J. Cell Sci. 25, 125-138 (1977) 125

Printed in Great Britain

DIFFERENTIATION OF THE TAPETUM

IN A VENA

I. THE CELL SURFACE

M. W. STEERBotany Department, The Queen's University, Belfast, N. Ireland

SUMMARY

The development of the tapetal cell surface and associated structures in Avena has beenfollowed from cell formation to senescence. Plasmodesmata initially connect the tapetal cells toeach other, the pollen mother cells, and the inner loculus wall cells. These connexions aresubsequently severed, those to the sporogenous cells being broken first at the pollen mothercell surface during callose wall formation. Loss of cellulose from the tapetal walls was followedusing the decline in the ability of the wall to bind the fluorescent brightener, CalcofluorWhite M2R New. Subplasma-membrane microtubules persist after loss of the cellulose wall.The tapetal plasma membrane facing the meiocytes then develops a series of depressions, orcups, over its surface, which are later the site of pro-orbicule formation. Sporopollenin is laiddown over the pro-orbicules, to form orbicules, and over other tapetal cell surfaces. No morpho-logical evidence was found for the intracytoplasmic formation of pro-orbicules or polymerizedsporopollenin precursors.

These observations on Avena are compared with those on other plants. The changes in thecell wall and associated structures, plasmodesmata and microtubules, are considered in detail,while the general significance of cell wall loss to the water relations of the tissue are assessed.Proposals that pro-orbicule formation results from non-specific accumulation of lipid at a freecell surface are rejected, instead this formation is considered to be related to the presence of aspecially modified plasmamembrane surface.

INTRODUCTION

The tapetum is one of the most complex plant cell types, showing a remarkablespecialization of its intracellular and extracellular components. The developmentalsequence of this specialized tissue can be studied in material that is readily accessible,the anther, and whose growth is easily monitored by reference to the stage of micro-spore formation. This makes the tapetum a particularly attractive model system forthe study of various aspects of plant cell activity. Of the 2 types of tapetum foundamong flowering plants, the secretory type is the more suitable for this type of studyas the cells are static and do not migrate into the anther loculus, as do cells of theperiplasmodial type (Echlin, 1971a).

General descriptions of Avena tapetal and microspore structure are given in Steer(1974) and Gunning & Steer (1975). This paper will be concerned with establishingthe development of the cell surface and associated structures, a further paper will bepresented on the internal development of the cell.

Echlin & Godwin (1968) first applied improved tissue preparation methods for9 C E L 25

126 M. W. Steer

electron microscopy to the study of tapetal development in a dicotyledon, Helleborus,reporting on many features of the cell surface and laying the groundwork for furtherobservations and discussions. This was followed by observations from other dicotyle-dons (Marquardt, Barth & von Rahden, 1968; Risueno, Gimenez-Martin, Lopez-Saez & Garcia, 1969; Hoefert, 1971; Horner & Lersten, 1971), monocotyledons(Heslop-Harrison & Dickinson, 1969; De Vries & Ie, 1970; Christensen, Horner &Lersten, 1972), and agymnosperm (Dickinson and Bell, 1972, 1976a). It is clear fromthese observations that the development of the secretory tapetum cell surface isbroadly similar in all these groups. The differences in development between thesemajor taxonomic groupings that have been reported so far are mainly concerned withthe relative timing of tapetal events compared with the stage of microspore develop-ment, and are difficult to assess in view of the limited number of genera examined.

In Avena it has been possible to follow formation of the cellulose wall and associatedplasmodesmata and their loss and replacement by a layer of orbicules that develop inassociation with a specialized plasma-membrane surface. The observations reportedhere are in very close agreement with those already made on another monocotyledon,Sorghum (Christensen et al. 1972), and appear similar to those from Zea mays (Skvarla& Larson, 1966) and wheat (De Vries & Ie, 1970).

All micrographs are arranged so that the pollen mother cell (microspore) lies to the left ofeach figure and the tapetal cell to the right.

