomes restricted to whorls 2 and 3.

Several alloplasmic cultivars have been produced within the species Nicotiana tabacum. Depending on the species chosen as the cytoplasmic donor for production of each alloplasmic line, a diversity of abnormal floral morphologies has been obtained (Bonnett et al., 1991). In this study, we describe and compare floral structures, morphology and ontogeny in an alloplasmic, male-sterile line of Nicotiana tabacum containing cytoplasm from Nicotiana repanda, with the corresponding male-fertile Nicotiana tabacum line. The male-sterile line exhibits strongly modified floral organs with the most drastic modifications of the petals, anthers and carpels. An altered pattern of organ inititation is found after initiation of the sepal primordia leading to development of short unpigmented petals and abnormal sterile stamens, which can be fused to the gyneocium. Aberrant development seems to stretch from late petal development involving corolla length and limb expansion, through stamen development to early carpel initiation and formation, while the later stages of carpel development seem appropriate. These aberrations result in development of mosaic organs and to male sterility, but female fertility. Since the floral organ genes are hypothesized to control organ departition and development, they were likely candidates for the disturbances. This was further supported by the similarities of the morphological aberrations obtained from mutations in the floral organ genes and the modifications established in the alloplasmic male-sterile lines.

To elucidate if the nuclear-mitochondrial regulation of the aberrant floral development in the male-sterile tobacco line modified the expression patterns of the tobacco homeotic genes, the expressions of NTGLO, NTDEF, NAG1 and NFL homologous to the Arabidopsis thaliana genes PISTILLATA, APETALA 3, AGAMOUS and LEAFY (Davies et al., 1996; Hansen et al., 1993; Kelly et al., 1995; Kempin et al., 1993; Weigel et al., 1992) were examined. This examination was based on in situ hybridisation in developing flower buds of the male-fertile and alloplasmic male-sterile lines and used to complement the anatomical and morphological characterisation of floral commitment in these plants.


Floral morphology

A morphological comparison of mature flowers of male-fertile Nicotiana tabacum and of the male-sterile alloplasmic line reveals that the calyx of the male-sterile line is quite similar but generally wider than the male-fertile line. The corolla differs with respect to the petals. They are shorter, and their limbs are less pigmented and less horizontally expanded in the alloplasmic male-sterile line (Figure 1a,b). The flowers differ considerably with respect to the male organs (Figure 1c,d). Normal male-fertile flowers develop five stamens with pollen containing anther bags on each filament (Figure 1c). In contrast, alloplasmic male-sterile flowers contain 4 or 5 shortened stamen filaments with shrivelled anthers usually capped with stigmatoid structures. Furthermore, the arrangement of the stamens in regular whorls is completely abolished. Stamens can be fused on both sides of the carpel, thus disrupting normal fusion of the style and leading to a bifurcated style and stigma (Figure 1d).

Figure 1.

Phenotype of flowers from the alloplasmic male-sterile and male-fertile tobacco plant.

(a) A mature male-fertile flower of N. tabacum. (b) A mature flower from the alloplasmic male-sterile tobacco line. Notice the larger calyx and a less pigmented corolla with less horizontal expansion of the limb compared with the male-fertile tobacco flower. Bar in (a) and (b), 10 mm. (c) Reproductive organs (stamens and carpel) from the male-fertile tobacco flower of 6-mm length. (d) Reproductive organs from the alloplasmic male-sterile tobacco of 6-mm length (sepals and petals removed). Specific features are aberrant stamens and fusions between stamens and the carpel. (e) Reproductive organs from the alloplasmic male-sterile tobacco flower of 16-mm length. Abnormal anthers are fused at the level of the style. External ovules are visible on the fused stamens. (f) Longitudinal section of the reproductive organs shown in (e). The arrow indicates a modified anther bag with ovules replacing the anther tissues. (g) Transverse section of the reproductive organs shown in (e). Stamens show several appendages and the gynoecium has additional ovary loculi. Bar in c–g is 1 mm. Abbreviations: c, carpel; st, stamen.

