Development 113, 931-939 (1991)Printed in Great Britain © The Company of Biologists Limited 1991

931

Laser microsurgery demonstrates that cytoplasm ic strands anchoring the

nucleus across the vacuoie of premitotic plant cells are under tension.

Implications for division plane alignment

KIM C. GOODBODY1, CATHARINA J. VENVERLOO2 and CLIVE W. LLOYD1*

'Department of Cell Biology, John Innes Institute, Colney Lane, Norwich NR4 7UH, UK^Department of Plant Molecular Biology, Botanisch Laboratonum, Rijksuniversiteit, Leiden, The Netherlands

* Author for reprints and correspondence

Summary

In epidermal cells of the plant Nautilocalyx lynchii,induced to divide by explanation, the nucleus undergoesa series of movements, on cytoplasmic strands, leadingto construction of a division plane across the vacuoie. Inthe early stage, the nucleus separates from the cortex,occupying an eccentric position in the cell, suspendedacross the vacuoie by few thin strands. In the centralstage, the nucleus occupies a central position anchoredto the cortex by more numerous, thicker strands.Finally, the phragmosome forms as a coalescence ofcytoplasmic strands across the cell, constituting thedivision plane within which mitosis and cytokinesis takeplace. The behaviour and alignment of these strands istherefore important since some are precursors of thedivision plane.

In a previous paper (Flanders et al (1990) J. Cell Biol.110,1111-1122), it was pointed out that the alignment ofcytoplasmic strands showed features common to avariety of elements under tension. That is, provided theyare free to move relative to the cortex, strands radiatingfrom the nucleus should tend to seek short rather thanlong routes to the cortex. In this way, strands undertension would move away from distant cell comerswhere two of the cell's edges make a three-way junctionwith a neighbouring wall. This provides a basis for theavoidance of four-way junctions and the maintenance ofthree-way junctions which are a characteristic feature ofmost plant tissues. In addition, tensile elements such as

soap bubble walls contact rigid surfaces perpendicu-larly. Perpendicular attachment of the cell plate to theside wall is embodied in Sachs' rule of cell division andthe existence of the premitotic strands in a state oftension would provide a basis for this rule.

In this study, laser microsurgery has been used toconfirm that strands connecting the premitotic nucleusto the cortex are under tension since the severed endsretract immediately upon severance. However,. theresponse of the nucleus to the breaking of a cytoplasmicstrand by laser depends upon the particular stage. Thenucleus is most likely to move during the early stagewhen it is eccentrically placed and has few cytoplasmicstrands. In the middle and phragmosomal stages thenucleus is more resistant. Computer-aided image recon-struction of anti-tubulin-labelled confocal sections indi-cates that the middle stage is characterised by thickmicrotubule (MT) bundles, which radiate from thenucleus and which seem to account for. the increasingstabilization of the nucleus. Although such strands aremore difficult to sever with the laser, their ends stilldisplay elastic properties when cut. It is proposed thattension is likely to influence the alignment of strands aswell as the position of the nucleus during division planeformation in vacuolated cells.

Key words: laser microsurgery, plant cells, cytoplasmicstrands, cell division, phragmosome, cytoskeleton.

Introduction

In vacuolated plant cells preparing to divide, thenucleus migrates into the centre of the cell, suspendedby cytoplasmic strands. Gradually these strands coa-lesce to form a transvacuolar sheet denning the divisionplane within which mitosis and cytokinesis occur(Sinnott and Bloch, 1940, 1941). The behaviour andproperties of these cytoplasmic strands are therefore an

important aspect of division plane determination andare the subject of microsurgical experiments in thispaper.

Subsequent to the classical studies of Sinnott andBloch, relatively little work has been performed ondividing vacuolated cells compared with numerousstudies on smaller meristematic cells. Nonetheless,investigations of vacuolated tissue cells and epidermalcells have established that these transvacuolar strands

932 K. C. Goodbody, C. J. Venverloo and C. W. Lloyd

contain actin filaments (Traas et al. 1987; Kakimoto andShibaoka, 1987; Lloyd and Traas, 1988) and micro-tubules (Bakhuizen et al. 1985; Flanders et al. 1990;Katsuta et al. 1990). Consistent with the idea that thesecytoskeletal elements are" involved in nuclear anchorageand positioning, anti-cytoskeletal drugs prevent thecentral location of the nucleus and alignment of thespindle in suspension cells (Lloyd and Traas, 1988;Katsuta et al. 1990). In epidermal cells of Nautilocalyxexplants, such drugs also perturb the accumulation ofcytoplasmic strands into the premitotic sheet (thephragmosome), and upset cell plate alignment (Venver-loo and Libbenga, 1987).

