Evidence for a non-tubulin spindle matrix and for spindle components immunologically related to tektin filaments

The machinery for cell division, the spindle, is formed by centrosomes, chromosomes and microtubules, where microtubules provide a link between centrosomes and chromosomes. In spindles of higher eukaryotes the microtubule organization appears fairly complex (Fuge, 1977) and it is generally believed that by polymerizing into microtubules, tubulin alone forms the structural framework of the spindle essential for chromosome movement. However, several pieces of evidence indicate the presence of additional structural components. The presence of such components is indicated by experiments where UV microbeam irradiation was used to produce localized microtubule discontinuities within a chromosome fiber. The disruption of the microtubule bundle caused by UV microbeam irradiation, however, did not prevent the chromosomes moving polewards in crane fly spermatocytes (Wilson and Forer, 1988). Furthermore, connections between microtubules within a half-spindle have been elegantly demonstrated by micromanipulation experiments on grasshopper spermatocytes (Nicklas et al. 1982). In those experiments, lateral movement of a chromosome produced with a micromanipulation nee dle not only had an effect on microtubules of the attached chromosome fiber, it also bent microtubules of the neighboring fibers. Our knowledge about components responsible for microtubule interconnections or a filamentous matrix is very limited.

Previous studies on echinoderms have suggested that mitotic spindles contain a non-tubulin matrix. Isolated spindles from sea urchin eggs, when treated with 1 mM free calcium, lose almost all their tubulin but still retain a spindle shape (Hays and Salmon, 1983; Leslie et al. 1987; Rebhun and Palazzo, 1988). Rebhun and Palazzo (1988) reported that Ca²⁺-extracted spindles of sea urchin eggs are mainly composed of a ˜55 kDa polypeptide with an amino acid composition similar to intermediate filament proteins. Possibly in relation to these observations, we found that affinity-purified, polyclonal antibodies raised against sea urchin tektins (Linck et al. 1987) cross-react with intermediate filament proteins (Steffen and Linck, 1989a). Tektins were originally described as integral, filamentous components of doublet microtubules of sea urchin sperm axonemes (Linck and Langevin, 1982). On the basis of the biochemical characterization (Linck and Langevin, 1982; Linck and Stephens, 1987) and predicted secondary structure as determined from cDNA sequence data (J. M. Norrander, L. A. Amos and R. W. Linck, unpublished), tektins are a distinct family of structural proteins having certain features in common with intermediate filaments such as a high αhelical content, the filamentous nature of the proteins, a similar amino acid composition, and a similar but distinct secondary structure.

If a non-tubulin matrix is a fundamental property of spindles of dividing cells, one might expect to find related components in a wide range of higher eukaryotes. In this study we provide evidence for the presence and localization of tektin-like spindle components in molluscan and mammalian cells. Furthermore, we demonstrate that these components are associated with tubulin-depleted spindle remnants.

Axoneme storage buffer (10 mM Tris-HCl, pH 8.0, 150 mM KC1, 5 mM MgSO₄, 0.5 mM EDTA), bovine serum albumin (BSA), Dulbecco’s modified Eagle’s medium (DMEM), ethylene glycol-bis(succinic acid N-hydroxysuccinimide ester) (EGS), fetal calf serum (FCS), lysis buffer (10 mM Pipes, pH 6.6, 5 mM EGTA, 1 mM MgCl₂, 20% glycerol, 1% Nonidet P-40), phosphate-buffered saline (PBS: 10 mM phosphate, 150 mM NaCl, 2.7 mM KC1, pH 7.4), pig kidney epithelia cells (LLC-PK₁), Pipes buffer (100 mM Pipes, pH 6.9, 2 mM EGTA, 1 mM MgCl₂), polyacrylamide gel electrophoresis (PAGE), sodium dodecyl sulfate (SDS), spindle storage buffer (50 mM Pipes, pH 6.8, 2 mM EGTA, 1 mM MgCl₂), Tris-buffered saline (TBS: 20 mM Tris-HCl, 500 mM NaCl, pH 7.5).

Polyclonal antibodies were raised against tektins A, B and C from sperm axonemes of two sea urchin species, Lytechinus pictus and Strongylocentrotus purpuratus, and affinity purified as described earlier (Linck et al. 1987). Briefly, tektins were isolated from sea urchin sperm axonemes by two extractions with 0.5% Sarkosyl/2 M urea, followed by a preparative SDS-PAGE purification of the individual proteins. Antibodies were raised in rabbits and affinity purified over a Sepharose-4B column covalently coupled with SDS-denatured tektin filaments. In this study only affinity-purified antibodies have been employed. The monoclonal anti-tubulin (α-tub) was raised against tubulin derived from S. purpuratus axonemal microtubules and was found to be specific for acetylated α- tubulin, like the antibody raised by Pipemo and Fuller (1985). A more detailed characterization of this antibody will be provided elsewhere (Steffen and Linck, in preparation).

