Cell Motility and the Cytoskeleton 34331-94 (1996) In Vivo Microtubule Dynamics During Experimentally Induced Conversions Between Tubulin Assembly States in A llogromia laticollaris Elizabeth A. Welnhofer and Jeffrey L. Travis Deparfment of Biological Sciences, University at Albany, SUNY, Albany, New York A distinctive property of foraminiferan tubulin is that, in addition to microtubules (MTs), it exists in an alternate assembly state, helical filaments. Here, we have examined in vivo MT dynamics during experimentally induced conversions between these two assembly states in the reticulopods of the marine foraminiferan Allogromia laticollaris. Exposure to high extracellular concentrations of Mg2 (165 mh4) resulted in a complete conversion of MTs into helical filaments. However, Mg2+ treatment also induced a retrograde movement of organelles and cytoplasm, and it was necessary to inhibit this response in order to assess the effects of assembly state changes on individual MTs. This was accomplished by simultaneous treatment with high extracellular Mg2+ and 2,4-dinitrophenol (DNP). The resulting loss in MTs was detected by video enhanced DIC (VECDIC) microscopy as either an endwise MT shortening (at an average rate of 474 p d m i n ) or transformation into one or more irregularly shaped fibrils, which we termed residual fibrils. Correlative immunofluorescence and video microscopy showed residual fibrils to be composed of helical filaments. Removal of extracellular Mg2+/DNPinitiated a reversal in assembly state, from helical filaments into MTs, which was completed within 5 min. VEC-DIC microscopy showed that MTs reformed by an endwise lengthening at an average rate of 216 pn!min. These results suggest that conversion between alternate tubulin assembly states provides a more rapid means to build and dismantle MTs than conventional subunit-driven pathways. 0 1996 Wiley-Liss, Inc. + Key words: helical filaments, tubulin lattice transformations,microtubule behavior, cytoskeletaldynamics, Foraminifera INTRODUCTION The reticulopodial networks elaborated by foraminifera undergo continuous and rapid microtubule-dependent changes in morphology. These dynamic cellular appendages, composed largely of an array of interconnected filopods, are continuously remodeled as individual filopods independently extend, retract, branch, or fuse with neighboring filopodia. An extensive network of microtubules (MTs) serves as the major cytoskeleton of the filopods and powers their motility [Travis and Allen, 1981; Travis et al., 1983; Travis and Bowser, 1986a,b]. The MT-mediated filopodial movements can be extremely rapid. For example, filopods may extend 0 1996 Wiley-Liss, Inc. and retract at rates from 1-10 p d s e c [Jahn and Rinaldi, 1959; Allen, 1964; Bowser and DeLaca, 19851, which is at least an order of magnitude faster than the actin-mediated filopod motility that occurs in vertebrate cells [reviewed in Condeelis, 19931. In order to effect such rapid Received September 27, 1995; accepted February 15, 1996. Elizabeth A. Welnhofer’s current address is Department of Anatomy and Cell Biology, School of Medicine, University at Buffalo, SUNY, Buffalo, NY 14224. Address reprint requests to Jeffrey L. Travis, Department of Biological Sciences, University at Albany, Albany, NY 12222. 82 Welnhofer and Travis filopod movements, foraminifera must employ mechanisms to reorganize the MT cytoskeleton quickly. Foraminiferan tubulin is unusual because it can exist in two distinct assembled states in vivo: typical 13 protofilament MTs and a novel polymer type, termed helical filaments, consisting of approximately 5 nm filaments wound into a coil approximately 30 nm in diameter [Rupp et al., 1986; Golz and Hauser, 19861. Unlike MTs, helical filaments are unable to support bidirectional organelle transport [Rupp et al., 19861, presumably because they lack the linear protofilament lattice required by known MT motors like dynein and kinesin [Kamimura and Mandelkow, 19921. A number of experimental studies [reviewed in Travis and Bowser, 19911 have suggested that these alternate assembly polymorphs transform from one polymer “state” to the other in vivo. For instance, Bowser et al.  observed that the cell bodies of juvenile Allogromia contain massive stores of helical filaments that diminish as the young cells form their pseudopodial MTs. Helical filaments can be observed sporadically throughout well-established reticulopodial networks, and treatments known to cause the breakdown of MTs (e.g., cold, colchicine, and high Mg2+) cause a loss of MTs and a corresponding accumulation of helical filaments in the reticulopodia [Koury et al., 1985; Travis and Bowser, 1986al. We hypothesized that the reversible transformation between the MT and helical filament states enables foram MTs to be formed and broken down at velocities sufficient to support observed pseudopod movements. To test this hypothesis, we have examined in detail MT behavior during experimentally induced changes in the assembly state of tubulin in the reticulopods of Allogromia laticollaris. The resulting loss and subsequent reformation of individual MTs could be followed in real time with video enhanced differential interference (VECDIC) microscopy. Here we present the kinetic analysis of these events. The results support the above hypothesis, suggesting that unique tubulin assembly state changes occurring in foraminifera may represent an adaptation that facilitates rapid MT-dependent changes in cellular morphology. acid etched coverslips cationized with 2% Alcian Blue or carbon stabilized formvar coated gold grids treated with polylysine. Both of these polycationic surfaces induce reticulopods to form extremely thin lamellipodial regions in which individual microtubules can be visualized using VEC-DIC or whole mount electron microscopy [Travis et al., 19831. In some instances, cells were examined in microperfusion chambers [McGee-Russell and Allen, 19711 that had been modified to fit the stage of a Zeiss (Thornwood, NY) IM-35 microscope. These chambers allow continuous observation of specimens throughout the following experimental treatments. Cells were exposed either to 165 mM MgCl, in CaFSW or 165 mM MgC1, and 2 mM 2,4-dinitrophenol (DNP) in CaFSW to induce microtubule disassembly. Recovery from this treatment was initiated by perfusion with CaFSW. Video Microscopy The behavior of individual microtubules was assayed by VEC-DIC microscopy. Specimens were viewed with a Zeiss IM-35 microscope, equipped with oil immersion DIC optics. Video images obtained with a Hamamatsu (2-2400 newvicon camera were digitally enhanced and averaged (2 frames) in real time as described previously [Travis and Bowser, 19901 and recorded on either 3/4 inch U-matic or 1/2 inch VHS videotapes. Micrographs were recorded on T-Max 100 or Plus X-pan film directly from paused video frames on the monitor. Antitubulin lrnmunofluorescence Cells were fixed, permeabilized, and reduced as described by Rupp et al.  before incubating with DMlA (Sigma, St. Louis, MO) monoclonal tubulin antibody [Blose et al., 19841 and subsequently staining with fluorescein isothiocyanate labeled goat anti-mouse IgG. These preparations were then mounted in either 50% glycerol containing 3% N-propyl gallate or in Slow Fade (Molecular Probes, Eugene, OR) and viewed by epifluorescence microscopy on a Zeiss IM-35. Fluorescent images were recorded on T-Max-400 film, exposed at an IS0 of 400, and developed in T-Max developer. Whole Mount Electron Microscopy A. laticollaris were individually plated on formvarcoated gold grids. These were fixed for 1 h in 5% glutAllogromia laticollaris were cultured as described araldehyde with .04% tannic acid in 0.1 M Na-cacodylby Travis and Allen . Cells were washed several ate buffer (pH 7.4), rinsed in 0.1 M cacodylate buffer, times in calcium free sea water (CaFSW) (390 mM and then transferred to a grid holder designed to fit into NaCl, 49 mM MgCl,, 26 mM Na2SO,, 8 mM KC1, 2 the chamber of a critical point dryer (Denton DCP-1). mM NaHCO,), buffered to pH 8.1 with 10 mM TRIS- The specimens were post-fixed in 0.5% OsO,, and proHCI. In all experiments, CaFSW was supplemented with cessed as described by Travis et al.  and then 2 mM EGTA in order to preserve the reticulopod mor- examined in either a Philips (Mahwah, NJ) 201 or a phology during fixation. Cells then were plated on either Zeiss 902 transmission electron microscope. In some METHODS Cell Preparation and Experimental Treatments Tubulin Lattice Transformations In Vivo cases, specimens were viewed with the high voltage transmission electron microscope at the Wadsworth Center for Laboratories and Research in Albany, NY. Measurement of the Rate of Microtubule Dynamics Twenty-two microtubule shortening events that occurred during Mg2+/DNP treatment and 47 microtubule lengthening events that occurred during the first 3 min after removal of Mg2+/DNP were analyzed with a Hamamatsu DVS-3000 image processor. The location of microtubule ends at two time points during shortening or lengthening was marked with cursors and the distance between the cursors was measured using the calibrated distance function on the Hamamatsu processor. The time/date generator used in these experiments (Thelnors Electronic Lab, Ann Arbor, MI) tracked time in sec:field and therefore the calculated rates have a maximum resolution of approximately 30 msec. The two-tailed Student’s t-test was used to assess statistical differences in the rates of MT shortening and lengthening. Correlative Video Enhanced DIC Microscopy/lmmunof luorescence Cells were plated in mini-perfusion chambers constructed by placing two plastic strips (0.5 mm thick) on 24 X 60 mm no. 0 coverslips. An 18 X 18 mm coverslip was placed on top of the spacers after the cells were plated onto the bottom coverslip. A strip of filter paper was placed in one opening of the chamber to withdraw liquid. In this way, cells could be observed with VECDIC microscopy throughout experimental treatments and subsequent fixation. At the appropriate time, the cells were fixed for 2 min by perfusion with 2% glutaraldehyde in CaFSW with 10% sucrose followed by fixation for 20 min in 0.2% glutaraldehyde in PHEM buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, 2 mM MgCl,, pH 6.9). Fixation was immediate as evidenced by the instantaneous cessation of motility, and no changes in reticulopod morphology were detected at the level of resolution afforded by VEC-DIC microscopy (resolution approximately 0.2 pm.). Specimens were then processed for immunofluorescence as described above. RESULTS Mg*+-lnduced In Vivo Microtubule Conversion Into Helical Filaments The addition of MgCl, at concentrations greater than 165 mM to sea water (Mg2+ treatment) induced a classical pseudopodial withdrawal response in Allogromia laticollaris, similar to that described in Allogromia species [McGee-Russell and Allen, 1971; Koury et al., 83 1985; Rupp et al., 19861. Withdrawal was first marked by a change in cytoplasmic transport from the characteristic saltatory, bidirectional organelle transport to a primarily unidirectional movement of organelles and cytoplasm toward the cell body. Next, pseudopods detached from the substrate and began to retract towards the cell body. The retracting pseudopods appeared to lose their characteristic rigidity and became bent repeatedly along their length. Within 3 to 5 min after Mg2+ treatment, all organelle and pseudopodial movement ceased. In addition to a pseudopodial withdrawal response, Mg2+-treatment also caused a change in the assembly state of tubulin in the reticulopodia. Electron microscopy showed that the immobile pseudopods of Mg2+ treated specimens contained few, if any, MTs but displayed prominent aggregations of helical filaments (data not shown). The pseudopodial withdrawal that accompanied Mg2+-treatment prevented continuous observation of individual MTs. In an effort to immobilize the pseudopods, we plated organisms on Alcian Blue coated substrates [Rupp et al., 19861. These cationized surfaces induced formation of lamellipods that remained extended during Mg2+ treatment and subsequent recovery. Video enhanced DIC microscopy showed that Mg2+ treatment caused some MT bundles (see Fig. 1 for definition) within these lamellipods to bend into exaggerated serpentine shapes, as shown in Figure 1. Retrograde movement of organelles continued as the bundles progressively bent (Fig. 1a-d), but ceased as they abruptly broke