EXPERIMENTAL
CELL
RESEARCH
192,
16-‘1
(1%1)
Abnormal Behavior of the Yolk Centrosomes during Early Embryogenesis of Drosophila melanogaster GIULIANOCALLAINI'ANDROMANO Department
of Evolutionary
Riology,
of Siena,
University
INTRODUCTION
Fertilization in Drosophila is followed by 13 nuclear divisions which occur without cytokinesis in rapid succession and lead to a syncytial blastoderm of about 6000 nuclei. The first seven mitoses occur synchronously in the central region of the egg. During the 8th and 9th mitoses most of the nuclei migrate to the embryo surface, and a smaller number remain in the interior [l-5]. The stationary nuclei, or yolk nuclei, still surrounded by the yolk particles, divide in synchrony with the somatic nuclei during nuclear cycles 8 and 9 and cease dividing after the mitosis of cycle 10, becoming polyploid [6]. At cycle 9 some nuclei arrive at the posterior pole of the egg and form the pole cells. Hence there are three different nuclear lineages before the formation of the cellular blastoderm: pole, somatic, and yolk nuclei. It is not clear how yolk nuclei segregate from somatic nuclei [7,8] nor is the mechanism controlling their division known. Although yolk nuclei cease dividing after the 10th stage, many other aspects of the mitotic cycle continue.
reprint
requests
should
4, Fi3100 Siena,
MATERIALS
Italy
AND
METHODS
Collection of embryos. Embryos of Drosophila melanogaster (Oregon-R strain) were collected at 25°C on agar plates, dechorionated in a 50% commercial bleach solution, and washed with distilled water. Excess liquid was removed by blotting on tissue paper and the embryos were processed for different preparations. Staging of embryos. For immunoffuorescence observations ages of the embryos were determined according to Foe and Alberts by direct observation with interference contrast or by counting somatic nuclei.
the [6] the
Fluorescence microscopy. The dechorionated embryos were fixed and their vitelline membranes were removed essentially as described by Warn and Warn [13], except for a final fixation with acetone for 5 min. The embryos were then washed in phosphate-buffered saline (PBS) and incubated 1 h in PBS containing 0.1% bovine serum albumin. For centrosome staining the embryos were incubated for 5 h at
be addressed.
0014.4827/91 $3.00 Copyright (0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
via Mattioli
The nuclear envelope undergoes the same changes observed in the last syncytial mitoses of the somatic nuclei [9], and DNA replication occurs despite the fact that the yolk nuclei are unable to divide further [6]. Recently, Millar et al. [lo] found an antigen in the yolk nuclei which occurs only in the spindles of the mitotic cells. The only apparent difference between yolk and somatic nuclei is the lack of a functional spindle and the failure of the chromosomes to separate. Previous studies on the cytoskeleton of the Drosophila embryo during the first stages of development focused attention on the syncytial blastoderm and little attention was paid to the cytoskeletal components of the yolk region. Spindle microtubules were observed in the center of the embryo only during the intravitelline mitoses; after the nuclei migrated to the periphery of the embryo, the microtubules were no longer visible in the center [ll, 121. Because the ability of most animal cells to divide depends on the presence of a juxtanuclear organelle, the centrosome, from which both spindle and interphase microtubules nucleate, we performed immunofluorescence studies with anti-tubulin and anti-centrosome antibodies and transmission electron microscopy to investigate the behavior and structure of the centrosomes during the segregation of the yolk nuclei.
After the 10th nuclear cycle the yolk centrosomes follow an irregular pathway. Unlike the somatic centrosomes, which move to the opposite poles of the nuclei to form the bipolar spindles, the yolk centrosomes remain as pairs at one pole of the yolk nuclei or shift feebly and nucleate irregular spindles, most of which have only one main pole. The yolk centrosomes are no longer observed near the yolk nuclei, but progressively move away into the surrounding cytoplasm. Despite the irregular behavior of the centrosomes and although the yolk nuclei cease to divide, the yolk centrosome duplication cycle continues. The early development of Drosophila thus provides an excellent natural system for the study of the uncoupling of the nuclear and centrosomal cycles. IX. 1991 Academic Press, Inc.
1 To whom
DALLAI
16
IRREGULAR
BEHAVIOR
embryos 1141 at a dilution of 1:400. For microtubule staining the embryos were incubated for 30 min at room temperature with a monoclonal antibody against n-tubulin (Amersham). After rinsing in PBS the embryos were incubated in secondary antibodies (goat anti-rabbit and goat anti-mouse fluorescein-conjugated IgG, both at a dilution of 1:lOOO (Cappel, West Chester, PA)) and incubated 3 min with 1 pg/ml of the DNA-specific dye Hoechst 33258, to visualize the nuclei. After washing in PBS the samples were mounted in 90% glycerol containing 2.5% n-propyl gallate to reduce photobleaching [15]. Fluorescence observations were carried out with a Leitz Aristoplan microscope equipped with fluorescein and uv filters. Photomicrographs were taken with Kodak Tri-X pan film and developed in Kodak HC 110 developer for 7 min at 20°C. Electron microscopy. For transmission electron microscopy observations the embryos were fixed in the trialdehyde solution of Kalt and Tandler [16] for 2 h. After rinsing in cacodylate buffer 0.1 M, pH 7.2, the embryos were postfixed in 1% osmium tetroxide for 2 h, dehydrated in a graded series of alcohols, and bulk-stained in 1% uranyl acetate for 1 h. After treatment with propylene oxide the embryos were embedded in an Epon-Araldite mixture and polymerized at 60°C for 48 h. Sections cut using an LKB Nova ultramicrotome and a diamond knife (Diatome Ltd, Switzerland) were collected on copper grids and stained with uranyl acetate and lead citrate. Sections were observed and photographed with a Phillips EM 400 electron microscope.
