Analysis of Ni²⁺-induced arrest of Paramecium axonemes

Paramecium is an excellent system for studying the effects of various agents on the control of ciliary activity, since biochemical and mutational analysis can be combined to dissect the regulatory pathways (Hinrichsen et al. 1986; Saimi and Kung, 1987; Bonini and Nelson, 1988; Satir et al. 1988). Since the original work of Gelei (1935) and Tartar (1950), it has been well established that Ni²⁺ gradually inhibits ciliary activity in different ciliated cells, including Paramecium, but the molecular basis of Ni ²⁺ action has not been understood. This may prove important, since Ni²⁺ is effective in reversing the high Ca²⁺ block in certain axonemes to a fixed beat position (bands down’), in which one axonemal switch that controls arm activity during beat is thought to be blocked (Satir et al. 1990). The present study has investigated whether Ni ²⁺ acts directly on the Paramecium ciliary axoneme and, if so, whether the site of inhibition is dynein itself. Detergent-permeabilized cells have proven valuable in addressing the first question (Naitoh and Kaneko, 1973; Bonini and Nelson, 1988; Lieberman et al. 1988). This treatment compromises the cell membrane so that, as we will demonstrate, low concentrations of Ni²⁺ can directly and reversibly block axonemal motility.

In order to clarify the mechanism of the immobilizing action of Ni ²⁺ on demembranated cilia, we first investigated the effect of Ni ²⁺ on the ATP-induced sliding of microtubules in trypsin-treated axonemes of Paramecium. We will demonstrate that Ni²⁺ directly and reversibly inhibits sliding. However, the uncertain effects of protease digestion and the complexity of structural interaction in the axonemes limit the information obtainable about the direct effect of Ni²⁺ on dynein itself in such studies. An in vitro motility assay has recently been introduced by which the properties of isolated dynein in translocating microtubules can be studied directly (Paschal et al. 1987; Vale and Toyoshima, 1988; Sale and Fox, 1988). This assay is an improvement over existing motility assays, because only purified dynein and microtubules are involved. We have used this assay to investigate the effect of Ni²⁺ on microtubule translocation induced by the two dyneins (22 S and 14 S) that we have isolated from Paramecium axonemes (Larsen et al. 1991). The effect of Ni²⁺ on dynein ATPase activity has also been determined. We will demonstrate that Ni²⁺ inhibits beat and sliding at the same concentration, that it inhibits microtubule translocation by one of the dyneins (14 S), while it has a much more limited effect on translocation by isolated 22S dynein, suggesting that Ni²⁺ inhibits axonemal switching by an effect on 14 S dynein.

Paramecium tetraurelia were grown axenically at 27 °C in a medium adapted from Soldo et al. (1966). Cells were harvested at late log or early stationary phase by gentle centrifugation. Paramecia were then washed three times in different buffers and allowed to equilibrate for at least 30 min prior to experimental use.

Ni²⁺ was added as nickel acetate from 10 mM or 100 mM stock solutions. When addition of Ni²⁺ or other components to the media caused a change in pH this was readjusted by addition of NaOH or HC1. To calculate pNi for different solutions, we have used the Calcon computer program developed by Goldstein (1979), and modified by J. S. Tash. Stability constants are from Martell and Smith (1974).

In preliminary experiments we investigated the effect of Ni²⁺ on swimming behavior of living paramecia in different buffers: (1) in 2mM KC1, 1 mM CaCl₂, 5 mM Tris-HCl, pH 7.2; (2) in a TECK buffer (4mM KC1, lmM CaCl₂, 0.1 mM EDTA, 10 mM Tris-HCl, pH 7.2); or (3) in a buffer without calcium (5 mM Pipes, 1 mM KC1, 20 mM MgCl₂, pH 7.2). For determination of the rate of motility, control and cation-exposed cells were placed in a small chamber and observed by dark-field or phase-contrast microscopy. Visual observations or motion analysis were used to estimate changes in swimming behavior caused by Ni²⁺

