Graviresponses in Paramecium Caudatum and Didinium Nasutum Examined Under Varied Hypergravity Conditions

All ciliates so far investigated respond to the natural gravity vector. The most obvious response, swimming away from the centre of gravity (‘negative gravitaxis’), has long been known and described in detail (see Machemer and Bräucker, 1992, for references). Responses to gravity at the cellular level have appeared in a new light since Naitoh and Eckert (1969) discovered the electrophysiological basis of mechanosensitivity in Paramecium and since the presence of an intricate gradient-type distribution of depolarizing and hyperpolarizing mechanoreceptor channels was established in the membranes of Stylonychia and Paramecium (De Peyer and Machemer, 1978; Ogura and Machemer, 1980). The topographical pattern of mechanosensitivity in a ciliate determines the sign and amplitude of the receptor potential following local stimulation. If the gravity vector can deform mechanically sensitive structures on the membrane, the resulting potential may be predicted to depend on cell orientation in the gravity field. Moreover, this hypothetical ‘gravireceptor potential’ carries valuable information on the position of ‘up’ and ‘down’ in the environment (Machemer et al. 1991). Extensive investigations of the control of ciliary activity in ciliates have established a close coupling between ciliary beating and membrane potential: in Paramecium, a hyperpolarization augments the rate of beating, whereas a small depolarization slows down ciliary frequency (see Machemer, 1986, 1988b). The mechanosensitive organization and electromotor coupling properties formed the basis of a previous hypothesis predicting that upward-swimming cells increase, and downward-swimming cells decrease, active locomotion in response to the gravity vector (Machemer, 1989; Baba et al. 1991). Kineses of this kind have been isolated repeatedly in Paramecium, Didinium and Loxodes (see Machemer and Bräucker, 1992, for references). We have termed this type of response ‘negative gravikinesis’ because it tends to offset, by analogy with ‘negative gravitaxis’, the sedimenting effect of the gravitational pull. A gravikinetic response of a cell results from (1) the mechanosensory organization of the cell and (2) its orientation in space. Like classical kinesis, gravikinesis has no direct cell-orientating capacity.

Using hypergravity is a promising approach to the investigation of gravikinetic responses. Enhancing gravitational acceleration may establish more pronounced graviresponses depending on type, number, location and activation state of receptor channels. In a previous investigation of vertically gliding Loxodes striatus, it was shown that constant velocities of vertical gliding at 1 g (Bräucker et al. 1992) gave way to diverging velocities under hypergravity: the rates of upward gliding were unchanged, but those of downward gliding rose with the gravity vector (Machemer-Röhnisch et al. 1993). The latter investigation also established that both graviorientation and gravikinesis in Loxodes prevent net upward or downward shifts of a cell population (‘neutral gravitaxis’).

The present study uses two species of ciliates, Paramecium caudatum and its predator Didinium nasutum, for a comparative study of the effect of hypergravity on the graviresponses. Didinium differs from Paramecium in two fundamental ways: (1) it can generate depolarizing, but not hyperpolarizing, mechanoreceptor responses (Hara and Asai, 1980; Hara et al. 1985); (2) modulation of its ciliary activity is limited to activation by depolarizing stimuli: induced hyperpolarization from the resting potential does not change ciliary frequency (Mogami et al. 1990; Pernberg, 1993). We have evaluated the effects of integration of gravikinesis and graviorientation on the position of cell populations of Paramecium and Didinium using the taxis coefficient for quantification. The results confirm the persistence of ‘neutral gravitaxis’ even under hypergravity.

Paramecium caudatum, line G3, were reared in Cerophyl solution (Cerophyl Laboratories, Inc., Kansas City, USA), 0.2% (w/v) cerophyl powder in double-distilled water that had been autoclaved, buffered at pH 7.0 by modified Sörensen buffer (1.38 mmol l⁻¹Na₂HPO₄⁺ 0.62 mmol l⁻¹ NaH₂PO₄) and bacterized with Aerobacter aerogenes. The Paramecium were cultured at 22°C in a 14 h:10 h light:dark regimen and harvested in the early (3 days, ‘well-fed cells’) or late (7 days, ‘starved cells’) stationary phase. The cells were collected by negative gravitactic accumulation in the experimental solution (in mmol l⁻¹: CaCl₂, 1; KCl, 1; MgSO₄, 0.1; Mops, 1.5; buffered at pH 7.0). Transfer to the experimental chamber (volume 0.86 ml) was performed by a syringe (1 mm terminal inner diameter), avoiding mechanical agitation.

