J. Cell Set. ai, 59-71 (1976)
Printed in Great Britain
THE CRYSTALLINE GLYCOPROTEIN CELL
WALL OF THE GREEN ALGA CHLOROGONIUM
ELONGATUM: A STRUCTURAL ANALYSIS
K. ROBERTS AND G. J. HILLS
John limes Institute, Colney Lane, Norwich NR4 7 UH,
Norfolk, England
SUMMARY
Members of the Chlamydomonaceae, mostly single-celled green algae, have been shown to
contain a crystalline glycoprotein cell wall component. Most of the species examined fall into
a class of algae whose walls have an identical crystalline unit cell. Chlorogonium elongatum
has been chosen as a representative of this class in order to investigate in more detail its cell
wall structure. The alga has a spindleshaped cell wall which retains its asymmetric shape on
isolation. Sections from walls fixed in the presence of tannic acid clearly reveal a regular
subunit monolayer, about 20 nm thick, within the wall. Sodium dodecylsulphate (SDS)
polyacrylamide gel electrophoresis shows the presence of at least 2 major glycoprotein species
in the wall. Negatively stained purified cell walls demonstrate the crystalline nature of the cell
wall. Optical diffraction of bright-field images and direct electron diffraction both give clear
diffraction patterns whose spacings extend out to 3 nm and fall on a reciprocal lattice whose
vectors describe a 2-dimensional unit cell within the wall 215 n m x 7 0 nm and an included
angle of 8o°. Lattice defects within the cell wall are revealed by both negative staining and
surface replication. Through-focal series were used to choose images with the optimal degree of
underfocus for image processing. Linear integration and optical filtering of such images gave
essentially the same result. A similar image was also obtained by computing the autocorrelation
function of the amplitudes in the electron-diffraction pattern and the optical-diffraction
pattern of the in-focus image. On the basis of these data a 2-dimensional model of the crystalline
cell wall layer is presented.
INTRODUCTION
Following the discovery that the unicellular green alga Chlamydomonas reinhardi
has a cell wall containing a complex crystalline glycoprotein lattice (Roberts, GurneySmith & Hills, 1972; Hills, Gurney-Smith & Roberts, 1973) a survey of numerous
related algae revealed that many other species within the Chlamydomonaceae also
have crystalline cell wall components (Roberts, 1974). These algae were classified
into 5 main classes, on the basis of the optical diffraction patterns of micrographs of
negatively stained cell walls. Chlamydomonas reinhardi and several multicellular
species (Volvox aureus, Eudorina elegans) were placed in Class I. The vast majority of
the remaining algae were placed in Class II, which was characterized by cell walls
having a 2-dimensional repeating morphological unit of sides 21-5 nm and 7-0 nm
and an angle of 8o°. Included in this class are species of Chlamydomonas, Chlorogonium,
Polytoma, Carteria, Haematococcus and Brachiomonas (Roberts, 1974). To study the
cell wall common to the algae of this class, Chlorogonium elongatum was chosen as the
60
K. Roberts and G. J. Hills
type organism. We present here a high-resolution study on the structure of this
particular cell wall based on electron microscopy and image-enhancement techniques,
together with data on its composition and assembly.
MATERIALS AND METHODS
Cultures of Chlorogonium elongatum (Dangeard) (Cambridge Culture Collection Strain 12/1)
were grown in sterile conditions in a defined salts medium (Sueoka, i960) supplemented with
0-2 % sodium acetate, at 25 °C, without aeration, and at 8000 lx continuous light. Walls were
prepared either by harvesting the cast off mother cell walls from the old growth medium, or by
Mickle disintegration and differential centrifugation, in the same way as has been described for
Chlamydomonas cell walls (Hills et al. 1973; Roberts, 1974). Light microscopy was performed
on a Zeiss photomicroscope II.
