ABSTRACT
During embryogenesis, the β3 tubulin gene of Drosophila is transcribed predominantly in the mesoderm. We have raised antibodies specific to the C-terminal domain of the β3 tubulin and analysed by immunostaining the distribution of this tubulin isotype during Drosophila embryogenesis. The protein is first detectable in the cephalic mesoderm at maximal germband extension. Shortly afterwards, β3 tubulin is expressed in single cells at identical positions of the thoracic and abdominal segments. We suggest that these cells represent muscle pioneer cells of Drosophila. During later embryonic development the somatic musculature, visceral musculature, dorsal vessel and macrophages contain β3 tubulin. In dorsalizing mutants dorsal, snail and twist, which do not form a ventral furrow during gastrulation, β3 expression is greatly reduced but not completely abolished. Our analysis shows that β3 tubulin immunostaining characterizes the differentiation of mesodermal derivatives during embryogenesis.
Introduction
Tubulins are the main structural components of the meiotic and mitotic spindle, neural processes, cilia and flagella, and form part of the cytoskeleton. In Drosophila, as in all higher organisms, α and β tubulins are encoded by gene families (for review see Cleveland & Sullivan, 1985). The β tubulin genes show a clear developmentally regulated mode of expression (Bialojan et al. 1984; Natzle & McCarthy, 1984). During early embryogenesis, the β1 tubulin mRNA is maternal in origin and homogeneously distributed, whereas in later stages it is found mainly in the developing nervous system, presumably due to embryonic expression (Gasch et al. 1988). In contrast, in situ localization of the β3 tubulin mRNA revealed restriction to the mesoderm (Gasch et al. 1988). Expression of the β3 tubulin gene in cephalic, somatic and visceral mesoderm, indicates a specificity of this tubulin isotype for mesodermal derivatives such as the musculature. The high tubulin mRNA content in mesodermal derivatives is in agreement with the cytological observation of numerous microtubules in differentiating muscles during pupal stages of insect development (Crossley, 1978).
In contrast to the comparatively well-characterized processes controlling pattern formation and differentiation of the ectodermal germ layer of Drosophila, little is known about the corresponding events leading to the formation of internal structures such as the muscle system (for review see Lawrence, 1985; Martinez-Arias & Lawrence, 1985; Akam, 1987; Technau, 1987). The availability of mutants greatly facilitates studies of differentiation processes. Mutants demonstrating a dorsalizing effect, and thus lacking ventral furrow formation during gastrulation, have been described though it is not clear if all muscle cells are eliminated in these mutants (Anderson & Nüsslein-Volhard, 1986).
We report here the use of an antibody specific for β3 tubulin to follow the differentiation of mesodermal cells in both wild-type embryos and in dorsalizing mutants. In the wild-type embryo, β3 tubulin expression is initiated in single cells of the mesoderm in every segment at maximal germband extension. During later embryogenesis, the β3 tubulin is characteristic of mesodermal derivatives such as the somatic and visceral musculature, the dorsal vessel and macrophages. In agreement with the proposed specificity for mesodermal derivatives, β3 tubulin expression is greatly reduced in the dorsalizing mutants dorsal, twist and snail.
Materials and methods
Generation and purification of isotype-specific antibodies
The 15 carboxyterminal amino acids were synthesized on an Applied Biosystems peptide synthesizer 430A using the FMOC-chemistry (Meierhofer et al. 1979) and purified by HPLC using gradients of acetonitrile/water containing 0·l% trifluoroacetic acid.
Peptides were covalently linked to Keyhole Limpet Hemocyanin (KLH, Calbiochem.) using l-ethyl-3-(3′-dimethylaminopropy)-carbodiimide as described by Tamura et al. (1983), using method 2. Rabbits were injected subcutaneously with these conjugates emulsified in complete Freund’s Adjuvant. After two boosts, serum was collected and affinity purified by passing over a peptide-Affigel 15 column as outlined by Caroil & Scott (1985). Collected fractions were tested by ELISA and Western blotting (see below).
Test for specificity of the antibodies
Enzyme-linked-Immunoabsorbent Assay (ELISA) was carried out essentially according to Engvall et al. (1971). Preimmunsera were used as negative controls.
