Differential behavior of centrosomes in unequally dividing blastomeres during fourth cleavage of sea urchin embryos

Cell division usually results in the production of two daughter cells similar in size and appearance. Instances of unequal cleavage, however, are relatively rare, with the exception that they commonly occur during embryonic differentiation in many animals. Examples of unequal division in early development include polar body formation during meiotic maturation of eggs (Wolpert, 1960), the first divisions of nematode (Albertson, 1984; Strome and Hill, 1988), oligochaete (Shimuzu, 1982) and mollusc (Allen, 1953; Dan and Ito, 1984) embryos, division of grasshopper neuroblasts (Carlson, 1977) and fourth division of sea urchin vegetal blastomeres (Okazaki, 1975; Dan, 1979, 1984). Unequal cleavage during embryogenesis is important in that it represents an initial event in cell differentiation. For example, sea urchin vegetal blastomeres divide unequally to produce micromeres, which are precursors of the skeletogenic mesenchyme, and macromeres.

Unequal division results from an eccentric positioning of the cleavage furrow, and the location of the cleavage furrow has been shown to be determined by the location of the spindle (Conklin, 1917; Harvey, 1935; Kawamura, 1960, 1977) or the spindle asters (Rappaport, 1969; Hiramoto, 1971). In sea urchin vegetal blastomeres, the spindle is assembled in an eccentric location, subsequent to a migration of the nucleus toward the vegetal pole (Dan, 1979). The peripheral spindle pole becomes closely associated with the plasmalemma; in fact, the peripheral aster is flattened and truncated (Dan and Nakajima, 1956), presumably due to the physical proximity of the plasma membrane. Dan and Ito (1984) suggested that the physical interaction of a spindle pole with the plasma membrane is an essential component of unequal division. In addition, the fact that sea urchin vegetal blastomeres possess dissimilar spindle poles raises the possibility that differential spindle pole behavior during mitosis and cytokinesis is also an integral part of unequal cleavage. Because centrosomes organize the spindle (Sluder and Rieder, 1985a), they may represent the organelles that engineer spindle pole behavior during unequal cleavage. To test this possibility, centrosomal dynamics during fourth cleavage of sea urchin embryos were examined by immunofluorescence methods in conjunction with laser scanning confocal microscopy and three-dimensional computer reconstruction techniques in order to correlate centrosomal morphology with the structural events characterizing this unequal cell division.

Eggs and sperm from the sea urchins Lytechinus pictus and Strongylocentrotus purpuratus were obtained by injecting 0.5 M KC1 into the coelomic cavity; eggs were shed into artificial seawater (AFSW) and sperm were collected ‘dry’ on ice. Fertilization envelopes were removed by inseminating eggs in 3–6 mM 3-amino-l,2,4-triazole (Sigma Chemical Co., St Louis, MO) and subsequently passing zygotes or embryos through 54–74 μm mesh nylon membranes (Showman and Foerder, 1979). Embryos were cultured in suspension in artificial seawater with stirring until fourth cleavage, at which time they were prepared for either fluorescent labeling or spindle isolation.

Specimens of L. pictus were transferred to Ca²⁺-free AFSW shortly after third cleavage (about 190min post-insemination). Because of the tendency of sea urchin embryos to dissociate in Ca²⁺-free AFSW, numerous isolated blastomeres, blastomere pairs and blastomere tetrads were obtained. At various timepoints from about 200–250 min post-insemination, blastomeres were pipetted onto polylysine-coated coverslips and allowed to adhere for only lmin or less in order to avoid cytoskeletal perturbations (Schroeder, 1987). Attached cells were either extracted with glycerol and Triton X-100 and fixed with ice-cold methanol as described by Balczon and Schatten (1983), or directly fixed with ice-cold methanol alone (Harris et al. 1980).