Cell walls were examined, using either 5 % aqueous ammonium molybdate or 5 % aqueous
sodium tungstate (adjusted to pH 6-8 with formic acid) as a negative stain, in a JEOL JEM
100B electron microscope. The parameters of the cell wall were measured by using the optical
transform of catalase crystals as a standard (Hills et al. 1973). Optical diffraction analysis
(Finch, Klug & Nermut, 1967) was carried out using the apparatus described by Home &
Markham (1972) and the apparatus used for linear integration was that described by Hitchborn
& Hills (1968). Cell wall surface carbon replicas were prepared by low angle (150) rotary
shadowing at 200 rev/min using 60/40 (w/w) gold/palladium alloy followed by carbon coating
(Hills et al. 1973). Cells were prepared for sectioning using the fixation method of Franke, Krien
& Malcolm Brown (1969), or a modified version (Tilney et al. 1973) of the tannic acid fixation
method of Mitzuhira & Futeasaku (1972). Selected area electron-diffraction patterns were
obtained using a cool beam Wehnelt cylinder in the JEM 100B microscope at 80 kV, with the
condenser set for the highest crossover point. For computer reconstruction of the autocorrelation function optical diffraction patterns of in-focus or slightly under-focus micrographs
were recorded on Uford R20 plates. The optical densities of the diffraction maxima were
measured on a Joyce Loebl microdensitometer. The relative positions and square roots of the
densities of the maxima were Fourier transformed on either an IBM 1130 or an IBM 370
computer using a standard fast Fourier transform programme. The output modulus was
squared and displayed on either an Optronics filmwriter or on a Tektronix oscilloscope. Purified
cell walls were examined by sodium dodecyl sulphate (SDS) polyacrylamide gel electrophoresis using the method previously described (Roberts et al. 1972). Digestion of wall samples
was carried out in 1 % SDS + i % mercaptoethanol in o-oi M phosphate buffer (pH 7 0 ) at
100 CC for 1 min. Gels were stained in Coomassie blue to reveal protein bands or by the modified
periodic acid/Schiff method to reveal sugar-containing bands (Roberts, 1974).
Fig. 1. Phase-contrast light micrograph of the living cell of Chlorogonium elongatum.
The 2 pointed ends of the cell wall reveal the gap between the wall and the cell body
itself, x 1500.
Fig. 2. Purified cell walls obtained from an old culture which contained mother
cell walls cast off in the course of cell division, x 1500.
Fig. 3. Low-power electron micrograph of the apical region of the cell in negative
stain. This shows the elongated cytoplasmic extension (ce) which swells apically (a)
to form a region from which the 2 flagella (/) emerge and pass through the cell wall.
(See Fig. 7.) x 8000.
Fig. 4. Electron micrograph of a region of purified cell wall negatively stained in
ammonium molybdate and showing the ordered crystalline lattice that is found
within the wall. Protein here is indicated white. X 450000.
The crystalline cell wall of Chlorogonium
61
f
K. Roberts and G. 7. Hills
dc
10 11
12
The crystalline cell wall of Chlorogonium
63
RESULTS AND DISCUSSION
Light microscopy
Chlorogonium elongatum is a highly elongated spindleshaped biflagellate unicellular
green alga about 20-25 Z"11 m length (Fig. 1). The protoplast within the cell wall is
more nearly round, leaving a large gap between the two pointed ends of the cell wall
and the plasmalemma (Fig. 1). A long cytoplasmic extension protrudes from one end
of the protoplast and has an apical swelling from which the 2 flagella arise, passing
out through 2 collars in the cell wall (Figs. 1, 3, 7). During the life-cycle the daughter
cells are released from the mother cell wall which splits, usually at one end, at right
angles to the long axis of the cell. The splitting of the mother cell wall, at least in
Chlamydomonas, is mediated by a specific factor released by the cells at this time
(Schlosser, 1966; Mihara & Hase, 1975). Cast off mother cell walls, or cell walls
prepared by gentle Mickle disintegration, retain their original shape (Fig. 2).
