Microtubule-organizing centres and assembly of the double-spiral microtubule pattern in certain heliozoan axonemes

There are two basic ways in which microtubule patterns are established during the assembly of highly ordered microtubule bundles (Tucker, 1977, 1979)-Pattern may be established by a highly ordered array of interconnected microtubule-nucleating elements that acts as a microtubule-nucleating template so that microtubules are situated in a specific pattern as their assembly begins. This is the case during assembly of centrioles and basal bodies (see Tamm & Tamm, 1980), and certain cytopharyngeal microtubule bundles (Tucker, Dunn & Pattisson, 1975; Pearson & Tucker, 1977). Alternatively, pattern may be generated by a self-linkage procedure in which microtubules are not positioned precisely when they start to assemble. Pattern is established as intertubule links join microtubules together.

The axopodial axonemes of actinophryidian heliozoans exhibit one of the most complex microtubule patterns so far described (MacDonald & Kitching, 1967; Tilney & Byers, 1969; Roth, Pihlaja & Shigenaka, 1970; Ockleford & Tucker, 1973; Cachon & Cachon, 1974). The 12-sectored double-spiral pattern is considerably more elaborate than the cylindrical arrangement of centriole microtubules or the hexagonal microtubule lattice in cytopharyngeal bundles. The double-spiral lattice of microtubules and links represents a non-equivalent pattern (Tucker, 1977) of microtubule packing and linkage. Microtubules at certain loci in the pattern are linked and positioned with respect to their immediate neighbours in a different way from those situated elsewhere in the pattern. Nevertheless, examination of axonemes reassembling after microtubule breakdown that has been induced by cold treatment indicates that the double-spiral pattern is generated by self-linkage (Tilney & Byers, 1969). However, serial cross-sectioning of the first-formed basal portions of developing axonemes has never been undertaken. The central portion of the double-spiral is the most complex part of the pattern and is constructed before the more peripheral regions. The possibility that it is established by a pre-existing template that ‘starts-off’ the double-spiral pattern has not been excluded. Other important questions also remain unanswered. Is assembly of axonemal microtubules initiated in discrete cytoplasmic localities by microtubule-organizing centres (MTOCs) so that microtubules are in fairly close proximity to begin with? There is no evidence for such MTOCs in these particular organisms to date. How do an axoneme’s microtubules become closely juxtaposed alongside each other so that self-linkage can be effected if a template is not employed? These issues are dealt with in this spatio-temporal analysis of axonemal assembly for 2 species of actinophryidian heliozoans.

Echinosphaerium nucleofilum and Actinophrys sol were obtained from the Culture Centre of Algae and Protozoa, 36 Storey’s Way, Cambridge CB3 oDT, England (reference nos. 1507/1 and 1502/2, respectively). They were cultured at room temperature (20 °C) as described by Ockleford & Tucker (1973), except that 1 ml of an aqueous solution of disodium EDTA (0·29 g l⁻¹) was added to each litre of culture medium for E. nucleofilum. All organisms examined were taken from cultures approximately 6 days after subculturing so that none of the organisms was packed with freshly formed food vacuoles (this occurs shortly after subculturing) or had been subjected to prolonged starvation.

Living organisms were examined using a Zeiss (Oberkochen Ltd) Stereomicroscope III fitted with a transilluminator or a Universal Microscope fitted with differential interference contrast optics. For high-resolution examinations a coverslip was supported by a ring of silicone grease that surrounded a drop of culture medium and organisms. This drop formed a column between slide and slip, and its sides did not contact the grease ring. Organisms survive in a healthy state for at least 24 h in these preparations, which are referred to as ring-preparations in the account that follows. The lengths of axopodia and diameters of cell bodies were assessed using an eyepiece micrometer.