into shorter, irregularly shaped fibrils (Fig. le). We operationally defined the fibrils formed from MT bundles by Mg2+ treatment as residual fibrils (Fig. le) because they persisted throughout this treatment. Besides transformation of MT bundles into residual fibrils, MT bundles also frequently disappeared by an apparent endwise shortening (not illustrated). However, this latter interpretation is suspect because there was still considerable cytoplasmic withdrawal, as evidenced by the retrograde movement of organelles and bulk cytoplasm. As such, MT movement could not be ruled out as the basis for MT behavior during Mg2+ treatment. Mg*+/DNP-lnduced Microtubule Conversion Into Helical Filaments In order to examine directly the effects of tubulin assembly state changes on MTs, it was necessary to inhibit the cytoplasmic withdrawal response that accompanied Mg2+ treatment. Previously we had shown that reticulopod motility (pseudopod movements, organelle and cell surface transport) depends on oxidative energy metabolism and is inhibited by simultaneous treatment with KCN and salicylhydroxamic acid (SHAM) [Travis and Bowser, 1986bl. We found that 2,4-dinitrophenol (DNP), a potent inhibitor of oxidative phosphorylation, 84 Welnhofer and Travis Fig. 1. The in vivo effect of Mg2+ treatment on MTs. In this video sequence, cells were plated on Alcian Blue to inhibit pseudopodial withdrawal during perfusion with 165 mM MgCl, in CaFSW (Mg2+ treatment). In this way, the behavior of MTs in lamellipodial regions could be followed using video enhanced DIC microscopy. Previous work has demonstrated that the long fibrils detected by video enhanced DIC microscopy in the lamellipodia of AlZogromia are composed of MTs [Travis et al., 19831, varying in number from 1-15 MTs. Throughout the rest of the paper, we will refer to these fibrils as MT bundles in recognition that they may be composed of one to several MTs. Immediately after perfusion with the Mg2+ solution, the majority of organelle transport was directed towards the cell body (in this sequence, towards the bottom of the figure). Within the next 30 sec, linear MT bundles (a, black arrowheads) progressively bend into a serpentine-shaped filament (b-d, black arrowheads), which abruptly snaps and transforms into several irregularly shaped fibrils (e, white arrowheads). We have defined these fibrils that derive from MTs after treatment with Mgz+ as residual fibrils. Time shown is in seconds. also inhibited reticulopod motility without affecting the assembly state of tubulin (data not shown). We reasoned that DNP might allow us to uncouple the Mg*+-induced tubulin assembly state transformations from the cytoplasmic withdrawal. Simultaneous treatment with 2 ,4-dinitrophenol and high extracellular Mg2+ (Mg2+/DNP) blocked the retrograde movement of organelles and bulk cytoplasm, but did not alter the effect of Mg2+ on the tubulin assembly state. Whole mount electron micrographs verified that the elaborate MT cytoskeleton found in untreated specimens (Fig. 2d) was transformed into helical filaments within 5 min in Mg2+/DNP (Fig. 2f). Aggregations of helical filaments were often arranged into wavy tracks that were thicker and shorter than MTs (Fig. 2e,f). The assembly state transformation could also be recognized with antitubulin immunofluorescence as indicated in Fig. 2a,b. Mg2+/DNPtreatment caused the pattern of antitubulin staining to change from the long, thin linear fibrils (Fig. 2a) typical of the MT bundles seen in control cells (Fig. 2d) to shorter, thicker, and irregularly shaped fibrils corresponding to aggregations of helical filaments (Fig. 2b and f). As such, antitubulin staining proved to be a convenient assay of the tubulin assembly state. At the level of video microscopy, Mg2+/DNP treatment resulted in dramatic structural changes in MT bundles. The type of structural change observed depended on the size of the MT bundle, which has been shown to vary between 1-15 MTs [Travis et al., 19831. Bundles composed of numerous MTs, as indicated by their large diameter (>200 nm) or more intense video signal, more frequently transformed into residual fibrils than did MT bundles composed of fewer MTs (Fig. 3). The Mg2+/DNP induced transformation of MT bundles into residual fibrils was not preceded by formation of serpentine-shaped MT bundles, but rather occurred abruptly (often in less than 1 sec) (compare Fig. 1 to Fig. 3). Video records played at slower than real time speed showed the transformation started at one end of the MT bundle and proceeded unidirectionally to the other end. Correlative video microscopy and anti-tubulin immunofluorescence showed that the Mg2+/DNP induced residual fibrils (Fig. 4b) represented accumulations of helical filaments (Fig. 4c). Smaller MT bundles disappeared by an endwise shortening after Mg2+/DNP treatment (Fig. 5). Because Mg2+/DNP treatment inhibited cytoplasmic withdrawal, this behavior most likely reflected a change in assembly state to helical filaments rather than movement of MTs. Consistent with this interpretation, particles that appeared to be associated with the MT bundles did not move as the bundles shortened. The rare occasions when both ends of the MT bundles were in the field of view provided further support that MT shortening did not result from MT movement (Fig. 6 ) . In these cases, the MT bundles did not move in the cytoplasm as they shortened from only one end. Fig. 2. Tubulin assembly states in the reticulopodia of untreated and Mg2+/DNPtreated Allogrornia laticollaris. a and b are epifluorescent micrographs of lamellipodial regions of the reticulopodial network stained with DMlA anti-tubulin; c-f are whole mount electron micrographs of similar regions. MTs can be observed throughout the reticulopodia in untreated specimens (a,c,d) while they are absent in cells exposed to Ca2+-free sea water supplemented with 165 mM MgCl and 2 mM DNP for 5 min (b,e,f). Instead, tubulin immunofluorescence (b) and whole mount electron microscopy (f) both reveal the presence of thick wavy fibrils (arrowheads in f) in the reticulopodia of the experimentally treated cells. A different, higher magnification whole mount electron micrograph (e) clearly shows that the wavy filaments are composed of rows of helical filaments. Here, the helical filaments exhibit the characteristic alternation of electron opaque and lucent bands, representing the coils and successive spaces between the coils in helical filaments. The scale bars for a and b; c and e; and d and fare equivalent. 