Negative staining
Cell walls, prepared either by harvesting cast off mother cell walls or by Mickle
disintegration, reveal, when negatively stained, a highly ordered crystalline component
(Fig. 4). This crystalline wall component is destroyed by such negative stains as
uranyl acetate or potassium phosphotungstate, but is preserved by either ammonium
molybdate or sodium tungstate. Very large areas of dislocation-free cell wall may be
seen in negatively stained preparations (up to several /tm2) and in general the main
longitudinal periodicities run more or less parallel to the long axis of the cell wall.
Fig. 5. Section of the cell wall from a specimen fixed by the method of Franke et al.
(1969). The ordered nature of the outer layer of the wall can be seen more clearly
by sighting along the length of the wall, x 105000.
Fig. 6. The periodic nature of the wall is revealed more clearly in this section of a
wall fixed in a tannic acid-glutaraldehyde fixative, x 105000.
Fig. 7. The region of the cell wall through which the flagellum passes, cw, cell wall;
/ , flagellum. x 105000.
Fig. 8. Part of a recently divided cell showing 2 naked daughter protoplasts (j>),
the old mother cell wall (mw), and material which appears to be the new cell wall
precursor (m). x 15000.
Fig. 9. This shows a later stage of cell division than Fig. 8 in which a new cell wall
(civ) has begun to be deposited on the naked daughter protoplast (p). The old
mother cell wall is still present (mw). x 60000.
Figs, io, 11. Polyacrylamide SDS gel electrophoresis of the cell wall glycoproteins
stained with Coomassie blue for proteins in Fig. 10 and with periodic acid/Schiff
reagent for polysaccharide material in Fig. 11.
Fig. 12. A disclination in the cell wall crystalline layer revealed by negative staining.
dc, disclination core, x 100 000.
Fig. 13. A rotary shadowed carbon replica of the posterior apical region of the cell
wall. The numerous lattice defects at the apex are clearly seen. Arrows indicate
lattice direction, x 30000.
64
K. Roberts and G. J. Hills
Similar cell wall periodicities have been seen in freeze-fractured replicas of Chlamydomonas moewusii (Nakamura, Bray, Costerton & Wagenaar, 1973), a species also
belonging to Class II (Roberts, 1974).
Sectioning
Cells have been examined in thin sections of material fixed either in a mixture of
glutaraldehyde and osmium tetroxide or in glutaraldehyde containing tannic acid.
The appearance of the cell wall in each case is distinctive. Fig. 5 shows the wall when
fixed using the method of Franke et al. (1969). In some respect the wall looks similar
to that of Chlamydomonas reinhardi and can be seen to comprise an amorphous inner
layer of variable thickness and an ordered outer layer of about 20 nm thickness. In
Fig. 5 the inner amorphous layer is substantial, whereas the presence of this layer
following tannic acid treatment (Fig. 6) is variable. A thin amorphous outer wall layer
of variable thickness is also seen in Fig. 6. The distinctive difference in appearance,
however, is that in cells fixed in the presence of tannic acid the wall can clearly be
seen to be constructed of periodic electron-dense elements (Fig. 6). These are about
10 nm across and are spaced at 17-20 nm apart, depending on the angle of the section.
Similar periodicities have also been seen in Chlamydomonas reinhardi cell walls
following tannic acid treatment (Roberts, Phillips & Hills, 1975). This would seem to
indicate that the same basic construction pattern, i.e. an inner amorphous layer on the
outside of which is a crystalline layer of sub-units, is used by the various algae
originally described as having a crystalline cell wall component (Roberts, 1974)- As
in Chlamydomonas reinhardi there are 2 specialized regions of the cell wall through
which the flagella emerge. In Chlorogonium however these 2 collars are not such
markedly differentiated organelles as in Chlamydomonas (Roberts et al. 1975) and
appear in fact to be constructed by a simple infolding of the cell wall itself (Fig. 7).