For immunofluorescence microscopy a biotin-conjugated goat anti-mouse IgG or a biotin-conjugated goat anti-rabbit IgG from Sigma (St Louis, MO) and Texas-Red-conjugated streptavidin from Molecular Probe (Eugene, OR) were employed as detection systems. For immunoblotting a peroxidase-conjugated goat anti-mouse IgG or a peroxidase-conjugated goat anti-rabbit IgG from BioRad (Richmond, CA) were used. For immunoelectron microscopy a 5 nm gold-conjugated goat anti-rabbit IgG (Steffen and Linck, 1989a) was used.

Ovaries of ripe surf clams (Spisula solidissima) were dissected and incubated in sea water to release their gametes. Eggs were twice filtered through a 100 pm nylon filter and washed several times with excess amounts of filtered sea water. Washed eggs could be stored on ice for 1-–2 days before use. They were artificially activated by adding up to 70 mM KC1 to an egg suspension of 5 ml settled eggs in 50 ml sea water (Allen, 1953). Spindle formation was monitored, using a microscope equipped with polarization optics. For immunofluorescence microscopy whole eggs were placed onto poly-L-lysine-coated coverslips 2–3 min before they reached metaphase. They were fixed either in −20°C methanol or in 5 mM EGS (100 mM EGS stock in dimethyl sulfoxide) in 50 mM Pipes, pH 6.8, 5 mM EGTA, 1 mM MgCl₂ for 10 min.

Isolated spindles were prepared according to Suprenant and Rebhun (1984). Briefly, artificially activated eggs were washed first with 1 M glycerol, 10 mM sodium phosphate, pH 7.8, and then washed with unbuffered 1 M glycerol. After the second wash, pelleted eggs were resuspended in lysis buffer (10 mM Pipes, pH 6.6, 5 mM EGTA, 1 mM MgCl₂, 20% glycerol, 1% Nonidet P-40). To pellet eggs between washes a hand-crank centrifuge (Curtin Matheson Scientific, Eden Prairie, MN) was used (˜1000 revs/min for 15 s). Isolated spindles were pelleted in a clinical table-top centrifuge (IEC-428, International Equipment Co., Needham Hts., MA) at setting no. 6 for 2 min and gently resuspended in lysis buffer to separate clean spindles from aggregated cell debris. To obtain clean spindle preparations for fractionation experiments, spindles were filtered through a 48 μm nylon filter and washed twice with lysis buffer without detergent. The purity of the spindle preparation was monitored by phase-contrast and polarization microscopy. For immunofluorescence microscopy isolated spindles were placed onto poly-L-lysine-coated coverslips and fixed in − 20°C methanol, Pipes-buffered 3.7% formaldehyde or Pipes-buffered 5 mM EGS for 10 min.

For SDS-PAGE analysis twice-washed spindles were divided into aliquots and incubated in spindle storage buffer containing various concentrations of Ca²⁺ for 30 min at room temperature or on ice. Spindle remnants were pelleted in a clinical table-top centrifuge at setting no. 7 for 5 min. Solubilized components were obtained by precipitation either on ice in 80% ice-cold acetone containing 4.5% ammonium hydroxide or in 10% trichloroacetic acid (TCA) and analyzed by SDS-PAGE. High-salt, Sarkosyl and urea extractions were carried out under similar conditions, i.e. on ice for 30 min. Remnants from these extractions were pelleted at 100,000 g_av for 60 min. All fractionation experiments were carried out on freshly prepared spindles.

To obtain axonemes, Spisula sperm was released by dissecting and incubating the gonad in sea water. All steps were performed at 4°C. Sperm were separated from tissue by filtering through a 100 μpm nylon filter and pelleting at 1,000 g_av for 5 min. Pelleted sperm were resuspended in axoneme storage buffer and homogenized until about 90% of the heads and tails were detached. Sperm tails were purified by differential centrifugation, involving alternating low-speed spins (1,000 g_av for 5 min to pellet heads) and high-speed spins (10,000 g_av for 12 min to collect tails). The membrane was removed by adding up to 1% Nonidet P-40 to resuspended sperm tails. Axonemes were washed twice with axoneme storage buffer to remove the detergent. They were fractionated into Sarkosyl-insoluble ribbons or Sarkosyl-urea tektin filaments by extraction either with 0.5% Sarkosyl or with 0.5% Sarkosyl/2 M urea in 50 mM Tris, pH 8.0,1 mM EDTA, 1 mM dithiothreitol (DTT), twice for 30 min.