Centrosomes were labeled with one of two monoclonal antibodies, Ah6 (described previously by Schatten et al. 1987) or 4D2. Monoclonal antibody 4D2 was raised against purified centrosomal material from S. purpuratus (Thompson-Coffe et al. unpublished) and was found to produce staining patterns similar to those obtained with the anti-centrosomal antibodies Ah6 and 5051 (Calarco-Gillam et al. 1983; Schatten et al. 1986, 1987). All of the immunofluorescent centrosomal images depicted in this report were taken from 4D2-labeled specimens because of the low background and strong signal exhibited by this antibody. Methanol-fixed or extracted and fixed embryos were incubated in undiluted 4D2 hybridoma culture medium for Ih at 37 °C. Following PBS washes, specimens were incubated for 1 h at 37 °C with goat anti-mouse polyvalent antibodies (anti-IgG, -IgA and -IgM) conjugated to fluorescein (Zymed Laboratories, Inc., San Francisco, CA) diluted 1:60. Embryos were counterstained with 5 pg ml^-1 Hoechst 33258 (Polysciences, Inc., Warrington, PA) to label DNA, washed in PBS and mounted in 90% glycerol and 10% PBS containing 0.1% p-phenylenediamine to retard photobleaching. Some embryos were also labeled with affinity-purified antitubulin rabbit polyclonal antibodies raised against pig brain tubulin in order to visualize microtubules. Immunocytochemistry and counterstaining were carried out as described for 4D2, with the exception that donkey anti-rabbit IgG conjugated to Texas Red (Jackson ImmunoResearch, West Grove, PA) was used as the secondary labeled antibody, diluted 1:200. Labeled L. pictus embryos were examined with either a Zeiss Axiophot epifluorescence microscope or a Bio-Rad MRC-600 laser scanning confocal microscope (LSCM).

Series of optical sections were collected through fluorescently labeled embryos and blastomeres or isolated spindles with the MRC LSCM and stored on 200 megabyte optical disks using a Nimbus VX 386 microcomputer and an IBM 3363 optical disk drive (Holy et al. 1991). Data sets (stacks of optical sections) were transferred to a Silicon Graphie 4D/70GT Iris graphics workstation and reconstructed using Voxel View volume rendering software (Vital Images, Inc., Fairfield, Iowa).

The amounts of immunofluorescent material within macromere, micromere and animal blastomere centrosomes were compared by analyzing stacks of optical sections collected under identical conditions with the LSCM. The collection aperture of the MRC LSCM was opened slightly from its smallest position, resulting in an optical slice thickness of approximately 1 μm. Optical sections were then collected at 1 μm intervals, thereby minimizing overlap or skipping of fluorescent signal during collection. The projected image therefore included all of the fluorescently labeled centrosomal material within a blastomere. The settings on the LSCM (aperture size, black level, gain and choice of neutral density filter) remained the same for each sample examined, and all of the data sets for statistical analyses were collected in the same session (and within 48 h of immunolabeling) to circumvent variability in fluorescent signal between samples due to storage.

Stacks of optical sections completely spanning both centrosomes of mitotic blastomeres were projected (i.e. overlaid) using MRC LSCM software to yield a single composite image, and individual centrosomes at metaphase were outlined using the ‘area’ command included in the MRC LSCM software. The mean pixel value (i.e. fluorescence intensity) and area in pm² were obtained for each centrosome examined. Because animal and vegetal blastomere centrosomes at metaphase are spherical (see Results), the spherical volume of each centrosome was calculated from its projected area. Volume in /an³ was multiplied by the mean pixel intensity of the projected image to obtain a quantitative measure for each centrosome.

At approximately 30 min prior to the onset of fourth cleavage, cultured S. purpuratus embryos were treated with 0.05% trypsin in Ca²⁺-free AFSW for 10 min at 12 °C. Embryos were then washed with 0.1 mg ml⁻¹ soy bean trypsin inhibitor, 5mM benzamidine and 100/ZM PMSF (phenylmethylsulfonyl fluoride) in Ca²⁺-free AFSW for 10 min at 12 °C. When most embryos appeared to reach metaphase of fourth cleavage, they were gently pelleted with a hand centrifuge and suspended in lysis medium consisting of 1M glycerol, 25 mM Pipes, 0.5 mM MgCl₂, 5mM EGTA and 100 μM PMSF at 25 °C. Spindles were collected from lysed cells by low-speed centrifugation and resuspended and stored at 4°C in lysis medium containing 25% glycerol.

Although animal and vegetal blastomeres are about the same size in eight-cell S. purpuratus embryos, vegetal blastomeres are slightly smaller than animal blastomeres at the same stage in L. pictus (Fig. 1A). In both dissociated and intact embryos, blastomere divisions are already somewhat asynchronous by fourth cleavage. By prophase of fourth cleavage, vegetal blastomere nuclei have migrated close to the vegetal pole, while animal blastomere nuclei remain positioned near the cell center (Fig. 1A,F). In early prophase, centrosomal material circumscribes the nucleus (Fig. 1B). As prophase progresses, bipolarization of centrosomal material occurs: the thin, diffuse nuclear blanket of centrosomal material condenses to form two aggregations on opposite sides of the nucleus (Fig. 1G). Bipolarization originates in an orientation predictive of the ensuing cleavage pattern of the animal and vegetal blastomeres (Fig. 1D,E).