The basal bodies of the flagella reside in a small volume of cytoplasm which forms an
apical swelling on a thin cytoplasmic extension from the main cell protoplast (Figs. 1,
3, 7). This cytoplasmic extension is about 10 fim long and contains numerous microtubules. Thin sectioning has also provided us with some evidence concerning cell
wall biogenesis. As in Chlamydomonas reinhardi, cell division involves the production
of several daughter protoplasts within the mother cell wall. A new cell wall is then
rapidly formed around each protoplast and the mother cell wall ruptures releasing
the daughter cells. The earliest stages of cell wall deposition that have been seen
indicate large amounts of diffuse material excreted within the space between the
daughter protoplasts and the mother cell wall (Fig. 8). This material coalesces to
form the amorphous inner layer of the complete cell wall, which in turn acts as the
site for the deposition of the outer crystalline wall component (Fig. 9). This process
is similar to that observed in vitro with Chlamydomonas reinhardi (Hills, Phillips, Gay
& Roberts, 1975). This similarity is not surprising as the 2 wall types both appear to
be constructed from high molecular weight glycoprotein components. These have
been resolved by SDS polyacrylamide gel electrophoresis followed by staining for
protein with Coomassie blue (Fig. 10) and for carbohydrate with the periodic acid/
Schiff method (Fig. 11). It should be pointed out that the other algae exhibiting a
The crystalline cell wall of Chlorogonium
65
Class II type crystalline wall (Roberts, 1974) have similar, but not identical, glycoproteins in their cell wall.
Defects in the crystalline cell wall layer
The overall shape of the Chlorogonium cell is dictated by the cell wall and not by the
enclosed protoplast (Fig. 1). The spindleshaped cell wall retains its shape after isolation (Fig. 2). The question arises of how the regular 2-dimensional crystalline component of the wall, revealed by negative staining (Fig. 4), relates to the 3-dimensional
spindle shape of the cell wall. As pointed out earlier (Roberts, 1974) the crystalline
lattice displays various local defects. The arrangement of these defects has been
clarified by examining rotary-shadowed carbon replicas of isolated cell walls. The
main longitudinal lattice periodicities are aligned with the long axis of the cell wall.
In any closed surface crystal however there must be disclinations and their rotations
must sum to 477 radians. In this case there are numerous partial disclinations (Fig. 12)
concentrated at the 2 apices. These are associated with numerous tilt boundaries
(Fig. 13) which are clearly revealed on the replicas. As suggested by Harris & Scriven
(1970) these dislocations in the cell wall probably act as growth centres for the addition
of new subunits. However, because the shape of the cell wall is determined by the
inner wall layer and not by the crystalline layer (by analogy with Chlamydomonas:
Hills et al. 1975) the actual pattern of dislocation does not influence the final shape of
the cell. Moreover, as there is some evidence that the wall glycoproteins are continually
being synthesized and excreted throughout the cell cycle (Lang & Chrispeels, 1976),
it follows that subunits are constantly being exchanged at the cell surface and that
therefore the pattern of dislocations is not static but mirrors the dynamic events
during the cell cycle.
Diffraction studies
The cell wall of Chlorogonium, when viewed by negative-staining techniques,
shows a highly ordered crystalline structure. Areas of dislocation-free cell wall are
easily obtained which display several thousand regularly repeating unit cells. Such
2-dimensional arrays are ideally suited for analysis by diffraction techniques and
image enhancement and have enabled us to describe the envelope of the glycoproteins
mapped out within the unit cell by the negative stain. It should be pointed out that the
small variations within individual unit cells, and the degree of disorder they display
set the limits to the maximum spatial frequencies we can usefully recover from the
image. In this case the maximum spatial frequency that we can detect regularly above
background noise, in both electron- and optical-diffraction studies, is about 3-0 nm.
This limit, characteristic of negatively stained specimens, is not surprising as we are
dealing with several high molecular weight glycoproteins within a relatively large
unit cell, and the most we can hope to achieve is to establish their gross morphology
and their packing geometry in 2 dimensions. A full 3-dimensional reconstruction of
the unit cell will depend on tilt series and computer image processing, experiments on
which are are in progress.