Pig kidney epithelial (LLC-PK₁) cells were cultured in DMEM, 10% FCS at 37°C and 10% CO₂. For immunofluorescence microscopy cells were cultured on poly-L-lysine-coated coverslips. Cells were fixed with −20°C methanol or with 3.7% formaldehyde in PBS for 10 min. Alternatively, they were first extracted with 0.1% Triton X-100 in Pipes buffer (100 mM Pipes, pH 6.9, 2 mM EGTA, 1 mM MgCl₂) for 30 s, incubated in Pipes buffer containing 0.1 to 3 mM free Ca²⁺ for 10 min and then fixed with 2% glutaraldehyde, 3.7% formaldehyde or −20°C methanol.

Cells cultured on polylysine-coated coverslips were rinsed with PBS, extracted with Pipes buffer containing 0.5% Triton X-100 and 5 μg/ml Taxol for 30 s at room temperature, and fixed in Pipes buffer containing 1% glutaraldehyde for 10 min. After three washes with Pipes buffer cells were treated with 0.5 mg/ml NaBH₄ in Pipes buffer containing 140 mM NaCl on ice 3 times for 10 min each to reduce the negative effect of glutaraldehyde on antibody binding. Immuno-gold labeling was performed as described elsewhere (Steffen and Linck, 1989a). Briefly, fixed cells were first incubated in 1% BSA/PBS for 20–30 min to block unspecific binding sites, followed by an incubation with primary antibody diluted in PBS containing 1% BSA, 0.05% goat serum and 0.02% NaN₃ for 1 h or overnight, depending on the antibody dilution used. Goat serum was employed to reduce possible unspecific labeling of the goat secondary antibody. 5 nm gold-conjugated goat anti-rabbit IgG or biotin-conjugated goat anti-rabbit IgG and 5 nm gold-conjugated streptavidin were used as detection systems. The gold complex was diluted in PBS containing 1% BSA and 0.5% fish gelatine (Birrell et al., 1987). Silver enhancement (Janssen, Piscataway, NJ) was employed to locate flat-embedded cells for sectioning. Cells were postfixed with 1% glutaraldehyde in Pipes buffer for 10 min and embedded in Vestopal (Polaron, San Jose, CA). Ultra-thin sections were stained with 0.4% uranyl acetate for 30 min and 0.27% lead citrate for 8 min and examined with a Jeol JEM 100CX at 80 kV.

Protein samples were separated by sodium dodecyl sulfate/polyacrylamide gel electrophoresis (SDS-PAGE) according to Laemmli (1970). Two types of SDS were used: SDS from BioRad (Electrophoresis Purity Grade) and SDS from Sigma (product no. L5750). Proteins were either stained with Serva Blue or transferred onto nitrocellulose (Bio Rad, Richmond, CA) under‘renaturing’ transblot conditions as described by Dunn (1986). The‘renaturing’ transblot was carried out as follows: gels were incubated in 20% glycerol, 50 mM Tris-HC1, pH 7.4, for 60 min and proteins were electrotransferred in 3 mM Na₂CO₃, 10 mM NaHCO₃, 20% methanol, pH 9.9, at 60 V for 2 h onto nitrocellulose sheets. Replicas of gels were first stained with Ponceau S and then immuno-stained as described elsewhere (Steffen and Linck, 1989b).

Since it had been reported that the integrity of sea urchin spindles is maintained following exposure to 1 mM Ca²⁺ (Hays and Salmon, 1983), it was necessary for our studies to determine the changes in the birefingence of Spisula spindles, following treatments that cause microtubule depolymerization. Fig. 1 shows the birefringence of an isolated spindle of Spisula solidissima at different time points while exposed to 0.5 mM free Ca²⁺ at room temperature. Within 10 min the birefringence was significantly reduced. A low level of birefingence was maintained even after extended exposure to calcium. An increase to 1 mM free Ca²⁺ did not seem to have an additional effect (not shown). Calcium has also been used to depolymerize cytoplasmic microtubules of Triton X-100-extracted mammalian cells (Schliwa et al. 1981). In our studies, when Triton X-100-extracted LLC-PK₁ cells were treated with 1 mM Ca²⁺, not all microtubules were depolymerized. While most cytoplasmic microtubules were depolymerized, a large fraction of the spindle microtubules was still present after exposure to 1 mM free Ca²⁺ in Pipes buffer for 10 min (Fig. 2). By immunofluorescence microscopy, all remaining microtubules were shown to contain acetylated a-tubulin (Fig. 2). An increase to 3 mM free Ca²⁺ depolymerized all cytoplasmic microtubules, but left centriole microtubules (Fig. 2b’) and many spindle microtubules (Fig. 2c’).

