Microtubule distribution reveals superficial metameric patterns in the early Drosophila embryo

The body of insects and other arthropods, and annelids, is spatially divided into a number of repeating units known as segments. In the Drosophila embryo, segment primordia become externally visible during stage 11, when transverse grooves divide the main part of the germ band into 14 uniform regions (Poulson, 1950; Turner & Mahowald, 1977). However, this is not the first visible manifestation of metamery. Early in stage 9, before the ectoderm is visibly segmented, the mesodermal cells become arranged in periodic bulges (Campos-Ortega & Hartenstein, 1985) and segmentation of the nervous system primordia is apparent (Brown & Shubiger, 1981; Hartenstein & Campos-Ortega, 1984). In embryos at stage 5, no morphological signs of segmentation can be seen, although cell lineage studies (Lawrence, 1981; Lohs-Schardin et al. 1979; Niisslein-Volhard & Wieschaus, 1980; Hartenstein et al. 1985) and transcription patterns for a number of genes affecting segmentation (for review Scott & O’Farrell, 1986; Akam, 1987; Nüsslein-Volhard et al. 1987; Scott & Carroll, 1987; Howard, 1988; Lehmann, 1988) have revealed that the blastoderm cells are spatially determined with respect to the segmental identity. At this stage, scanning electron microscope investigations have not revealed repeating surface domains (Turner & Mahowald, 1977).

The aim of this study is to examine the embryo surface from the formation of the cellular blastoderm to the completion of gastrulation in order to visualize three-dimensional changes eventually occurring in a repeating pattern. For this purpose, careful scanning electron microscope observations, undertaken with accurate tilting of the specimens to increase surface contrast, were combined with immunofluorescence staining to visualize microfilament and microtubule distributions. These cytoskeletal elements constitute two networks, apically and subapically located with respect to the embryo surface (Warn & Magrath, 1983; Warn, 1986; Karr & Alberts, 1986; Warn & Warn, 1986) and provide a good tool to visualize spatial irregularities in cell disposition.

Embryos of Drosophila melanogaster (Oregon-R strain), staged according to Campos-Ortega & Hartenstein (1985), were collected at 25°, rinsed in standard phosphate-buffered saline (PBS) or Tris-buffered saline (TBS), dechorionated in a 50% commercial bleach solution and washed with distilled water.

For F-actin identification, the dechorionated eggs were permeabilized with heptane for 2-3 min and fixed in 4% freshly prepared paraformaldehyde in PBS. Twenty minutes later the eggs were transferred for 10 min to the same solution containing 0-1% Triton X-100. The vitelline envelope was removed with fine needles. Selected stages were stained for 30 min with ljugml^-1 phalloidin-labelled with rhodamine (Molecular Probes Inc.). After rinsing in PBS the embryos were mounted in 90% glycerol containing 2-5% n-propyl gallate to reduce photobleaching (Giloh & Sedat, 1982).

For indirect immunofluorescence staining against tubulin, the dechorionated embryos were fixed and their vitelline envelope was removed essentially as described by Warn & Warn (1986), except for a final fixation with acetone for 5 min. The devitellinized embryos were then washed extensively in TBS, incubated Ih in TBS containing 10mgml^-1 bovine serum albumin, washed again in TBS and incubated for 30 min with a monoclonal antibody against α-tubulin (Amersham). After rinsing in TBS, the embryos were incubated for 30 min with fluorescein isothiocyanate-conjugated rabbit antimouse 1gG serum (Miles). The eggs were then washed in TBS and mounted in glycerol containing n-propyl gallate. To ensure the validity of the results some embryos were incubated with the second antibody; in no case was the fluorescence significant. Sometimes groups of blastoderm cells were dissected with fine needles to simplify the interpretation of microtubule distribution.

Fluorescence observations were carried out with a standard Leitz microscope equipped with filters for fluorescein and rhodamine. Photomicrographs were taken with Kodak Tri-X Pan film.