Electron diffraction patterns of the negatively stained cell wall were recorded, and
66
K. Roberts and G. J. Hills
10
1.0
0,0
1.0
2.0
3.0
4.0
U
The crystalline cell wall of Chlorogonium
67
a long and short exposure pair of the same specimen are shown in Fig. 14. Taken
together, both the high and low spatial frequency reflexions are clearly visible and
can be placed on the reciprocal lattice shown in Fig. 15. From this lattice we can deduce
that the unit cell forms a primitive two-dimensional Bravais lattice that is oblique
(space group p i ; a — "721'5 nm"1, b = ^ y o run"1, y = 80°). The intensity of the
strongest reflexion in the electron-diffraction pattern as a fraction of the intensity of
the undeviated beam is ~5 x io~3. This means that we are dealing with a weakly
scattering object, and that the intensities of the reflexions provide a meaningful
measure of the structure factor amplitudes. Unlike the optical-diffraction patterns
the electron-diffraction pattern is unaffected by the contrast transfer function of the
microscope.
For optical image processing it is necessary to ensure that the micrograph chosen
has all spatial frequencies up to the limit chosen (i.e. 3 nm) transferred with the same
contrast into the image. For this reason we have regularly taken focal series, at
300-nm steps, of the negatively stained cell walls, and used the optical transforms of
such images (Figs. 16-27) t 0 assess both the quality of the image and position of the
phase-contrast transfer function (Erickson & Klug, 1971). Micrographs chosen for
processing were those with an under-focus of 300-600 nm (see Figs. 20, 21). It is
worth pointing out that many people would choose Fig. 16 as being aesthetically the
better image, although it can be seen from the optical transform that the 4,2 reflexion,
for example, is missing, being on the zero of the first transfer function and hence not
contributing to the final image.
It is of interest that the relative intensities and positions of the diffraction maxima
in the electron-diffraction pattern are almost identical to those found in the optical
transform of the in-focus (or slightly under-focused) image and are significantly
different from the transforms of the under- and over-focused images shown in Figs. 17
and 27.
An unusual feature of the optical transform which has never been seen in the
electron-diffraction pattern is the presence of regular extra spots which, although
seemingly related to the major spots, do not fall on the same reciprocal lattice. These
are most clearly seen in Fig. 19. We do not yet know the origin of these spots and in
both our optical filtering and computer image processing we have ignored them.
Fig. 14. Low-angle selected area electron-diffraction pattern obtained from a single
piece of cell negatively stained in ammonium molybdate. The top half has been
exposed to reveal the diffraction maxima on the second layer line.
Fig. 15. The reciprocal lattice on which the electron-diffraction and optical-diffraction
maxima can be placed. Only half the diffraction pattern is shown. The lattice vectors
are a = V 2I 'S nm"1, b = 1/y-o nm-1, y = 8o°. The figures above the indexed
points are the square roots of the measured intensities (the amplitudes) of the optical
diffraction pattern used to produce the image in Fig. 31.
Figs. 16-27. Part of a through-focal series of a negatively stained cell wall ( x 105000)
together with their respective optical transforms. Fig. 22 represents the close-to-focus
image. Figs. 20, 18 and 16 are taken with 600-nm underfocus steps respectively and
Figs. 24 and 26 are taken with 600-nm overfocus steps respectively.
5-2
68
K. Roberts and G. J. Hills
1
The crystalline cell wall of Chlorogonium
69
Image processing
Knowing the symmetry and lattice vectors from optical and electron diffraction,
we have processed selected micrographs by both translational superposition and optical
filtering. Figs. 28-30 show respectively a selected area of cell wall, an optically filtered
image, and a linearly integrated image, all at the same magnification. The optically
filtered image and the linearly integrated image show remarkable agreement in their
enhancement of the structure of the unit cell. Figs. 28-31 have been printed with an
opposite contrast to the negatively stained image shown in Fig. 20, so that organic
matter is shown black. The main structural features are the predominant longitudinal
periodicities composed of globular subunits every 7-0 nm. Each globular subunit is
joined to its adjacent neighbours by an oblique link and also shows some evidence of
being composed of 2 smaller subunits. More indistinct material is found between
the main rows of periodicity.
