The plant Spc98p homologue colocalizes with γ-tubulin at microtubule nucleation sites and is required for microtubule nucleation

The molecular mechanisms responsible for microtubule (MT) nucleation have not yet been identified in higher plants. Unlike other eukaryotic cells, no centrosome-like organelles are present to nucleate MT assembly. In animal cells, the centrosome functions as a microtubule-organizing center (MTOC), and in fungi, the spindle pole body (SPB) plays this role. MT nucleation is initiated by γ-tubulin ring complexes, or γ-TuRCs(Zheng et al., 1995), which are recruited from the cytoplasm to the MTOC and activated(Schiebel, 2000). The smallest complex unit capable of MT nucleation, the γ-tubulin small complex(γ-TuSC), was identified in yeast(Knop and Schiebel, 1997) and in Drosophila (Oegema et al.,1999; Gunawardane et al.,2000). The γ-TuSC, which is thought to be a γ-TuRC precursor, contains γ-tubulin, known as a universal nucleator(Oakley, 1992) and two additional proteins, spindle pole body components Spc98p and Spc97p or their homologues. These proteins are essential for the nucleation activity of the complex. SPC98 was characterized as a dosage-dependent suppressor of a mutant of TUB4 (Geissler et al.,1996), the S. cerevisiae γ-tubulin homologue(Sobel and Snyder, 1995). Spc98p interacts with γ-tubulin, and Spc97p interacts with Spc98p. Spc98p or its homologue can therefore be considered as a marker for MT nucleation complexes (Martin et al.,1998; Tassin et al.,1998).

Compared with other eukaryotes, the situation in higher plants is unique and surprising. Two major features of plant cells should be underlined. First,diverse MT arrays assemble successively, with different orientations during the cell cycle and/or developmental controls. The cortical MTs, the preprophase band and the phragmoplast are not found in other eukaryotic cells(Staiger and Lloyd, 1991;Lambert and Lloyd, 1994;Shibaoka and Nagai, 1994). It is not known whether a single MTOC or whether multiple MTOCs are involved in generating the plant MT cytoskeleton(Canaday et al., 2000;Vantard et al., 2000), and it is not clear how assembly of an acentrosomal spindle could be regulated(Marc, 1997;Vaughn and Harper, 1998).

Secondly, in higher plant cells, γ-tubulin is distributed along all MT arrays (Liu et al., 1993;Joshi and Palevitz, 1996;Endlé et al., 1997;Canaday et al., 2000). This localization is enigmatic as γ-tubulin, which is considered to be a universal nucleator in other eukaryotes(Oakley, 1992), would not be expected to nucleate along MTs. The nuclear surface is the only functionally characterized MT nucleation site in plants(Mizuno, 1993;Stoppin et al., 1994), andγ-tubulin is detected there. Our aim was to identify proteins involved in MT nucleation in higher plant cells and to use them as markers of plant MT nucleation sites.

In the present report, we have identified and characterized SPC98orthologues in rice (Oryza sativa) and Arabidopsis thaliana,indicating that higher plants possess γ-tubulin complex components,although no centrosome-like organelle is present. Unlike γ-tubulin,plant Spc98p is not detected along MTs, suggesting that plant γ-tubulin may have a role that is independent of MT nucleation at these sites. We show that Spc98p and γ-tubulin colocalize at MT nucleation sites on the nuclear surface. In addition, both proteins are also found close to the cell membrane, suggesting that cortical MTs are either nucleated at these sites or stabilized by a minus-end anchorage complex as suggested for pericentriolar MTs in animal cells (Mogensen et al.,2000).

To identify plant homologues of SPC98, database searches were done using BLASTN programs. A rice (O. sativa)-expressed sequence tag(EST) (GenBank accession number c26482) was detected, but sequencing showed that this EST clone was incomplete at the 5′ end. A complete A. thaliana gene, PGC95 (GenBank accession number BAB09802), was identified and amplified by PCR from A. thaliana (ecotype Columbia)genomic DNA. To obtain clone pCK.AtSPC98::GFP, the gene was amplified by PCR using primers 1 (CCATGGAAGACGACGATCAGCAG) and 2 (CCATGGCTCCTTTGGAATGCAATCG),which introduced NcoI sites at each end of the AtSPC98 gene. The 2600 nucleotide fragment amplified was digested by NcoI and inserted into a pUC-based plasmid, pCK.EGFP, which is a modified version of pCK.GFPS65C. In our construct, pCK.AtSPC98::GFP, the PGC95 initiation codon was conserved, and the GFP termination codon was replaced by GCC.