Figure 1(e) shows male and female organs of an alloplasmic mature flower in which two abnormal stamens are free standing and one stamen is fused to the female organ. The filament is fused along the carpel wall and the anther is fused at the junction between the style and the ovary. Longitudinal and transverse sections (Figure 1f,g) show the internal structure of the reproductive organs, including the two types of stamen. One type has free filaments and abnormally folded anther bags covered with stigmatic cells, resulting in several appendages that contain sporogenous cells. These can produce a few abnormal pollen grains. The other type of stamen is fused with the carpel wall and contains both male and female specific reproductive structures, that is, pollen and ovules. The ovule develops in the place of an anther sac (Figure 1f). The fusion of anther filaments with the carpel wall results in a structural modification of the ovary loculus and formation of a pseudo loculus with or without ovules (Figure 1g). The pseudo loculus is often empty at the base. When ovules are present, they are generally located at the top of the pseudo loculus in the region of the style. The style and stigma are also modified by fusion with the stamen. The style is usually curved, accounting for the observation that its length varies considerably among flowers of the same plant. The stigma is generally larger than the stigma in male-fertile tobacco and is often bifurcated (Figure 1d). Nevertheless, the alloplasmic male-sterile plants are female fertile. A comparison of the seed set obtained in male-fertile tobacco (74.8 ± 40.4 mg capsule−1) and the alloplasmic male-sterile line pollinated with pollen from the male-fertile tobacco (102.6 ± 55.4 mg capsule−1) revealed that the amount of seeds varied considerably, but did not differ in a significant way between the lines.

Organogenesis of flower organs

To determine the origin of abnormal development in internal whorls, early floral meristems were examined. The nomenclature for N. tabacum flower development, as described by Koltunow et al. (1990), was used to describe the developmental stages investigated during flower organogenesis. In the male-fertile line, the earliest stage described is stage −7 corresponding to a bud length of about 0.75 mm. In addition, we have assigned stage −8 to the floral meristem of buds of 0.6 mm and stage −9 to the floral meristem which corresponds to a bud of 0.3 mm.

At stage −9 the sepals develop along the edges of the floral meristem, establishing the first whorl. In male-fertile N. tabacum, the floral meristem consists of the three concentric layers L1, L2 and L3. The sepals have grown to a size large enough to cover the floral meristem. On the flank of the meristem the cells are smaller than in the center and divide actively in a controlled and regulated fashion (Figure 2a,b). These cells are recruited to form the next primordia, which are the petals of whorl 2 followed by the stamen primordia of whorl 3. Cells in the meristem center have prominent nuclei and are slightly vacuolated. In buds of the same size (0.3 mm) in the alloplasmic line, the floral meristem has a general organisation of L1, L2 and L3 similar to the male-fertile line, but the flanking zone of the meristem consists of very few dividing cells (Figure 2c,d). Compared with the male-fertile line the cells of the dome are larger, more vacuolated and have larger nuclei. An estimate of the number of cell divisions was made by counting all cells in mitosis in serial sections of 3 µm thickness of the complete meristems in the two lines. From this calculation we could show that at stage −9 in the alloplasmic male-sterile line, fewer than 10 divisions were found compared with more than 200 divisions in the male-fertile line at the same stage.

Figure 2.

Figure 2.

    Flower development at stages −9 and −8.

    (a) Floral meristem at stage −9 of male-fertile tobacco. (b) A magnification of the flanking zone of the floral meristem shown in (a). (c) Floral meristem at stage −9 of the alloplasmic male-sterile line. Notice the absence of cell divisions in the meristem. (d) A magnification of the flanking zone of the alloplasmic male-sterile floral meristem shown in (c). (e) Floral meristem at stage −8 of the male-fertile tobacco line showing initiation of petal and stamen primordia. (f) Floral meristem at stage −8 of the alloplasmic male-sterile line showing initiation of stamen primordia, which have fused with the carpel primordium (arrow). (g) Confocal laser scanning of the male-fertile tobacco floral meristem at stage −7 where primordia of sepals and petals were removed. (h) Confocal laser scanning of the alloplasmic male sterile floral meristem at stage −7 where primordia of sepals and petals were removed. The bud length is indicated by l (length) in mm in (a), (b), (d) and (e). Bar in (a) and (c) is 75 µm. Bar in (e), (f), (g) and (h) is 100 µm.The following abbreviations are used for the flower structures: c, carpel; p, petals; s, sepals; st, stamens. L1 to L3 denotes the different meristematic layers.