On the grounds that the transform of the in-focus image is almost identical to the
electron-diffraction pattern, which is a function of the density distribution in the
object, and in the absence of facilities for full Fourier reconstruction, we computed
the autocorrelation function by Fourier transforming the amplitudes (the positive
square root of the measured intensities) of the reflexions in the optical diffraction
pattern (shown in Fig. 15). The centre spot was given an amplitude of 40, which is
just larger than the sum of the amplitudes in the reflexions. The output modulus was
squared and displayed as a set of intensity levels as shown in Fig. 31. The similarity
of this output to the optically filtered image (Fig. 29) is immediately apparent. An
extra feature which is seen on the autocorrelation function is the elaborate ladderlike
structure between the main lines of periodicity. That this is also present, but less
distinct, in Fig. 29 may be seen by sighting up the page along the main lines of periodicity (Figs. 29, 31). The autocorrelation function and optically filtered images
resemble each other in this case, presumably because the majority of the lattice peaks
have the same phase. This situation could arise because the main scattering units are
small compared to the lattice dimensions. The Fourier transform is the product of
their shape and the lattice transform. The shape transform is a broad thing because
Fig. 28. Portion of a negatively stained single piece of cell wall, x 1000000.
Fig. 29. Fig. 28 which has been optically filtered using a mask letting through those
spots of the diffraction pattern shown in Fig. 15. x 1000000.
Fig. 30. Fig. 28 which has had the periodic structures present reinforced by linear
integration (15 unit cells integrated along the main line of periodicity), x 1000 000.
Fig. 31. The autocorrelation function obtained by computer fast Fourier transforming
the amplitudes of the optical diffraction pattern maxima, x 1000 000. A single primitive unit cell is outlined in common with a similar unit cell shown in Figs. 28-30.
Fig. 32. Rotary shadowed replica of the cell wall. Inset is the optical diffraction
pattern derived from it. x 100000.
Fig. 33. An identical area to Fig. 32 which has been optically filtered using a mask
containing the 8 diffraction maxima shown in Fig. 32 inset, x 1000000.
70
K. Roberts and G. J. Hills
the units are small and consequently a lot of the lattice points fall within the central
maximum where the phase is constant.
The images discussed above are all related to 2-dimensional projections of the
3-dimensional cell wall and one would like to know how the various components
found within the unit cell are arranged in space. Clues from thin sections have already
been discussed, and in the absence of a full 3-dimensional Fourier analysis we have
tried to get some preliminary information from surface replicas of the cell wall.
Fig. 32 shows a typical rotary-shadowed carbon replica together with its opticaldiffraction patterns. The spots fall on a reciprocal lattice whose equatorial vector is
identical to that of the reciprocal lattice of the negatively stained cell wall, but whose
other vector is half the original value. This suggests therefore that only every other
one of the 7-o-nm repeats is being imaged in the surface replica. Optical filtering of
this image is shown in Fig. 33 and confirms that the unit cell is now 21-5 nm by
14-0 nm. Our conclusion from this is that the surface of the cell wall is corrugated at
an angle of 80° to the main periodicities, alternate 7-nm spaced subunits being slightly
raised. Work is now in progress on constructing a full 3-dimensional model of the
components within the unit cell.
We wish to acknowledge Dr M. W. Johnson of this Institute and Dr W. O. Saxton of the
Cavendish Laboratory, Cambridge, for help in establishing our computer image processing
system, and for producing the Optronics output shown in Fig. 31. We thank Dr N. Unwin of
the M.R.C. Laboratory of Molecular Biology, Cambridge, for critical comments on the final
manuscript.
REFERENCES
H. P. & KLUG, A. (1971). Measurement and compensation of defocusing and
aberrations by Fourier processing of electron micrographs. Phil. Trans. R. Soc. Ser. B 261,
ERICKSON,
105-118.
