Motile Characteristics of 9+2 and 9+1 Flagellar Axonemes of Crithidia Oncopelti

The demembranated flagellum of Crithidia oncopelti can be reactivated after removal of one of the central microtubules, together with its associated central structures and radial spokes (Marchese-Ragona & Holwill, 1980; Marchese-Ragona et al. 1983). When both central microtubules have been removed from this flagellum, reactivation has not proved possible. In two cell types, Anguilla sperm (Baccetti et al. 1979; Gibbons et al. 1985) and Chlamydomonas mutants (Brokaw & Luck, 1985), intact flagella with nine peripheral doublets but containing no central microtubules are known to be motile. Other motile flagella lacking the central pair, but with different peripheral microtubular arrangements (3+0, Prensier et al. 1980, and 6+0, Schrével & Besse, 1975) have also been observed. The flagellum of C. oncopelti is the only flagellum from which central structures can be removed chemically without destroying the capacity for movement. The preparation thus has great potential for the elucidation of the role in motility of the central apparatus of flagella.

Although the central complex does not form an integral part of the mechanism required for axonemal motility, observations on Chlamydomonas flagella and Anguilla sperm suggest that this structure determines the symmetry of bending, being associated with the sensitivity of the organelles to calcium ions (Brokaw et al. 1982; Gibbons et al. 1985). The motile response of an axoneme to the removal of a central microtubule and radial spokes is therefore of importance, since it may provide more information about the function of these structures in the intact axoneme. If the active binding of radial spokes to the central complex is a source of resistance leading to the conversion of sliding into bending in the axoneme (Warner & Satir, 1974), a change in wavelength, amplitude or frequency might be expected to accompany the degradation. Mechanical considerations alone suggest that the removal of some of the axonemal structure would lead to an observable modification of the flagellar beat parameters, if only because the gross elastic properties of the system would be altered. Since flagellar bending involves the conversion of chemical energy into mechanical energy, a study of the chemical kinetics of the system might be expected to reveal functional changes in the dynein-tubulin interaction caused by degradation of the active regions of either, or both, of these components.

To assess the role of the central complex in motility, we report here the results of a series of experiments in which the wave parameters of axonemes with intact central complexes are compared with those of axonemes where one of the central microtubules is degraded. Our results support earlier conclusions (Marchese-Ragona & Holwill, 1980; Marchese-Ragona et al. 1983) that the central complex is not necessary for motility and has no significant effect on wave shape.

Cells from a 5-to 6-day-old culture of Crithidia oncopelti were harvested by gentle centrifugation and rinsed three times in a wash solution containing 100molm⁻³ (= lmmoll⁻¹) MgCl₂ adjusted to pH7·0 with mercaptoacetic acid (thioglycolic acid). Cells were demembranated using 0·1 % (w/v) of the non-ionic detergent Nonidet P42 in the wash solution. The demembranation was performed with 1·0 cm³ of the Nonidet solution for every 1·0 mg of cell protein (measured by the method of Lowry et al. 1951), at a temperature of 18°C for 25 s. The preparation was then cooled rapidly to 0°C and stored at this temperature for various intervals (see Results). To determine the effects of divalent-cation chelating agents on motility, EGTA or EDTA at a concentration of 5 mol m⁻³ was added to the wash, extraction and reactivation solutions.

Axonemes were reactivated at different ATP concentrations, with the magnesium concentration maintained at 4·5 molm⁻³. At the ATP concentrations used (within the range 10⁻³ to 10⁻¹molm⁻³) an almost constant proportion (approx. 80%) of the ATP existed as MgATP²⁻ while the magnesium ion concentration remained essentially constant (Storer & Cornish-Bowden, 1976).

The reactivated samples of axonemes were observed using dark-field optics on a Zeiss WL research microscope, and flagellar images were recorded using a Jackson high-speed (100 frames s⁻¹) video system (Jackson Electronics, Newark, Nottinghamshire, UK). Beat frequencies above 8Hz were measured using a stroboscopic technique, whereas those below 8 Hz were determined by counting the number of waves propagated by a flagellum in 10 s. Suitable flagellar images were traced from the video monitor onto transparent plastic, and Cartesian coordinates generated using a bit pad. Samples remained active for a period of about 5 min, in which time 10–20 organisms were examined.