    In male-fertile tobacco at stage −8, when the length of the floral bud is approximately 0.6 mm, the sepals are well developed and almost closed around the internal bud structures. Petal and stamen primordia emerge at approximately the same time from the flank of the meristem to form whorls 2 and 3 (Figure 2e). The primordia contain actively dividing meristematic cells. At the time, when the carpel will develop, the central zone of the meristem is still flat with clear L1, L2 and L3 layers. In the alloplasmic male-sterile line, whorl 3 is initiated in floral buds of approximately 0.9 mm compared with 0.6 mm for the male-fertile line, at a stage when the petals have already developed (Figure 2f). Thus, initiation of stamens is delayed and disorganized compared with the male-fertile line. The lack of clear organization of cells into discrete units of stamen primordia is confirmed by confocal analysis of developing buds from the male-fertile and the alloplasmic male-sterile line (Figure 2g,h, respectively).

    To complement anatomical and morphological data SEM and histological analyses were performed. At stage −6 (1.6 mm bud length) the male-fertile buds have five stamens with the characteristic bilobed shape and adaxial indentation forming a distinct longitudinal cleft along the front of each anther (Figure 3a). The two carpels are formed as two distinct horseshoe-shaped fused primordia with a central placental column (Figure 3a,b). The development of alloplasmic male-sterile flowers differs significantly from that of the male-fertile line by the time the third whorl organ primordia appears. When stamen primordia surround the carpel in the alloplasmic line, at the corresponding stage −6, the bud is 2 mm. The longer bud length confirms that stamen primordia are delayed in their development relative to that of the petals (Figure 3b,e). They are not properly placed in the third whorl and can be fused with the carpel primordia (Figure 3d). Stamen primordia that are not fused grow and become dorsiventrally flattened (Figure 3d,e). The stamen primordia appear to accumulate cells without clear cell organisation and differentiation (Figure 3f), rather than developing into the typical stamen primordia formed in the male-fertile line, in which anther lobe differentiation has started (Figure 3c). The fused stamens have a ‘horse ear shape’(Figure 3d). Gynoecial growth proceeds with the characteristic rapid vertical growth even though normal loculus formation is disturbed by the interference of stamen primordia.

    Figure 3.

    Figure 3.

      Flower development at stage −6.

      (a) Scanning electron micrograph of reproductive organ primordia at stage −6 of the male-fertile tobacco flower. (b) Longitudinal section of the male-fertile flower at stage −6. (c) Anther of the male-fertile flower at stage −6 shown in (b). (d) Scanning electron micrograph of reproductive organ primordia at stage −6 of the alloplasmic male-sterile line. (e) Longitudinal section of the alloplasmic male-sterile flower at stage −6. (f) Anther of the alloplasmic male-sterile flower at stage −6 shown in (e). The following abbreviations are used for the flower structures: c, carpel; p, petals; s, sepals; st, stamens. The bud length is indicated by l (length) in mm in (a), (b), (d) and (e). Bar in (a), (b), (d) is 200 µm. Bar in (e) is 100 µm.

      Analysis of floral meristematic cells from the L1 and L2 layers in buds corresponding to stage −9 of the alloplasmic and male-fertile lines was performed by electron microscopy. In male-fertile plants (Figure 4a) the cytoplasm contained numerous ribosomes and few vacuoles. The endoplasmic reticulum was well developed and regularly distributed throughout the cell as a dense network of dilated tubules with a fine granular content. Numerous small spherical mitochondria containing large dilated cristae were observed. In the alloplasmic male-sterile line the endoplasmic reticulum tubules were more expanded than in the male-fertile plants. The most dramatic difference from the male-fertile plants concerned the mitochondria. Most of the mitochondria were twice as large in diameter compared with the male-fertile line (Figure 4b). When analyzed in detail the surrounding membranes showed the characteristic triple lamellar structure of mitochondria, but the intermembrane space was reduced compared with the fertile plants. The mitochondrial matrix was clear (not electron dense) and the cristae were scarce or swollen, some being clearly disorganized. Thus, the mitochondria of the alloplasmic male-sterile line seemed to be altered in their morphology.