In a previous paper (Flanders et al. 1990), the MTs instrands radiating from the nucleus of Datura stramo-nium epidermal cells were studied by computer-aidedimage-reconstruction methods. First, this showed thatstrands which initially radiated from the nucleus invariable directions were increasingly relocated towardsa division plane also defined by an ever-tighteningcortical preprophase band (PPB) of MTs. That is, thephragmosomal plane became more defined as the PPBnarrowed. Second, in isodiametric cells which appearedas regular rather than elongated polygons in opticalsection, the strands radiating from the central nucleuswere often seen to take the shorter path to the middle ofa wall, rather than the longer path to the vertex. At suchcorners of a cell, two of the cell's facets form a Y-shapedjunction with a neighbouring wall. Avoidance of thesecorners by the cytoplasmic strands, then the phragmo-some, and hence by the cell plate, would appear to beimportant for maintaining the three-rayed junctionsseen in normal tissue in section, and for avoiding theformation of four-way junctions that would give tissue acheckerboard appearance. This behaviour of strandscan be modelled by tensile elements, such as springsand bubble walls, held in a flexible hexagonal frame,from which it was proposed that cytoplasmic strandswould also tend to seek the minimal, corner-avoidingpaths (a) if free to move and (b) if they, too, were undertension (Flanders et al. 1990). Hahne and Hoffman(1984) demonstrated by laser microsurgery that thestrands anchoring nuclei in freshly prepared protoplastsare under tension because the severed strand recoiledtowards the nucleus whilst cortical in-pullings disap-peared. It cannot be assumed, however, that strands inthickly walled cells behave similarly to those in freshlyisolated protoplasts, particularly since cytoskeletalcomponents may change as cells progress towardsmitosis. The aim of this study was therefore to test fortension in pre-mitotic strands of vacuolated tissue cells.Nautilocalyx lynchii was chosen because its cells aremore stable to laser treatment, and because thebehaviour of their strands during progression towardsdivision has already been carefully described (Venver-loo et al. 1980, Venverloo, 1990).

Materials and methods

Plant materialLeaves of Nautilocalyx lynchii (Hook f.) Sprague, which were

just fully expanded, were surface sterilized in 4% (v/v)hypochlorite solution. Explants consisting of the epidermisand up to 4 collenchyma layers were made from- the basal 3 cmof the midrib as described by Venverloo et al. (1980). Theexplants were cultured on modified Murashige and Skoog'smedium (Flow Laboratories) pH5.7, containing 1% (w/v)glucose; 0.8 % (w/v) agar and supplemented with 1 /JM indole-3 acetic acid (LAA), in small sealed Petri dishes with theepidermis facing the lid. Cultures were kept at 25°C in a 12 hlight cycle.

Light microscopyExperiments were performed on the epidermal cells of 3-6dold explants by transferring explants onto microscope slideswith a drop of water, and covering them with a coverslip forshort periods of time. Cells induced to divide by explantationproceed by the nucleus migrating into the centre of the cell,across the vacuole, on transvacuolar strands. The phragmo-some forms across the cell, either periclinally or anticlinally,depending on the thickness of the explant and the proximityof the cell to the wound surface. A Zeiss photomicroscopewith Nomarski optics was used for observation, either withx40 or a xlOO oil immersion objective.

Laser equipmentA nitrogen laser (Photochemical Research Associates Inc.,Model LN100) with a wavelength of 337 nm and a pulseduration of 800 psecs. was set up beside the microscope andthe emitted beam was guided throught the objective to focuson the specimen. The beam was positioned using a graticuleso that the target could be subsequently placed in the laserbeam, using the stage controls. The maximum average powerof the laser was 50 mW and the maximum energy per pulsewas 2.5 mJ. The x40 objective on the microscope was used toburst cells since the beam was not brought to such a narrowfocus. The x 100 objective enabled the beam to be focussed oncytoplasmic strands without bursting the cell. That the laserdid or did not puncture cells using a particular objective wasconfirmed by adding the colloidal dye ruthenium red to thewater. Dye only entered cells when the x40 objective wasused (ie. the cell was holed by the laser pulse), never when thexlOO objective was used to focus on a single transvacuolarstrand.

Data collectionThe microscope was fitted with a video camera linked to a TVmonitor and a RCA time-lapse video recorder. Photographsof video data were taken on Ilford PAN F 32 film directly fromthe monitor screen.

Immune'fluorescenceExplants were fixed in formaldehyde and stained with YOL1/34 anti-tubulin (Kilmartin et al. 1982) essentially as describedby Flanders et al. (1990).