For Western blotting, protein extracts prepared from staged Drosophila embryos were fractionated on a 11 % SDS-polyacrylamide gel and were transferred to nitrocellulose (Towbin et al. 1979). Western blots were first incubated with 0·l % Tween 20 in PBS (30 min) and then incubated in the same solution containing β3 antibody overnight at 4 °C. Blots were washed (10 min room temperature) three times in the Tween–PBS solution and then incubated with phosphatase-conjugated anti-rabbit antibody (DIANOVA, 1:2000). After repeated washes in PBS–Tween the colour was developed with bromo-chloro-indolyl-phosphate (Boehringer) and nitroblue tetrazoHum (Sigma) in substrate buffer (100mm-NaCl, 10mm-Tris-HCl, pH8 · 8, 5 mm-MgCl2).
Antibody staining of embryos
Embryos were dechorionated, permeabilized and fixed essentially as described by Freeman et al. (1986). After washing and blocking in BBT (0 · 15% crystalline BSA, 10mm-Tris-HCl, pH7 · 5, 50mm-NaCl, 40mm-MgCI2, 5mm-CaCl2, 20mm-glucose, 50mm-sucrose, 0 · l% Tween 20) they were treated with anti-β3 antibody overnight (1:50). The bound antibody was detected with a biotinylated secondary antibody and stained with Vectastain ABCkit (VectorLabs) using diaminobenzidine following the manufacturers’ instructions except that 30 μl of 1 % nickel sulphate were added per 600 μl staining mix (essentially according to Lawrence et al. 1987).
Individual embryos were mounted on slides in Epon and viewed under a Leitz microscope. All photographs (Kodak, Ektachrome 50) were taken under Nomarski optics.
Results
Generation of specific antibodies against β3 tubulin The amino acid sequences of Drosophila β tubulins have been deduced from the DNA sequence (Michiels et al. 1987; Rudolph et al. 1987) revealing 90–95 % homologies between the individual β tubulins. The β3 tubulin isotype differs markedly from other Drosophila β tubulins by an insertion of six amino acids in the TV-terminal region (Rudolph et al. 1987; Leiss & Renkawitz-Pohl, unpublished observation). However, the region most generally divergent between the β tubulins is the COOH-terminus (Table 1). Furthermore, it is known from pig β tubulins that the C-terminal domain is accessible to antibodies in the intact microtubules (Breitling & Little, 1986). Therefore, we decided to raise antibodies against synthetic peptides corresponding to the 15 extreme C-terminal amino acids of the β3 tubulin isotype (Table 1). The antibody was affinity purified (for details see Materials and methods) and tested for specificity. ELISA (data not shown) as well as Western blot experiments clearly showed specificity of the antibody for the β3 tubulin. Total protein extracts of early embryos (Fig. 1A), which contain maternally derived β1 tubulin and no β3 tubulin mRNA (Gasch et al. 1988), do not react with the antibody. In contrast, proteins from 6 – 10 h old embryos or pupae, which abundantly express β3 tubulin (Gasch et al. 1988), strongly react with the antibody (Fig. 1B,C). The reacting protein is 55 × 103Mr in size and comigrates with a tubulin (Fig. 1D). These data and the comparison of mRNA and protein in situ localization (see below) show that the antibody is specific for the β3 tubulin isotype.
At the end of gastrulation, β3 tubulin appears in single large mesodermal cells
Our previous in situ hybridization studies, the resolution of which was limited by the autoradiographic techniques used, revealed that the expression of the β3 tubulin gene during embryogenesis is limited to mesodermal cells (Gasch et al. 1988). The comparatively high resolution achieved by whole-mount staining of embryos with the β3 specific antibody enabled us to investigate in more detail the precise pattern and progression of β3 tubulin expression in early mesodermal cells and their derivatives.