RESULTS
During the first seven intravitelline mitoses the centrosomes, which are the foci of aster and spindle microtubules, undergo changes in shape like those observed during the four rounds of mitoses that precede cellularization. During this period the distinction between yolk and somatic nuclei is not evident, however, during the 6th and 7th mitoses some nuclei are located more internally and others more externally. During the 8th and 9th mitoses the somatic nuclei segregate from the yolk nuclei, which remain stationary in the center of the embryo. When the somatic nuclei reach the embryo surface, the yolk nuclei divide once again and the centrosome shape undergoes the same changes in both yolk and somatic nuclei (Figs. la-d). The embryo in Fig. le is in late telophase of the 10th nuclear division, judging by the number of mitotic spindles and the presence of midbodies which are characteristic of telophase. Abnormal telophase spindles, probably due to the damaging of the centrosome cycle, appear in the central region of the embryo (Fig. If). At this moment, the somatic centrosomes, as visualized by immunofluorescence with a rabbit antiserum that recognizes a centrosome-associated antigen in Drosophila [14], split into two separate units and begin to move toward the opposite poles of the reforming nuclei to organize the interphase and, then, the mitotic arrays of microtubules (Figs. lg and h). Yolk, unlike somatic centrosomes, does not follow a normal pathway. After nuclear cycle 10 they remain as pairs closely associated with the undividing yolk nuclei (Figs. li and j). When the somatic nuclei enter their 11th mitosis, the centro-
OF
CENTROSOMES
17
somes are no longer closely associated with the yolk nuclei, but gradually move away from the nuclear region. Figure 2a is a detail of an embryo treated during the 11th nuclear division with Rb188. This antibody gives a feeble staining of the nuclear region [ 141, allowing visualization of the position of the centrosomes to the nuclei. Most of the centrosomes visible in this field did not remain associated with the nuclei, but moved away (Fig. 2a), and we can see single and paired centrosomes in the same region of the embryo. The moving centrosomes may duplicate even if they are far from the nuclei and even if the nuclei do not divide. Careful observation with the Rb188 antibody, during the 11th nuclear division, reveals tight pairs of centrosomes, indicating that they have just replicated after their dissociation from the nuclei, but have not yet separated (Fig. 2b). Ultrastructural analysis of the free centrosome pairs demonstrates that each centrosome consists of daughter and parent centrioles (Fig. 2~). The centrioles no longer appear as orthogonally arranged cylinders, but gradually separate. The relative position of the centrioles in Fig. 2c is similar to that observed in the somatic centrosome cycle, during the metaphase-anaphase transition (unpublished results), indicating that the centriole cycle continues in the absence of nuclear replication. Lower magnification of embryos during nuclear cycle 12 (Fig. 2g) shows that the number of the centrosomes is higher than that expected after the 10th mitosis, when the yolk nuclei ceased to divide. This confirms that yolk centrosomes undergo several rounds of replication independent of nuclear division. We are unable to give the exact number of the centrosomes in the yolk region at this stage, because the centrosomes were observed in different planes of focus. Focusing at the surface and in the middle of the embryo it was apparent that most of the yolk centrosomes are localized between the yolk nuclei and the periplasm (Figs. 2f and g). However, the number and the distribution of the centrosomes vary among the areas within the egg, and clusters of 4-8 centrosomes were sometimes observed (not shown). Such an arrangement further confirms that the yolk centrosomes replicate independently of the nuclear cycle. There was no apparent relation between the somatic and the yolk centrosome cycles, and the replication of the free yolk centrosomes does not appear to be a synchronous event, to judge from the simultaneous presence of single and paired centrosomes in the same embryo (Fig. 2a). As observed in Fig. lf, yolk centrosomes can nucleate normal mitotic spindles during nuclear cycle 10, even if some of them have small irregularities. During nuclear cycles 11 and 12, despite their unusual behavior the yolk centrosomes are still able to nucleate microtubules. Anti-tubulin staining shows that at this stage they organize abnormal spindles, most of which have only one main pole (Figs. 2d and e). Double labeling with anti-
18
CALLAINI
AND
DALLAI
FIG. 1. l1rosophila embryos stained during the nuclear cycle 10 with Rb188 (b, d, g, j) and anti-tubulin (e, f) antibodies to reveal centrosomes and microtubules and counterstained with Hoechst 33258 (a, c, h, i) to reveal the nuclei. (a-d) Whole mount showing the distribution of the centrosomes during anaphase, focusing at the surface (a, b) and in the interior (c, d) of the embryo; arrows and arrowheads indicate nuclei and centrosomes of the same spindles. (e, f) Spindle morphology during telophase in the periplasm and in the yolk region; arrow and arrowhead point to defective and normal spindles, respectively. The periplasm is partially removed to make the yolk region evident. (g-j) Centrosome behavior during the transition from telophase to the next interphase in periplasm (g, h) and yolk region (i, j); arrowheads indicate centrosome pairs inherited by the daughter nuclei (arrows); yn, yolk nuclei, sn, somatic nuclei. Bar, 70 pm in a-d; 100 Km in e, f; 40 pm in g, h; 20 Frn in i, j.