The procedure devised by Roth, Pihlaja & Shigenaka (1970) was used during preparation of organisms for electron microscopy. Fixatives were all at 20 °C prior to addition to culture medium and organisms (even if the latter had been cooled), because this provided better preservation than when the temperature of the fixative was lowered tq that of the culture medium. When cold-treated organisms were fixed, a duplicate control preparation (of organisms from the same culture as those destined for fixation and contained in identical watchglasses) was always cooled simultaneously and examined as it warmed at room temperature, shortly after the other organisms were fixed, to ascertain that more than 90% of the organisms in the control preparation had survived the cold treatment and retained the ability to form new axo-podia. This was undertaken to minimize the possibility that organisms were in a moribund state and/or potentially unable to generate new axonemes and axopodia when fixed.

Organisms and culture medium contained in watchglasses or ring-preparations were cooled by placing them on ice-chips in a polystyrene container or by using a cooled incubator (set at — 2 °C). The rise in temperature of the culture medium after removal from the ice-container or incubator to room temperature was monitored using a thermocouple as described by Ockleford & Tucker (1973).

E. nucleofilum was cooled to o °C for 6 h. The organisms did not usually survive cooling to lower temperatures or being kept for more than 6 h at o °C. Hence the o °C/6 h regime was the most severe form of low-temperature treatment available. Over 90% of the organisms survive this treatment. For A. sol the most severe cold treatment giving at least 90% survival was also used; organisms were cooled to —2 °C for 17 h.

Three untreated organisms, 3 organisms fixed immediately after 6 h cold treatment at o °C, and 3 organisms fixed 7 min after termination of this treatment when the temperature had risen to 10 °C were examined ultrastructurally.

The spherical multinucleate cell body is highly vacuolated and consists of 2 distinct cytoplasmic regions (Fig. 1, arrows). Most of the vacuoles in the centrally situated endoplasm are much smaller than those in the peripheral ectoplasm. Nuclei are distributed more or less evenly in a single layer around the ecto-endoplasmic border. Axonemes terminate proximally at this level. In some cases the base of an axoneme appears to make contact with the surface of a nucleus (Tilney & Porter, 1965; Roth & Shigenaka, 1970); most bases do not but are usually situated within 1·5 μm of the nearest nucleus.

Sections of axonemes close to their basal extremities (within 1–2 μm) reveal the presence of small patches of apparently amorphous dense material positioned along the sides of axonemes. This material appears to make contact with the walls of some of the microtubules at the peripheries of axonemal cross-sectional profiles (Fig. 3, arrows). Whether these patches are spatially discrete entities or, alternatively, represent interconnected portions of a loose anastomosing network has not been ascertained. Such patches are also apparent in a micrograph (Fig. 9, Roth & Shigenaka, 1970) but did not attract comment. They are distinct from the less substantial dense material clustered at the basal tips of axonemal microtubules (Tilney & Porter, 1965; Roth & Shigenaka, 1970).

The cross-sectional profile of a mature axoneme consists of 2 main interlocking microtubule spirals (Fig. 3). In some instances cross-sectional profiles of axonemes on opposite sides of an individual were examined because Tilney & Porter (1965) suggested that such axonemes might have opposite spiral handedness and that this might be related to axopodial activity during locomotion. All spirals followed anticlockwise courses (progressing from centre to periphery of a cross-sectional profile) for profiles viewed distally (looking from tip to base of an axoneme). This is also the case for mature axonemes in A. sol and for developing axonemes shortly after cold treatment in both organisms. There are no micrographs published showing axonemes of opposite hand in the cell bodies of these organisms. However, Tilney & Porter (1965) have shown for E. nucleofilum that a few axopodia contain a pair of oppositely spiralling axonemes. Such axopodia might have bent over on themselves and portions proximal and distal to a bend might have fused together (during or just prior to initial fixation) so that the same axoneme passes through a section twice and with opposite polarities.

