QUICK SEARCH:   [advanced] Author: Keyword(s):
Home     Help     Feedback     Subscriptions     Archive     Search     Table of Contents

First published online 17 July 2006
doi: 10.1242/jcs.03039

This Article Summary Figures Only Full Text (PDF) Supplementary Material All Versions of this Article: jcs.03039v1 119/15/3238    most recent Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Related articles in JCS Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Venkei, Z. Articles by Szabad, J. Search for Related Content PubMed PubMed Citation Articles by Venkei, Z. Articles by Szabad, J.

4-Tubulin is involved in rapid formation of long microtubules to push apart the daughter centrosomes during earlyx Drosophila embryogenesis

¹ Maternal Effect and Embryogenesis Research Group of the Hungarian Academy of Sciences at the University of Szeged, Faculty of Medicine, Department of Biology, Somogyi B. u. 4, H-6720 Szeged, Hungary
² University of Szeged, Faculty of Medicine, Department of Medical Chemistry, Szeged, Hungary

^* Author for correspondence (e-mail: szabad{at}mdbio.szote.u-szeged.hu)

Accepted 8 May 2006

	Summary

Top Summary Introduction Results Discussion Materials and Methods References

 
Although 4-tubulin comprises only about one-fifth of the -tubulin pool in every Drosophila egg, in the absence of 4-tubulin - in eggs of the kavar⁰/- hemizygous females - only a tassel of short microtubules forms with two barely separated daughter centrosomes. We report that 4-tubulin is enriched in the long microtubules that embrace the nuclear envelope and suggest that they push apart daughter centrosomes along the nuclear perimeter during the initial cleavage divisions. In vitro tubulin polymerization showed that 4-tubulin is required for rapid tubulin polymerization. Since tubulin polymerization is slow inside eggs of the kavar⁰/- females, only short microtubules can form within the 4 to 5 minutes allowed for the process. A tassel of short microtubules with two barely separated centrosomes forms in every egg of the Kavar^18c/+ females, in which the cytoplasm contains both wild-type and Kavar^18c-encoded 4-tubulin with an E82K amino acid substitution (E82K-4-tubulin). E82K-4-tubulin is incorporated into the microtubules and renders them unstable. When injected into wild-type early cleavage embryos E82K-4-tubulin slows down the formation of long microtubules and the separation of the daughter centrosomes. Surprisingly, when injected into late cleavage embryos E82K-4-tubulin is non-toxic. Similarly, in the neuroblasts, ectopically expressed E82K-4-tubulin becomes incorporated into the microtubules that grow sufficiently long and function normally.

Key words: -tubulin, Interpolar microtubules, Centrosome separation, Dominant-negative mutation, Drosophila embryogenesis

	Introduction

Top Summary Introduction Results Discussion Materials and Methods References

 
Embryos in the animal kingdom rely almost exclusively on maternally provided molecules at the very beginning of their life (DeRenzo and Seydoux, 2004). Constituents of maternal effect are synthesized by the maternal genes and are deposited into the cytoplasm of the egg cell during oogenesis for future use in embryogenesis (Tadros and Lipshitz, 2005). To identify components of the egg cytoplasm required for the commencement of embryogenesis, we made use of the `genetic-dissection' technique, in which we induced and isolated dominant female-sterile (Fs) mutations of Drosophila melanogaster that allowed the formation of normal-looking and fertilized eggs; however, embryogenesis came to an arrest shortly after fertilization (Erdélyi and Szabad, 1989; Szabad et al., 1989). It was anticipated that, if an Fs mutation prevents the commencement of embryogenesis, the product of a wild-type (+) gene that has been identified as Fs is involved in the process and molecular analysis of its function, and would therefore shed light on the beginning of a new life.

The Kavar^18c and the Kavar^21g Fs mutations were used to identify the Tub67C gene (full name -Tubulin at 67C) of Drosophila melanogaster (Venkei and Szabad, 2005), which encodes 4-tubulin, the so-called maternal isoform of the four -tubulin isoforms (Kalfayan and Wensink, 1982; Matthews et al., 1989; Theurkauf, 1992; Matthews et al., 1993; Máthé et al., 1998; Matthies et al., 1999) (for a comprehensive list of the 4-tubulin-related references see the FlyBase at http://flybase.bio.indiana.edu/). Although eggs of the Kavar^18c/+ females appear normal and are fertilized as in wild type, embryogenesis soon comes to a standstill and a monopolar spindle appears in each of the `die-at-start' embryos, with two nearby located centrosomes embedded in a tassel of 3-5 µm long, short and straight microtubules (Venkei and Szabad, 2005). Kavar^18c has been shown to be dominant-negative (Erdélyi and Szabad, 1989; Venkei and Szabad, 2005), implying that the Kavar^18c-encoded E82K-4-tubulin molecules participate in the same process as wild-type 4-tubulin, such that E82K-4-tubulin hinders function of wild-type 4-tubulin. It appears thus, that 4-tubulin is required for the formation of long microtubules.

Second mutagenesis (by X-rays) of Kavar^21g led to the formation of kavar^rX21 (hereafter referred to as kavar⁰) the only known null-allele of the Tub67C gene (Venkei and Szabad, 2005). A tassel of short and straight microtubules forms in every egg of the kavar⁰/- females that do not carry functional Tub67C genes (-indicates a short deficiency, which removes Tub67C and a few adjacent genes) (Venkei and Szabad, 2005). The short microtubules are nucleated by two barely separated daughter centrosomes. The kavar⁰/- mutant phenotype clearly shows involvement of 4-tubulin in the formation of long microtubules and appropriate separation of the daughter centrosomes.

