The Mitotic Apparatus-associated 51-kDa Protein from Sea Urchin Eggs Is a GTP-binding Protein and Is Immunologically Related to Yeast Polypeptide Elongation Factor la *

We investigated the biochemical characteristics of the Bl-kDa protein that is a major mitotic apparatusassociated basic protein of sea urchin eggs (Toriyama, M., Ohta, K., Endo, S., and Sakai, H. (1988) Cell Motil. Cytoskeleton 9,117-128). The amino acid composition of the 51-kDa protein was apparently different from those of tubulin, actin, histones, and myelin basic protein; yet it was similar to those of polypeptide elongation factors la! (EFla). In addition, antibody to EFlcu from yeast cross-reacted with the Bl-kDa protein. [3H] GTP binding activity was detected in the phosphocellulose-purified fraction (PC fraction) which predominantly contained the 51-kDa protein and was shown to be specific to GTP, GDP, guanylyl imidodiphosphate, and ITP. Photo-affinity labeling using [cx-~~P]& azidoguanosine triphosphate (8-azido-GTP) demonstrated that a 51-kDa polypeptide in the PC fraction specifically bound 8-azido-GTP. This GTP-binding polypeptide was bound to a GTP affinity column, could be eluted by the addition of GTP, and was immunoreactive with anti-51-kDa protein antibodies. When the PC fraction was applied to a gel filtration chromatography column, GTP binding activity was completely coeluted with the 51-kDa protein. Furthermore, the PC fraction and the gel filtration-purified fraction had EF-lcu activity: [‘%]Phe-tRNA transferring activity to ribosomes in the presence of poly(U) and ribosome-dependent GTPase activity. The results indicate that the mitotic apparatus-associated 51-kDa protein is a GTP-binding protein and suggest that it is structurally and functionally related to yeast EF-lc~.

The amino acid composition of the 51-kDa protein was apparently different from those of tubulin, actin, histones, and myelin basic protein; yet it was similar to those of polypeptide elongation factors la! (EF-la). In addition, antibody to EF-lcu from yeast cross-reacted with the Bl-kDa protein.
[3H] GTP binding activity was detected in the phosphocellulose-purified fraction (PC fraction) which predominantly contained the 51-kDa protein and was shown to be specific to GTP, GDP, guanylyl imidodiphosphate, and ITP. Photo-affinity labeling using [cx-~~P]& azidoguanosine triphosphate  demonstrated that a 51-kDa polypeptide in the PC fraction specifically bound 8-azido-GTP. This GTP-binding polypeptide was bound to a GTP affinity column, could be eluted by the addition of GTP, and was immunoreactive with anti-51-kDa protein antibodies. When the PC fraction was applied to a gel filtration chromatography column, GTP binding activity was completely coeluted with the 51-kDa protein.
Furthermore, the PC fraction and the gel filtration-purified fraction had EF-lcu activity: ['%]Phe-tRNA transferring activity to ribosomes in the presence of poly(U) and ribosome-dependent GTPase activity.
The results indicate that the mitotic apparatus-associated 51-kDa protein is a GTP-binding protein and suggest that it is structurally and functionally related to yeast EF-lc~.
The mitotic apparatus is an indispensable cell organelle in the distribution of chromosomes into daughter cells in eukaryotes, and microtubules play essential roles in the formation of the mitotic apparatus. The centrosomes mainly govern the temporal and spatial arrangement of astral and spindle microtubules, and the kinetochores help organization of the spindle by capturing microtubules from the poles. In addition to these structural elements,   pointed out the importance of the guanine nucleotide-dependent behavior of microtubules (dynamic instability) in the formation of the mitotic apparatus.
However, the precise molecular mechanisms of the formation of the mitotic apparatus have not fully been resolved yet.