      Figure 4.

      Transmission electron micro graphs of floral meristematic cells.

      (a) Detail of a meristematic cell of the male-fertile tobacco plant. (b) Detail of a meristematic cell of the alloplasmic male-sterile plant. The following abbreviations are used: cw, cell wall; er, endoplasmic reticulum; n, nucleus; m, mitochondria; pl, plastid. Bar in a and b is 1 µm.

      Microsporogenesis: from stage −4 to +4

      To compare microsporogenesis in the two lines, we investigated the characteristic events at stage −4 (bud length 4 mm) and stage +4 (bud length 16 mm). At stage −4, the anthers in the male-fertile line have established a typical organisation with four separate pollen sacs (Figure 5a). Each pollen sac contains normally organized epidermis, endothecium, a middle layer and a tapetum. The cell division activity of the tapetum results in a characteristic multinucleate cell layer, formation of sporogenous cells with callose deposition and initiation of meiosis (Figure 5b). The microsporogenesis process leads to the production of microspores at stage + 4 in a pollen sac in which the tapetum is degenerating (Figure 5c). Subsequently, the microspores mature and anthers dehisce via the stomium to release pollen grains.

      Figure 5.

      Figure 5.

        Anther development in male-fertile and alloplasmic male-sterile plants.

        (a) Transverse section of an anther from a male-fertile flower of 4-mm length (stage −4). (b) Pollen sac of an anther shown in (a). (c) Pollen sac from an anther from a male-fertile flower of 18-mm length (stage 4). (d) Transversal section of an anther from an alloplasmic male-sterile flower of 13-mm length. (e) Pollen sac of an anther from the male-sterile flower shown in (d). (f) Pollen sac of an anther from an alloplasmic male-sterile flower of 21-mm length. e, epidermis; en, endothecium; pg, pollen grains; s, stomium; sp, sporogenous cells; t, tapetum; v, vascular tissue. Bar in (a) and (d) is 170 µm. Bar in (b), (c), (e), (f) is 80 µm.

        In the alloplasmic male-sterile line, the microsporogenesis process is severely impaired. The stage corresponding to stage −4, when the corolla emerges from the calyx (Figure 5d), represents buds of 13 mm. In the male-fertile flowers, this developmental stage corresponds to flower buds of 4 mm. The typical ‘butterfly’ shape of the male-fertile anthers is severely modified generating the teratologic morphology previously described. The number of pollen sacs varies, resulting in abnormal structures of reduced size. When a pollen sac develops somewhat normally, the characteristic tissues of the male-fertile anther can be observed. However, they are not as well organised. For example, the endothecium contains more than 2 cell layers, while the tapetum remains mononuclear, even though callose has been deposited around the sporogenous cells (Figure 5e). In addition, the cell number in each pollen sac is drastically reduced. Anthers in 21-mm buds contain free microspores corresponding to stage +4, while the tapetum is multinucleate, which is characteristic of stage −2. The microspores are tightly packed in the small loculi and surrounded with callose. These small loculi are enclosed within a thick endothecium containing twice as many cell layers as the male-fertile line. The endothecial cells are smaller, less differentiated and do not contain plastids. The middle cell layer is absent and the epidermis is not fully differentiated as expected at this stage. In such anthers, most of the pollen sacs collapse (Figure 5f). Furthermore, the anthers are unable to dehisce, since the stomium is not properly placed.

        Temporal and spatial expression patterns of NFL and the organ identity genes NTGLO, NTDEF, NAG1 in developing male-fertile flowers and alloplasmic male-sterile flowers.

