J. Cell Sci. 32, 493-5°! 0976) 493Printed in Great Britain

POLLEN ULTRASTRUCTURE IN ANTHER

CULTURES OF DATURA INNOXIA

III. INCOMPLETE MICROSPORE DIVISION

J. M. DUNWELL AND N. SUNDERLANDJohn Innes Institute, Colney Lane, Norwich, U.K.

SUMMARYDuring the microspore division in Datura innoxia, the mitotic spindle is oriented in planes

both perpendicular (PE) and oblique (OB) to the spore wall against which the nucleus issituated. However, irrespective of polarity, the usual type of hemispherical wall is laid down atcytokinesis and isolates the generative cell from the rest of the pollen grain (type A). In PEspores the vegetative nucleus initially occupies a central position in the pollen grain, whereas inOB spores the vegetative nucleus lies at the periphery of the grain close to the generative cell.In anther cultures initiated just before the microspore division is due to take place, no markedchange can be observed in either orientation or symmetry of the mitotic spindle when thespores divide. In some, however, cytokinesis is disrupted and deposition of the hemisphericalwall arrested. In the absence of a complete wall, differentiation of the generative cell cannot takeplace and binucleate pollen grains are formed having 2 vegetative-type nuclei (type B). The2 nuclei in the B pollens are always situated against the pollen-grain wall, suggesting that thedisruption phenomenon is related to the OB spores. The incomplete wall always makes contactwith the intine on the intine-side of the spindle. Wall material may be represented merely asshort stubs projecting out from the intine into the cytoplasm, in which event the 2 nuclei lieclose to each other and are separated by only a narrow zone of cytoplasm. In other grains thewall is partially developed between the nuclei and terminates at varying distances from thetonoplast; in these, the nuclei are separated by a wider zone of cytoplasm. The significance ofthese binucleate grains in pollen embryogenesis is discussed.

INTRODUCTION

The formation of atypical pollen grains lacking the usual generative-within-vegetative cell arrangement has been reported in anther cultures of several species,among which Datura innoxia is the most notable (for review see Sunderland, 1974).These pollens (designated B pollens to distinguish them from the normal A type)provide one of the recognized pathways of pollen embryogenesis and are thought bysome workers to be the principal route in the formation of haploid embryos (Nitsch &Norreel, 1973 ; Nitsch, 19740,6 ; Rashid & Street, 1974). The B pollens are generallyequated with anomalous pollen grains that arise in vivo in certain genotypes, and areinducible by extreme-temperature treatments (Sax, 1935 ; La Cour, 1949). Theformation of these anomalous pollens is ascribed to a switch in orientation of themitotic spindle at the microspore division such that the spindle lies parallel to themicrospore wall instead of in the more usual perpendicular plane. As a result thespindle is symmetrical; the wall that develops is straight and extends across the sporeso that if the nucleus is suitably positioned the spore is partitioned into 2 equal cells.

494 J- M. Dunwell and N. Sunderland

This interpretation has been adopted by Nitsch & Norreel (1973) to account for theformation of B pollens in Datura cultures.

In this paper we describe the ultrastructure of the B pollens produced in anthercultures of D. innoxia and discuss the mechanism of their formation.

MATERIALS AND METHODS

Observations were made on. the anther cultures of Datura innoxia Mill, described in the firstpaper of this series (Dunwell & Sunderland, 1976a). Spindle orientations were assessed in thelight microscope, scoring being restricted to spores in which the mitotic figure could be seen intrue median view in relation to the long axis of the spore (Figs. 1, 2).

RESULTS

Microspore mitosis

In accordance with the observations of Nitsch & Norreel (1973), spindle orientationsare seen in planes both perpendicular to the microspore wall (Fig. 1) and oblique to it(Fig. 2). The relative frequencies of the 2 forms vary from one anther to another ; onaverage, perpendicular spindles occur slightly more frequently in both control andcultured anthers (Table 1). Spindles oriented in a plane parallel to the microspore wallcan occasionally be observed but they occur in both control and cultured anthers andin similar frequencies. Parallel spindles occur too infrequently to be located in theelectron microscope. No difference can be discerned between control and culturedanthers with respect to the structure and symmetry of the mitotic apparatus duringdivision of the microspore nucleus. An example typical of many seen in vitro is shownin Fig. 5.