Didinium nasutum, wild type, was grown in Sörensen-buffered Pringsheim solution [in mmol l⁻¹: Ca(NO₃)₂, 0.85; KCl, 0.35; MgSO₄, 0.08; Na₂HPO₄, 0.41; NaH₂PO₄, 0.18; pH 7.2] at 19°C and fed every 24 h with cells of Paramecium caudatum. Prior to an experiment, Didinium cells were starved for 12–16 h. Gravitactic accumulation and transfer were carried out in the experimental solution (see Paramecium).

The acrylic chamber enclosed an experimental fluid space of 25 mm×25 mm×1.6 mm, equivalent to 1 ml. The upper (=centripetal) and lower (=centrifugal) borders of this space were lined by agar blocks (1.5% in experimental solution; 0.8 ml each) to increase the buffering volume of the medium.

After transfer into the chamber, the cells were equilibrated for 4 h in experimental solution. The O₂ level within the closed chamber was ⩾93% air saturation (polarographic sensor and amplifier, type 170, Ingold, Germany). In a large-scale reference chamber with the same surface-to-volume ratio, the O₂ concentration decreased by ⩽5% over 4 h at the same cell density [cell/water: 1/16 000 (v/v)].

For the recording of cellular behaviour at gravity values above 1 g, the chamber was mounted in a horizontal centrifuge (NIZEMI, Dornier GmbH, Germany) designed to generate highly controlled accelerations of microorganisms (Briegleb and Hemmersbach, 1987; Kreuzberg et al. 1991). The centrifuge consisted of a belt-driven disc, to which the chamber was fixed in a radial plane so that the resultant acceleration vector was parallel to the plane of the chamber (Fig. 1). The disk-shaped chamber was free to pivot about its central axis (perpendicular to the chamber surface) in the radial plane of the centrifuge. An eccentric mass connected to the chamber could swing out depending on the angular velocity of the centrifuge. The design of the centrifuge ensures that the major axis of the chamber is in line with the axis of the pendulum so that the resulting acceleration acts parallel to that axis (Machemer-Röhnisch et al. 1993). The angle of centrifugal swing (γ) was used to calculate the resulting factor (x) of terrestrial acceleration (g):

where l is the distance between the centres of mass of the chamber and the pendulum and R represents the distance between the centre of the chamber and the axis of the centrifuge.The accuracy of equation 1 for the determination of acceleration was checked against an equation using the angular velocity of the centrifuge and R.

The central recording area (8.5 mm×11.5 mm) was illuminated by a ring of 48 light-emitting diodes generating a dark-field illumination in the experimental plane of 800 lx. The emission wavelength (565 nm; half-width 28 nm) is near the minimum spectral sensitivity for photoresponses in Paramecium caudatum (Iwatsuki and Naitoh, 1982). Spectral sensitivity in the colourless Didinium nasutum is yet to be determined. The macrolens of a high-resolution video camera (HR600 M, 25 Hz) was focused on the central plane of the fluid volume 0.8 mm away from the inner surfaces of the chamber. Images (8.5 mm×11.5 mm) of the field were projected into the video camera via a mirror. The video signals were transmitted to a recorder outside the centrifuge. The positions of cells in the vertical plane were defined with respect to the angle made by the trajectory of the cell with the vector of acceleration.

