Localization of transcripts from the wingless gene in whole Drosophila embryos

Insects are a classic example of metameric metazoans whose bodies comprise a series of homologous segments. In Drosophila the overt segmentation of the trunk region of the embryo depends on the action of segmentation genes. Mutations in these genes alter the number or polarity of segments formed (Nüsslein-Volhard & Wieschaus, 1980). A hierarchy of segmentation genes divides the central portion of the embryo into successively smaller units, culminating in the expression of some genes in a specific part of every segment (reviewed in Akam, 1987). One such gene expressed in every segment is wingless (wg), which is transcribed in the posteriormost cells of each parasegment (Baker, 1987). The wg gene is thought to control the proper organization of each segment through an intercellular signalling mechanism (Nüsslein-Volhard & Wieschaus, 1980; Baker, 1987, 1988a; Rijsewijk et al. 1987; Cabrera et al. 1987; Martinez-Arias et al. 1988).

Segmentation is best understood in the trunk region, where fourteen segments are recognized between the mandibular and the eighth abdominal segments. The head and caudal regions are also segmented, but their organization is not clear in Drosophila embryos (Poulson, 1950; Turner & Mahowald, 1979; Campos-Ortega & Hartenstein, 1985; Martinez-Arias & Lawrence, 1985). However, comparative studies have suggested that all insects have a similar underlying body plan (reviewed in Anderson, 1973). On this basis the head anterior of the mandibular segment is expected to contain intercalary, antennal and labral segments; the most anterior part of the embryo invaginates to form the foregut. Posterior to A8 the number of further abdominal segments varies in different insects, the maximum number attained apparently being twelve. The variation is thought to be due to the loss or fusion of various of these caudal segments in different groups (reviewed in Matsuda, 1976). Evolutionarily, the head segments are probably most ancient. The trunk segments could have evolved by the serial reduplication of a pre-existing unit. This hypothesis is consistent with the molecular mechanisms involved in formation of the trunk segments, which do not seem to apply to the head (reviewed in Akam, 1987; Scott & Carroll, 1987). Thus the trunk segments are only a part of insect segmentation. If head segmentation is more ancient, then it may be more pertinent to understanding the origins of segmentation and the relationship of insect segmentation to that of other metameric organisms. For example, it appears that genes involved in head development may have closer counterparts in vertebrates than other segmentation and homeotic genes (Regulski et al. 1987; Joyner & Martin, 1987; Rijsewijk et al. 1987).

Since segmentation gene expression precedes morphological segmentation, it can be useful in mapping segmental primordia during early development (e.g. Hafen et al. 1984; Kornberg et al. 1985; DiNardo et al. 1985; Ingham et al. 1985a,b; Carroll & Scott, 1985; Martinez-Arias & Lawrence, 1985). This paper further describes the accumulation of transcripts from the wg gene during embryogenesis. Previous descriptions of wg expression have been based on detection of transcripts in tissue sections (Baker, 1987; Rijsewijk et al. 1987; Ingham et al. 1988). These studies showed that wg is also transcribed outside the trunk, in the head and caudal regions. The striped pattern of transcription of the trunk region is clearly revealed by sagittal and parasagittal sections, but patterns in the dorsoventral axis and in the head are more difficult to reconstruct. This information can be obtained more readily by in situ hybridization to whole-mounted embryos. In this paper, an improved procedure has been used to detect wg transcription in whole embryos, leading to a more detailed description of wg transcription outside the trunk region. This is discussed with respect to the putative segmentation of these regions. In addition, it is reported that the pattern of wg transcription does not remain constant throughout embryogenesis. Midway through the extended germband stage, marked changes in wg expression occur in the dorsoventral axis of all segments.