Paramecia were permeabilized with 0.01% Triton X-100 on ice according to the method of Naitoh and Kaneko (1972). When swimming and ciliary beating had ceased (25 min), the cells were washed free of detergent. Thin sections of such permeabilized cells indicate that the ciliary, cell and outer alveolar membranes are greatly disrupted or entirely missing (Lieberman et al. 1988). Permeabilized cells were reactivated at room temperature in a MgATP solution (4mM MgSO₄, 4mM ATP, 1.5 mM EGTA, 10mM Tris-maleate, 50 mM KC1, pH 7.0). In some cases, cyclic AMP was added to the reactivation medium to increase the swimming speed (Bonini and Nelson, 1988). Samples of 0.4ml were withdrawn and exposed to Ni²⁺ and other divalent cations (Ca²⁺, Mg²⁺, Co²⁺) to determine their effect on swimming behavior and ciliary activity in reactivated cell models. After 2 mm exposure the effect was observed by phase-contrast and dark-field microscopy. In some experiments EGTA was added following cation exposure, in which case the effect on swimming behavior and ciliary activity was determined again 2 min after EGTA addition.

Swimming behavior was examined by dark-field microscopy and recorded on videotape for quantitative motion analysis. The videotapes were analyzed using a system introduced by Motion Analysis Systems (MAS), Inc. (Santa Rosa, CA), following the procedure of Lieberman et al. (1988), modified after Sundberg et al. (1986).

Preparation for SEM was performed as described by Lieberman et al. (1988) with minor modifications. Living and permeabilized cells were quick-fixed for 30 s in 1 % OsO₄ in 10 mM cacodylate buffer (pH 7.0), followed by rapid addition of 2% glutaraldehyde in 10 mM cacodylate buffer. After 10 min the OsO₄-glutaraldehyde mixture was replaced by fresh glutaraldehyde (2 %) for 1 h, and washed twice with 10 mM cacodylate buffer. Fixed cells were then transferred to polylysine-coated coverslips and dehydrated through a graded series of ethanol. After dehydration the samples were critical-point dried and gold coated before examination in a JEOL S35 scanning electron microscope (SEM). Micrographs are representative of the sample of undistorted, well-ciliated cells.

To eliminate the possible effect of high Ca²⁺ in the calcium shock treatment, in some cases axonemes were prepared by mechanical deciliation of permeabilized cells as described by Hamasaki et al. (1989). Results using this preparation are essentially identical to those obtained by a calcium shock deciliation (Adoutte et al. 1980), which was used routinely. After deciliation, cilia were isolated from cell bodies by centrifugation and purified according to methods of Hamasaki et al. (1989). Cilia derived from Ca²⁺ shock treatment were then treated with 0.8% Triton X-100 for 20 min on ice to remove membranes.

The axonemes were washed twice with ‘activation buffer’ (4mM MgS0₄, 1.5 mM EGTA, 50 mM KC1, 10 mM Tris-maleate, pH 7.0 or 7.5) to remove Triton, resedimented, and resuspended in activation buffer. Following measurement of absorbance (O.D. units) of the suspension at 350 nm, a sample corresponding to 0.4 unit A₃₅₀ was obtained and placed on ice for immediate use. In some experiments, this axonemal suspension was exposed to Ca²⁺, Mg^2*, Co²⁺ or Ni²⁺ for 2min and introduced into a 15 μ l perfusion chamber (Larsen et al. 1991). The axonemes attached to the glass were then activated by perfusion with activation buffer containing trypsin (1 μ g ml⁻¹), ATP (0.1 mM – 4mM) and appropriate cations. In other sliding experiments, the axonemal suspension was first digested with trypsin. The digestion process was monitored turbidimetrically and halted with excess soybean trypsin inhibitor, when the A₃₅₀ decreased to 80 % of the initial value. The axonemes were then applied to the perfusion chamber and unadsorbed axonemes were removed by perfusing activation buffer through the chamber. The adsorbed axonemes were exposed to the appropriate cations for 2 min and then perfused with activation buffer containing ATP. In some cases, experiments were performed in which EGTA was added following cation exposure. The proportion of disintegrated axonemes was estimated by direct observation of sliding in dark-field and recorded on videotape.