The 51-kDa protein is a major nontubulin protein component in the mitotic apparatus of sea urchin eggs (5)(6)(7)(8). The 51-kDa protein was first demonstrated to be closely correlated to the nucleation of astral microtubules induced by microtubule-organizing granules (MTOGs)' in vitro (5,6). The aster forming activity of MTOGs could be solubilized from the isolated mitotic apparatus fraction in a solution containing 0.6 M KC1 and 50% glycerol. When the extract was dialyzed against a low ionic strength solution, granular assemblies which could form asters when incubated with tubulin at 37 "C were formed. Phosphocellulose column chromatography enabled us to separate the aster-forming protein fraction which contained the 51-kDa protein as a major component.
Using immunofluorescence staining, the 51-kDa protein was shown to be localized in the mitotic apparatus, especially in the centrosomes, the spindle, and the basal region of the asters (7). Monoclonal antibody to the 51-kDa protein, which did not inhibit the aster forming activity of the MTOGs in uitro, totally blocked the formation of the mitotic apparatus when antibody was injected into the living eggs before prophase. If antibody was injected at prometaphase, an extremely short spindle (birefringent mass) appeared, causing failure of nuclear division as well as cytokinesis (6,8). These results indicated that the 51-kDa protein is an essential regulator in the formation of the mitotic apparatus of sea urchin eggs. In this study, we investigate the biochemical characteristics of the 51-kDa protein and show that this protein is a GTPbinding protein structurally and functionally related to polypeptide elongation factor la. For GTP affinity chromatography, we used the high-speed supernatant fraction just before the phosphocellulose chromatography step and the PC fraction, which were obtained as described before.
These fractions were diluted ten times by PEM3 to reduce the concentration of glycerol to 5%. Reducing the glycerol concentration was necessary for binding of the Bl-kDa protein to GTP-Sepharose like EF-lol (11 Seventeen days after the second injection, the last injection was made intramuscularly, and the rabbit was bled 2 weeks after the last injection. The antiserum was subjected to ammonium sulfate precipitation (50 and 40% successively) and stored at -80 "C. This antiserum was used for the screening of yeast EF-la cDNA by Nagata et al. (14).
Miscellaneous-Protein concentration was determined by the method of Bradford (15). SDS-PAGE was performed according to the method of Laemmli (16). Gels were stained with Coomassie Brilliant Blue R-250. Immunoblotting was performed by the method of Towbin et al. (17), and detection of positive immunoreaction was by horseradish peroxidase-linked second antibodies (Cappel) at l:l,OOO dilution and by 4-chloro-1-naphthol.
Anti-51-kDa protein antibodies were prepared as described elsewhere (7,8). Antiyeast EF-la serum was affinity-purified by Durapore sheets to which the 5l-kDa protein of H. pulcherrimus was transferred according to the method described before (7) GTP-binding Protein in Mitotic Apparatus absorption experiment, the PC fraction was electrophoresed, and the 51-kDa protein band was cut off and homogenized in 20 mM phosphate/KOH (pH 7.4), 150 mM NaCl containing 3% (w/v) bovine serum albumin (fraction V; Seikagaku-kogyo), and 1 mg/ml goat immunoglobulin (Sigma). Anti-EF-lol antiserum was diluted (1:lOO) in this gel homogenate and incubated overnight at 0 'C. The mixture was centrifuged at 30,000 x g for 20 min, and the supernatant was used as the absorbed antiserum. Aster forming activity was measured as described elsewhere (18), and isolation of the mitotic apparatus from H. pulcherrimus was performed as described (6,18).

RESULTS
Amino Acid Composition of 51 -kDa Protein-To investigate chemical properties of the 51-kDa protein, we purified the 51-kDa protein as described under "Materials and Methods." Most of the 51-kDa protein did not bind to the hydroxylapatite column (Fig. lD, lanes FT), in contrast with other proteins which could be eluted by the addition of 0.6 M KH,PO,. Since the 51-kDa protein in the flow-through fractions was close to homogeneity, we used these fractions for amino acid analysis. We could obtain 0.88 mg of the 51-kDa protein from 50 ml of packed sea urchin egg cells.