IRREGULAR
BEHAVIOR
OF
CENTROSOMES
19
[31, 321. Moreover, exposure to low concentrations of diazepam causes inhibition of centrosome shifting and induces the formation of polyploid figures in the early Drosophila embryo [331. Another intriguing question is the differential migration of the nuclei toward the embryo surface during stages 8 and 9, leading to the segregation of somatic, yolk, and pole nuclei. It has been suggested that centrosomes can organize the cytoskeletal network of the syncytial blastoderm [II, 12, 34, 351 and there are data to DISCUSSION indicate that the movement of the nuclei to the cortex is One of the main findings of this study is that the cen- mediated by forces acting on the centrosomes themtrosomes gradually dissociate from the yolk nuclei and selves [36]. Hence the centrosomes seem to play a crumove into the surrounding cytoplasm after the nuclei cial role in determining nuclear segregation. Transplanceased dividing. This is not a general feature in the ani- tation experiments suggest that the developmental mal kingdom, but is consistent with the findings on cen- fates of the nuclei are not predetermined [37], but durtrosome behavior in gnu [17] and aphidicolin-treated ing nuclear cycles 8 and 9 all the centrosomes in the Drosophila embryos [ 181, in which there are dissociated embryo are structurally and functionally equivalent, as centrosomes, and apparently supports an earlier hy- they are able to nucleate functional mitotic spindles. Since only the behavior of the yolk centrosomes is pothesis that the inability of the yolk nuclei to form functional mitotic spindles, and therefore to divide, is affected, there must be a specific mechanism that the due to the loss of their centrosomes [9]. It has been embryo uses to selectively perturb yolk centrosome activity, while leaving the activity of the somatic centrosuggested that the disorganization of the centrosomal structure is related to the loss of parthenogenetic activsomes intact. Because the proper behavior of the yolk ity in Xenopus eggs [19, 201 and that the functional in- centrosomes is affected during the formation of the synactivation of the maternal centrosome in starfish egg is cytial blastoderm, one is left with the conclusion that due to the absence of one centriole [21]. Rb188 antibody this may be due to cytoplasmic factors differentially disand electron microscopy show that yolk centrosomes tributed throughout the egg. Alternatively, the fate of may undergo replication even if the yolk nuclei cease the yolk centrosomes might be directly controlled by dividing and anti-tubulin staining shows that they early expressed zygotic transcripts; however, significant maintain their microtubule nucleating properties. levels of transcripts do not occur until the cellular blasTherefore, the lack of functional mitotic spindles in the toderm forms [38-401 and several studies have shown yolk region is not due to the altered structure of the yolk that the mitoses leading to the syncytial blastoderm can centrosomes. However, yolk centrosomes, unlike the so- be perturbed by maternal-effect mutations [41] but not matic ones, do not move to opposite poles of the daugh- by alterations in the zygotic genome [42,43]. In any case ter nuclei after the 10th mitosis, but initially remain such evidence does not exclude the expression in very small amounts of zygotic genes. Localization of ftz RNA closely associated at one pole and nucleate irregular spindles, most of which have only one main pole. Double was first detected in embryos during the 10th nuclear labeling with anti-tubulin and Rb188 antibodies reveals cycle [44], and in engrailed mutants the yolk nuclei conthe presence of a pair of centrosomes at the pole of the tinue to divide after the 10th nuclear division cycle [45], monopolar spindles (unpublished results). There are suggesting that engrailed gene is required for the correct data to suggest that the centrosomes are attached to the timing of the mitosis of these nuclei. nuclear envelope [22] via intermediate filaments [23] Centrosomes are usually close to the nuclear envelope and that microtubules and microfilaments have been and segregate with chromatin during mitosis, and their implicated in determining centrosome shape, position, number increases in proportion to the number of nuclei. and separation [24-281. A local cytoskeletal alteration During nuclear cycle 12 the ratio of yolk centrosomes to may thus disturb the separation or shifting of the yolk yolk nuclei is higher than expected, demonstrating that centrosomes. The alteration of the centrosome cycle yolk centrosomes continue to duplicate even if the nuwill be the consequence of the arrest of the yolk nuclear clei cease dividing. The uncoupling of centrosomes and cycle, as observed in aphidicolin-treated Drosophila em- the nuclear cycle has also been observed ingnu Drosophbryos [18]. However, there have been reports of materila mutants. These embryos, lacking a gene whose prodnal mutations affecting the structure and behavior of uct is needed for nuclear division during early embryothe centrosomes in Drosophila. Several mitotic mutants genesis, develop giant nuclei as a result of continued in Drosophila display polyploid figures (see [29, 301 for DNA duplication in the absence of nuclear division and reviews), and of these mgr and polo mutations seem to the centrosomes proceed through multiple rounds of diaffect the behavior and structure of the centrosomes visions [17, 461. In gnu embryos the free centrosomes tubulin antibody and the DNA-binding fluorochrome Hoechst 33258 allowed us to compare nuclear and spindle microtubule positions. We observed that all the spindles in the yolk region are associated with the chromatin and far from the nuclei no aster microtubules are visible (Figs. 2d and e). However, the egg is a large opaque structure and we cannot exclude the presence of free microtubules in the yolk region.
