Basal body and flagellar development during the vegetative cell cycle and the sexual cycle of Chlamydomonas reinhardii

The unicellular biflagellate green alga Chlamydomonas reinhardii is the most suitable eukaryote for combined genetical, electron-microscopic, physiological and biochemical studies of flagella and basal bodies (Randall, Cavalier-Smith, McVittie, Warr & Hopkins, 1967; Rosenbaum, Moulder & Ringo, 1969). The study reported in this paper was carried out in the belief that it is essential to have as complete an understanding as possible of the ultrastructural changes which occur in wild type cells during the development of flagella and basal bodies, as a basis for hypotheses about the mechanisms and control of flagellar and basal body development.

The fine structure of the mature flagellar apparatus of C. reinhardii has been described in detail by Ringo (1967a, b) and by me (Cavalier-Smith, 1967). This paper describes the developmental cycle of flagella and basal bodies in synchronous cultures of C. reinhardii and during the sexual cycle. Further details of the structure of basal bodies and associated structures as seen in cell homogenates are described. My observations on synchronous cultures and the sexual cycle were summarized briefly in the review by Randall et al. (1967). Since their completion a paper on cell division in ‘logarithmically’ grown cells has appeared (Johnson & Porter, 1968). Johnson & Porter’s findings agree with those reported here in some, but not all, respects; in particular their interpretation of flagellar regression is substantially different.

Wild type strain 32D (minus mating type) of Chlamydomonas reinhardii from the Culture Collection, Botany School, Cambridge, England, was synchronized by a method based on that used by Bernstein (1960, 1964) for Chlamydomonas moewusii, the major difference being that aeration was by filtered air instead of 5% carbon dioxide in air. Cells were grown at 25 °C in alternating 12-h periods of light and darkness, the light being provided by ‘cool white’ and ‘daylight’ fluorescent lamps at an intensity of about 6500 lux (measured by a Gossen Lunasix CdS exposure meter). Cells were grown in 2–1. Erlenmeyer flasks in 1 1. of a liquid medium based on Medium I of Sager & Granick (1953), but with ferric chloride replaced by 0·01 g/1. ferric citrate and 0·01 g/1. citric acid.

To assess the degree of synchrony cells were counted in a haemocytometer after fixation in glutaraldehyde or formaldehyde, and growth curves obtained by plotting cell numbers/ml against time (Fig. 1). All observations were made after at least one (and usually two or more) full light-dark cycles (each of 24 h) had elapsed since inoculation of the culture.

Plus and minus wild type strains (32D and 32C) of C. reinhardii were grown in synchronous culture, and gametes were prepared by transferring cells, after 6 h in the light period, into sterile 5 × 10⁻⁴ M CaCl₂ solution and incubating in continuous light for 18–24 h. Equal quantities of plus and minus cells were mixed and incubated in the light to allow mating to occur. After zygote formation, the cells were either suspended in fresh liquid medium or plated out on 2% agar plates and then incubated in the light for 1 day and in the dark for 5 days. The resulting zygospores were placed in the light on fresh agar plates for germination to occur. Because zygospores ‘matured’ in liquid were very inhomogeneous and usually failed to germinate (Cavalier-Smith, 1967), only agar-matured zygospores were used for studies of zygospore germination. Zygospores matured on agar by a method known to lead subsequently to very synchronous germination (Lawrence, 1965) were kindly provided by Dr Lawrence, and then germinated at 25 °C and 5000 lux by transferring them on to fresh 1-5% agar in a medium with the following composition, per litre: K₂HPO₄, 1-44 g; KH₂PO₄, 0·72 g; NH₄NO₃, 0·4g; CaCl₂.6H₂O, 0·07 g; MgSO₄.7H₂O, 0 ·1 g; ferric citrate, 0·01 g; citric acid, 0·01 g; sodium citrate, 0·05 g; and trace element solution (Sager & Granick, 1953), 10 ml.

Living cells were examined at all stages by phase-contrast and bright-field microscopy, and cells were sampled and fixed for electron microscopy in the middle of the light period and at °> J> 3. 5> 6, and 8 h after the beginning of the dark period and at various intervals during the sexual cycle. Cells were either centrifuged into a pellet to which fixative was added, or an equal volume of fixative was added directly to the suspension and the cells subsequently centrifuged and fresh fixative added to the resulting pellet. Fixation was for 1 h with 1%glutaraldehyde in 0·02 M phosphate buffer pH 7·4 (total osmolarity 140 mosmol, measured on an Advanced osmomcter). Higher osmolarities caused cell shrinkage. Cells were washed over night in 0·05 M phosphate buffer (102 mosmol), postfixed in 1% osmium tetroxide in 0·02 M phosphate pH 7·4 and embedded in Araldite (Luft, 1961). Sections were stained in 2% uranyl acetate or uranyl magnesium acetate (in distilled water or 50% ethanol) for from 15 min to several hours and post-stained in lead citrate (Reynolds, 1963) for 5–30 min. A Siemens Elmiskop I was used at 80 kV with 50μm objective apertures.