Parts of existing hybridization protocols (Cox et al. 1984; Kornberg et al. 1985; Ingham et al. 1985a) were combined in the detection of transcripts in whole embryos. Embryos were dechorionated with generic bleach (1:1 with H₂O), washed in H₂O, and prefixed [three parts 4 % paraformaldehyde in PBS (130mm-NaCl, 10 mm-sodium phosphate pH 6·8), freshly made, 1 part n-heptane] for 15 min over ice with frequent agitation, then removed from the vitelline membrane as previously described (Karr & Alberts, 1986). Postfixation and rehydration were accomplished by 5 min washes in each of the following: (1) 90% aqueous methanol, 50mm-EGTA pH 7·0; (2) 7:3 MeOH/ EGTA:PBS + 4% paraformaldehyde; (3) 5:5 MeOH/ EGTA: PBS + 4 % paraformaldehyde; (4) 3:7 MeOH/EGTA: PBS + 4 % paraformaldehyde; (5) PBS +4 % paraformaldehyde; (6) PBS. Embryos were either hybridized directly or dehydrated through alcohol to 70% ethanol, 30 % PBS and stored at −20°C, but storage for more than a few days was found to be detrimental. Before hybridizing, embryos were transferred from PBS to 0·2 M-HCI (20 min), washed five times in PBS and digested with pronase (0·15 mg ml^-1 in 50mm-Tris pH7·5, 5mm-EDTA, for 7·5 min at room temperature). Proteolysis was stopped with glycine (added to 2 mg ml^-1, then washed in 2 mg ml^-1 glycine in Tris/EDTA) and embryos washed five times in PBS. Hybridization was overnight at 53°C in the buffer described by Ingham et al. (1985a), using ³⁵S-labelled antisense-RNA probes derived from cDNA sequences using the T3 promoter ofpwg-cl4a (Baker, 1987) or the T7 promoter of T7(2)en (a gift of Dr M. Weir). Probes were reduced to a mean (mass average) length of 100-200 bp using base hydrolysis (Cox et al. 1984). Embryos were washed in ten changes of a buffer comprising 50 % formamide, 0·06m-NaCl, 5mm-Tris pH 7·4, 5mm-sodium phosphate pH6·8, 5mm-EDTA, 1 xDenhardt’s Solution, 14 mm-β-mercaptoethanol at 53°C over a 24 h period, then washed once in NTE (NTE = 0·5m-NaCl, 10mm-Tris pH8·0,1 mm-EDTA), incubated with RNase A (20 μg ml^-1 in NTE, 30 min, 37°C) and washed twice with N IE at 37°C and twice with PBS at 45 °C. In the experiment shown in Fig. 1A, embryos were then dried onto slides and autoradiographed as described by Kornberg et al. (1985) for imaginal discs. For the other experiments, this procedure was replaced with the following: an approximately equal volume of molten Kodak NTB2 emulsion was added to the final PBS wash and the embryos incubated in this mixture for 20 min with occasional shaking. The emulsion/embryo mixture was spread dropwise on gelatin-subbed slides and allowed to air-dry. After 21 days exposure at 4°C, slides were developed in Kodak D19 (2min 15s, 15°C), stopped in 2 % acetic acid (30 s, 15°C) and fixed in Kodak fixer (1h), washed extensively in H₂O, air-dried and viewed by Nomarski-interference microscopy.

The sectioned material has been described previously (Baker, 1987), except for Fig. 3D, which was provided by Dr P. Ingham.

Although most in situ hybridization studies have used tissue sections, Kornberg et al. (1985) described the use of in situ hybridization to detect transcripts from the engrailed (en) gene in whole imaginal discs. In an initial experiment, a wg probe was hybridized to whole embryos following a similar protocol (see Materials and methods), and others have also described similar experiments (St. Johnston & Gelbart, 1987; Baumgartner et al. 1987). An example is illustrated in Fig. 1A; autoradiography revealed the striped pattern of wg transcription, but the morphology was poor and uneven coating with emulsion made the signal irreproducible. Both these problems were offset by a straightforward modification of the procedure. Hybridized embryos were transferred directly into liquid emulsion after washing, and the emulsion/embryo mixture allowed to air dry onto microscope slides. This results in more even-coating with emulsion than if specimens are first dried onto slides before dipping, and morphology is improved when the ethanol dehydration prior to drying is eliminated (Fig. 1). A series of embryos hybridized following the new method is shown in Fig. 2. It is important to mention shortcomings and sources of artefact that remain in the technique. (1) Signal is frequently stronger around the edge of the specimen and weaker over the highest point. This bias in sensitivity could be due to the thickness of the emulsion coat, which is probably thinner over the top of the embryo. This artefact is circumvented by examining embryos mounted in different orientations. (2) Silver grains are usually present where there are deep grooves on the surface of the specimen, such as the cephalic furrow or segment boundaries. This could be caused by incomplete washing of probe from these grooves. In the present experiments, this has not caused difficulties; rather, the interpretation of hybridization signals is made easier by this highlighting of morphological features. (3) Both wg and en transcripts are localized predominantly in external parts of the embryo. Although wg transcripts are detected internally in the anal region and hindgut (Fig. 2K), this signal is weaker than might be expected on the basis of sectioned material (see Fig. 3). Therefore, it is not certain how readily transcripts in internal tissues will be detected by this method.