Fig. 1. Longitudinal section, 1 /tm thick, of glycol methacrylate-embedded antherstained with 001 % Calcofluor White M2R New for 20 min and photographedusing a Zeiss fluorescence microscope. The layer of binucleate tapetal cells lies betweenthe pollen mother cells, at prophase I, and the loculus wall cells. Note intensefluorescence of the callose walls and of the cellulose walls in the outer layers of the an-ther. By comparison there is a complete lack of fluorescence from the tapetal cell walls.X950.

Fig. 2. As Fig. 1. Longitudinal section tangential to the anther sac showing layer ofbinucleate tapetal cells. Fluorescence is absent from the tapetal cell walls, while it ispresent from the cellulose around the loculus wall cells, x 1100.

Fig. 3. Early pollen grain development, otherwise as Fig. 1. Exine formation is almostcomplete and the 3 layers can. be resolved due to autoftuorescence of sporopollenin inthe ncxine and tcctum. Only the layer of orbicules on. the adjacent tapetal wall showsevidence of fluorescence, while other walls show a strong fluorescence, x 950.

Fig. 4. The tapetum-pollen mother cell wall (czo) at the time of tapetal cell formation.The tapetal nucleus (n) had just been reconstituted at the end of mitosis and the cell wasin cytokinesis at the time of fixation. A plasmodesma (pd) interconnects the 2 cell types,x 9500.

Fig. 5. Two tapetal cells (top and right) abut a pollen mother cell which is at the pre-callose wall stage. The cell wall has a lightly stained fibrillar appearance and is traversedby a plasmodesma (pd) which runs out of the plane of section at the top left. Micro-tubules (nit) occur just below the tapetal plasma membrane and in the adjacentcytoplasm. They are also present in the pollen mother cell cytoplasm, x 19000.

Avena tapetum. I. Cell surface

9-2

128 M. W. Steer

MATERIALS AND METHODS

Seeds of Avena sativa L. var Stormont Sceptre were grown as described previously (Steer,IO-75)- Anthers at different stages of development were excised and processed for either lightor electron microscopy. Sections i-/tm thick, from material embedded in glycol methacrylate(Feder & O'Brien, 1968) were routinely stained with toluidine blue or periodic acid-Schiff'sreaction for examination by transmitted light microscopy. Glycol methacrylate sections werealso stained with o-oi % Calcofluor White M2R New in distilled water (Hughes & McCully1975) and examined in a Zeiss fluorescence microscope. Thick sections (1 fim) cut from Epon-embedded anthers were routinely stained with toluidine blue for light microscopy, while thinsections from the same anthers were stained with uranyl acetate and lead citrate for electronmicroscopy.

OBSERVATIONS

The following observations are based on examination of Avena sativa anthers by

light and electron microscopy. Observations on other oat species have been made

(Gunning & Steer, 1975) and these do not appear significantly different from the

following account. The accompanying figures, representing only a small fraction of

the material available, have been selected to illustrate specific stages of tapetal cell

surface development and concentrate on the earlier stages, since later ones are already

widely reported in the literature. In describing events in the tapetum reference will be

made to the corresponding stage of microspore development, so that comparisons can

be made with tapetum development in other plants.

Fig. 6. Subplasma-membrane microtubules (mt) lie on either side of the cellulose wallseparating a tapetal cell (right) from a pollen mother cell which is laying down calloseat the opposite (innermost) face. There is a noticeable halo around each microtubulein the tapetal cell, suggesting the presence of a non-staining component, x 57000.

Fig. 7. Early stage of cell wall breakdown at the beginning of prophase I. The tapetalplasma membrane is slightly withdrawn from the wall which is still distinctly fibrouson the tapetal side of the middle lamella (ml). Note plasmodesmata (pd). x 15000.

Fig. 8. Callose wall formation has extended around the pollen mother cell to the walladjacent to the tapetum. This results in an increased thickness of the wall on the mothercell side of the middle lamella (ml). The outer leaflet of the tapetal plasma membraneis heavily stained, x 62300.

Fig. 9. Loss of cellulose fibrils from tapetal side of middle lamella and replacement bya system of granules and fibrils. The pollen mother cells in the loculus of this antherare at anaphase-telophase I. Microtubules (arrows) are present beneath the tapetalplasma membrane, x 28500.