In wild-type Drosophila embryo, the sperm-introduced basal body takes centrosome function and organizes the formation of long microtubules (Foe et al., 1993). Long microtubules compose the sperm aster and serve as route for shipment of the female pronucleus to close vicinity of the male pronucleus (Callaini and Riparbelli, 1996). They are also involved in separation of the daughter centrosomes, which migrate along the nuclear perimeter (Foe et al., 1993), as well as in the formation of the spindle apparatus in the early embryo (Scholey et al., 2003). (4) Their involvement has also been documented in the central and in the peripheral nervous systems of Drosophila embryos (Máthé et al., 1998).

The lack of long microtubules in eggs of the kavar⁰/- females is surprising, considering that 4-tubulin comprises only about 20% of the -tubulin pool in the Drosophila eggs. The lack of 4-tubulin can not be substituted by another -tubulin isoform (Matthews et al., 1989). Although there is plenty of 1-tubulin and 3-tubulin in every Drosophila egg - encoded by the constitutively expressed and evolutionary highly conserved Tub84B and Tub84D genes - they do not seem to support the formation of long microtubules. What makes 4-tubulin special? What is its function at the beginning of embryogenesis? Where is it localized in the early embryo? How do E84K-4-tubulin molecules block the formation of long microtubules? Here, we used two mutant Tub67C alleles (Kavar^18c and kavar⁰) to understand the role of 4-tubulin in the formation and function of long microtubules. We describe that 4-tubulin is accumulated in the long microtubules that embrace the nuclear envelope, and hypothesize that the vast majority of the force that pushes apart the daughter centrosomes to opposite poles along the nuclear perimeter comes from the fast-growing interpolar microtubules in the early cleavage embryos. It appears - based on results of in vitro tubulin-polymerization assays - that 4-tubulin is required for rapid formation of long microtubules. Since the time available for tubulin polymerization is very short during the initial cleavage divisions (Edgar and Datar, 1996; Ji et al., 2004; Tadros and Lipshitz, 2005) and tubulin polymerization proceeds slowly in absence of 4-tubulin, only short microtubules form, which do not support proper centrosome separation. We also report that the need for 4-tubulin is limited to the initial cleavage divisions. Once the cleavage nuclei populate the egg cortex, forces other than the growing interpolar microtubules accomplish separation of the daughter centrosomes (Cytrynbaum et al., 2003). The ectopically expressed Kavar^18c-encoded E82K-4-tubulin molecules become incorporated into the microtubules, they do not disrupt microtubule-associated functions in imaginal disc and neuroblast cells.

	Results

Top Summary Introduction Results Discussion Materials and Methods References

 
Eggs of the Kavar^18c/+ females are fertilized as in wild type, the centrosome divides and daughter centrosomes form. In wild-type embryos the daughter centrosomes move apart to opposite poles over the perimeter of the nuclear envelope and form a spindle (Foe et al., 1993). In eggs of the Kavar^18c/+ females separation of the daughter centrosomes comes to a standstill when they are 3-4 µm apart and, although the centrosomes nucleate microtubules, they remain short (at 5-8 µm), straight and form a tassel that appears as a monopolar spindle (Venkei and Szabad, 2005). Embryogenesis is terminated at this point inside eggs of the Kavar^18c/+ females irrespective of the genotype of the embryo.

Very similar, if not identical, defects emerge inside eggs of the kavar⁰/- females, which lack functional Tub67C genes (Venkei and Szabad, 2005). The mutant phenotype, i.e. the formation of a single monopolar spindle in every egg of the Kavar^18c/+, the kavar⁰/- and the Kavar^18c/- females may originate through failure of the daughter centrosomes to separate or the collapse of the first cleavage spindle (Sharp et al., 1999; Scholey et al., 2003; Rogers et al., 2004). This implies involvement of 4-tubulin in separation of the daughter centrosomes or in maintenance and/or elongation of at least the first cleavage spindle. The similar defects seen in eggs of the Kavar^18c/+ and the kavar⁰/- females show that 4-tubulin and E82K-4-tubulin molecules participate in the same process, and since E82K-4-tubulin hinders function of 4-tubulin, Kavar^18c is a dominant-negative mutation (Erdélyi and Szabad, 1989; Venkei and Szabad, 2005).

Cytoplasm injections
To reveal the role of 4-tubulin, we injected small cytoplasm samples from eggs of wild-type (control) and Kavar^18c/- hemizygous females into live embryos whose microtubules were highlighted by GFP-tagged -tubulin. Two types of cytoplasm injections were done. In the first set of injections, the detailed effect of the injected cytoplasm was not followed over time. The injected embryos were engaged in the seventh to ninth cleavage cycle. Toxicity of the Kavar^18c/- derived egg cytoplasm was apparent: while larvae hatched from almost all of the 185 embryos injected with wild-type egg cytoplasm, not a single larva hatched from the 138 embryos that had been injected with E82K-4-tubulin-containing cytoplasm. In the second set of experiments, we injected the cytoplasm samples into embryos in the tenth to eleventh cleavage cycle that had their nuclei already in the egg cortex and analyzed the effects of the injected cytoplasm in time-lapse optical sections. In this experiment 20 embryos were injected with-wild type egg cytoplasm, and 30 embryos with cytoplasm from Kavar^18c/- females. The Kavar^18c/- egg cytoplasm had no effect on the already established mitotic spindles and did not disturb events of meta-, ana- or telophase (Fig. 1). The observation excludes the possibility that the monopolar spindles originate through spindle collapse.