To generate rabbit polyclonal antibodies against plant γ-tubulin, two peptides were selected. The first was a maize peptide EDFATQGGDRKDVFFYQ, (p1),which is conserved in eukaryotes (Joshi et al., 1992), and the second, CESPDYIKWGMEDP, (p2), is a plant-specific sequence from the C-terminus. Antibody specificity was shown using extracts from tobacco BY-2 cells, mammalian 3T3 cells and from Escherichia Coli expressing plant γ-tubulin.

Rabbit polyclonal antibodies were raised against three rice Spc98p peptides. The first peptide, (pA), LETAIRASNAQYDDRDIL, is from the central domain, which is conserved in eukaryotes(Fig. 1, residue 625 to 647). The second peptide, (pB), DLDSIAKDYTSSLDA, is a plant-specific peptide from the C-terminus. The third peptide, (pC), FRLDFTEYYSRVSSNK, is from the C-terminal domain and is less conserved than pA or pB. All antibodies were affinity purified against the corresponding peptide.

For MT labeling, commercial mouse monoclonal antibodies againstα-tubulin (Amersham RPN 356) were used. Alexa-488- and -568-labeled anti-rabbit or anti-mouse IgGs were purchased from Molecular Probes (A-11029 and A-11036).

A suspension of the tobacco BY-2 cell line derived from Nicotiana tabaccum L. cv. Bright Yellow 2 was subcultured as described by Nagata et al. (Nagata et al., 1981). Transient expression of SPC98::GFP was obtained after bombardment of cDNA-coated tungsten particles using a particle inflow gun. 1 μg of cDNA,mixed with 1 mg of particles in the presence of 0.75 M CaCl₂ and 1.5 M spermidin, was used per assay. Cells expressing Spc98p-GFP fusion proteins were collected under a fluorescence-equipped binocular microscope. Cells were centrifuged at 60 g for 3 minutes and mixed (v/v)with sample buffer (250 mM Tris, 20% glycerol, 100 mM DTT and 1.5 M SDS). After 3 minutes of sonication, the sample was boiled for 5 minutes and stored in aliquots at -20°C. Cell extracts were analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) according to Laemmli(Laemmli, 1970) using 0.75 or 1 mm thick mini-slab gels and an acrylamide concentration of 6.5% or 12.5%. After electrotransfer to Immobilon (Millipore) membranes, samples were probed using various antibodies. Non-specific binding sites were saturated with 5%dry milk, 1% acetylated BSA (Aurion) and 0.2% Triton X-100 in 20 mM Tris-HCl pH 7.4 for 1 hour at 37°C. Primary antibodies were diluted in the saturating buffer. Secondary horseradish-peroxidase-conjugated antibodies(Pierce) were detected using a chemiluminescence kit (Roche).

Nuclei were isolated from 3.5-day-old cultured tobacco BY-2 cells(Stoppin et al., 1996). BY-2 cells were grown with 1.5% sucrose for 24 hours. To avoid plastid contamination, cells were grown without sucrose for the last 12 hours before harvesting. Protoplasts were then prepared using 3% cellulase RS (Onozuka),0.2% macerozyme R10 (Serva 28302) and 0.15% pectolyase Y23 (Seishin), 0.45 M mannitol, 8 mM CaCl₂ in 25 mM MES buffer pH 5.5 for 2 hours at 30°C with slow agitation. After three washes with 250 mM sucrose, 0.5 mM EDTA, 1 mM EGTA, 1 mM MgCl₂ in 25 mM MES pH 5.5, protoplasts were suspended in washing medium supplemented with 0.025% NP40, 10 μM leupeptin,10 μM pepstatin, 1 mM PMSF, 1 mM DTT, 10 μg/ml aprotinin for 20 minutes at 4°C and agitated slowly. Protoplasts were then broken by passage through a 10 μm nylon mesh. Purified nuclei were centrifuged for 5 minutes at 100 g at 4°C and conserved in 50% (v/v) glycerol in liquid nitrogen.

