Calcium ions initiate the selective depolymerization of the pellicular microtubules in bloodstream forms of Trypanosoma brucei

Most of the biochemical studies of microtubules appearing in the literature over the last 15 years have centred on the mechanism of assembly and disassembly of mammalian brain microtubules (for reviews, see Roberts & Hyams, 1979; Pachter, Liem & Shelanski, 1984; Dustin, 1984; Ponstingl, Little & Krauhs, 1984). However, a number of recent biochemical studies on microtubules in eukaryotic microorganisms have revealed significant and interesting differences from their mammalian counterparts (Messier, 1971; Haber, Peloquin, Halvorson & Borisy, 1972; Maekawa & Sakai, 1978; Roobol, Pogson&Gull, 1980; Ireland etal. 1982; Dawson, Gutteridge & Gull, 1983; Lataste et al. 1984). These differences probably reflect at least in part the numerous specialized structures constructed from microtubules in eukaryotic microorganisms that are absent from mammalian cells. For example, all kinetoplastid flagellates and a number of other unicellular organisms contain a pellicular array of microtubules of various configurations, but all attached to the inner surface of the plasma membrane (Vickerman, 1969; Angelopoulos, 1970; Messier, 1971; Holberton, 1981; Roth, 1957; D’Haese, Mehlhorn & Peters, 1977; Russell & Sinden, 1982; Hofmann & Bouck, 1976; Murray, 1984a). At least in the case of Trypanosoma brucei the pellicular array of microtubules and the flagellum remain attached to the plasma membrane during its purification (Voorheis, Gale, Owen & Edwards, 1979), providing a convenient source of relatively pure material for the biochemical study of both of these tubulin-based structures.

Both pellicular and flagellar microtubules are relatively stable structures; they are not disrupted by cold and are resistant to most of the common drugs that affect microtubular assembly and disassembly in mammalian systems. In particular, nothing that is known about the assembly or disassembly of the pellicular array of microtubules (Anketell & Lagnado, 1983; Steiger, Wyler & Seebeck, 1984; Russell, Miller & Gull, 1984) could reasonably be expected to operate physiologically in vivo.

In this study we report that Ca²⁺ initiates the specific depolymerization of the pellicular microtubules of T. brucei, using purified plasma membrane-microtubule complexes as a source of the pellicular array. The disassembly process released assembly-competent pellicular tubulin without the release of any detectable trace of a microtubule-associated protein.

Our results suggest that a Ca²⁺-dependent mechanism may operate in vivo for the specific, reversible and controlled depolymerization of individual pellicular microtubules that probably accompanies certain stages of cell growth and cell division.

The source and history of Trypanosoma brucei 427-12/ICI-060 (MITat 1 · 1) has been described elsewhere (Martin, Voorheis & Kennedy, 1978; Voorheis et al. 1979; Holder & Cross, 1981).

A specific anti-tubulin rat monoclonal antibody (YL1/2) raised against yeast tubulin was a kind gift from Dr John Kilmartin, Medical Research Council, Molecular Biology Unit, Hills Road, Cambridge, England. Peroxidase-conjugated goat anti-rat immunoglobulin G (IgG) was obtained from Cappel Laboratories, 237 Lacey Street, West Chester, Pennsylvania, U.S.A. ¹²⁵I-labelled protein A was a generous gift from Dr Cormac Kilty and Dr Paddy Joyce, Department of Zoology, University College, Dublin, Ireland.

All other chemicals were obtained from either Sigma Chemical Co. Ltd, Fancy Road, Poole, Dorset, England or BDH Chemicals Ltd, Poole, Dorset, England.

Collodion bags for concentration of protein solutions were obtained from Sartorious Gmbh, 3400 Göttingen, Federal Republic of Germany.

The ELISA wells used were ‘Removastrip Greiner’ from Medlabs Ltd, Unit IC, Stillorgan Industrial Park, Stillorgan, Dublin, Ireland.

Bloodstream forms of T. brucei were grown in laboratory rats and harvested according to the method of Lanham (1968), incorporating a minor modification (Owen & Voorheis, 1980). It also has been found that trypanosomes survive longer if the buffer described by Lanham (1968) is supplemented with sucrose (2·8%, w/v).