Shortly after bipolarization, the nuclear envelope breaks down and spindle formation occurs. Spindles form centrally within animal blastomeres but are assembled in an eccentric position close to the vegetal pole in vegetal blastomeres (Fig. 2A). During prometaphase, centrosomal material continues to condense and eventually forms compact spheres at late metaphase (Fig. 2B). Isolated spindles from animal and vegetal blastomeres display distinct differences with respect to astral organization. Animal blastomere spindles possess symmetrical, radiate asters resembling those formed during the first three equal cleavages. By contrast, the aster adjacent to the vegetal pole plasmalemma in vegetal blastomeres, i.e. the micromere aster, is highly flattened (Fig. 2C,D).

A total of 80 maximally compacted centrosomes from both animal and vegetal blastomeres were measured for fluorescence intensity (mean pixel value) and volume. The product of the volume and mean fluorescence intensity was used as a quantitative measure of centrosomal material present within each spindle pole; by this measure, spindle poles segregating to micromeres contained significantly less 4D2-immunoreactive centrosomal material than did the macromere spindle poles (Fig- 3). Additionally, the sum of the mean values for macromere and micromere poles was slightly less than twice the sum of the mean animal centrosomal value (Table 1). This indicates that micromere spindle poles contain less immunoreactive centrosomal material than macromere spindle poles (36% and 64% of total vegetal blastomere immunoreactive centrosomal material, respectively), and that the total amount of immunoreactive centrosomal material within vegetal blastomeres is somewhat less than the amount within animal blastomeres (47% and 53% of combined amount of vegetal and animal blastomere immunoreactive centrosomal material, respectively) in L. pictus.

As anaphase progresses, micromere spindle pole centrosomes begin to elongate, while macromere pole centrosomes remain spherical (Fig. 4B). The micromere centrosome continues to elongate throughout late anaphase and telophase and eventually resembles a long thread (Fig. 4E,H,K). Three-dimensional reconstruction indicates that the initial elongation of micromere centrosomes results in the formation of a thin oval plate that progressively becomes more filiform as elongation continues (Fig. 5). By telophase, macromere centrosomes flatten slightly, giving rise to a thick disk-like structure (Fig. 5).

Immunofluorescent labeling of microtubules indicates that the microtubule organizing centers of both macromeres and micromeres split by late telophase (Fig. 4J). By early interphase following fourth cleavage, centrosomal material decondenses and begins to spread out and cover the nuclear surface once again.

Unlike the differential behavior of micromere and macromere centrosomes throughout anaphase and telophase, both spindle pole centrosomes of equally dividing animal blastomeres follow a common pattern of structural reorganization. During anaphase, the previously spherical centrosomes elongate slightly and generally form oval disks (Fig. 4L); these disks lengthen somewhat throughout telophase and eventually decondense to envelope the nucleus by early interphase following fourth cleavage in a manner similar to that of the vegetal centrosomes.

The products of the unequal fourth cleavage of sea urchin embryos are macromeres and micromeres (reviewed by Okazaki, 1975; Horstadius, 1975; Dan, 1984; Davidson, 1989). Micromere progeny differentiate to form the primary, skeletogenic mesenchyme, and unequal division is thought to constitute the initial step in this differentiation (Dan et al. 1983). Tanaka (1976) found that when fourth cleavage of Hemicentrotus pulcherrimus, Temnop- leurus toreumaticus and Pseudocentrotus depressus embryos were made equal by detergent treatment, deficiencies in primary mesenchyme formation occured. Some species of sea urchins, however, are still able to produce skeletal structures following equalization of the fourth division by detergent treatment (Langelan and Whitley, 1985), presumably by the conversion of secondary mesenchyme cells into skeletogenic mesenchyme (Etten- sohn and McClay, 1988).

A hallmark of unequally dividing cells is a peripherally positioned spindle that closely adjoins the plasma membrane. Specific mechanical linkages between the spindle and plasma membrane have been demonstrated for a number of unequally dividing cell types (Conklin, 1917; Harvey, 1935; Lutz et al. 1988), and Dan and Ito (1984) suggested that spindle–membrane interaction constitutes a fundamental, prerequisite for unequal division. It appears that a spindle can attain an eccentric location within the cell by two distinct mechanisms: (1) peripheral migration of a centrally formed spindle, and (2) peripheral migration of the nucleus prior to spindle assembly.