We first detect the β3 tubulin at stage 10 in the mesoderm anterior to the cephalic fuuow (Fig. 2A) while the other cells of the mesodermal layer are still free of label. Soon after, the antigen is detectable in additional mesodermal cells, being most prominent in the gnathal segments (i.e. mandibular, maxillary and labial bud) as well as in the posterior region of the mesoderm (Fig. 2B). A lateral view of a slightly older embryo clearly shows staining in evenly spaced single cells (Fig. 2C). Although the β3 tubulin seems to be initially expressed synchronously in single large cells of every segment, there are subsequent distinct differences. The segments T1 and A8/A9 soon show several stained cells in comparison to the other segments. The β3-positive cells differ from other yet unlabelled mesodermal cells by their large size (Fig. 2C) and have characteristic extended processes, as can be seen by the β3 tubulin distribution. The cellular distribution of the β3 tubulin reveals a concentration at the cell periphery. We suggest that these cells, which become more abundant later, might represent the Drosophila homologue of the muscle pioneer cells as described for the grasshopper (Ho et al. 1983; Ball et al. 1985).
At stage 12, the splanchnopleura, giving rise to the visceral musculature, has separated from the somatopleura, the progenitor of the somatic musculature and some other mesodermal tissues (Campos-Ortega & Hartenstein, 1985). As is shown in Fig. 2D, this separation process can be followed by the β3 tubulin distribution rather well. The outer layer, the somatic mesoderm, is arranged in a segmental order (Fig. 2D). The number of β3 stained cells has increased considerably from stage 10, but the cells have not yet elongated. In this embryo, the differentiation state of the splanchnopleura can also be followed by the distribution of β3 tubulin, which again concentrates close to the periphery of the cells. In contrast to the somatopleura, the splanchnopleura is visible as a continuous epithelium of palisade-like cells (Fig. 2D).
During proceeding germband shortening (stage 12) further differentiation of myogenic cells occurs in the somatopleura (data not shown, for later stages see below). Three major groups of myoblasts are distinguishable, which will give rise to ventral, pleural and dorsal groups of muscles.
All mesodermal derivatives express β3 tubulin
In later embryonic stages, the β3 antibody allows one to follow the differentiation of the mesodermal derivatives rather well (Figs 3–5). By stage 14, the midgut is continuous along its entire length and flanked by the visceral mesoderm. In both the visceral and somatic mesoderm, strong expression of the β3 tubulin gene is evident by the RNA and the protein distribution (Fig. 4A and B). This comparison of RNA and protein distribution again confirms the specificity of our antibody. The separation between visceral and somatic muscle primordia is clearly visible (Fig. 4A and 4B). Myoblasts of every segment are stained by the antibody. Differentiation of the embryonic muscles by fusion of the myoblasts to syncytial cells starts at stage 13 and continues until stage 15. Differentiation of these muscle fibres gives rise to the final muscle pattern clearly recognizable at stage 16 (Fig. 3B). In a lateral view, the ventral, pleural and dorsal group of muscles are distinguishable as three different groups, the ventral muscles having not yet obtained their final fibrillar organization by this stage (compare Fig. 3C to 3B). The complete somatic musculature of Drosophila is characterized by β3 tubulin expression (stage 16 and 17, see Fig. 3B and 3C). It is clearly visible that the intersegmental apodemes do not contain detectable levels of β3 tubulin. Furthermore the cephalic musculature contains β3 tubulin, the parallel arrangement of the pharyngeal muscles being clearly visible (Fig. 3D).
Other cell types also derive from the mesoderm, including macrophages, which occur predominantly in regions where cell death occurs during embryogenesis (Campos-Ortega & Hartenstein, 1985). In a ventral view of a stage-15 embryo, β3 tubulin expression can be seen in distinct large cells, which we suggest to be macrophages (Fig. 5A). Again, the antigen concentrates close to the periphery of the cells and in the fibrillar extensions. For another mesodermal derivative, the fat body, staining could not be clearly visualized due to the strong staining of the muscles. At dorsal closure (stage 14), two rows of cardioblasts migrate to join midsagittally to form the dorsal vessel. These cardioblasts are also of mesodermal origin, and most express β3 tubulin as demonstrated in stage 16 when the dorsal vessel is formed (Fig. 5B). The pericardioblasts, which originate from the amnioserosa, are free of label. Four cardioblasts in register with the segments are stained and separated by two cardioblasts free of detectable β3 tubulin. In these rather large cells, the β3 tubulin is again concentrated at the periphery as observed for the presumptive muscle pioneer cells and the macrophages.