Axopodia retract completely during cold treatment (Tilney & Porter, 1967) but numerous short pointed protuberances that are not present usually project from the surfaces of cell bodies (Fig. 2). The ectoplasm contains considerable numbers of macrotubules with diameters of about 40 nm (Tilney & Porter, 1967; Toyohara, Shigenaka & Mohri, 1978). Most of these appear to be arranged randomly but inside the protuberances the majority are oriented along the lengths of the protuberances. Hence, these macrotubules may help to maintain the elongated (about 10 μm long) shapes of the protuberances.

Organisms still contain a few double-spiral microtubule arrays near the endo-ectoplasmic border. This is in contrast to the apparent lack of any double-spiral arrays after cooling to 0 °C for 3 h, as reported previously (Tilney & Byers, 1969). These axonemal remnants are very much shorter, far less numerous, and include considerably fewer microtubules (only the central portion of the double-spiral pattern) than the mature axonemes of untreated organisms (compare Figs. 3, 5). Dense material is concentrated around their sides at certain points along their lengths. This material is similar in appearance to that of the dense patches at the bases of mature axonemes and to that which forms small, approximately spherical, clumps (0·2–0·4 μm in diameter) that are not associated with double-spiral microtubule arrays immediately after cold treatment. Each of these clumps is situated usually within 1–2 μm of a nucleus and clumps are encountered with about the same frequency in sections near the ecto-endoplasmic border as the bases of mature axonemes in untreated organisms. The spacing of clumps (and axonemal bases in untreated organisms) corresponds closely to that of nuclei. Microtubules project at an apparently random variety of orientations from the surfaces of the clumps (Fig. 6). Apart from these microtubules and the axonemal remnants no other groupings of microtubules were detected. A few microtubules are sparsely distributed among the macrotubules in the ectoplasm.

Small axopodia (5–10 μm long) are first detectable at the surfaces of organisms contained in ring-preparations examined using high-resolution interference contrast optics about 8 min after removal from the cold, when preparations have wanned to 20 °C. Prior to this point it is difficult to distinguish projections that may represent axopodia from the pointed protuberances that are still present. However, the fixation procedure for electron microscopy requires that organisms are cooled in 5 ml of culture medium contained in watchglasses. Warming after removal to room temperature proceeds more slowly than it does for ring-preparations, which reach 20 °C after 4 min, and recovery of axopodia has to be monitored with a dissecting microscope that provides less resolution than interference contrast optics. Under these conditions, axopodia (which probably have lengths of 15–40μm) are first detectable 12 min after removal from the cold, when watchglasses and their contents have warmed to 13·5 °C. Organisms were fixed for electron microscopy 7 min after removal from the cold, when the temperature had risen to 10 °C. These organisms contain a range of stages in the assembly of new axonemes.

The frequency with which doublespiral axonemal microtubule arrays were encountered in sections cut through the ectoplasm was considerably greater than for organisms fixed immediately after cold treatment. They all include fewer microtubules than those of untreated organisms and exhibit the patterns found at the centres of mature axonemes. Most of them are more or less radially oriented with respect to the spherical cell body. No indications were obtained of the way in which this orientation is accomplished. A few of the developing axonemes project into the macrotubule-containing surface protuberances (Fig. 4). Examination of ring-preparations reveals that new axopodia often extend from the tips of protuberances during recovery from cold treatment.

Clumps of dense material are still present near nuclei. They are considerably larger, more irregularly shaped, and are associated with more microtubules than those in organisms fixed immediately after cold treatment (compare Figs. 6, 7).

Serial sectioning revealed 2 main types of clumps. Some only have microtubules radiating in apparently random directions from their surfaces (Fig. 7). The dense material of the others, in addition to exhibiting this type of microtubule association, extends around and along the sides of small groupings of well-aligned microtubules (Fig. 8), which when cross-sectioned exhibit a packing pattern identical with that at the centre of a double-spiral pattern (Fig. 9). Whether the clumps associated with an individual developing axoneme are all interconnected by thinner strips and strands of dense material has not been ascertained. The clumped material extends for up to 5 μm along the sides of the basal portions of these developing axonemes.