        In situ hybridizations were carried out in order to compare temporal and spatial expression patterns of NFL, NTGLO, NTDEF and NAG1 in the alloplasmic male-sterile line and in the male-fertile tobacco lines. The expression patterns were determined on sections of buds at an early stage of development, using Dig-labelled antisense probes. The expression of NFL is temporally and spatially identical in the alloplasmic male-sterile and the fertile lines (data not shown). The NFL transcripts were detected at a very early stage (stage −8) in floral primordia in the male-fertile line. At this stage, NFL transcripts were sequestered to the presumptive petal regions.

        The expression patterns of NTGLO mRNA are shown in Figure 6 (a–d). In the male-fertile tobacco line, the earliest stage at which NTGLO mRNA expression was detected was in the floral meristem when the sepal primordia differentiate. Before stage −9 (bud length 0.3 mm), no NTGLO transcripts were detected (Figure 6a). Shortly after the sepal primordia are formed, NTGLO mRNA can be observed in cells that will give rise to petals and stamens (Figure 6a). In the alloplasmic male-sterile line, NTGLO expression is usually first established in buds that have reached the size of 0.7 mm when the reproductive organs are still at stage −9 (Figure 6c). At this stage the sepals are more expanded than in the male-fertile line. During stage −7, which for the male-fertile line corresponds to a bud length of 1.1 mm, petal and stamen primordia are prominent and the sepals have closed. At this stage, NTGLO transcripts have accumulated in cells that will develop into petals and stamens (Figure 6b). In the alloplasmic male-sterile line, at a stage corresponding to stage −7, the third whorl primordia are delayed in their development and look like petalodes rather than stamens. NTGLO transcripts are detected at high levels in the second and third whorls (Figure 6d). The petals and stamens increase rapidly in size between stages −7 and −2. NTGLO transcripts continue to be present at high levels in all cells of the stamens and throughout the petals (Figure 6d). In older alloplasmic buds, during development of the abnormal stamens, the NTGLO transcripts are restricted to the upper part of the stamen, where some archesporial cells can be found (data not shown).

        Figure 6.

        Figure 6.

          In situ analysis of organ identity gene expression in developing flowers of the male-fertile and alloplasmic male-sterile plants.

          (a–d) Longitudinal sections of flowers from male-fertile (a,b) and alloplasmic male-sterile (c,d) plants, probed with antisense NtGLO. Bar is 200 µm. (e,f) Longitudinal sections of flowers from male-fertile (e) and alloplasmic male-sterile (f) plants, probed with antisense NtDEF. Bar in (e) and (f) is 400 µm. (g,h) Longitudinal sections of flowers from male-fertile (g) and alloplasmic male-sterile (h) plants, probed with antisense NAG1. Bar in (g) and (h) is 200 µm. The bud length is indicated in mm in (a)–(h).

          In early developmental stages the expression pattern of NTDEF is more widespread than NTGLO both in the male-fertile line and the alloplasmic male-sterile line (data not shown). At later stages, when carpel primordia emerge, NTDEF expression becomes enhanced in the developing whorls 2, 3 and 4 (Figure 6e). The expression pattern of NTDEF in the alloplasmic line is comparable with the male-fertile tobacco line (Figure 6f).

          In male-fertile tobacco plants NAG1 transcripts are detected before stage −9 of flower development (bud length of 0.2 mm). During stage −9 (bud length 0.3 mm), NAG1 transcripts are detected in the center of the meristem in those cells that will later develop into stamens and carpels (Figure 6g). Later during development, NAG1 transcripts accumulate in whorls 3 and 4 with the strongest expression found in the region forming the ovaries (data not shown). In the alloplasmic male-sterile line, NAG1 expression is established in the meristem in 0.3 mm buds. No NAG1 transcripts were detected in younger meristems (Figure 6h). Later in development, NAG1 transcripts accumulate in the abnormal stamens and carpels, especially in the ovary cells. The spatial expression of NAG1 in the alloplasmic line is comparable with the male-fertile line. For each gene investigated control experiments were performed, with Dig-labelled sense probes, which confirmed that the patterns visualized in Figure 6 were due to true expressions from the genes and not unspecific signals (data not shown).