MT nucleation assays were done (Stoppin et al., 1994; Stoppin et al.,1996) in the presence of 10^-5 M oryzalin to avoid elongation of plant MT seeds on the nuclei. Oryzalin is a potent inhibitor of the assembly of plant MTs but does not interfere with neurotubulin assembly. The concentration of purified neurotubulin used (7 to 10 μM) was below the critical concentration for autoassembly. Isolated nuclei were observed by DIC and fluorescence microscopy.

Isolated nuclei were incubated in the presence of anti-Spc98pA, pB or anti-γ-tubulin antibodies (diluted 1/50-1/200) for 20 minutes at 4°C before the nucleation assays. Incubation with anti-GFP or anti-Spc98pC antibodies, which do not crossreact with tobacco Spc98p, were used as negative controls. Competition was performed by incubation of anti-Spc98p antibodies together with their corresponding peptides for 1 hour at room temperature before the nucleation assays. Each experiment was done three times. 200 nuclei were counted in each assay and compared with controls done simultaneously using the same set of tubulin and nuclei. The proportion of nuclei unable to nucleate MTs in controls was subtracted from the inhibition measurements. This allows direct comparison between the 100% nucleation in controls and the various nucleation rates shown in inhibition assays.

Arabidopsis or BY-2 cells were fixed for 15 minutes in 3.7%formaldehyde in MT-stabilizing buffer (5 mM EGTA, 2 mM MgCl₂, 50 mM PIPES, pH 6.9), post-fixed for 5 minutes in cold methanol and washed three times with stabilizing buffer. Cell walls were permeabilized for 5 minutes in a 1/10 dilution of the enzyme mixture used for protoplast isolation and incubated overnight with anti-γ-tubulin (1/1000 in PBS: 0.14 M NaCl, 2.7 mM KCl, 1.5 mM KH₂PO₄, 8,1 mM Na₂HPO₄, pH 7.4) or anti-Spc98p antibodies (1/500),followed by secondary antibodies (1/300 in PBS supplemented with 0.1%acetylated bovine serum albumin from Aurion). Controls were performed both using preimmune sera and by preincubation of the antibodies with their respective corresponding peptides.

Immunofluorescence images were obtained using a Sony camera connected to Visolab 200 software adapted to a Leica DMBR microscope equipped with various filters (SP505-550 and LP560), which avoid crosstalk. TIFF images were treated using Adobe Photoshop software.

Living BY-2 cells were attached to polylysine-coated glass coverslips,mounted in a perfusion chamber and perfused for 15 to 30 minutes with culture medium in the presence or absence of 0.2 M mannitol. Cells were monitored during plasmolysis and recovery. Distribution of AtSpc98p-GFP fusion protein in living or fixed cells was analysed using a Zeiss LSM 510 Confocal Laser Scanning Microscope equipped with argon and helium/neon lasers. Pinholes were adjusted to obtain 0.4 μm optical sections. Projections of serial optical sections using Z-series of 0.4 μm intervals were obtained using the LSM software. Optical planes were scanned with the 488 nm ray of the Argon laser and using a 505-550 nm barrier filter to detect GFP fluorescence. For dual imaging of AtSpc98p-GFP and Alexa-568-labeled MTs, the 543 nm ray of the helium/neon laser was used in a multitrack configuration to avoid crosstalk. A dichroic mirror at 545 nm separated GFP (short pass filter, 505-545 nm) and Alexa (long pass filter >560 nm) channels. Time-lapse series of the same confocal plane were taken during 30 minutes with 10 second intervals. LSM images were converted into TIFFs for treatment with Adobe Photoshop software. A Zeiss microscope equipped for conventional fluorescence microscopy was used to capture the corresponding Dapi images using a 3CCD digital camera (Axiocam)associated with Axiovision software.

O. sativa EST c26482 was partially sequenced, and the 630 amino-acid C-terminal sequence was compared using the BLAST program to the NCBI database, which revealed homology with A. thaliana and Spc98p-like proteins. For A. thaliana, a unique mRNA of 2514 nucleotides obtained by splicing of an intron of 116 nucleotides codes for a presumptive 95 kDa protein we called AtSpc98p. Analysis by the CLUSTAL W program showed that plant Spc98p-like proteins (A. thaliana and O. sativa) show homology to γ-tubulin-interacting proteins from other organisms (Homo sapiens, Xenopus laevis, Drosophila melanogaster and Saccharomyces cerevisiae). The 95 kDa AtSpc98p shows 64.5% similarity to human GCP3 and 57% similarity to yeast Spc98. The multiple sequence alignment (Fig. 1) shows that the region from position 319 to 809 is conserved in different species (45% similarity), whereas the N-terminal region is divergent(24% similarity).