J. T., KLUG, A. & NERMUT, M. V. (1967). The structure of the macromolecular units
on the cell walls of Bacillus polymyxa. J. Cell Set. 3, 587-590.
FRANKE, W. W., KRIEN, S. & MALCOLM BROWN, R. (1969). Simultaneous glutaraldehydeosmium tetroxide fixation with post osmication. Histochemie 19, 162-164.
HARRIS, W. F. & SCRIVEN, L. E. (1970). Function of dislocations in cell walls and membranes.
Nature, Land. zz8, 827-829.
HILLS, G. J., GUHNEY-SMITH, M. & ROBERTS, K. (1973). Structure, composition and morphogenesis of the cell wall of Chlamydomonas reinhardi. II. Electron microscopy and optical
diffraction analysis. ,7. Ultrastruct. Res. 43, 179—192.
HILLS, G. J., PHILLIPS, J. M., GAY, M. R. & ROBERTS, K. (1975). Self assembly of a plant cell
wall in vitro. J. molec. Biol. 96, 431-441.
HITCHBORN, J. H. & HILLS, G. J. (1968). A study of tubes infected with a strain of turnip
yellow mosaic virus. Virology 35, 50-70.
HORNE, R. W. & MARKHAM, R. (1972). Application of optical diffraction and image reconstruction techniques to electron micrographs. In Practical Methods in Electron Microscopy,
vol. 1 (ed. A. M. Glauert), pp. 327-434. Amsterdam: North-Holland/Elsevier.
LANG, W. C. & CHRISPEELS, M. J. (1976). Biosynthesis of cell wall proteins during the vegetative
cell cycle of Chlamydomonas reinhardii. Planta (in Press).
MIHARA, S. & HASE, E. (1975). Studies on the vegetative life cycle of Chlamydomonas reinhardii
Dangeard in synchronous culture. III. Some notes on the process of zoospore liberation.
PI. Cell Physiol. 16, 371-375.
MITZUHIRA, V. & FUTAESAKU, Y. (1972). New fixation for biological membranes using tannic
acid. Ada Histochem. Cytochem. 5, 233-236.
FINCH,
The crystalline cell wall of Chlorogonium
71
K., BRAY, D. F., COSTERTON, J. W. & WAGENAAR, E. B. (1973). The eyespot of
Chlamydomonas eugametos: a freeze etch study. Can.J. Bot. 51, 817-819.
ROBERTS, K. (1974). Crystalline glycoprotein cell walls of algae: their structure, composition
and assembly. Phil. Trans. R. Soc. Ser. B 368, 129-146.
ROBERTS, K., GURNEY-SMITH, M. & HILLS, G. J. (1972). Structure, composition and morphogenesis of the cell wall of Chlamydomonas reinhardi. I. Ultrastructure and preliminary
chemical analysis. .7. Ultrastruct. Res. 40, 599-613.
ROBERTS, K., PHILLIPS, J. M. & HILLS, G. J. (197s). Structure, composition and morphogenesis
of the cell wall of Chlamydomonas reinhardi. IV. The flagellar collar. Micron 5, 341-357.
SCHLOSSER, U. (1966). Enzymatisch gesteuerte Freisetzung von Zoosporen bei Chlamydomonas
reinhardii Dangeard in Synchronkultur. Arch. Mikrobiol. 54, 129-159.
SUEOKA, N. (i960). Mitotic replication of deoxyribonucleic acid in Chlamydomonas reinhardi.
Proc. natn. Acad. Sci. U.S.A. 46, 83-91.
NAKAMURA,
TILNEY, L. G., BRYAN, L., BUSH, D. J., FUJIWARA, K., MOOSEKER, M. S., MURPHY, D. B. &
SNYDER, D. H. (1973). Microtubules: evidence for 13 pro to filaments. J. Cell Biol. 59, 267-
275(Received 31 October 1975)