Samples were prepared for thin-section electron microscopy by fixation in 2·5 % (v/v) glutaraldehyde in 100 mol m⁻³ sodium phosphate buffer at pH7·0 for 2 min. The cells were then washed three times in phosphate buffer, postfixed with 1·0 % (w/v) O_SO₄ in buffer, washed a further three times in buffer solution, and suspended in a fibrin clot (Furtado, 1970). Dehydration was performed in a graded series of ethanol, and was followed by embedding in SPURR resin (EM scope Laboratories Ltd, Kingsworth Industrial Estate, Wotton Road, Ashford, Kent, UK). Silver to gold thin sections were collected on an uncoated 400-mesh copper grid and stained in the dark for 60 min in a 50:50 ethanol: water mixture containing 1·0% (w/v) uranyl acetate, followed by Reynolds lead citrate for 3·0min. Specimens were examined on a transmission electron microscope (Philips 300 or Jeol 100C) operated at 80 kV with a 30 μm objective aperture. All micrographs were taken near optimum defocus.

Storage of Crithidia cells in demembranation solution at 0°C results in progressive degradation of the axoneme (Marchese-Ragona & Holwill, 1980) to an extent which depends on storage time and on concentration of detergent relative to protein mass (Marchese-Ragona et al. 1983). At higher concentrations than the optimum (see Materials and methods) axonemes are degraded more rapidly, but the resulting preparation has poor reactivation qualities.

When extraction was performed in the absence of a chelating agent (i.e. the standard demembranation solution), solubilization of the central microtubules proceeded as shown in Table 2 and Fig. 1A,B. The axonemes exhibited four distinct axonemal configurations which could be classified according to the pattern of microtubules remaining. The four categories occurred in a specific temporal sequence starting with the 9+2 configuration of the intact axoneme (Fig. 1A). Thq sequence of degradation began with the loss of one central microtubule to give a 9+1 microtubule configuration (Fig. 1B) followed by the removal of the second central microtubule to leave a 9+0 microtubule configuration (Fig. 1C). Most of the B microtubule of each doublet was then solubilized to produce a cylinder of nine A microtubules each connected to a small portion of the B microtubule; the cylinder was presumably held together by nexin links (Fig. ID). In some instances, a few A microtubules also showed signs of degradation.

After a 10 min storage period, 74 % of axonemal cross-sections examined by electron microscopy retained the 9+2 microtubular configuration, so samples stored for between 0 and 10 min were used as the control sample of 9+2 axonemes. After 60 min of storage, 72 % of axonemal cross-sections had a 9+1 microtubular configuration, hence samples stored for between 55 and 70 min were used as the 9+1 sample of axonemes. The percentage of cells in a sample which could be reactivated upon the addition of ATP was 90 % after 10 min of storage and 80 % after 60 min of storage. Motility observed at the times specified is thus characteristic of axonemes of the two structural types. After storage times longer than 60 min, the percentage of axonemes with the 9+1 structure increased, but the proportion of motile cells in the sample decreased.

If the wash and extraction were performed in the presence of EGTAor EDTA, the rate of microtubule solubilization was reduced to about one-tenth of its value in the absence of these agents, so that the time required to produce a sample containing predominantly 9+1 axonemes would be in the region of 10 h rather than 1 h. Samples prepared in the presence of EGTA or EDTA also retained the capacity for reactivation for a time which was about 10 times that of samples prepared without the chelating agents. However, the chelating agents induced aggregation of cells to such an extent that it was difficult to observe the behaviour of individual organisms. Samples treated with EGTA exhibited the property of tip-to-base wave propagation, provided they were reactivated with a solution containing the chelating agent within about 15 min of detergent treatment; after this period the direction of wave propagation was from base to tip. A similar effect was seen with EDTA, but the time for which tip-to-base wave propagation was sustained was considerably shorter than for EGTA. Since the time required to degrade one central microtubule was significantly longer than 15 min, no additional information regarding motility could be obtained by adding chelating agents, so all samples were prepared in the absence of EGTAand EDTA, thereby effecting a considerable saving in time.