The tipa of irregularly arranged microtubules that surround each axoneme are embedded in the clumps. These microtubules splay out from the clumps at a variety of orientations with respect to the longitudinal axis of a developing axoneme (Fig. 8).

Individual developing axonemes can occasionally be recognized and followed in serial sections for distances of up to 25 μm along their lengths. They can be distinguished because of the extent to which assembly of the double-spiral pattern has progressed. The organization of 3 axonemes in 2 organisms was examined in this way. They showed that the developing double-spiral array includes more microtubules close to its base than it does at more distal levels (Figs. 10–12). However, some loci in the pattern that are occupied by microtubules at distal levels are not so occupied at more basal levels (Figs. 10–12, short arrows). Thus portions of tubules are added to the pattern at both distal and basal levels, although the former is more common. Such addition might be achieved by microtubule elongation as microtubule assembly progresses or, alternatively, because portions of tubules splay out from the pattern as indicated in Fig. 13 and are subsequently drawn into the pattern. Some of the microtubule profiles that are grouped around and well-aligned with developing double-spiral arrays may represent such portions (Figs. 8, 11), but in no case was it possible to demonstrate unequivocally that tubules providing such profiles joined the pattern. It is difficult to follow these tubules in sequential sections because they are not positioned very regularly and oriented with respect to each other and the double-spiral array. Because such tubule portions are much less numerous in the vicinity of mature axonemes, and because it is difficult to see how they could become precisely positioned simultaneously at all points along their lengths, it is reasonable to suppose that microtubules are incorporated progressively into the pattern in both distal and proximal directions (Fig. 13). If this is not the case then these tubules must be disassembled at some point and can have little direct relevance to axonemal assembly. This possibility seems most unlikely, bearing in mind the large numbers of microtubules involved (Fig. 8). Fine strands of dense material that are situated

Although the central portion of the double-spiral is constructed before its periphery, the 2 spirals are not only produced by addition of microtubules one-by-one to the most peripheral end of each spiral. Portions of spirals can be built up at (apparently) any region around the edge of a developing double-spiral array (Fig. 10).

Three untreated organisms, 3 organisms fixed immediately after 17 h cold treatment at — 2 °C, and several organisms fixed at intervals after this treatment (3 organisms at 2 min, 12 min and 16 min; 2 organisms at 5, 6·5 and 10 min; 1 organism at 15 min) were examined ultrastructurally.

Axonemes radiate from the outer surface of the single central nucleus. The basal portions of their microtubules terminate in lateral register with one another where they contact the nuclear envelope (Fig. 14). The double spiral pattern is apparently identical with that of E. nucleofilum but the largest axonemes in the cell body and proximal portions of axopodia include fewer microtubules than those of E. nucleofilum (Ockleford & Tucker, 1973; Patterson, 1979). No patches of dense material were detected at axonemal bases.

A few short axopodia remain after cold treatment. They contain microtubules aligned parallel to their longitudinal axes but these are not arranged in a double-spiral pattern. A few sparsely distributed macrotubules with diameters of about 40 nm are also present in the ectoplasm near the surface of the cell body. Numerous microtubules project from the outer surface of the nuclear envelope. Unlike the microtubules of mature axonemes, these microtubules are not well-aligned with their neighbours, nor are they grouped in compact arrays and confined to certain surface regions of the envelope (compare Figs. 14, 15). They are distributed more or less evenly over the entire surface of the envelope. None of these microtubules are arranged in a double-spiral pattern. The proximal ends of most (perhaps all) of them are attached to the nuclear envelope.

Examination of ring-preparations reveals that new axopodia start to extend from cell bodies 12 min after removal from the cold. These preparations reach room temperature (20 °C) 4 min after such removal but organisms cooled in watchglasses for fixation have only warmed to 10 °C after 12 min.

