This paper describes the cytochemistry and ultrastructure of the developing tapetum in Primula obconica, a plant with a heteromorphic, sporophytic self-incompatibility system. The tapetum is of the secretory type and cytochemical tests have shown that when it breaks down proteinaceous (esterase) and lipidie components are deposited on the developing pollen grains. Acid phosphatase, a marker of gametophytic enzyme activity, is confined to the cytoplasm and intine of the developing pollen. Ultrastructural studies show that prior to its dissolution the tapetum undergoes a number of changes. In the early stages of development the tapetum is rich in ribosomes and rough endoplasmic reticulum, but following the breakdown of the tapetal cell wall the main components of the cytoplasm are densely staining spherical bodies surrounded by ribosomes and orbicular bodies, which appear to be confined to the cell periphery. As the cells break down, rod-like fibrils can be seen amongst the degenerate organelles and within the bacular cavities of the pollen. On dehiscence the pollen has a lipidie coating in addition to the fibrillar material in the pollen wall and the remnants of the tapetum can be seen adhering to the fibrous layer of the anther wall. Thus the mature, binucleate pollen of P. obconica is demonstrated to carry wall materials of sporophytic origin.
The development of haploid pollen grains occurs within the anther, where the sporogenous cells are surrounded by diploid sporophytic tissue. This sporophytic tissue is called the tapetum and its structure and function have been investigated in a limited number of plant species. Echlin (1971) has categorized tapeta into 2 types, amoeboid and secretory. In the first type the tapetal cells become amoeboid and extend into the anther loculus; they surround the developing spores and retain a functional ultrastructure until just prior to anthesis. In the second type the tapetal cells are secretory and disintegrate long before the pollen is mature. These cells release a variety of components into the anther loculus, which subsequently become attached to the pollen grains (Echlin & Godwin, 1968; Dickinson, 1973; Dickinson & Lewis, 1973). Vithanage & Knox (1976) have demonstrated, using enzymic cytochemical techniques, that materials are transferred from the tapetum to the pollen exine in Brassica olerácea, and were able to show that in this species the intine and exine proteins are of gametophytic and sporophytic origin, respectively. The importance of this material of tapetal origin in the self-incompatibility reaction was first demonstrated by Dickinson & Lewis (1973) in Raphanus sativus. They were able to show that in this species, which has a homomorphic, sporophytic self-incompatibility system, material of tapetal origin can bring about the rejection reaction in the stigmatic papillae in incompatible combinations. Thus at least one aspect of the recognition reaction of self and non-self pollen is mediated by material derived from the tapetum.
Primula obconica has a heteromorphic self-incompatibility system, in which the dimorphism of pollen grain size, anther position, style length and form of the stigmatic papillae as well as the self-incompatibility system are controlled by a sporophytically expressed super-gene (Dowrick, 1956). This study was designed to investigate whether materials of tapetal origin are transferred to the developing pollen grain of P. obconica, and forms part of a study of the mechanism of self-incompatibility in heteromorphic flowering plants. The transfer of material from the tapetum to the pollen grain in no way confirms that these substances have a role to play in the self-incompatibility reaction, as they may be important in other aspects of pollination or in inter-specific recognition as in Populus (Knox, Willing & Ashford, 1972). On the other hand the absence of tapetally derived materials from the walls of the pollen would indicate that there are major differences in the mechanisms of homomorphic and heteromorphic self-incompatibility systems.
MATERIALS AND METHODS
Preparation of frozen sections
Anther tissue was placed in a small hole in a 15% gelatin medium containing 2% glycerol. Once the gelatin had reset, a cube containing the tissue was cut out and rapidly attached to a chuck cooled to — 12 °C. The block was frozen immediately in Arcton 12 (ICI) and transferred to the cryostat. Sections 4 Jim thick were cut and removed from the knife onto clean glass slides. At least 10 slides were prepared for each anther stage considered: 4 to test for esterase activity, 4 for acid phosphatase activity and 2 for lipids.
Location of esterase activity
A modification of the method of Pearse (1961) was used. The staining medium was prepared as follows: 10 mg alpha napthyl acetate was dissolved in 0·25 ml acetone, and 20 ml 0·1 M-phosphate buffer (pH 7·4) was added. The mixture was shaken and filtered to remove any precipitate. The filtrate was pipetted directly onto the sections to be stained and left at room temperature for 15 min. Slides were then washed in running water for 3 min, excess water was allowed to drain off, and the sections were then mounted in dilute glycerine. Control sections were incubated in a medium from which the substrate, alpha napthyl acetate, had been omitted. Regions of esterase activity stain black.