Results

Severing of cytoplasmic strandsUsing the xlOO objective, the laser could be focussedprecisely upon a cytoplasmic strand. In this way, strandswere severed without bursting the cell as confirmed bycontinued cellular exclusion of ruthenium red added tothe medium. The advantage of the large Nautilocalyxlynchii epidermal cells is that the laser 'hot spot' can be

Laser microsurgery of cytoplasmic strands 933

V

Fig. 1. Phase-contrastmicrograph of Nautilocalyxlynchii epidermal peel afterculturing on nutrient agar. Thenuclei have migrated into thecentre of the vacuoles oncytoplasmic stands. In placesthe strands are broadening,particularly around thenucleus, which eventually leadsto formation of thephragmosome. Note theoccasional co-alignment ofstrands between neighbouringcells. Scale bar=10//m.

focussed well away from the plasma membrane. Arepresentative field of epidermal cells, 2.5 days afterexplantation, is shown in Fig. 1. This illustrates thecentral nuclei and their associated strands which inplaces are beginning to web.

The aim was to study the response of the strands tobeing broken and the resulting effect on the position ofthe nucleus. It became apparent, however, that thequality of response depended on the stage of pro-gression towards mitosis. Epidermal cells in an explantare only loosely synchronized (Venverloo et al. 1980)and they may not necessarily be following the same timecourse as cells in an adjacent explant. For this reason,stages were set according to physical features ratherthan to strict timing following explantation. Initially,the nucleus may migrate within the cortical cytoplasmas part of the traumatotactic response (Goodbody andLloyd, 1990). Then the nucleus begins to separate fromthe cortex to which it is connected by relatively fewcytoplasmic strands. Later, in a distinctly differentstage, the nucleus occupies a central location supportedby more and thicker strands, with organelles clusteredaround it. Finally, the phragmosome forms as a planarsheet of anastomosed cytoplasmic strands althoughnon-phragmosomal strands radiate away from thenucleus, out of the division plane. The four stages willtherefore be referred to as: traumatotactic (migrationof the nucleus, usually through the cortex, to a walladjacent to the wound), early, eccentric (lifting of thenucleus away from the cortex), central (nucleus in acentral location, supported across the vacuole byconspicuous cytoplasmic strands), and phragmosomal(when the cytoplasmic strands have subsequentlyreorganized to form the transvacuolar phragmosome).

The precise response of the nucleus to the severing ofa cytoplasmic strand was dependent on the stage, assummarized in Fig. 2. Initially, the nucleus underwenttraumatotactic migration in the cortical cytoplasm, in

100 -i

80 -

60 -

40 -

20 -

Early Middle PS

Fig. 2. Percentage of cells, at different stages, showingnuclear movement as a result of severing a cytoplasmicstrand by laser irradiation. First column. Traumatotactic(Tc) stage when nucleus is still within the corticalcytoplasm (22 cells). Second column. Early stage of nuclearmovement into the vacuole; eccentric nucleus with fewcytoplasmic strands (39 cells). Third column. Middle stage,when the nucleus is central, supported by thicker, morenumerous cytoplasmic strands (77 cells). Fourth column.Phragmosomal stage in which most of the strands haveaccumulated in one plane and webbed to form atransvacuolar sheet (15 cells).

response to explantation. While in the cortex, suchnuclei were not always associated with transvacuolarstrands but of 22 such cells selected for laser treatment,only one responded by nuclear movement afterseverance of a strand. In the early, eccentric stage(column 2) the response of the nucleus to the breakingof a strand seemed to depend upon its location, the

934 K. C. Goodbody, C. J. Venverloo and C. W. Lloyd

number of strands and on their particular arrangement.However, laser microsurgery of a single strand inducednuclear movement in 28 out of 39 cases. Of thosepositive responses, 25 nuclei moved away from thebreakage, 3 moved towards. Analysis of the videoindicated that this movement to a new position did notoccur immediately but over tens of seconds. In both thetraumatotactic and early, eccentric stages, cytoplasmicstrands could be easily broken by the laser, often with asingle pulse.

In the central stage (column 3), the thicker strandswere more difficult to sever, often taking several laserpulses to cause complete breakage. One commonlyobserved response, short of breakage, was the swellingof the strand at the focal point of the laser beam. It wasnot always possible to sever a strand in this middlestage. However, successful breakage did result inretraction of the cut ends. Compared to the previousstage in which the nucleus is eccentric (column 2), asmaller percentage (21 %, 16 out of 77) of nucleiunderwent movement in response to severance of asingle strand.

In the final stage, the strands had mostly accumulatedto form the webbed phragmosome with the nucleus atits centre. Venverloo (1990) has described the processwhereby cytoplasmic strands are interconnected bysheets of cytoplasm to form the phragmosome as'webbing'. As more or less continuous sheets, there-fore, phragmosomes are structurally different from un-interconnected single strands and could not be brokenby a few pulses. However, non-phragmosomal (i.e.polar) strands which are not webbed together behavedmuch as strands did in the central stage in that theirends retracted upon severance. However, only 2 out of15 nuclei were displaced as a result.