Plasma membranes containing the attached pellicular microtubule complex and flagellum were isolated according to the method of Voorheis et al. (1979).

Plasma membrane-microtubule complexes were washed three times in phosphate-glutamate buffer (20mM-Na₂ HPO₄, lOOmM-glutamic acid, 1 mM-β-mercaptoethanol, adjusted to pH7·0 with 1 mM-NaOH) at 4°C by centrifuging in a minifuge for 20 s at 9000 g and resuspended at 37°C in phosphate-glutamate buffer containing CaCl₂ (100μM) and incubated at 37°C for 30 min. The membranes and flagella were separated from the released pellicular tubulin dimers by cooling the mixture for 20 min on ice and then centrifuging at 9000 g for 10 min. The supernatant and the pellet were used for experiments immediately.

The supernatant from the disassembly procedure for pellicular microtubules was concentrated at 4°C in collodion bags under reduced pressure until the concentration of tubulin reached 5 mg ml⁻¹. At this stage the released tubulin appeared essentially pure as assessed by sodium dodecyl sulphate/polyacrylamide gel electrophoresis (SDS/PAGE), with only very minor amounts of other proteins present. High molecular weight microtubule associated proteins (HMW-MAPS) and τ-factor were not detected. The final step in the purification of the pellicular tubulin was accomplished by chromatography on phosphocellulose columns (Weingarten, Suter, Littman & Kirschner, 1974).

Tubulin was purified from bovine brain along with the microtubule-associated proteins according to the cycle procedure of Shelanski, Gaskin & Cantor (1973), as modified by Berkowitz, Katagiri, Binder & Williams (1977). Tubulin was further purified on phosphocellulose columns as described by Weingarten et al. (1974).

Microtubule seeds from bovine brain were prepared by resuspending the final pellet of microtubules from the cycle purification method in MES buffer (0·l M-MES, lmM-EGTA, 0·5mM-MgCl₂, 1 mM-β-mercaptoethanol, pH6 · 4) at 25°C using a small glass homogenizer and then cooling on ice for 30 min to depolymerize the microtubules before centrifuging at 0—4°C for 3 h at 250 000 g. The pellet, containing microtubule seeds, was resuspended in phosphate-glutamate buffer (100μl for every 100g brain cortex starting material).

Calmodulin was purified from bovine brain by the method of Kakiuchi et al. (1981).

The assay was based on the method of O’Farrelly et al. (1983). Samples containing only soluble protein were used directly. Samples containing particulate protein were solubilized in 20 μl of a solution of 30 M-urea before use. In both cases the final soluble solutions were diluted with coating buffer (15mM-Na₂CO₃, 35mM-NaHCO ₃, 3mM-NaN ₃, pH9 · 6) so that the final concentration of tubulin was in the range 0·l-l·0μgml⁻¹. Portions (100μl) of these coating solutions were incubated overnight at 4°C in standard polycarbonate ELISA wells. The wells were then washed with four successive portions (100μl each) of phosphate-buffered saline (2mM-KH₂PO₄ 4, 8mM-Na2HPO4, 3 mM-KCl, 150mM-NaCl, pH7·4) containing Tween-20 (0·05%, w/v). The wells were then incubated (25°C) for 2·5h with 100μl each of rat anti-tubulin IgG (0·2μgml⁻¹ in phosphate-buffered saline-Tween), washed as before and then incubated (25°C) for a further 2·5-h period with 100μl each of peroxidase-linked anti-rat IgG (0·2μgm l⁻¹> in phosphate-buffered saline-Tween) and once again washed as before. The wells were finally incubated (25°C) with 100μl each of substrate solution (0·00012% H₂O₂, w/v; 0·4mgml⁻¹o-dianisidine,25mM-citric acid, 50mM-NaH₂PO4, pH5 ·6) for 30 min followed by the addition to each well of 100 μl of 2·5 M-H₂SO₄. The absorbance of the solution in each well was measured at 492nm and compared with a standard curve prepared from phosphocellulose-purified tubulin from T. brucei. On a number of occasions parallel standard curves, using phosphocellulose-purified tubulin from beef brain cortex and from T. brucei, were compared. On each occasion the standard curves were identical, regardless of the source of tubulin. Additional controls showed that in all cases under the conditions of this assay the polycarbonate well sites were far from saturated following the incubation with tubulin samples, even when they were derived from impure sources such as solubilized fractions of cells.