Cells of strain 89+ from the Indiana culture collection were suspended at 4 °C in a 15% (w/v) sucrose solution containing 1 mM ethylenediaminetetra-acetic acid (EDTA), pH 8, and passed through a French press at a pressure of 1·38 × 10⁴ kN m⁻². The resulting homogenate was either negatively stained with 2% potassium phosphotungstate (pH 70) after 5 min fixation in OsO₄ vapour, or, alternatively, centrifuged to give a pellet, which was fixed in glutaraldehyde and OsO₄, embedded in Epon, sectioned and stained as described for whole cells.

The overall behaviour of flagella and basal bodies during the vegetative cell cycle is shown in Fig. 2. The flagella regress before cell division begins, by gradually becoming shorter over a period of approximately 30 min, just as in C. moewusii (Lewin, 1952). No flagellar fragments sufficiently large to be seen under phase contrast are broken off and cast into the external medium; moreover, during regression the remaining part of the flagellum may beat apparently normally, which suggests that it is fairly intact.

The cell apex remains pointed until the 2 flagellar stumps have completely regressed; it then becomes rounded and about 15 min later a cleavage furrow begins to form in the colourless apical cytoplasm at the point where the flagella had been. At some stage during cytokinesis (and not ‘just prior to nuclear division’ as reported by Buffaloe, 1958) the protoplast rotates through 90° relative to the cell wall, so that by the end of cytokinesis (which takes about 7 min) the cleavage furrow lies at right angles to the former long axis of the cell.

Under the growth conditions used here a second cleavage (and often later a third) begins about 12 min later; thus 4 or 8 daughter cells are formed and remain enclosed by the mother cell wall until daughter cell liberation occurs. Each daughter cell grows 2 flagella, which become motile before the mother cell wall breaks down; indeed their ‘writhing’ inside the swollen wall possibly helps to break it open. Rupture occurs at any point on the cell surface, and the flagellar tunnels (see below) are visible as 2 dark spots on the cell wall, which later disintegrates.

Although all the cells divide during each dark period, the fact that the successive cell divisions are spread throughout several hours of the dark period (Fig. 3), coupled with the short duration of each cell division, means that only a small proportion of the population is in the same stage of cell division at any one time. Flagellar regression, likewise, takes place during a period of about 5 h, so that at any one time not more than 10% of the population are regressing. Daughter cell liberation (Figs. 1, 3) is somewhat better synchronized, usually occurring in a 1 –3 h period.

Fig. 4 summarizes the overall behaviour of the flagellar apparatus during the sexual cycle. Flagellar regression begins a few hours after the biflagellate + and −gametes have fused to form a quadriflagellate zygote. As in vegetative cells, regression occurs by the gradual shortening of the flagella and not by breakage. Shortening is easier to observe than in vegetative cells, because in a good mating it can occur more or less simultaneously in many cells. Shortening takes about 0 ·5 h and all 4 flagella shorten at the same rate. Electron microscopy (see below) shows that after the disappearance of the flagella, the basal bodies, striated connexions and microtubular roots also disappear from the cell. By 6 ·5 h after mating all trace of flagellar apparatus and associated structures have disappeared. Nuclear fusion has occurred and chloroplast fusion (Cavalier-Smith, 1970) is nearly complete; the diploid zygote is growing rapidly in size and the sticky outer layer of the zygospore wall is already laid down (Cavalier Smith, 1967). Throughout the remaining 18 h in the light, and the 5 days in the dark, needed for the maturation of the zygospore, basal bodies and associated structures are completely absent. During this period the complex multilayered zygospore wall is formed and the cytoplasm of the zygospore (including the chloroplast) undergoes drastic dediffercntiation.

After 5 days in the dark the now mature zygospores are transferred to fresh agar and placed in the light. They germinate synchronously and undergo meiosis, which begins about 6 h, and is complete about 9 ·5 –10 h after the transfer to fresh medium in the light (Lawrence & Davies, 1967). During germination the cytoplasm undergoes a dramatic redifferentiation which restores its structure to that characteristic of vegetative cells (Cavalier-Smith, 1967). The zygospore wall is gradually broken down during meiosis, and the 4 or 8 daughter cells (zoospores) acquire flagella and cell walls before the zygospore wall finally bursts (at about 12 –18 h after the beginning of germination) and liberates them to the exterior. The first traces of new basal bodies (see below) are visible at 6 h.

Thus during the sexual cycle of Chlamydomonas basal bodies and flagella regress completely in the early zygotes and only reappear 6 days later, during zygospore germination. This clearcut separation in time between regression and assembly (and also the relative synchrony of each process throughout the population) makes electron-microscopic study of the events much simpler and interpretation less liable to error than in the case of the vegetative cycle.

Fig. 5 summarizes the structure of a mature Chlamydomonas flagellum and basal body and the names of the parts. Further details are given by Ringo (1967 a, b), Cavalier-Smith (1967), and Hopkins (1970).