CALLAINI
AND
DALLAI
FIG. 2. Immunofluorescence with Rb188 (a, h, g) and anti-tuhulin antibodies (d), Hoechst fluorochrome (e, f), and electron microscopy (c) observations of Drosophila embryos. (a) Field of an embryo fixed during the 1 lth nuclear division showing single (arrowheads) and paired (arrows) centrosomes moving away from the yolk nuclei (yn). (h) Detail from a different embryo at the same stage as that in (a), showing replicated centrosomes in paired configuration (arrowheads) or just separated (arrows). (c) High magnification of two centrosomes (arrows) showing a centriole pair on the top of the picture and a single centriole on the bottom. (d, e) Detail of an embryo showing spindle morphology in the yolk region during nuclear cycle 11; arrows and arrowheads point to spindles and nuclei, respectively. (f, g) Whole mount of an embryo during the 12th mitosis; yn, yolk nuclei, sn, somatic nuclei, yc, yolk centrosomes, SC, somatic centrosomes. Bar, 20 pm in a, h; 0.5 pm in c; 40 Frn in d, e; 90 @cm in f, g.
nucleate apparently regular shaped spindles, whereas yolk centrosomes form irregular mitotic spindles when they are close to the nuclei, but lose their nucleating properties away from the nuclear region. This suggests that yolk centrosomes are inactive after their dissocia-
tion from the nuclear periphery. Centrosome replication without nuclear division has been induced by mercaptoethanol in sea urchin embryos [47-491 and in PtK, cells by colcemid [50]. The process of yolk nuclei formais therefore the first example of a tion in Drosophila
IRREGULAR
BEHAVIOR
OF
21
CENTROSOMES
natural system in which centrosome replication is normally uncoupled from nuclear division and constitutes a useful tool for studying the relationship between centrosome and nuclear cycles.
21.
Sluder, G., Miller, C. L. (1989) Dev.
22.
Bornens,
23.
The authors are indebted to Dr. W. Whitfield ofthe Biochemistry, Imperial College of Science, Technology T,ondon, for his generous gift of the Rb188 antibody. project was supported by Grant 40% M.P.I. to R.D.
24. 25.
Katsuma, Y., Swierenga. S. H. H., Marceau, N., and French, s. w. (1987) Biol. Cell. 59, 193-204. Gotlieb, A. I., Subrahmanyan, L., and Kalnins, V. (1983) J. Cell Rid. 96, 1266-1272. Euteneuer, U., and Schliwa, M. (1985) J. Cell Bid. 101,96~103.
26.
Paweletz,
Department of and Medicine, This research
F. J., Lewis, K., Davison, Riol. 131,567~579.
M. (1977)
Nature
N., Mazia,
(London)
D., and Finze
E. D., and Rieder,
270,80-82.
E.-M.
(1987)
Eur.
J. Cell Hid.
44.195-204. REFERENCES 1.
2. 3.
Rabinowitz, Rabinowitz,
M. (1941a) M. (1941b)
J. Morphol. 69, l-49. Anat. Rec. 81, 80-81.
Paulson, D. F. (1950) in Biology of L)rosophila Ed.), pp. 168-274, Wiley, New York.
(Demerec,
M.,
Sonnenblick. B. P. (1950) in Biology of Drosophila M., Ed.), pp. 1622167, Wiley, New York.
6. 6. 7.
Zalokar, M., and Erk, I. (1976) J. Microsc. Aiol. Cell. 25,97-106. Foe, V. E., and Alberts, B. M. (1983) J. Cell Sri. 61, 31-70.
9. IO.
S. E., Freeman,
M., and Glover,
Warn,
12.
Karr, 1509. Warn,
13. 14. 15. 16.
R. M. (1986)
J. Cell Sci. Suppl.
T. L., and Alberts, R. M.,
and Warn,
B. M. (1986) A. (1986)
Exp.
Cell Res.
18. 19. 20.