For scanning electron microscopy, the dechorionated embryos were treated with a solution containing 25 % glutaraldehyde plus 5% acrolein in 0·lM-sodium cacodylate buffer pH 7·2 in an equal volume of heptane (Zalokar & Erk, 1977). After 3 min the embryos were transferred to the solution of Kalt & Tandler (1971) for 2h and the vitelline membrane was removed with tungsten needles. After rinsing in cacodylate buffer, the embryos were postfixed in osmium tetroxide for 2h and dehydrated in a graded series of alcohols. The embryos were then dried by the critical-point method and observed with a Philips SEM 505 at 30 kV.

Indirect immunofluorescence with antibody against a- tubulin and rhodamine-labelled phalloidin for F-actin shows that, during stage 5, microtubules and microfilaments form two networks, composed of hexagonally arranged meshes and ring-shaped structures, respectively. The actin meshes are thinner than the tubulin ones. F-actin is detected below the lateral cell membrane, from the apical to the basal region, and underlies the periphery of the blastoderm cells (Fig. 1A,B). Microtubule distribution is quite complex. Sagittal images show that two groups of microtubules radiate from two widely separated centrosomes located between the nucleus and the plasma membrane (Fig. 1C). The first group is composed of short microtubules crossing the subapical region of the cells, and the second consists of very long microtubules, arranged parallel to the longitudinal axis of the nucleus and forming a cylindrical envelope around the nucleus itself (Fullilove & Jacobson, 1971; Karr & Alberts, 1986; Warn & Warn, 1986; Callaini & Anselmi, 1988). In superficial observations, microtubules are therefore detected at two distinct levels of focus. Images focused at the embryo surface show many pairs of distinct centrosomes (Fig. ID), whereas slightly deeper observations reveal fluorescent ring-shaped structures, corresponding to transverse sections of the microtubular cylinders enveloping the nuclei (Fig. IE).

Immunofluorescence and scanning electron microscope did not reveal surface irregularities during stage 5, but sagittal views of the dorsal region of embryos stained with antibodies against α-tubulin during early stage 6 showed that the centrosome pairs are placed at different levels, despite the linearity of the embryo surface, and form an undulating dotted line (Fig. IF). This discontinuity is clearly visible in surface observations which show irregular areas where the fluorescence is weaker. At this time, the dorsal region of the embryos stained with antibodies against α-tubulin is seen to be divided into transverse bands of unequal size (Fig. 1G). This pattern is not equally clear in rhodamine-labelled embryos and is not evident by scanning electron microscope. A ventral view of the same embryo at level focus near the nuclei shows that the fluorescence of the microtubular network is not uniform as it was during stage 5. The embryo surface posterior to the cephalic fold is divided into 14 equal and evenly spaced bright fluorescent stripes, separated by as many areas of fainter fluorescence (Fig. 1H). Simultaneous lower power observations with scanning electron microscopy did not reveal a similar surface pattern (Fig. II). The pattern of dark and bright areas is also visible in the lateral part of the embryo (Fig. 2A). Details of the ventral region of the embryo lining the longitudinal furrow show that the fluorescent rings seem to be disposed at different levels because of a pattern of surface bulges (Fig. 2B). This suggests that the stripes may be related to the different relative positions of the microtubular baskets enveloping the nuclei. The microtubule arrays pass in and out of the plane of focus, determining the variations in labelling. However, SEM observations do not reveal surface bulges (Fig. II) and the disposition of the ‘zonulae adhaerentes’, as shown by rhodamine-labelled phalloidin, was almost uniform (Fig. 2G). The zebra-like pattern of tubulin distribution is unchanged during late stage 6 when the cell plate carrying pole cells reaches the dorsal region of the embryo (Fig. 2C) but the alternating pattern disappears from the embryo surface during stage 7, when amnio-proctodeal invagination occurs. However, at this moment, an unclear zebra-like fluorescence is still observed for a short time in the mesodermal region (not shown). In embryos exposed to low temperature for 60min, most of the microtubules constituting the basket-like structures depolymerize. When the embryos were recovered at 25°C, the microtubules form again and the nuclei become irregularly arranged below the plasma membrane (unpublished results). When the embryos were cold treated at the beginning of the stage 6 and recovered for 30 min at 25°C, germ band extension occurred, but the transverse stripes were not as obvious as they were in untreated embryos (Fig. 2D).