Fig. 10. Interdigitating ends of tapetal cells showing plasmodesmatal connexionsbetween them (arrowed). The tapetal wall inside the middle lamella has almostdisappeared, while the pollen mother cell callose wall (ca) has continued to increase inthickness, x 20300.

Fig. 11. Several plasmodesmatal connexions (pd) from the tapetum persist aftercallose wall formation around the pollen mother cell. The lower one probably passesout of the plane of section, whilst the upper clearly traverses the middle lamella andterminates in a slightly swollen end at the callose wall. The plasmodesmata are filledwith a homogenous, stained, material and are lined by a tripartite membrane. Micro-tubules are present below the tapetal plasma membrane at this tangential wall and theadjacent radial wall (arrows), x 30000.

Avena tapetum. I. Cell surface 129

' • ' 1

:L> . * , . < 10:

130 M. W. Steer

Tapetal cells are formed when cell division takes place in the cylindrical layer ofcells immediately surrounding the pollen mother cells. The division is slightlyasymmetric, the cell plate being laid down tangentially to the anther cylinder, cuttingoff a slightly larger tapetal cell, adjacent to the pollen mother cells, from a smallerloculus wall cell. Subsequent growth of the tapetal cell increases its volume by afactor of 10, from about 650 to 6500 /mv3. Most of the increase in cell size is due toan increase in length of the cell (i.e. parallel to the anther long axis), with smallerincreases occurring in the lengths of the radial and tangential walls.

The newly formed tapetal cell is bounded by a thin cell wall (0-1-0-2/tm thick),that binds the fluorescent brightener Calcofluor White M2R New (Hughes & McCully,1975). It contains numerous fine, lightly stained fibrils, and appears to be a typicalprimary cellulose wall (Figs. 4, 5). Plasmodesmata connect adjacent tapetal cells toeach other and to the pollen mother cells and the loculus wall cells (Figs. 4, 5, 7, io, 11).Just beneath the plasma membrane lie a series of microtubules. These are usuallyoriented parallel to the cell's (and anther's) long axis on the tangential walls (Fig. 6),and either in this direction or radially on the radial walls (Figs. 12, 13). Later, justbefore mitosis occurs to give binucleate tapetal cells (Figs. 1, 2), aggregates of micro-tubules occur that are similar to those described as preprophase bands (Fig. 12).

With the extension of the callose walls, from the centre of the pollen mother cellmass to the outer wall, the tapetal cell walls start to break down. The tangential tapetalwall adjacent to the pollen mother cells is the first to be affected, followed by changesto the outer tangential wall and radial walls.

Loss of the wall against the pollen mother cells is accompanied by a pronounced

Fig. 12. Preprophase band microtubules in tapetal cell before mitosis. Other micro-tubules run parallel to them, and at right angles, along the radial wall separating 2adjacent tapetal cells. Note the tenuous nature of the intervening cell wall. The outermembrane of the nuclear envelope is marked by arrows, x 40000.Fig. 13. Radial wall between 2 tapetal cells at metaphase I in the mciocytes. Only themiddle lamella (ml) remains of the original cellulose wall. Microtubules (mt) run bothalongside the wall and out into the cytoplasm. The outer leaflets of the plasma mem-branes are darkly stained, x 80000.Fig. 14. Adjacent tapetal cells (left) have withdrawn from the cellulose wall of theinner loculus wall cell. Large lipid droplets (arrowed) have been deposited along theinner face of the loculus wall, especially at the corners of the tapetal cells. The middlelamella (»//) can be traced through the lipid and along the loculus wall, x 27000.Fig. 15. Tapetal cell plasma-membrane with a series of cups (arrows). The cell wallbetween this membrane and the middle lamella consists of a series of granules inter-spersed with fibrils. On the other side of the middle lamella is the meiocyte callose wallenclosing a cell at anaphase II meiosis. x 60000.Fig. 16. Pro-orbicule (po) formation on the tapetal surface. The lipid droplets appearin the plasma membrane cups without any indication of similar droplets in the cyto-plasm. Note profiles of the endoplasmic reticulum (er) containing a stained matrix;some of these (unlabelled arrows) approach the plasma membrane and are bifacial(lacking ribosomes on one face). Part of the granular tapetal cell wall can be seen here,elsewhere in this section are newly formed microspores at the end of meiosis II.x 48 000.