Location of acid phosphatase activity
Sections were stained in drops of substrate solution prepared in the following manner: 10 ml distilled water, 10 ml 0·2 M-Tris-acetate buffer (pH 5) and 10 ml 1·25% sodium β- glycerophosphate were mixed together. Twenty ml 0·2% lead nitrate solution was added slowly with constant stirring. Control sections were incubated in the above medium from which the substrate, sodium β-glycerophosphate, had been omitted. After 15 min the slides were rinsed in 0·05 M-Tris-acetate buffer (pH 5) and placed in 1% ammonium sulphide solution for 2 min. The slides were then rinsed in running water for 5 min and the sections subsequently mounted in dilute glycerine. Regions of acid phosphatase activity stain brown.
Location of lipids
Slides were placed in 50% ethanol for a few minutes and then the sections were stained with drops of Sudan IV (saturated solution in 70% ethanol) for 10 min (Jensen, 1962). Sections were differentiated in 50% ethanol for 1 min before mounting in dilute glycerine. Fats, oils and waxes stain orange.
Preparation of anther material for transmission electron microscopy
Excised anthers were cut into approximately 1 mm lengths and fixed in 3% glutaraldehyde in 0·05 M-phosphate buffer (pH 7·2) for 4 h at room temperature. The material was then rinsed in 3 changes of buffer over 4 h and post-fixed in 1% osmium tetroxide (aqueous) for 3 h at 4 °C. Following washing in distilled water the material was dehydrated through an acetone series (15, 30, 60, 90%) with 2 changes of absolute acetone before embedding in Epon 812.
‘Thick’ sections (0·5g μm) were cut with a glass knife from blocks at each stage of development studied and then stained with methylene blue (2% aqueous) for observation with the light microscope. Gold sections of suitable regions were cut on a Cambridge Huxley microtome, using a diamond knife, and mounted on copper grids. Sections were post-stained with 2% aqueous uranyl acetate for 15 min and then with Reynolds lead citrate (Weakley, 1972) for 5 min. Grids were examined using an AEI 801 transmission electron microscope at 60 kV.
Anatomical changes in the anther during development
Stages in anther development were studied using 0·5 μm sections observed with the light microscope. Similar patterns of development were seen in both pin and thrum anthers. Five stages of development could be distinguished readily and are represented diagramatically in Fig. 1. The anther eventually differentiates into a complex structure that is made up of a number of different cell types. In this study, sections of anther that included the epidermis were used wherever possible as this provided a consistently recognizable region to which changes in other layers could be referred.
The first developmental stage that can be recognized (stage 1) represents the time at which pollen mother cells can be distinguished from the surrounding tapetum. At this stage the tapetal cells have prominent nuclei and the cytoplasm stains densely with methylene blue. The endothecium, the tissue external to the tapetum and internal to the epidermis, is composed of 3 or 4 layers of cells. These are smaller than the tapetal cells and are partially vacuolated. At stage 2 the structure of the tapetum and endothecium is similar to stage 1, but at stage 2 the pollen mother cells have completed meoisis and appear as tetrads enveloped in a layer of callose. By stage 3 the microspores, which still contain a single nucleus, have been released from the callose and lie free in the anther loculus. Stage 3 is also characterized by the differentiation of the endothecium. The outer 1 or 2 layers of vocuolated cells adjacent to the epidermis have much enlarged since stage 2 and the inner endothecial layers now contain numerous starch-like inclusions. At stage 4 the microspore contains a single prominent nucleus and has become highly vacuolated. The thickness of the tapetum is reduced due to the breakdown of the cellular structure and it is no longer possible at this stage to distinguish between the tapetal layers and the inner layers of the endothecium. The outer endothecial layers have developed considerably thickened anticlinal walls and now comprise the fibrous layer of the anther wall. At dehiscence (stage 5) the epidermis and fibrous layer can still be distinguished but the inner endothecial layers and the tapetum have condensed further and no longer have a distinct morphology. The pollen grains, which are no longer vacuolated, now contain 2 nuclei, one of which stains more deeply than the other and is probably the nucleus of the generative cell. The baculate exine of the pollen wall can also be distinguished clearly at this stage as it stains deep blue with methylene blue, whereas the intine remains unstained. The pollen grains tend to adhere to each other at this stage.