Paramecium cells from the experimental solution were exposed to 0.5 mmol l⁻¹ NiCl₂ dissolved in experimental solution and were immediately transferred into the chamber. After allowing 20 min for immobilization, sedimenting cells were recorded. Didinium cells were treated with 1 mmol l⁻¹ NiCl₂ in experimental solution and transferred to the chamber 45 min after nickel application. Sedimenting cells were recorded following 1 h of immobilization. For both species, recording time was limited to approximately 20 min. A criterion for completeness and quality of cell immobilization, apart from direct inspection of the sedimentation traces, was the shape of the velocity distribution: samples showing a steady and preferably Gaussian distribution were considered as acceptable (see inset to Fig. 6). Uneven distributions of velocities may be caused by Ni²⁺-induced cell deformations (swelling) and/or residual ciliary activity.

A sample of 150–200 cells was transferred into the chamber. The temperature inside the chamber was between 22 and 23°C. Swimming in Paramecium was recorded for periods of 2 min in two sequences of g values: ‘type 1’, 1→1.5→2.1→3.3→4.3→5.4→1; ‘type 2’, 1→5.4→4.3→3.3→2.1→1.5→1. Changes in gravitational acceleration between two predetermined g levels were set at 1 g min⁻¹. A 2 min recording of cell locomotion was evaluated to generate five separate representations of 4 s tracks using digital image analysis (Machemer et al. 1991). From each of these images, at least 10, and commonly about 30, tracks were evaluated. With Paramecium, 12 experiments were carried out for each value of hypergravity (1 g: 24 experiments). Assuming that each specimen entered the recording area and was measured 1.3 times, data points are based on at least 1200 individual cells. With a similar protocol, about 1000 individual Didinium cells were evaluated up to gravities of 2.1 g. Beyond 2.1 g, cell numbers declined because they showed rapid downward swimming activity and escaped from the recording area (see Figs 3, 9).

For evaluations of tracks, we used 16 markers on each track separated by the same time interval. Velocity and orientation were determined for each trace. The averaging of velocities within an orientation class was carried out irrespective of the cell count. Non-parametric statistics (calculation of median values; confidence ranges; U-test) were applied because Gaussian distribution of the data could not be guaranteed (see Figs 3 and 9, for example).

which determines the value and sign of Δ. Variables in this equation are velocities, which are proportional to forces of active propulsion or gravitational pull according to the Stokes equation of flow at low Reynolds numbers (Wu, 1977). An observed downward swimming rate may exceed, correspond to, or fall below the upward swimming rate. Unless the rate of sedimentation (S) is determined, an observed response to gravity has only descriptive value. With the value of S known, the amount and sign of gravikinesis can be calculated. The term ‘negative gravikinesis’ indicates, by analogy with an orientational (=tactic) response, that upward or downward velocity is regulated so that the cell moves away from the centre of gravity. With a ‘positive gravikinesis’, the probability of locomotion towards the centre of gravity is raised.

the sign and size of the kinetic responses may be determined. The arithmetic mean of Δ_U and Δ_D gives the generalizing term Δ as represented in equation 2.

Graviorientation, integrated over time and with many individuals, is expressed by the orientation coefficient, r_o, which incorporates the stimulus direction and can vary between +1 (all cells strictly orientated upwards) and −1 (all cells strictly orientated downwards), with zero designating no orientation. The combined effects of swimming velocity and orientation are expressed by the taxis coefficient, r_t, which weighs individual swimming orientation by individual swimming rate. A negative taxis coefficient indicates a net upward shift of a cell population, and a positive taxis coefficient indicates a net downward shift. A taxis coefficient close to zero indicates that the centre of a population of cells probably maintains a stable vertical position (see Machemer and Bräucker, 1992).

The velocities of cells from 3-day cultures were plotted using polar histograms for orientation (class width, 15°; Fig. 2). The rates of locomotion within these sectors are represented by medians. The velocity polarograms show a minor increase in downward rates over the upward rates at 1 g. This tendency is much enhanced with rising acceleration. At 5.4 g the velocities tend to grow continuously with a more ‘downward’ orientation of the cell. This would be expected from vector addition of increasing sedimentation to cellular propulsion.