For determination of the effect of Ni²⁺ on negatively stained preparations of digested axonemes two samples were withdrawn from the above mentioned cuvette. Ni²⁺ was added to one sample, whereas the other served as control. Digested axonemes were applied to a sheet of Formvar-coated carbon-stabilized, copper grids. Sliding was initiated by diluting the sample with an equal volume of activation buffer containing ATP. The axonemes were incubated for 5 min in these solutions at room temperature, after which excess fluid was removed and 1 % aqueous uranyl acetate pipetted onto the grids for negative staining. The gride were drained, allowed to dry in air under cover, and observed in a JEOL 100CX electron microscope. Sliding images were identified by criteria discussed by Sale and Satir (1977) and Larsen et al. (1991).

This procedure essentially follows Larsen et al. (1991). Axonemes from 4-liter stationary-phase cultures were isolated as above, and resuspended in axoneme buffer (30 mM Hepes, 5 mM MgS0₄, 0.5 mM EDTA, 20 mM KC1, lmM dithiothreitol (DTT), 50KIUml⁻¹ aprotinin, 10 μ gml^{− 1} leupeptin, pH 7.6) with 0.6 M KC1 to extract the dyneins. Samples of 0.2 ml containing high-salt extracted protein were layered on top of linear 12 ml 5 % to 30 % sucrose gradients prepared in axoneme buffer without leupeptin and aprotinin, and with DTT at 0.1 mM. The gradient was centriftiged for 15 h at 35 000 revs min ^{− 1} (4°C) in a SW4l Ti rotor. Fractions were collected from top to bottom of the tube. Each fraction was assayed for protein and ATPase activity, and those containing the 14 S and 22 S dynein were pooled separately. The purified dynein solutions were frozen for later use in in vitro motility assays or for measurement of ATPase activity.

Motility assays were carried out essentially as described by Vale and Toyoshima (1988) and Larsen et al. (1991). 14 S or 22 S dynein samples were adjusted to approximately 0.1mg ml⁻¹ protein in axoneme buffer and a 20 μ l sample was applied to the assay chamber in two successive portions for 2-min incubation periods each. The unadsorbed dynein was removed by perfusing 20 μ l translocation buffer (50mM K*-acetate, 10 mM Tris-acetate, pH7.5, lmM EGTA, 3mM MgS0₄) through the chamber. Subsequently, translocation buffer containing taxol-stabilized bovine brain microtubules and 1 mM ATP was perfused through the chamber. This assay system has the advantage that the chamber that has dynein bound to the glass can be perfused several times to examine movement under different conditions. To test the effects of Ni²⁺ on microtubule motility, Ni²⁺ was added to the translocation medium in which the motility assay was performed. Translocation of microtubules was examined at room temperature using video-enhanced dark-field microscopy. Reversibility of inhibition was tested by perfusing additional translocation buffer without nickel through the chamber.

ATPase activities of the 22 S and 14 S dyneins were tested in the presence of Ca^2*, Mg²⁺, Co²⁺, Ni²⁺ and the dynein inhibitor vanadate. In these experiments the dyneins were always washed free from sucrose using a 30000M_r cutoff centricon and resuspended in activation or translocation buffer before addition of the metal ions.

Dynein ATPase activity (in the absence of microtubules) was analyzed by the orthophosphate determination method following Hayashi and Takahashi (1979), modified after Murphy and Riley (1962). Protein concentration was determined using the BioRad Bradford reagent (BioRad, Rockville Centre, NY) using bovine serum albumin as a standard.