Nicotiana tabaccum L. cv. Bright Yellow 2 (BY-2) culture cells were used for transient expression of Spc98p-GFP. Protein extracts were prepared from wild-type and bombarded populations of cells. Immunoblots were probed with antibodies directed against α, β and γ-tubulin,and polypeptides of the expected sizes (50 to 55 kDa) were detected(Fig. 2A, lane 1-3). Monoclonal anti-GFP antibodies labeled the 125 kDa Spc98p-GFP fusion protein(Fig. 2A, lane 4). Free GFP was not detected, indicating the absence of proteolysis in Spc98p-GFP cells. Antibodies were raised against three rice peptides (pA, pB and pC) and used to probe BY-2 cell extracts (Fig. 2B) and extracts from cells expressing Spc98p-GFP(Fig. 2A, lane 5). Anti-Spc98pB labeled the 125 kDa Spc98p-GFP fusion protein and revealed endogenous tobacco Spc98p (Fig. 2A, lane 5;Fig. 2B, lanes 7), as did the anti-Spc98pA antibodies (Fig. 2B, lanes 6). The third one, anti-Spc98pC, did not crossreact with the tobacco Spc98p orthologue and was used as negative control(Fig. 2B, lanes 8). Anti-Spc98pA labels centrosomes in human HEP cells (data not shown).

To investigate the role of plant Spc98p, we used a functional assay for MT nucleation. Isolated tobacco BY-2 nuclei were incubated with purified neurotubulin, which was below the critical concentration for MT autoassembly in the absence of stabilizing agents such as taxol. Oryzalin, a specific inhibitor of plant MT assembly, was used to prevent elongation from MT seeds,which could be present on the nuclear surface after nuclei isolation. Neurotubulin assembly is not affected by oryzalin at the concentration used. In our in vitro assay, MTs are specifically nucleated on the surface of plant nuclei and form a sun-like MT pattern (Fig. 3B,C,E,F). The average nucleation efficiency reached 68%. MT-nucleated nuclei were labeled with the same antibodies used for immunoblots, either polyclonal antibodies raised against plant γ-tubulin or a mixture of anti-Spc98pA and pB. MTs were labeled simultaneously using commercial monoclonal anti-α-tubulin. Optical sections obtained by laser scanning microscopy showed that the plant Spc98p homologue(Fig. 3A) as well asγ-tubulin (Fig. 3D) are distributed over the entire surface of the plant nucleus. Both proteins are found at sites where MTs are nucleated, showing that they remain at the nuclear surface during nuclei isolation and that their MT nucleation activity is conserved (Fig. 3C,F). However, Spc98p and γ-tubulin immunolabeling is decreased by treating nuclei with detergent or salt (data not shown). When isolated nuclei were preincubated with antibodies against γ-tubulin(Fig. 3G) or against Spc98pB(Fig. 3H,I), sun-like nucleated nuclei decreased dramatically, showing 63 to 96% inhibition(Fig. 4). Each inhibition assay was simultaneously performed, without antibodies, using the same batch of purified brain tubulin and isolated tobacco nuclei. 200 nuclei were analysed in each experiment, and the proportion of nucleated nuclei was compared with the corresponding controls in which inactive nuclei were subtracted from active ones. Thus, the control nucleation rate was adjusted to 100%, and nucleation inhibition appeared as a decrease in the nucleation rate.

Increasing antibody concentration proportionally decreased the nucleation capacity. 70 to 93% inhibition was observed using γ-tubulin antibodies,and 80 to 85% inhibition was observed using Spc98pA antibodies. Increasing the time of incubation of antibodies before nucleation assays also enhanced the inhibition of MT nucleation, which varied from 72% after 20 minutes of incubation to 96% after 60 minutes. Controls using either non-crossreactive anti-Spc98pC antibody or an anti-GFP antibody at comparable levels of concentration and time of incubation did not significantly affect the nucleation process.