          Several alloplasmic lines of N. tabacum develop flowers with abnormal organs (Bonnett et al., 1991; Kofer et al., 1991; Zubko et al., 1996). Similar modifications are found in other alloplasmic lines including Brassica (Delourme and Budar, 1999; Forsberg et al., 1994; Liu et al., 1995) revealing the importance of nuclear-mitochondrial interaction for floral organ development. In the alloplasmic male-sterile line containing N. repanda cytoplasm the morphological aberrations appear after floral transition and are mainly restricted to the floral organs. An altered pattern of floral organ initiation was found in the floral meristem after initiation of the sepal primordia leading to the development of short unpigmented petals and abnormal sterile stamens, which can be fused to the gyneocium. As early as stage −9, when the stamen primordia were formed from the floral meristem, we observed a difference in the development of the central region. The region remained as an undeveloped meristematic structure, from which the abnormal stamens developed at an irregular position in whorls 3 and 4. The cells of stamen primordia had merged with the cells of the gyneocium and the boundary between the whorls was inappropriately defined. These morphogenic abnormalities might, thus, reflect a failure to establish boundaries between whorls 3 and 4 and/or to control the distribution of cell proliferation activity in the central region for a discrete period of time during flower development.

          Since the diameter of the entire floral meristem is only 0.3 mm it was difficult to identify the precise whorl boundaries in the alloplasmic line accurately. However, examinations of temporal and spatial RNA expression patterns of the floral organ identity genes revealed that they constituted excellent markers in specifying organ primordia and the whorl boundaries in the floral meristem. Studies of in situ hybridization experiments showed that the pattern of spatial expression of NFL and the tobacco organ identity genes, NTDEF, NTGLO and NAG1 was comparable in normal male-fertile and alloplasmic male-sterile lines. NTGLO is expressed in whorls 2 and 3, NTDEF in whorls 2, 3 and 4 and NAG1 in the cells of the inner whorls. These results are consistent with the investigations performed by Kelly et al. (1995) and Davies et al. (1996) in tobacco. The localization and distribution of gene expression are well defined and pattern the floral meristem in an appropriate way. Consequently the floral abnormalities in the alloplasmic line do not result from faulty expression of the organ identity genes NTGLO, NTDEF and NAG1. The abnormal development found in the alloplasmic male-sterile line appears instead to be the consequence of a developmental disorder in the floral meristem before its subdivision into domains in which the activation of organ identity genes occurs.

          During development of the floral meristem, cells are continuously produced, displaced to the periphery and subsequently incorporated into organ primordia. In order to maintain such an ordered, continuously developing structure, the behaviour of individual cells (that is, cell division, expansion and differentiation) is strongly coordinated. This is illustrated by the regular distribution and pattern of cells in the floral meristem of male-fertile tobacco shown by confocal microscopy. In contrast the ordered distribution and specific pattern of cell division and division plane alignment is abolished in the floral meristem of the alloplasmic male-sterile line. In the meristems different sets of genes determine the identity of the organs produced and regulate the basic mechanisms that are active in shoot apical meristems and floral meristems (Laufs et al., 1998a; Laux and Mayer, 1998; McSteen and Hake, 1998). A balance between cell production and cell differentiation can be, at least partially, achieved by controlling the rate of division of meristematic cells. If such controls are perturbed, the meristem may not maintain itself. This is illustrated by mutants affected in the function of the shoot apical meristem or floral meristem in Arabidopsis such as SHOOT MERISTEMLESS (STM) (Barton and Poethig, 1993; Endrizzi et al., 1996) WUSCHEL (WUS) (Laux et al., 1996; Mayer et al., 1998), PINHEAD (McConnell and Barton, 1995) and CLAVATA (CLV) (Clark et al., 1993, 1995). While none of these genes has yet been shown to affect the proliferation of cells specifically, in particular histogenic layers, some of them, such as CLAVATA1 and WUSCHEL, have layer-specific expression patterns and are possibly involved in coordinating proliferation of the different cell layers. An interesting observation and hypothesis brought forward by Jenik and Irish (2000) is that there is more plasticity in the L2 and L3 layers than in L1. Layer invasions are more frequent in plants in which a clone of cells has some growth disadvantage (Stewart et al., 1974) or in plants that are chimeras of different species (Marcotrigiano and Bernatzky, 1995). The fact that the number of dividing cells in the flanking zone of the meristem was improperly regulated in the alloplasmic line would lead to a reduced number of cells upon which the ground plan genes can act. Our conclusion is that a discrepancy between timing and control of cell proliferation in the central region of the meristem does not affect the expression of the organ identity genes, but does affect the coordination and rate of division in the different floral organs.