Cytokinesis

As previously described (Dunwell & Sunderland, 1976 a), cytokinesis is for the mostpart regular in cultured anthers and gives rise to typical pollen grains (type A) having

Figs. 1-4. Light-microscope views of the microspore division in D. innoxia in vivo.Anthers fixed overnight in acetic acid : ethanol (1 : 3 v/v) and after hydrolysis stainedin Feulgen reagent for 2 h. Anthers squashed and pollen mounted in acetocarmine.x 1900.

Fig. 1. Anaphase in a spore (J>e) predisposed to divide in a plane perpendicular to thewall against which the nucleus is situated.

Fig. 2. Anaphase in a spore (ob) predisposed to divide in a plane oblique to the wallagainst which the nucleus is situated.

Fig. 3. Young pollen grain showing the disposition of the generative and vegetative(vn) nuclei just after completion of the microspore division in a PE spore.

Fig. 4. Young pollen grain showing the disposition of the generative and vegetative(vn) nuclei just after completion of the microspore division in an OB spore.Fig. 5. Electron micrograph of the microspore nucleus at metaphase in a PE spore(cf. Fig. 1) from an anther of D. innoxia cultured just prior to the microspore division.The culture procedure does not appear to affect this mitosis in any way. Stage 3 antherafter 24 h in culture at 28 °C in darkness, x 15000.

Pollen ultrastructure in anther cultures. Ill 495

496 J. M. Dunwell and N. Sunderland

the 2 sister nuclei separated by the usual hemispherical wall. Immediately aftercompletion of cytokinesis the pollen grains have the vegetative nucleus in either acentral (Fig. 3), or peripheral, position (Figs. 4, 6). The central position results fromperpendicular spindle formation, whereas the peripheral position results mostly fromoblique spindle formation. Counts made in the light microscope indicate that theperipheral position occurs slightly more frequently than the central in both controland cultured anthers (Table 1). Since oblique spindles are slightly less frequent, theexcess of grains showing the vegetative nucleus in the peripheral position suggestsmigration of the centrally positioned nucleus in some instances. Migration of thenucleus probably takes place slightly more frequently in the cultured anthers. Veryoccasionally the vegetative nucleus migrates to the other side of the pollen grain oppo-site to the generative cell.

Table 1. Percentages of (i) perpendicular and oblique mitotic spindles during the microsporedivision in Datura innoxia in vivo and in vitro, and of (it) centrally and peripherallydisposed nuclei in the vegetative cell just after completion of the division

In vivoIn vitro

Orientation of spindle

Totalcounted

18895

Perpendicular Oblique

57 4353 47

Position

Totalcounted

488187

of vegetative

Central

4437

nucleus

Peripheral

5663

Type B pollen grains can be identified by incomplete wall formation. The 2 nuclei,which in the examples illustrated here were shown by serial sections to be the onlynuclei present, lie adjacent to each other in a peripheral position close to the pollenwall (Figs. 7-9). They both have similar interphase profiles and each contains aprominent nucleolus. The extent of anomalous wall development varies from onegrain to another. In some, an incomplete wall separates the nuclei; it makes contact onthe intine side of the spindle and terminates in the cytoplasm on the vacuole side ofthe spindle at varying distances from the tonoplast (Figs. 7, 8). The incomplete waJlhas the same appearance as the hemispherical wall produced in type A grains, butfrequently follows a sinuous course (Fig. 8). In grains showing incomplete walls thenuclei are separated by a relatively wide zone of cytoplasm occupied by all types of

Fig. 6. Electron micrograph of a young pollen grain (type A) derived from an OB sporeto show the peripheral position of the vegetative nucleus (vn). From the same culture asthe spore illustrated in Fig. 5. x 7000.Fig. 7. Electron micrograph of an embryogenic B pollen grain in D. innoxia showing arelatively extensive wall between the two sister nuclei (n). The wall reaches almost tothe tonoplast. The nuclei are separated by a wide zone of cytoplasm and each containsa prominent nucleolus. Note the osmiophilic deposit (arrows) that coats the tonoplast,and is characteristic of embryogenic grains in this species. Stage 3 anther after culturefor 48 h at 28 °C in darkness, x 9100.