Greater numbers of microtubules radiate from the surface of the nuclear envelope, have a more precise radial orientation, are more exactly aligned with their neighbours, and are less evenly distributed where they contact the envelope than the juxtanuclear microtubules, in organisms fixed immediately after cold treatment (compare Figs. 15, 16). Many of them are concentrated together to form fairly compact bundles. Within bundles they are not packed in well-defined patterns except within 2–3 μm of the nuclear envelope where some of them are arranged as short slightly curved rows of tubules. Some of these rows are grouped in pairs in configurations closely approximating to those at the centre of the double-spiral axonemal pattern (Fig. 17). Serial sectioning did not reveal any configurations that were exactly identical with the central portion of a mature double-spiral; presumably they represent stages in the self-linkage of microtubules to form axonemal centres. They were not detected in organisms fixed 2, 5, 6·5 or 10 min after cold treatment. Fine strands of dense material that may interconnect tubules are sometimes situated within these doublespiral groupings (Fig. 17).

Numerous groupings of radially oriented microtubules are present in the vicinity of the nucleus in these organisms. Many of these groupings include the central portions of the double-spiral arrays of developing axonemes (Figs. 18, 19). The spatial organization of the basal portions of 4 developing axonemes that could be followed in sequential sections through 2 organisms was examined. More microtubules were found at basal levels than at more distal levels and this provided one instance in which a row of microtubules splays out distally from an axonemal centre (compare Figs. 18, 19, arrows). No examples were found of the absence of microtubules from loci at basal levels in these arrays that are occupied by microtubules at more distal levels. It is perhaps to be expected that this feature would be less common than it is in E. nucleofilum because the microtubules are initially most closely juxtaposed at a single level (the surface of the nuclear envelope), so that pattern generation is most likely to begin at tubule bases. The spatio-temporal sequence of events as axonemes start to assemble is summarized in Fig. 20. No clumps of dense material were detected at the bases of developing axonemes.

The double-spiral pattern is rapidly established after cold treatment, within 7 and 16 min for E. nucleofilum and A. sol, respectively. In both organisms this appears to be facilitated by nucleation of microtubule assembly in highly localized juxtanuclear cytoplasmic regions.

In E. nucleofilum clumps of dense material act as MTOCs. They apparently nucleate microtubule assembly, because microtubules grow out from them. Continuation of a highly localized supply of microtubules as axonemes assemble seems to be achieved because clumps spread along the sides of the first-formed portions of axonemes. The clumps appear to do little in the way of ‘organizing’ apart from ensuring that microtubules are concentrated in a number of discrete cytoplasmic regions. The dense patches at the base of each mature axoneme of untreated organisms probably represent remnants of an MTOC clump. They may provide the material basis for maintenance of axoneme number and initiation of an active MTOC following situations such as exposure to low temperatures in which most axonemes are completely disassembled.

MTOC clumps are situated close to nuclei although direct structural connection between them was not detected. This raises the question of whether there is a precise numerical correlation between nuclei, MTOCs, axonemes and axopodia. Possibly, cell size, nuclear number and the number of axopodia stay in register during interfission growth because each nucleus is involved in the maintenance of one axonemal MTOC. Under the culture conditions employed in this study the basal portions of all axonemes of untreated organisms exhibited cross-sectional profiles of similar size, included similar numbers of microtubules, and their spacing corresponded closely to that of the nuclei. The majority (perhaps all) of these axonemes extended into axopodia and each axopodium included a single axoneme. However, this situation may not be typical. It is clear from the micrographs of Tilney (1968), Allison et al. (1970), Roth & Shigenaka (1970) and Roth, Pihlaja & Shigenaka (1970) that, sometimes, under both normal and experimental conditions considerably more than 2 axonemal bases are positioned near each nucleus. In these instances axonemes vary markedly in cross-sectional size and it is possible that, as in A. sol (Ockleford & Tucker, 1973), many of the smaller axonemes are confined to the cell body and do not extend into axopodia.

