y. Cdi Sci. 36,97-107 (1979) 97Printed in Great Britain © Company of Biologists Limited 1970

MARGINAL BANDS IN CAMEL

ERYTHROCYTES

WILLIAM D. COHENDepartment of Biological Sciences, Hunter College of the City University of New York,New York, New York 10021, U.S.A.

AND NORA BARCLAY TERWILLIGER*Department of Anatomy, University of Wisconsin, Madison, Wisconsin 53706, U.S.A.

SUMMARYThe elliptical, enucleate erythrocytes of camels have been examined for the presence of

marginal bands and their constituent microtubules. Lysis of erythrocytes under microtubule-stabilizing conditions readily revealed marginal bands in at least 3 % of the cells, as observedby phase-contrast and darkfield light microscopy. Microtubules plus a marginal band-encompassing network of material are visible in lysed cell whole mounts with transmissionelectron microscopy. Marginal band microtubules are also evident in electron micrographs ofthin-sectioned camel erythrocytes identifiable as reticulocytes on the basis of submaximalelectron density (reduced haemoglobin iron content) and presence of polysomes. The resultssuggest that marginal bands may be involved in morphogenesis of camel erythrocytes but arenot required for maintenance of their ellipticity after cells are fully differentiated.

INTRODUCTION

The marginal band (MB) is a discrete circumferential bundle of microtubules witha probable role in alteration and perhaps maintenance of cell shape. MBs occur in theelliptical, nucleated erythrocytes of non-mammalian vertebrates (Dehler, 1895;Meves, 1911; Fawcett, 1959), in the thrombocytes of both mammalian and non-mammalian vertebrates (Fawcett & Witebsky, 1964; Behnke, 1965; Sandborn,LeBuis & Bois, 1966), and in blood cells of certain invertebrates (Cohen, Nemhauser& Jaeger, 1977). MBs have not been observed in the mature anucleate, diskoidalferythrocytes of mammals.

Among the mammals, members of the family Camelidae (camels, vicunas, guanacos,llamas, alpacas) are unique in that their erythrocytes, though anucleate, are elliptical(Andrew, 1965). The question thus arises as to whether MBs play a role in cell shapegeneration and/or maintenance in these species. Barclay (1966), in an abstract,reported the occurrence of MB microtubules in thin-sectioned camel erythrocytes.Recently, however, Goniakowska-Witalinska & Witaliriski (1976) were unable to

• Present address: Department of Biology, University of Oregon, Oregon Institute ofMarine Biology, Charleston, Oregon 97420, U.S.A.

f Terminology in the literature is often confusing with respect to erythrocyte shape,referring to 'round disks', 'elliptical disks', and 'disk-shaped' elliptical cells. The terms'disk' and 'diskoidaF should be reserved for cells which are flattened and circular; a diskcannot be elliptical.

98 W. D. Cohen and N. B. TerwiUiger

verify Barclay's observation after examining more than 2000 thin-sectioned camel and

llama cells. In order to help resolve this issue a somewhat different approach appeared

desirable, one which would avoid potential fixation and sampling problems associated

with thin-sectioning. In the work reported here, camel erythrocytes have been lysed in

a microtubule-stabilizing medium, permitting direct visualization of MBs in large

numbers of lysed cells by means of phase-contrast and darkfield light microscopy, with

subsequent TEM on whole mounts (Cohen et al. 1977). These results thus support

Barclay's (1966) report. Thin-section ultrastructural data on intact cells, upon which

Barclay's abstract was based, are presented for further documentation of the presence

of MB microtubules in camel erythrocytes.

MATERIALS AND METHODS

Lysed erythrocytes

Observations were made on blood samples from 2 camels (Camelus dromedaritu), 2 guanacos(Lama guanicoe), and 1 llama (Lama glama). Whole blood was drawn by syringe from thejugular veins of donor animals at the Bronx Zoological Park by Dr Emil P. Dolensek of theNew York Zoological Society. The blood was collected in either citrated or heparinized tubesand kept warm in the hand until samples were lysed (within 10-30 min of collection).

In order to observe and photograph individual erythrocytes, whole blood was diluted approxi-mately 1:100 with mammalian Ringer solution (Krebs formula, Cavanaugh, 1975). The lyticmedium, based upon microtubule-polymerization conditions (Weisenberg, 1972; Rebhun,Rosenbaum, LeFebre & Smith, 1974), consisted of 100 mM PIPES (piperazine-iV-W-bis[2-£thanesulphonic acid]), 1 mM MgCl,, 5 mM EGTA, 10 mM TAME (£-tosyl argininemethylester HC1), and 0-4% (w/v) Triton X-100 brought to pH 6-8 with KOH. The TAMEprotects against proteolysis, while the EGTA reduces Ca1+ below polymerization-inhibitinglevels (Rebhun et al. 1974).