Gravity-dependent distributions of velocity are independent of previous stimulation. This applies to both experimental series employing rising (type 1) and falling (type 2) acceleration. A small time-dependent decrease in absolute velocity is apparent in both type 1 and type 2 experiments (Fig. 2). This decay may be due to the partial loss of rapid swimmers, which were able to leave the central recording area (Fig. 3, 1 g before, 1 g after).

Distributions of orientation of Paramecium applying gravitational acceleration between 1 g and 5.4 g are shown in Fig. 4. The polar histograms suggest a small tendency for upward orientations (‘negative gravitaxis’) at normal gravity. With acceleration rising to 5.4 g, negative graviorientation is clearly evident. Correspondingly, the orientation coefficient (r_o), which integrates all swimming directions, rises from near zero to almost +0.2 (see Table 2).

Comparison of the polarograms of velocity and orientation (Figs 2, 4) suggests an inverse relationship between the g-dependent proportion of cells found in a given orientational sector and their speed: many ‘slow’ cells swim ‘upwards’ and a smaller proportion of ‘fast’ cells swim ‘downwards’. This diverse behaviour of a cell population is integrated by the taxis coefficient (r_t; see Machemer and Bräucker, 1992). At a taxis coefficient close to zero (see Table 2), the probability of a gravity-dependent shift of the cell population is low. Thus, the data suggest that the orientational and kinetic behaviours in Paramecium tend to neutralize the effects of gravity on a cell population at all g values.

Medians of the vertical and horizontal velocities (±15°) plotted as a function of acceleration appear to change in a linear fashion (Fig. 5), especially if the values at 1 g are not used in the regressions (Table 1). We have tentatively excluded the 1 g data from the calculation of the g-dependent slopes of velocity because an unknown mechanical factor (possibly vibration) might have influenced the locomotor activity of Paramecium while the centrifuge was running (open symbols in Fig. 5). A comparison of the swimming rates in well-fed and starved cells (Fig. 5A,B) reveals differences in the slopes of the regression lines of velocities: the positive slope of downward swimming (V_D) increased, and the negative slope of upward swimming (V_U) decreased, in starved cells. No change occurred in the slopes of horizontal swimming rates.

The observed relationship between swimming velocity and acceleration suggests velocities below 1 g can be estimated by extrapolation. The inferred swimming rate at 0 g then equals the propulsion rate, P, of Paramecium, which is defined as being unaffected by gravity. Theory predicts that the curves of horizontal and vertical swimming rates must intersect at 0 g. Fig. 5 shows only an approximation to this prediction. In order to minimize errors in determinations of P from the regression lines, we calculated the velocity at the intersection of the V_D and V_U slopes near 0 g (arrows in Fig. 5). These inferred velocities during weightlessness agree with the observed horizontal swimming velocity of Paramecium at 1 g within close limits (Fig. 5A, well-fed cells, −1%; Fig. 5B, starved cells, −5%).

An interesting observation is that hypergravity was virtually ineffective in modulating horizontal velocity: at 5 g, the horizontal swimming velocity was reduced by 6.5% (in well-fed cells; 6.2% in starved cells). This strengthens the conclusion from previous work (Machemer et al. 1991, 1993) that the bipolar gravisensory input to horizontally swimming Paramecium is neutralized at 1 g. Assuming that the slopes of V_H in Fig. 5A,B are real, the swimming velocity, P, during weightlessness will, nevertheless, exceed horizontal velocity at 1 g conditions by 1.5%.

Swimming velocities of cells are not directly amenable to determinations of gravikinesis because they include the sedimentation rate. Fig. 6 shows that immobilized cells from 3-day cultures settled at a median rate of 90 μm s⁻¹. We were unable to determine sedimentation in starved, 7-day cells because they tended to disintegrate during the nickel treatment. Individual sedimentation rates were quite different owing to varying cell size and cytoplasmic inclusions (inset Fig. 6). The distribution histogram of sedimentation rates may even include contributions from rudimentary ciliary activity of the gullet region of the immobilized Paramecium. With rising acceleration, the sedimentation rate rose in a linear fashion (correlation coefficient 0.995). The extrapolated regression line, however, does not intersect with the origin, suggesting the possibility that the sedimentation curve is non-linear below 1 g.