Electron microscopy shows that during mitosis the basal bodies remain in the cell, perpendicular to the plasma membrane, and still apparently attached to it, presumably by their transitional fibres. The flagellar axoneme and transitional region, however, have disappeared completely. Electron micrographs of cells from cultures where regression is occurring show no axonemes free in the cytoplasm. Therefore the flagella do not regress by the withdrawal of their axonemes into the cytoplasm. Since no visible fragments were liberated into the medium the mechanism is probably one of gradual breakdown. In vegetative cells at 3 h after onset of darkness, longitudinally sectioned flagella include some (Fig. 6) which may reasonably be interpreted as regressing. The membrane at their tips is sectioned perpendicularly, so it is unlikely that such flagella appear short simply because they curve out of the plane of the section. The 9 + 2 axoneme is apparently intact right to the tip of the flagellum, though the membrane ‘balloons’ away from it more frequently than at other stages, possibly because its connexions to the axoneme are already broken. Nor is it likely that these are broken flagella, since accidental breaks (e.g. during specimen preparation) usually occur (as in artificial deflagellation, Figs. 10 and 11; and see Randall, 1969) at the junction between the axoneme proper and the transitional region, just below where the central pair stops: this leaves a stump consisting only of the transitional region and basal body (Figs. 10, 11). Such stumps occur frequently in regressing populations; though many of them probably do represent an intermediate stage of lysis, at least some must be stumps left after breakage, since they are also found to a small extent when regression is not taking place.

In zygotes the much greater synchrony of flagellar regression results in a very much greater frequency of accurately longitudinal sections through short, presumably regressing, flagella. These reveal that the axoneme is always complete, except at its tip where amorphous material often occurs (Figs. 24, 27). This material is probably protein, derived from the disassembly of the axonemal tubules, which has not yet passed into the body of the cell. The intactness of the axonemes and the fact that the flagella can beat normally during shortening indicate that disassembly probably occurs only at the tip of the axoneme, and proceeds sequentially to the base of the flagellum. As in vegetative cells, no axonemes were ever seen free in the cytoplasm.

It seems that in zygotes the transitional fibres connecting the basal body to the plasma membrane, and the annular connexion between the transitional region and the flagellar membrane are both broken before the transitional region itself is dissolved, since the transitional region is often partly withdrawn into the cell (Fig. 25). However, these connexions are not broken until the 9 + 2 axoneme is completely dissolved. The transitional region, the striated connexions and the microtubular roots are then broken down and the basal body is left free in the cytoplasm some distance from the plasma membrane. Such basal bodies are often much shorter than their mature length (Fig. 28). Sometimes one basal body appears longer than its partner (Fig. 26). By 5 · 5 h after mixing the gametes only traces of basal bodies and/or roots remain. At 6 · 5, 7 · 5, 9 · 5, 12, and 24 h, I have been unable to detect any trace of basal bodies or related structures despite careful examination of many sections for each time point. At these times the cytoplasm is not markedly dense and one would have expected to see very short ‘probasal bodies’, like those reported by Renaud & Swift (1964) in Allomyces, if any were present.

The cytoplasm of mature zygospores (matured on agar) is so dense that it would be possible to overlook the presence of very short probasal bodies. However, any full sized basal bodies should have been visible, but none was seen. By 3 h after the beginning of germination, the cytoplasm has become noticeably less dense (Cavalier-Smith, 1967) but still no traces of basal bodies can be seen. By 6 h the cytoplasm has become much less dense and structures stand out with much better contrast; most cells still appear to lack basal bodies. I have seen an indistinct basal body in only 2 cells. At 7 h, although prophase of meiosis has begun (Cavalier-Smith, 1967), basal bodies are very rare. Even at 8 h (when some cells are in metaphase or anaphase and a few have completed the first meiotic nuclear division) basal bodies are rare, and usually much shorter than their mature length. They are commoner at 9 h, particularly at the newly formed surfaces of daughter cells. By 10 ·5 h, when most zygospores contain 4 highly granular daughter cells, most daughter cells have basal bodies.

At 9 h and 10 h (during the peak of basal body assembly), 9-singlet structures, consisting of 9 single microtubules surrounding a central ‘cartwheel’, occur frequently just below the cell surface. They occur in zygospores prior to cytokinesis (Fig. 30), in dividing cells (Fig. 32) and in daughter cells which have already grown flagella (Fig. 31). There can be no doubt that these structures are developing basal bodies and not some stage in mature or regressing ones, since they occur at a time of extensive basal body assembly following a period of 6 days when basal bodies were entirely absent. Similar 9-singlet structures are found in vegetative cells only at stages when new basal bodies are being formed. They cannot be confused with structures other than basal bodies since they have such a very well defined geometry, which corresponds closely with part of the structure of the proximal end of a mature basal body. Their overall diameter is about 0 ·16 μ m, i.e. the same as that of the A-fibres alone, and not that of a complete basal body (0·22 μ m).

The 9 tubules are connected to the 9 spokes in the same way as are the A-subfibres of the mature basal body (see below), which shows that the 9 singlets actually are A-subfibres. Their diameter and the subunit structure of their 5-nm-thick dense walls is also the same as those of the A-subfibres. Dense material is usually present between the 9 singlets but it usually does not take the form of discrete links. The structure of the hub and inner parts of the spokes more nearly resembles the diagrammatic interpretation given in Fig. 5 (i.e. a central ring and 9 radiating lines) than does that of mature basal bodies, because the hub and spokes appear intrinsically denser and have less obscuring background material.