Avena tapetum. I. Cell surface igr

132 M.W. Steer

swelling and loss of staining properties, including the ability to bind Calcofluor(Figs. 1-3) and to stain with periodic acid-Schiff's reaction. By late meiosis I, thefine ordered fibrils of the primary wall are replaced with a loose meshwork of coarsefibres and granules which lies adjacent to the callose wall that surrounds the meiocytes(Fig. 9). Microtubules are still present beneath the tapetal plasma membrane, evenwhen all remnants of the original wall have disappeared (Figs. 9-11). The plasmodes-matal connexions to the pollen mother cell are broken by the formation of the inter-vening callose walls, leaving plasmodesmata projecting from the surface of thetapetal cell (Fig. 11). The lumen of these plasmodesmata is filled with a substancethat stains uniformly. Later they disappear completely, either due to absorption bythe cell or to becoming detached from this suface and lost in the loculus cavity.

At the inner loculus wall the tapetal plasma membrane becomes withdrawn,cutting off the plasmodesmata. Lipoidal material accumulates along this wall andfrequently forms large droplets appressed to the wall and running round the cornersand up the radial walls between adjacent tapetal cells (Fig. 14). The radial wallsbecome more tenuous and loose their fine fibrils as the plasma membrane on eachside shrinks away from the wall. Again plasmodesmatal connexions are broken,though some plasma membrane connexions have been found stretched betweenadjacent tapetal cells at maturity. The plasma membrane of the radial walls (Fig. 13)and the lipid lining of the loculus wall (Figs. 14, 20) become coated with a thin,dark-staining layer which may be a thin lamella of sporopollenin (Dickinson &Bell, 1972).

The most prominent feature of the tapetum at maturity is the carpet of orbiculesthat coat the cell surface of the inner tangential wall. Each orbicule consists of aglobular lipid core (about 150 nm diameter) surmounted by a decorated sporopollenincoat (about 200 nm thick). The orbicules develop on the cell surface following aseries of discrete changes in the plasma membrane.

Fig. 17. Dense plaques of sporopollenin forming around pro-orbicules on the tapetalcell surface (arrows). The wall between this cell and the microspore is distinctlybipartite, with the callose layer becoming more diffuse (see Fig. 15). x 51000.Fig. 18. Accretion of sporopollenin at distinct nodes over the pro-orbicule surfaceinitiates development of the orbicule spines. The adjacent cell wall layers are almostcompletely disintegrated leaving the orbicules in close proximity to the developingmicrospore exine. x 5500.Fig. 19. Development of the characteristic channelling in the layer of sporopolleninencrusting the pro-orbicules can be seen. Note the microbodies in the adjacent tapetalcytoplasm (arrows), x 36500.Fig. 20. Nodules of sporopollenin accumulating along the inner loculus wall (s). Theyappear to be attached to a thin, dark-staining layer (arrows), probably also of sporo-pollenin. x 39000.Fig. 21. Orbicule formation almost complete at the time of exine completion and pollengrain mitosis, x 39000.Fig. 22. Senescence of tapetal cell (right) results in the complete loss of cell contents(compare Fig. 21). The orbicules, which have lost their lipid cores, remain abutting theadjacent microspore exine (lower left). Sporopollenin deposits are clearly seen alongthe radial wall between adjacent tapetal cells (arrows), x 35000.

Avena tapetum. I. Cell surface

134 M. W. Steer

Following loss of the cell wall the inner facing tapetal cell surface is seen as arelatively smooth expanse of plasma membrane. During the early stages of meiosis IIthis smooth membrane develops a series of small depressions, or cups, 50 nm deepand 100 nm across, at intervals over the cell surface (Fig. 15). At the end of meiosis alipid droplet, or pro-orbicule, is formed in each pocket which is initially 50-75 nm indiameter (Fig. 16). These grow in size until they reach 150 nm across, the same size asthe lipid cores in the mature orbicules. The whole process is strictly synchronizedover the cell surface, all pro-orbicules appearing at the same time. The origin of thepro-orbicules is not clear. There are no comparable lipid droplets in the cytoplasm atthis time, nor are any of the cell organelles specifically associated with the cell surfaceat this stage more than at preceding or subsequent stages.