86 Welnhofer and Travis Fig. 3. The in vivo effect of MgZf/DNP treatment on MTs. These two video micrographs illustrate the differential effects of Mg’+/DNP treatment on MT fibrils. a shows a lamellipodial region immediately prior to Mg2+/DNP treatment and (b) the same region after 1:OO min in this medium. In a, the MT bundles marked by black arrows exhibit a greater degree of contrast than the MT bundles marked by the white arrowheads. The higher contrast most likely results from a greater number of MTs. The larger bundles of MTs (a, black arrows) transformed into residual fibrils (b, black arrows) whereas the smaller MT bundles (a, white arrowheads) no longer were readily detectable by video microscopy along most of their length. In Vivo Conversion of Helical Filaments Into MTs Allogromia laticollaris rapidly and completely recovered from the effects of Mg2+/DNP treatment after being returned to normal medium. Immunofluorescent micrographs confirmed that the tubulin assembly state changed from helical filaments into MTs during recov- Fig. 4. Residual fibrils represent accumulations of helical filaments. In a, several MT bundles are detected immediately prior to perfusion with Mg2+/DNP. Within 30 sec (b), the MT fibrils marked by black arrowheads transform into residual fibrils. Other MT bundles (a, white arrow) disappear (b) by an endwise shortening. Tubulin immunofluorescence (c) of the same region shown in b shows a direct correlation between the residual fibrils and the tubulin staining pattern. The pattern of fluorescence corresponds to that established for helical filaments (compare c to Fig. 2b). Note that short tubulin containing fragments remained in the area of the MT bundles that disappeared (a, white arrow). The tubulin material is not readily detectable in the DIC images (a,b). Scale bar in c = 5 Fm. ery. As early as 30 sec after removal of Mg2+/DNP (Fig. 7a), long MTs could be detected adjacent to the short helical filament containing fibrils. The relative proportion of MTs increased (Fig. 7b) during the next 5 min, after which they were the major tubulin assembly state (Fig. 7c). Video microscopy showed that MT bundles reappeared by an endwise lengthening during recovery from Mg2+/DNP treatment (Fig. 8). Intracellular particles contacted by lengthening MT bundles remained stationary (Fig. 9) or transported in the direction opposite to that in which the MT end lengthened (Fig. 8). This observation is consistent with the interpretation that length- Tubulin Lattice Transformations In Vivo 87 treatment. Because Mg2 +/DNP treatment inhibited cytoplasmic withdrawal, the observed MT behavior was unlikely to be the result of MT movement associated with retrograde movement of organelles and cytoplasm. We restricted our analysis to events in which MT bundles shortened from their ends rather than transformation into residual fibrils because the former changes in length could be readily defined and measured with the methods used in this study. MT bundles shortened at an average rate of 7.9 pm per second; the maximum rate observed was 17.7 pm/sec (Table I). To approximate the rate that MTs form from helical filaments, we analyzed video records of MT bundle lengthening during recovery from Mg2+/DNP treatment. The average rate of MT bundle lengthening was 3.6 p-m/ sec, while the maximum observed lengthening rate was 11.6 p-m/sec. The difference between the rates of MT lengthening and shortening is statistically significant ( P < .001, Table I). DISCUSSION Fig. 5 . Endwise MT shortening induced by MgZ+/DNP. After perfusion with MgZ+/DNP, the MT bundle shown in a shortens at 7.2 km/sec. Notice that the particle (short arrow) that appears to be associated with the MT fibril in a remained in place as the fibril both shortened (thin arrow in b marks the visible end of the fibril) and then disappeared (c). Elapsed time is shown in minutes: seconds. The speed at which foraminifera remodel their reticulopodial networks indicates they have adopted mechanisms to rapidly build, dismantle, and reorganize their MT cytoskeleton. Foraminiferan tubulin exists in two distinct assembly states in vivo, MTs and helical filaments. In this study, we have measured the rates at which MTs are formed and taken apart during transformations between these two polymer states in vivo. Experimental Modulation of Tubulin Assembly States In Vivo Previous investigators have identified a number of experimental treatments that induce reversible tubulin asening represents reformation of MTs, not movement of sembly state transformations in foraminifera [reviewed preexisting MTs. in Travis and Bowser, 19901. However, these treatments There was no discernible pattern to MT reforma- also result in cytoplasmic movements , complicating obtion. Rather, MTs reformed randomly at numerous sites servation and analysis of MT behavior. For example, in throughout the reticulopod. The pools of helical filaments Figure 1, it was unclear whether the pronounced bending that had formed during Mg2+/DNP treatment were the of MT bundles was a consequence of cytoplasmic withonly sites where MTs could be predicted to form. As is drawal or a necessary stage in the formation of helical illustrated in Figure 10, MT bundles typically reformed filaments. Furthermore, apparent MT shortening during directly from residual fibrils during recovery from Mg2+/ Mg2+ treatment could have resulted from either the DNP. MT reappearance, however, was not restricted only movement of MTs during cytoplasmic withdrawal or the to these regions (Fig. 8). In addition, the direction in formation of helical filaments. which MTs lengthened with respect to the cell body varMg2+/DNP treatment allowed us to separate the ied, even within the same lamellipod (Fig. 9). tubulin assembly state transformations from cytoplasmic withdrawal. Our current observations suggest that MT Kinetics of MT Behavior During Mg2+/DNP bending is caused by the movements associated with