          The timing and rate of cell division in different parts of a plant are crucial to normal growth, development and maintenance, and are regulated by several factors and genes. The frequency of cell division varies with the type of cell. The cell cycle is timed by rhythmic fluctuations in the activity of different protein kinases (CDKs) which together with distinct cyclins seem to regulate the subsequent progression through check-points preceding cell cycle borders (Sherr and Roberts, 1999). One critical check-point determinant is thought to be an appropriate ratio of cellular biomass to cellular volume. For a cell to divide, the content of biomolecules, such as ribosomal and cytoskeletal proteins, must have reached a certain concentration otherwise the cell chooses to halt division in order to complete the synthesis of biomolecules. Interestingly TEM analysis revealed that the mitochondria in alloplasmic floral meristem cells were modified in their morphology and presumably in their function. These modifications correlate with the investigations of Bergman et al. (2000) on ATP and ADP levels in young floral buds. A direct measurement of steady-state ATP and ADP levels in stage −3 of non-green tissues of floral buds showed that alloplasmic cells had an ATP + ADP pool 55% smaller than the male-fertile tobacco cells. Moreover the alloplasmic line also had a significantly lower ATP to ADP ratio, probably reflecting a less active energy metabolism. Thus, a disorganized and slowly dividing meristematic flanking zone resulting in fewer cells, as shown in this study, might be due to a growth disadvantage caused by the low levels of cellular ATP due to mitochondrial dysfunction.

          A consequence of this growth disadvantage in the alloplasmic line is that the development of organs takes a longer time than in the male-fertile line. Since expression of the organ identity genes is coordinated with the development of organs, the expression of these genes is extended over a longer period of time in the alloplasmic line. In the alloplasmic line, the slow development of stamens, in comparison with the development of adjacent floral organs, implies that the cascade of genes expressed during flower organ formation could be out of synchrony with other floral genes. For instance the coordination between cadastral genes, such as SUPERMAN in Arabidopsis (Bowman et al., 1992) and the floral organ identity genes might be affected. SUPERMAN is thought to link the regulation of class B gene expression with the maintenance of whorl boundaries (Meyerowitz et al., 1995). Consequently, absence of coordinated expression could perturb establishment of proper boundaries between whorls 3 and 4 in the floral meristem and result in partial or complete fusion between organs derived from these two whorls.

          Experimental procedures

          Plant material

          The experimental material consists of male-fertile Nicotiana tabacum cv SC 58 denoted as the male-fertile line. The alloplasmic male sterile-line has been produced by sexual hybridization between Nicotiana tabacum and Nicotiana repanda followed by consecutive back crossings with N. tabacum as pollinator resulting in an introgression of the Nicotiana tabacum nucleus into the cytoplasm of Nicotiana repanda (Burk, 1967). The alloplasmic male-sterile line has been propagated as a cultivar for at least 20 generations by utilising the N. tabacum Sc58 cultivar as pollinator. Thus, the alloplasmic male-sterile line, which constitutes the research material, exhibits stable phenotypic flower modifications.

          Scanning electron microscopy and confocal laser scanning microscopy

          Young buds and early floral meristems were dissected and fixed overnight in 3% glutaraldehyde in 0.025 m sodium phosphate (pH 7.0) at 4°C. They were rinsed in buffer, dehydrated in a graded ethanol series and critical point dried in liquid carbon dioxide. Individual buds were mounted on stubs, and coated with gold, and viewed in a scanning electron microscope, SEM. Confocal laser scanning microscopy was performed as described by Laufs et al. (1998b).