To determine whether the plant Spc98p is located at MT nucleation sites, we labeled tobacco BY-2 and cultured A. thaliana cells at different stages of the cell cycle with anti-Spc98p, anti-γ-tubulin and anti-α-tubulin antibodies, as shown for BY-2 cells inFig. 5. Both anti-Spc98p(Fig. 5A) and anti-γ-tubulin (Fig. 5C)densely label the nuclear surface that functions as a MT nucleation site in plants. The anti-Spc98p antibody gives some punctate cytoplasmic staining. Neither preimmune sera nor antibody depletion by peptide competition produced such labeling.

γ-tubulin was found at the nuclear surface and along all MT arrays:nuclear-associated MTs, cortical MTs and the preprophase band(Fig. 5C,D). Spc98p is not co-distributed with the γ-tubulin associated with MTs. The same results were obtained in both tobacco and Arabidopsis cells.

The AtSPC98::GFP fusion construct was introduced by bombarding tobacco BY-2 cells or by electroporating protoplasts. In both BY-2 cells and protoplasts, the fusion protein was found mainly on the nuclear surface(Fig. 6C), whereas, when unfused GFP is transiently expressed in BY-2 cells, a diffuse distribution is detected in both the cytoplasm and the nucleus(Fig. 6A). In addition to perinuclear labeling, AtSpc98p-GFP was regularly distributed in the cortical cytoplasm, close to the plasma membrane(Fig. 6C,G-L). These cortical labelings were particularly visible in elongated cells where the cortical MTs are well organized. To determine whether the cortical labelling is linked to the plasma membrane, we followed the movement of fluorescent fusion proteins. Time-lapse images were taken every 10 seconds for 30 minutes on turgescent cells. Then, cells were plasmolyzed for 15 minutes, and recovery was analysed during the following 15 minutes. The GFP fluorescent signals moved slightly during plasmolysis within the focal area and followed the speed of membrane displacement during the 6 minutes necessary for turgescence recovery, as shown by the increasing distance between fluorescent dots marked by an arrow and an arrowhead (Fig. 6G-M). Some of the GFP signals observed using confocal microscopy coincide with ends of MT bundles as observed after immunolabeling(Fig. 6M,N). These data suggest that Spc98p is linked to the plasma membrane and localized at sites that could be involved in cortical MT nucleation and/or anchoring.

The subcellular distribution of Spc98p-GFP confirms the results obtained by immunolabeling, although some fusion protein aggregates were observed when the protein was overexpressed. Using both techniques, a punctate labeling is observed at the nuclear surface and in the cytoplasm. The cortical labeling at the plasma membrane is clearer when the Spc98p-GFP fusion protein is expressed than when immunolabeled, which is mainly because of membrane permeabilization in the latter case.

We show that there are plant homologues of yeast Spc98, a component of theγ-TuSC which is the minimal unit for MT nucleation in yeast and Drosophila (Knop and Schiebel,1997; Oegema et al.,1999; Gunawardane et al.,2000).

The presence of both γ-tubulin and Spc98p suggests that plant cells contain γ-TuSC-like nucleation complexes. In addition, Spc97, anotherγ-TuSC complex component, has a putative homologue in the Arabidopsis thaliana database. Do plant cells possess larger MT nucleating complexes, such as the γ-TuRCs(Murphy et al., 1998;Jeng and Stearns, 1999;Wiese and Zheng, 1999;Moritz and Agard, 2001), found in other eukaryotes? Large soluble γ-tubulin-containing complexes were identified in plants (Stoppin-Mellet et al., 2000), but their activity has not yet been characterized. Further experiments will be necessary to isolate and characterizeγ-tubulin complexes containing plant Spc97p and Spc98p and to determine whether additional plant-specific proteins are present in these nucleation complexes.

The colocalization of Spc98p and γ-tubulin on isolated BY-2 nuclei where MT nucleation is initiated favors the hypothesis that MT nucleation complexes are present at the nuclear surface. Inhibition of MT nucleation by Spc98p and γ-tubulin antibodies argues that Spc98p/γ-tubulin-containing complexes are involved in MT nucleation. Isolated centrosomes, like plant nuclei, are capable of MT nucleation in vitro. Addition of plant cellular extracts to urea-inactivated mammalian centrosomes rescues nucleation activity(Stoppin-Mellet et al., 1999),suggesting that cytosolic γ-tubulin complexes containing Spc98p may be recruited and activated at MT nucleation sites, as in mammalian cells(Moudjou et al., 1996).