Experimental limitations make it impossible to investigate the association of swimming velocity with sedimentation at the single-cell level. Therefore, determinations of median rates of swimming and sedimentation are important for the calculation of gravikinesis. Median sedimentation rates at various levels of hypergravity and the establishment of a gravity–sedimentation relationship limit experimental errors. We use interpolations of the sedimentation rate from the regression line for measurement of gravikinesis under hypergravity. The 1 g data are again excluded from this procedure because they were not obtained while the centrifuge was running.

Hypergravity between 1.5 and 5.4 g increased the negative gravikinetic response of Paramecium, that is its ability to counteract sedimentation (Fig. 7, see generalized value of Δ). Gravikinesis rose markedly with hypergravity in starved cells (Fig. 7B). This positive correlation was less pronounced in well-fed cells (Fig. 7A). The basis of this difference in responsiveness to hypergravity is, primarily, the gravikinetic response of the upward swimmers (Fig. 7, Δ_U). Δ_U decreased with rising acceleration in well-fed cells (Fig. 7A), whereas in starved cells, it rose with hypergravity running nearly parallel to the gravikinetic response of the downward swimmers (Δ_D; Fig. 7B).

The experiments in the vertical chamber at 1 g show that downward swimming rates are slightly greater than upward velocities (Fig. 8A). The decrease in upward swimming rates and the increase in downward swimming rates grew with rising acceleration. At 5.4 g, the downward velocity was 150% of the upward velocity. In contrast to Paramecium caudatum, a plot of the distributions of velocities revealed an over-representation of comparatively higher velocities up to accelerations of 3.3 g (Fig. 9).

Polarograms plotting the orientations of Didinium at different accelerations (Fig. 8B) are the inverse of the distributions of velocities (Fig. 8A). A small preference for upward orientation at 1 g (r_o=0.031) was continuously transformed to obvious negative gravitactic orientation above 3.3 g (r_o⩾0.15; Table 2). The taxis coefficients (r_t) show that the orientational response effectively counterbalanced the effects of the enhanced downward velocities: during hypergravity, r_t shifted from near zero at 1 g to small positive values, maximally +0.08. These observations on the gravitactic behaviour in Didinium correspond to those in Paramecium (Table 2).

Plotting swimming rates as a function of acceleration gives linear relationships with different correlation coefficients (upward swimmers, r=−1.00; downward swimmers, r=0.65; horizontal swimmers, r=−0.88) and varying ranges of confidence (Fig. 10). The reason is that, as hypergravity increased, an increasing number of cells swam upwards, whereas horizontal and downward swimmers were registered less frequently (Fig. 8; Table 1). However, the regression lines intersect with the ordinate (0 g) at very similar velocities, suggesting that the gravity-free propulsion rate can be determined reliably. Intersection of the ‘downward’ with the ‘upward’ regression is at 1592 μm s⁻¹ (the definition of P). The intersection of the ‘horizontal’ regression line with 0 g is at 1583 μm s⁻¹. In contrast to Paramecium, the horizontal swimming rate is strongly dependent on acceleration. The value of P cannot be determined from V_H because the distribution of gravireceptor conductances in Didinium is still unknown (see Discussion).

Median sedimentation rates of Ni²⁺-immobilized Didinium are a linear function of acceleration (Fig. 11; correlation coefficient: r=+0.98). The good correlation enables the use of interpolated sedimentation rates between 1 and 5.4 g to compensate for experimental limitations and errors (Table 1). As in the experiments with Paramecium, the regression line does not intersect with the origin (zero sedimentation at 0 g) and thereby restricts our conclusions to the range of accelerations used in the experiments.

Using interpolations of the vertical velocities (V_U, V_D) and sedimentation (S), a generalized gravikinesis (Δ) of -23 μm s⁻¹ results for normal gravity (Fig. 12, Table 1). Gravikinesis was virtually absent in upward swimmers (Δ_U=−1μm s⁻¹ ), whereas it was substantial in downward swimmers (Δ_D=-45 μm s⁻¹ ). These data suggest that Didinium does not actively respond to gravity while swimming upwards at 1 g.