The completed pro-orbicule is coated with sporopollenin at the time of microsporeseparation from the tetrads and exine formation. First discrete plaques of sporo-pollenin are laid on the lipid surface (Fig. 17), which then enlarge by accretion ofmore material (Fig. 18), and coalesce to form a complete coating (Figs. 19, 21).Sporopollenin is also laid down in small quantities along the radial walls (Fig. 22)and in larger accumulations on the lipid lining of the inner loculus wall (Fig. 20).

At pollen mitosis the tapetal cell cytoplasm starts to senesce, so that by the time thegenerative cells are formed the tapetal cells are devoid of contents and are onlyoutlined by the orbicules on the tangential wall and other sporopollenin depositsalong the radial walls (Fig. 22).

DISCUSSION

This account of the Avena tapetum cell surface can be compared with those pre-viously published on other monocotyledons and with accounts of secretory tapetumdevelopment in various dicotyledons. The basic developmental sequence is broadlysimilar in all these accounts. However, there are variations in the timing of tapetal cellsurface changes when related to the stage of microspore development, for example inBeta loss of the inner tangential wall occurs at the tetrad stage (Hoefert, 1971), whileit is much earlier in Avena. Further, in Helleborus, pro-orbicule formation does notoccur until after exine formation has commenced (Echlin and Godwin, 1968), butin Avena it occurs at the end of meiosis. Detailed comparisons between this and theother accounts of tapetum development reveal a number of similarities and differences,both in the observations and their interpretation, which will be discussed further.

The early stages of tapetal wall development in Helleborus have been described byEchlin & Godwin (1968). They interpreted the increased thickness of the innertapetal tangential wall (abutting the pollen mother cell callose wall) as being due to thedeposition of cellulose. However, in Avena the cellulose component is very rapidly lostfrom this wall as judged by fluorescence microscopy and the increase in wall thicknessappears to be accompanied by a net loss of material visible in electron micrographs.The overall sequence of tapetal wall loss found in Avena has also been reported fromBeta (Hoefert, 1971), Sorghum (Christensen et al. 1972) and Pinus (Dickinson & Bell,1972). The formation of granules and coarse fibrils as an intermediate stage in the

Avena tapetum. I. Cell surface 135

breakdown of cellulose walls has also been seen in cotyledons of germinating seeds(Smith, 1974, and personal communication). The granules may be protein, as they donot stain for carbohydrates, in which case they could be extracellular enzymes.

Presumably the tapetal cells' physiology changes when the cellulose wall is lost.This removes the mechanical constraint to cell size and shape, and might be thought toupset the cells' osmoregulatory system. However, loss of the wall is accompanied byloss of the cell vacuoles, thus removing one solute reservoir that would normallycontribute to the cells' water potential. The surrounding tissues, although capable ofimposing mechanical constraint on the tapetal cells, are not in contact with its plasmamembrane. Hence the solute concentration of the loculus must balance the internalsolute concentration of the tapetum, presumably the loculus solute concentration isincreased by the accumulation of tapetal cell wall and callose wall breakdown pro-ducts. Movement of water from the apoplast of the surrounding anther tissue into theloculus may be controlled by the lipoidal coating of the inner loculus wall.

The subject of plasmodesmatal connexions between the various tissues of themicrosporangium has given rise to numerous comments in other publications. Insome of these confusion has been caused by failure to qualify statements about theiroccurrence with the corresponding stage of development. For example in a recentreview on the occurrence of plasmodesmata in reproductive structures of plants it isasserted that they are absent from the tapetum - male meiocyte cell wall (Carr, 1976),a similar statement can also be found in Robards (1975). Both these references pre-sumably refer to later stages of development, since plasmodesmata have been foundin this situation by Heslop-Harrison (1966), Horner & Lersten (1971), Christensenet al. (1972) and, more recently, by Dickinson & Bell (1976a). Their presence andloss in Avena have been reported here and it is of interest that the connexions arefirst broken at the pollen mother cell surface at callose formation, leaving the plasmamembrane-lined structure projecting from the tapetal cell surface.