The effect of Ni²⁺on ciliary activity in living paramecia In TECK buffer Paramecium tetraurelia swam forward in a gently curving spiral path with a swimming velocity of about 360±26 μ ms⁻¹ (mean of 3 experiments; swimming velocity determined by motion analysis of at least 30 randomly chosen cells in each experiment). These forward swimming cells showed characteristic metachronal wave patterns with a wavelength of approximately 11 μ m and an effective stroke toward the posterior end of the cell (Fig. 1). Ni ²⁺ immobilized cell motility in a dose-dependent manner. In TECK buffer, concentrations above 0.1 mM Ni²⁺ slowed and finally stopped swimming within 5 min as a result of uncoordinated ciliary activity. Such immobilized cells showed some sporadic ciliary motion, but had lost the coordinated beating characteristic of metachronal waves. After prolonged exposure to Ni²⁺ concentrations above 0.1 mM an increasing deciliation was observed, but some cilia still beat, and the cells exhibited rocking movements. This residual ciliary activity was, however, limited and the cilia appeared stiff, in that their beat was restricted to a limited arc relative to the body surface.

Ca²⁺ has been reported to abolish the immobilizing effect of Ni²⁺ in Paramecium (Kuznicki, 1963; Andrivon, 1972) and it could be that the residual beating we observed was due to the presence of 1 mM CaCl₂ in TECK buffer. Therefore the effect of Ni²⁺ on cell motility was tested in a buffer without added Ca²⁺. This buffer did not affect the results.

In an appropriate reactivation medium, permeabilized paramecia were immobile in the absence of externally supplied ATP, After reactivation with Mg²⁺ and ATP, ciliary activity resumed and most cell models swam. The few models that did not swim were reactivatable, since their cilia were beating rapidly, but they appeared to be attached firmly to the glass of the chamber. The average swimming velocity of the free-swimming cell models was approximately l70±28 μms⁻¹ (mean of 3 experiments) and the models swam continuously forward with the metachronal wave pattern characteristic of living, forward-swimming’cells (Fig. 2A). Addition of cyclic AMP to the reactivation medium caused an increase in the swimming speed of the free-moving permeabilized cells, and an increase in the number of free-moving cells in the population. When the cells were reactivated in the presence of 10 μ M cyclic AMP the swimming speed increased to about 200 μm s ^{− 1} and virtually all of the cells in the population swam in straight or curved paths (Fig. 3). Ni²⁺ at low free-ion concentration (30 μM) caused rapid arrest of all ciliary activity, resulting in a complete immobilization of the permeabilized cells (Fig. 3). Other divalent cations such as Co²⁺ did not cause a comparable inhibition. When permeabilized cells were reactivated and exposed to approximately 30 μM free Ni²⁺ (pNi 4.6), there was no deciliation. The metachronal wave was no longer observed in SEM and cilia were captured primarily in a single-stroke position (Fig. 2B). For most of their length, the cilia curved uniformly to point toward the posterior end of the cell in a position corresponding to the end of the effective stroke (‘hands down’, in the terminology of Wais-Steider and Satir, 1979).

The permeabilized cells resumed ciliary activity when EGTA was added, raising the pNi to 6.4 or higher (Fig. 3; Table 1). Although beat was reactivated in the whole population, both the number of swimming cells and their average swimming velocity fell substantially (Fig. 3). Even after all treatment, however, some cells resumed swimming at rates greater than 200 μm s ^{− 1}, suggesting that for these cells, Ni²⁺ inhibition was completely reversible (Fig. 4).

The axonemes obtained from isolated cilia of Paramecium were not uniform in length and short fragments were commonly found. Without trypsin treatment, in ATP Paramecium axonemes did not normally resume beating and only slid to a limited extent. Trypsin treatment caused an increase in the proportion of axonemes that disintegrated by sliding upon ATP addition. Ni²⁺ had a marked effect on the ATP-induced disintegration of trypsintreated axonemes. When pNi was lowered to about 5, the number of axonemes sliding was greatly reduced or sliding was abolished. In contrast, marked sliding occurred at pCa 4. Sliding also occurred when Co²⁺ was substituted for Ni ²⁺ (Table 2). The Ni²⁺ inhibition of sliding could be reversed by subsequent addition of EGTA.