The amino acid composition of the 51-kDa protein from P. depressus is shown in Fig. 1 together with those of chicken @tubulin (19), rabbit skeletal actin (29), myelin basic protein (21), histone Hl (22), and human (23) and yeast (14,24,25) EFs-lcu. The amino acid composition of the 51-kDa protein was plotted in the order of the molar content of each amino acid residue. Corresponding molar contents of the other proteins were also plotted on Fig. 1. EF-lcu is the a-subunit of elongation factor 1 which is involved in the binding of aminoacyl-tRNAs to 80 S ribosomes during the elongation of the polypeptide chain (26-28). Unfortunately, the amino acid composition of EF-la from sea urchin is not known because EF-lol from sea urchin eggs was not identified yet. However, the amino acid sequences of EFs-lcr which were shown to be highly conserved (80-90%) (14,(23)(24)(25)29)  The 51-kDa protein from P. depressus was purified by hydroxylapatite column chromatograph, and the amino acid composition of the purified 51-kDa protein (0) was analyzed as described under "Materials and Methods." Relative amounts of amino acids (mole percent) were plotted in the order of the molar content of each amino acid residue. A, comparison of amino acid compositions of the 51.kDa protein, P-tubulin, and actin; B, comparison of the 51-kDa protein, myelin basic protein, and histone Hl; C, comparison of the 51-kDa protein and EFs-la from human and yeast. Amino acid compositions of P-tubulin (A) and of EPs-la from human (x) and yeast (0) are from the sequences deduced from cloned cDNA (tubulin (19), human EF-lol (23), and yeast EF-la (14,24,25) had a higher content of lysine (10.9 mol %) and valine (9.4 mol %). This feature was distinct from other cytoskeletal proteins (tubulin and actin) and some basic proteins (histones and myelin basic protein), but was similar to EFs-lcu. The greater number of lysine residues implies the basic property of both proteins. Actually, the isoelectric points of the 51-kDa protein (9.8) and EFs-lot (8.5-9.5) are considerably high. Furthermore, EFs-la have molecular weights (49,000-53,000) similar to that of the 51-kDa protein. Despite the similarity of the contents of most of the amino acid residues between the 51-kDa protein and EFs-la, some differences were observed, i.e. the 51-kDa protein has a greater number of serine residues and a lower number of threonine and aspartic acid + asparagine residues than do EFs-lcu.
Anti-EF-la Antibody Cross-Reacts with 51.kDa Protein-Antiserum to yeast (S. carlsbergensis) EF-la cross-reacted with the 51-kDa polypeptide in the whole egg homogenate (Fig. 24, lane l'), the isolated mitotic apparatus fraction (lane 2'), the PC fraction (lane 3'), and the gel filtration-purified fraction (lane 4') from H. pulcherrimus eggs. The antigenic Cross-reactivity of anti-EF-lcr antiserum with 51-kDa protein and localization of antigen to anti-EF-la antibody in sea urchin eggs. A, the whole egg homogenate (lanes 1, I', and I"), the isolated mitotic apparatus fraction (lanes 2, 2', and 2"), the PC fraction (lanes 3, 3', and 3"), and the gel filtration-purified fraction (lanes 4, 4', and 4") were prepared from H. pulcherrimus as described under "Materials and Methods." These fractions were analyzed by SDS-PAGE (lanes 1-4) and by immunoblotting with anti-EF-lo( antiserum (lanes I'-4') and with monoclonal anti-51-kDa protein antibody HP1 (lanes 1"-4"). B, antibody immunoreactive with the 51-kDa protein was affinity-purified as described under "Materials and Methods." Sections of paraffin-embedded H. pulcherrimus