J. Cell Sci. 36, 121-136 (1979) I2I

Printed in Great Britain @ Company of Biologists Limited 1979

ULTRASTRUCTURE OF THE ENDOPLASMIC

FACTOR RESPONSIBLE FOR CYTOPLASMIC

STREAMING IN CHARA INTERNODAL CELLS

REIKO NAGAI AND TOMIO HAYAMADepartment of Biology, Faculty of Science, Osaka University, Toyonaka,Osaka, 560 Japan

SUMMARY

Previous investigators have proposed that cytoplasmic streaming in Chara internodal cellsresults from the interaction between an endoplasmic factor and fibrils composed of microfilamentsin the stationary cortex. Using the internal perfusion technique, we confirmed the observationthat organelles which had been attached to the fibrils by decreasing the internal concentrationof ATP moved along the fibrils after ATP was introduced. Thin-sectioned specimens revealedthat endoplasmic organelles of various shapes were linked to microfUament bundles in theabsence of ATP. Linkage was effected by regularly arranged electron-dense materials with aspacing of 100 - 130 nm, at definite regions on each organelle. The organelles in question werestudied in negatively stained preparations of endoplasm. The organelles had some commonfeatures. (1) They were all membrane-limited. (2) Their sizes and configurations variedlargely. (3) One or more protuberances were present on them. (4) The protuberances wereusually rod- or horn-like. (5) Small globular bodies 20 - 30 nm in diameter were found inordered array with the same spacing as those in thin sections at the surface of the protuberances.(6) Many fine filaments were always attached to the surface of the protuberances. These finefilaments differed from F-actin in diameter (less than 4 nm) and inability to react with heavymeromyosin from rabbit skeletal muscle. The role of such components of the organelles incytoplasmic streaming is discussed.

A paracrystalline array of microfilaments with a transverse periodicity of about 38 nm ispresented, together with its optical diffraction pattern.

INTRODUCTION

It has been generally accepted that rotational cytoplasmic streaming in the characeancell is caused by the active shearing force generated at the interface between themoving endoplasm and the stationary ectoplasm (Kamiya & Kuroda, 1956; Kamiya,1959). Fibrils running parallel to the direction, of the streaming are located on theinner surface of the chloroplast files in contact with the streaming endoplasm (Kamit-subo, 1966). Each'of the fibrils is a bundle of microfilaments similar in appearance tomuscle F-actin (Nagai & Rebhun, 1966). They reversibly bind rabbit skeletal muscleheavy meromyosin to form arrowhead structures (Palevitz, Ash & Hepler, 1974;Williamson, 1974; Palevitz & Hepler, 1975). The arrowheads are oriented oppositeto the direction of the streaming (Kersey, Hepler, Palevitz & Wessells, 1976).

When the chloroplasts, together with the fibrils, are locally dislodged by stronglocal irradiation, the streaming endoplasm becomes stagnant or moves slowly, whilestreaming continues in other, intact areas. However, the streaming returns to normal

122 R. Nagai and T. Hayama

after the regeneration of the fibrils (Kamitsubo, 19726). These facts indicate thatfibrils composed of microfilaments are indispensable for endoplasmic streaming.

Factors which interact with microfilament bundles to produce the active shearinghave been studied by several investigators. Bradley (1973) discussed the possibilitythat motive force might be generated by reaction of actin with myosin, which maybe anchored at suitable sites on the endoplasmic reticulum. Based on differentialtreatment of the cell with an SH-reagent, JV-ethylmaleimide, Chen & Kamiya (1975)suggested thatthe factor or the putative myosin is localized in the streaming endoplasm.Williamson (1975), using a vacuolar perfusion technique (Tazawa, 1964), showedthat organelles which had been immobilized on the fibrils in the absence of ATP(inactive state) started to move along them after ATP was introduced (reactivation).Williamson's interpretation was that the streaming was caused by interaction betweenthe microfilaments and a myosin-like protein which he supposed was linked to theendoplasmic organelles. The motility required, in addition to ATP, millimolar levelsof Mg^ and free Caa+ at io~7M or less (Williamson, 1975). Tazawa, Kikuyama &Shimmen (1976) succeeded in controlling the cytoplasmic streaming by removingthe tonoplast through replacement of the natural cell sap with EGTA-containingsolution. They also showed that ATP and Mg^ were indispensable for the cytoplasmicstreaming in the tonoplast-free cells.