Electron microscopy of negatively stained preparations of disintegrated axonemes showed that both the dynein arms and the spoke groups were easily identified after trypsin treatment. Sliding configurations of ciliary axonernes were readily identified (Larsen et al. 1991) (Fig. 5A). When pNi was lowered to about 5, sliding configurations could no longer be found, but axonemes sometimes opened to show periodic interdoublet linkages (Fig. 5B).

When dyneins were extracted from Paramecium axonemes with 0.6M KC1 and further purified by sucrose density gradient centrifugation, two peaks of ATPase activity, corresponding to 22 S dynein and 14 S dynein, were obtained. In in vitro motility assays, the translocation velocity depends on the buffer used. Under standard conditions with the translocation buffer (see Materials and methods) the isolated 22 S dynein induced attachment and translocated MAP-free, taxol-stabilized, calf brain microtubules with an average velocity of 2.7l(±0.92) μ m s⁻¹ at 22 °C in the presence of 1 mM ATP (Table 3) (Larsen et al. 1991). Addition of Ni²⁺, to pNi 4.6, sufficient to inhibit beat and sliding, affected translocation velocity significantly, but did not stop translocation (Table 3). In some cases, translocation by 22 S dynein could still be observed at free Ni ²⁺ concentrations at least 10 times higher than those necessary to inhibit beat completely in permeabilized cells. In contrast, under the same standard conditions at pNi>7, 14 S dynein translocated microtubules at velocities of 0.79(±0.33) μm s⁻¹ (Table 3) (Larsen et al. 1991). However, translocation was inhibited at pNi 5.6 and was almost completely abolished at pNi 4.6, i.e. the same concentration of Ni²⁺ that blocked ciliary beat in MgATP-reactivated permeabilized cells and sliding of isolated trypsin-treated axonemes. When pNi was raised by repeated perfusion with translocation buffer (containing EGTA), translocation by 14 S dynein resumed at control velocities.

The Mg²⁺-ATPase activity of purified 22 S or 14 S dynein was measured before and after addition of Ni²⁺. In translocation buffer, at pNi values below 5, dynein activity decreased to about 80 % of the control value for both 22 S and 14 S dynein (Table 4). Comparable addition of Ca²⁺ had no effect on ATPase activity. Addition of 10 μM vanadate nearly abolished activity, as anticipated. Thus at pNi values where motility, sliding and microtubule translocation by 14 S dynein were greatly inhibited, ATPase activity of dynein was much less affected.

Ni ²⁺ has long been known to inhibit ciliary activity of ciliates (Gelei, 1935; Kuznicki, 1963) and other cells, although the precise mechanism involved was unknown. In living paramecia, Ni²⁺ inhibits ciliary activity in a gradual and diffuse way, probably acting first at the cell membrane, by competing with other divalent cations at membrane channel sites. The use of detergent-permeabilized, MgATP-reactivated cells has permitted experimental access to the axonemal effects of Ni²⁺. In contrast to its effects on living cells, Ni²⁺ rapidly arrests ciliary activity in reactivated Triton-extracted cells (Naitoh and Kaneko, 1973; Andrivon, 1974). Hence Ni²⁺ appears to act directly on the axoneme. In SEM, almost all nonbeating cilia of Ni²⁺-immobilized permeabilized paramecia point posteriorly (‘hands down’). Naitoh and Kaneko (1973) showed that Ni²⁺-immobilized cilia could be moved into a ‘hands up’ position by addition of Ca²⁺ (above 10^{− 4}M). This behavior would be consistent with the switching of mammalian sperm cells by Ca²⁺ and Ni²⁺ (Lindemann and Goltz, 1988) and of ‘hands down’ mussel gill cilia (Wais-Steider and Satir, 1979; Reed and Satir, 1986). In cilia, Ni ²⁺ affects a switch point of axonemal activity so that cilia are blocked in a position corresponding to the final stage of an effective stroke, while Ca²⁺ acts to block motility at the end of a recovery stroke (Satir et al. 1990).