cyTreatment and Subsequent Recovery toplasmic withdrawal, not by the conversion of MTs into To estimate the speed that MTs are taken apart as helical filaments. In Mg2 /DNP treated cells, larger MT they are converted into helical filaments, we calculated bundles could be observed transforming directly into rethe average rate of MT shortening during Mg*+/DNP sidual fibrils (see Fig. 3,4) without first undergoing + 88 Welnhofer and Travis Fig. 6. Mg2+/DNP induces MT shortening from one end. A MT fibril with both ends visible (a) shortens (6.0 F d s e c ) from one end (b) and disappears (c). This example clearly demonstrates that this behavior is not the result of MT movement, as the other end of the MT fibril (marked by white arrowhead) remained stationary throughout the entire shortening event. A particle (thick arrow) apparently associated with the shortening end of the MT fibril in a remained in place after the fibril disappeared (c). Time frame: seconds. bending) MT ‘‘translocations’ ’ mediate reorganization of the reticulopodial MT cytoskeleton [Travis et al., 1983; Travis and Bowser, 1988, 19901. Axial MT “translocations” may be due either to the movement of preformed MTs (sliding or gliding), or the formation and break down of MTs. The rates reported for axial MT “translocations’’ [Travis and Bowser, 19881 are very rapid (2-10 p d s e c ) and have been considered to be too fast to be accounted for by more conventional subunit-driven MT assembly and disassembly, which typically occurs at rates between 0.1-0.5 p d s e c in vivo [Cassimeris et al., 1988; Sheldon and Wadsworth, 19931. Previous workers therefore argued that MT movements mediate MT reorganization and subsequent remodelling of reticulopods [reviewed in Travis and Bowser, 19901. However, the speeds measured for MT shortening (average rate = 7.9 pmhec) and lengthening (average rate = 3.6 p d s e c ) during Mg2+/DNPtreatment and recovery show that this behavior occurs fast enough to account for axial MT “translocations. ” Foraminifera may exploit changes in tubulin assembly state to rapidly build, dismantle, and remodel their MT cytoskeleton during reticulopod motility and morphogenesis. Rapid transformations between alternate assembly states of tubulin may be a common mechanism employed by protozoans to accelerate MT formation and break down. There are at least two other known examples of protozoans that undergo rapid MT-dependent changes in Implications for Models of Reticulopod Motility cellular morphology and possess an alternate tubulin asIn order to accommodate the rapid and continuous sembly state. An extensive arrangement of MTs supports remodeling of foram reticulopods, the MT cytoskeleton the stalk in the protozoan Actinocoryne contractillis. must be reorganized equally as fast. Previous investiga- This appendage is remarkable in that it can contract 60% tions have established that axial (defined as apparent MT of its length (about 150 pm) within 2-8 msec [Febvrelengthening or shortening) and lateral (defined as MT Chevalier, 1980, 19811. The extremely rapid stalk conu bending as in Figure 1. Furthermore, in Mg2+/DNP treated cells, we have interpreted the apparent MT shortening as a manifestation of the transformation of MTs into HFs, rather than movement of MTs that is associated with the cytoplasmic withdrawal process. Inhibitors of energy metabolism, such as DNP (see Results) and KCN/SHAM [Travis and Bowser, 1986bl are effective at blocking motility (organelle and pseudopod movements) in foraminifera, presumably by lowering the ATP concentration within the reticulopods. Although we have not been able to measure the ATP content in pseudopods, measurements have shown that DNP reduces whole cell ATP levels by 40% (Travis and Bernhard, unpublished observations). We assume that DNP reduces pseudopodial ATP concentration by at least as much because bidirectional organelle transport and pseudopod motility is blocked in the energy-poisoned pseudopods. Thus, our results suggest that while cytoplasmic withdrawal is an energy-dependent process, the transformation of MTs into helical filaments may be an energy-independent process. In contrast, MTs in cultured metazoan cells are stabilized in energy poisoned cells, and do not disassemble in the presence of colchicine, vinblastine, and nocodazole [Bershadsky and Gelfand, 1981; DeBrander et al., 19811. These differences in MT behavior may point to fundamental differences between the MTs of foraminifera and metazoan cells. v U I Tubulin Lattice Transformations In Vivo 89 bvre-Chevalier and Febvre, 19921. These investigators suggest that this rapid disassembly occurs as MTs are severed along their lengths and the fragments are transformed into coiled ribbons. Helical filaments have also been detected in the dynamic pseudopods of the freshwater protozoan Reticulomyxu, where the reported rates for in vivo MT lengthening and shortening are nearly as rapid as those in Allogromiu [Chen and Schliwa, 19901. Possible Mechanisms for Rapid Conversion of Tubulin Assembly States Fig. 7. The MT cytoskeleton reforms after recovery from Mg2+/DNP treatment. Specimens were maintained in Mg2+/DNP for 5 min and subsequently fixed after transfer to normal CaFSW for (a) 30 sec, (b) 1 min, and (c) 3 min. The tubulin assembly state was assayed by tubulin immunofluorescence. The long, thin fibrils (marked by arrowheads in a,b) are composed of MTs (see Fig. 2). The thicker, shorter and wavy fibrils (marked by thick arrow in b) are composed of helical filaments (see Fig. 2). With increasing time after recovery, the propoaion of tubulin in a helical filament state decreased as that in the MT state increased. By 3 min, MTs represented the major tubulin assembly state. traction is accompanied by a breakdown in the MT cytoskeleton and the formation of coiled ribbons, similar in morphology to the helical filaments in foraminifera [Fe- The conversion between MTs and helical filament states may be direct or indirect. As discussed below, a direct transition can better account for both the observed kinetics of MT formation and break down and the behavior of MTs during assembly state changes. In an indirect transition, tubulin subunits are intermediates in the conversion between