Longitudinal sections of basal bodies are much less easy to interpret, but it is clear that at 9 and 10 · 5 h, many basal bodies are shorter than mature ones. It is often not possible to tell whether they consist of singlets or triplets, though sometimes basal bodies are seen which appear to consist of singlets throughout their length (Fig. 29). The ‘singlet’ basal body in Fig. 29 appears to be full length and to have a cartwheel hub and spokes throughout at least three quarters of its length, in marked contrast to mature basal bodies.

Most mid-light period cells contain only the 2 flagella-bearing basal bodies, but a few already have a third basal body. More cells have extra basal bodies at the beginning of the dark period, and 3 h later (when most cells are preparing to divide and about 13% have partly done so) a high proportion have one or two extra basal bodies. Since the 8 daughter cells derived from each parent contain altogether at least 16 basal bodies, and since I have seldom seen more than four (never more than five) basal bodies in one cell, many basal bodies must be made after the onset of the first division, i.e. between, during or after subsequent divisions.

At 3 and 5 h after the beginning of the dark period one observes transverse sections of basal bodies (in daughter cells) which consist of 9 singlet tubules surrounding a central cartwheel (Fig. 8), like the developing basal bodies seen in germinating zygospores, rather than the 9 triplets (Fig. 18) characteristic of the proximal cartwheel containing region of mature basal bodies. Occasionally, amorphous ‘disks’ similar in diameter to a basal body are seen (Figs. 9 – 11); although some may be grazing sections across the ends of mature basal bodies it is more likely that most are transverse sections at some level through developing ones, especially those that have the smaller diameter (0 ·16 μ m) characteristic of 9-singlet basal bodies.

In vegetative cells flagellar development does not begin until cell division is over and basal bodies of mature length and apparently mature structure are formed (Fig. 7). Developing flagella are distinguishable from regressing ones since they are attached to daughter cells still inside the mother cell wall. At the mid-dark period many daughter cells have flagella, some of which are shorter than usual (Fig. 9). Some of these show an only partially complete transitional region and no (9 + 2) axoneme, while others have a complete transitional region but a poorly organized (9 + 2) axoneme and dense amorphous material may be present inside the often-irregular membrane. If longer flagella are assumed to represent later stages in development, these observations indicate that the transitional region is formed before the (9 + 2) axoneme, which suggests that the flagellum is assembled sequentially, beginning at the base and growing by adding material to the tip. The existence of a disorganized region at the tip of developing flagella and not throughout their length may also be taken as evidence for growth at the tip rather than intercalary growth. Daughter cells at this stage frequently have more than two basal bodies, though the extra ones may be incomplete.

In germinating zygospores, as in vegetative cells, flagellar development does not begin until the basal bodies are fully mature, and are joined together by striated con nexions and associated with microtubular roots. Development is sequential, the transitional region being formed before the 9 + 2 axoneme proper. A few daughter cells already have flagella at 9 h and many do so at 10 ·5 h. From the beginning, new basal bodies are found close to the plasma membrane, but microtubular roots and striated connexions are generally not visible at the earliest stages when basal bodies appear. However, microtubular roots, at least, are sometimes seen before the zygospore undergoes cytokinesis. As in vegetative cell division, pairs of basal bodies, not yet joined by striated connexions, are frequent. After basal bodies have become joined in pairs, new ones form nearby, linked to them by amorphous material, so that groups of up to four basal bodies can occur as in vegetative cells (Figs. 29, 32).

The homogenization procedure breaks the flagellar axoneme away from the transitional region, confirming the evidence from artificial deflagellation (Figs. 10, 11; Randall, 1969; Rosenbaum et al. 1969), that the axoneme-transitional region junction is the weakest part of the flagellar apparatus. The 9 + 2 part of the flagellum is grossly disrupted, presumably partly by the homogenization itself and partly by the EDTA which solubilizes certain components (Jacobs & McVittie, 1970). The transitional region and basal bodies, on the other hand, are remarkably well preserved and more details of their fine structure are visible than in intact cells (Figs. 13–23).

The transitional region normally remains firmly attached to the basal bodies, but very occasionally becomes detached and after negative staining lies flat on the grid (Figs. 13 –23).