Next, SDS-PAGE was used to quantitate the loss of tubulin from spindles and to identify possible nontubulin spindle components. Except when stated, fractionation experiments on isolated Spisula spindles were carried out on ice. Twice-washed spindles were divided into aliquots and exposed to various microtubule-depolymerizing and tubulin-extracting conditions such as 0.5 to 2 mM free Ca²⁺ at room temperature or on ice, increasing salt concentration, increasing Sarkosyl concentration and increasing urea concentration. Each extraction was carried out for 30 min. Fig. 3 demonstrates by SDS-PAGE the selective removal of tubulin from the spindle by 0.5 mM free Ca²⁺ at room temperature (lanes 3 and 3’), by an incubation on ice (lanes 4 and 4’), and by 0.5 mM free Ca²⁺ at 4°C (lanes 5 and 5’); most other components were unaffected. The reduction of tubulin allowed the identification of a ˜52 kDa polypeptide band in SDS-PAGE using BioRad SDS (lanes 2 to 5, indicated by a dot).

The two SDS-PAGE gels in Fig. 3 demonstrate the difference in protein separation in the presence of BioRad SDS versus Sigma SDS. The use of Sigma SDS resulted in a wider separation of α- and β-tubulin, allowing the identification of two polypeptide bands between α- and βtubulin after most of the tubulin was removed (lanes 2’ to 5’)-In BioRad SDS-PAGE gels, however, these two polypeptide bands appeared to comigrate with or just in front of β-tubulin. Impurities in SDS are known to effect the mobility of α- and β tubulin (Stephens, 1981).

To determine the depletion of α and βtubulin, the Serva Blue-stained gel lanes shown in Fig. 3 (lanes 2’ to 5’) were scanned using the JAVA system from Jandel (Corte Madera, CA). Fig. 4a shows the average intensity traces in the tubulin region of lanes 2’ to 5’, and Fig. 4b shows the integrated peak areas for α- and β-tubulin. An exposure of the spindles to 0.5 mM free Ca²⁺ for 30 min at room temperature reduced the amount of α and β-tubulin by about 50%. A much greater reduction was observed when the spindles were exposed to 0.5 mM free Ca²⁺ at 4°C. These conditions resulted in a loss of 92 to 96% of the tubulin. When isolated spindles were incubated with increasing concentrations of Sarkosyl or urea, no further specific reduction in tubulin was observed (not shown).

An exposure to high salt concentrations resulted in the extraction of several polypeptide bands, most noticeably a ˜100 kDa polypeptide band (Fig. 3, lanes 6 and 6’, and Fig. 5, lanes 3 and 4). By comparing pellet and supernatant fractions, a 55 kDa polypeptide band could be identified between α- and β-tubulin of the supernatant fraction. This polypeptide band was not detectable in whole spindles (Fig. 5, lane 1) or Ca²⁺/cold-resistant spindle remnants (Fig. 5, lane 2), probably because it was masked by the relative abundance of tubulin. Several of these polypeptide bands will be discussed in more detail below.

Polyclonal anti-tektins were raised against SDS-PAGE-purified tektins A, B and C from two different sea urchin species (Strongylocentrotus purpuratus and Lytechinus pictus). On denaturing SDS-PAGE immunoblots (Towbin et al. 1979) each anti-tektin revealed a high degree of mono-specificity for its respective tektin within a species and a cross-reactivity with the corresponding tektin in the other species (Linck et al. 1987; Steffen and Linck, 1988). However, when tektins were transferred to nitrocellulose under‘renaturing’ conditions (Dunn, 1986), the anti-tektins also showed a cross-reaction with non-corresponding tektins, indicating a degree of similarity in their structure. Thus, to maximize the possibility of detecting tektins or tektin-related components in systems other than sea urchin, we used the‘renaturing’ transblot conditions for immunoblots from BioRad SDS-PAGE.

