How Does the Contractile Vacuole of Paramecium Multimicronucleatum Expel Fluid? Modelling The Expulsion Mechanism

The contractile vacuole is the organelle responsible for osmoregulation in many protozoa (Kitching, 1956) and freshwater sponges (Jepps, 1947). Kitching (1952, 1956, 1967), in pioneering studies on the physiology of the contractile vacuole, pointed out that a pressure on, or a tension at, the vacuole membrane could produce the pressure required to discharge the fluid contents of the vacuole through the vacuole pore.

On the basis of their analysis of cinematographically recorded pulsating contractile vacuoles in Amoeba proteus, Wigg et al. (1967) argued that the vacuole fluid is discharged by the pressure produced by the cytosolic gel surrounding the vacuole and not by contraction of the vacuole membrane or its associated cytoskeletal elements. They therefore introduced the term ‘water expulsion vesicle’ to replace the term ‘contractile vacuole’. This work provoked a long-standing debate on whether an active contractile process in the vacuole membrane (or a joint membrane–cytoskeletal system) is involved in the pulsation of the vacuole in amoebae and ciliates or whether other forces cause fluid discharge (Patterson, 1977, 1980).

Recently, Doberstein et al. (1993) demonstrated that, when antibodies raised against a synthetic peptide with the sequence of the myosin-IC phosphorylation site were introduced into a living Acanthamoeba sp., these antibodies interfered with the activity of its contractile vacuole. They suggested that myosin-IC is involved in generating the force required to empty the vacuole in this cell.

Recent exocytosis studies of smaller vesicles have focused on the very early events of pore formation and on the first escape of the vesicle contents to the cell’s exterior. The use of patch-clamp techniques demonstrated a ‘flickering’ type of exocytotic discharge of neurotransmitter substances through a forming pore even before stable membrane fusion has occurred (Alvarez de Toledo et al. 1993; Fernández et al. 1984; Neher, 1993). It has usually been assumed, for small vesicles, that the vesicular contents are expelled by diffusion. These assumptions have now been questioned by Khanin et al. (1994) and Parnas and Parnas (1994), who have reported that even for small vesicles in fast synapses the discharge is too rapid to be due solely to diffusion.

In the light of such reports, it seems expedient to re-examine the forces required to expel the exocytotic contents from vesicles of a wide range of diameters. For this study, we chose the contractile vacuole of P. multimicronucleatum, which is large enough to allow accurate determination of the time course of the change in vesicle diameter during systole, a variable essential for examining the putative forces that could lead to discharge of the fluid contents. We propose several plausible models by which this pressure may be generated and test how well each model corresponds to the actual observed rate of fluid discharge.

Cells of Paramecium multimicronucleatum (syngen 2) (Allen et al. 1988) were grown in an axenic culture medium (Fok and Allen, 1979) and were harvested at the mid-logarithmic phase of growth. These cells were washed with a standard saline solution containing (final concentration in mmol l⁻¹): 0.5 NaCl, 2.0 CaCl₂, 1.0 MgCl₂, 1.0 KCl and 1.0 Tris–HCl buffer (pH 7.4). Cells were then transferred into each experimental solution and equilibrated in the solution for more than 30 min prior to experimentation. The experimental solutions were prepared by adding the chemicals to be tested (80 mmol l⁻¹ sorbitol, 33 μg ml⁻¹ cationized ferritin and 0.29 mmol l⁻¹ cytochalasin B) to the standard saline solution.

Equilibrated cells were slightly compressed in a thin (20 μm) space between a glass slide and a coverslip. Profile views of the contractile vacuole were video-recorded (Panasonic AG-6300) at 30 frames s⁻¹ using a phase-contrast objective (Olympus 40×). The time course of the change in vacuole diameter during systole was measured directly from recorded images of the vacuole displayed frame by frame on a monitor screen. The contractile vacuole is almost spherical in the early and middle phases of systole. In the late phase of systole, it takes on a shape more like an ‘Erlenmeyer flask’ owing to the presence of microtubules that extend from the pore region and pass over the part of the vacuole next to the pore (Hausmann and Allen, 1977). Therefore, we omitted measurements of the diameter of the vacuole in very late systole.