These findings lead us to conclude that the factor which interacts with micro-filaments in the presence of ATP and Mg24" is localized in the moving endoplasm. AsBradley and Williamson suggested, the factor may be carried by the endoplasmicorganelles. The next problem is to identify the organelles in question and to show howand where the factor is arranged on them. The purpose of the present study was tothrow light on this problem. Endoplasmic organelles are found to be equipped withmany fine filaments, apparently different from F-actin and with globular bodies whichseem to be the factor interacting with the microfilaments.

MATERIALS AND METHODS

Cells. Internodal cells of Chara australis were used. The cells were cultured outdoors in alarge plastic bucket filled with rainwater with soil at the bottom. In winter, the alga was keptin plastic boxes under artificial illumination in the laboratory. Cells 8-10 cm long wereisolated from adjacent cells and kept for at least 1 day before use in pond water.

Internal perfusion. Each cell was cut and internally perfused after Ta2awa et al. (1976).The solution for the internal perfusion was composed of 5 mM EGTA, 6 mM MgCl., 290 mMsorbitol and 5 mM Tris-maleate buffer (pH 7, adjusted with KOH). The cell was ligated atboth open ends after the natural cell sap had been completely replaced with the artificialmedium. Within 10 - 20 min after perfusion, the tonoplast disintegrated (tonoplast-free cells,Tazawa et al. 1976). In these tonoplast-free cells active streaming was observed, and they wereused as the starting materials for both light and electron microscopy.

Light microscopy. As the chloroplasts anchored in the cortical gel layer interfere with thelight-microscopic observation of the fibrils, they were dislodged by the 'window technique' ofKamitsubo (19726) before cells were internally perfused. A tonoplast-free cell was then placedon an apparatus similar to that used by Williamson for light-microscopic observation. Afterreplacing the bathing solution (pond water) with 0-3 M sorbitol solution, which is approximatelyisotonic with the perfusion medium, both ligated ends of the tonoplast-free cell were amputatedagain for the second and third perfusion. The streaming in the window area was observed undera Zeiss photomicroscope II with differential interference optics.

Endoplasmic factor in cytoplasmic streaming 123

Thin sectioning. Tonoplast-free cells were perfused again with the same medium used for thefirst perfusion. After perfusing the cell interior with an amount of medium 2 to 3 times the cellvolume, the cell was ligated at both ends. Endoplasmic flow was not observed after the secondperfusion. After confirming the cessation of streaming, the cell was fixed for thin sectioning.Cells in this situation were immersed in 2 % glutaraldehyde containing 6 mM MgClj and 50mMphosphate buffer (pH 7). A few minutes later, the cells were cut into small segments to acceleratefixation, then kept in the fixative for 1 h at room temperature.

Cells which had been perfused twice were perfused again with the same medium supple-mented with ATP. After restoration of streaming they were fixed with the same mediumcontaining 1 mM ATP to avoid extreme decrease in ATP concentration when the cells werecut into segments in the fixative.

All cells were postfixed with 1 % Os04 solution containing 6 mM MgCl2 and 50 mM phos-phate buffer for 1 h.

The specimens were embedded in Spurr's medium (1969) or Epon 812 (Luft, 1961) afterdehydration in an ethanol series. Thin sectioning was done on a LKB-ultratome with a diamondknife. Grids containing sections were stained with uranyl acetate dissolved in methanol andlead citrate before being examined on a JEM-100C electron microscope at 80 or 100 kV.

Negative staining. Streaming endoplasm in normal internodal cells was collected by cent-rifuging the cell at 130 - 140 g for 10 min. As the shifted endoplasm always started to moveimmediately after the cessation of centrifugation, the cells were chilled promptly to preventmovement by replacing the bathing solution in the centrifuge tube with a chilled one. Thecentrifugal region of the cell was ligated while the cell was chilled. The accumulated endoplasmstarted rotational streaming in the small segment of the cell when the temperature rose. Theligated endoplasm-enriched segment (volume: o-z — 0-5 fil) was placed in a drop (ca. 50 fil) ofthe perfusion medium containing 3 mM dithiothreitol (DTT) and o-i mM phenylmethylsulphonyl fluoride (PMSF). The endoplasm was suspended in the solution by cutting thesegment. Pieces of cell wall were removed with forceps. After gentle mixing, drops of theendoplasmic suspension were negatively stained with 1 % aqueous uranyl acetate on coppergrids coated with Formvar-carbon.