As demonstrated above, the meiotic apparatus of 5. solidissima still maintained a spindle shape after 95% of the tubulin had been removed. By SDS-PAGE immunoblot, three different polyclonal anti-tektins were shown to cross-react strongly with the tubulin-depleted spindles remnant: anti-(L.p.)-tektin A and anti-(L.p.)-tektin C raised against tektins from L. pictus and anti-(S.p.)-tektin C raised against tektin from S. purpuratus. Anti-(L.p.)-tektin C recognized three polypeptide bands of about 64 kDa, 100 kDa and >200 kDa (Fig. 6, lane 1). Anti-(L.p.)-tektin A recognized a component of about 48 kDa (Fig. 6, lane 2). Anti-(S.p.)-tektin C recognized a polypeptide band of about 52 kDa (Fig. 6, lane 3) and faintly cross-reacted with a band of about 97 kDa; however, anti-(S.p.)-tektin C did not recognize the 52 kDa polypeptide band in whole spindles, most likely because it was masked by tubulin on the nitrocellulose paper (see Fig. 7d). By comparing whole spindles and tubulin-depleted spindles, the removal of spindle tubulin did not have an effect on the presence of the tektin-related antigens. However, when tubulin-depleted spindles were treated with 1 M KC1 (Fig. 6, pellet and supernatant), 2 M NaCl or 2 M urea (not shown), the components recognized by anti-(L.p.)-tektin C and anti-(L.p.)-tektin A were solubilized (Fig.6, lanes 1” and 2”). The 52 kDa component detected by anti-(S.p.)-tektin C remained with the salt-insoluble fraction (Fig. 6, lane 3’). As mentioned earlier, the anti-tektins recognized components in replicas derived from BioRad SDS-PAGE; in contrast, the polypeptide bands could not be recognized in replicas obtained from Sigma SDS-PAGE.

It was important to understand how the tektin-like spindle components compared with bona fide tektins from Spisula sperm axonemal microtubules. For this purpose Spisula sperm flagellar axonemes were treated to yield a 0.5% Sarkosyl-insoluble fraction and a 0.5% Sarkosyl, 2 M urea-insoluble fraction. The polypeptide composition of 0.5% Sarkosyl-insoluble fraction of Spisula axonemes is very similar to that of the Sarkosyl-insoluble protofilament ribbons from sea urchin sperm (compare with Fig. 4 of Linck et al. 1987), consisting of a- and α-tubulin, a pair of polypeptide bands of about 80 kDa and several polypeptide bands in the tubulin region. The polypeptide composition of the Sarkosyl/urea-insoluble fraction of Spisula axonemes is reduced to only two polypeptide bands of about 50-52 kDa of unequal amounts, compared to sea urchin axonemal tektin filament preparations, which consist of equimolar amounts of tektins A, B and C.

The Sarkosyl-insoluble fraction, the Sarkosyl/urea-insoluble fraction and whole axonemes were then analyzed by SDS-PAGE immunoblotting (Fig. 7) using BioRad SDS and antibodies against sea urchin tektins. In the Sarkosyl-insoluble fraction of Spisula a total of five polypeptide bands were recognized. In whole axonemes, however, two of these polypeptide bands could not be detected; these polypeptides co-migrate with α- and β-tubulin and most likely their immunostaining was blocked by the high level of tubulin.

Finally, when Spisula axonemal tektins were compared with the spindle components, a clear distinction could be made (Fig. 7). While anti-(L.p.)-tektin C recognized several polypeptides bands between 50 kDa and 58 kDa in sperm axonemes (Fig. 7b, lanes 1–3), this antibody recognized three polypeptide bands of ˜64, ˜100 and >200 kDa in whole and calcium-extracted spindles (lane 4). Anti-(L.p.)-tektin A also recognized several polypeptides bands between 50 kDa and 58 kDa (Fig. 7c, lanes 1-3) but only one polypeptide band of about 48 kDa in isolated spindles (lane 4). Only the anti-(S.p.)-tektin C recognized a ˜52 kDa spindle component that co-migrated with tektin-related components of sperm axonemes (Fig. 7d). These results indicate that, with the possible exception of the 52 kDa polypeptide, the tektin-like components of the spindle are different from axonemal tektins.

When artificially activated eggs of the surf clam Spisula solidissima were stained with any of the anti-tektins for immunofluorescence microscopy, only anti-(L.p.)-tektin C revealed a distinct labeling of the spindle (Fig. 8). The immunosignal was obtained in whole eggs as well as in isolated spindles (compare Figs 8 and 9). Immunofluorescence microscopy of whole eggs indicated that the tektin C-related antigen co-localized with spindle microtubules (Fig. 8). The immuno-signal could be observed from early prometaphase until anaphase (not shown). Initially, the signal of the tektin-like spindle staining was rather weak, mainly because of some background staining of the egg cytoplasm. However, a clear image could be obtained, when EGS was used to fix the cells and hypersensitized Kodak Technical Pan film (Sliva, 1981a,b) was employed to increase the film contrast.