The results of investigations into the activity of esterase and acid phosphatase and the presence of lipids in anthers at a number of developmental stages are presented diagrammatically in Fig. 2. No staining for esterase, acid phosphatase or lipid was detected at the pollen mother cell or early tetrad stages (Fig. 2A, B) in any of the tissues of the anther, but staining for esterase and lipids was noted in the tapetal tissue later in the tetrad stage and following the release of the spores from the callose wall (Fig. 2C, D, E). NO acid phosphatase activity was present until after the dissolution of the tapetum when staining was found to be restricted to the grain cytoplasm only (Fig. 2F, G). On the breakdown of the tapetum, esterase activity became distributed throughout the anther loculus, and at dehiscence staining was concentrated around the pollen grains, and apparently within them. This apparent cytoplasmic activity may, in fact, only be dense wall staining. Lipid globules were also released into the anther loculus following tapetal breakdown and became attached to the pollen grain walls (Fig. 2F, G).
It can be seen from these studies that esterase and acid phosphatase activity are associated with mature pollen grains. Esterase is present in the tapetum of young anthers and on breakdown is found in the loculus and on the surface of the grains. Lipid distribution also begins in the tapetum and, after its dissolution, lipid is found in the loculus and attached to the grains. This material of sporophytic origin is present on pollen grains at dehiscence. Acid phosphatase is associated only with the grains, indicating gametophytic origin.
Ultrastructure changes in the tapetum during anther development
Similar developmental sequences were observed in the anthers of pin and thrum plants. At stage 1, the tapetum may be distinguished easily from the sporogenous tissue and is more deeply stained. The plasma membrane is continuous and the cell wall present, the latter being similar in appearance to that surrounding the sporogenous cells. The nuclei are prominent and chromosomal material is evident. The cytoplasm of the tapetal cells contains many mitochondria, ribosomes, long strands of endoplasmic reticulum and dictyosomes, and appears to be highly active (Figs. 3-4).
By stage 2, the sporogenous cells are enclosed within a layer of callose. The tapetal cell wall is still distinct at the loculus face, but walls between tapetal cells appear to be breaking down (Fig. 5) and in some regions no plasma membrane can be found. Changes have taken place within the tapetal cells since stage 1. The most obvious feature is that the ribosomes and endoplasmic reticulum are associated, producing very electron-dense areas (Fig. 6). Many vesicles are present and nuclei and mitochondria may still be distinguished in some cells.
Radical changes take place in the tapetum between stages 2 and 3. The tapetal walls break down and a number of cytoplasmic organelles are released into the anther loculus. Fig. 7 shows that the remaining tapetal cells are bound only by a membrane. Darkly staining bodies may be seen inside the tapetal cells and at the inside face of the membrane, and bodies with dark centres and paler edges are attached to the outer surface. The tapetal cells also contain many other less densely staining spherical bodies, each surrounded by structures of similar dimensions to ribosomes.
The mitochondria are rounded and no endoplasmic reticulum is evident. There are numerous dictyosomes in a secretory state. Large vesicles containing membranes are found and the general appearance of the tapetal cytoplasm suggests a state of autolysis. In the later part of stage 3 the tapetal cells undergo further disorganization. In Fig. 8 the spherical vesicles surrounded by ribosome-like bodies predominate and the rest of the cytoplasm consists of degenerate membranous structures and fibro-granular material. Darkly staining bodies line the inner surface of the plasma membrane and orbicules, resembling those described by Echlin & Godwin (1968) and Heslop-Harrison (1971), are present on the extracellular surface. These orbicules are frequently found all around the tapetal cell membranes, not only on the face adjacent to the anther loculus. The form and location of these bodies suggests that they are coated with sporopollenin. Some cell membranes are broken and cytoplasmic components can be seen in the loculus. At this stage the pollen exine is well established (Fig. 8) and the pollen contains many polysomes, dictyosomes and vesicles.
Further tapetal disintegration takes place and at stage 4 the layer of cells adjacent to the tapetum also appears degenerate with the fibrillar thickening of the walls staining darkly (Fig. 9). The tapetal remains consist of fragments of membrane, orbicules and many short rod-like fibrils. These fibrils may also be discerned in the exine cavities (Fig. 10). The intine layer is now present in the pollen wall and contains many vesicles. Fibrillar and orbicular material condense further at the periphery of the anther loculus with more electron-dense material (Fig. 11). At dehiscence the anther loculus is lined with a very thin layer of cellular debris of which darkly staining lipid bodies are the most noticeable feature (Fig. 12). Material of a similar appearance adheres to the pollen-grain wall, which suggests that it has a similar origin. Thus, material from the breakdown of tapetal and anther wall cells, of sporophytic origin, is deposited on the haploid pollen grains.