helical filaments and MTs (Fig. 11). The apparent length changes of MT bundles during conversion of tubulin assembly states then would result from the endwise loss or addition of tubulin subunits. MT lengthening at the rates we report here for Allogromiu would require tubulin subunits to add onto the end of the MT lattice at an average rate of 5,850 per second. MT shortening at Allogromiu rates would require subunits to dissociate from the MT lattice at a rate of 12,800 per second (Table 11). These values are at least an order of magnitude higher than those reported in the literature for in vivo MT assembly/disassembly involving subunit associatioddissociation (Table 11). The parameters that influence subunit association and dissociation in MT assembly and disassembly in metazoan cells have been defined from in vitro tubulin polymerization studies [Voter and Erikson, 1984; Walker et al., 19881. The rate of MT shortening depends on the rate that tubulin-GDP (KoffjDp)dissociates from the MT lattice. Three parameters govern the rate for MT lengthening by endwise subunit association: (1) tubulin subunit concentration, (2) association constant for tubulin-GTP (K onGTP),and (3) dissociation constant for tubulin-GTP (K ofpTP).It is possible that foraminifera have evolved mechanisms to modulate these parameters so as to increase the rate of MT elongation or shortening. Weakening of tubulin-tubulin subunit interactions in the MT lattice is one way to accelerate dissociation of tubulin subunits from shortening MTs. Gal et al. [ 19881 observed that tubulin subunits dissociate faster in vitro from MTs when divalent cations are present at high concentrations. These investigators argued that saturation of low affinity sites for divalent cations on the MT lattice destabilized the tubulin subunit interactions and facilitated the increase in tubulin subunit dissociation. However, the rates reported by Gal et al.  cannot ac- 90 Welnhofer and Travis Fig. 8. MTs reform by an endwise lengthening. After specimens were incubated in Mg2+/DNP for 5 min to induce the formation of helical filaments, CaFSW was perfused through the chambers to initiate recovery (a). A MT bundle appears in a region in which residual fibrils were absent. The MT fibril progressively lengthens, at a velocity of 2.8 p d s e c , from the top of micrograph (b) towards the bottom (c). After apparently being contacted by the lengthening MT fibril, two particles (marked by arrowheads in b) remained in place for a brief moment and then were subsequently transported in a direction opposite that in which the fibril lengthened. The time frame is displayed in the upper left-hand corner in sec:fields. Fig. 9. The direction of MT lengthening varies during recovery from Mg2+/DNP treatment. In this sequence of video micrographs, arrowheads mark the end of a lengthening MT bundle. In a, a MT bundle extends (6.7 p d s e c ) from the lower left-hand corner until it reached the edge of the membrane (b). After being contacted by this MT bundle, the particle (a) realigns relative to the MT bundle (b, arrow) but remains stationary as the MT bundle lengthens. Immediately following this event, another MT bundle (arrowhead) emerges from the upper right-hand corner (c) and lengthens until it reaches the edge of the membrane at the bottom of the micrograph (d). Time elapsed is shown in seconds in the lower left-hand corner. Tubulin Lattice Transformations In Vivo Fig. 10. Transformation of a residual fibril into an MT bundle. a shows a residual fibril that formed as a result of MgZ+/DNPtreatment. The particles associated with the residual fibril (arrowheads) remained immobile throughout this treatment. After perfusion with normal medium, the residual fibril abruptly transformed into a linear fibril (b) 91 and then lengthened (c). Bi-directional organelle transport resumed along the linear fibril. This suggests the fibril was composed of MTs rather than helical filaments, which do not support bidirectional particle movement. TABLE I. Rates of MT Behavior During Experimentally Induced Changes in Tubulin Assembly State Average rate ? s.d. (wdsec) Lengtheninga S horteningb 3.6 7.9 k 1.4 k 3.5 Maximum rate (wdsec) nc 11.6 17.7 41 22 "Endwise lengthening of MT bundles that occurred during the 5 min time frame when tubulin polymer was converted from helical filaments into MTs. bEndwise shortening of MT bundles that occurred during the first 5 min after MgZf/DNP treatment when MTs converted into helical filaments. 'n = number of observations. count for the rates of MT shortening in Allogromia laticollaris. MTs were estimated to shorten as fast as 100 pm/min in vitro in the presence of divalent cations, whereas the fastest shortening in Allogromia laticollaris was 1,300 p d m i n . Furthermore, the observation that MT bundles sometimes directly reorganize into aggregates of helical filaments is difficult to explain by an indirect transition as it would require not only unusually rapid disassembly of MTs into tubulin subunits, but also the rapid assembly of tubulin subunits into helical filaments. The rate of MT elongation can be increased by suppressing the dissociation of tubulin-GTP (KofpTP). In vitro studies have shown that the microtubule associated proteins MAP2 and tau increase the rate of MT assembly onto flagellar axoneme templates by this mechanism [Pryer et al., 19921. However, complete suppression of KofPTPduring elongation would result in only a twofold increase in elongation velocity and this is still I Microtubule Helical filament 1 ., Tubulin subunits Fig. 11. Diagrammatic representation of alternative mechanisms for conversion between MTs and helical filaments. In the direct transition model (A), the MT lattice transforms directly into the coiled structure of the helical filament as a result of modifications in tubulin-tubulin subunit interactions. In the indirect transition model, conversion of tubulin protein from MTs into helical filaments occurs in at least two distinct steps. MTs first depolymerize into tubulin subunits (B), followed by reassociation of tubulin subunits into helical filaments ( C ) . For the formation of MTs by the direct transition model, a subset of tubulin subunit bonds reform as the helical filament