Particular examples from the aggregated data areillustrated in Figs 3 and 4. The upper right cell in Fig. 3contains a nucleus in the early eccentric stage of nuclearmigration. It has relatively few thin cytoplasmic strandsin various focal planes. The slight inflection in thestrand to the right of the arrow (which indicates thefocal point of the laser beam) represents a branch pointwhere another-strand splits towards the left anticlinalwall, but at a deeper level. Within 3-4 s of laserirradiation (single pulse) the strand to the left of the siteof severance can be seen on the video to havecompletely retracted to the anticlinal wall. In Fig. 4 thestrand in the upper right is severed 7 s after thisparticular frame. The cut ends of the strand retractwithin a few seconds and are no longer visible in 4b butthese two frames demonstrate the more extendedeffects of induced nuclear migration. Between 17 (timeof severance) and 40 s, the nucleus migrates towards thelower anticlinal wall. During this time two strands to theupper left combine (small arrows).

Anti-tubulin immunofluorescenceTo' investigate the possibility that the microtubularcomponent of the cytoplasmic strands was changingwith progressiontowards mitosis, explants were stainedwith anti-tubulin antibody. Stained cells were then

Fig. 3. Strand retraction following severance by the laser.Four seconds following the laser pulse (arrow) the strandhas retracted to the left anticlinal wall. The stub ofcytoplasm associated with the nucleus is attached to asubsidiary strand which travels in a lower focal planetowards the left anticlinal wall.

optically sectioned in 2-3/xm steps by confocal laserscanning microscopy, the sections reconstituted by theStardent image processing computer which projectedthe cytoskeleton as stereo-pairs, rotating about the yaxis.

Fig. 5a shows the upper half of an early stage cell.The interphase MTs can be seen at the cortex of thedomed outer epidermal wall. The stereopair shows thatthe nucleus is some distance away from the corticalcytoplasm. Anti-tubulin staining can be seen around thenucleus and in a few of the associated transvacuolarstrands but such staining is weak compared to thefollowing stage.

In the central stage, the nucleus (Fig. 5b) is muchmore heavily stained with thick bundles of MTs whichradiate from it. These bundles connect with the cortexwhich, at this stage, often show signs of incipient PPBformation - although the cortical MTs are not yettightly bunched,

A more or less completed phragmosome is illustratedin Fig. 5c. This is of a cell dividing in a slightly

Laser microsurgery of cytoplasmic strands 935

Fig. 4. Nuclear movement following severance of a strandby the laser. The focal point of the laser is arrowed in theupper frame. As a result of irradiation, the nucleusmigrates towards the opposite, lower anticlinal wall. Thesmaller arrows indicate two other strands that combinewhile the nucleus undergoes this induced movement.

oblique/periclinal plane but has been rotated to presentthe phragmosome parallel to the picture plane. ThePPB has bunched at this stage; it encircles the cell andits component MTs can be seen to the left, along theanticlinal wall. Some strands are coming from thenucleus towards the viewer. These strands to the outerepidermal wall are the polar strands. Most of the otherMT bundles around the nucleus belong to the phragmo-some and radiate within the phragmosomal plane toconnect with the PPB. Features characteristic ofphragmosome formation, as described by Venverloo(1990), can be seen in Fig. 5c. Near the nucleus, strandsshow evidence of webbing; near the cortex can be seensigns of fanning as the cytoplasm around strands splaysout as it contacts the cortex.

Effects of laser microsurgery on neighbouring cells(a) Co-aligned strands in neighbouring cells

Using the xlOO objective, individual strands weresevered to test the reaction of neighbouring strandswith which they lined up. Of twenty-four pairs of cells

showing co-aligned cytoplasmic strands, there was onlyone instance in which severing one resulted inmovement of the other.

(b) The effects of ablating one cell, upon itsneighbours

With the x40 objective, the laser could be used topuncture a cell. Target cells were generally in themiddle, cytoplasmic star stage but because synchrony isonly loose in the explants, neighbouring cells were in avariety of stages. This may account for the heterogen-eity of responses observed. Of 91 cells holed with asingle pulse, adjacent nuclei and/or neighbouringcytoplasmic strands were seen to move in 59 cases. In 12other instances there was no dramatic movement ofneighbouring nuclei despite the fact that their anticlinalwalls could be clearly seen to bulge into the damagedcell.

Fig. 6 shows the effect of holing one cell upon itsneighbour. The irradiated cell is marked with a crossand over the 22 s separating the frames, the nucleus inthe adjacent cell (arrowed) has moved towards theseparating wall.