Centrally formed spindles migrate peripherally prior to cytokinesis during meiotic maturation or early development in the eggs and embryos of nematodes (Albertson, 1984), annelids (Shimizu, 1982; Hamaguchi et al. 1983; Lutz et al. 1988) clams (Longo and Anderson, 1970; Dan and Ito, 1984) and mice (Longo and Chen, 1985). In contrast, the fourth division of sea urchin vegetal blastomeres begins with migration of the nucleus toward the vegetal pole (Dan, 1979, 1984; Lutz and Inoue, 1982; Schroeder, 1987). In Strongylocentrotus purpuratus, nuclear migration is well advanced by late prophase, at which time one centrosome (the macromere centrosome) is centrally located while the other (micromere centrosome) fies subjacent to the vegetal pole plasma membrane (Schroeder, 1987). The mitotic spindle at fourth cleavage in sea urchins is therefore assembled in an eccentric location.

In many types of unequally dividing cells, peripheral spindles are oriented orthogonal to the plasma membrane. In contrast, mouse oocytes display anastral spindles that lie tangential to the plasma membrane. During polar body formation, however, mouse oocyte spindles rotate to an orthogonal position (e.g. see Sato and Blandau, 1979). In orthogonal spindles, the spindle pole adjoining the plasmalemma (the peripheral pole) frequently displays a truncated or flattened aster, and it has been suggested that differential astral growth or disassembly represents the mechanism by which these spindles attain their eccentric location (Ishizaka, 1969; Czihak, 1973; Kawa- mura, 1977). Dan and Ito (1984) argue, however, that, at least in Spisula oocytes, both asters are similar, one merely becoming deformed as a result of its close apposition to the plasma membrane. The asters of Chaetopterus sp. eggs also appear to be identical: the flattened peripheral aster is able to regain a radiate shape when mechanically displaced from the membrane by micromanipulation methods, and either aster is capable of forming a mechanical linkage with the plasmalemma (Lutz et al. 1988).

Quantitative differences in the amount of immunoreactive centrosomal material between the macromere and micromere poles at metaphase suggest that spindle poles of sea urchin vegetal blastomeres at fourth cleavage are distinct. Furthermore, micromere asters remain flattened after removal from the cell, indicating that either specific structural elements are present within the aster that help maintain its flattened shape, or the micromere aster is intrinsically radiate in only two dimensions. The centrosome model advanced by Mazia (1984) could conceivably produce such a two-dimensio- nally radiate aster if the microtubule nucleating subunits were appropriately arranged. Interaction of the micromere pole with the plasmalemma appears to be required for the assembly of a two-dimensionally radiate aster, because the detergent treatment used to equalize fourth cleavage produces centrally formed spindles that display asters similar in size and shape (Tanaka, 1976).

Two differences were noted between the peripheral (micromere) and central (macromere) spindle pole centrosomes in sea urchin vegetal blastomeres at fourth cleavage: (1) micromere centrosomes contain less 4D2- immunoreactive material than macromere centrosomes at metaphase, and (2) micromere centrosomes undergo a unique filiform elongation during late anaphase and telophase. Quantitation of 4D2 immunofluorescence shows that the fourth cleavage of vegetal blastomeres is unequal not only in the division of cytoplasm, but also in the partitioning of centrosomal material. Although the centrosomal component carrying the 4D2-binding epitope has not been identified biochemically, the labeling pattern produced by this monoclonal antibody resembles that obtained with the well-characterized anti-centrosomal antibodies 5051 and Ah6 (Calarco-Gillam et al. 1983; Schatten et al. 1986, 1987, 1988). The labeling pattern produced with all three of these antibodies appears to correspond to the distribution of peri-centriolar material; for this discussion, we will simply refer to this as ‘centrosomal material’. Notwithstanding the lack of biochemical information regarding the 4D2-binding epitope, this antibody has revealed significant behavioral differences between the centrosomes of sea urchin blastomeres at fourth cleavage.