Echlin & Godwin (1968) reported on the occurrence of microtubules beneath thetapetal plasma membrane and their persistence until the tetrad stage. Similarly inAvena they persist until long after the loss of the cellulose cell wall. These observationsraise questions about their function in the cell at this time. Traditionally subplasma-membrane microtubules have been associated with orientation of cellulose fibrils asthey are formed in the adjacent plant cell wall (for a review see Hepler & Palevitz,1974), however, cellulose production only occurs at the earliest stages of tapetal celldevelopment. Microtubules seen at later stages may only be relics of the earlieractivity, although it seems possible that they might contribute to the maintenance oftapetal cell shape, as in many animal cells (Tilney, 1968).

The origin of the lipid cores of the orbicules, first termed pro-orbicules by Heslop-Harrison & Dickinson (1969), has attracted a great deal of attention. Echlin & Godwin(1968) were the first to attempt an analysis of pro-orbicule formation. In their material,Helleborus, conspicuous lipid droplets are present in the tapetal cytoplasm and it wasconcluded that these are extruded through the plasma membrane on to the tapetalcell surface forming the pro-orbicules. However, these authors reported 2 observa-tions which cast doubt on this interpretation, they noted that the lipid bodies in the

136 M.W. Steer

cytoplasm were larger than the pro-orbicules and that they were present at all stagesof development, even after orbicule formation had ceased. A subsequent report byHeslop-Harrison & Dickinson (1969) supported this model for pro-orbicule formation,but in later work by Horner & Lersten (1971) and Dickinson & Bell (1972) no firmevidence was found for the formation of pro-orbicules by this method. Risueno et al.(1969) presented evidence for the formation of pro-orbicules in elements of theendoplasmic reticulum and their subsequent movement to the surface, however,their micrograph (fig. 4 in the above paper) illustrating this latter event seems to showparts of three different cells (see Fig. 10 of this paper for a similar view, and Fig. 13for a view of the radial wall which could be confused with an intracytoplasmic channel).In Avena no evidence was found for the formation of lipid cores in the tapetal cyto-plasm. The strict synchrony found in development of Avena pro-orbicules, alsofound in Pinns (Dickinson & Bell, 19766), contrasts with the continuous productionof these bodies reported for Helleborus (Echlin & Godwin, 1968) and Beta (Hoefert,1971).

The formation of pro-orbicules in cup-shaped depressions of the plasma membranewas first observed in Sorghum by Christensen et al. (1972). The presence of thesecups has probably been responsible for some of the confusion over the intracyto-plasmic origin of pro-orbicules (Echlin & Godwin, 1968) and they have been inter-preted as discharging dictyosome vesicles (Horner & Lersten, 1971). Cup formationappears to be a specialization of the plasma membrane at one particular tapetalsurface. In animal cells the formation of such plasma membrane configurations is oftenassociated with systems of microfilaments below the cell surface (Reaven & Axline,1973) but these cannot be positively identified in the micrographs from Avena. InSorghum a much clearer association between the cups and subsurface endoplasmicreticulum was found than in the present work. In both Sorghum and Avena a granularstaining material is present in the endoplasmic reticulum that could be involved inorbicule formation (Fig. 16). However, other studies indicate a quite different fatefor these intracisternal contents (Steer, 1974; Gunning & Steer, 1975 and unpublishedwork). Lipid synthesis is known to occur in the endoplasmic reticulum of plants(Moore, Lord, Kagawa & Beevers, 1973) as well as animals (Jungalwala & Dawson,1970) and the possibility that such lipid ultimately finds its way to the pro-orbiculesis not excluded by the present work. Once on the surface of the tapetal cell the lipidaccumulates as discrete spheres within the cups, rather than as a continuous layer.Christensen et al. (1972) suggested that this is due to the absence of mechanicalrestraint on lipid accumulation around the tapetum, contrasted with the presence of acallose wall around the microspore, which they believe restricts such accumulation toa thin continuous layer. However, in Avena a structurally similar primexine developsin the absence of a callose wall, which disappears after meiosis II.