Discussion

Cytoplasmic strands are under tension during themovements leading to phragmosome formationThe main aim of this study was to determine whethertransvacuolar cytoplasmic strands anchoring the pre-mitotic nucleus are under tension. Observations on thebehaviour of strands following laser ablation confirmedthat this is the case. Immediately following severance,the cut ends retracted back to the cortex or thecytoplasmic mass around the nucleus. The strands aretherefore under tension. A further example of elasticitywas provided by Y-shaped cytoplasmic strands (notshown). The co-equal angles often adopted by the limbsindicated that each was exerting equivalent tension, inthe manner of bubble walls that meet in a froth, and asshown by other minimal area networks (see D'ArcyThompson, 1942). Laser severance of one of the strandsattached to the cortex caused the instantaneousstraightening of the two residual limbs, which remainedto bridge the nucleus to the cortex. This thereforeconfirms Hahne and Hoffman's (1984) elegant lasermicrosurgery study of Hibiscus protoplasts, whichindicated that the strands anchoring the nucleus wereunder tension and could cause in-pullings of the plasmamembrane. The present observations extend thatconclusion to tissue cells in the process of dividing. Itwas important to perform such experiments on tissuecells since the presence of a wall might reasonably beexpected to alter behaviour. Indeed, strands attachedto a wall, as here, were unable to cause the in-pullingsseen by Hahne and Hoffman (1984) in freshly isolatedprotoplasts. Furthermore, we wished to see how thebehaviour of the strands changed as they reorganized toform the phragmosome.

These observations, and those of Ashkin and

936 K. C. Goodbody, C. J. Venverloo and C. W. Lloyd

Fig. 5. Computer-aided imagereconstruction of opticallysectioned anti-tubulin-stainedepidermal cells. Sections at2^m intervals werereconstructed and rotatedabout the y axis as stereopairs.The upper pair presents thedomed outer epidermal walland part of the anticlinal walls.The cortical MTs are still inthe interphase configuration.Stereoscopy shows that at thisearly stage the nucleus hasbegun to migrate into thevacuole and has relatively littleassociated anti-tubulinimmunofluorescence. In themiddle stereopair, only half ofthe cell is projected to show amid-stage, central nucleus. Thenucleus is at the centre of acytoplasmic star composed ofconspicuous, thick bundles ofMTs which connect with thepericlinal wall where thecortical MTs are perhaps onlyjust beginning to bunch up toform the PPB. The lower pairis of an oblique periclinalphragmosome rotated to beparallel to the picture plane.In the phragmosomal stage,most of the MTs are in thisplane, connecting with thecortical PPB to form a bicycle-wheel configuration. The fewMT bundles coming from thenucleus towards the viewer,constitute the perpendicularpolar strands.

Dziedzic (1989) on cytoplasmic strands snapping backafter release from a laser trap, indicate that cytoplasmicstrands exist in a state of tension. However, theproperties of the strands, and the response of cells todamaged neighbours, appears to be modified during thegradual process that leads to phragmosome formation.In an initial study on the timing of phragmosomeformation in Nautilocalyx lynchii explants, Venverloo etal. (1980) contrasted the paucity of nucleus-associatedcytoplasmic strands just after the start of nuclearmigration, with the conspicuous three-dimensionalsystem of strands which, later, gives the central nucleusthe impression of being embedded in a cytoplasmic star.We have categorised the phases as traumatotactic, early

eccentric, central and phragmosomal. However, of thelatter three stages in which the nucleus is suspendedacross the vacuole, it is during the early phase that thenucleus is most likely to be displaced by severance of astrand. The early stage is classified by an eccentricnucleus suspended by few strands and so breakage ofany one of these strands is likely to have a proportion-ately large effect on the position of a nucleus which hasnot completed migration to the centre. By contrast,nuclei classified as central are likely to have achieved astable position in which they are supported by thicker,more numerous cytoplasmic strands. These featuresprobably explain the relative insensitivity of the centralnucleus to the severance of a single strand.

Laser microsurgery of cytoplasmic strands 937

Fig. 6. The effect of holing a cell upon a neighbouringnucleus. The upper cell (marked with a X) is holed usingthe laser with the x40 objective. Twenty two seconds later,the nucleus in the lower cell (arrowed in the lower frame)has migrated to the separating anticlinal wall.