C., and (1988)
Glover,
J. Cell Bid.
D. M.
(1986)
136, 321-329. Received Revised
March version
30, 1990 received
CELL
RESEARCH
192,
16-‘1
(1%1)
Abnormal Behavior of the Yolk Centrosomes during Early Embryogenesis of Drosophila melanogaster GIULIANOCALLAINI'ANDROMANO Department
of Evolutionary
Riology,
of Siena,
University
INTRODUCTION
Fertilization in Drosophila is followed by 13 nuclear divisions which occur without cytokinesis in rapid succession and lead to a syncytial blastoderm of about 6000 nuclei. The first seven mitoses occur synchronously in the central region of the egg. During the 8th and 9th mitoses most of the nuclei migrate to the embryo surface, and a smaller number remain in the interior [l-5]. The stationary nuclei, or yolk nuclei, still surrounded by the yolk particles, divide in synchrony with the somatic nuclei during nuclear cycles 8 and 9 and cease dividing after the mitosis of cycle 10, becoming polyploid [6]. At cycle 9 some nuclei arrive at the posterior pole of the egg and form the pole cells. Hence there are three different nuclear lineages before the formation of the cellular blastoderm: pole, somatic, and yolk nuclei. It is not clear how yolk nuclei segregate from somatic nuclei [7,8] nor is the mechanism controlling their division known. Although yolk nuclei cease dividing after the 10th stage, many other aspects of the mitotic cycle continue.
reprint
requests
should
4, Fi3100 Siena,
MATERIALS
Italy
AND
METHODS
Collection of embryos. Embryos of Drosophila melanogaster (Oregon-R strain) were collected at 25°C on agar plates, dechorionated in a 50% commercial bleach solution, and washed with distilled water. Excess liquid was removed by blotting on tissue paper and the embryos were processed for different preparations. Staging of embryos. For immunoffuorescence observations ages of the embryos were determined according to Foe and Alberts by direct observation with interference contrast or by counting somatic nuclei.
the [6] the
Fluorescence microscopy. The dechorionated embryos were fixed and their vitelline membranes were removed essentially as described by Warn and Warn [13], except for a final fixation with acetone for 5 min. The embryos were then washed in phosphate-buffered saline (PBS) and incubated 1 h in PBS containing 0.1% bovine serum albumin. For centrosome staining the embryos were incubated for 5 h at
be addressed.
0014.4827/91 $3.00 Copyright (0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
via Mattioli
The nuclear envelope undergoes the same changes observed in the last syncytial mitoses of the somatic nuclei [9], and DNA replication occurs despite the fact that the yolk nuclei are unable to divide further [6]. Recently, Millar et al. [lo] found an antigen in the yolk nuclei which occurs only in the spindles of the mitotic cells. The only apparent difference between yolk and somatic nuclei is the lack of a functional spindle and the failure of the chromosomes to separate. Previous studies on the cytoskeleton of the Drosophila embryo during the first stages of development focused attention on the syncytial blastoderm and little attention was paid to the cytoskeletal components of the yolk region. Spindle microtubules were observed in the center of the embryo only during the intravitelline mitoses; after the nuclei migrated to the periphery of the embryo, the microtubules were no longer visible in the center [ll, 121. Because the ability of most animal cells to divide depends on the presence of a juxtanuclear organelle, the centrosome, from which both spindle and interphase microtubules nucleate, we performed immunofluorescence studies with anti-tubulin and anti-centrosome antibodies and transmission electron microscopy to investigate the behavior and structure of the centrosomes during the segregation of the yolk nuclei.
After the 10th nuclear cycle the yolk centrosomes follow an irregular pathway. Unlike the somatic centrosomes, which move to the opposite poles of the nuclei to form the bipolar spindles, the yolk centrosomes remain as pairs at one pole of the yolk nuclei or shift feebly and nucleate irregular spindles, most of which have only one main pole. The yolk centrosomes are no longer observed near the yolk nuclei, but progressively move away into the surrounding cytoplasm. Despite the irregular behavior of the centrosomes and although the yolk nuclei cease to divide, the yolk centrosome duplication cycle continues. The early development of Drosophila thus provides an excellent natural system for the study of the uncoupling of the nuclear and centrosomal cycles. IX. 1991 Academic Press, Inc.
1 To whom
DALLAI
16
IRREGULAR
BEHAVIOR
embryos 1141 at a dilution of 1:400. For microtubule staining the embryos were incubated for 30 min at room temperature with a monoclonal antibody against n-tubulin (Amersham). After rinsing in PBS the embryos were incubated in secondary antibodies (goat anti-rabbit and goat anti-mouse fluorescein-conjugated IgG, both at a dilution of 1:lOOO (Cappel, West Chester, PA)) and incubated 3 min with 1 pg/ml of the DNA-specific dye Hoechst 33258, to visualize the nuclei. After washing in PBS the samples were mounted in 90% glycerol containing 2.5% n-propyl gallate to reduce photobleaching [15]. Fluorescence observations were carried out with a Leitz Aristoplan microscope equipped with fluorescein and uv filters. Photomicrographs were taken with Kodak Tri-X pan film and developed in Kodak HC 110 developer for 7 min at 20°C. Electron microscopy. For transmission electron microscopy observations the embryos were fixed in the trialdehyde solution of Kalt and Tandler [16] for 2 h. After rinsing in cacodylate buffer 0.1 M, pH 7.2, the embryos were postfixed in 1% osmium tetroxide for 2 h, dehydrated in a graded series of alcohols, and bulk-stained in 1% uranyl acetate for 1 h. After treatment with propylene oxide the embryos were embedded in an Epon-Araldite mixture and polymerized at 60°C for 48 h. Sections cut using an LKB Nova ultramicrotome and a diamond knife (Diatome Ltd, Switzerland) were collected on copper grids and stained with uranyl acetate and lead citrate. Sections were observed and photographed with a Phillips EM 400 electron microscope.