When eggs from 3 different sea urchin species (Arbacia punctulata, L. pictus and 5. purpuratus) were examined under similar conditions, none of our polyclonal anti-tektins labeled the spindle; surprisingly, however, in cultured cell lines certain anti-tektins again revealed a staining of the spindle. In an initial characterization of our antibodies we noticed that anti-(S.p.)-tektin A and C recognized a component associated with the spindle in whole and Triton-extracted LLC-PKi cells (Steffen and Linck, 1989c). To compare the spindle component of mammalian cells with those of the surf clam, LLC-PKj cells were extracted with 0.1% Triton under microtubule-stabilizing conditions and then treated with increasing concentrations of calcium. As described above for Spisula spindles, isolated spindles of LLC-PKj cells revealed a considerable resistance to the microtubule-depolymerizing effect of calcium (Fig. 2). The tektin-like antigen was still present in the spindle when mitotic cells were treated with 1 or 2 mM Ca²⁺. Fig. 10 shows immunofluorescence micrographs of mitotic LLC-PKi cells labeled with anti-(S.p.)-tektin C. The tektin-like antigen co-localized with the Ca²⁺-resistant spindle remnant. Furthermore, a strong immunosignal was also observed at the spindle poles. In initial immunoblots of Triton X-100-extracted LLC-PKi cells anti-(S.p.)-tektin C recognized an >100 kDa polypeptide band (not shown).

To preserve the ultrastructure of microtubules glutaraldehyde must be employed rather than formaldehyde. When isolated spindles of S. solidissma were fixed in the presence of 1% or 2% glutaraldehyde for 10 min, anti-(L.p.)-tektin C no longer recognized the spindle component. Postembedment labeling of Lowricryl sections has so far failed to produce a signal. However, it was noted that the tektin-like component in LLC-PK, cells could still be recognized by the antibody when glutaraldehyde was used to fix the cells. By immunofluorescence microscopy the tektin-like antigen was localized within the spindle (Fig. 10). As shown in Fig. 11, the gold labeling with anti-(S.p.)-tektin C co-localized with the chromosome fibers. The co-localiz-ation was characterized by a higher density of colloidal gold particles in the areas of microtubule bundles. The gold particles were not always closely associated with the spindle microtubules. They could also be found in spaces between bundles of microtubules, however, at a lower concentration. Several control experiments were performed to ensure the specificity of the immuno-gold signal. According to Birrell et al. (1987), when gold conjugates were diluted in a buffer containing fish gelatine, there was a lower background signal. Further more, no spindle-specific signal was obtained when the secondary antibody was used alone or when the anti-(S.p.)-tektin C was replaced by anti-(L.p.)-tektin B (not shown). The immuno-chemical data obtained with the affinity-purified anti-tektins are summarized in Table 1.

As reviewed above, several pieces of evidence suggest the existence of a spindle matrix in dividing cells (Hays and Salmon, 1983; Leslie et al. 1987; Nicklas et al. 1982; Rebhun and Palazzo, 1988; Wilson and Forer, 1988). An undefined spindle matrix has also been invoked to explain gaps in our understanding of anaphase A chromosome movement (Koshland et al. 1988). At the outset one could imagine that a spindle matrix could have one or more important functions: e.g. purely structural interactions with spindle microtubules, poles or kinetochores, anchorage of the spindle within the cytoplasm, or even a more active role in chromosome movement. Our present investigation has focussed on several aspects of the spindle matrix: the conditions for its isolation, its polypeptide composition, and a com parison of spindle proteins with tektins from axonemal microtubules.

Since the spindle matrix is operationally defined as the remnant left after microtubule depolymerization, it is important first to consider the stability/lability of spindle microtubules. Schliwa et al. (1981) were able to demonstrate that micromolar concentrations of free calcium are capable of depolymerizing interphase microtubules in Triton X-100-extracted cells. Spindle microtubules, on the other hand, display a higher resistance to calcium (Fig. 2) and cold (Brinkley and Cartwright, 1975). So far it is not understood why certain microtubules are more resistant to calcium-induced depolymerization. It has been suggested that a post-translational modification of tubulin, such as acetylation or detyrosination, might correlate with increased microtubule stability (Greer and Rosenbaum, 1989). In fact, all microtubules of LLC-PK₁ cells expressing a higher calcium stability contain acetylated a-tubulin (Fig. 2). Acetylation, however, cannot be the only factor influencing the stability of spindle microtubules. The labile, cytoplasmic microtubules of echinoderm eggs are known to contain only nonacetylated tubulin (Pipemo and Fuller, 1985). Furthermore, in artificially activated eggs of the surf clam Spisula solidissima no spindle microtubules could be detected with our monoclonal antibody specific for acetylated a-tubulin, even though ˜50% of the tubulin remained after spindles were incubated with 0.5 mM free calcium for up to 30 min (Figs 3 and 4, lane 4). In our experiments we used a glycerol-containing lysis buffer. It is known that glycerol stabilizes microtubules (Na and Timasheff, 1981), and furthermore, that it tightly binds to tubulin (Detrich et al. 1976). Thus, we cannot completely rule out a stabilizing effect due to glycerol, even though spindles were washed twice to remove it. In an independent study, however, we were not able to remove all the tubulin, even when the spindles were isolated using glycerol-free isolation buffer (Steffen and Linck, unpublished data).