ATP-induced sliding of trypsin-treated axonemes is inhibited by the same concentration of Ni²⁺ that inhibits the motility of demembranated cells under the same conditions. When pNi is lowered to about 5,^* sliding is reduced dramatically or abolished. In contrast, sliding occurs at pCa 4 or in comparable Co²⁺ concentrations. Therefore, Ni ²⁺ specifically affects either dynein mechanochemistry and/or some trypsin-insensitive structural components of the axoneme. Lindemann et al. (1980) showed that low concentrations of Ni²⁺ did not completely inhibit microtubule sliding in reactivated bull sperm flagella, but did inhibit beat. This suggests that the concentration thresholds for Ni²⁺ inhibition of sliding and beat may not coincide exactly, although in our experiments they are in the same range.

To investigate whether Ni²⁺ affects dynein directly, we isolated 22 S and 14 S dynein from Paramecium axonemes. 22 S dynein is a three-headed bouquet with three heavy chains that are u.v. photocleavable in the presence of vanadate, while 14 S dynein is a single-headed species that may be heterogeneous (Larsen et al. 1991). Two u.v. photocleavable heavy chains are found in the 14 S region in our preparations. Travis and Nelson (1988) reported that a shoulder of the 22 S peak could be identified as 19 S dynein, but this is not apparent in our preparations where the fraction size is larger. Their 12 S peak probably corresponds to our 14 S dynein (see Larsen et al. 1991, for further details).

Both 22 S dynein and 14 S dynein support translocation of purified bovine brain microtubules in an in vitro motility assay using a standard translocation buffer, but 22S dynein translocates microtubules at a rate about three times as fast as 14S dynein. Addition of Ni²⁺ at concentrations sufficient to inhibit beat and sliding abolishes translocation by 14 S dynein in in vitro assays but only partially affects translocation by 22 S dynein. Translocation by 22 S dynein is observed at Ni²⁺ concentrations at least ten times higher than those required for complete beat inhibition. In contrast, Vale and Toyoshima (1988) demonstrated that addition of vanadate completely inhibited 22S dynein-induced microtubule translocation in Tetrahymena, indicating that Ni²⁺ does not inhibit ciliary motility in the same way as vanadate, by abolishing general dynein ATPase activity. In fact ATPase activities of purified 22 S and 14 S dyneins are only weakly inhibited by Ni ²⁺, although the extent of inhibition seems dependent on the exact assay conditions. Using somewhat different conditions, Travis and Nelson (1988) showed that 14 S dynein ATPase activity was inhibited by Ni²⁺ to a much greater extent than 22S dynein. Therefore, inhibition of beat, sliding and probably ATPase activity by Ni ²⁺ correspond to the specific inhibition of 14 S dynein, as demonstrated by inhibition of microtubule translocation in in vitro assays. This is supported by the observation that the effect of Ni ²⁺ on translocation, as well as on beat and sliding, is reversible when EGTA is added to raise the pNi. Unfortunately, the axonemal localization of 14 S dynein, which now seems critical to an understanding of switching activity in the axoneme, and of the Ni²⁺ effects, is uncertain. Many investigators believe that, on the basis of rebinding studies, 14 S dynein is a component of the inner dynein arm (Warner et al. 1985). However, dyneins often rebind promiscuously to several structures in the axoneme (Satir et al. 1981). Another possibility is that 14 S dynein is a component of the spoke head. There is some suggestion that antibodies to 14S Tetrahymena dynein are localized at either the spoke head or the outer mid-wall of the doublets. It is unlikely that dynein at the outer midwall could be responsible for switching arm activity, but the inner arm-spoke relationships are probably fundamental to this process. It would be useful to have the appropriate axonemal mutants in Paramecium to be able to pursue this question further. It is also interesting that 22S dynein, which is found in the outer arm, is evidently not critically inhibited by Ni²⁺ at low concentrations.

This work was supported in part by grants from the USPHS to P.S. and from the Danish Natural Science Research Council to J.L. We are grateful to J. Avolio, K. Barkalow and T. Hamasaki for their help with various aspects of the study and to M. AnnHolland for secretarial assistance. Assistance and support for the motion analysis work reported was kindly provided by Dr J. L. Spudich and his laboratory.