RESULTS

Cytoplasmic streaming in cells without vacuolar membrane

Active streaming in the tonoplast-free cells after first perfusion was probably dueto the presence of endogenous ATP still left in sufficient concentration to causestreaming (Shimmen, 1978). When the tonoplast-free cells were perfused again withthe perfusion medium without ATP, however, the organelles ceased to move im-mediately, or within several seconds. Then the fibrils appeared thick due to theattachment of many organelles along their whole length (Fig. IA). The situationcorresponds to the inactive state of Williamson. A few minutes after entry into theinactive state, the perfusion medium containing 1 mM ATP was introduced. Organellesanchored on the fibrils promptly started moving along the fibrils. The direction ofthe movement was the same as before. The number of organelles coming into theobserved area diminished with time. The fibrils finally appeared naked and clean(Fig. IB).

It is clear from these observations that we could set up the rigor combinationbetween the organelle and the subcortical fibril, and could reactivate the system toproduce active movement with Mg-ATP. Using this system, the ultrastructure ofthe organelle and the linkage between the organelle and the fibril were investigated, asdescribed in the next section.

I 2 4 R. Nagai and T. Hayama

Linkage between fibrils and organelles

Thin sections of cells which had been fixed in the inactive state were made alongthe long axis of the chloroplast files. Fig. 2 shows one of these sections. Rows ofbundles (mf), some of which are associated with chloroplasts (chl), and several otherorganelles (arrows) anchored to the bundles, can be seen. The arrangement of thebundles and their spacing coincide well with that observed previously (Nagai & Rebhun,

Fig. 1. Fibrils in the' windowed region' of a Chara cell, A, fibrils appear thick due to theanchoring of many endoplasmic organelles in the absence of ATP. B, fibrils appearingnaked and clean after the restoration of movement of organelles by introduction of1 mM ATP. x 750.

1966). It is thus reasonable to conclude that the bundles are identical with subcorticalfibrils composed of F-actin (Palevitz & Hepler, 1975). Most of the endoplasmiccontents were dispersed due to the lack of tonoplast. The organelles shown in thefigure can reasonably be assumed to be the same as those seen tightly bound to thefibrils observed under the light microscope, since such organelles were not observed

Endoplasmic factor in cytoplasmic streaming 12§

in cells fixed after the restoration of cytoplasmic flow produced by introduction ofATP, as will be described later. The organelles are linked to the microfilaments bymeans of electron-dense structures projecting from their surface. These are arrangedin periodic order along the surface of the organelles.

Thicker sections were prepared to find whether or not the appearance in Fig. 2represents the configuration of the organelles as a whole. In sections of around 0-15 /tm

Fig 2. A longitudinal thin section of a tonoplast-free cell which is ATP-defkient.Rows of microfilament bundles {mf), chloroplasts (cht) and several organelles (arrow-heads) anchored to the bundles are seen. Notice electron-dense materials projectingfrom the surface of each organelle. They are arrayed regularly at the surface of theorganelles. x 44000.

§ CEL 36

R. Nagai and T. Hayama

3 A

\

- /",.

B

Fig. 3. A balloon-shaped organelle with 3 long protuberances (A), and with a singleprotuberance (B) observed in a thicker section (about 0-15 fim) of ATP-deficienttonoplast-free cells. Both are linked to microfilament bundles through electron-densematerials. The electron-dense materials in (B) are arrayed at intervals of ioo, 100, 100and 110 respectively from the right (arrows), A, x 51000; B, X 74000.