Between 5 and 10 % of the axonemes had structures which fell between two of the four distinct categories of micro tubular configuration. In these cases, axonemes were assigned to the higher level of structural integrity; for example, axonemes with one complete central microtubule and only part of the second microtubule were classified as 9+2 axonemes. Axonemes with only part of one central microtubule were classified as 9+1 axonemes. The proportions of axonemes which fell into these classes after various times are presented in Table 2. This rule of structural assignment, together with the definition that an axoneme would only be recorded as motile if it propagated bends along its entire length,ensures that, on comparing structure and motility percentages at a given time, the proposition that axonemes lacking a central microtubule are motile can be addressed in the most unfavourable conditions.

The average percentage of axonemes with a 9+1 microtubular configuration after a 1 h treatment period was 71·7 % (maximum, 79·3 % ; minimum, 65·7 %) on six separate preparations. In all the preparations incubated for 60 min, 80–90 % of axonemes, as estimated by visual counting, were capable of propagating bends along the total length of the flagellum when reactivated. All transverse sections of 9+1 axonemes examined had lost the central sheath projections and between three and five of the radial spokes that were associated with the solubilized central microtubule. The complement of dynein arms on the peripheral doublets had been reduced by about 10 %.

In both groups of cells two distinct types of waves were observed. On freely swimming cells the axonemes propagated three-dimensional waves which became two-dimensional when the organelle came in contact with the slide.

Both 9+2 and 9+1 samples showed a variation in waveshapes among cells, but throughout the range of ATP concentrations used all reactivated axonemes conformed roughly to a basic waveform of between one-and-a-half and two wavelengths per flagellum with wave parameters as shown in Table 3. An objective comparison of the two-dimensional waveforms on the two structural types of flagella, at an ATP concentration of 25 mmol m⁻³, was made using Fourier techniques (see Materials and methods). For the 9+2 sample, 50 flagella centrelines were traced from nine different cells, giving a total of 354 quarter-wave sections whereas, for the 9+1 sample, 28 flagella centrelines were traced from seven cells for a total of 202 quarter-wave sections. From these data the means and standard errors of the first seven coefficients of the best fitting Fourier harmonic series were calculated (Table 4).

Beat frequencies of reactivated demembranated cells were analysed as mean values over 5-min periods elapsed since demembranation, at the different ATP concentrations (Table 5). Representative data from Table 5 show a slight decrease in beat frequency with time for each ATP concentration (Fig. 2). For the period when axonemal structures were predominantly 9+2 (T₁ in Table 5) or 9+1 (T₂ in Table 5), the reciprocal of beat frequency showed a non-linear relationship with the reciprocal of ATP concentration (Fig. 3), approximating one linear region at concentrations between 5 and 20 mmol m⁻³ and another linear portion between 20 and 250mmolm⁻³.

Our results indicate that one of the central microtubules of Crithidia flagella is chemically more stable than the other; a property found in several organisms (Chlamydomonas reinhardii, Jacobs et al. 1968; Jacobs & McVittie, 1970; Tetrahy-mena cilia, Chasey, 1969; Aequipecten irradians cilia, Linck, 1973a,b; spermatozoa of the squid Loligo pealei, Linck et al. 1981). In Crithidia we have not been able to identify unequivocally which is the more stable of the two central microtubules, since there is no structural marker to allow the determination of microtubule number; in organisms in which the remaining tubule has been identified, it has proved to be Cl (notation of Linck et al. 1981).

The reduced rates of axonemal degradation in the presence of EGTAor EDTA suggest that divalent cations, in particular Ca²⁺, play an important role in the degradation process. Since calcium ions are known to prevent polymerization of brain tubulin (Scheele & Borisy, 1979) and have been shown here to increase the rate of axonemal degradation, it is possible that the concentration of these ions governs the rate of depolymerization of microtubules of the Crithidia axoneme. Microtubules of Chlamydomonas are also solubilized by exposure to some detergents (Witman et al. 1972), an effect which could possibly be attributed to the presence of calcium ions in the preparation.

The small reduction in frequency that occurs with storage time can be explained in terms of the degradation of dynein arms over this period. Over the initial 1-h period of treatment, there is a reduction of about 10 % in the number of dynein arms (Table 2), and a similar percentage decrease in beat frequency (Fig. 2), suggesting that the beat frequency is related to the number of functional dynein arms in the axoneme. This is consistent with the reduction in frequency which parallels the loss of dynein arms in reactivated sea urchin sperm (Gibbons & Gibbons, 1973). Similar results are reported for mutant Chlamydomonas flagella having missing or modified outer arms (Brokaw & Luck, 1985; Kamiya & Okamoto 1985); the beat frequency for these organelles is about 50% of that of wild-type cells, a reduction corresponding to the percentage loss of dynein arms.