The 4·5-nm globular subunits comprising the outer doublets are then clearly seen (the circular A-tubule has about 13 subunits, and the C-shaped B-tubule about 9). In addition, it can be seen that the V-shaped projections from the basal cylinder, which form the ‘star’ pattern, and the link attaching the apex of the V to the A fibres, are also composed of globular subunits, somewhat smaller than those of the tubules themselves. The link from the A-tubule to the star points probably consists of 3 sub-units, the third of which forms the apex of the ‘star’. Each arm of the V consists of 5 subunits in addition to this common apical one. Positively stained sections of transitional region (Figs. 14, 15) show that at least some B-tubules have a hitherto unnoticed dense ‘knob’ (comparable in size to one, or at most two, microtubular sub-units), which projects into their lumen from the wall diametrically opposite to the A-tubule. Jacobs & McVittie (1970 and personal communication) have seen identical ‘B-tubule knobs’ in isolated flagella of C. reinhardii. In their pictures only 3 out of the 9 B-tubules appear to have knobs; a pair of adjacent knob-bearing doublets is separated from the third one by two groups of 3, apparently knob-free, doublets. It is unlikely that these B-tubule knobs are merely an artifact of isolation in EDTA since 3 such knobs with the typical arrangement can clearly be seen in Ringo’s (1967b) fig. 7 (and fig. 9), though usually they are not clearly visible in intact cells. In the transitional region there sometimes appear to be more than 3 B-tubule knobs (e.g. about 5 in Fig. 15). I have seen at least 2 B-tubule knobs in triplet basal bodies, so they are present throughout the basal-body-flagellar axoneme.

Further details can also be seen in the dense ‘annular connexion’ which joins the transitional region doublets to the flagellar membrane. Each of the 9 doublets bears a dense 10-nm ‘doublet outer projection’, which projects radially outwards from the junction point of the A and B tubules (Figs. 4, 13, 15). In longitudinal sections (Fig. 17) densities visible at 25-nm intervals along the length of the outer doublets are probably best explained as end-on views of doublet outer projections. The dense material of the annular connexion itself is not homogeneous, but contains densities placed in pairs opposite each doublet outer projection (Figs. 14, 15).

Sections through the apex of isolated basal bodies (Fig. 19) show that each transitional fibre attaches at its broader end not simply to the B-tubule of each triplet (Ringo, 1967a), but also to the A-fibre, and to the C-fibre of the adjacent triplet. The tapering ends often remain securely attached to fragments of plasma membrane, so that there can be little doubt that the function of the transitional fibres is to anchor the basal body firmly perpendicularly to the plasma membrane.

The lumen of the basal bodies often contains the cartwheel structure at its proximal end (Figs. 18, 20) and amorphous material of medium density at its distal end, as in intact cells. Frequently, however, this material is lost (Fig. 21), and because the lumen is empty a series of projections spaced at 15-nm intervals on the inner side of the A-tubule stand out clearly (Fig. 21). These projections are essentially comparable to the dense ‘feet’ seen in mammalian centrioles (Stubblefield & Brinkley, 1967) and also in Paramecium (Dippel, 1968), so I shall call them ‘A-tubule feet’. In their orientation and attachment to the A-tubule they resemble the radial ‘spokes’ of the outer doublets of flagella (Hopkins, 1970), but are only 12 nm long instead of 33 nm. Like the flagcllar spokes, they consist of a filament terminated by a knob. Ringo (1967a) interpreted these knobs as A-C connexions. This cannot be the case, because A-C connexions would not project into the lumen in median longitudinal sections such as Fig. 22, and can only be seen in tangential longitudinal sections which pass through 2 outer fibres (Fig. 17). In transverse sections through the distal end of the basal body the ‘feet’ tend to be obscured by the dense material in the lumen, but are normally visible in the cartwheel region. In monkey oviduct centrioles Anderson & Brenner (1971) have interpreted the links between the A-tubules and terminal blob of the cartwheel spokes as sections through an ‘A-tubule attachment sheet’ instead of separate filaments. However, the very close resemblance in detailed structure between the basal bodies of Chlamydomonas and Paramecium (Dippell, 1968) and Chinese hamster centrioles (Stubblefield & Brinkley, 1967) and Anderson & Brenner’s actual micrographs encourages the expectation that in mammals, too, these links will eventually turn out to be filaments rather than sheets.

Thus there arc 3 quite distinct kinds of inwardly pointing radial projections from the A-tubule: (1) the A-tubule feet; (2) the A-tubule-star point links in the transitional region; and (3) the spokes in the 9 + 2 region of the flagellum. It is conceivable, however, that the proximal part of all 3 kinds of projections (corresponding to the 3 globular subunits of the A-tubule-star point links) is homologous or even identical.

The structure of the cartwheel (Figs. 4, 16, 18) is the same as in mammals (Stubblefield & Brinkley, 1967) and in Paramecium (Dippell, 1968). Each spoke is terminated by a knob which is distinct from, but attached to, the knob of an A-tubule foot; each spoke is somewhat thicker at the hub end and has another thickening near its mid-point.

Basal bodies in homogenates usually remain attached together in pairs by the distal and 2 proximal striated connexions; this supports the idea that these connexions have a structural role. The dense plate underlying the distal striated connexion is also present (Figs. 21, 22) but is often (like parts of the distal connexion) at least partially dissolved. The 4 microtubular roots are always broken at some point, but part of each remains connected to the basal bodies by amorphous material (Fig. 23). Daughter basal bodies are also physically attached to the old basal body complex by similar amorphous material of medium density (Fig. 16). A very dense material is associated with the proximal end of the basal bodies. This material is much more pronounced on the side of each basal body opposite to that where the distal connexion and micro-tubular roots attach (Figs. 21, 22); the triplets on this side are longer than on the other, but are often indistinct because they are embedded in the dense material.