Whole blood samples were lysed at a ratio of 1 vol. blood to 9 vol. medium (generallyo-i m l+ 0-9 ml). Some samples were diluted further with the same medium to facilitate countsof MBs versus cell number in non-overlapping fields in phase contrast. Light-microscopicobservations were made in a Zeiss phase-contrast microscope. To achieve a darkfield effect,the 100 x condenser annulus was used in conjunction with the 16 x phase objective. Slidesand coverslips were detergent-cleaned and washed to remove oily films, and petroleum jellywas used to seal coverslips for long-term observation. For TEM of whole mounts one drop oflysed cell suspension was placed on each Formvar-coated grid for 2 min, drawn off with filterpaper, and replaced by a drop of deionized, distilled water. The water was immediately drawnoff and replaced by a drop of 2 % aqueous uranyl acetate for 15-30 8. After removal of excessstain, the grid was allowed to air dry. Grids were examined in an Hitachi HS-8 transmissionelectron microscope operating at 50 kV.

Thin sections

Whole blood was obtained as above from the jugular veins of 2 mature camels at the VilasPark Zoo, Madison, Wisconsin. The blood was collected in tubes containing acid-citrate-dextrose anticoagulant and fixed within 30 min in 5 % glutaraldehyde, 0-075 M phosphatebuffer, pH 7-4, for 30 min. The cells were rinsed in 0075 M phosphate buffer and then post-fixed in similarly buffered 1 % osmium tetroxide for 1 h. The cells were centrifuged into a pelletbefore each solution change and then resuspended. After postfixation, the pellets were cut intosmall pieces, dehydrated in an ethanol series, and embedded in Epon 812. Thin sections weretransferred to carbon-reinforced parlodion-coated grids and stained with uranyl acetatefollowed by lead hydroxide. The grids were examined in an RCA EMU 3E electron microscope.

Marginal bands in camel erythrocytes 99

RESULTS

Lysed cells

Large numbers of elliptical MBs were visible under phase-contrast (Fig. 1) inpreparations of lysed erythrocytes from a young camel (10 months). The majority oflysed cells, however, did not display MBs but appeared as partially collapsed, roughlyelliptical ghosts. The MBs were of approximately the same axial dimensions as intactcells (Figs. 2-5). Many MBs had dense, thickened regions along their inner surface,often but not always near the ends of the ellipse (Figs. 4, 5). Some MBs were twistedabout their long axis so as to form 'figure-eights' in certain views (Fig. 6A, B). Withinthe regions circumscribed by most MBs there was no material visible, giving theinitial impression that the MBs were completely free of other cellular material. Largemultilobed nuclei, presumably derived from leucocytes, formed small clumps in themedium (Fig. 1). Counts made in non-overlapping fields chosen at random showedthat there was approximately one clearly visible MB per 35 cells for this animal (totalcount: 19 MBs, 664 cells). This is probably a minimum estimate as some relativelythin MBs may not be observable in phase-contrast.

To test the possibility that this concentration of MBs might be typical only of younganimals, comparison was made with preparations of erythrocytes from a much oldercamel (27 years; Fig. 7). In general there was little difference, with approximately oneclearly visible MB per 33 cells (total count: 22 MBs, 717 cells). On these MBs,however, the dense, thickened regions seen in the lysed cells of the younger camel werenot observed. The MBs varied to some extent in size and in apparent thickness withina given preparation, and on rare occasions large MBs, almost twice the length of mostothers (major axis of ellipse), were noted (Fig. 8). In a single instance an elliptical MBwas observed surrounding an elliptical nucleus. This nucleus and MB were apparentlyattached to one another by material not visible in phase-contrast (Fig. 9), as verified byobserving their conjoined movement in medium flowing under the coverslip. Severalother MBs contained within their boundaries large granules or droplets (Fig. 10)which remained trapped in position as material moved in a flow.

For comparative purposes, observations were also made on llama and guanacoerythrocytes. The llama blood contained many cells with readily identifiable MBs, butwith fewer MBs per total number of cells than in the camel blood. Blood from oneguanaco also had many MB-containing cells, though again in lower percentage thanthe camels, while in that of another guanaco scarcely any MBs were observed.