The difference in the amount of centrosomal material present within each metaphase spindle pole primarily reflects differences in the volumes of the spherical centrosomes, because the mean pixel value, i.e. intensity of fluorescence, is closely comparable between animal, macromere and micromere centrosomes (Fig. 3). It might be argued that micromere centrosomes possibly contain the same amount of material as macromere centrosomes but are more condensed, resulting in steric hindrance during immunolabeling and artifactually low pixel values. This appears not to be the case for three reasons: (1) by this argument, a correlation should exist between centrosomal spherical volume (i.e. degree of condensation of centrosomal material) and pixel value, and such a correlation is not observed (Fig- 3); (2) elongate telophase micromere centrosomes, which present much more surface area that is available for immunolabeling, display comparable pixel values to those of spherical metaphase micromere centrosomes (data not shown); and (3) measurements of pixel values from individual optical sections taken through the midplane of spherical centrosomes show that fluorescence intensity progressively increases and reaches maximal values toward the center of the centrosome.

The micromere spindle pole centrosome is different from the other centrosomes of the embryo at fourth cleavage in that it undergoes a striking filiform elongation during late anaphase and telophase. This elongation may be microtubule-dependent, because colchicine has been shown to affect centrosomal shape during the first cell cycle (Schatten et al. 1988). In this study, however, obvious differences between the microtubule organizing centers of the macromere and micromere poles have not been observed at telophase, with the important exception that the micromere microtubule organizing, center is closely apposed to the vegetal plasmalemma (Fig. 4J). It is possible that elongation of the micromere centrosome is mediated by the combined influences of the microtubular cytoskeleton and the cortical cytoplasm, including the plasma membrane.

During the first three cleavage divisions, all centrosomes undergo similar structural rearrangements following bipolarization (see Boveri, 1901; Paweletz et al. 1984, 1987a,b;,Schatten et al. 1986, 1987, 1988). During the fourth division, by contrast, micromere centrosomes behave differently from macromere and animal blastomere centrosomes. It therefore appears that in sea urchin embryos equal cleavages are characterized by equivalent centrosomes, whereas unequal cleavage involves specialized, divergent centrosomal behavior. The striking elongation exhibited by the micromere centrosome at late anaphase and telophase suggesta that a unique event is occurring at this pole. It is tempting to presume that this event involves a specific interaction of this spindle pole with the plasmalemma, and it is intriguing that centrosomal elongation begins at about the same time as cytokinesis. Thus, although models of cytokinesis that rely upon forces produced by the aster have been largely supplanted, unequal division may represent a special case where microtubule dynamics are closely involved in the physical division of cytoplasm.

Quantitation of immunofluorescence indicates that vegetal centrosomes are already distinct from each other by metaphase, and it is possible that these quantitative differences arise during bipolarization. Unequal centrosomal inheritance may serve to maintain the proportionality of the centrosome-to-cytoplasm ratio in unequally dividing cells. It is interesting that centrosomal material is controlled in this fashion, because the essential reproductive capacity of sea urchin centrosomes has been shown to be correlated with the presence of centrioles, and not peri-centriolar material (Sluder and Rieder, 1985b). This suggests that, in addition to receiving the necessary elements for centrosomal replication, the total amount of centrosomal material a daughter cell receives is also important. It is particularly intriguing that, in addition to the inequality of macromere/micromere centrosomal inheritance, vegetal blastomeres, which are slightly smaller than animal blastomeres in L. pictus, also receive slightly less centrosomal material (Table 1). It may be that the amount of centrosomal material a daughter cell inherits is closely correlated with the amount of cytoplasm it receives.

In summary, micromere spindle pole centrosomes exhibit quantitative and structural differences from macromere and animal pole centrosomes at fourth cleavage. The behavior of micromere centrosomes indicates that a unique reorganization of the micromere spindle pole and centrosome occurs within the vegetal pole cortex during fourth division. Centrosomes may therefore play a fundamental role in unequal cell division during fourth cleavage; such a role is of considerable developmental importance because unequal division in sea urchin embryos constitutes a prelude to cell differentiation.

The authors are grateful for the insightful discussions held with Dr Heide Schatten and Dr Daniel Mazia regarding centrosomal events during cleavage, and for the helpful advice of Cathy Thompson-Coffe, Dr Gerard Coffe and Calvin Simerly regarding fertilization, culture and handling of sea urchin embryos. Dr Stephen Paddock and Peter DeVries of the Integrated Microscopy Resource, University of Wisconsin-Madison, provided training in laser scanning confocal microscopy and three-dimensional computer reconstruction methods. Monoclonal antibody 4D2 was generously provided by Cathy Thompson-Coffe. This research was supported by grants from the NIH: HD12913 to G.S., and postdoctoral NRSA GM12835 to J.H. The Integrated Microscopy Resource is funded as an NTH Biomedical Research Technology Resource.