For fluorescence microscopical examination of the contractile vacuole complex of P. multimicronucleatum, formaldehyde-fixed (3 % in 50 mmol l⁻¹ phosphate buffer; pH 7.4) and cold (−20 °C)-acetone-permeabilized cells were treated using two monoclonal antibodies, one raised against the decorated spongiome (DS-1) and the other raised against a pool of membranes including the smooth spongiome and the plasma membrane (SS-1) (Allen et al. 1990). This incubation was followed by an incubation in fluorescein-isothiocyanate-(FITC)-and Texas-Red-conjugated rabbit anti-mouse IgG and IgM (Miles Laboratories, Naperville, IL, USA), respectively. Unbound antibodies were washed away using excess buffer solution. The cells were observed using a Zeiss microscope equipped with epifluorescence illumination and filters appropriate for FITC (decorated spongiome) (B-2E, Nikon) and for Texas Red (smooth spongiome) (Texas Red filter, Zeiss). Photographs were obtained using photographic film (Kodak Tri-X film) (Fok et al. 1995). Intracellular injection of the antibody, DS-1 or IgG_2b, was performed as described previously (Ishida et al. 1993).

A cell in a small droplet of standard saline solution was fixed instantaneously by squirting a fixative containing 2 % glutaraldehyde against the cell through a fine pipette placed close to the cell. The precise moment of fixation was monitored using a video camera, and the phase of systole was determined. Conventional techniques were employed for obtaining the subsequent electron micrographs (Allen and Fok, 1988). All experiments were performed at a room temperature ranging from 24 to 26 °C.

Results and discussion

Fig. 1 shows three representative plots of the change in diameter (D) during systole (filled circles) of contractile vacuoles with three different initial diameters (the diameter immediately before the start of systole). These contractile vacuoles were from three different cells equilibrated in three different solutions. The initial diameter was largest in the cell equilibrated in saline solution containing 33 μg ml⁻¹ cationized ferritin (A; 24.0 μm) and smallest in saline solution containing 80 mmol l⁻¹ sorbitol (C; 8.0 μm). In standard saline solution (B), the initial diameter of the contractile vacuole was 14.6 μm. Consecutive video images of the contractile vacuole in the ferritin-exposed cell are shown in the upper portion of Fig. 1.

where D is the diameter of the vacuole, n is the total number of activated myosin–actin cross-bridges in the network (n is assumed to be constant if equal numbers of cross-bridges in the network are activated at any one time), S is the surface area of the vacuole membrane and κ is a constant that relates to the force generated by a single cross-bridge.

where R is the radius of the cross section of the pore, L is the length of the pore and η is the viscosity of the fluid. According to McKanna (1973), the pore is almost cylindrical, and R and L are approximately 0.5 μm and 1.2 μm, respectively. η is assumed to be equal to the value for water (1×10⁻³ N m⁻² s). The value for γ therefore approximates 2.1×10⁻¹⁷ N⁻¹ m⁵ s⁻¹.

where D₀ is the initial diameter (the diameter at time 0) and D is the diameter at time t. Equation 7 implies that D⁵ is proportional to t. This model will be termed the ‘contraction model’.

Equation 9 implies that D⁴ is proportional to t. This model will be termed the ‘surface tension model’.

which implies that D³ is proportional to t. This model will be termed the ‘cytosolic pressure model’.

To examine which of the three proposed models fits the in vivo decrease in vacuole diameter best, values for D³, D⁴ and D⁵ were plotted against time (Fig. 1). It is clear that the D³versus t plot was linear (Fig. 1, thick lines) for all initial vacuole diameters. This implies that the cytosolic pressure model most accurately describes the contractile vacuole in vivo, i.e. the fluid contents of the contractile vacuole of P. multimicronucleatum are expelled by the cytosolic pressure.

The values for the rate of discharge determined from the slopes of D³versus t in Fig. 1 were 8.9×10³ μm³ s⁻¹ for A, 5.4×10³ μm³ s⁻¹ for B and 2.0×10³ μm³ s⁻¹ for C. The rate was higher for larger vacuoles, which implies that the effective diameter of the contractile vacuole pore is larger in larger vacuoles. We found that the rate of discharge was proportional to D₀² (data not shown). The values for the rate of discharge after equilibration in four different external solutions (standard saline solution, saline containing 80 mmol l⁻¹ sorbitol or 33 μg ml⁻¹ cationized ferritin and axenic culture medium) are shown in Table 1. The normalized value for the rate of discharge was only significantly affected by the cationized-ferritin-containing saline solution (see below), implying that the cytosolic pressure remains approximately constant in the different external solutions used. The internal pressure of the vacuole was calculated from the rate of discharge in standard saline solution, according to the Law of Hagen and Poiseuille, to be approximately 2.0×10² N m⁻² from control values for the 80 mmol l⁻¹ sorbitol and ferritin experiments shown in Table 1.