TEM observation of uranyl acetate-stained whole mounts confirmed the presenceof MBs in some lysed camel cells and their absence in the majority. TEM revealedthat the MBs were not really cell-free structures as suggested by phase-contrastmicroscopy, but rather were associated with a network of stained material (Fig. 12).This material appeared to extend just beyond the boundary of the MB, as if the MBwere trapped within a collapsed sac (Fig. 13). In some lysed cells there were regionsin which the material coating the MB was less dense and the MB somewhat flattened,generally near or within the highly curved ends of the ellipse. In such cases it was

100 W. D. Cohen and N. B. Terwilliger

Fig. i. Low-magnification view of lysed cell preparation from a young camel. Manyelliptical MBs are present (arrows), but most cells lack MBs and appear as smallerghosts. Some MBs (fainter, bright contrast) are slightly out of plane of focus.Clumped nuclei («) are apparent remnants of leucocytes. Phase-contrast, x 510.

Fig. 2. Camel erythrocytes in mammalian Ringer's solution (dilution of whole bloodrequired for observation of individual cells). Typical cells are flattened, anucleate,and elliptical, with the long axis of the ellipse in the range of 7-9 micrometres.Phase-contrast, x 1420.

Fig. 3. Lysed camel cell preparation at same magnification as Fig. 2, showing MBwith axial dimensions similar to intact cells, and smaller ghosts lacking MBs (arrows).Phase-contrast, x 1420 (mag. bar as in Fig. 2).

Figs. 4, 5. Examples of MBs with associated dense, thickened regions typical ofpreparations from this young camel. Different MBs contained from 1 to 4 suchenlarged areas always protruding inward. Most ghosts lacking MBs appear empty(arrow) but some contain dense granules (g). Phase-contrast, x 1420 (mag. bar as inFig. 2).

Marginal bands in camel erythrocytes 101

possible to resolve the longitudinal array of microtubules constituting the MB bundle(Figs. 14, 15). Where individual microtubules could be distinguished, their diameterwas in the range of 22-0-24-0 nm (Fig. 14). Material comprising those lysed cellswhich lacked MBs was qualitatively different from that encompassing MBs in that ittypically exhibited denser, more uniform staining (Fig. 16). Thickened regions wereobserved along many MBs of the young camel's erythrocytes in the uranyl acetate-stained whole mounts (Fig. 12), probably corresponding to those observed underphase contrast in the same preparation.

Thin sections

Marginal band microtubules, 22-5-25-0 nm in diameter, are clearly visible in thinsections of some camel blood cells (Fig. 17). A clear zone or halo of lighter densitythan the rest of the cytoplasmic matrix surrounds each microtubule. Longitudinalsections of microtubules indicate that the zone of lighter density extends along thelength of the microtubule. The combined diameter of the microtubule plus the clearzone as measured in cross-section is about 3i-5-35-o nm. The marginal band micro-tubules were readily apparent in cells identified as circulating reticulocytes, which areless electron-dense than the mature erythrocytes due to a lower haemoglobin concen-tration. The number of microtubules seen in cross-section generally ranged from 7-22,although clusters of as many as 35-55 were occasionally seen (Fig. 17). There did notseem to be any correlation between the number of polysomes and the number of MBmicrotubules visible in reticulocytes; cells with very few polysomes had as manymicrotubules as did less mature reticulocytes with many polysomes. MB micro-tubules were rarely observed in fully differentiated erythrocytes, and the contrastbetween microtubule and background haemoglobin was less striking than in theyounger cells, perhaps due to a diminution of the halo or clear zone around themicrotubules.

Fig. 6. One of many camel MBs twisted into 'figure-8' shape, A, FOCUS at uppersurface of MB, showing cross-over from upper right to lower left; B, focus at lowersurface, with cross-over reversed. Phase-contrast, x 1420 (mag. bar as in Fig. 2).Fig. 7. Low-magnification view of lysed cell preparation from older camel, as seen indaikfield. MBs appear as bright ellipses. Note bright granules in some of the ghostsin background, x 510 (mag. bar as in Fig. 1).Fig. 8. One of several extra large MBs observed in camel blood preparations; com-pare with typical MB nearby (arrow). Phase-contrast, x 960.Fig. 9. An elliptical nucleus (n) surrounded by large, elliptical MB. A typical camel MB(arrow) and presumed leucocyte nuclei (In) are present. Phase-contrast, x 960 (mag.bar as in Fig. 8).Fig. 10. Lysed camel erythrocytes with MBs; one contains a dense droplet, apparentlytrapped between transparent sheets of material traversing MB. Phase-contrast, x 960(mag. bar as in Fig. 8).Fig. 11. MB of lysed guanaco erythrocyte. Phase-contrast, x 960 (mag. bar as inFig 8).