transforms into a MT lattice (A). In the indirect transition model, MT reformation would first require helical filaments to disassemble into tubulin subunits (C), which could then assemble into MTs (B). too small to account for the accelerated rates of MT reformation in Allogromia. One possible way to explain the increased rate for MT lengthening in Allogromia laticollaris is that the intracellular tubulin concentration may be higher than that found in metazoan cells (10-20 pM). Localized high concentrations of tubulin could be generated in foraminifera by the depolymerization of helical filaments into 92 Welnhofer and Travis TABLE 11. A Comparison of Microtubule (MT) Dynamics in Allogrornia and Fibroblasts Average rate of MT shortening Average rate of MT lengthening Allogromia laticollaris CHO Fibroblasts' pdmin Subunitsisec" pmimin Subunits/secb 216 5,850 489 414 30 12,800 813 18 "If MT assembly results from the association of tubulin dimers to the end of the MT polymer, this is the predicted rate based on a 13 protofilament MT with 1,625 subunits/ w . bAssuming MT disassembly results from the loss of tubulin dimers from the ends of the shrinking polymer, this is the predicted rate based on a 13 protofilament MT with 1,625 subunits/pm. 'These measurements were taken from Sheldon and Wadsworth . They measured the in vivo rates for MT shortening and lengthening during observations of MT dynamic instability in Chinese Hamster Ovary fibroblasts injected with rhodamine-labeled tubulin. These are the fastest in vivo rates for MT dynamics reported in metazoan cells. tubulin subunits. However, if one uses the values for the association rate and dissociation rate of tubulin generated from the in vitro assembly kinetics of purified brain tubulin [Walker et al., 19881, the concentration of tubulin required to elongate a MT at a rate of 180 pm/min would be 0.555 mM. We calculate that the concentration of tubulin in a MT is 9 mM. As such, tubulin concentrations greater than 0.5 mM could be achieved in localized areas if helical filaments spontaneously disassemble into subunits. However, it should be noted that aggregates of helical filaments, seen as residual fibrils with video microscopy, did not spontaneously disappear before MTs formed. Rather, MTs appeared to form directly from residual fibrils, as is illustrated in Figure 10. Perhaps the most attractive model that has been proposed for tubulin assembly state changes in Allogromia is the direct transition model, originally proposed by Hauser and Schwab [ 19741. In this model, a local change in tubulin subunit-subunit interactions causes the tubulin lattice to transform directly from one state to the other (Fig. 11). This change would occur without a loss or addition of tubulin subunits and as such could ensue faster than an indirect transition because fewer intersubunit bonds would need to be broken or formed. In the direct transition model, the apparent MT shortening during Mg2+/DNP treatment would result from the MT lattice transforming into helical filaments that are not detectable at the level of VEC-DIC microscopy. Consistent with this interpretation, we observed small patches of antitubulin staining remaining along the path where MTs had shortened (Fig. 4). On the other hand, larger bundles of MTs would yield a greater number of helical filaments and we suggest these are detected by VEC-DIC microscopy as residual fibrils. If MTs and helical filaments directly convert, then the observations made in this study indicate that this structural change occurs in an endwise manner. This may be important, for instance, in forming MTs with the correct polarity or dismantling MTs in the distal regions of the reticulopod first during pseudopod withdrawal. In a direct transition model, MT bundles would appear to lengthen as the tubulin lattice transformed from helical filaments into MTs. As such, the location of MT reformation would be determined by the location of helical filaments and would occur at multiple sites, not from one centralized region as in metazoan cells. Indeed, the regions where accumulations of helical filaments had formed and were detected as residual fibrils by video microscopy were where MTs reformed after removal of Mg2 IDNP. Oligomeric forms of tubulin may also have a role in MT assembly/disassembly pathways in vertebrate cells. Under certain buffer conditions in vitro, tubulin protofilaments can form extensively coiled structures, and time resolved cyro-electron microscopy studies have shown that MTs disassemble in vitro by releasing coiled protofilament fragments [Mandelkow et al., 19911. However, a number of observations make it unlikely that the in vitro coiled tubulin oligomers are identical to the Allogromia helical filaments. Electron microscopic studies have revealed that Allogromia MTs transform into a single helical filament [Golz and Hauser, 1986; Travis and Allen, 19811 (Welnhofer and Travis, manuscript in preparation) whereas one would expect to see as many as 13 if they were protofilaments that had peeled from MT lattice. Allogromia helical filaments appear to be much more stable than the coiled protofilaments formed in vitro from vertebrate tubulin. Helical filaments are observed in vivo in untreated cells and are stable enough to remain intact in detergent extracted cell models of reticulopodia (Welnhofer and Travis, manuscript in preparation). This stability may allow Allogromia to employ helical fila+ Tubulin Lattice Transformations In Vivo ments as a storage and transport form of prefabricated, or at least preorganized, microtubule proteins. CONCLUSIONS The foraminifera appear to have evolved mechanisms to build and dismantle their MT cytoskeleton at rates much faster than those observed in metazoan cells. These mechanisms involve changes in the tubulin assembly state between MTs and helical filaments. The rate of MT behavior during tubulin assembly state changes in forams suggests that they are mediated by a direct lattice transformation. Furthermore, a direct transition between MTs and helical filaments is supported by in vitro experiments (Welnhofer and Travis, manuscript in preparation) in which we observed a reversible transformation between MTs and helical filament in detergent lysed cells. ACKNOWLEDGMENTS We are indebted to Drs. Sam McGee-Russell and Sam Bowser for their advice and helpful discussions, and we thank Drs. Bob Hard and Roger Sloboda for critically reading the manuscript. We would also like to thank Dr. Joan Bernhard for assistance with ATP assays. This work was supported in part by grant MCB 95-05855 from the National Science Foundation. REFERENCES Allen, R. D. (1964): Cytoplasmic streaming and locomotion in marine foraminifera. In Allen, R. D. and Kamiya, N. (eds.): “Primitive Motile Systems in Cell Biology.” New York: Academic Press, pp. 407-432. Bershadsky, A. D. and Gelfand, V. I. (1981): ATP-dependent regulation of cytoplasmic microtubule disassembly. Proc. Natl. Acad. Sci. U S A . 