Johnson & Porter (1968), by using the phrase ‘dissociation of basal bodies from flagella’, imply that Chlamydomonas reinhardii flagella are lost from the cell by breakage or detachment prior to cell division. My observations by light microscopy show that this is not the case, but that the flagella regress gradually by shortening (as Lewin (1952) observed in C. moewusii), at least until the flagella are too short to project from the flagellar tunnels, and therefore disappear from view. I suggest that this shortening is brought about by the solubilization of the axoneme proteins (and possibly also the membrane) by a sequential process starting at the tip of the flagellum and ending at the transitional region.

What happens to the transitional region itself seems to be more variable. In the synchronous cultures studied what usually happens is that the sequential dissolution of the axoneme continues through the transitional region and dissolves it. The connexion between the base of the flagellar tunnel and the base of the flagellar membrane is then broken (this connexion must be fairly strong since it resists homogenization in some cells at least). Fragments of flagellar membrane remain stuck in the flagellar tunnel, and may persist even after the mother wall is cast, but usually the tunnel is empty. In Johnson & Porter’s (1968) ‘logarithmic’ cultures, however, it seems that the entire transitional region (and sometimes some of the 9 + 2 axoneme also) remains trapped in the tunnels. I have also seen an entire (or more usually a partly degraded) transitional region (but never also a 9 + 2 axoneme) trapped in the flagellar tunnels of cells recently washed off agar into liquid medium.

The reason for this variability may lie in the variable timing of the 90⁰ rotation of the protoplast with respect to the cell wall. This rotation probably breaks the connexion between the cell membrane and the base of the flagellar tunnel. If the rotation does not begin until regression is complete, the flagellar tunnels in the mother cell wall will be empty, or have only membrane fragments, as in my synchronous cultures. If rotation begins before regression is complete, short lengths of flagella will be broken away from the basal bodies and remain in the tunnels (Johnson & Porter, 1968).

Regression in developing zygospores (which possess no cell wall or flagellar tunnels to complicate matters) very clearly occurs by gradual shortening, first of the axoneme and then the transitional region. The withdrawal of the transitional region into the cytoplasm before it is lysed, frequently observed in developing zygospores, was not observed in vegetative cells. I suggest that this is because the transitional fibres joining the basal body to the plasma membrane remain unbroken in vegetative cells (the basal body remains attached to the plasma membrane throughout cell division), but are broken down in zygospores.

Flagellar regression before cell division, and in developing zygospores, takes about 0-5 h, like regression following unilateral deflagellation (Rosenbaum et al. 1969). It is likely that the same mechanism is involved in all three cases. The evidence that flagellar protein is conserved by the cell during regression following unilateral de flagellation (Coyne & Rosenbaum, 1970) is consistent with the light -and electron microscope evidence reported here for regression by shortening and not by breakage.

Basal bodies appear to be absent for 6 days in Chlamydomonas zygospores yet new basal bodies appear during germination. Thus Chlamydomonas can be added to the growing list of organisms in which basal bodies appear to be entirely absent for some part of the life cycle (Fulton, 1971). It is clear that basal bodies can be formed in the complete absence of existing mature basal bodies. Therefore, the widespread current notion (e.g. see Pitelka, 1969; Gibor & Granick, 1967; Jinks, 1964; Wilkie, 1964) that basal bodies are ‘self-replicating’ is, in its simplest and most obvious sense, simply untrue for many cells.

Since, as Fulton (1971) and I (Cavalier-Smith, 1967) have argued, there is no good evidence cither for the presence of DNA in basal bodies or for the self-replication or genetic continuity of basal bodies, it would be better if authors ceased to use these terms with respect to basal bodies and centrioles; Beisson & Sonneborn (1965) and Sonneborn (1970), point out that even if DNA were present in basal bodies this would not explain their results. Indeed their results, and comparable ones for Tetra hymena (Nanney, 1968), and the electron-microscopic observations on basal body development in numerous organisms can be explained without postulating self-replication. All these results are compatible with the coding of all basal body proteins by nuclear genes and their synthesis on ordinary cytoplasmic ribosomes. Moreover, there is no reason to postulate that old basal bodies contribute any information about the pattern of assembly of new ones into the 9-triplet plus cartwheel structure. This can be explained by a sequential assembly process (see below) with no necessity for any kind of template.

All that the electron-microscopic and experimental observations require is that there is a mechanism for ensuring that the assembly of new basal bodies begins only at a definite location and with a definite orientation of the new basal body. Such mechanisms must exist for many different organelles (cf. Tucker, 1971) in most eukaryotic cells, but their molecular basis is a mystery. Where ‘old’ basal bodies are present (e.g. in Chlamydomonas vegetative cells) amorphous material close to existing basal bodies appears to serve as a nucleating centre for the first steps of assembly (Dippell, 1968). In Chlamydomonas my observations on homogenates show that this material is actually attached to both new and old basal bodies. In some cases where new basal bodies are not formed adjacent to old ones fibrous or granular material seems to act as a nucleation centre (e.g. Anderson & Brenner, 1971), but in others no obvious material is present (Pickett-Heaps, 1971). It is clear that in germinating Chlamydomonas zygospores new basal bodies are always formed just below the plasma membrane. Clearly something must ensure this precise location and orientation. Even though basal bodies themselves are absent, it is conceivable that the amorphous material seen in vegetative cells remains, perhaps attached to the plasma membrane throughout the sexual cycle.