102 W. D. Cohen and N. B. Terwilliger

b

16

Marginal bands in camel erythrocytes 103

DISCUSSION

Direct light-microscopic observation of camel erythrocytes lysed under micro-tubule-stabilizing conditions shows that MBs are present in at least 3% of the cells.Correspondingly, MB microtubules are consistently seen in thin sections of those redblood cells identified as reticulocytes, but only rarely in those which appear fullymature as judged by high cytoplasmic electron density due to haemoglobin iron andby absence of ribosomes or other organelles. The possibility exists that the high back-ground density, and/or lack of a clear zone, obscures the presence of MB micro-tubules in mature cells, and that all of the elliptical erythrocytes actually containMBs. However, one would then have to assume that a majority of MBs solubilize inthe lytic medium. This is unlikely, as experiments with a wide range of vertebratesshow that the same medium stabilizes MBs in all of the erythrocytes in a given prep-aration (Cohen et al. 1977; Cohen, 1978).

The observed occurrence of MBs in both lysed cell preparations and thin sectionsof intact cells is consistent with a role of MBs in establishing the elliptical morphologyduring differentiation, with possible subsequent loss of MBs as the cells mature andage. This is supported by the observation that all of the lysed or thin-sectionedcells are initially elliptical whether or not MBs are visible, suggesting that the MB isnot required for maintenance of ellipticity in this system. Behnke (1970) also reportedconditions under which ellipticity of chick erythrocytes was retained in the apparentabsence of MB microtubules.

The proposed interpretation is in agreement with that put forward by Barrett &Dawson (1974) for chicken erythrocytes. Here the chick cells will lose their ellipticalshape shortly after differentiation in response to certain agents, but are resistant toshape change after a period of maturation. In addition, microtubule number dimin-ishes after the differentiating cells attain an elliptical morphology. The number ofMB microtubules is similarly reduced during differentiation of larval erythrocytes inthe rainbow trout (Yammoto & Luchi, 1975). Although there does not seem to beany reduction in number of microtubules during maturation of the camel reticulocytein peripheral blood, microtubules have rarely been identified in fully mature erythro-cytes. Furthermore, a comparison has not been made with more immature cells which

Fig. 12. Uranyl-acetate-stained whole mount of lysed camel erythrocyte, as seen inTEM. The MB is a continuous, densely staining peripheral band, traversed by a net-work of irregularly stained material in which there is a fold (/). Dense enlargement oninner side of MB (b) is distinguished from stained contaminant (X). TEM, x 13000.Fig. 13. Higher-magnification view of the area within the rectangle in Fig. 12. Outersurface of the MB is coated with network material (arrow). TEM, x 92000.Fig. 14. Region at one end of MB in which overlying material is less dense and theMB flattened, permitting view of individual microtubules (mt). The MB here isabout 10 microtubule diameters in width. Uranyl acetate staining; TEM, x 92000.Figs. 15, 16. Comparable views of lysed cells with and without MBs. A rough networkof material appears to encompass and coat the MB (Fig. 15); material comprisingghosts which lack MBs is more electron-dense and uniform (Fig. 16). Uranyl acetatestaining; TEM, x 58000.

W. D. Cohen and N. B. Terwilliger

mt

Marginal bands in camel erythrocytes 105

would be found in the bone marrow. However, based upon the results reported inother species, one would predict that erythropoietic cells in camel bone marrowcontain MB microtubules, possibly in greater number per cell and in a greaterpercentage of cells than in the peripheral blood. Goniakowska-Witalinska & Witalinski(1976) state that microtubules occur temporarily in the course of erythropoiesis in thellama and suggest that ellipticity is induced by the temporary presence of these micro-tubules. One indication that camels may typically release immature erythrocytes intothe peripheral blood is the occurrence of nucleated red cells in smears stained withWright's stain (Andrew, 1965). The nucleated MB-containing lysed cell observed inthis study (Fig. 9) was probably the product of one such cell.