With acceleration exceeding 1 g, the upward swimmers showed an increasing positive value of kinesis (5.4 g: Δ_U=+214 μm s⁻¹ ), i.e. the upward swimming rate was depressed by both sedimentation and an active reduction in forward propulsion (Fig. 12). In the downward-swimming cells, negative gravikinesis gradually increased in response to rising acceleration (5.4 g: Δ_D=−70 μm s⁻¹ ). Consequently, the generalized term changed from a negative value at 1 g to a positive value at 5.4 g (+72 μm s⁻¹ ). Note that this bipolarity of the gravikinetic response was associated with artificial accelerations.

Isolation of the physiological response to gravity from behaviour in ciliates requires standardized experimental conditions and large cell numbers because swimming velocity and direction, the rate of intrinsic propulsion and sedimentation cannot all be determined in the same cell. At terrestrial gravity level, the kinetic response in Paramecium and Didinium is a small fraction of locomotory velocity (Table 1). Since gravikinesis is calculated from three observed variables (equations 2–4), possible experimental errors in their determination are summed. An additional problem is the unknown individual variability in a population of cells. In spite of these complications, the present study confirms the existence and magnitude of gravikinesis in Paramecium and Didinium at natural gravity (Table 2 in Machemer and Bräucker, 1992). Moreover, by using hypergravity up to 5 g, we have been able to study the modulation of both gravikinesis and gravitaxis as a function of acceleration.

The bipolar organization of mechanosensitivity in Paramecium ensures that the sign of gravikinesis remains negative in upward and downward swimmers even under hypergravity (see Fig. 13, left-hand column). Under normal gravity, the unipolarity of the mechanoreceptor response in Didinium leaves upward swimmers without kinesis; downward swimmers are slowed down by negative gravikinesis (Fig. 13, right-hand column). We believe that this system, which is less effective in antagonizing sedimentation, is disturbed under hypergravity under which the upward swimmers generate a positive gravikinesis, enhancing the effect of passive sedimentation. Hypergravity was not experienced in the evolution of ciliates so that this behaviour has no impact on survival in common environments. Moreover, it may be seen from Fig. 8 and from the taxis coefficients in Table 2 that, in Didinium, negative graviorientation appears successfully to offset the less favourable gravikinesis. A downward drift of a cell population is therefore less probable even under hypergravity.

We have proposed that bipolar gravistimulation in horizontally swimming Paramecium cancels gravikinesis at 1 g (Machemer et al. 1991). In confirmation of this view, the swimming rate of Paramecium under weightless conditions equalled the swimming rate in a horizontal direction at 1 g (Machemer et al. 1992). Our present experiments test this hypothesis under hypergravity (Fig. 5). The horizontal velocity decreased by only 6% between 1 and 5.4 g in both well-fed and starved cells, confirming our hypothesis. It can be concluded from electrophysiological experiments in Paramecium that the ratio of summed mechanically induced conductances, g_Ca/g_K, corresponds to the conductance ratio of the unstimulated cell (Ogura and Machemer, 1980; Machemer, 1988a). This provides further support for the observed gravitational zero balance of horizontally orientated Paramecium (Machemer, 1994).

Inspection of the data from horizontally swimming Didinium (Fig. 10) immediately reveals that conclusions valid for Paramecium may not apply to other types of cells. In Didinium, the horizontal swimming rate decreased with rising gravitational acceleration, suggesting the existence of a depolarizing input in horizontal swimmers inhibiting ciliary frequency. Depolarizing mechanoreceptor potentials in Didinium were established, whereas hyperpolarizing mechanoreceptor potentials were not detected (Hara et al. 1985). In addition, hyperpolarizing voltage steps did not modify the ciliary frequency of unstimulated Didinium (Mogami et al. 1990). Under weightless conditions, Didinium swam faster than horizontally oriented specimens did under 1 g (Machemer et al. 1992). Fig. 13 (right-hand column) accounts for these data by assuming that a gradient of hyperpolarizing mechanoreceptors is missing in Didinium. The scheme explains why V_H<P, why, under rising gravity, horizontally swimming Didinium cells slowed down and why the upward swimming rate was strongly decreased (Fig. 10).