However, it is clear from the present immunofluor-escence observations that the cytoskeletal compositionof the cytoplasmic strands changes considerably as thenucleus progresses from an early, eccentric position to acentral location. Comparison of the cytoplasmic strandsseen by phase contrast microscopy, with the immuno-fluorescence image, has established that every strandcontains MT bundles in N. lynchii epidermal cells with acentral nucleus (Venverloo, unpublished observations).It is quite likely, therefore, that this increase in thenumber of MTs in the strands contributes to theincreasing resistance of the nucleus to move when astrand is severed. This is also likely to be the basis of thethickening of the strands reported by Venverloo et al.(1980). These thicker strands are certainly moredifficult to sever with the laser. Nevertheless, afterbeing cut the ends begin to retract immediately whichseems unlikely if rigid, stable MTs were still present inthe strand. But because several pulses are required tobreak these strands, the treatment takes tens ofseconds. If MTs begin to depolymerize immediatelyafter the first pulse and not just when the strand finallybreaks, then MT shrinkage rates of 20^mmin~(Schulze and Kirschner, 1987) could be sufficient toallow for the cut ends of the strands to shrink back.Recently, Ashkin and Dziedzic (1989) used infraredlaser traps to tease out cytoplasmic strands in plant

cells; they found that released strands could snap backto the point of origin with speeds of up to 100/an s"1.

In addition to MTs, actin filaments are known to bepresent in cytoplasmic strands associated with thedividing nucleus. In dividing carrot cells, actin filamentswere shown to occur in the strands caging the nucleus atone end and contributing to the preprophase band atthe other, cortical end (Traas et al. 1987. Lloyd andTraas, 1988). This was also shown for BY-2 tobaccocells by Kakimoto and Shibaoka (1987) and Katsuta etal. (1990). Apart from vacuolated cells such as these,actin filaments have also been shown to be present inthe cytoplasmic strands of epidermal tissue cells. In aprevious study (Goodbody and Lloyd, 1990), it wasdemonstrated that when the nucleus began to undergotraumatotactic migration to the wound wall in Tra-descantia albovittata cells, it became associated with aperinuclear mass of actin filaments; F-actin was in thestrands during all subsequent stages as the nucleusmoved into the centre of the cell to divide. The entireprocess could be inhibited with cytochalasin D. Ideallyfor the laser studies we would have wished to have useda tissue whose cells can be presented to the laser andwhose actin filaments and MTs can be stained routinely.MTs in D. stramonium epidermis can be stained(Flanders et al. 1990) but not the actin. Because of thedifficulties of staining F-actin in some green tissues withrhodamine-phalloidin (RhPh) Tradescantia albovittatawas selected (Goodbody and Lloyd, 1990) but,although the white areas of leaves stain well for actin,we have had little success in visualizing the MTs. Theselarge epidermal cells frequently burst during lasertreatment. No one epidermal tissue that we havestudied combines all requirements. However, thesmaller N. lynchii epidermal cells do withstand lasertreatment using the X100 objective. The cells can alsobe stained well with anti-tubulin and the lack of stainingwith RhPh is partly compensated for by the fact that theresponse of these explants to cytochalasin D has beenwell characterized. Venverloo and Libbenga (1987)found that all stages leading to phragmosome formation- including the formation of cytoplasmic strands - couldbe inhibited by the drug, from which it seems probablethat F-actin is a component of the strands throughout.

Combining the information from the three epidermalsystems, it can be concluded that initial migration of thenucleus in the cortex involves actin filaments and can beinhibited by CD. Early movement of the nucleus awayfrom the wall also involves actin filaments althoughMTs may begin to appear in the strands at this stage. Bythe time the nucleus is in the centre of the cell, it issurrounded by thicker strands which now containbundles of MTs. Other studies have indicated thatnuclei becomes increasingly resistant to displacementby centrifugation as cells progress towards prophase(Pickett-Heaps, 1969; Mineyuki and Furuya, 1980).This is consistent with the increased stability of thenucleus to the effects of the breakage of strands by lasertreatment and it is likely that this increasing stabiliz-ation of the nucleus is due to the bundles of MTs thatappear in the strands.

938 K. C. Goodbody, C. J. Venverloo and C. W. Lloyd

Implications of tensile strands for the siting of thedivision planeThe specific aim of this paper was to test an hypothesis(Flanders et al. 1990) that premitotic strands invacuolated cells are under tension. In that study it wasobserved in isodiametric cells of D. stramonium, whichapproximated regular hexagons in optical section, thatcytoplasmic strands radiating from the nucleus tendedto connect with the middle of walls rather than corners.Plant tissues are packed in such a way that, in section,only 3 walls meet at a point (D'Arcy Thompson, 1942),indicating that there must be some avoidance mechan-ism preventing the attachment of a cell plate to a cornerwhere 2 of the cell's own edges plus a neighbouring wallmeet in three-way junction. Since the premitotic strandseventually reorganize to form the transvacuolar phrag-mosome, factors affecting their alignment must alsoaffect the positioning of the division plane. Flanders etal. (1990) demonstrated that the behaviour of strands inpremitotic epidermal cells could be modelled by springsor bubble walls held across a flexible hexagonal frame.The key point about elements under tension is,provided they are free to move, that they will tend toattach to stiffer edges (ie. edges that cannot be distortedby the tension) by the shortest route, usually attachingto such edges at right angles. Bubble walls, for instance,exert equal tension in a foam and therefore meet inthrees, forming equal 120° juncture angles, whereasthey attach to the stiffer walls of a container at rightangles. This behaviour is characteristic of a range ofelastic materials as described at length by D'ArcyThompson (1942).