AtSpc98p-GFP was detected in living tobacco BY-2 cells on the nuclear surface as expected. In addition, in G1 phase and in elongated cells, a fluorescent signal is detected close to the plasma membrane, at ends of MT bundles. γ-tubulin is also present at sites where cortical MTs contact the cell membrane (McDonald et al.,1993; Canaday et al.,2000), and perinuclear MTs do not directly form cortical arrays(Nagata et al., 1994). The co-distribution of Spc98p and γ-tubulin suggests that cortical MTs are nucleated in the cortex at the plasma membrane. A 49 kDa component of the centrosphere, identified as elongating factor E2F-1α, is also found both at the nuclear envelope and at the plasmalemma(Hasezawa and Nagata, 1993). This result is indicative of multiple MT nucleation sites, but E2F-1αcould be involved in signal transduction rather than nucleation.

Cortical MTs turn over more rapidly than centrosome-nucleated MTs in animal cells, suggesting that cortical MTs are assembled by de novo nucleation(Wasteneys et al., 1993;Yuan et al., 1994;Hush et al., 1994). This cortical MT instability and the localization of AtSpc98p-GFP at sites where MT bundles start, strongly argue for nucleation at the cortex. Spc98p was not clearly immunodetected at cortical sites, but the fixation and permeabilization procedures used could affect the plasma membrane, leading to the loss of small structures such as γ-tubulin-nucleating complexes. In vivo detection using GFP fusion protein enhances the resolution of membrane nucleation sites and provides evidence to support the hypothesis that there are multiple MT nucleation sites in plant cells.

Neither AtSpc98p-GFP nor antibody-labeled endogenous Spc98p are co-distributed with γ-tubulin along MT arrays. This indicates that theγ-tubulin associated along the length of MTs has an alternative activity that may affect MT properties. Cytosol plant γ-tubulin is present in high molecular weight complexes containing Hsp70 and TCP1 chaperones(Stoppin-Mellet et al., 2000),suggesting that the different γ-tubulin complexes present in higher plant cells may have different functions. γ-tubulin may be involved in MT stabilization instead of nucleation or be maintained in a storage form before its recruitment at nucleation sites(Dibbayawan et al., 2001). Perinuclear and perhaps cortical sites where γ-tubulin and Spc98p colocalize would correspond to activatable nucleation sites.

On the basis of our present data, we suggest that plantγ-tubulin/Spc98p-containing complexes are involved in nucleation of plant MTs and are functionally homologous to the γ-TuRCs found in metazoans and fungi. The localization of Spc98p-GFP in vivo in combination with the results of our MT nucleation assays leads us to propose a model for the dynamics of plant MT nucleating complexes during the cell cycle and development (Fig. 7).

Cytoplasmic MT-nucleating complexes containing γ-tubulin and plant Spc98p could be recruited to various MT nucleation sites during the cell cycle or development. Complexes situated at cortical sites could be activated during G0 and G1, that is, at stages where cortical MTs are assembled. The complexes located at perinuclear sites would be activated in G2, when MTs radiating from the nuclear surface are predominant. The pre-prophase band assembled at this stage could originate from perinuclear and/or cortical nucleation sites. In most eukaryotes, the γ-TuRC is targeted to a structured organelle such as the centrosome or the SPB, which nucleates and organizes all MTs. In higher plants, γ-TuRC-like complexes could be recruited to different sites and coordinately activated to organize the successive MT arrays.

We acknowledge funding by the Centre National de la Recherche Scientifique(CNRS), the Ministère de la Recherche for its `Action Concertée Incitative (ACI) en Biologie du Développement et physiologie intégrative' (N° 289 to A.C.S.), the European Community(BIO4-CT98-5036 to M.S.), and the Körber Foundation, Hamburg. The Inter-Institut Confocal Microscopy Platform was co-financed by the CNRS, the Université Louis Pasteur, the Région Alsace and the Association pour la Recherche sur le Cancer (ARC). We are indebted to C. Ritzenthaler for setting up the time-lapse microscopy perfusion system.