Endoplasmic factor in cytoplasmic streaming 127

we obtained more extensive views of their configuration, as can be seen in Fig. 3Aand B. The organelle shown in Fig. 3 A resembles a balloon with 3 long protuberances;2 are closely associated with a bundle of the microfilaments over their whole lengthand the third bends partly but finally comes into contact with the microfilaments atits tip. Notice that there are regularly arranged bridges between the protuberancesand microfilaments. Fig. 3B shows a balloon with a single protuberance. Here alsoelectron-dense bridges (arrows) are noticeable. The spacings between the bridgesshown are 100, 100, 100 and 110 nm, respectively from the right. We also observedan organelle attached to 2 adjacent microfilament bundles simultaneously at 2 sites.Based on the distinctive morphology of the 2 organelles shown above, the organellesin Fig. 2 may show part of organelles linked to the microfilaments.

20 -

10 -

A

r

- 1

-

- 1

1

40 -

20 -

B

-

—1

6 8 10 12 14 16

Spacing, nm X 10"'

6 8 10 12 14 16

Spacing, nm X 10"'

Fig. 4. Histograms showing the frequency distribution of spacing A, of the electron-dense materials on sectioned specimens (n = 84), and B, of the globular bodies onnegatively stained specimens (n = 124).

These electron micrographs revealed that (1) there must be large variations inorganelle size and morphology; (2) the organelles are linked, in the absence of ATP,to the microfilament bundles through special region(s), the protuberance(s), whichare functionally and morphologically differentiated in each organelle; (3) in thesestructures, electron-dense bridges are regularly arrayed; (4) 64% of the electron-dense bridges are located at intervals of 100 - 130 nm. Variation of spacing is shownin Fig. 4A. The average value (with S.D.) is calculated to be 111 ± 15 nm (n = 84).

When, the cell in the inactive state was perfused again with the perfusion mediumsuplemented with 1 mM ATP, the cytoplasmic streaming was restored due to restoralof an adequate ATP concentration (Shimmen, 1978). In sections of such cells,organelles linked to the microfilament bundles were not observed. This may be

9-2

128 R. Nagai and T. Hayama

interpreted to mean that the organelles in question could not remain linked to themicrofilaments during fixation because of the presence of internal and external ATP.

Structures of organelles revealed by negative staining

When characean internodes are centrifuged at moderate force, the streamingendoplasm collects at the centrifugal end of the cell. The shifted endoplasm alwaysstarts to move immediately after centrifugation along the chloroplast files, in the samedirection as before centrifugation (Kamiya & Kuroda, 1956; Hayashi, 1963). There-fore, the endoplasm collected by centrifugation was expected to contain many of theorganelles in question.

In negatively stained specimens, we observed many organelles of characteristicappearances, as expected from the sectioned specimens. Fig. 5A-D shows examplesof organelles of various sizes and shapes. They had features in common with those insectioned specimens. Further it was found that: (1) one or more protuberances werepresent on their bodies; (2) protuberances were usually rod- or horn-like; (3) smallglobular bodies were attached in ordered array on the surface of the protuberances(arrows); and (4) many fine filaments were always attached to and frequently frayedout from the surface of the protuberances (to be shown later). The organelle in Fig. 5 Ahas the same configuration as that in Fig. 3B. The organelle in Fig. 5D is similar tothat in Fig. 3 A.

The globular bodies on the horn-like protuberance are shown in Fig. 6 underhigher magnification. One globule is always found at the pointed end of the horn,(arrow 1). Except for the tip area, sets composed of 2 globules (arrows 2-4), perpen-dicular to the axis of the horn, are arrayed periodically at 120 run on average. Similarglobular bodies, 20-30 nm in diameter, arranged along the surface of the rod-likeprotuberance are shown in Fig. 7.

The spacing of the globules agrees well with that of the electron-dense material insectioned specimens. About 85% of the globular bodies lie within 100-130 nm apart.The variation of spacing is presented in Fig. 43. The average value (with SD) wascalculated to be 120 ±9 nm (n = 124).

Notice a filament (Fig. 7, small arrows) originating from the globular body andother long filaments running parallel to the long axis of the rod. Some of the longfilaments (large arrows) spread out from the area where the globular body is absent.It is not clear whether these filaments are identical or not. The diameter of an individualfilament is less than 4 nm. They did not react with heavy meromyosin from rabbitskeletal muscle, indicating that they differ from F-actin (data not shown).

The endoplasmic organelles rarely had globular bodies along the whole length ofthe rod-like protuberance. However, one globule was usually located in the regionof the tip, even when globules were rare or absent elsewhere. It is not clear whetherthe globules are detached during preparation from the surface of the rod-likeprotuberances or are intrinsically rare or absent.