In tissue cells, routes from the central nucleus to thecell's corners are likely to represent long rather thanshort paths. Since the corner or vertex is where aneighbouring cell plate has already attached, theavoidance by strands of the most distant points from thecentre provides a mechanism for the avoidance offorming four-way junctions. The results of lasermicrosurgery experiments in this study confirm thatnucleus-associated, premitotic strands are under ten-sion as required by the minimal path hypothesis. Inaddition to avoidance of four-way junctions, a corollaryof this hypothesis is that in attaching to cell edgesperpendicularly, tensile strands help to explain theobservations crystallized by Sachs' rule: that the newplane of division intersects the mother wall at rightangles (see D'Arcy Thompson, 1942; also see Lloyd,1991 for a review on cell geometry).

In a previous paper (Goodbody and Lloyd, 1990), itwas demonstrated that division planes can be inducedto take long rather than short routes across cells, forinstance, parallel to the long sides of onion epidermalcells when the wound that induced the division wassimilarly parallel to the long axis. Under woundconditions factors evidently cadse the attachment ofstrands (and subsequently cell plates) along long pathsand at oblique angles, apparently overriding conditionsprevailing in the unwounded state. The N. lynchiiexplants are, in a sense, a wound system; cells close to acut edge are influenced to divide parallel to the edge.

However, the influence of the wound diminishes withdistance, particularly in explants containing severallayers of cells, and in all cases reported here theepidermal cells were in the centre of the explant wherethe overriding effects of the wound were not evident(Venverloo and Libbenga, 1987).

Another objection to the minimal path hypothesis isthat adjacent cells in wounded tissue can dividesimultaneously, with aligned cell plates, forming four-way junctions. Avoidance of four-way junctions bytensile strands can only work where there is a corner, orindeed any other distant point, which would beenergetically unfavourable for the attachment of astrand that was free to move relative to the cortex. Korn(1980) noted that new cell plates did not expand for onecell cycle, during which time all other older wallscontinued to expand. Attachment of a non-expandingplate to an expanding wall was said to cause the latter tobuckle, converting a perpendicular joint towards threeco-equal angles of 120°. D'Arcy Thompson (1942)stated that the cell plate could only begin to bend theolder wall, to which it was attached perpendicularly,when the former had become equally strong. Eitherway, the acquisition of a new vertex depends on one celldividing ahead of its neighbour. When the other celllater divides there is a new vertex, marking theattachment of the neighbour, to be avoided bypremitotic strands if they are free to move towardsminimal paths. By contrast, because wound-induceddivisions occur more or less synchronously (eg. Venver-loo et al. 1980; Goodbody and Lloyd, 1990) there ispresumably insufficient time for one cell plate to bucklethe common wall before the adjacent cell divides. In theabsence of a vertex there is no reason why adjacent cellplates should not line up. Indeed, in the latter study onT. albovittata epidermal cells dividing parallel to a razorslit, it was shown that actin cables co-aligned acrossgroups of cells. The actin cables anticipated the path tobe taken by the cell plates that later formed four-wayjunctions between adjacent cells. Actin cables cantherefore form mirror images between cells in the waythat cytoplasmic thickenings form supracellular pat-terns that pass across groups of cells (Sinnott and Bloch,1945).

Co-alignment of strands between neighbouring cellsCo-aligned strands were sought in the present study inorder to test their connectivity by laser treatment. It isformally possible that two aligned strands in adjacentcells are somehow dynamically interlinked. This doesnot seem likely from observing the behaviour of strandsin living cells since they show no evidence of movingeither side of the separating wall in pairs. Severing onestrand by laser did not generally cause the other tomove. This indicates that they are independently andstably attached, albeit to the same point on theseparating wall. However, when entire cells were burst,by focusing the laser down the x40 objective, nuclearmovement and/or realignment of cytoplasmic strandswas seen in over two thirds of cases. Quantification ofthis response was complicated by the fact that mitotic

Laser microsurgery of cytoplasmic strands 939

synchrony is only loose in these explants and soneighbours are in a variety of stages with - as has beenshown - differing degrees of nuclear stabilization.Nevertheless, this does tend to show that the positionsof nuclei in sheets are maintained in a dynamicequilibrium, sensitive to the local topology.

This work was funded by The Agriculture and FoodResearch Council by way of a grant in aid to the John InnesInstitute.

We are indebted to Dr John White of the Laboratory forMolecular Biology, Cambridge for allowing us access to thelaser. We thank Dr Peter Shaw for advice and assistance onimage reconstruction, and Isobel Brewin for secretarialassistance.