View larger version (104K): [in this window] [in a new window]   Fig. 1. Time-lapse series of optical sections throughout the twelfth cleavage cycle following injection of egg cytoplasm into embryos. Microtubules are highlighted by 1-tubulin-GFP. The embryos were injected during anaphase of the eleventh cleavage cycle. The injected cytoplasm samples were taken from wild-type eggs (Wild type) or eggs from Kavar18c/- females (Kavar18c/-). E82K-4-tubulin, introduced in the Kavar18c/- egg cytoplasm, slowed down the separation of the daughter centrosomes (white arrows). The partially separated centrosomes either organize the formation of short microtubules that never assemble to a spindle apparatus (middle column of panels), or fail to localize the nuclei properly (right column of panels; adjoining nuclei in the boxed area); as a consequence tripolar spindles form. Asterisks indicate sites of cytoplasm injection; time after injection is shown in the first column of panels but applies to the entire row; arrows indicate centrosomes, open arrows indicate interpolar microtubule bundles.

View larger version (34K): [in this window] [in a new window]   Fig. 2. Specificity of the monoclonal DM1A antibody to 1-tubulin and 3-tubulin, the constitutively expressed and evolutionary highly conserved -tubulin isoforms, and of rabbit anti-4-tubulin polyclonal antibody that recognizes only 4-tubulin. Notice that DM1A does not recognize 4-tubulin. There is no 4-tubulin in egg cytoplasm of kavar0/- females. (Dilutions were 500-fold for DM1A and 2000-fold for anti-4-tubulin.)

Localization of 4-tubulin
To visualize 4-tubulin throughout the cleavage cycles, we generated a polyclonal anti-4-tubulin antibody by making use of the unique 14 amino acid long C-terminal segment of 4-tubulin. This antibody is specific for 4-tubulin and does not recognize 1-tubulin and 3-tubulin, the evolutionally highly conserved and constitutively expressed isoforms (Fig. 2). The monoclonal anti--tubulin antibody DM1A, however, recognizes 1-tubulin and 3-tubulin but not 4-tubulin and, thus, the two antibodies allow side-by-side detection of the different -tubulin types (Fig. 2).

As summarized on Figs 3, 4 and 5, all three -tubulin isoforms are components of all types of microtubules throughout the cleavage divisions. However, there is a significant enrichment of 4-tubulin in the so-called interpolar microtubules that embrace the nuclear envelope during interphase and early prophase, and push the daughter centrosomes along the nuclear perimeter to opposite poles (Fig. 4) (see also Scholey et al., 2003; Cytrynbaum et al., 2003).

View larger version (116K): [in this window] [in a new window]   Fig. 3. Localization of 1-tubulin and 3-tubulin as well as 4-tubulin throughout the eleventh cleavage cycle in a wild-type embryo. The constitutively expressed 1-tubulin and 3-tubulin isoforms were detected with the monoclonal DM1A primary and a Texas-Red-labeled fluorescent secondary antibody (merged images). 4-tubulin was detected by an affinity-purified rabbit polyclonal anti-4-tubulin primary antibody and an FITC-labeled fluorescent secondary antibody (merged images). DNA appears in blue (merged images). Arrowheads indicate interpolar microtubules.

View larger version (43K): [in this window] [in a new window]   Fig. 4. The maternal 4-tubulin isoform is enriched in the interpolar microtubulues. The 1-tubulin and 3-tubulin and also the 4-tubulin isoforms were stained in fixed cleavage embryos as described in Fig. 3 (1-tubulin and 3-tubulin, red; 4-tubulin, green, DNA, blue) and analyzed in optical sections prepared in an Olympus FV1000 confocal microscope. To reveal the concentrations of 1-tubulin and 3-tubulin as well as 4-tubulin over the nuclear region, we set one line across the centrosomes and another line crossing the interpolar microtubule region. An arc was set to follow the region of the interpolar microtubules. We then determined the distribution of signal intensities (ranging from 0 to 255 arbitrary units) in the pixels along the two lines and the arc. The pixel intensities reflect fluorescence intensities and the amounts of the different types of tubulins. The curves over the lines illustrate - as examples - intensity distributions for a single nuclear region. The intensity distributions along the arc appear above the straight lines below the images. Altogether, 20 nuclear regions were analyzed: five nuclei of four embryos each. The peak intensities (average ± s.d.; n=40) over the centrosome regions and at the intersections in the interpolar microtubules (along the arches) are shown in the figure. The analysis revealed highly significant accumulation of 4-tubulin in the interpolar microtubules (P<0.01, t-test). Merged image shows 4-tubulin, green; 1-tubulin and 3-tubulin, red; DNA, blue.

View larger version (66K): [in this window] [in a new window]   Fig. 5. Localization of 1-tubulin and 3-tubulin (red), 4-tubulin (green) and DNA (blue) in the first mitotic spindle in wild-type embryos, Kavar18c/- and kavar0/- females. The first and only monopolar spindle that appears in eggs of the mutant females is a tassel of short microtubules.

In vitro tubulin-polymerization assays
To determine the role of 4-tubulin in microtubule formation and how E82K-4-tubulin exerts its toxic effect, we carried out two sets of in vitro microtubule polymerization assays. We prepared extracts from 0-hour- to 1-hour-old eggs of +/- females (control) and Kavar^18c/- virgin females, and induced polymerization of tubulin in the egg extracts by addition of GTP. The forming microtubules were stabilized with taxol. The microtubules were pelleted and the types of proteins in the pellets were separated by SDS-PAGE. We then used western blot analysis to detect 1-tubulin and 3-tubulin with DM1A antibody, and 4-tubulin with its specific anti-4-tubulin antibody. To distinguish the different -tubulin isoforms we used fluorescently labeled secondary antibodies. Results of the in vitro tubulin-polymerization assay are summarized in Fig. 6 and show that egg cytoplasm of the Kavar^18c/- and the +/- females contain roughly equal amounts of total -tubulin. Western blot analysis did not detect -tubulin in the supernatant, indicating that both the 4-tubulin and E82K-4-tubulin become incorporated into the forming microtubules. Apparently, and in agreement with results presented in Fig. 5, E82K-4-tubulin does not exert its toxic effect by blocking microtubule formation: it does become incorporated into the microtubules but, however, keeps them short.