Fig. 8 shows part of a large mat-like organelle. Globular bodies (arrows) of constantspacing on the organelle surface are noticeable and fine filaments are also numerous.

In some instances, the protuberance is composed of some smaller vesicles, which

Fig

. 5.

Var

ious

kin

ds o

f or

gane

lles

in

neg

ativ

ely

stai

ned

spe

cim

ens

of e

ndop

lasm

-ric

h su

spen

sion

. O

ne

(A) o

r m

ore

(B

,c,D

)

pro

tub

eran

ces

are

pres

ent

on e

ach

orga

nell

e.

Th

e p

rotu

bera

nces

are

of

rod-

like

(A

,B) o

r ho

rn-l

ike

(B,c

,D)

appe

aran

ce.

Y

N

Sm

all

glob

ular

bod

ies

(arr

owhe

ad)

are

arra

ng

ed a

t th

e su

rfac

e of

th

e p

rotu

ber

ance

in

per

iodi

c o

rder

wit

h a

co

nst

ant

spac

ing.

A,

x 5

00

w; B, x

23000; C,

x 4

70

~);

D

, x

180~0.

R. Nagai and T. Hayama

are interconnected by many fine filaments (Fig. 9). The fine filaments unite to formthicker curly filament bundles.

Bundles of microfilaments

In negatively stained preparations of the endoplasm, microfilament bundles werefrequently seen. They may come from the ectoplasmic layer of the endoplasm-enriched cell segment. The bundles were associated with large dense structures,which were probably chloroplasts. Some bundles were associated with large mem-branous structures and others lay free on the grids. Organelles linked to microfilamentswere rarely observed, perhaps because the ATP concentration of the endoplasmic

4r 3

r 2r 1

Fig. 6. Enlargement of the tip area of a horn-like protuberance. One globule is alwaysfound at the pointed end of the horn (arrow 1). Except for the tip area, sets composedof 2 globules (arrows 2-4), lying perpendicular to the axis of the horn, are arrayedperiodically, x 137000.

suspension was not extremely low, i.e. 2 - 5 /tM. This value was estimated from theratio of the volume of an endoplasm-enriched segment to a drop of the medium inwhich the endoplasm was suspended (cf. Methods), and the cytoplasmic ATP con-centration, which is around 0-5 raM (Hatano & Nakajima, 1963). Fig. IOA shows amicrofilament bundle embedded in a membranous structure. Microfilaments in thebundle are parallel to each other and a transverse periodicity of about 38 nm isobvious. Fig. IOB shows another example of a bundle in which the twisted and beadednature of each microfilament is more obvious. The transverse periodicity is also 38-39 nm. These filament bundles closely resemble a paracrystal formed in vitro frompurified F-actin or formed from actin, tropomyosin and troponin in the presenceof Mg2+ (Moore, Huxley & DeRosier, 1970; Spudich, Huxley & Finch, 1972; Ohtsuki &Wakabayashi, 1972; Gillis & O'Brien, 1975; Wakabayashi, Huxley, Amos & Klug,1975). In addition, naturally occurring actin paracrystals have been reported in theacrosomal process of Limulus sperm (Tilney, 19750,6). Fig. 10c and D shows the

Endoplasmic factor in cytoplasmic streaming

ii

Fig. 7. Enlargement of a rod-shaped protuberance. Notice globular bodies arrangedperiodically, fine filaments originating from the globule (small arrowheads) and longfilaments running parallel to the long axis of the protuberance. Some of the longfilaments spread out from the area where the globular body is absent (large arrowheads),x 118000.

Fig. 8. Part of a large mat-like organelle equipped with globular bodies (arrowheads)and many fine filaments, x 138000.

I 3 2 R. Nagai and T. Hayama

B

Fig. 9. Protuberance composed of small vesicles interconnected by many fine filaments.Fine filaments unite to form thicker curly filament bundles, x 118000.Fig. 10. Microfilament bundles observed in endoplasm-rich suspension, A, a bundleof microfilaments embedded in a membranous structure. Microfilaments in the bundlelie parallel to each other with transverse periodicity of about 38 nm. The twisted andbeaded nature of each microfilament is more obvious in B. C, optical diffraction patternfrom (A); I-O cm corresponds to 9-17 nm. D, optical diffraction pattern from B; i*o cmcorresponds to 7-47 nm. A, X 127000; B, x 168000.