In order to examine further the possibility of involvement of an actin-mediated contractile process in the discharge of vacuole fluid, we determined the effects of external application of 0.29 mmol l⁻¹ cytochalasin B (a potent inhibitor of actin polymerization) on the rate of fluid discharge from the vacuole. This drug did not alter the rate of discharge (4.6×10³±0.8×10³ μm³ s⁻¹, N=10, compared with the control rate of 4.6×10³±0.8×10³ μm³ s⁻¹, N=7) or the linear D³versus t relationship, although it inhibited actin-mediated food vacuole formation in the cells examined (Allen et al. 1995; Allen and Fok, 1983). This result indicates that an actin-mediated contractile process involving a membrane-associated cytoskeleton is not involved in the generation of the internal pressure required to expel vacuole fluid in P. multimicronucleatum. The lack of any ultrastructural or immunological evidence for the presence of a fibrous network system around the contractile vacuole of Paramecium (Allen and Fok, 1988, for P. multimicronucleatum; Cohen et al. 1984, for P. caudatum) that could account for contraction of the vacuole is also consistent with this finding.

The fluid-segregation mechanism of the contractile vacuole complex in P. multimicronucleatum is situated in the decorated spongiomes (Allen et al. 1990). To determine whether this mechanism affects the mechanism used to generate the fluid-expulsion force, fluid discharge was examined in cells in which fluid segregation had been partially blocked by injection of a monoclonal antibody (DS-1) raised against the decorated spongiomes in P. multimicronucleatum (Ishida et al. 1993). As shown in Table 1, injection of this antibody reduced the rate of fluid segregation by approximately 55 % and the pulsation frequency by approximately 34 %, which resulted in a sharp reduction in the total fluid output from these cells. In spite of this functional retardation of fluid segregation, the linear D³versus t relationship and the rate of fluid expulsion during systole were not affected (Table 1).

Morphological changes correlated with functional retardation of the fluid-segregation mechanism in response to injection of DS-1 were sought using fluorescence microscopy. Fig. 2B shows representative micrographs of a cell injected with DS-1 and Fig. 2A shows a control cell injected with nonspecific IgG_2b, an immunoglobulin of the same serotype as monoclonal antibody DS-1. The micrographs clearly demonstrate that the decorated spongiomes were markedly disrupted by DS-1. This result is consistent with our previous observations (Ishida et al. 1993).

Increased external osmolarity caused by the addition of 80 mmol l⁻¹ sorbitol to the standard saline solution resulted in decreases in D₀, pulsation frequency and rate of fluid segregation (Table 1), which together resulted in a pronounced decrease in the total fluid output by the cell. However, the linearity of the D³versus t relationship (Fig. 1) and the rate of fluid discharge during systole (Table 1) were not affected. Using fluorescence microscopy, the label for the decorated spongiomes, but not the smooth spongiomes, was markedly reduced in cells placed in the sorbitol-containing saline (compare Fig. 2D with Fig. 2C) (Ishida et al. 1996).

Cationized ferritin (33 μg ml⁻¹) added to standard saline solution caused a significant increase in D₀ and decrease in the pulsation frequency. However, the rate of fluid segregation over the relatively short duration of the experiments was unaffected by the presence of cationized ferritin (Table 1). This is consistent with the observation that the decorated spongiome was little affected by the presence of cationized ferritin (compare Fig. 2E with Fig. 2C). It should be noted, however, that the normalized rate of fluid discharge during systole was significantly reduced by cationized ferritin even though the linearity of the D³versus t relationship remained unchanged (Fig. 1). The lower rate of discharge could correspond to a decrease in the functional diameter of the pore. We assume that cationized ferritin somehow inhibits the mechanism of pore opening, in which fusion of the vacuole membrane with the pore membrane is critical.

It should be noted that morphological and physiological disruption of the fluid-segregation mechanism has little effect on the linearity of the D³versus t relationship. This observation strongly supports the hypothesis that the fluid-segregation mechanism per se is not intimately tied to the mechanism for generating the internal pressure of the contractile vacuole used to expel the fluid.

It is assumed that the cell’s cytosolic pressure is greatly reduced or effectively eliminated by rupturing the cell. We therefore examined the effects of cell rupture on the time course of the change in vacuole diameter during systole. Cells equilibrated in standard saline solution and placed in a thin space formed between a glass slide and a coverslip were gradually squeezed by pushing the coverslip against the glass slide using a micromanipulator, until the cells were flattened and eventually ruptured. In some cases, the contractile vacuole of the disrupted cell underwent systole after cell rupture.

Fig. 3 gives two representative graphs showing the change in diameter during systole of two contractile vacuoles with different initial diameters (22.9 μm in A and 14.3 μm in B) derived from two different ruptured cells. Values calculated for D², D³, D⁴ and D⁵ are also shown. Consecutive video images of the larger vacuole are shown in the upper portion of the figure.