View larger version (28K): [in this window] [in a new window]   Fig. 6. E82K-4-tubulin is incorporated into the microtubules during in vitro tubulin-polymerization. Microtubule formation was induced in egg extracts of +/- (contains one wild-type Tub67C gene) and Kavar18c/- females. Formed microtubules were pelleted. Egg extracts, supernatant and microtubule-containing pellet were analyzed by SDS PAGE stained with Coomassie Blue and western blot. The 1-tubulin and the 3-tubulin isoforms were detected with monoclonal DM1A primary and a Texas-Red-labeled fluorescent secondary antibody, 4-tubulin by a polyclonal anti-4-tubulin primary and an FITC-labeled fluorescent secondary antibody.

Unexpectedly, the microtubules grew to an almost equal size in egg extract of control (+/-) and kavar⁰/- females (11.5 µm versus 11.7 µm; Table 1). It appears that equally long microtubules form if the time available for tubulin polymerization is sufficiently long, irrespectively of the presence or absence of 4-tubulin.

To test the effect of 4-tubulin on microtubule polymerization, we studied the kinetics of tubulin polymerization in egg extracts of virgin +/- (control) and kavar⁰/- females and analyzed the length-distribution of the forming microtubules plotted against the time allowed for tubulin polymerization. There is a remarkable difference in kinetics of microtubule formation in the two egg extracts (Fig. 7) (details of tubulin polymerization are summarized in supplementary material Fig. S1). By 2 minutes after the initiation of tubulin polymerization, nucleation seeds appear in both preparations and their numbers do not differ significantly over a field of view (207±42, n=3 in control versus 225±29, n=5 in preparation without 4-tubulin) (Fig. 7). However, whereas microtubules with measurable length appear after 5 minutes in the control, at this time still only nucleation seeds are present in the kavar⁰/- egg extract (199±66, n=5 in control versus 171±47, n=5 in null mutant). During the early phase of elongation even the longest microtubules forming in absence 4-tubulin are shorter than most of the microtubules forming in the presence of 4-tubulin. Analysis of polymerization kinetics curves revealed very similar average microtubule growth rates (maximum 0.35 µm/minute in the control and 0.32 µm/minute under the null conditions). However, whereas in the control preparations the maximum rate is reached by 15 minutes, it took 25 minutes in absence of 4-tubulin, indicating that 4-tubulin accelerates the formation of microtubules. The difference in the distribution of microtubules of different length in the two types of preparations gradually decreases and disappears by about 30 minutes (Fig. 7). Results of the in vitro tubulin-polymerization kinetics indicate that 4-tubulin is required for rapid tubulin polymerization.

View larger version (19K): [in this window] [in a new window]   Fig. 7. Kinetics of in vitro tubulin-polymerization. Length distribution of the forming microtubules plotted against the time allowed for tubulin polymerization. White and grey arrows and bars correspond to egg extracts of females with one wild-type Tub67C gene (+/-, control) and egg extracts of females not carrying functional Tub67C genes (kavar0/-). The wide arrows represent preparations in which only nucleation seeds appeared. The five horizontal lines in the bars correspond to 10th, 25th, 50th, 75th and 90th percentiles, respectively (as shown on the right). For details of tubulin polymerization see supplementary material Fig. S1. White and gray diamonds represent average microtubule lengths for control and null-mutant, respectively. Curves (double line, control; single line, null) were fitted to these points.

The toxic effect of E82K-4-tubulin is limited to the early cleavage cycles

A possible explanation for nontoxicity of the ectopically expressed E82K-4-tubulin is that it is not incorporated into the microtubules. To decide whether the ectopically expressed E82K-4-tubulin (and also the wild-type 4-tubulin) molecules are incorporated into microtubules of neuroblasts and whether they affect the length of microtubules, we generated elav-Gal4; UAS-Kavar^18c larvae and, as a control, elav-Gal4; UAS-Tub67C, dissected the ventral ganglia and analyzed the neuroblast cells. As shown in Fig. 8, E82K-4-tubulin (like wild-type 4-tubulin) becomes incorporated into the spindle microtubules and does not alter spindle shape or size. The longest microtubules are 6-8 µm and appear fully functional.

View larger version (90K): [in this window] [in a new window]   Fig. 8. Ectopically expressed 4-tubulin and E82K-4-tubulin are incorporated into the microtubules in the neuroblast as shown by the encircled spindle apparatus. Ectopic expression of 4-tubulin and E82K-4-tubulin was achieved by driving UAS-Tub67C and UAS-Kavar18c transgenes with the nervous-system-specific elav-Gal4 driver. 1-tubulin and 3-tubulin (red) were detected with monoclonal DM1A primary and a Texas-Red-labeled fluorescent secondary antibody, 4-tubulin and E82K-4-tubulin (both green) with a polyclonal primary anti-4-tubulin and a FITC-labeled fluorescent secondary antibody; DNA is stained blue.