Christensen et al. (1972) have considered the possible evolutionary origins for thetapetum's ability to form orbicules and favour the view that it represents a vestigialcapacity of a tissue that was once sporogenous. This may be true of the ability to laydown sporopollenin on external surfaces and could explain the general accumulationof sporopollenin around the tangential and radial walls of the tapetum. But it is

Avena tapetum. I. Cell surface 137

inadequate to account for either the specialization of one face of the plasma membrane,to form a series of cup-shaped depressions, or the specific accumulation of lipiddroplets within them. Surely pro-orbicule formation is an example of evolutionaryspecialization producing a specific product that forms the base for sporopollenindeposition and orbicule formation, not a vestigial capacity to produce a pollengrain wall. The problem lies in trying to elucidate the function of orbicules so that thenature of the evolutionary pressure can be understood (for a review see Echlin, 1971 a).Suggesting that the sporopollenin provides a non-wetting surface so that pollengrains may be more easily dispersed (Heslop-Harrison & Dickinson, 1969) ignoresthe fact that the radial walls are so covered without recourse to orbicule formation.The answer may lie in the particular physical properties of the three-dimensionalsurface generated by the orbicules.

The discovery that sporopollenin contains polymerized carotenoids (Brooks, 1971)initiated a search for the cytoplasmic source of these precursors. Early stages ofcarotene polymerization to form sporopollenin have been claimed to occur in theendoplasmic reticulum in Allium (Risueno et al. 1969) and in the endoplasmicreticulum and associated vesicles in Pinus (Dickinson & Bell, 19766). There is noevidence for such intracellular formation of sporopollenin in the tapetum of Avena,supporting the view expressed by Echlin (19716) that such polymerization is alwaysextracellular. In particular there are no deposits in the cytoplasmic organelles citedabove that resemble either carotenoid pigment globules or sporopollenin. Doubt hasalready been expressed in this section over the interpretation of the micrographs inRisueno et al. (1969). Dickinson & Bell (19766) provide only one example of theirintracytoplasmic vesicles containing precursors of sporopollenin with no evidence forits movement 'intact' into the loculus. Clearly electron microscopy alone cannotprovide an understanding of these dynamic events.

The expert technical assistance of Mr D. Kernoghan and Mr G. McCartney is gratefullyacknowledged. Part of this work was supported by a grant from the U.K. Science ResearchCouncil.

REFERENCES

BROOKS, J. (1971). Some chemical and geochemical studies on sporopollenin. In Sporopollenin(ed. J. Brooks, P. R. Grant, M. D. Muir, P. van Gijel & G. Shaw), pp. 305-351. Londonand New York: Academic Press.

CARR, D. J. (1976). Plasmodesmata in growth and development. In Intracellular Communicationin Plants: Studies on Plasmodesmata (ed. B. E. S. Gunning & A. W. Robards), pp. 243-289.New York, Heidelberg and Berlin: Springer-Verlag.

CHRISTENSEN, J. E., HORNER, H. T. & LERSTEN, N. R. (1972). Pollen wall and tapetal orbicularwall development in Sorghum bicolor (Gramineae). Am. J. Bot. 59, 43-58.

DEVRIES, A. PH. & IE, T. S. (1970). Electron-microscopy on anther tissue and pollen of malesterile and fertile wheat {Triticum aestivum L.) Euphytica 19, 103-120.

DICKINSON, H. G. & BELL, P. R. (1972). The role of the tapetum in the formation of sporo-pollenin-containing structures during microsporogenesis in Pinus banksiana. Planta 107,205-215.

DICKINSON, H. G. & BELL, P. R. (1976a). Development of the tapetum in Pinus banksianapreceding microsporogenesis. Ann. Bot. 40, 103—113.

138 M. W. Steer

DICKINSON, H. G. & BELL, P. R. (19766). The changes in the tapetum of Pimis banksianaaccompanying formation and maturation of the pollen. Ann. Bot. 40, 1101-1109.