The period of systole in contractile vacuoles from ruptured cells was markedly prolonged. The plot of D²versus t was always linear irrespective of the initial diameter of the vacuole (thick lines in Fig. 3), whereas the D²versus t plots for contractile vacuoles in unruptured cells were distinctly curved (Fig. 1). These results suggest that the cytosolic pressure model is no longer applicable to contractile vacuoles after cell rupture. Electron micrographs of cells fixed immediately after disruption demonstrate a clear rupture of the surface membrane and outflow of a considerable amount of cytosol through the rupture (data not shown). Disruption undoubtedly brings about a sudden reduction in cytosolic pressure.

As described above, contractile vacuoles from ruptured cells showed systole with a time course that differed from that predicted by any of the models presented in equations 7, 9 and 11. This suggests the presence of a novel mechanism for fluid discharge in ruptured cells.

where α is the slope of the line. Equation 12 implies that the rate of decrease in the vacuole’s membrane area (πα/4) is constant during vacuole discharge. There is strong morphological evidence, at least in ciliates, to show that the membrane of the contractile vacuole in vivo is transformed during systole into tubules that do not separate from the vacuole membrane (Allen and Fok, 1988). An electron micrograph showing this kind of membrane tubulation in a ruptured cell is presented in Fig. 4. Although the mechanism for transformation of an approximately planar membrane into tubules is not completely understood, the rate of this physical event is assumed to be constant under a given physical condition. Such tubulation would presumably lead to a constant reduction in the vacuole membrane area surrounding the undischarged fluid.

Equation 16 implies that D² is proportional to t. This model will be termed the ‘membrane area-proportional tension model’. This model fits the observed change in contractile vacuole diameter in ruptured cells and suggests that there is a causal relationship between membrane tubulation and the membrane area-proportional tension.

where V₀ is the vacuole volume immediately before the start of discharge. The work done by the vacuoles shown in Fig. 3A,B is therefore 7.4×10⁻¹³ N m and 1.8×10⁻¹³ N m, respectively.

where ΔE is the energy stored per unit area, δ is the bending modulus of the membrane and S_o is the curvature of the membrane. Although this formula is intended for isotropic curvature, we applied it, taking a value of 0.05 nm⁻¹ for the curvature of the membrane obtained from the reciprocal of the radius of a cross section of the tubule (approximately 20 nm; see Fig. 4). The calculated energy storage values were 7.6×10⁻¹³ N m for the vacuole shown in Fig. 3A and 3.0×10⁻¹³ N m for that in Fig. 3B. These values correspond extremely well with the work done by each vacuole determined from the rate of decrease in the volume of the vacuole. This coincidence suggests that the bending energy stored in the flattened vacuole membrane is sufficient to discharge the vacuole contents in ruptured cells as the membrane transforms into tubules.

We clearly demonstrated above that the fluid contents of the in vivo contractile vacuole of P. multimicronucleatum were discharged by cytosolic pressure and that tension of the vacuole membrane generated in association with membrane tubulation may be responsible for fluid discharge when the cytosolic pressure is eliminated. The rates of discharge calculated from the slope α of the D²versus t plots shown in Fig. 3 are much lower than those found prior to rupturing the cell and calculated from the slope of the corresponding D³versus t plots (4.7×10² μm³ s⁻¹versus 4.1×10³ μm³ s⁻¹ in A and 1.9×10³ μm³ s⁻¹versus 6.1×10³ μm³ s⁻¹ in B; D₀ is normalized to 12.0 μm). The times required for complete discharge were calculated from each value for the rate of discharge and found to be 2.9 s versus 0.22 s in A and 0.70 s versus 0.15 s in B. This indicates that, in a normal (unruptured) cell, tubulation of the vacuole membrane is always preceded by fluid discharge due to cytosolic pressure. Therefore, membrane tubulation is not likely to facilitate fluid discharge in vivo where fluid discharge is dependent solely on the cytosolic pressure.

We have described a novel mechanism that may facilitate some types of exocytic discharge. As exocytic vesicles vary in size and content, it is highly probable that some cells and organisms employ mechanisms similar to those described here for the contractile vacuole for discharging their vesicular contents. In addition, forces generated by the bending of the membrane of exocytic vesicles may contribute to the mechanism of exocytosis by causing the membrane to tubulate. These forces should be considered, along with other mechanisms, when the discharge of exocytic contents is under investigation.

Appendix

This section presents the derivation of the equations that appear in the text. Symbols have the same meaning as in the text.

C and C′ in the above equations are integration constants for each calculation.

This work was supported by NSF Grant MCB 95 05910. We thank Dr Y. Hiramoto and Dr M. Yoneda for their valuable comments on the modeling and Dr H. Seki and Dr H. Yamagishi for help with the equipment.