In the head region, we found two additional cell types to be labelled, neither of which are mesodermal in origin. These are a small number of cells of the clypeolabrum, presumably ectodermal in origin, and two prominent axons leading from anterior sense organs to the brain. As described below these structures are much clearer in mutants, lacking the majority of mesodermal cells, than they are in the wild type. The majority of the embryonic nervous system does not contain β3 tubulin.
In order to determine if the specificity of β3 tubulin for mesodermal derivatives is conserved between different Drosophila species, we analysed the β3 tubulin distribution in the distantly related species D. hydei. As in D. melanogaster, the antibody stains the mesodermal derivatives, for example the visceral, somatic and pharyngeal muscles (Fig. 5C). This not only indicates a comparable distribution between the two species, but also demonstrates clearly that the C-terminal domain of the β3 tubulin is conserved. Conservation has also been shown for the testis-specific β2 tubulins of D. melanogaster and D. hydei, here the amino acid sequence is completely identical (Michiels et al. 1987).
The β3 tubulin is present in abundance in mesodermal cells, and is characteristic of the differentiation of mesodermal derivatives during embryogenesis in D. melanogaster as well as in D. hydei. This may be indicative of a functional role of this tubulin isotype during myoblast migration and fusion.
Distribution of β3 tubulin in the dorsalizing mutants, dorsal, snail and twist
The dorsal-ventral pattern of the Drosophila embryo is determined by both maternally and zygotically expressed genes (Anderson, 1987; Anderson & Nüsslein-Volhard, 1986). Mutations of any one of these genes results in failure to form a ventral furrow (mesodermal invagination) during gastrulation. All cells continue to develop but the majority obtain epidermal fate (Campos-Ortega, 1983). We addressed the question of whether these mutations completely eliminate cells of the mesodermal pathway or if some mesodermal cells are present despite the inability to form the ventral furrow. Representative of the dorsalizing maternal effect mutants is the mutant dorsal 1 (for review see Anderson & Nüsslein-Volhard, 1986). From molecular studies, it is known that the dorsal (dl) gene product is homologous to the c rel protein, a vertebrate oncogene known to have a nuclear localization (Steward, 1987). Homozygous dl1/dl1 female flies lay eggs differentiating to an essentially hollow epidermal tube. Besides the posterior midgut and a small group of nerve cells in the head, no organs are built (Campos-Ortega, 1983). In addition to maternal effect genes, two zygotically expressed genes, twist (twi) and snail (sna), show a similar dorsalizing effect (Simpson, 1983; Nüsslein-Volhard et al. 1984). The ventral cells fail to invaginate in embryos homozygous for the mutations twist and snail (Simpson, 1983; Nüsslein-Volhard et al. 1984).
The analysis of the β3 tubulin distribution in the mutants dorsal, twist and snail, revealed the common feature of greatly reduced, but surprisingly not completely abolished, β3 tubulin expression. As is shown for snail as an example (Fig. 6F and 6G) all three mutants demonstrate two stained axonic cells in the head region, as was observed in the wildtype, and an additional staining of the posterior spiracles. A specific feature of the dorsal 1 mutants are some fibrillar aggregates visible at the anterior and posterior poles (Fig. 6A). The origin of these structures is not clear. Further β3 expressing cells are missing, a feature that distinguishes dorsal clearly from the zygotic-effect mutant genes twist and snail. Embryos homozygous for these latter mutations reveal β3 tubulin-stained structures organized in a segmental fashion (Fig. 6D and 6F). Higher magnifications show patches of spindle-shaped cells in a parallel arrangement (Fig. 6E for twist, Fig. 6H for snail). The origin of these cells remains to be clarified. From the analysis of the β3 tubulin distribution in dorsal, snail and twist mutants, we conclude that, with minor exceptions in neural cells, the β3 tubulin expression is specific for the mesodermal germ layer.