eggs were prepared by the method described before (7). Immunofluorescence labeling using affinity-purified antibody (aEFap51) was performed as described (7). Note the accumulation of the antigen in the mitotic apparatus. Bar, 10 pm. polypeptide coincided with the 51-kDa protein detected by monoclonal anti-51-kDa protein antibody HP1 (lanes 1"-4"). When anti-EF-lcu antiserum was absorbed by the 51-kDa protein, immunoreaction was greatly reduced. Anti-EF-ln antibody affinity-purified by the 51-kDa protein from If. pulcherrimus (aEFap51) cross-reacted only with the protein band corresponding to the 51-kDa protein in the whole egg homogenate (Fig. 2C, lane 5'). In addition, aEFap51 and anti-51-kDa antibody HP1 were also shown to react with EF-la from S. curlsbergensis. Immunofluorescence labeling of the sections of paraffin-embedded H. pulcherrimus eggs by aEFap51 revealed the same localization pattern of the 51-kDa protein in the mitotic apparatus (Fig. 2B) as that detected by monoclonal and polyclonal antibodies against the 51-kDa protein (7). Anti-EF-la antiserum presented a similar staining. When we used antiserum absorbed by the 51-kDa protein, labeling of the mitotic apparatus was largely reduced. No significant labeling was detected in the control experiments using nonimmunized y-globulin or second antibody only. These results indicated that the 51-kDa protein has at least one common antigenic determinant with EF-lcu of S. car&bergensis EF-la. Furthermore, limited chymotryptic digestion of the 51-kDa protein left a stable 43-kDa polypeptide whose molecular mass was similar to those of EFs-la previously reported (12,(30)(31)(32)(33). From these, we concluded that the 51-kDa protein shares some structural resemblance with EF-la. 51.kDu Protein Binds GTP-During binding of aminoacyl-tRNAs to 80 S ribosomes, GTP is hydrolyzed into GDP by EF-lor (28). To perform this function, EF-lcu has the guanine nucleotide-binding site and the GTPase domain whose amino acid sequences are homologous to those of elongation factors Tu and G, initiation factor 2, G-proteins, and RAS proteins (23). In addition, EF-la interacts with EF-10 to exchange bound GDP for GTP (34). Thus, EF-la shares a common property with G-proteins. From the structural resemblance described above, it was plausible that the 51-kDa protein binds GTP as does EF-la. Therefore, we tested the possibility of GTP binding to the 51. kDa protein.
We had succeeded in purifying the 51-kDa protein by hydroxylapatite column chromatography as described before (Fig. lD), so we examined whether or not the 51-kDa protein in this fraction could bind ['HIGTP.
When we digested the hydroxylapatite-purified 51-kDa protein with chymotrypsin, it could no longer generate the stable 43-kDa polypeptide described before, although the 51-kDa protein in this fraction was immunoreactive with all anti-51-kDa protein antibodies. Therefore, we considered that the hydroxylapatite-purified 51.kDa protein was denatured by a irreversible conformational change. Then, we studied the binding of GTP to the 51-kDa protein fraction fractionated by phosphocellulose column chromatography. We found that proteins in 30 ~1 of the PC fraction (containing 17 pmol of the 51-kDa protein) bound 13 pmol of ['lH]GTP, showing a ratio of 1:0.76. This suggested that 1 mol of the 51-kDa protein would intrinsically bind 1 mol of GTP. When the 51-kDa protein bound to the phosphocellulose column was eluted by a linear concentration gradient of KCl, the elution of the GTP binding activity was superimposed with that of the 51-kDa protein.