	Discussion

Top Summary Introduction Results Discussion Materials and Methods References

However, to conclude that 4-tubulin is required for the formation of long microtubules, is superficial. In vitro polymerization (Fig. 7) of tubulins revealed that, once sufficient time is available for tubulin polymerization, microtubules of the similar length will form whether 4-tubulin is present or not. It might thus be concluded that 4-tubulin is required for rapid formation of the microtubules. It appears that, although tubulin polymerization starts in eggs of the kavar⁰/- females, it goes on far too slowly and, thus, the forming microtubules can not grow sufficiently long within the time available for the process. The time allowed for tubulin polymerization is limited by the Cyclin-Cdc-controlled cyclic sets of events that start during egg activation, whether the egg is fertilized or not (Edgar and Datar, 1996; Donaldson et al., 2001; Ji et al., 2004; Tadros and Lipshitz, 2005). The lack of formation of sufficiently long microtubules leads to arrest of the cleavage cycles and the eventual death of the embryos. Similar `out-of-phase' and `behind-schedule-and-lost' types of events have already been reported during Drosophila embryogenesis. For example, in embryos defective in cytoplasmic dynein-heavy-chain function the replication cycle comes to an end soon after fertilization, whereas the centrosome cycles proceed normally (Belecz et al., 2001).

A remarkable feature of 4-tubulin is its apparent enrichment in the so-called interpolar microtubules that embrace the nuclear envelope (Figs 3 and 4). The interpolar microtubules interconnect the centrosomes and were proposed to participate in separation of the daughter centrosomes (Cytrynbaum et al., 2003; Scholey et al., 2003). In fact, since diameter of the nuclei is about 10 µm up to the ninth cleavage cycle, the interpolar microtubules need to be as long as 15-16 µm, when perfect geometrical parameters during interphase is a concern (Fig. 9). Since duration of an early cleavage cycle is about 8 minutes, of which the interphase comprises about 4 minutes (Foe et al., 1993; Ji et al., 2004), the interpolar microtubules must grow `fast' (with a speed of about 4 µm/minute) to fulfill their functions. Along progression, beyond the ninth cleavage cycle size of the nuclei decreases to about 5 µm and, correspondingly, the interpolar microtubules need to grow to only approximately 8 µm to extend over the opposite poles. Since length of the interphase in the cleavage cycle ten is approximately 6 minutes (Ji et al., 2004), the slow growth rate of microtubules of 1.3 µm/minute may be adequate for the formation of sufficiently long interpolar microtubules.

View larger version (14K): [in this window] [in a new window]   Fig. 9. Diameter of late interphase cleavage nuclei throughout the cleavage cycles in Drosophila melanogaster wild-type embryos. The distance between opposite poles was calculated assuming ball-shaped nuclei.

Separation of the daughter centrosomes
The lack of 4-tubulin (in embryos of the kavar⁰/- females) leads to partial separation of the daughter centrosomes. In wild-type Drosophila embryos, the centrosomes become duplicated during telophase, leading to the production of two adjacent daughter centrosomes (Foe et al., 1993; Sullivan and Theurkauf, 1995; Debec et al., 1999). The daughter centrosomes are moved apart to opposite poles along the nuclear envelope during the subsequent interphase and prophase as a result of a shift in the balance between outward and inward forces (Cytrynbaum et al., 2003). In the egg cortex (beyond the tenth cleavage cycle) the outward forces that separate the daughter centrosomes are generated first by the pushing force exerted by the growing microtubules, and afterwards by the cytoplasmic dynein associated with the actin microfilament network. The pushing forces are compensated by inward force generated by the C-terminal kinesin Ncd (Robinson et al., 1999; Sharp et al., 2000a; Sharp et al., 2000b; Scholey et al., 2003; Cytrynbaum et al., 2003). It was reported recently that Ncd does not play a role in daughter centrosome separation in course of the first 12 cleavage divisions (Cytrynbaum et al., 2005). Cytrynbaum et al. proposed in a model that at the beginning of daughter centrosome separation almost all the force that move the daughter centrosomes apart originate from the pushing force of the polymerizing tubulin (Cytrynbaum et al., 2003). The forming microtubules are nucleated by the daughter centrosomes. While some of the forming microtubules bump into the other centrosome they exert a pushing force and the mutual thrust leads to separation of the daughter centrosomes. In their model, Cytrynbaum et al. assumed a radial array of microtubules emanating from the centrosomes and concluded that, the polymerization force decreases rapidly with increasing distance between the centrosomes and becomes ineffective by about 3-4 µm for two reasons (Cytrynbaum et al., 2003). First, the number of microtubules encountering the other centrosome drops off together with the increasing separation distance, and the receding centrosomes disappear on the nuclear horizon. Second, in the egg cortex the task of centrosome separation is taken over by cytoplasmic dynein that is linked to the actin cytoskeleton (Scholey et al., 2003; Cytrynbaum et al., 2003; Cytrynbaum et al., 2005).

While the nuclei are still deep down in the egg cytoplasm, cytoplasmic dynein associated with the cortical actin microfilament network cannot contribute to separation of the daughter centrosomes; thus the task of moving the daughter centrosomes to opposite poles seems to be left - as suggested here - to the interpolar microtubules. Remarkably, the interpolar microtubules that contain 4-tubulin are apt to meet the task, because they (1) bend around the nuclear envelope and (2) grow fast and sufficiently long to push the daughter centrosomes towards opposite poles, along the nuclear envelope, over a distance of more than 3-4 µm. It may be a sacrilege to propose that the early cleavage nuclei are kept large (with small curvature) such that while growing and pushing the daughter centrosomes apart the interpolar microtubules can bend along the nuclear envelope. Once the nuclei reach the egg cortex and cytoplasmic dynein joins in the task of centrosome separation, the microtubules do not need to grow for much longer and the embryos can afford to have small nuclei.

View larger version (39K): [in this window] [in a new window]   Fig. 10. Simplified model to explain the mechanism of daughter centrosome separation and migration in the early cleavage (cycles 2-10) Drosophila embryos. The interpolar microtubules, nucleated by the daughter centrosomes (large arrowheads) exert pushing force and begin to separate the daughter centrosomes. Microtubules containing 4-tubulin bend around the nuclear envelope while they grow. Dyneins fixed to the nuclear envelope (not shown) keep the microtubules attached and use them as a migration route to pull the centrosomes in the indicated directions. Centrosome migration stops when the interpolar microtubules bend away from the nuclear envelope and the centrosomes organize symmetric array of microtubules equalizing inward and outward forces.