The Fourier coefficients calculated for 9+2 and 9+1 demembranated axonemes (Table 4) represent averaged quarter-waves for the two structural types. Inspection of corresponding coefficients for the two types shows that, within the limits of experimental error, the waveshapes are the same, indicating that the structural degradation had little effect on the mechanism responsible for wave formation and propagation. In particular, this result suggests that the central complex does not make a significant contribution to the passive elastic resistance of the system. This is consistent with recent studies (Holwill, 1989) which indicate that structures other than the microtubules provide a significant resistance to bending.

Although the waveshapes for flagella of the two structural types appear to be identical, comparison of amplitudes and wavelengths (Table 3) suggests that the more degraded axoneme bears smaller waves. Further experimentation is needed to establish whether the difference is significant.

Apart from b₁, the Fourier coefficients for both the 9+2 and 9+1 axonemes are close to zero, indicating that the flagellar waves are more closely matched by sinegenerated waves, or meanders, than by sinusoidal or arc-line waves (compare the coefficients in Tables 1 and 4). The differences between a sine-generated curve and the meander are slight, and would not be detected using the present technique. The waveshape of reactivated flagella is therefore different from that for in vivo flagella on forward-swimming Crithidia (Johnston et al. 1979). In this latter case the bends were characterized as arc-line, with between 76 % and 92 % circular arc; some minor departures from the ideal arc-line curve occur as a result of the mechanical properties of the system. Preliminary studies suggest that waves propagating from base to tip on intact Crithidia flagella also appear to be arc-fine in form (M. R. Hirons & M. E. J. Holwill, unpublished). A possible explanation for the differing shapes exhibited by intact and reactivated flagella is that the membrane, which is absent in the reactivated cells, exerts some degree of control over the waveshape. Current theoretical models of bend formation and propagation relying only on the mechanical properties of the flagellum and its environment can generate flagellar beat patterns that are an excellent first-order approximation to those observed experimentally (e.g. Brokaw, 1985; Sugino & Naitoh, 1982, 1983). It may be, however, that subtle modifications to the flagellar waveshape are controlled by the membrane. Although the doublets are linked to the membrane in Crithidia oncopelti (Marchese-Ragona, 1981), there is little evidence that the membrane could modify the flagellar wavelengths by direct mechanical means (Sleigh & Silvester, 1983). It is more reasonable to suppose that the waveshape is governed to some extent by an active process under cellular control, with ionic fluxes being mediated by the membrane. For convenience of observation and measurement the waveshapes of reactivated 9+2 and 9+1 axonemes were studied at relatively low ATP concentrations. It is possible that higher ATP concentrations may alter the waveshape, but observations on glycerol-extracted Crithidia flagella at high and low levels of ATP indicate that there is little alteration in this parameter (Douglas & Holwill, 1972). The meander shape observed on demembranated axonemes suggests that the passive elastic properties of the organelle, which are provided mainly by the peripheral microtubules, play an important role in determining the bend patterns in this case (e.g. Johnston et al. 1979).

The difference in waveshape between in vivo and demembranated Crithidia flagella is at variance with the finding of Weaver & Hard (1985) that the waveshapes of intact and demembranated newt lung cilia were the same. However, these authors based their conclusions on angular measurements of the various phases of the beat cycle, which may not reveal the subtle changes in waveshape reported here. A study of ciliary bend patterns using the Fourier analysis method would be of interest in this regard.

In Chlamydomonas flagella, the radial spoke/central sheath interaction is believed to be involved in the control of wave symmetry (Brokaw et al. 1982; Brokaw & Luck, 1985), since mutants lacking these structures appear incapable of propagating highly asymmetric bends. The central complex may therefore be the site of calcium ion action, since variations in Ca²⁺ levels produce changes in wave symmetry of Chlamydomonas flagella. This conclusion is supported by the observation that Anguilla sperm, which naturally lack the central complex, also have no calcium sensitivity (Gibbons et al. 1985). For Crithidia, the wave symmetry does not appear to be regulated by the radial spokes, since it is unchanged when some of the spokes, together with the associated central microtubule, appear to be disrupted. However, the response of Crithidia flagella to changes in Ca²⁺ concentration is not associated with wave symmetry, but with the direction of wave propagation (Holwill & McGregor, 1976). In vivo Crithidia, and certain in vitro preparations of the system, maintain proximally directed waves, but reverse the direction of wave propagation when stimulated appropriately. Whereas living cells are able repeatedly to reverse the propagation direction in response to environmental changes, once a reactivated flagellum begins propagating waves from base to tip this propagation direction is maintained; conditions in which the organelle will revert to tip-to-base beating have not been discovered. It has therefore not been possible to test whether the central complex is involved in controlling wave direction.