          Inflorescences from both male-fertile and alloplasmic male-sterile plants were collected and fixed for 1 h in 2% paraformaldehyde, 0.25% glutaraldehyde in 0.1 m phosphate buffer, pH 7.0 at 4°C. They were transferred to 4% paraformaldehyde, 0.25% glutaraldehyde in 0.1 m phosphate buffer, pH 7.0 and incubated in a vacuum for 2 h. Samples were rinsed in buffer and dehydrated in ethanol series. Infiltration was carried out at room temperature for 1 h each with 25%, 50%, 75% and 100%, and 100% in Technovit's resin. Tissues were left overnight at 4°C and embedded in Technovit resin. Three-micron sections were cut with glass knives and stained in a solution of toluidine blue (0.5% w/v).

          In situ hybridisation

          Probes were labeled using Digoxigenin labeling mix (Boehringer, Mannheim, Germany) according to the manufacturer's protocol. NTGLO, NTDEF and NAG1 contain a putative DNA-binding region termed the MADS box. To avoid the possibility of cross-hybridisation with other genes, the MADS box sequences were removed from the plasmids used to make the probes. Southern blot hybridization of tobacco DNA with NTGLO and NTDEF was performed and a single band was detected.

          Flower buds at early stages of development were dissected and fixed in 3.7% formaldehyde, 5% acetic acid and 50% ethanol. Fixed tissue was dehydrated with ethanol, cleared with histoclear and embedded in paraffin (Paraplast Plus, Sigma-Aldrich, Steinheim, Germany). Embedded tissue was sliced into serial 8 µm sections and attached to microscope slides that were coated with APES (Aminopropyltriethoxysilane; Sigma, Sigma-Aldrich, Steinheim, Germany). Paraplast was removed by immersion in histoclear. Sections were rehydrated, incubated at 37°C for 30 min in a solution containing 1 µg ml-1 Proteinase K (Sigma) 100 mm Tris (pH 7.5), 50 mm EDTA followed by two washes in H2O for 5 min each. After dehydration by an ethanol series, slides were air dried before application of the hybridisation solution containing 50 ng slide−1 of DIG labelled RNA probe, 50% formamide, 300 mm NaCl, 10 mm Tris (pH 7.5), 1 mm EDTA, 5% Dextran sulfate, 1% Blocking reagent, 150 µg ml-1 tRNA). After incubation in a humid box at 37°C overnight, slides were washed in 0.2 X SSPE for 1 h at 55°C. After incubation with 5 µg ml-1 RNAse A for 30 min at 37°C, slides were washed twice in 0.2 X SSPE for 30 min at 55°C, then 5 min in 2 X SSPE at 37°C and 5 min in PBS (1,3 m NaCl, 30 mm NaH2PO4, 70 mm Na2HPO4), pH 7.4 at room temperature. Slides were incubated in 1% blocking reagent (Boehringer) in maleic buffer (100 mm maleic acid, 150 mm NaCl, pH 7.5) for 45 min, then in BXT (1% BSA, 100 mm Tris, 150 mm NaCl and 0.3% triton-X-100, pH 7.5) for 45 min. Anti-digoxigenin-alkaline-phosphatase-coupled antibody (Boehringer) was diluted 1 : 600 in BXT solution. A volume of 150 µl was applied to each slide, the sections covered with a cover slip and incubated for 2 h. Slides were washed twice, 25 min each time with BXT solution. Fresh Color-Substrate solution (NBT and BCIP in 100 mm Tris, 50 mm MgCl2 and 100 mm NaCl, pH 9.5) was applied to each slide and incubated in the dark at room temperature. When the color reaction was complete, slides were washed twice with TE for 5 min and mounted with Cristal/Mount (Biomeda, Hayward, CA, USA).


          We thank I. Eriksson for skilful technical assistance and M. Ingouff, M. Landgren and P. Bergman for helpful discussions. This work was supported by grants from the Swedish Natural Science Research Council. A. Bereterbide was supported by a French MENSR-DGRT fellowship. The homeotic probes were kindly provided by the laboratory of H. Schwarz-Sommer.