SEM images were unable to reveal surface modifications during late stage 6 (Fig. 2E), but during stage 7 evident irregularities of the embryo surface were observed. At this stage, SEM observations carried out in the ventral region of the embryo revealed seven groups of more tightly packed cells separated by areas of less tightly packed cells (Fig. 2F). The smooth appearance of these areas may mean that they were flattened at the time of fixing because of their proximity to the vitelline envelope.

Rhodamine-labelled embryos at the early stage 6 did not show any alternating pattern resembling that observed with the tubulin stain (Fig. 2G). The irregularities become clearly visible in the lateral regions of slightly older embryos, during late stage 6, and are visible as dark transverse stripes, 2–3 cells wide, interposed with larger areas of brighter fluorescence (Fig. 2H). The transverse bands where the fluorescence is fainter, one anterior and seven posterior to the cephalic furrow, may correspond to the areas of more tightly packed cells observed in slightly older embryos in SEM images. These cells are at lower level than the neighboring ones. As in the case of the tubulin zebralike pattern, the alternate stripes of the actin network disappear in late stage 7 embryos (not shown).

Despite the uniform appearance of the embryo surface at stage 5, the blastoderm cells express selected genes involved in regulating embryogenesis (for review see Scott & O’Farrell, 1986; Akam, 1987; Nüsslein-Volhard et al. 1987; Howard, 1988). Because scanning electron microscope observations are unable to reveal significant surface modifications at this time (Turner & Mahowald, 1977; present observations), the distribution of the two major cytoskeletal proteins, actin and tubulin, was examined to obtain new morphological information about the spatial organization of the blastoderm cells. Rhodamine-labelled phalloidin provides a convenient way of visualizing the actin network below the plasma membrane of the blastoderm cells, and fluorescent conjugated antibodies against α-tubulin make it possible to observe the microtubular network closely related to the blastodermic nuclei. In this way, the network visible using phalloidin may be compared with surface scanning observations, and the microtubular network provides a useful tool to examine nuclear position. This strategy for identifying irregularities in cellular disposition nevertheless has several limitations: the impossibility of observing actin distribution in embryos fixed by the methanol-acetone procedure and the doubt that the embryo surface, as observed by scanning electron microscopy, might be a consequence of the critical point drying. To verify the real or artifactual nature of the data obtained, several control investigations were attempted. The outlines of the blastoderm cells were visualized with Con A both in methanol-acetone- and paraformaldehyde-treated embryos, but nothing significant was found (data not shown). Scanning electron microscope observation of the methanol-acetone-treated embryos did not reveal any obvious pattern comparable to that observed with tubulin fluorescence and the embryo surface was uniform in appearance (data not shown). Tubulin staining in paraformaldehyde-treated embryos revealed a pattern of fluorescence similar to that observed in methanol-treated embryos (data not shown).

The major finding of this study is that the fluorescent tubulin network is not uniform during early stage 6 as it was at the stage 5, but shows several discontinuities that repeat in a metameric manner. The alternating pattern of fluorescence is a surprising feature and does not seem to have any direct morphological correspondence, but sagittal observation of the embryo surface shows that the centrosomes form an undulating dotted line. As the microtubules constituting the cylindrical envelope around the nucleus radiate from these centrosomes, superficial observation of the resulting undulating network shows that it consists of areas of different brightness. An intriguing question, which has yet to be resolved, is whether the undulating disposition of the tubulin network is dependent on the nucleus via centrosomes, or whether nuclear position is related to the cellular cytoskeleton.