E. nucleofilum apparently stores much of its tubulin in the form of macrotubular tubulin aggregates during cold treatment. It has been suggested that macrotubules may convert directly to microtubules by rearrangement of protofilaments during recovery from cold treatment (Tilney & Porter, 1967; Toyohara, Shigenaka & Mohri, 1978) rather than dissociating completely into soluble tubulins, which then reassemble to form microtubules. There is direct evidence for such conversions (Suzaki, Toyohara, Watanabe, Shigenaka & Sakai, 1980). The concentration of microtubules at the surfaces of MTOC clumps while macrotubules remain scattered throughout the ectoplasm indicates that, at least initially, axonemal microtubules assemble from soluble tubulin.

The outer surface of the nuclear envelope of A. sol may act as an MTOC. After cold treatment microtubules assemble in its immediate vicinity. Microtubulenucleating elements might be firmly bound to, and oriented at, the surface of the envelope so that microtubules are radially oriented alongside each other as they start to assemble. This would facilitate subsequent production of the parallel, more closely packed, array of axonemal microtubules in axonemes. However, the possibility that assembly is initiated close to, but not in contact with, the envelope and is followed by attachment of tubule tips to the envelope has not been eliminated in this study. In either case the nuclear surface clearly plays some part in organizing the microtubules. If attachment and/or nucleation of tubule tips is related to tubule polarity, the association will define the polarity of the axoneme as well as its radial orientation. Tip attachment to the outer membrane of the envelope presumably must be such that tips can move in the plane of the membrane. If this were not the case most of the tubules could not be included in the compact double-spiral pattern that is finally established right down to the level at which tubules contact the membrane.

In A. sol there does not seem to be any nucleus-mediated specification of axonemal number. Large numbers of axonemes start to assemble near the nucleus. They are finally of very variable cross-sectional size (consisting of correspondingly variable numbers of microtubules). Only the largest project into axonemes (Ockleford & Tucker, 1973). Thus it appears that production of central axonemal portions is not under strict numerical control, that they may ‘compete’ for ‘free’ microtubules on a ‘first come, first served’ basis in which the ‘large get larger* because they have more peripheral sites available for self-linking free microtubules, and that only those above a certain critical size continue to elongate and promote outgrowth of axopodia.

The central portion of the double-spiral pattern is more complex in terms of microtubule packing and linkage than its periphery (MacDonald & Kitching, 1967; Tilney & Byers, 1969; Harris, 1970; Cachon & Cachon, 1974). Establishing the central portion may require more detailed spatial instructions than are needed to extend the 2 spirals once the central portion has been constructed. Bearing in mind the speed with which central portions are established (7–16 min) and their intricacy, it is not unreasonable to consider whether the double-spiral pattern is initiated by a pre-existing microtubule-nucleating template. It has been suggested that a template is not involved in E. nucleofilum because sections of developing axonemes reveal microtubule configurations that appear to represent stages in the self-linkage and positioning of microtubules to form the central portions of double-spiral arrays (Tilney & Byers, 1969). However, the present study (see below) and that conducted on Raphidiophrys (Tilney, 1971) show that in developing axonemes pattern can be established at one level in the microtubule bundle prior to its appearance at another. Patterning is apparently achieved as adjacent microtubules are ‘zipped’ together by a self-linkage procedure at progressively more distant levels from those at which pattern is first established. Hence, even if stages in pattern establishment are detected at a particular level in an axoneme they may not represent the initial establishment of pattern. Examination of cross-sectioned axonemes at all levels during pattern establishment is needed to eliminate the possibility of template specification of microtubule positioning.