One can only speculate as to why MBs were not observed by Goniakowska-Witalinska & Witalinski (1976) in camel erythrocytes. Apart from the possibility offixation or thin-section sampling artifacts, it may be that not all camels are identicalwith respect to the percentage of cells containing MBs or that there is variation withinthe same camel at different times. Although in the present study of lysed cells bothcamels exhibited similar numbers of MBs, one of the two guanacos examined hadconsiderably greater numbers of MBs than the other.

The probable loss of MBs during final stages of erythrocyte differentiation impliesthat changes occur in surface-associated cellular material, perhaps by alteration ofmolecular constituents or molecular cross-linking, so as to maintain elliptical morpho-logy once it is established by the MB system. Such a process could account for thedifference in appearance between the network of MB-associated material in lysed cellwhole mounts and that of more uniformly electron-dense cell ghosts or remnantslacking MBs, as observed with TEM. An alternative possibility for this difference inappearance, however, could be shrinkage or contraction of the cell surface materialin lysed cells lacking MBs. This would also account for the reduced size of such ghostsas compared with intact erythrocytes.

The network of material enclosing MBs of lysed camel erythrocytes is morpho-logically similar to that observed in the semi-lysed, nucleated elliptical erythrocytesof fish and amphibians, and referred to previously as trans-band material or TBM(Cohen, 1978; Cohen et al. 1977). It has been postulated that a TBM network isnormally under tension in such cells, applying force asymmetrically across the MB soas to deform an otherwise more circular MB into an ellipse. The presence of a TBMcorrelate in elliptical mammalian erythrocytes is consistent with this hypothesis.Similarly, the figure-8 camel MB configuration corresponds to figure-8 MBs observedin semilysed erythrocytes of many non-mammalian species and interpreted as excessivedeformation of the MB due to extreme TBM shrinkage or contraction (Cohen, 1978).

Fig. 17. Thin section through whole, fixed camel erythrocytes. MB microtubules (mt)are evident in cross- and oblique section at opposite ends of central cell. A clear zoneor halo of lighter density than the rest of the cytoplasmic matrix surrounds many of themicrotubules. Note difference in haemoglobin concentration (electron density) betweenthis cell and others, indicating that MB-containing cell is a reticulocyte. TEM,x 28000. Inset: Higher-magnification view of microtubules in upper right of Fig. 17.TEM, x 48000.

106 W. D. Cohen and N. B. Terurilliger

The occurrence of MBs in camel erythrocytes is possibly correlated with ontogenyof distinctive physiological properties. Camels are adapted to survive extreme de-hydration and rapid rehydration. Their erythrocytes can withstand considerableosmotic stress, responding in a manner more similar to the elliptical, nucleatederythrocytes of non-mammalian vertebrates than to the biconcave diskoidal cellstypical of other mammals (Ponder, 1942; Trotter, 1956). Camel erythrocytes arehighly resistant to hypotonic haemolysis (Perk, 1963; Yagil, Sod-Moriah & Meyerstein,1974), exhibit a low rate of water transport (Naccache & Sha'afi, 1974), and are alsorelatively stable under hypertonic conditions, in which crenation was not observed(Yagil et al. 1974). In addition, very young camels (6 months or less) apparentlypossess 2 populations of erythrocytes with respect to osmotic resistance: one popu-lation with adult-type response, the other with still greater haemolytic resistance (Perk,1966). Direct studies of the possible correlation between occurrence of MBs andosmotic resistance in camel erythrocytes would therefore appear to have potentialvalue for understanding MB and erythrocyte function.

It is of interest to consider whether the family Camelidae is unique among themammals in having MBs associated with erythrocyte structure, or whether theCamelidae have simply come under closer scrutiny because their mature erythrocytesare elliptical. Grasso (1966) reported that MBs were present in nucleated foetalerythroblasts of rabbits, a species in which mature adult erythrocytes have theanucleate biconcave diskoidal structure typical of most mammals. More extensiveinvestigation of the possible role of MBs during mammalian erythrogenesis in generalwould therefore seem warranted.

W. D. Cohen wishes to express his gratitude to Dr Emil P. Dolensek, Veterinarian of theNew York Zoological Society, for his interest and generous cooperation in providing bloodsamples from animals at the Bronx Zoological Park. The helpful discussion and assistance ofI. Nemhauser, J. Hoffman, R. Mawe, and R. I. Sha'afi, as well as support of this work bygrant 11619 of the PSC/BHE City University of N.Y. Research Award Program and byNIH grant HL 20902 from the National Heart, Lung, and Blood Institute to Dr Cohen, isgratefully acknowledged. Part of this work was supported by NIH Training Grant 5-707-GM-00723 to the University of Wisconsin Anatomy Department and N. Barclay Tenvilliger isgrateful to Dr David B. Slautterback for his encouragement.