Discussion
We have used antibodies against the C-terminal domain of the β3 tubulin to follow the distribution of this antigen in the mesoderm and its derivatives.
The main results of our analysis concern the following topics: (1) existence of presumptive muscle pioneer cells in Drosophila, (2) the distribution of β3 tubulin in differentiating mesodermal tissues.
The β3 tubulin expression starts at stage 10 in the cephalic mesoderm, other mesodermal cells being free of staining. This indicates that the process of mesoderm formation during gastrulation is independent of this tubulin species. By the time segmentation is clearly visible, single cells in every metameric unit express β3 tubulin. Soon thereafter a group of mesodermal cells become stained. We suggest that these cells are homologous to the muscle pioneer cells described in the grasshopper embryo (Ho et al. 1983; Ball et al. 1985). In the grasshopper, the muscle pioneer cells erect a scaffold for developing muscles and for motomeurone growth cones. These muscle pioneer cells are distinguishable from their neighbouring mesodermal cells by their large size, a feature that is shared by the β3-positive cells. So far there is no information about the way in which these β3-positive cells, presumably muscle pioneers, are specified from the other mesodermal cells.
At early stage 12, the splanchnopleura, which give rise to the visceral musculature, is clearly distinguishable from the cells of the somatopleura, which give rise to the somatic musculature. At this stage the cells of the somatic mesoderm start myoblast differentiation, which is characterized by growth, fusion and migration, and which can be followed by β3 tubulin staining.
Another antigen has recently been localized in the mesoderm, the invertebrate integrin homologue PS2. PS2 is present in the somatic and visceral mesoderm, and is subsequently found to be concentrated in muscle attachment sites, while the muscles themselves are free of PS2 (Bogaert et al. 1987). In contrast, β3 tubulin is also expressed in fully differentiated visceral and somatic musculature, as well as cardioblasts and macrophages. The intracellular distribution of β3 tubulin at the periphery and in cellular extensions may suggest a functional role during cell migration and cell fusion.
In Drosophila, analysis of mutants provides a powerful tool to examine the determination of embryonic cells. Several mutants have been isolated that fail to form a ventral furrow during gastrulation and thus show no invagination of mesodermal cells (for review, see Anderson & Nüsslein-Volhard, 1986; Anderson, 1987). The genes concerned fall in two groups, maternal effect genes (e.g. dorsal) and genes transcribed during embryogenesis (twist and snail).
The snail and twist genes have been cloned, and the molecular analysis suggests a nuclear localization of the gene products in mesodermal cells (Boulay et al. 1987; Thisse et al. 1987b, 1988). The twist and snail gene products are present earlier than the β3 tubulin mRNA. Thus twist and snail are likely to be involved in a regulatory pathway leading to β3 tubulin expression, although the residual β3 tubulin we found in some nerve cells of these mutants suggests that the β3 tubulin gene can be expressed in non-mesodermal derivatives by a pathway independent of twist and snail.
Little is known concerning the differentiation of muscles in Drosophila, but few approaches have been taken previously. For example, the homeotic gene Ultrabithorax specifies the types of muscles developing in the thoracic segments of the larvae (Hooper, 1986) and for adult muscles it is proposed that there is a dependence on neural cells for the type of muscle developing (Lawrence & Johnston, 1986). Further analysis of β3 tubulin expression will enable us to investigate the effect of mutants on the differentiating mesodermal derivatives.
ACKNOWLEDGEMENTS
We highly thank Dr Ch. Nüsslein-Volhard and Dr J. Campos-Ortega as well as Dr G. Technau for stimulating discussions and providing us with mutant fly strains. We greatly acknowledge Dr Campos-Ortega and Dr M. Cross for critical discussion of the manuscript. We highly acknowledge A. Mausolf for advice in raising antibodies and Dr H. Saumweber for the generous gift of the α tubulin antibody. We thank B. Keβ1er and U. Otto for their excellent technical assistance. We are indebted to Dr H. Hollander for making his microscope facilities available to us. This work was supported by grants from the Deutsche Forschungsgemeinschaft (DFG Re628/2-2) and from the Bundesministerium for Research and Technology to R. Renkawitz-Pohl.