RESULTS
During the first seven intravitelline mitoses the centrosomes, which are the foci of aster and spindle microtubules, undergo changes in shape like those observed during the four rounds of mitoses that precede cellularization. During this period the distinction between yolk and somatic nuclei is not evident, however, during the 6th and 7th mitoses some nuclei are located more internally and others more externally. During the 8th and 9th mitoses the somatic nuclei segregate from the yolk nuclei, which remain stationary in the center of the embryo. When the somatic nuclei reach the embryo surface, the yolk nuclei divide once again and the centrosome shape undergoes the same changes in both yolk and somatic nuclei (Figs. la-d). The embryo in Fig. le is in late telophase of the 10th nuclear division, judging by the number of mitotic spindles and the presence of midbodies which are characteristic of telophase. Abnormal telophase spindles, probably due to the damaging of the centrosome cycle, appear in the central region of the embryo (Fig. If). At this moment, the somatic centrosomes, as visualized by immunofluorescence with a rabbit antiserum that recognizes a centrosome-associated antigen in Drosophila [14], split into two separate units and begin to move toward the opposite poles of the reforming nuclei to organize the interphase and, then, the mitotic arrays of microtubules (Figs. lg and h). Yolk, unlike somatic centrosomes, does not follow a normal pathway. After nuclear cycle 10 they remain as pairs closely associated with the undividing yolk nuclei (Figs. li and j). When the somatic nuclei enter their 11th mitosis, the centro-
OF
CENTROSOMES
17
somes are no longer closely associated with the yolk nuclei, but gradually move away from the nuclear region. Figure 2a is a detail of an embryo treated during the 11th nuclear division with Rb188. This antibody gives a feeble staining of the nuclear region [ 141, allowing visualization of the position of the centrosomes to the nuclei. Most of the centrosomes visible in this field did not remain associated with the nuclei, but moved away (Fig. 2a), and we can see single and paired centrosomes in the same region of the embryo. The moving centrosomes may duplicate even if they are far from the nuclei and even if the nuclei do not divide. Careful observation with the Rb188 antibody, during the 11th nuclear division, reveals tight pairs of centrosomes, indicating that they have just replicated after their dissociation from the nuclei, but have not yet separated (Fig. 2b). Ultrastructural analysis of the free centrosome pairs demonstrates that each centrosome consists of daughter and parent centrioles (Fig. 2~). The centrioles no longer appear as orthogonally arranged cylinders, but gradually separate. The relative position of the centrioles in Fig. 2c is similar to that observed in the somatic centrosome cycle, during the metaphase-anaphase transition (unpublished results), indicating that the centriole cycle continues in the absence of nuclear replication. Lower magnification of embryos during nuclear cycle 12 (Fig. 2g) shows that the number of the centrosomes is higher than that expected after the 10th mitosis, when the yolk nuclei ceased to divide. This confirms that yolk centrosomes undergo several rounds of replication independent of nuclear division. We are unable to give the exact number of the centrosomes in the yolk region at this stage, because the centrosomes were observed in different planes of focus. Focusing at the surface and in the middle of the embryo it was apparent that most of the yolk centrosomes are localized between the yolk nuclei and the periplasm (Figs. 2f and g). However, the number and the distribution of the centrosomes vary among the areas within the egg, and clusters of 4-8 centrosomes were sometimes observed (not shown). Such an arrangement further confirms that the yolk centrosomes replicate independently of the nuclear cycle. There was no apparent relation between the somatic and the yolk centrosome cycles, and the replication of the free yolk centrosomes does not appear to be a synchronous event, to judge from the simultaneous presence of single and paired centrosomes in the same embryo (Fig. 2a). As observed in Fig. lf, yolk centrosomes can nucleate normal mitotic spindles during nuclear cycle 10, even if some of them have small irregularities. During nuclear cycles 11 and 12, despite their unusual behavior the yolk centrosomes are still able to nucleate microtubules. Anti-tubulin staining shows that at this stage they organize abnormal spindles, most of which have only one main pole (Figs. 2d and e). Double labeling with anti-
18
CALLAINI
AND
DALLAI
FIG. 1. l1rosophila embryos stained during the nuclear cycle 10 with Rb188 (b, d, g, j) and anti-tubulin (e, f) antibodies to reveal centrosomes and microtubules and counterstained with Hoechst 33258 (a, c, h, i) to reveal the nuclei. (a-d) Whole mount showing the distribution of the centrosomes during anaphase, focusing at the surface (a, b) and in the interior (c, d) of the embryo; arrows and arrowheads indicate nuclei and centrosomes of the same spindles. (e, f) Spindle morphology during telophase in the periplasm and in the yolk region; arrow and arrowhead point to defective and normal spindles, respectively. The periplasm is partially removed to make the yolk region evident. (g-j) Centrosome behavior during the transition from telophase to the next interphase in periplasm (g, h) and yolk region (i, j); arrowheads indicate centrosome pairs inherited by the daughter nuclei (arrows); yn, yolk nuclei, sn, somatic nuclei. Bar, 70 pm in a-d; 100 Km in e, f; 40 pm in g, h; 20 Frn in i, j.