At the beginning of daughter centrosome separation, the centrosomes organize an asymmetric array of microtubules that are in contact with the nuclear envelope (Fig. 10) - a similar condition has recently been suggested for the astral microtubules by Cytrynbaum et al. (Cytrynbaum et al., 2005). While dyneins - fixed to the nuclear envelope - move along the nearby microtubules they pull the daughter centrosomes apart. Contribution of dynein to centrosome separation ceases by the time when the centrosomes are distantly located and nucleate symmetrical arrays of microtubules (Fig. 10).

There seems to be no well-defined trail for the migration of centrosomes along the nuclear envelope because orientation of at least the first cleavage spindle, which is organized by the pushed apart daughter centrosomes, is random. Random orientation of the first cleavage spindle has been long known from random orientation of the XX/X0, female/male borderline in gynandromorphs (Janning, 1978).

The mode of action of E82K-4-tubulin
In embryos of the Kavar^18c/+ females, E82K-4-tubulin becomes incorporated into microtubules soon after fertilization (Fig. 5). However, microtubules containing E82K-4-tubulin remain short and can not push the daughter centrosomes far apart; consequently, the embryos die and the females are sterile (Venkei and Szabad, 2005). Why can microtubules containing the E82K-4-tubulin not grow long enough? The tubulin molecules form globular structures with characteristic helices and ß sheets (Nogales, 1999). GTP is a structural component of both - and ß-tubulin (see supplementary material Fig. 3). In the evolutionary conserved -tubulins the Glu residue at position 71 is involved (together with a number of amino acid side chains) in GTP-binding through an Mg²⁺ ion (Nogales, 1999; Löwe et al., 2001). Mg²⁺ controls stability and structure of the tubulins (Menéndez et al., 1998). Replacement of Glu71 (which is Glu82 in 4-tubulin) by Lys in E82K-4-tubulin does not prevent the formation of heterodimers between E82K-4-tubulin and ß-tubulin or the assembly of microtubules, as shown by incorporation of E82K-4-tubulin into the microtubules (Figs 5 and 6). However, microtubules containing E82K-4-tubulin become instable, which is probably the reason for the formation of only short microtubules in the presence of E82K-4-tubulin. The destabilizing effect of E82K-4-tubulin is best shown by the finding that, while at least short microtubules form in egg extract of the kavar⁰/- females, there is no microtubule formation in egg extracts of the Kavar^18c/- females. The discrepancy between the in vivo observations (short microtubules form) and in vitro observations (microtubules do not form) can possibly be accounted for by some of the microtubule-associated proteins (MAPs) that are present in vivo, and much of which were absent in the in vitro tubulin-polymerization assays. The destabilizing effect of E82K-4-tubulin is supported by the observation that, in presence of taxol, which is known to stabilize the microtubules, short microtubules grow in the presence of E82K-4-tubulin. Thus taxol, like some of the MAPs, can overcome to some extent the destabilizing effect of E82K-4-tubulin.

Surprisingly, E82K-4-tubulin, which is very toxic to early cleavage embryos, is harmless to late cleavage embryos and does not disturb microtubule functions in the dividing cells (Fig. 8). Explanation for the unexpected behavior of E82K-4-tubulin is most probably related to a major difference between the early- and the late-cleavage cycles and the dividing cells. As discussed above, during the last cleavage divisions in the egg cortex, centrosome separation is slower than during the early divisions because there is more time - longer interphase (Ji et al., 2004) - and less distance - smaller nuclear diameter (Fig. 9) - to travel for daughter centrosomes. The composition of contributing forces is also different; cytoplasmatic dynein anchored to the cortical actin network exerts the majority of outward forces, Ncd starts to generate an inward force that increases with the length of interpolar microtubules (Robinson et al., 1999; Sharp et al., 2000a; Sharp et al., 2000b; Scholey et al., 2003; Cytrynbaum et al., 2003; Cytrynbaum et al., 2005). The above findings suggest that the interpolar microtubules lose their importance in separating the daughter centrosomes.

Although the microtubules containing E82K-4-tubulin grow shorter than those containing wild-type 4-tubulin, they are long enough to `hand over' the daughter centrosomes to cytoplasmic dynein, which will carry on and accomplish centrosome separation. The situation is very similar in the imaginal disc and in neuroblast cells, where incorporation of E82K-4-tubulin into the microtubules has no apparent consequences (Fig. 8). In summary, 4-tubulin is required during early embryogenesis, when only short time intervals are available to accomplish the consecutive events including the rapid formation of long microtubules.

	Materials and Methods

Top Summary Introduction Results Discussion Materials and Methods References

 
Fly stocks
Kavar^18c is one of the ethyl methanesulphonate (EMS)-induced dominant female-sterile mutations (Erdélyi and Szabad, 1989) and kavar⁰ is an X-ray-induced null allele of the Tub67C gene (Venkei and Szabad, 2005). The Kavar^18c/+, Kavar^18c/- and the kavar⁰/- females were generated by standard genetic crosses. The - symbol stands for Df(3L)55, a short deficiency that removes the Tub67C and three of the adjacent genes (Venkei and Szabad, 2005). The Drosophila cultures were raised on standard food and kept at 25°C. For explanation of the genetic symbols see the FlyBase at http://flybase.bio.indiana.edu.