Pollen ultrastructure in anther cultures. Ill 497

33

4Q8 J. M. Dunwell and N. Sunderland

Pollen ultrastructure in anther cultures. Ill 499

organelles (Figs. 7, 8). In other B pollens, there is no wall development between thenuclei, but wall material can be seen as ' stubs' which project a short distance from theintine into the cytoplasm (Fig. 9). In such grains the nuclei lie close to each other withtheir bounding membranes separated by only a narrow layer of cytoplasm. The closeproximity of the nuclei suggests that they move together during late telophase. In thelight microscope, closely apposed nuclei often give the impression of being joinedtogether especially if one nucleus overlaps the other, but in the electron microscope noevidence was found of bridge connexions between nuclei as have been described, forexample, in multinucleate protoplasts of soya bean (Fowke, Bech-Hansen, Gamborg &Constabel, 1975).

DISCUSSION

The data show that the atypical pollen grains produced in anther cultures of Daturainnoxia consist of binucleate cells in which sister nuclei share a common cytoplasmderived from the mother microspore. This is true even of cases where there is anincomplete wall reaching almost to the tonoplast (Fig. 7), although in the lightmicroscope such grains might well be mistaken for bicellular structures. Incompletecytokinesis and anomalous wall development at the microspore division must inevitablydivert the normal course of pollen development - in the absence of the hemisphericalwall a generative cell cannot be formed. In this situation considerations of spindleorientation and spindle symmetry are of secondary importance.

However, while we have not found evidence of a change in the orientation of thespindle when Datura spores divide in culture, the observation that the 2 nuclei in theB pollen grains invariably occupy a peripheral position close to the pollen grain walldoes suggest that disrupted cytokinesis cannot be entirely divorced from considera-tions of spindle orientation. As has been seen, peripheral disposition of the nuclei isclosely related to obliqueness of the spindle at the microspore mitosis. Granted thatmigration of the presumptive vegetative nucleus from a central to a peripheral positiondoes sometimes occur, we think the complete absence of B pollens showing onenucleus in a central position, as though it had originated from a perpendicular spindle,is highly suggestive of a causal relationship between disrupted cytokinesis and spindleobliqueness.

The tendency to spindle obliqueness during the microspore division in D. innoxia isinnate, and determined prior to culture, presumably at meiosis (for further discussion

Fig. 8. Section of another B pollen grain of Datura innoxia, similar to that shown inFig. 7 but having a shorter wall partially separating the nuclei. The nuclei (n) arecloser together than in Fig. 7. The wall follows a more sinuous course than thatseen in normal development and appears to be interrupted at intervals by cytoplasmicbridges, x 26000.Fig. 9. Section of a B pollen grain taken from the same anther as the grain illustratedin Fig. 8. In this case the wall is represented by only a short stub ; the 2 nuclei (n) lie inclose apposition and are separated by an exceedingly narrow zone of cytoplasm,x 41000.

33-2

500 J. M. Dunwell and N. Sunderland

see Sunderland, 1974). Our contention is that this predetermined and atypicalpolarity, which is present in roughly 45 % of the spore population in an anther, is thebasis for the irregularity in culture. The conclusion has support from Nicotianatabacum which shows little atypical polarity at the microspore division and little Bpollen formation in vitro.

On the assumption that disrupted cytokinesis and spindle obliqueness are causallyrelated two immediate questions arise: (i) why all OB spores (those predisposed todivide in an oblique plane) do not show the disruption ; and (ii) why there should besuch a variation in the amount of wall laid down ? A plausible hypothesis can be putforward in terms of (i) the asynchrony of the spore population at inoculation, and (ii)the onset of embryogenesis of which disrupted cytokinesis is the first sign. It isenvisaged that the switch in programme is not effected immediately at anther inocula-tion and that the delay is sufficiently long to allow the most advanced OB spores atinoculation to pass through nuclear division and normal wall formation before theyare arrested. Other OB spores are actually in cytokinesis when the switch is effected,some being slightly more advanced (Figs. 7, 8) than others (Fig. 9). Not all cytokineticprocesses are, however, arrested. Assembly of the nuclear membranes is completedand the nuclei pass into interphase. Similarly, wall deposition is completed in thoseareas where presumably phragmoplast development, and aggregation of the plate-forming vesicles, has begun. A situation can be visualized where cytokinesis isinterrupted before the plate has made contact with the intine. Yet such situations werenot observed. In this connexion the B pollens as illustrated in Fig. 9 may be pertinent.It is possible that the stubs represent unattached portions of wall material which havefailed to aggregate correctly owing to disruption of the spindle microtubules and havebeen pushed against the intine as the nuclei reform and migrate together. They may,on the other hand, represent a short centripetal development of wall material homo-logous with a primitive type of cell-cleavage found in some non-vascular plants (seee.g. Pickett-Heaps, 1969). If such is the case, however, one would expect to see twostubs emanating from the intine rather than one, but this was not observed in the Bpollens.