["HIGTP binding was greatly reduced by preincubation of the PC fraction with 2 mM cold GTP, GDP, GMP-PNP, and ITP, but not with 2 mM ATP, UTP, and CTP (Fig. 3). GMP-PNP and ITP are structural analogues of GTP; therefore, binding is reasonable. This binding specificity was similar to  In addition, we observed only a small reduction of GTP binding to the PC fraction after preincubation of ATP, as shown before (Fig. 3). These results mean that the 42-and 41-kDa polypeptides contributed little to ["HIGTP binding to the PC fraction, and so the efficiency of labeling to the 51-kDa polypeptide by [y-32P]8-azido-GTP may be much less than that of the 42-and 41-kDa polypeptides. In fact, we needed much longer exposure (1 week) to detect incorporation by [-y-32P]8-azido-GTP than that by [(Y-3'P]8-azido-GTP (6 h). Possibly, this low efficiency of labeling to the 51-kDa polypeptide by [y-32P]8-azido-GTP seen in Fig.  4 (lanes 1'-4') was due to GTP hydrolysis by the 5l-kDa protein described below. From these results, we concluded that the 51-kDa polypeptide specifically binds [32P]8-azido-GTP.
To examine whether or not the 51-kDa polypeptide identified by 8-azido-GTP is the mitotic apparatus-associated 51-kDa protein, we carried out GTP affinity column chromatography, followed by immunoblotting using anti-51-kDa protein antibodies. The fraction from the purification step just before phosphocellulose column chromatography was diluted 10 times to reduce the glycerol concentration to 5%. If dilution was eliminated, no protein was adsorbed by GTP-Sepharose because of the change in hydrophobicity in the presence of 50% glycerol. After extensive washing, 2 mM GTP was added to the column for specific elution of GTP-binding proteins. demonstrated that the 51-kDa polypeptide was the 51-kDa protein. We examined the aster forming activity of this GTP eluate according to the method reported previously (5,18). Unfortunately, the aster forming activity was lost after dilution of glycerol. When the concentration of glycerol is decreased below 50%, the aster forming activity becomes quite labile (5,18). The 51-kDa protein in the fraction from the purification step just before phosphocellulose chromatography could not fully combine with GTP-Sepharose (lane FT).
The yield of the 51-kDa protein was 40% of the initial amount of the 51-kDa protein from densitometric analyses. This may be due to denaturation by low concentrations of glycerol. The 45-kDa polypeptide in the same fraction was shown not to bind [*'P]8-azido-GTP (data not shown), and so the 51-kDa protein did not bind to GTP-Sepharose via the 45-kDa polypeptide. The 45-kDa polypeptide was not a kind of degradation product of the 51-kDa protein from immunoblotting.
The 45-kDa protein may be a protein which binds to the 5l-kDa protein.
We further tested whether or not the 51-kDa protein in the PC fraction could bind to GTP-Sepharose (Fig. 5B). The phosphocellulose-purified 5l-kDa protein seemed to bind to GTP-Sepharose more tightly because a higher concentration of GTP (6 mM) or KC1 (1 M) was necessary for elution of the bound 51-kDa protein. The yield of the 51-kDa protein was 30%. In this case, no apparent polypeptide could be seen other than the 51-kDa protein eluted by GTP or KC1 (lanes 7 and 1 I ), ruling out the possibility that the 5l-kDa protein bound via another protein that binds GTP.
In addition to these experiments, we fractionated the PC fraction by Sephacryl S-300 gel filtration column chromatography and studied whether or not GTP binding activity coeluted with the 51-kDa protein (Fig. 6). Elutions of the 51-kDa protein and GTP binding activity coincided perfectly. The peak fractions, 42-44, predominantly contained the 51-kDa protein (70% at the maximum level from densitometric analysis). The ratio of the 51-kDa protein to bound GTP was -1:0.7-0.8, again suggesting that the 51-kDa protein intrin- sically has one binding site for GTP. The stoichiometry was in good agreement with the result of phosphocellulose chromatography.
The yield of the 51-kDa protein was -8O%, showing good agreement with the yield of GTP binding activity (74%). Under this condition, the 51-kDa protein was eluted between bovine serum albumin and ovalbumin.
From the data so far accumulated, we concluded that the 51-kDa protein is a GTP-binding protein.