IRREGULAR
BEHAVIOR
OF
CENTROSOMES
19
[31, 321. Moreover, exposure to low concentrations of diazepam causes inhibition of centrosome shifting and induces the formation of polyploid figures in the early Drosophila embryo [331. Another intriguing question is the differential migration of the nuclei toward the embryo surface during stages 8 and 9, leading to the segregation of somatic, yolk, and pole nuclei. It has been suggested that centrosomes can organize the cytoskeletal network of the syncytial blastoderm [II, 12, 34, 351 and there are data to DISCUSSION indicate that the movement of the nuclei to the cortex is One of the main findings of this study is that the cen- mediated by forces acting on the centrosomes themtrosomes gradually dissociate from the yolk nuclei and selves [36]. Hence the centrosomes seem to play a crumove into the surrounding cytoplasm after the nuclei cial role in determining nuclear segregation. Transplanceased dividing. This is not a general feature in the ani- tation experiments suggest that the developmental mal kingdom, but is consistent with the findings on cen- fates of the nuclei are not predetermined [37], but durtrosome behavior in gnu [17] and aphidicolin-treated ing nuclear cycles 8 and 9 all the centrosomes in the Drosophila embryos [ 181, in which there are dissociated embryo are structurally and functionally equivalent, as centrosomes, and apparently supports an earlier hy- they are able to nucleate functional mitotic spindles. Since only the behavior of the yolk centrosomes is pothesis that the inability of the yolk nuclei to form functional mitotic spindles, and therefore to divide, is affected, there must be a specific mechanism that the due to the loss of their centrosomes [9]. It has been embryo uses to selectively perturb yolk centrosome activity, while leaving the activity of the somatic centrosuggested that the disorganization of the centrosomal structure is related to the loss of parthenogenetic activsomes intact. Because the proper behavior of the yolk ity in Xenopus eggs [19, 201 and that the functional in- centrosomes is affected during the formation of the synactivation of the maternal centrosome in starfish egg is cytial blastoderm, one is left with the conclusion that due to the absence of one centriole [21]. Rb188 antibody this may be due to cytoplasmic factors differentially disand electron microscopy show that yolk centrosomes tributed throughout the egg. Alternatively, the fate of may undergo replication even if the yolk nuclei cease the yolk centrosomes might be directly controlled by dividing and anti-tubulin staining shows that they early expressed zygotic transcripts; however, significant maintain their microtubule nucleating properties. levels of transcripts do not occur until the cellular blasTherefore, the lack of functional mitotic spindles in the toderm forms [38-401 and several studies have shown yolk region is not due to the altered structure of the yolk that the mitoses leading to the syncytial blastoderm can centrosomes. However, yolk centrosomes, unlike the so- be perturbed by maternal-effect mutations [41] but not matic ones, do not move to opposite poles of the daugh- by alterations in the zygotic genome [42,43]. In any case ter nuclei after the 10th mitosis, but initially remain such evidence does not exclude the expression in very small amounts of zygotic genes. Localization of ftz RNA closely associated at one pole and nucleate irregular spindles, most of which have only one main pole. Double was first detected in embryos during the 10th nuclear labeling with anti-tubulin and Rb188 antibodies reveals cycle [44], and in engrailed mutants the yolk nuclei conthe presence of a pair of centrosomes at the pole of the tinue to divide after the 10th nuclear division cycle [45], monopolar spindles (unpublished results). There are suggesting that engrailed gene is required for the correct data to suggest that the centrosomes are attached to the timing of the mitosis of these nuclei. nuclear envelope [22] via intermediate filaments [23] Centrosomes are usually close to the nuclear envelope and that microtubules and microfilaments have been and segregate with chromatin during mitosis, and their implicated in determining centrosome shape, position, number increases in proportion to the number of nuclei. and separation [24-281. A local cytoskeletal alteration During nuclear cycle 12 the ratio of yolk centrosomes to may thus disturb the separation or shifting of the yolk yolk nuclei is higher than expected, demonstrating that centrosomes. The alteration of the centrosome cycle yolk centrosomes continue to duplicate even if the nuwill be the consequence of the arrest of the yolk nuclear clei cease dividing. The uncoupling of centrosomes and cycle, as observed in aphidicolin-treated Drosophila em- the nuclear cycle has also been observed ingnu Drosophbryos [18]. However, there have been reports of materila mutants. These embryos, lacking a gene whose prodnal mutations affecting the structure and behavior of uct is needed for nuclear division during early embryothe centrosomes in Drosophila. Several mitotic mutants genesis, develop giant nuclei as a result of continued in Drosophila display polyploid figures (see [29, 301 for DNA duplication in the absence of nuclear division and reviews), and of these mgr and polo mutations seem to the centrosomes proceed through multiple rounds of diaffect the behavior and structure of the centrosomes visions [17, 461. In gnu embryos the free centrosomes tubulin antibody and the DNA-binding fluorochrome Hoechst 33258 allowed us to compare nuclear and spindle microtubule positions. We observed that all the spindles in the yolk region are associated with the chromatin and far from the nuclei no aster microtubules are visible (Figs. 2d and e). However, the egg is a large opaque structure and we cannot exclude the presence of free microtubules in the yolk region.