Endoplasmic factor in cytoplasmic streaming 133

optical diffraction patterns of the bundles in Fig. IOA and B, respectively. Bothpatterns show features characteristic of F-actin bundles: the layer line at a spacing ofabout 37 nm arises from the double-stranded nature of F-actin and one with a spacingof about 5-8 nm is the reflexion from the basic helix. The spacings of these layerlines are 578 and 37-0 nm in Fig. 10c and 5-84 and 37-8 nm in Fig. IOD. The spacingratio of the 2 layer lines is 6-40 in Fig. 10c and 6-47 in Fig. IOD. These values indicatethat the basic helix of F-actin in our preparation has 28 subunits per 13 turns.

The nature of F-actin in the microfilament was also confirmed with specimensprepared by squeezing out the cytoplasm of the perfused cells. The microfilamentswere decorated with heavy meromyosin from rabbit skeletal muscle to form arrow-heads (data not shown) which were shown previously by Palevitz et al. (1974).

DISCUSSION

It has been suggested (Palevitz & Hepler, 1975) that microfilaments are packedin vivo in a paracrystalline array in characean internodes. A paracrystalline array ofmicrofilaments with a transverse periodicity of about 38 nm is clearly shown in thepresent paper. The paracrystal of microfilaments was probably not formed in the per-fusion medium supplemented with DTT and PMSF, since the Mg2+ concentrationin the medium is lower (6 mM) than that necessary for paracrystal formation fromactin in vitro (25-50 misi).

DeRosier et al. (1977) showed that paracrystals consisted of hexagonal arrays ofactin filaments cross-linked by a second protein having a molecular weight of about55000. Bradley (1973) observed bridges connecting each microfilament in Nitella.Allen & Condeelis (unpublished) found a 55000 Dalton component in Nitella extracts.It may be possible that a protein of 55000 Daltons has the role of maintaining theparacrystalline array of the microfilaments in vivo.

The endoplasmic organelles which we observed here were quite different fromother endoplasmic organelles from the structural point of view. First, they borecharacteristic protuberance. Secondly, the organelles were equipped with globularbodies arrayed in periodic order on the surface of their protuberances and wereequipped also with fine filaments attached to the surface of the protuberances.

We mentioned above that in the inactive state produced by the absence of ATP, inwhich actin and myosin would be in rigor combination, the organelles were linkedto the microfilaments by electron-dense material. This must correspond to the globularbodies seen in negatively stained specimens, because of the similarity in size andspacing on the surface of the organelles (cf. Fig. 4A, B). We can reasonably supposethat the globular bodies act as functional units when the endoplasmic organelles slidealong the microfilaments.

Myosin-like protein has been extracted iromNitella and studied(Kato & Tonomura,1977). The protein formed bipolar aggregates resembling those of myosin from rabbitskeletal muscle and other sources. Although we have no direct evidence, such asimmunological identification, the globular bodies may be composed at least of the

134 R- Nagai and T. Hayama

functional head of myosin or myosin aggregates, possibly with some other unknownprotein. Their periodic array itself suggests that the period might be determined bythe tail portion of the myosin molecule, or other unknown protein(s).

Fine filaments were also seen usually on the surface of the protuberances or on acertain other region of the organelles. In this connexion, it is interesting to touch uponthe endoplasmic filaments in Nitella revealed by scanning electron microscopy (Allen &Reinhart, 1976; Allen, 1977). They are thinner than subcortical fibrils and interwindto form loose networks. Some of them seem to originate from granular cytoplasmicstructures much smaller than the chloroplast (fig. 2 in Allen, 1977). Since the finefilaments observed in the present study often fray out from the organelle and havea property of uniting to form thicker curly filament bundles, they morphologicallylook like the endoplasmic filaments of Allen. He has suggested that the endoplasmicfilaments are composed, in part, of F-actin. Our fine filaments, however, are apparentlydifferent from F-actin in their inability to react with HMM from rabbit skeletal muscleand in diameter (less than 4 nm). It is unlikely that these properties come fromdamage to F-actin during preparation of the endoplasm, since in the same preparationactin filaments usually keep their native appearances and exist in the form of bundles(Fig. IOA, B, and at upper left corner of Fig. 8).