Using calcium and/or cold temperature, we demonstrated here that about 95% of the spindle tubulin can be removed. Still, meiotic spindles of the surf clam Spisula solidissima maintain their spindle shape. One might argue that the remaining tubulin could be sufficient to provide a framework for the spindle. Leslie et al. (1987) were able to demonstrate by immuno microscopy that no microtubules were left behind in Ca²⁺/cold-extracted sea urchin spindles. On the basis of these observations, i.e. the removal of >95% tubulin and the lack of microtubules in Ca²⁺-extracted echinoderm or molluscan spindles, it can be assumed that the spindle structure is not only maintained by tubulin.

Following extraction with high salt, a ˜55 kDa component could be isolated from sea urchin spindles as the dominant non-tubulin component (Rebhun and Palazzo, 1988). It was proposed (Leslie et al. 1987; Rebhun and Palazzo, 1988) that the 55 kDa component of sea urchin eggs might provide a filamentous framework for the spindle. From Spisula spindles we failed to isolate a dominant 55 kDa component under similar conditions. Nevertheless, depending on the brand of SDS used, a ˜52 kDa (BioRad SDS) or a ˜55 kDa (Sigma SDS) component was noted in calcium- and KC1-extracted spindles (Fig. 3, lanes 6 and 6’) and found to be recognized by anti-(S.p.)-tektin C (Fig. 6, lanes 3 and 3’).

The structural organization of the 55 kDa sea urchin polypeptide within the cell is not clear. Antibodies raised against this component failed so far to localize specifically with the spindle: instead, staining of the whole egg cytoplasm was obtained (Rebhun and Palazzo, personal communication). As with the results of Rebhun and Palazzo, the anti-(S.p.)-tektin C, which cross-reacts with the 52 kDa component, also failed to label the spindle by immunofluorescence microscopy. An alternative distribution of this component has to be considered (see below). In contrast, spindle staining in artificially activated Spisula eggs could be obtained with anti-(L.p.)-tektin C (Figs 8 and 9).

We demonstrated that fractionation of Ca²⁺/cold-stable spindle remnants of Spisula eggs are of limited use. Therefore, an alternative method to characterize the spindle remnant must be considered. As stated earlier, tektins are filamentous components obtained from microtubules but possibly related to intermediate filaments (IFs). By SDS-PAGE immunoblot several tektin-related components (˜48 kDa, ˜52 kDa, ˜64 kDa, ˜100 kDa, and >200 kDa) were identified as being associated with tubulin-depleted spindle remnants (Fig. 6 and Fig. 7). The anti-tektins that recognized the 48 kDa and 52 kDa polypeptides were not able to label a specific cell structure. However, the anti-(L.p.)-tektin C-like antigens, ˜64 kDa, ˜100 kDa and >200 kDa polypeptides, could be localized within the spindle by immunofluorescence microscopy of whole eggs and isolated spindles (Table 1). The antigens appeared to co-localize with the spindle microtubules (Fig. 8) but stayed with the spindle remnant when 95% of the spindle tubulin was removed (Fig. 6 and Fig. 7). So far, we do not know whether the tektin-like spindle components represent microtubule-associated proteins or whether they are independent structural components. Immunoelectron microscopy of the Spisula components has not been accomplished and immunoelectron microscopy of LLC-PK₁ spindles is not conclusive (Fig. 11). Further studies must be done to elucidate the structural organization of the tektin-like components within the spindle.