CALLAINI
AND
DALLAI
FIG. 2. Immunofluorescence with Rb188 (a, h, g) and anti-tuhulin antibodies (d), Hoechst fluorochrome (e, f), and electron microscopy (c) observations of Drosophila embryos. (a) Field of an embryo fixed during the 1 lth nuclear division showing single (arrowheads) and paired (arrows) centrosomes moving away from the yolk nuclei (yn). (h) Detail from a different embryo at the same stage as that in (a), showing replicated centrosomes in paired configuration (arrowheads) or just separated (arrows). (c) High magnification of two centrosomes (arrows) showing a centriole pair on the top of the picture and a single centriole on the bottom. (d, e) Detail of an embryo showing spindle morphology in the yolk region during nuclear cycle 11; arrows and arrowheads point to spindles and nuclei, respectively. (f, g) Whole mount of an embryo during the 12th mitosis; yn, yolk nuclei, sn, somatic nuclei, yc, yolk centrosomes, SC, somatic centrosomes. Bar, 20 pm in a, h; 0.5 pm in c; 40 Frn in d, e; 90 @cm in f, g.
nucleate apparently regular shaped spindles, whereas yolk centrosomes form irregular mitotic spindles when they are close to the nuclei, but lose their nucleating properties away from the nuclear region. This suggests that yolk centrosomes are inactive after their dissocia-
tion from the nuclear periphery. Centrosome replication without nuclear division has been induced by mercaptoethanol in sea urchin embryos [47-491 and in PtK, cells by colcemid [50]. The process of yolk nuclei formais therefore the first example of a tion in Drosophila
IRREGULAR
BEHAVIOR
OF
21
CENTROSOMES
natural system in which centrosome replication is normally uncoupled from nuclear division and constitutes a useful tool for studying the relationship between centrosome and nuclear cycles.
21.
Sluder, G., Miller, C. L. (1989) Dev.
22.
Bornens,
23.
The authors are indebted to Dr. W. Whitfield ofthe Biochemistry, Imperial College of Science, Technology T,ondon, for his generous gift of the Rb188 antibody. project was supported by Grant 40% M.P.I. to R.D.
24. 25.
Katsuma, Y., Swierenga. S. H. H., Marceau, N., and French, s. w. (1987) Biol. Cell. 59, 193-204. Gotlieb, A. I., Subrahmanyan, L., and Kalnins, V. (1983) J. Cell Rid. 96, 1266-1272. Euteneuer, U., and Schliwa, M. (1985) J. Cell Bid. 101,96~103.
26.
Paweletz,
Department of and Medicine, This research
F. J., Lewis, K., Davison, Riol. 131,567~579.
M. (1977)
Nature
N., Mazia,
(London)
D., and Finze
E. D., and Rieder,
270,80-82.
E.-M.
(1987)
Eur.
J. Cell Hid.
44.195-204. REFERENCES 1.
2. 3.
Rabinowitz, Rabinowitz,
M. (1941a) M. (1941b)
J. Morphol. 69, l-49. Anat. Rec. 81, 80-81.
Paulson, D. F. (1950) in Biology of L)rosophila Ed.), pp. 168-274, Wiley, New York.
(Demerec,
M.,
Sonnenblick. B. P. (1950) in Biology of Drosophila M., Ed.), pp. 1622167, Wiley, New York.
6. 6. 7.
Zalokar, M., and Erk, I. (1976) J. Microsc. Aiol. Cell. 25,97-106. Foe, V. E., and Alberts, B. M. (1983) J. Cell Sri. 61, 31-70.
9. IO.
S. E., Freeman,
M., and Glover,
Warn,
12.
Karr, 1509. Warn,
13. 14. 15. 16.
R. M. (1986)
J. Cell Sci. Suppl.
T. L., and Alberts, R. M.,
and Warn,
B. M. (1986) A. (1986)
Exp.
Cell Res.
18. 19. 20.
C., and (1988)
Glover,
J. Cell Bid.
D. M.
(1986)
136, 321-329. Received Revised
March version
30, 1990 received