Cells taken from our cultures at 3 days and 7 days show different kinetic behaviour under hypergravity (Fig. 7). Cells from these cultures differ in individual age. Interestingly, the data show the same value for generalized gravikinesis, (Δ= −60 μm s⁻¹ ) for well-fed and starved Paramecium at normal gravity (Table 1). Under hypergravity, the slopes of are primarily determined by the inversion of the slope of Δ_U (Fig. 7A). This cannot be explained by assuming saturation of the response of the hyperpolarizing gravireceptor, but a gravireceptor at the anterior cell end must be activated under hypergravity. The simultaneous activation of antagonizing (i.e. depolarizing and hyperpolarizing) mechanoreceptors is a fundamental property of Paramecium (Ogura and Machemer, 1980) and is likely to apply to gravireception (Machemer et al. 1991). We interpret the slope of Δ_U in well-fed cells (Fig. 7A) by assuming an additional depolarizing deformation of the anterior membrane of upwardly oriented Paramecium (filled arrow in Fig. 13A). The mechanically induced conductance ratio, g_Ca/g_K, decreases as a consequence of such dual gravireceptor activation. Hence, the resulting (hyperpolarizing) receptor potential is depressed, the frequency response of the cilia declines and the upward swimming rate is reduced. Surprisingly, our data do not suggest a similar conclusion for starved cells, where gravikinesis was uniform in upward-and downward-swimming cells (Fig. 7B). This cannot be explained by assuming an age-dependent shift in the proportion of mechanically sensitive channels because the horizontal swimming rates were unchanged (Fig. 5). Possibly, these are age-dependent changes in the mechanical properties of the cell cortex, which may affect the gravireceptor conductances in different ways. Microtubules and microfilaments of the cortex of Paramecium (Allen, 1988) may change in ageing cells, and these elements of the cytoskeleton may play a role in mechanotransduction (Wang et al. 1993).

The taxis coefficients integrate the active and passive components of velocity and orientational responses of cells. They are therefore a measure of the displacement of a cell population in the gravitational field (Machemer and Bräucker, 1992). The taxis coefficients of the present data are close to zero under natural gravity conditions in Paramecium and Didinium, and they vary little from zero (Table 2) under hypergravity. We conclude that the swimming ciliates so far investigated are able to offset sedimentation effects. Sedimentation of cells in the absence of any counterbalancing mechanism irreversibly moves them to the bottom of fluid environments. In addition, sedimentation of cells interferes with the identification of stimulus gradients in water (e.g. chemical, photic or thermic), which may be important for survival. It has been shown that Loxodes maintains the same gliding velocity irrespective of its orientation in space (Bräucker et al. 1992). The compensation of sedimentation effects by a negative gravikinesis response in Loxodes broke down under hypergravity. Nevertheless, the overall taxis was neutral up to 5 g (Machemer-Röhnisch et al. 1993).

Our conclusion that gravitational pull is neutralized by orientational and velocity responses in Paramecium and Didinium applies only to fully equilibrated cells in their physiological steady state. Previously stimulated cells (for instance, by mechanical disturbances, transfer to different solutions, shifts in temperature) may show a transient, pronounced negative gravitaxis, which is not neutralized by velocity responses. As is well known to experimenters, this spectacular phenomenon is commonly limited to some minutes or tens of minutes. Neither experimentally induced gravitaxis nor swimming rate under steady-state conditions is fully representative of the behaviour of a ciliate in its natural environment.

We gratefully acknowledge Dr Dorothea Ch. Neugebauer for helpful comments on the manuscript. We thank Ute Nagel, Petra Ullrich and Gerhard Krumbach for their devoted assistance. This work was supported by the Deutsche Agentur für Raumfahrtangelegenheiten (DARA, grant 50WB93193) and the Minister für Wissenschaft und Forschung of the state of Northrhine-Westfalia (grant IV A1-21600588).