The procedure of Vaessen, Krieke & Groot (1981) was used for immunoblotting, with one modification. After transfer of the proteins from the gel to the nitrocellulose sheet, the sheet was incubated in phosphate-buffered saline (10mM-Na₂HPO₄, 150mM-NaCl, pH7·2) containing bovine serum albumin (8%, w/v) for 1 h and then transferred to a freshly prepared urea solution (8M) for 1 h. The urea was removed by washing the sheet overnight in 200 ml of phosphate-buffered saline followed by five successive fresh washes for 1 h each. The sheet (3 cm × 6cm) then was immersed in the solution (5 ml) of primary antibody (20 μg antibody/ml phosphate-buffered-saline containing bovine serum albumin, 8%, w/v). The remainder of the procedure was exactly as described by Vaessen et al. (1981).

Samples were prepared for electrophoresis by precipitation with trichloroacetic acid (5%, w/v) and centrifuging for 5 s at 9000 g in a minifuge. In the case of isoelectric focusing (IEF/PAGE) after treatment with trichloroacetic acid, the pellets were dissolved in 10 μl of 8 M-urea solution and after 1 h mixed with 40μl of the sample buffer of O’Farrell (1975). In the case of SDS/PAGE the pellets of sample protein, following treatment with trichloroacetic acid, were dissolved in the sample buffer of Laemmli (1970) and boiled for 3 min.

Both IEF/PAGE and two-dimensional gel electrophoresis (first dimension, IEF/PAGE; second dimension, SDS/PAGE) were carried out according to the method of O’Farrell (1975).

SDS/PAGE was performed according to the method of Laemmli (1970), using gels (0 ·7mm thickness) of either 10% (w/v) acrylamide or 5% to 15% acrylamide gradients.

Bands of protein were visualized by staining with Coomassie Brilliant Blue.

Electron microscopy of whole cells has been described previously (Voorheis et al. 1979). Pellets of isolated plasma membrane-microtubule complexes, either before or after treatment, were suspended in 0·05 M-phosphate buffer (pH 6·8) containing 2% (w/v) glutaraldehyde and fixed by incubating at 25°C for 2h. After six washes with phosphate buffer alone, the pellets of complexes or membranes were post-fixed in phosphate buffer containing 0·5% (w/v) osmium tetroxide at 25°C, washed once in phosphate buffer and stained in 1% (w/v) uranyl acetate for 12h at 40°C. The stained pellets were dehydrated by passage through successively increasing concentrations of acetone and finally propylene oxide before embedding in Epon-Araldite. Thin sections were stained in lead citrate (Reynolds, 1963).

Microtubules were negatively stained with 1% (w/v) aqueous uranyl acetate on Formvar grids.

Thin sections and negatively stained specimens were examined using an Hitachi HU-12A electron microscope.

The assembly of tubulin from the disassembled pellicular microtubules was measured at 30°C by the increase in light scattering at 350 nm in a medium composed of phosphate-glutamate buffer (for composition see Materials and Methods), containing lmM-EGTA, 1 mM-MgCl₂ and 1 mM-GTP. The final concentration of tubulin used was 1 mg m l⁻¹. The reaction was initiated by adding microtubule seeds from beef brain (final concentration, 10μgm l⁻¹).

The morphology of the assembled microtubules was assessed using electron microscopy of samples of negatively stained material taken from the incubation mixtures.

Incubation of isolated plasma membrane-microtubule complexes with Ca²⁺ (100μM) for 30 min at 37°C resulted in the complete depolymerization of the pellicular microtubules and complete retention of the flagellar microtubules, as assessed by electron microscopy of thin sections (Fig. 1).

The time course of the Ca²⁺-stimulated release of tubulin from purified plasma membrane-microtubule complexes of T. brucei demonstrated an initial rapid rise in the release of tubulin followed by a plateau region without further release that was reached by 20min at 37°C in the presence of 100μM added Ca²⁺ (Fig. 2). The final extent of release under these conditions approached 53% of the total tubulin in the membrane complexes.