Such examinations in the present study show unequivocally that the positions of microtubules at the periphery of the double-spiral pattern are not template-specified because these tubules join the patterned array at several levels as axonemes assemble, and in E. nucleofilum most microtubules in the vicinity of the axonemal MTOC clumps are not precisely positioned when they start to assemble. Stages in the initial positioning of microtubules at the centre of a double-spiral were not found in E. nucleofilum but may have been obscured by the dense MTOC material. Crosssectional profiles that appear to represent such stages were obtained for A. sol. These microtubule groupings did not exhibit perfect double-spiral microtubule positioning at any points along their lengths. Thus, the entire double-spiral pattern is probably established without the exploitation of a microtubule-nucleating template. Lack of precise positioning when microtubules start to assemble also eliminates ‘linkernucleation’ (Bárdele, 1977) as the mechanism for generating pattern.

The ‘template controversy’ (Tilney, 1971) includes the centrohelidian Raphidiophrys where the axonemal microtubule pattern is less complex than it is in an actino-phryidian double-spiral. In Raphidiophrys a single dense body called the centroplast nucleates axonemal microtubule assembly, but pattern is established initially at some distance from its surface and so, presumably, it does not act as a template and self-linkage generates pattern (Tilney, 1971). However, Rieder (1979) demonstrated intertubular rod-like components where axonemes emanate from the centroplast and suggested that they form part of a template that establishes axonemal microtubule pattern.

Neighbouring microtubules need to be aligned and situated within a ‘link’s length’ of each other before self-linkage can proceed. Neither of these conditions initially obtain to any great extent as microtubules start to assemble in E. nucleofilum after cold treatment. In A. sol some of the microtubules are fairly well-aligned. High-resolution light microscopy does not reveal any detectable Brownian movement in the cytoplasm (excluding the vacuoles) of these 2 organisms. The degree of cytoplasmic gelation is probably such that microtubules need to approach each other with the assistance of some agency other than random diffusion if self-linkage and pattern formation are to be effected as rapidly as observed. Fine strands are situated between microtubules during pattern formation. These might form part of a contractile meshwork that draws microtubules together to facilitate self-linkage. The filaments in such a meshwork would not necessarily need to have a very well organized spatial arrangement to do this. Provided some portions of some microtubules are drawn to within a link’s length of each other, self-linkage of those that are aligned could begin. It may not be necessary to align and position microtubules precisely along their entire lengths before self-linkage can start. Once short portions of 2 adjacent microtubules are connected by links, regions immediately ‘above and below’ these portions would be more suitably positioned for self-linkage. Thus self-linkage could proceed progressively along tubules moving away from the initially linked portions and by a ‘zipper-like action’ draw the flexible microtubules (Ockleford & Tucker, 1973) together along their entire lengths eventually. The evidence that some of the microtubules splay out from the patterned array as axonemes assemble is compatible with such a procedure, which is similar in some respects to that envisaged in the ‘zipper hypothesis’ for microtubule repositioning in certain spindles (Bajer, 1973).

Serial sectioning reveals that microtubules are incorporated into the doublespiral pattern at various levels, proceeding both proximally and distally along developing axonemes in E. nucleofilum. Hence it is difficult to account for pattern generation in the entire axoneme unless individual tubules can be zipped together in both directions. It is not known if all tubules in an axoneme have the same polarity. The distal migration of bends (produced by manipulation with a microneedle) during axonemal repair in A. sol (Ockleford & Tucker, 1973) suggests that all the microtubules do have the same polarity and may be an indication that selflinkage proceeds more rapidly in a distal than in a proximal direction.

The ways in which various microtubule/link interactions might establish the 12-sectored double-spiral pattern during self-linkage have been considered in detail by previous investigators (Tilney & Byers, 1969; Roth, Pihlaja & Shigenaka, 1970; Cachon, Cachon, Febvre-Chevalier & Febvre, 1973; Cachon & Cachon, 1974; Bárdele, 1977).

J.C. R.J. held a Science Research Council (U.K.) Studentship during this study and grant support from the S.R.C. to J.B.T. is gratefully acknowledged.

It has recently been shown that most of the microtubules (more than 98%) in the axonemes of E. nucleofilum have the same polarity.