78:3610-3613. Blose, S., Meltzer, D., and Feramisco, J. (1984): 10 nm filaments are induced to collapse in living cells microinjected with monoclonal and polyclonal antibodies against tubulin. J. Cell Biol. 98:847-858. Bowser, S. S. and DeLaca, T. E. (1985): Rapid intracellular motility and dynamic membrane events in an Anatarctic foraminifer. Cell Biol. Int. Rep. 9:901-910. Bowser, S. S., McGee-Russell, S. M., and Rieder, C. L. (1984): Multiple fission in Allogromia sp.. , strain NF (Foraminiferida): Release, dispersal and ultrastructure of offspring. J. Protozool. 28~151-157. Cassimeris, L., Pryer, N. K., and Salmon, E. D. (1988): Real time observations of microtubule dynamics in living cells. J. Cell Biol. 107:2223-2231. Chen, Y.-T. and Schliwa, M. (1990): Direct observation of microtubule dynamics in Reticulomyxa: Unusally rapid length changes and microtubule sliding. Cell Motil. Cytoskel. 17:214-226. Condeelis, J. (1993): Life at the leading edge: the formation of cell protrusions. AMU Rev. Cell Biol. 9:411-444. DeBrander, M., Geuns, G., Nuydens, R., Willebrords, R. and De- 93 Mey, J. (1981): Microtubule assembly in living cells after release from nocodazole block: The effect of metabolic inhibitors, tax01 and pH. Cell Biol. Intern. Rep. 5:913-920. Febvre-Chevalier, C. (1980): Behavior and cytology of Actinocoryne contractilis, Nov. gen., Nov. Sp., a new stalked helizoan (Centrohelidia): Comparison with the other related genera. J. Mar. Biol. 60:909-928. Febvre-Chevalier, C. (1981): Preliminary study of the motility processes in the stalked helizoan Actinocoryne contractilis. BioSystems 14:337-343. Febvre-Chevalier, C. and Febvre, J. (1992): Microtubule disassembly in vivo. Intercalary destabilization and breakdown of microtubules in the Helizoan Actinocoryne contractilis. J. Cell Biol. 1181585-594. Gal, V., Martin, S., and Bayley, P. (1988): Fast disassembly of microtubules induced by magnesium or calcium. Biochem. Biophys. Res. Commun. 155:1464-1470. Golz, R. and Hauser, M. (1986): Polymorphic assembly states of Allogromia tubulin under physiological conditions. Eur. J Cell Biol. 40:124-129. Hauser, M. and Schwab, D. (1974): Microtubules and helical microfilaments in the cytoplasm of the foraminifer Allogromia laricollaris Arnold, investigations with vinblastine and deuterium oxide to demonstrate a close relationship. Cytobiologie 9:263279. Jahn, T. L. and Rinaldi, R. A. (1959): Protoplasmic movement in the foraminiferan, Allogromia laticollaris, and theory of its mechanism. Biol. Bull. 117:loO-118. Kamimura, S. and Mandelkow, E. (1992): Tubulin protofilament and kinesin-dependent motility. J. Cell Biol. 118:865-875. Koury, S. T., Bowser, S. S., and McGee-Russell, S. M. (1985): Ultrastructural changes during reticulopod withdrawal in the foraminiferan protozoan, Allogromia sp., strain NF. Protoplasma 129:149-156. Mandelkow, E.-M., Mandellcow, E., and Milligan, R. (1991): Microtubule dynamics and microtubule caps: A time-resolved cryo-electron microscopy study. J. Cell Biol. 114:977-991. McGee-Russell, S. M. and Allen, R. D. (1971): Reversible stabilization of labile microtubules in reticulopodial networks of Allogromia. Adv. Cell Mol. Biol. 1:153-184. Pryer, N., Walker, R., Skeen, V., Bourns, B., Soboeiro, M., and Salmon, E. (1992): Brain microtubule-associated proteins modulate microtubule dynamic instability in vitro. J. Cell Sci. 103~965 -976. Rupp, G., Bowser, S. S., Mannella, C., and Reider, C. L. (1986): Naturally occurring tubulin-containing paracrystals in Allogromia: Immunocytochemical identification and functional significance. Cell Motil. Cytoskeleton 6:363-375. Sheldon, E. and Wadsworth, P. (1993): Observation and quantification of individual microtubule behavior in vivo: Microtubule dynamics are cell-type specific. J. Cell Biol. 120:939-945. Travis, J. L. and Allen, R. D. (1981): Studies on the motility of the foraminifera. I. Ultrastructure of the reticulopodial network of Allogromia laticollaris (Arnold). J. Cell Biol. 90:211-221. Travis, J. L. and Bowser, S. S. (1986a): A new model of reticulopodial motility and shape: Evidence for a microtubule based motor and an actin skeleton. Cell Motil. Cytoskeleton 6:2-14. Travis, J. L. and Bowser, S. S. (1986b): Microtubule-dependent reticulopodial motility: Is there a role for actin? Cell Motil. Cytoskeleton 6: 146-152. Travis, J. L. and Bowser, S. S. (1988): Optical approaches to the study of foraminiferan motility. Cell Motil. Cytoskeleton 10: 126-136. Travis, J. L. and Bowser, S . S. (1990): Microtubule-membrane inter- 94 Welnhofer and Travis actions in vivo. Direct observation of plasma membrane deformation mediated by actively bending cytoplasmic microtubules. Protoplasma 154:184-189. Travis, J. L. and Bowser, S . S. (1991): The motility of foraminifera. In Lee, J. J. and Anderson, 0. R. (eds): “Biology of the Foraminifera.” London: Academic Press, pp. 91-155. Travis, J. L., Kenealy, J. F. X., and Allen, R. D. (1983): Studies on the motility of the foraminifera. 11. The dynamic microtubular cytoskeleton of the reticulopodial network of Allogromiu luticolluris. J. Cell Biol. 97:1668-1676. Voter, W. A. and Erikson, H. P. (1984): The kinetics of microtubule assembly. J. Biol. Chem. 259:10430-10438. Walker, R. A., O’Brien, E. T., Pryer, N. K., Soboeiro, M. F., Voter, W. A., and Erickson, H. P. (1988): Dynamic instability of individual microtubules analyzed by video light microscopy: Rate constants and transition frequencies. J. Cell Biol. 107: 1437-1448.