References

ASHKIN, A. AND DZIEDZIC, J. M. (1989). Internal cellmanipulation using infrared laser traps. Proc. natn. Acad. Sci.U.S.A. 86, 7914-7918.

BAKHUIZEN, R., VAN SPRONSEN, P. C , SLUIMAN-DEN HERTOG, F.A. J., VENVERLOO, C. J. AND GOOSEN-DE-ROO, L. (1985).Nuclear envelope radiating microtubules in plant cells duringinterphase mitosis transition. Protoplasma 128, 43-51.

FLANDERS, D. J., RAWUNS, D. J., SHAW, P. J. AND LLOYD, C. W.(1990). Nucleus associated microtubules help determine thedivision plane of plant epidermal cells: Avoidance of four-wayjunctions and the role of cell geometry. J Cell Biol. 110,1111-1122.

GOODBODY, K. C. AND LXOYD, C. W. (1990). Actin filaments lineup across Tradescantia epidermal cells, anticipating wound-induced division planes. Protoplasma. 157, 92-101.

HAHNE, G. AND HOFFMAN, F. (1984). The effect of lasermicrosurgery on cytoplasmic strands and cytoplasmic streamingin isolated plant protoplasts. Europ. J. Cell Biol. 33, 175-179.

KAHMOTO, T. AND SHTBAOKA, H. (1987). Actin filaments andmicrotubules in the preprophase band and phragmoplast oftobacco plants. Protoplasma. 140, 151-156.

KATSUTA, J., HASHIGUCHI, Y. AND SHIBOAKA, H. (1990). The roleof the cytoskeleton in positioning of the nucleus in pre-mitotictobacco BY-2 cells. / . Cell Sci. 95, 413-422.

KILMARTIN, J. V., WRIGHT, B. AND MILSTEIN, C. (1982). Ratmonoclonal antitubulin antibodies derived by using a new non-secreting rat cell line. / . Cell Biol. 93, 576-582.

KORN, R. W. (1980). The changing shape of plant cells:transformations during cell proliferation. Ann. Bot. 46, 649-666.

LLOYD, C. W. (1991). How does the cytoskeleton read the laws ofgeometry in aligning the division plane of plant cells?Development Supplement 1, 55-65.

LLOYD, C. W. AND TRAAS, J. A. (1988). The role of F-actin indetermining the division plane of carrot suspension cells: drugstudies. Development 102, 211-222.

MINEYUKI, Y. AND FURUYA, M. (1980). Effect of centrifugation onthe development of premitotic positioning of the nucleus inAdiantum protonemata. Develop., Growth and Differ. 22,867-874.

PICKETT-HEAPS, J. D. (1969). Preprophase microtubules andstomatal differentiation: some effects of centrifugation onsymmetrical and asymmetrical cell division. J. Ultrastruct. Res.27, 24-44.

SCHULZE, E. AND KIRSCHNER, M. (1987). Dynamic and stablepopulations of microtubules in cells. / . Cell Biol. 104, 277-288.

SINNOTT, E. W. AND BLOCH, R. (1940). Cytoplasmic behaviorduring division of vacuolate plant cells. Proc. natn. Acad. Sci.U.S.A. 26, 223-227.

SINNOTT, E. W. AND BLOCH, R. (1941). The relative position ofcell walls in developing plant tissues. Am. J. Bot. 28, 607-617.

SINNOTT, E. W. AND BLOCH, R. (1945). The cytoplasmic basis ofintercellular patterns in vascular differentiation. Am. J. Bot. 32,151-156.

THOMPSON, D. W. (1942). On Growth and Form. Camb. Univ.Press. Camb. 1116.

TRAAS, J. A., DOONAN, J. H., RAWLINS, D. J., SHAW, P. J.,WATTS, J. AND LLOYD, C. W. (1987). An actin network ispresent in the cytoplasm throughout the cell cycle of carrot cellsand associates with the dividing nucleus. J. Cell Biol. 105,387-395.

VENVERLOO, C. J. (1990). Regulation of the plane of cell divisionin vacuolated cells. II. Wound-induced changes. Protoplasma155, 85-94.

VENVERLOO, C. J., HOVENKAMP, P. H., WEEDA, A. J. AND

LIBBENGA, K. R. (1980). Cell division in Nautilocalyx explants.I. Phragmosome, preprophase band and plane of division. Z.Pflanzenphysiol. 100, 161-174.

VENVERLOO, C. J. AND LIBBENGA, K. R. (1987). Regulation of theplane of cell division in vacuolated cells. I. The function ofnuclear positioning and phragmosome formation. J. PlantPhysiol. 131, 267-284.

{Accepted 22 July 1991)