The non-linearity of the double-reciprocal plots of frequency and ATP concentration (Fig. 3) indicates that the response of the system does not obey Michaelis-Menten kinetics over the entire range of ATP concentrations used in our studies. However, the curves of Fig. 3 can be matched reasonably by two linear sections, corresponding to concentrations in the ranges 5–20 mmolm⁻³ and 20–250mmolm⁻³, showing that Michaelis-Menten type behaviour is followed over these ranges considered separately. For each range K_m values are the same for the 9+2 and 9+1 samples, within the limits of experimental error, suggesting that the ATPase kinetics are the same in the two systems. We conclude that the removal of one central microtubule does not alter the enzyme kinetics of the ATP-dynein interaction (which we assume to be reflected by the beat frequency) over the range of ATP concentrations considered.

Other work on Crithidia oncopelti (Douglas & Holwill, 1972; McGregor, 1976) and a variety of spermatozoa from marine organisms (e.g. Gibbons & Gibbons, 1972; Brokaw, 1975; Gibbons et al. 1985) has shown that at ATP concentrations above about 30mmolm⁻³ the variation of flagellar beat frequency with ATP concentration (more strictly with MgATP concentration) follows Michaelis-Menten kinetics with K_m values in the range 0·l–0·6molm⁻³ (Table 6). Precise agreement between the K_m values for different systems is not expected, since the nature of the enzyme and its chemical environment will vary from one organism to another. Departures from Michaelis-Menten behaviour at ATP concentrations lower than 30mmolm⁻³ have been noted by Gibbons & Gibbons (1972), Brokaw (1975) and McGregor (1976), but it is not clear whether their results could be matched by two linear regions as discussed above. Reactivated cilia from the newt lung do show a biphasic behaviour similar to that reported for Crithidia, and the K_m values are also similar (Table 6), but the junction of the two linear regions occurs at about 500mmolm⁻³. The present study did not extend to this concentration range, but earlier work (McGregor, 1976) indicates that a third linear region exists for Crithidia at higher ATP levels (Table 6).

K_m values derived from measurements of the sliding velocities of microtubules as a function of ATP concentration have been obtained for sea-urchin sperm (Yano & Miki-Noumura, 1980) and Chlamydomonas (Okagaki & Kamiya, 1986), and are similar to the values obtained from beat frequency measurements (Table 6). K_m values for microtubule sliding velocities in a bending axoneme correspond closely to those for beat frequencies (Okuno & Brokaw, 1979). The similarity between the K_m values obtained from the frequency and microtubule sliding experiments indicates that the mechanism which converts sliding into bending does not involve a separate enzymic process from that which induces sliding; if different processes were effective, it would be expected that the K_m values derived from beat frequency measurements and sliding velocity observations would differ. It is, therefore, not unreasonable to suppose that the reactions responsible for the production of sliding may also be involved in its resistance; that is, while some dynein arms are inducing active sliding others are responsible for restricting the sliding and thereby causing bending to occur.

Despite extensive tests to detect a significant difference between the motile characteristics of 9+1 and 9+2 axonemes, the results presented here indicate that they are essentially identical in terms of wave characteristics and enzymic activity. This supports the conclusion reached on the basis of studies on flagellar mutants and deviants that neither the central complex nor the radial spokes is essential for the conversion of sliding into bending, or for wave propagation. The possibility that the central complex is concerned with controlling the direction of wave propagation requires further study, but is a reasonable suggestion in view of the Ca²⁺ sensitivity noted earlier. On the basis of the motile response of Crithidia to Ca²⁺, Holwill & McGregor (1976) note the need for a site of calcium sequestration within the organelle, and the central complex would be an appropriate location. The different waveshapes observed on intact and demembranated flagella suggest that the membrane plays a significant role in determining this parameter.

This work was supported in part by grants from the SERC.