When the level of added Ca²⁺ was varied, the extent of tubulin released during a 30-min incubation period also varied. The release of tubulin from the pellicular microtubules was 90% complete at 100μM-Ca²⁺ added (Fig. 3). At very high levels of Ca²⁺ (1-2 mM) the pellicular microtubules were depolymerized, as assessed by electron microscopy, but no tubulin was released to the suspending medium (data not shown).

When the experiments studying either the time course (Fig. 2) or the effect of calcium concentration (Fig. 3) on the release of pellicular tubulin were repeated in the presence of purified calmodulin (50μgm l⁻¹) from bovine brain, no effect could be detected. Under similar experimental conditions in our hands calmodulin led to the rapid depolymerization of bovine brain microtubules (data not shown). In addition, the inhibitors of calmodulin-mediated processes, trifluoperazine (10⁻⁵μM) and compound 48/80 (10⁻⁵gm l⁻¹) did not inhibit or stimulate release of pellicular tubulin from the membrane complexes when incubated in phosphate-glutamate buffer at 37°C in the presence of Ca²⁺ (10μM and 100μM) for 30 min (Table 1).

When the supernatant from the incubation of plasma membrane-microtubule complexes with Ca²⁺ was subjected to SDS/PAGE, a single protein of apparent molecular weight 55 000 was usually detected. On some occasions (Fig. 4) the released protein migrated as a doublet with apparent molecular weights of 52 000 and 55 000. Numerous attempts to identify the parameter affecting the separation of these two components on SDS/PAGE, including systematic alteration of the pH of the running gel, source and purity of the SDS as suggested by Best, Warr & Gull (1981), purity of the acrylamide and bisacrylamide, amount of protein loaded per gel track, batch of plasma membrane-microtubule complex and use of the gel system of Yang & Criddle (1970), as recommended by Stieger et al. (1984), all proved unsuccessful. The release of other proteins, including HMW-MAPS and τ-factor has never been detected.

Samples of supernatant, following incubation of complexes with Ca²⁺, were also subjected to IEF/PAGE. In this case, regardless of the separation, on SDS/PAGE, three peptides were always resolved, having isoelectric points of 5·40, 5·38 and 5·8 (Fig. 5). Two-dimensional gels (1st dimension, IEF/PAGE; 2nd dimension, SDS/PAGE) revealed that whenever separation in the SDS dimension was achieved, the pl values of 5·40 and 5·38 belonged to the 55 000 M_r species (presumably α-tubulin), while the pl of 5·08 belonged to the 52000 M_r species (presumably β-tubulin) (Fig. 6).

Western blots of the protein released from the plasma membrane-microtubule complex revealed that it strongly cross-reacted with a specific rat monoclonal IgG anti-yeast-tubulin (Fig. 6). The same antibody also recognized a single protein of the same molecular weight as the released protein in blots of whole cells and isolated plasma membrane—microtubule complexes.

After the plasma membrane-microtubule complexes had been treated with Ca²⁺ for a period sufficient for release of tubulin to be complete the membranes and flagella were removed by centrifugation. The Ca²⁺ in the supernatant solution was chelated with excess EGTA. In the presence of GTP and Mg ²⁺ the tubulin assembled into microtubules in a phosphate-glutamate buffer when mammalian brain tubulin seeds were added (Fig. 7). The resulting assembled microtubules had a normal morphology (Fig. 8).

The most striking finding to emerge from this study was the specific depolymerization of the pellicular microtubules that occurs in the presence of calcium ions. After 30 min of incubation with 100μM-Ca²⁺ the pellicular microtubules were removed completely from the plasma membrane-microtubule complex, as judged by electron microscopy of multiple thin sections. In contrast, the flagellar microtubules remained completely intact. In all of the many thin sections examined, neither the central pair of flagellar microtubules nor any of the nine flagellar outer doublet microtubules were missing following treatment with calcium ions.

The unique specificity of this treatment has made it possible to separate the pellicular and flagellar tubulin easily and completely. The pellicular tubulin appears to be composed of one beta and two alpha isotypes as judged from isoelectric focusing gels. These results are different from those reported by Russell et al. (1984) for the composition of the pellicular tubulin of Crithidia fasciculata. These workers found that one beta and three alpha isotypes composed the pellicular tubulin that was released by differential sonication of the Triton-extracted pellet of broken cells of C. fasciculata. One possible explanation offered by Russell et al. (1984) for the heterogeneous composition of the pellicular alpha tubulin found in their study was that of contamination of their preparation by tubulin from the flagellum. However, these authors considered that only one of the alpha isotypes could have arisen in this manner. Therefore, there is a distinct possibility that the pellicular microtubules from both organisms have a similar content of two α- and one β-tubulin isotypes.