Because of the immunological similarities between tektins and intermediate filament (IF) proteins (Chang and Pipemo, 1987; Steffen and Linck, 1989a), the question arises: are the spindle components recognized in our studies tektin-like or intermediate filament-like? Several antibodies raised against intermediate filamentlike components from plant cells have been found to stain the spindle (Parke et al. 1987; Hargreaves et al. 1989), whereas in mammalian cells, IFs are excluded from the mitotic spindle and form a cage-like structure around the spindle (Blose, 1979; Zieve et al. 1980). Thus, from the limited information available there is no conclusive evidence that intermediate filaments are part of the spindle. On the basis of their amino acid sequence, IFs represent a heterogeneous group of filamentous proteins, yet all known classes of IF proteins have two features in common: a highly conserved secondary structure and consensus sequences at the amino and carboxy termini of the a- helical rod domain (review: Steinert and Roop, 1988). As indicated earlier, IF proteins and tektins represent distinct but possibly related protein families, as determined by comparison of their predicted secondary structures (Norrander et al. unpublished): importantly, they share 7 of the 15 residues of the IF consensus sequence in the C terminus (Norrander et al. unpublished). A monoclonal antibody specific for the carboxy-terminal consensus sequence of intermediate filaments (anti-IFA; Pruss et al. 1981) recognizes fibrillar bundles in higher plant cells that co-distribute with interphase and spindle microtubules (Dawson et al. 1985). The anti-IFA also cross-reacts weakly with tektins (Amos et al. 1986). So far we have not been able to determine whether spindles contain intermediate filament-like or tektin-like components. Immunological studies cannot provide definite evidence about the group to which these spindle components belong. We hope that direct sequence data or cDNA sequence data will answer this question.

The fractionation experiments of isolated spindles of clams and sea urchin strongly suggest the presence of a non-tubulin spindle matrix. What might be the importance of such a matrix? The observation that antibodies against the 55 kDa sea urchin component (Rebhun and Palazzo, personal communication) and our 52 kDa/55 kDa specific anti-(S.p.)-tektin C react with the whole cytoplasm could indicate that the‘spindle matrix’ organized by the 55 kDa component is an extension of a cytoplasmic, filamentous network penetrating the whole egg cytoplasm. However, the possibility cannot be excluded that the weak cytoplasmic staining with these two antibodies may be due to an unspecific interaction. Using the anti-IFA (Pruss et al. 1981) St-Pierre and Dufresne (1990) have identified an IF-like, 117 kDa component in sea urchin eggs and embryos; this component is localized throughout the cytoplasm of unfertilized eggs and is associated with microtubules of isolated spindles. Since immunolabeling of the 117 kDa component was carried out on extracted eggs, the possibility cannot be ruled out that the spindle staining with anti-IFA may be the result of extraction with Triton X-100. In Spisula, however, our anti-tektin recognized a 100 kDa/>200 kDa, spindle-associated component in whole eggs (Fig. 8). A filamentous cytoplasmic matrix could easily provide an anchorage for the spindle within the cell. On the other hand, a spindle-specific matrix could provide a selective framework on which the spindle microtubules could operate.

Sillers and Forer (1981) have demonstrated that in crane fly spermatocytes areas of reduced birefringence (ARBs) move polewards. By electron microscopy they could demonstrate that the microtubules in these ARBs were destroyed (Wilson and Forer, 1988). Two explanations can be offered for this behavior: microtubules could depolymerize on the pole side of the ARBs and polymerize simultaneously on the equatorial side of the ARBs. For such a process to take place, one has to assume that the minus and plus ends of the microtubules are in a steady state. A steady state of microtubules, however, does not seem to be very likely during anaphase when chromosomes are moving polewards. If the observation of polward moving ARBs is correct and minus and plus ends of microtubules are not in a steady state, one has to expect a non-microtubule connection between proximal and distal parts of the chromosome fibers spanning at least the ARBs. Forer (1985) postulated the presence of actin filaments within the spindle fibers, but the presence of actin has not been corroborated. Perhaps a tektin-like spindle matrix component could provide a structural network spanning the spindle. If such a matrix is important for spindle structure and function, one would expect to find spindle matrix components in all higher eukaryotes. As demonstrated here, antibodies raised against sea urchin tektins labeled the spindles of an echinoderm (Spisula) and a mammalian cell line (LLC-PKx), indicating the existence of a more universal tektin-like spindle component.

We thank Dr. Ray E. Stephens for providing us with a polarization microscope, Dr. Lionel I. Rebhun for providing us with the monoclonal antibody against the 55 kDa Spisula spindle component, Drs. Sid Tamm and Ray Stephens for hosting us in their laboratories during our stay in Woods Hole, and Dr. Julie L. Hodgkinson for critical reading of the manuscript. This study was largely conducted during a stay at the Marine Biological Laboratories at Woods Hole supported by a National Science Foundation grant DCB-8811015 to R.W.L. and a fellowship from the Marine Biological Laboratory (Woods Hole, MA) to W.S.