REFERENCES

ANDREW, W. (1965). Comparative Hematology. New York: Grune & Stratton.BARCLAY, N. E. (1966). Marginal bands in duck and camel erythrocytes. Anat. Rec. 154, 313.BARRETT, L. A. & DAWSON, R. B. (1974). Avian erythrocyte development: microtubules and

the formation of the disk shape. Devi Biol. 36, 72-81.BEHNKE, O. (1965). Further studies on microtubules. A marginal bundle in human and rat

thrombocytes. J. Ultrastruct. Res. 13, 469-477.BEHNKE, O. (1970). A comparative study of microtubules of disk-shaped blood cells. J. Ultra-

struct. Res. 31, 61^75.CAVANAUGH, G. H. (ed.) (1975). Formulae and Methods VI of the Marine Biological Laboratory

Chemical Room. Woods Hole, Mass.: MBL.COHEN, W. D. (1978). Observations on the marginal band system of nucleated erythrocytes.

J. Cell Biol. 78, 260-273.COHEN, W. D., NEMHAUSER, I. & JAEGER, R. (1977). Rapid visualization of the marginal band

system in blood cells of marine species. Biol. Bull. mar. biol. Lab., Woods Hole 153, 420.

Marginal bands in camel erythrocytes 107

DEHLER, A. (1895). Beitrag zur Kenntnis des feineren Baues der roten Bliitkorperchen beimHiihnerembryo. Arch, mikrosk. Anat. EnttvMech. 46, 414-430.

FAWCETT, D. W. (1959). Electron microscopic observations on the marginal band of nucleatederythrocytes. Anat. Rec. 133, 379.

FAWCETT, D. W. & WITEBSKY, F. (1964). Observations on the ultrastructure of nucleatederythrocytes and thrombocytes with particular reference to the structural basis of theirdiscoidal shape. Z. Zellforsch. mikrosk. Anat. 62, 785-806.

GONIAKOWSKA-W1TALIN8KA, L. & WITALINSKI, W. (1976). Evidence for a correlation betweenthe number of marginal band microtubules and the size of vertebrate erythrocytes. J. CellSci. 2a, 397-401.

GRASSO, J. A. (1966). Cytoplasmic microtubules in mammalian erythropoietic cells. Anat. Rec.i56, 397-4I4-

MEVES, F. (1911). Gesammelte Studien an den roten Blutkorperchen der Ampibien. Arch.mikrosk. Anat. EntwMech. yj, 465-540.

NACCACHE, P. & SHA'AFI, R. I. (1974). Effect of PCMS on water transfer across biologicalmembranes. J. cell. Physiol. 83, 449-456.

PERK, K. (1963). The camel's erythrocyte. Nature, hand. 200, 272-273.PERK, K. (1966). Osmotic hemolysis of the camel's erythrocytes. I. A microcinematographic

study. J. exp. Zool. 163, 241-246.PONDER, E. (1942). Shape transformations in the red cells of camels, birds, reptiles, amphibia,

and fishes. J. exp. Biol. 19, 215-219.REBHUN, L. I., ROSENBAUM, J., LEFEBRE, P. & SMITH, G. (1974). Reversible restoration of the

birefringence of cold-treated, isolated mitotic apparatus of surf clam eggs with chick braintubulin. Nature, Lond. 249, 113-115.

SANDBORN, E. B., LEBUIS, J. & Bois, P. (1966). Cytoplasmic microtubules in blood platelets.Blood 27, 247-252.

TROTTER, W. D. (1956). Banding in Salamander erythrocytes. A shape change corresponding todisc-sphere transformation in mammalian red cells. Expl Cell Res. 11, 587-603.

YAGIL, R., SOD-MORIAH, A. & MEYERSTEIN, N. (1974). Dehydration and camel blood. II.Shape, size, and concentration of red blood cells. Am. J. Physiol. 226, 301-304.

WEISENBERG, R. C. (1972). Microtubule formation in vitro in solutions containing low calciumconcentrations. Science, N.Y. 177, 1104-1105.

YAMMOTO, M. & IUCHI, I. (1975). Electron microscopic study of erythrocytes in developingrainbow trout, Salmo gairdnerii irideus, with particular reference to changes in the cell line.J. exp. Zool. 191, 407-426.

{Received 30 May 1978)