Diameter of embryonic nuclei
To determine the diameter of the cleavage nuclei, we collected embryos from two strains, the w¹¹¹⁸ strain and the histone-GFP strain, in which a transgene encodes a fusion protein of histone H2Av and GFP (Clarkson and Saint, 1999). The diameter of nuclei of w¹¹¹⁸ embryos was measured throughout the cleavage cycles after fixation with methanol and staining with DAPI, and also in tenth to the thirteenth cycle in live histone-GFP embryos. For cycles one to five, all nuclei of at least twenty embryos were measured, and for the later stages at least ten nuclei of at least ten embryos. Diameters of nuclei from fixed and live embryos were basically identical and were pooled for the images shown in Fig. 9.

Cytoplasm injections
To elucidate effects of the mutant cytoplasm on cleavage embryos, we injected about 300 pico liters cytoplasm (3% of total egg volume) from eggs of wild-type females (control) and Kavar^18c/- hemizygous females into the posterior region of wild-type embryos (see Tirián et al., 2000) whose microtubules were highlighted with GFP-tagged 1-tubulin (Grieder et al., 2000). The donor embryos were at the most 30 minutes old and the injected embryos were either in the seventh to ninth cleavage cycle of embryogenesis, with nuclei still deep down in the egg cytoplasm, or in the eleventh to thirteenth cycle in the egg cortex (Foe et al., 1993). The effect of the injected cytoplasm was followed over time by series of optical sections generated with an Olympus VS1000 confocal microscope. The injections were carried out at 25°C.

Immunological techniques
For immunological detection of tubulins we followed standard techniques (Sambrook et al., 1989). Briefly, protein samples were prepared from 0-hour-old to 2-hours-old embryos. The eggs were dechorionated by chlorox, homogenized in SDS buffer and boiled for 10 minutes before SDS denaturing gel electrophoresis. To detect -tubulin, both in western blots and in confocal microscopy, we used the DM1A monoclonal mouse anti--tubulin (T6199; Sigma), which recognizes 1-tubulin and 3-tubulin but not 4-tubulin. The secondary anti-mouse antibodies were labeled either with alkaline phosphatase (Roche), FITC or Alexa Flour-633 (Sigma). To specifically detect 4-tubulin, we raised an olygopeptide-based polyclonal anti-4-tubulin antibody in rabbits. The olygopeptide represented the last 14 amino acids of the 4-tubulin C-terminal and is specific for 4-tubulin (DNAEEGGDEDFDEF). The antibody was affinity-purified on a protein-A column and on nitrocellulose strips, followed by low-pH elution according to Smith and Fisher (Smith and Fisher, 1984). For confocal microscopy, non-specific activities from the protein-A-purified antibody samples were depleted by incubating the antibody with fixed and permeabilized eggs that had derived from kavar⁰/- females and lacked 4-tubulin before use (Venkei and Szabad, 2005). Anti-4-tubulin antibody allows clear distinction - in both western blots and confocal microscopy - between 4-tubulin and the two other -tubulin isoforms in the embryos, 1-tubulin and 3-tubulin (see Fig. 2). Since the C-terminal part is identical in 4-tubulin and E82K-4-tubulin, the anti-4-tubulin antibody recognizes the two types of molecules equally efficient. In western blot analyses the labeled proteins were detected in a Typhoon 8600 device (Amersham).

To analyze structures in the eggs and/or embryos, the chorion was removed by either a brief chlorox treatment or with double-sided sticky tape. After removal of the chorion the embryos were fixed in a mixture of 1:1 4% paraformaldehyde:heptane or in a 1:1 mixture of methanol:heptane. The vitelline membrane was removed subsequently by agitation in a mixture of heptane and methanol. Before processing for analysis, the embryos were stored in methanol at -20°C. For microscopy, embryos were rehydrated and rinsed in PBST. To block nonspecific staining, embryos were incubated in 1% BSA (Sigma) in PBST for 90 minutes at room temperature. For detection of microtubules, embryos were incubated with DM1A mouse monoclonal anti--tubulin antibody (1:1000, overnight at 4°C) and/or with anti-4-tubulin primary antibody (1:200). The primary antibodies were applied in 1% BSA in PBST. After several rinses in PBST, embryos were incubated in secondary antibodies for 3 hours at room temperature or overnight at 4°C. The following secondary antibodies were either anti-mouse or anti-rabbit IgG (Sigma) and were labeled with FITC, Texas-Red or Alexa Flour-633. To detect DNA, embryos were stained with DAPI following incubation with the secondary antibody. Following several rinses in PBST the embryos were mounted in Aqua PolyMount (Polysciences Inc). Optical sections were generated with an Olympus FV1000 confocal microscope.

In vitro microtubule-polymerization assay
In the in vitro tubulin-polymerization experiments, we collected eggs from three types of females: +/- control females that carry one copy of the wild-type Tub67C gene, kavar⁰/- females that do not carry a functional Tub67C gene and Kavar^18c/- females that carry only the dominant female-sterile mutant allele encoding E82K-4-tubulin. To achieve maximum similarity between egg cytoplasms and also a high egg-production rate, females were collected as virgins and were kept together with sterile X0 males (Liu and Kubli, 2003). For isolation of tubulins, we collected eggs for 1 hour and processed them according to Cullen et al. (Cullen et al., 1999). Briefly, after removal of the chorion with bleach, the eggs and/or embryos were homogenized in BRB80 buffer (80 mM Pipes, 2 mM MgCl₂ and 1 mM EGTA) containing a cocktail of protease inhibitors (PI) and 1 mM dithiothreitol (DTT). The homogenate was incubated at 4°C for 30 minutes and spun at 140,000 g at 4-8°C for 30 minutes to remove debris and heavy particles. The supernatant was kept as egg extract.