Whatever the mechanisms ensuring the precise location and orientation of new basal bodies and other organelles (referred to by Sonneborn as ‘cytotaxis’), they alone would be sufficient to explain the inheritance of cortical patterns in ciliates. The simplest type of mechanism is the sequential assembly of molecules, starting with the attachment of the first components to a site with a specific location and orientation on a relatively rigid pre-existing structure. In this way the location and orientation of the new organelle would depend on 2 quite separate factors. One is the geometry and chemical specificity of the component molecules of both the new and of the old structure; this structure is probably coded by normal DNA genes. The second factor is simply the orientation that the nucleation site happens to have with respect to the other cell structures at the time of assembly. If this orientation is experimentally altered (Beisson & Sonneborn, 1965) then such alteration will automatically be inherited, because of the inherent specificity of the molecules and the assembly process.

Thus, in the case of cortical inheritance it is as correct to speak of the cortical pattern, or even individual kineties, as ‘showing genetic continuity’ or ‘self-replication’ as it is for a DNA molecule. But there is no more justification for saying that elements of the pattern such as individual basal bodies, kinetodesmata and so on, are ‘self-replicating’ than there is for saying that deoxyribonucleotides are self-replicating. Of course, some elements of a cortical pattern may indeed, in some sense, be ‘self replicating’, as mitochondria are, but this requires independent evidence (which is lacking for basal bodies) and is neither necessary nor sufficient for the inheritance of the overall pattern.

The observations on homogenates show that new basal bodies are firmly attached to the old pair by ‘amorphous’ material. I suggest that this material serves as a nucleus for the initial stages of assembly and determines the site and orientation of new basal bodies through its geometry and chemical specificity.

It is clear that the 9-singlet structure represents a definite stage in the development of basal bodies. In strain 32 D all such structures contained a well defined central cart wheel, which led me to postulate (Cavalier-Smith, 1967) a sequential mechanism for the assembly of basal bodies: (1) the tubular hub of the cartwheel might be assembled first and (2) each of its 9 subunits visible in transverse sections (Gibbons & Grimstone, 1960) could serve as a site for the assembly of a single spoke; (3) one A-tubule is assembled at the end of each spoke to produce the observed 9-singlet structure; (4) the B and C tubules are added, and connexions made between A and C tubules; and (5) the basal body elongates, transitional fibres form and attach it to the plasma membrane. It is clear that this sequence occurs only in part in Paramecium, since Dippel (1968) found that the A-tubules are assembled before the cartwheel. In C. reinhardiistrain 89+ I have once seen a 9-singlet structure lacking a cartwheel; this may have been a section through a developing basal body whose A fibres had already lengthened beyond a cartwheel region not included in the section, but it could equally well mean that in Chlamydomonas also the cartwheel is not formed first but is added after the formation of a ring of A-tubules.

Some of the dense disk-like objects seen (Figs. 9 – 11) in the same position as developing basal bodies may be still earlier stages in basal body assembly, corresponding to the morphologically similar ‘generative disks’ which are the first basal body precursor to appear in Paramecium (Dippell, 1968). Similar structures have also been observed during basal body formation in monkey oviduct (Anderson & Brenner, 1971) and in chick trachea (Kalnins & Porter, 1969).

I have not seen developing basal body outer fibres consisting of doublets, though developing basal bodies consisting partly of singlets and partly of triplets do occur. This suggests that the C-tubule is added to the B-tubule immediately the B-tubule is itself added to the A-tubule. Johnson & Porter (1968, fig. 19) show a ‘basal body’ consisting of 8 doublets and 1 triplet; because it lacked transitional fibres they considered it to be a developing basal body and not simply a section through the junction of the basal body with the transitional region. However, the absence in their micrograph of transitional fibres is to be expected (apart from the fact that they sometimes fix poorly), since they attach to the extreme upper end of the triplet region of the basal body and not to the doublet fibres of the transitional region. The closeness of the flagellar membrane except on the side facing the triplet, is also characteristic of sections through the extreme base of the transitional region.

Although my observations are less complete than those of Dippell (1968) for Paramecium, they are sufficient to show that the sequence of assembly differs in at least one respect in Chlamydomonas: the cartwheel is assembled in Chlamydomonas before any B-fibres are complete, whereas in Paramecium B-fibres and at least some C-fibres arc formed before the cartwheel appears. This means that one should not expect to find an identical sequence of assembly in all organisms, and implies that the detailed mechanisms may differ. Developing basal bodies in Tetrahymena (Allen, 1969) and Tetraspora (Pickett-Heaps, 1973) also go through a 9-singlet stage.