To eliminate the possibility that stripes are apparent because of a pattern of surface bulges, the embryos were processed for SEM observations and treated with rhodamine-labelled phalloidin to visualize the actin network just below the embryo surface. The pattern of F-actin distribution observed during early stage 6, does not seem to relate to the data obtained with anti-tubulin antibodies, and the transverse areas observed during late stage 6 may be related to the superficial movements occurring during the unfolding of the posterior midgut rudiment. SEM observations showed that a secondary superficial folding is not responsible for the pattern of fluorescence observed with anti-tubulin antibodies, but the possibility that the forward movement of the posterior midgut rudiment causes extensive rearrangement of the periplasm cannot be excluded. To verify this hypothesis several embryos were exposed to low temperature at the beginning of stage 6. Cold treatment results in the partial depolymerization of the microtubular baskets, which, after recovery, form again and become irregular in size. In this way, the height of the nuclei from the embryo surface becomes modified and immunofluorescence microscopy does not reveal the same metameric pattern observed in untreated embryos, but only a few irregularly sized stripes. The transverse stripes therefore appear to be closely related to the integrity of the tubulin network. However, even assuming that the regularly repeated stripes are the consequence of germ band shifting, it is an intriguing question why and how the forces involved in this process determine such a metameric pattern.

The formation of segmental grooves and boundaries is the most evident event of segmentation, but does not seem to be the fundamental step. Genetic analysis has shown that each segment is composed of an anterior compartment and a posterior compartment (Garcia-Bellido et al. 1976) and that the distinction between these groups of cells occurs at the stage 5 (Wieschaus & Gehring, 1976). At this stage, the embryo is divided into 14 domains called parasegments, each consisting of the posterior compartment of one segment and the anterior compartment of the adjacent posterior segment (Martinez-Arias & Lawrence, 1985; Lawrence, 1988).

These studies have been confirmed by the expression pattern of the segment polarity genes engrailed (Di Nardo et al. 1985,1988; Fjose et al. 1985; Kornberg et al. 1985; Ingham et al. 1985b; Weir & Kornberg, 1985), gooseberry (Bopp et al. 1986) and wingless (Baker, 1987) and of the pair-rule genes fushi tarazu (Hafen et al. 1984; Carroll & Scott, 1985; Hiromi et al. 1985; Weir et al. 1985), hairy (Ingham et al. 1985a, Ingham & Gergen, 1988), paired (Kilcherr et al. 1986), and even skipped (Macdonald et al. 1986; Harding et al. 1986; Frash et al. 1987) but, despite the fundamental importance of compartmentalization in the development of the embryo (see Brower, 1985), morphological evidence of this phenomenon has until now been sought during stage 6. The boundary between the anterior and posterior compartments is not characterized by any discernible morphological discontinuity at this stage. The spatial alterations in the tubulin network expressed in the 14 alternating pairs of darker and brighter transverse areas may constitute the first morphological evidence of these molecular and genetic events. The zebra-striped pattern observed with antibodies against α-tubulin during early stage 6 recalls the alternating pattern of engrailed transcripts (Weir & Kornberg, 1985; Howard & Ingham, 1986) and the localization of the nuclear engrailed protein (DiNardo et al. 1985). In this way, according to the findings of Ingham et al. (1985b) and Lawrence et al. (1987), which use the engrailed gene products as a marker for the posterior compartment, and by comparing the tubulin network with Fig. 4 of DiNardo et al. (1985), the bright areas observed in fluorescent images may be considered to correspond to the posterior compartment and the dark areas to the anterior one. However, apart from the spatial correspondence between the tubulin and engrailed patterns, there is no evidence of such a close relationship. Although there is no evidence to support this correspondence, or any relationship between the tubulin network, and engrailed distribution, it is surprising that similar spatial organization is observed simultaneously in very different situations. The reason and the significance of these periodic surface irregularities in fluorescence images is unknown, but, because of their constant appearance at the beginning of stage 6, it might be thought that they are the external manifestation of an important phenomenon occurring at the molecular level.

The hypothesis that the transient tubulin pattern might be related to the first mechanism involved in embryonic pattern formation is an attractive, but yet unverified possibility, and only similar investigations on segment polarity and pair-rule mutants will answer the question. Furthermore, the microtubules that organize the nuclear baskets during cellularization appear very important in localizing the fushi tarazu RNA to the proper sites of synthesis (Edgar et al. 1987).

I thank R. M. Warn for advice and critical reading of the manuscript. This work was support by a 40% grant from Ministero Pubblica Istruzione.