Very few cytochemical or ultrastructural studies have been done on tapetal development, but from the information available there appears to be considerable variation between families in the form and distribution of organelles during the sequence of growth and development. Sporogenous and tapetal tissue may be distinguished within the anther of P. obconica during the early stages of meiosis, the tapetal cells appearing much denser than the pollen mother cells. Similar distinctions have also been found in Olea europaea (Pacini & Juniper, 1979), where the tapetal cells were found to have a higher ribosome content than the sporogenous tissue. In P. obconica, however, the differences appear to be mostly in the density of the ground cytoplasm and variations in organelle distribution. Breakdown of the tapetal cell walls begins during the tetrad period and it also occurs during this period in Helleborus foetidus (Echlin & Godwin, 1968), but not until the first haploid mitosis in Olea europaea (Pacini & Juniper, 1979). By the time the pollen grains are released from the callose into the anther loculus, the tapetal cells in P. obconica are bounded only by a membrane and appear to be undergoing autolysis. However, in other genera, including Olea (Pacini & Juniper, 1979), Avena sativa (Gunning & Steer, 1975) and Raphanus (Dickinson & Lewis, 1973), the tapetal cells are rectangular at this stage and still have ‘normal’ cytoplasm but with the addition of new plastids and vacuoles. The cellular degeneration of the tapetum occurs at an earlier period in the development of P. obconica pollen than in the other species cited. In a number of angiosperm species, conspicuous elaioplasts are present in the tapetal cells (e.g. Olea and Raphanus) and these may contain lipid droplets, which are released into the loculus following rupture of the tapetal protoplasts as in Raphanus (Dickinson & Lewis, 1973). No structures resembling these elaioplasts were observed in Primula, nor were there any large vacuoles containing fibro-granular material, which are prominent features in the tapetal cells of Raphanus. The ultrastructure of the tapetum of Primula, following release of the pollen grains from the tetrad, reveals the presence of some degenerating mitochondria and other organelles, but small spheroids, surrounded by bodies resembling ribosomes, and fibrous material are the predominant constituents of the cytoplasm. Deeply staining bodies are located at the cell peripheries, attached to the inner side of the membrane and orbicular structures, and are found on the outer surfaces on all faces of the tapetal cells. The cytochemical methods used show that esterase activity and lipidie material are present in the tapetum at the end of the tetrad stage. The electron micrographs of stage 2 indicate much synthetic activity and deeply staining bodies are seen lining the tapetal cell membrane. The location of these bodies suggests that they may be pro-orbicular bodies, which have been shown to be lipidie in Oxalis by Carniel (see Heslop-Harrison, 1971). Echlin & Godwin (1968) describe the formation of pro-orbicular bodies in Helleborus and show them surrounded by a zone of ribosomes. It is possible that the spheroids, surrounded by what appear to be ribosomes, may represent the same organelles. However, as the tapetal cells at this stage have high esterase activity, the spheroids may be postulated as likely sites for this. Few other organelles are present at stage 3, so unless the enzyme is free in the cytoplasm, it is likely to be located either in the spheroids or the darkly staining droplets.
After the tapetal membranes disintegrate lipid and esterase are detected in the loculus andon the pollen grains. Both cytochemical and ultrastructural evidence suggest transfer of material from the tapetum to the pollen grain walls ; rod-like fibrils being found in tapetal debris and in the cavities of the exine, and esterase activity and lipid changing location from tapetum to pollen after tapetal dissolution. Thus sporophytic tapetal material is transferred to the pollen-grain wall so that on dehiscence the haploid pollen bears a coating of protein (esterase) and lipid of diploid origin. This pattern of transfer of esterase activity and appearance of acid phosphatase activity only in the pollen is almost identical to that described by Vithanage & Knox (1976) in Brassica olerácea. However, the tapetal material in P. obconica is very different in appearance from that of Raphanus and presumably Brassica (since these 2 genera are very closely related), and its function appears to be quite different. Callose deposition in the stigmatic papillae is not characteristic of the self-incompatibility response in Primula (Stevens, 1980) as it is in many species of the Cruciferae and Compositae. Thus it would appear that the tapetally derived substances in the pollen-grain wall of this heteromorphic species have a different role to play from that of species with homomorphic, sporophytic self-incompatibility systems.
A postgraduate studentship from the SRC to the senior author is gratefully acknowledged. We would like to acknowledge the members of the Westfield College Electron Microscope Unit for their help at various stages of this investigation.