It is clear that in Chlamydomonas as in many other organisms (e.g. Dippell, 1968; Dingle & Fulton, 1966; Allen, 1969), flagellar growth does not begin until the basal bodies are completely assembled and attached to the plasma membrane of daughter cells by their transitional fibres. Furthermore, flagellar growth does not begin until the 2 basal bodies are at their characteristic angle to each other. Presumably at this stage the distal striated connexion, dense plates and 2 proximal striated connexions are also completely assembled, though one cannot be sure of this without complete serial sections of several daughter cells with stubby flagella. The 4 microtubular roots are assembled (at least in the immediate vicinity of the basal bodies) before flagellar growth begins.

My observations indicate that the transitional region is assembled first, before the axoneme proper. The presence of amorphous material at the tip of short, presumably growing, flagella suggests that the axoneme itself grows sequentially at the tip, which is supported by autoradiographic evidence (Rosenbaum et al. 1969). Indeed, my observations are consistent with the view that the whole flagellar-basal body axoneme complex is assembled in a base-to-tip sequence, beginning at the proximal end of the basal body and ending at the tip of the flagellar axoneme. Flagellar growth after artificial dcflagellation (Randall, 1969; Rosenbaum et al. 1969) appears to be essentially the same as in daughter cells after cell division, except that after deflagellation the transitional region remains and does not have to be formed anew.

The manner of association between the flagellum and the cell wall is of especial interest in Chlamydomonas. The proximal regions of mature flagella lie in tunnels in the cell wall and their membranes are fairly firmly attached to the base of the tunnels at the level of the proximal end of the basal cylinder in the transitional region. It appears that the flagella grow a certain length before the cell wall and tunnels are formed. It may well be that the tunnel is laid down around the flagellum. Unfortunately, mutants completely lacking a flagellar axoneme and transitional region have not yet been found (Randall et al. 1967; McVittie, 1972); such mutants would be very interesting since they could show whether the flagella serve in any way as a template or mould for the formation of the flagellar tunnels or whether their structure is an intrinsic property of the cell walls. Even if the latter is true, the flagella must presumably play a role in determining where in the cell wall tunnels are formed (unless one postulates that the cell rotates within its new wall, and the flagella somehow ‘find’ the tunnels and wriggle into them).

Basal bodies are not found directly at the spindle poles in C. reinhardii, though sometimes one or more is not far from one of the poles (Fig. 12; and Cavalier-Smith, 1967; Johnson & Porter, 1968). Indeed, they remain attached to the plasma membrane throughout mitosis and cytokinesis, whereas the spindle poles are usually 1 μ m or more from the cell surface. Basal bodies should therefore not be called ‘centrioles’ in Chlamydomonas.

It is unclear whether the 2 daughter cells each receive one old and one new basal body, or whether the old basal bodies remain joined together and both pass to the same daughter cell. Nor is it clear whether every daughter cell necessarily receives a basal body in cases where 2 or more successive cell divisions occur. However, in some cases, the first cell cleavage clearly passes exactly between 2 pairs of basal bodies, so that 2 do go to each daughter cell.

Since in predivision cells (as shown above) the 2 new basal bodies and the 2 old ones all appear to be attached together, it follows that at least some of these links must be broken prior to cytokinesis to allow separation, whatever the manner of distribution of new and old basal bodies to daughter cells. Moreover, since the old basal bodies are linked to each other and to the 4 microtubular roots which underlie an extensive area of the cell surface, at least some of the links involving only ‘old’ organelles must be broken to allow the cleavage furrow to pass. Thus the ‘basal body cycle’ in vegetative cells involves, not merely the assembly of new organelles and the movement of both new and old organelles, but also at least a partial alteration of existing structures; it would be very difficult to establish the extent of such an alteration; it might be as little as the temporary breakage of links between old basal bodies and old roots, and between old and new basal bodies, or as great as the complete lysis of all existing roots, striated connexions and the dense plate.

Daughter cells often have 4 basal bodies but grow only 2 flagella. Such cells never have more than one distal striated connexion or 4 microtubular roots. This indicates that the number of these structures is controlled much more rigidly than that of basal bodies. This control of flagella number seldom breaks down in diploid cells, where I have only very rarely observed 3 or 4 flagella at the anterior end; but Starling (1969) has shown that 2-5% of diploid cells have 3 flagella (and occasionally more). In the ‘twin’ C. reinhardii mutant (Warr, 1968) the number of nuclei and of pairs of flagella are closely correlated. Interphase cells have a fibrous area of cytoplasm (the ‘fibrous band’ (Cavalier-Smith, 1967)), which is free of ribosomes and links the ribosome-free fibrous matrix surrounding the basal bodies and associated structures to the nuclear envelope. Warr (1968) has suggested that the fibrous band could be the physical basis for the correlation between nuclear and flagellar number. The fibrous band is often visible during mitosis joining the nuclear envelope in the region of one of the spindle poles to the basal bodies which remain at the cell surface.

The observations on synchronous cultures and sexual stages were included in a thesis submitted in part fulfilment of the requirements for the Ph.D. degree of London University (Cavalier-Smith, 1967); the observations on homogenates were made during the tenure of a Damon Runyon Cancer Research Fellowship at the Rockefeller University, New York. I thank Professors Sir John Randall and David Luck for encouragement and support.