ECHLIN, P. (1971a). The role of the tapetum in microsporogenesis. of angiosperms. In PollenDevelopment and Physiology (ed. J. Heslop-Harrison), pp. 41-62. London: Butterworth.

ECHLIN, P. (19716). Production of sporopollenin by the tapetum. In Sporopollenin (ed.J. Brooks, P. R. Grant, M. D. Muir, P. van Gijzel & G. Shaw), pp. 220-248. London andNew York: Academic Press.

ECHLIN, P. & GODWIN, H. (1968). The ultrastructure and ontogeny of pollen in Helleborusfoetidus L. I. The development of the tapetum and Ubisch bodies. J. Cell Sci. 3, 161-174.

FEDER, N. & O'BRIEN, T. P. (1968). Plant microtechnique: some principles and new methods.Am. J. Bot. 55, 123-144.

GUNNING, B. E. S. & STEER, M. W. (1975). Ultrastructure andthe Biology of Plant Cells. London:Arnold.

HEPLER, P. K. & PALEVITZ, B. A. (1974). Microtubules and microfilaments. A. Rev. PI. Physiol.25, 309-362.

HESLOP-HARRISON, J. (1966). Cytoplasmic connexions between angiosperm meiocytes. Ami.Bot. 30, 221-230.

HESLOP-HARRISON, J. & DICKINSON, H. G. (1969). Time relationships of sporopolleninsynthesis associated with tapetum and microspores in Lilium. Planta 84, 199-214.

HOEFERT, L. L. (1971). Ultrastructure of the tapetal cell ontogeny in Beta. Protoplasma 73,397-406.

HORNER, H. T. & LERSTEN, N. R. (197I). Microsporogenesis in Citrus Union (Rutaceae). Am.J.Bot. 58, 72-79-

HUGHES, J. & MCCULLY, M. E. (1975). The use of an optical brightener in the study of plantstructure. Stain Technol. 50, 319-329.

JUNGALWALA, F. B. & DAWSON, R. M. C. (1970). The origin of mitochondrial phosphatidyl-choline within the liver cell. Eur. J. Biochem. 12, 399-402.

MARQUARDT, H., BARTH, O. M. & VON RAUDEN, U. (1968). Zytophotometrische und elektronen-mikroskopische Beobachtungen uber die Tapetumzellen in den Antherea von Paeoniatennifolia. Protoplasma 65, 407-421.

MOORE, T. S., LORD, J. M., KAGAWA, T. & BEEVERS, H. (1973). Enzymes of phospholipidmetabolism in the endoplasmic reticulum of castor bean endosperm. PI. Physiol., Lancaster52, 5°-53-

REAVEN, E. P. & AXLINE, S. G. (1973). Subplasmalemmal microfilaments and microtubules inresting and phagocytizing cultivated macrophages. J. Cell Biol. 59, 12-27.

RISUENO, M. C, GIMENEZ-MARTIN, G., LOPEZ-SAEZ, J. G. & GARCIA, M. I. R. (1969). Originand development of sporopollenin bodies. Protoplasma 67, 361-374.

ROBARDS, A. W. (1975). Plasmodesmata. A. Rev. PI. Physiol. 26, 13-29.SKVARLA, J. J. & LARSON, D. A. (1966). Fine structural studies of Zea mays pollen. I. Cell

membranes and exine ontogeny. Am.J. Bot. 53, 1112-1125.SMITH, D. L. (1974). A histological and histochemical study of the cotyledons of Phaseolus

vulgaris L. during germination. Protoplasma 79, 41-57.STEER, M. W. (1974). The development of tapetal cells in Avena sativa L. In Electron Micro-

scopy 1974, vol. 2. (ed. J.V.Sanders & D. J. Goodchild), pp. 594-595. Canberra: TheAustralian Academy of Science.

STEER, M. W. (1975). Evolution in the genus Avena: inheritance of different forms of ribulosediphosphate carboxylase. Can. J. Genet. Cytol. 17, 337-344.

TILNEY, L. G. (1968). Ordering of subcellular units. The assembly of microtubules and theirrole in the development of cell form. Devi Biol., Suppl. 2, 63-102.

{Received 15 October 1976)