We conducted two types of in vitro polymerization assays. In the first assay we used taxol, a microtubule-stabilizing agent, to incorporate all tubulins into microtubules. One mM GTP and a final concentration of 20 µM taxol were added to the egg extract before incubation on room temperature to induce polymerization of the tubulins and to stabilize the forming microtubules. The microtubules and microtubule-associated proteins were pelleted by spinning at 80,000 g for 30 minutes through a 30% sucrose cushion in BRB80 buffer. Egg extract, pellet (i.e. the microtubules) and supernatant were analyzed by SDS-PAGE and western-blot with anti--tubulin primary and fluorescently labeled secondary antibodies. In the second assay we tested the effect of E82K-4-tubulin and wild-type 4-tubulin on polymerization and analyzed microtubule formation in extracts of unfertilized eggs. Briefly, PI cocktail, DTT and EGTA to a 1 µM final concentration were added to 50-100 µl of 0-hour-old to 1-hour-old eggs before homogenization. The homogenized samples were stored at 4°C for 30 minutes to disassemble the microtubules. Next, homogenates were centrifuged three times (22,000 g at 4°C for 10 minutes) until they were clear. Tubulin concentration of the egg extracts was about 1 µM according to a semi-quantitative analysis of Coomassie-Blue-stained SDS-PAGEs. Samples were stored at -70°C until use. To produce labeled microtubules, mixed 5 µl of extract were mixed with 0.5 µl of 10xEMM mix [final concentrations: 5% DMSO, 0.5 mM GTP, 4 mM MgCl₂, 1 mM EGTA, 0.3 µM Oregon-Green 514 (Molecular Probes) or 0.3 µM Rhodamine-labeled (Cytoskeleton) bovine tubulin]. To test the effect of E82K-4-tubulin on microtubule formation we transferred the preparations (from eggs of +/-, Kavar^18c/- and kavar⁰/-) to 37°C and incubated for 30 minutes in the presence or absence of taxol. `Presence of taxol' actually means 0.2 µM taxol at the beginning, to stabilize the forming microtubules. After a 15-minute incubation the taxol concentration was raised to 2 µM according to the Mitchison laboratory protocols (http://mitchison.med.harvard.edu). To test the effect of 4-tubulin on polymerization kinetics, we incubated two types of egg extracts (from eggs of +/- and kavar⁰/- females) for 2, 5, 7.5, 10, 12.5, 15, 20 and 30 minutes at 32°C. Polymerization reactions were terminated by diluting the reaction 20-fold with 80T (BRB80+10 µM taxol). Samples (3 µl) were immediately mixed with 3 µl Aqua Polymount on a slide and covered with a coverslip; images were collected with an Olympus IX71 fluorescent microscope. Microtubule length was measured by using the ImageJ image analyzer program. Data were analyzed by using the SPSS^TM statistical program.

Ectopic expression of E82K-4-tubulin
To see whether ectopic expression of the Kavar^18c-encoded E82K-4-tubulin has the same effect on cells as on early cleavage embryos, we generated both UAS-Kavar^18c and, as control, UAS-Tub67C transgenes and expressed those with tissue-specific Gal4 drivers (Duffy, 2002) (FlyBase). For generation of UAS-Kavar^18c and UAS-Tub67C transgenes we inserted cDNAs that contained either the Kavar^18c mutation or the wild-type Tub67C coding sequences into the pUASp2 vector, a derivative of the CaSpeR vector family with multiple copies of the UAS sequence and the mini-white marker gene, and generated germ-line transformants by using standard procedures. Features of the UAS-Tub67C and the UAS-Kavar^18c transgenes are summarized in supplementary material Table 1.

For ectopic expression of E82K-4-tubulin (or wild-type 4-tubulin as control) we generated zygotes, through genetic crosses, in which different Gal4 drivers assured tissue specific expression of a UAS-Kavar^18c (or the UAS-Tub67C control) transgene and determined whether the Gal4; UAS-Kavar^18c flies are viable, possess developmental abnormalities or female-sterility. To detect the ectopically expressed E82K-4-tubulin (or 4-tubulin) we dissected eye discs of late third instar ey-Gal4; UAS-Kavar^18c (and ey-Gal4; UAS-Tub67C) larvae in which the ey-Gal4 eye disc specific driver (Bonini et al., 1997) ensured expression of E82K-4-tubulin (or 4-tubulin) in the eye primordia. The eye discs were analyzed for the presence of E82K-4-tubulin (or 4-tubulin) by the western blot technique. To determine whether E82K-4-tubulin (or 4-tubulin) become incorporated into the spindle apparatus in the neuroblasts we generated larvae in which the elav-Gal4 driver (Luo et al., 1994) drove a UAS-Kavar^18c (or a UAS-Tub67C) transgene. The ventral ganglion was dissected from late third instar larvae, fixed in 4% paraformaldehyde, incubated in both anti-4-tubulin - to label E82K-4-tubulin (or 4-tubulin) - as well as in the DM1A antibody to label -tubulin¹ and -tubulin³ for microscopic analysis and determined whether E82K-4-tubulin (or 4-tubulin) became incorporated into the microtubules of the spindle apparatus.

	Acknowledgments

 
We thank Allan Spradling for the transgenic line expressing GFP:TUB, Kissné Ani, Kiaspátiné Margó and Revész Kati for excellent technical assistance. Support for our work came from several sources: Maternal-Effect and Embryogenesis Research Group of the Hungarian Academy of Sciences, a grant of the Hungarian Ministry of Education (No. 1348), a grant of the Hungarian Scientific Research Fund T 043158, Agency for Research Fund Management and Research Exploitation 3.2.1-2004-04-0411/3.0 and the Graduate Student Program of the University of Szeged.

	Footnotes

 
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/119/15/3238/DC1

	References

Top Summary