It was significant that treatment with Ca²⁺ did not release any proteins other than tubulin. We have never observed the presence of any microtubule-associated proteins similar to those found in mammalian microtubule preparations, such as HMW-MAPS or τ-complex. It is clear from electron micrographs of whole cells (Meyer & de Souza, 1976; Vickerman, 1969; Messier, 1971; Bordier, Garavito & Armbruster, 1982; Souto-Padrón, De Souza & Heuser, 1984) and of isolated plasma membrane-microtubule complexes (Voorheis et al. 1979) that at least one associated protein must exist, because the pellicular microtubules are cross-linked to each other and to the membrane by a regular array of bridges. The proteins corresponding to these bridges must have remained attached to the plasma membrane during release of the pellicular tubulin in our experiments. Consequently, at least one bridge protein remains to be identified. Since the intertubule bridge protein(s) were not released it is also possible that they are independently linked to the membrane.

An alternative explanation for the absence of any microtubule-associated proteins in our preparations is degradation as a result of protease activity. However, this possibility seems unlikely, for the following reasons. First, the serine protease inhibitor, phenylmethylsulphonyl fluoride, was present at all stages. Second, the pH of the incubation medium was never low enough to permit acid proteases to act. Third, although the release of tubulin is conducted in the presence of 2-mercapto-ethanol and Ca²⁺, which could activate either a thiol protease or a divalent metal iondependent protease, when whole complexes were prepared in the absence of these components and in the presence of EDTA the polypeptide pattern was unchanged (Voorheis et al. 1979).

It is interesting that the preparation of pellicular tubulin purified from C.fasciculata by sonication and successive cycles of assembly/disassembly also did not contain any accessory proteins (Russell et al. 1984).

Recently, the removal of the pellicular microtubules from the unicellular green alga, Distigmaproteus, has been the subject of an elegant study (Murray, 1984b). In this case Ca²⁺ also was found to be a sufficient initiator of the release process, although much higher concentrations were required (1mM) than in the case of T. brucei. Murray (1984b) does not comment on the condition of the pellicular tubulin from D. proteus following its removal from the membrane. In the case of T. brucei the tubulin remains assembly-competent after its removal from the plasma membrane. Single microtubules of normal morphology and dimensions are formed with a random orientation in free solution when the released pellicular tubulin is assembled in phosphate-glutamate buffer with added GTP, Mg²⁺ and EGTA in the presence of brain microtubule seeds (i.e. rings). However, despite numerous attempts, we have not succeeded in assembling and organizing either the released disassembled pellicular tubulin from T. brucei or brain tubulin onto the surface of stripped membranes in the pattern of a regular array. In the case of D. proteus reassembly of the microtubular array from bovine brain tubulin on the membrane of this alga has been accomplished (Murray, 1984b). It could be that the microtubule assembly sites on the plasma membrane of D. proteus maintain a rigidly fixed orientation in the membrane during the release process, since at least some of the surface antigens of the related organism Euglena gracilis are known to be extremely immobile (Hofmann & Bouck, 1976). In contrast to this view of the surface of these algae it is possible that the microtubule assembly sites on the inner surface of the plasma membrane of T. brucei may be free to move in the plane of the membrane after removal of the microtubular array, as is the major surface antigen on the outer surface of this organism (Barry & Vickerman, 1977; Barry, 1979). The mobility of the microtubular attachment sites produced by our treatment might be expected to interfere with an assembly process in vitro that under conditions in vivo normally results in the assembly onto the membrane of a regularly spaced array.