Our interpretation of this study of OB and PE spores in D. innoxia is that, after thestimulus for embryogenesis becomes operative, the former respond slightly earlierthan the latter. While this may seem an insignificant difference it does have importantimplications in pollen embryogenesis. Like the pollen grains that enter the A pathway,the B pollens start to divide after only a short interphase. The nuclei divide eithersimultaneously or independently (Sunderland, Collins & Dunwell, 1974). In the lightof the present results, simultaneous divisions are probably associated with pollens inwhich there is no barrier between nuclei, and independent divisions with pollens inwhich the barrier is incomplete. After simultaneous division, lack of a barrier willallow mixing of the two closely apposed groups of chromosomes and these may beexpected to divide on one spindle and thus give rise to proembryos having diploidnuclei with genes all in the homozygous state. The situation is like that in the Cpathway of embryogenesis (see Dunwell & Sunderland, 19766) which generates stillhigher numbers of identical chromosome sets in the same pollen grain. Confirmation

Pollen ultrastructure in anther cultures. Ill 501

of fusion must await a more detailed analysis and the finding, in the electron micro-scope, of B pollens actually in division.

Information is also required on the orientation of the spindle upon which the twoB nuclei divide. As indicated in relation to nuclear fusion in the C pathway (Dunwell &Sunderland, 19766), evidence from light-microscope observations suggest that, afterfusion of the two peripherally disposed nuclei, the spindle may in some instances beoriented in a plane parallel to the pollen-grain wall. The cell plate thus developsacross the grain to give a proembryo with two equal cells. A situation is thus estab-lished which is consistent with Sax's (1935) mechanism, but it has the subtle differencethat the proembryo is diploid and not haploid.

REFERENCES

DUNWELL, J. M. & SUNDERLAND, N. (1976a). Pollen ultrastructure in anther cultures of Daturainnoxia. I. Division of the presumptive vegetative cell. J. Cell Sci. 22, 469—480.

DUNWELL, J. M. & SUNDERLAND, N. (19766). Pollen ultrastructure in anther cultures of Daturatmtoxia. II . The generative-cell wall. J. Cell Sci. 22, 481-491.

FOWKE, L. C , BECH-HANSEN, C. W., GAMBORG, O. L. & CONSTABEL, F. (1975). Electron-microscope observations of mitosis and cytokinesis in multinucleate protoplasts of soybean.J. Cell Sci. 18, 491-507.

LA COUR, L. F. (1949). Nuclear differentiation in the pollen grain. Heredity 3, 319-337.NITSCH, C. (1974a). La culture de pollen isol6 sur milieu synth^tique. C. r. hebd. Sianc. Acad.

Sci., Paris 278, 1031-1034.NITSCH, C. (19746). Pollen culture - a new technique for mass production of haploid and

homozygous plants. In Haploids in Higher Plants : Advances and Potential (ed. K. J. Kasha),pp. 123-135. Guelph : The University of Guelph.

NITSCH, C. & NORREEL, B. (1973). Effet d'un choc thermique sur le pouvoir embryogene dupollen de Datura innoxia cultive dans l'anthere ou isole de l'anthere. C. r. hebd. Sianc. Acad.Sci., Paris 276, 303-306.

PICKETT-HEAPS, J. D. (1969). The evolution of the mitotic apparatus : an attempt at comparativeultrastructural cytology in dividing plant cells. Cytobios 3, 257-280.

RASHID, A. & STREET, H. E. (1974). Segmentations in microspores of Nicotiana sylvestris andNicotiana tabacum which lead to embryoid formation in anther cultures. Protoplasma 80,323-334.

SAX, K. (1935). The effect of temperature on nuclear differentiation inmicrospore development.J. Arnold Arbor. 16, 301-311.

SUNDERLAND, N. (1974). Anther culture as a means of haploid induction. In Haploids in HigherPlants: Advances and Potential (ed. K. J. Kasha), pp. 91-122. Guelph : The University ofGuelph.

SUNDERLAND, N., COLLINS, G. B. & DUNWELL, J. M. (1974). The role of nuclear fusion inpollen embryogenesis of Datura innoxia Mill. Planta 117, 227-241.

{Received 29 March 1976)