The role of the fine filaments in the movement of the organelles must be important,although it is unlikely that they play the leading role in the sliding mechanism. Theymay (1) stiffen the protuberances by attaching to them; (2) provide chemical sites forthe myosin-like molecules to locate properly, with a distinct polarity; or (3) help propelviscous endoplasm when frayed out into the endoplasm. Also, we cannot exclude thepossibility that'the rod parts of myosin lacking functional heads might assemble them-selves longitudinally to form fine filaments.

In some specimens in which the direction of the streaming had been confirmedbefore fixation, the tips of the protuberances pointed downstream, in the same direc-tion as the streaming, and the body part upstream. This appears very effective forcarrying viscous endoplasm together with the sliding organelles. Several protuberancesaround a single'organelle, as shown in Fig. 5D, probably point in the same directionand interact with a single bundle of the microfilaments, as in Fig. 3A. They may alsointeract with a few bundles simultaneously. When larger mat-like organelles movealong the microfilaments, the endoplasm could be more effectively dragged. Cyto-plasmic streaming could be explained solely by this mechanism. The idea of 'undu-lating endoplasmic filaments' as a cause of streaming (Allen, 1974) does not seem tobe pertinent.

The ultrastructural basis for rotational cytoplasmic streaming now appears to havebeen established. It remains to discover the molecular mechanism for the generationof the sliding force.

We wish to express our sincere thanks to Professor N. Kamiya of National Institute ofBasic Biology for continuous interest and valuable criticism throughout the period of this study,and also to Dr Y. Nonomura for profitable suggestions. We are indebted to Dr R. Kamiya fortaking the photographs of the optical diffraction pattern, and also to Professor Y. Tonomura

Endoplasmic factor in cytoplasmic streaming 135

and Dr A. Inou6 for the supply of rabbit heavy meromyosin. This work was partly supportedby grants-in-aid from the Mitsubishi Foundation and the Japanese Ministry of Education,Science and Culture.

REFERENCESALLEN, N. S. (1974). Endoplasmic filaments generate the motive force for rotational streaming

in Nitella. J. Cell Biol. 63, 270-287.ALLEN, N. S., ALLEN, R. D. & REINHART, T. E. (1976). Confirmed: the existence of abundant

endoplasmic filaments in Nitella. Biol. Bull. mar. biol. Lab., Woods Hole 154, 398.ALLEN, R. D. (1977)- In International Cell Biology (ed. B. R. Brinkley & K. R. Porter), pp. 403-

406. New York: Rockefeller University Press.BRADLEY, E. G. (1973). Microfilaments and cytoplasmic streaming: inhibition of streaming

with cytochalasin. J. Cell Set. 12, 327-343.CHEN, J. C. W. & KAMIYA, N. (1975). Localization of myosin in the internodal cell of Nitella

as suggested by differential treatment with iV-ethylmaleimide. Cell Struct. Funct. 1, 1-9.DEROSIER, D., MANDELKOW, E., SILLIMAN, A., TILNEY, L. G. & KANE, R. E. (1977). Structure

of actin-containing filaments from two types of non-muscle cells. J. molec. Biol. 113,6

GILLIS, J. M. & O'BRIEN, E. J. (1975). The effect of calcium ions on the structure of recon-stituted muscle thin filaments. J. violec. Biol. 99, 445-459.

HATANO, S. & NAKAJIMA, H. (1963). ATP content and ATP-dephosphorylating activity ofNitella. A. Rep. scient. Works, Fac. Set. Osaka Univ. n , 71-76.

HAYASHI, T. (1963). Role of the cortical gel layer in cytoplasmic streaming. In Primitive MotileSystems in Cell Biology (ed. R. D. Allen & N. Kamiya), pp. 19—29. New York: AcademicPress.

KAMITSUBO, E. (1966). Motile protoplasmic fibrils in cells of Characeae. II. Linear fibrillarstructure and its bearing on protoplasmic streaming. Proc. Japan Acad. 42, 640-643.