51-kDa Protein Fractions
Have EF-la-like Activity-The GTP binding ability of the 51-kDa protein suggested that the 51-kDa protein is functionally related to EF-la. We examined EF-la activity of the 51-kDa protein: [14C]Phe-tRNA transferring activity to ribosomes and ribosome-dependent GTPase activity (Fig. 7). Since authentic EF-lcu from sea urchin eggs has not been identified or purified, we used purified yeast la as a reference for assays. Marked [14C]Phe-tRNA binding activities were found in fractions containing the 51-kDa protein: the PC fraction (Fig. 7A, bar b) and the gel filtrationpurified fraction (bar f). We adjusted the amount of the 51-kDa protein and yeast EF-101 to -3 pmol in the reaction mixtures. The activities of these fractions were 60% (the PC fraction) and 53% (the gel filtration-purified fraction) of the activity of yeast EF-la, as normalized to the amount of the 51-kDa protein in both fractions. Other fractions from gel filtration (bars c-e, and g) presented no activity. In addition, the PC fraction revealed ribosome-dependent GTPase activity (Fig. 7B, bar h), as did yeast EF-lcu (bar d). We estimated that the PC fraction had 50% (0.26 mol/mol/ min) of the yeast EF-lcu GTPase activity in the presence of ribosomes, Phe-tRNA, and poly(U). If ribosomes and/or Phe-tRNA was absent, GTPase activities of both yeast EF-la (bars a-c) and the PC fraction (bars e-g) were reduced. In contrast with yeast EF-la, it seemed that the PC fraction had slight GTPase activity independent of ribosomes and Phe-tRNA (bars e-g). These results suggest that the 51-kDa protein is functionally related to EF-la.

DISCUSSION
The results of this paper show that the mitotic apparatusassociated 51-kDa protein is a GTP-binding protein which is structurally and functionally related to EF-lo(. The PC fraction had [3H]GTP binding activity suited for the amount of We used 3 pmol of purified yeast EF-lo and the Bl-kDa protein in the PC fraction or gel filtration-purified fraction for the assays as well as the corresponding volume of the gel filtration-purified fractions which do not contain the 51-kDa protein. B, GTPase activity of yeast EF-la (bars a-d) and the PC fraction (bars e-h). We used 20 pmol of purified yeast EF-la and 10 pmol of the phosphocellulose-purified Bl-kDa protein for the assays.
The assay was performed at 30 "C for 30 min as the 51-kDa protein in this fraction. Binding was specific for guanine nucleotides, corresponding to the specificity of EFla. Results of photoaffinity labeling experiments, GTP affinity chromatography, and gel filtration show that the 51-kDa protein specifically binds GTP with a stoichiometry of 1:l. The amino acid composition, the chymotrytic digestion pattern of the 51-kDa protein, and antigenicity to anti-EFla antibody showed structural resemblance of the Bl-kDa protein to EF-la. Interestingly, the fractions containing the Bl-kDa protein as a main component could transfer charged aminoacyl-tRNA to ribosomes and hydrolyze GTP in a ribosome-dependent manner. Since these fractions did not have any potent GTP-binding proteins other than the 51-kDa protein, it is most likely that the 51-kDa protein can act as EF-la. Together with the GTP binding ability of the 51-kDa protein, it was suggested that the 51-kDa protein is also functionally related to EF-lol. It seems possible that the 51-kDa protein is EF-la from sea urchin eggs. However, some discrepancy arises if the 51-kDa protein is EF-la itself. For instance, the 51-kDa protein was purified from an insoluble fraction of sea urchin eggs and was shown to exist mainly in insoluble components as detected by subcellular fractionation. 3 Preliminary results indicated that the relative amount tions. In dynamic instability theory, microtubules undergo phase transitions between the growing phase in which GTP subunits associate to microtubule ends and the shrinking phase in which unstable polymers consisting of GDP-tubulin depolymerize at a fast rate following the loss of stabilizing terminal GTP subunits (2)(3)(4) Besides the functional participation of the 51-kDa protein in microtubule