During the normal reproductive life of any individual cell of T. brucei, the presence of a cross-linked pellicular microtubule complex completely surrounding the inner surface of the organism poses special problems with regard to the mechanism of cell division. It seems reasonable that cytokinesis could only be completed by depolymerizing a minimum of two of the pellicular microtubules, one on each side of the cell or, alternatively, detaching all of the cross-bridges between two pairs of adjacent microtubules, one pair on each side of the organism, in order to permit propagation of the division furrow through the cell. It is not known which of these two alternative mechanisms is used by T. brucei. In fact, we do not know yet whether the bridge proteins or the pellicular tubulin itself has the receptor site for Ca²⁺ and, therefore, initiates the depolymerization of the microtubules. However, the inherent polarity of microtubules could impose a simplifying constraint on our concept of the process of cell division if the initiation of cytokinesis was found to occur at the minus end of the array. Selective exposure of the minus end of as few as two of the pellicular microtubules might be expected to lead to their rapid depolymerization, thus establishing the eventual cleavage plane for cell division. Cytokinesis then would proceed toward the plus ends of the microtubules, simplifying our concept of the concerted growth of the new microtubules during cell division required to complete the full complement of pellicular microtubules for both daughter cells. Such a model depends critically upon the polarity of the pellicular microtubules and is, therefore, testable. Consequently, we suggest that determination of the polarity of the pellicular microtubules in T. brucei should receive high priority. In addition, Baum et al. (1981) have shown that cell division in T. cruzi will not proceed in the presence of the microtubule-stabilizing agent, Taxol. We note that the loop of endoplasmic reticulum always associated with the four pellicular microtubules adjacent to the flagellum (Fuge, 1968; Taylor & Godfrey, 1969; Vickerman, 1969) could act as a separate compartment, allowing the concentration of Ca²⁺ to rise locally in order to initiate cytokinesis. It is interesting that synthesis of the new flagellum and also the new endoplasmic reticulum loop embracing the four microtubules adjacent to the new flagellum occurs before cytokinesis (see, e.g. fig. 3 of Vickerman, 1969), as would be required by our hypothesis. Furthermore, this observation leads to the suggestion that the polarity of the flagellar microtubules will be found to be opposite to that of the pellicular microtubules where the two structures lie adjacent and parallel to each other.

Calcium has already been shown to regulate the assembly and disassembly of microtubules in mammalian cells (Marcum, Dedman, Brinkley & Means, 1978; Welsh, Dedman, Brinkley & Means, 1978). Both cold-sensitive microtubules (Marcum et al. 1978) and cold-stable microtubules (Job, Fischer & Margolis, 1981) disassemble in the presence of Ca²⁺. In both cases,, calmodulin greatly potentiates the effect of Ca²⁺ and, interestingly, in both cases trifluoperazine fails to inhibit the process (Perry, Brinkley & Bryan, 1980; Jobet al. 1981). However, the disassembly of cold-sensitive microtubules requires approximately 1 μM-Ca²⁺ for half-maximal disassembly while mammalian cold-stable microtubules require 100μM-Ca²⁺ for half-maximal disassembly (Job et al. 1981). The pellicular microtubules in T. brucei are cold-stable and the process of their disassembly requires 72μM-Ca²⁺ for halfmaximal effect (Dolan & Voorheis, 1985), which is close to the 100μM required for mammalian cold-stable microtubules. Furthermore, the disassembly process in T. brucei is also resistant to inhibition by trifluoperazine. However, one important difference between the pellicular microtubules in T. brucei and the cold-stable microtubules in rat brain does exist: the disassembly of pellicular microtubules is not potentiated by the addition of bovine brain calmodulin (Table 1). One word of caution is required in interpreting these last experiments with calmodulin. It remains possible, although perhaps unlikely, that the binding sites for calmodulin in T. brucei are sufficiently distinct from their mammalian homologues to make the addition of bovine brain calmodulin ineffective in this protozoal system. Certainly a number of structural differences do exist between the calmodulin from mammals and that from trypanosomes (Ruben, Egwuago & Patton, 1983). Although they are both effective in stimulating several mammalian enzymes it remains to be determined unequivocally whether they are both sufficiently potent in stimulating an appropriate target system in trypanosomes. It is interesting to note that the cAMP phosphodiesterase in T. cruzi (Goncalves, Zingales & Colli, 1980) and in T. brucei (C. Lowry, unpublished) is not sensitive to mammalian calmodulin, although its expected sensitivity to trypanosomal calmodulin has not been tested.

This work was supported in part by a grant from the Provost’s Appeal Fund, Trinity College Dublin and by a studentship to M.T.D. from the Department of Education, Eire.