KAMITSUBO, E. (1972a). Motile protoplasmic fibrils in cells of Characeae. Protoplasma 74,53-7°-

KAMITSUBO, E. (19726). A 'window technique' for detailed observation of characean cyto-plasmic streaming. Expl Cell Res. 74, 613-616.

KAMIYA, N. (1959). Protoplasmic streaming. Protoplasmatologia 8, 3a.KAMIYA, N. & KURODA, K. (1956). Velocity distribution of the protoplasmic streaming in

Nitella cells. Bot. Mag., Tokyo 69, 544-554.KATO, T. & TONOMURA, Y. (1977). Identification of myosin in Nitella flexilis. J. Biochem.,

Tokyo 82, 777-782.KERSEY, Y. M., HEPLER, P. K., PALEVITZ, B. A. & WESSELLS, N. K. (1976). Polarity of actin

filaments in characean algae. Proc. natn. Acad. Set. U.S.A. 73, 165-167.LUFT, J. H. (1961). Improvements in epoxy resin embedding methods. J. biophys. biocliem.

Cytol. 9, 409-414.MOORE, P. B., HUXLEY, H. E. & DEROSIER, D. J. (1970). Three-dimensional reconstruction of

F-actin, thin filaments and decorated thin filaments. J. molec. Biol. 50, 279-295.NAGAI, R. & REBHUN, L. I. (1966). Cytoplasmic microfilaments in streaming Nitella cells.

J. Ultrastruct. Res. 14, 571-589.OHTSUKI, I. & WAKABAYASHI, T. (1972). Optical diffraction studies on the structure of troponin-

tropomyosin-actin paracrystals. J. Biochem., Tokyo 72, 369-377.PALEVITZ, B. A., ASH, J. F. & HEPLER, P. K. (1974). Actin in the green alga, Nitella. Proc. natn.

Acad. Sci. U.S.A. 71, 363-366.PALEVITZ, B. A. & HEPLER, P. K. (1975). Identification of actin in situ at the ectoplasm-endo-

plasm interface of Nitella. Microfilament-chloroplast association. J. Cell Biol. 65, 29-38.SHIMMEN, T. (1978). Dependency of cytoplasmic streaming on intracellular ATP and Mgt+

concentrations. Cell Struct. Funct. 3, 113-121.SPUDICH, J. A., HUXLEY, H. E. & FINCH, J. T. (1972). Regulation of skeletal muscle contrac-

tion. II. Structural studies of the interaction of the tropomyosin-troponin complex withactin. J. molec. Biol. 72, 619-632.

136 R. Nagai and T. Hayama

SPURR, A. R. (1969). A low-viscosity epoxy resin embedding medium for electron microscopy.J. Ultrastruct. Res. 26, 31-43.

TAZAWA, M. (1964). Studies on Nitella having artificial cell sap. 1. Replacement of the cell sapwith artificial solutions. PI. Cell Physiol., Tokyo 5, 33-43.

TAZAWA, M., KIKUYAMA, M. & SHIMMEN, T. (1976). Electric characteristics and cytoplasmicstreaming of Characeae cells lacking tonoplast. Cell Struct. Fund. 1, 165-176.

TILNEY, L. G. (1975a). Actin filaments in the acrosomal reaction of Limtdus sperm. Motiongenerated by alterations in the packing of the filaments. J. Cell Biol. 64, 289-310.

TILNEY, L. G. (19756). The role of actin in nonmuscle cell motility. In Molecules and CellMovement (ed. S. Inou£ & R. E. Stephens), pp. 339-388. New York: Raven Press.

WAKABAYASHI, T., HUXLEY, H. E., AMOS, L. A. & KLUG, A. (1975). Three-dimensional imagereconstruction of actin-tropomyosin complex and actin-tropomyosin-troponin T-troponinI complex. J. molec. Biol. 93, 477-497.

WILLIAMSON, R. E. (1974). Actin in the alga, Chara corallina. Nature, London. 248, 801-802.WILLIAMSON, R. E. (1975). Cytoplasmic streaming in Chara: A cell model activated by ATP

r.nd inhibited by cytochalasin B. J. Cell Set. 17, 655-668.

(Received 24 July 1978 — Revised 17 October 1978)