Previous studies of wg transcription have concentrated on the trunk region of the embryo (Baker, 1987; Rijsewijk et al. 1987; Ingham et al. 1988). In combination with sectioned material (Fig. 3), the wholemounts shown in Fig. 2 have provided a more complete picture of wg transcription in the whole embryo. Details are described below, and summarized in Fig. 4.

wg transcripts in parasegments 0–13 are first detected during the cellular blastoderm stage. The stripes appear first at the anterior, and in even-numbered parasegments before odd. At this stage, signal seems stronger ventrally than dorsally (see Fig. 2C,D). Expression of the segmentation genes engrailed and gooseberry also develops with pair-rule modulation in the trunk region (Weir & Kornberg, 1985; DiNardo et al. 1985; Coté et al. 1987; Baumgartner et al. 1987). However, wg is unique amongst these in that the earliest expression is outside the trunk region. First to appear are transcripts at the anterior tip of the embryo and in a posterior ring, which occupy, respectively, the primordia of the foregut and of the hindgut and anal region. The next step is the appearance of blocks of wg transcription on either side of the dorsal head (Fig. 2C), then followed by the stripes of the trunk region (Baker, 1987; Ingham et al. 1988).

Thus when gastrulation begins regions of wg transcription are spread over most of the embryo, except for the endodermal regions (the primordia of the anterior and posterior midguts) and the germline (the pole cells).

After gastrulation, the movements of germband extension occur and the segmentation of the trunk region becomes morphologically apparent (Poulson, 1950; Campos-Ortega & Hartenstein, 1985; Martinez-Arias & Lawrence, 1985). During this period the distribution of wg transcripts continues to elaborate. By the end of the rapid phase of germband extension, two small patches of wg transcription have appeared on either side of the dorsal head, behind the foregut signal and anterior to the large dorsal patches (Fig. 2E,I). As the stomodeum invaginates and cells in the most anterior wg domain move into the foregut, these two patches fuse at the dorsal midline and move anteriorly to occupy the most dorsal part of the clypeolabral lobe. In some embryos at this stage, another two small patches of wg transcription are weakly detected on the ventral lateral head, just above parasegment 0 (Fig. 2J). These could correspond to the −1 signal previously observed in sections of the extended germband stage (Baker, 1987). This transcription is also sometimes detectable at blastoderm (Fig. 2D).

During germband extension, the pattern of wg transcription also becomes further resolved in the caudal region behind parasegment 13. wg tanscripts in parasegment 14 are first detected at the end of the rapid phase of germband extension (stage 9, 4h after fertilization according to Campos-Ortega & Hartenstein, 1985; Fig. 3A,B). Posterior to this, signal extends over the whole of the proctodeum and hindgut, as in the blastoderm. By the time the stomodeum invaginates (stage 10; ∽5 h after fertilization; Fig. 3C) the anus and the hindgut have resolved into separate wg domains. The signal in the hindgut is at the boundary with the posterior midgut and is later found in the malpighian tubules, which are the most posterior ectodermal structures (Fig. 3G,H; Lawrence & Johnston, 1986).

Midway through the extended germband stage, when the gnathal lobes have appeared (stage 11, ⁠), it is apparent that changes in wg expression have occurred in the dorsal-ventral axis. Most strikingly, wg transcripts become excluded from the lateral part of the epidermis of the thoracic and abdominal segments and are restricted to the ventral and dorsal parts of the original stripes (Fig. 2F,G). In addition, the shape of this remaining wg domain is altered. Midventrally the region containing wg transcripts is narrower than previously, but the lateral boundary of the domain is broader, so that the stripes have a dumbbell appearance. This is more pronounced in the anterior trunk segments (Fig. 2K). This change in wg transcription during the extended germband stage is in contrast to the constancy of the pattern of en expression in the trunk region, which seems to label the epidermal cells of posterior compartments constantly throughout development (Kornberg et al. 1985; Fjose et al. 1985; DiNardo et al. 1985; compare Fig. 2B with Fig. 2F).

In the labial and maxillary segments, transcripts are now restricted to a strip across the lobes themselves and do not extend to the ventral midline. The −1 stripe moves close to the transcripts in the mandibular segment and can no longer be found when germband shortening begins. The dorsal patch becomes less extensive and, as stomodeal invagination continues, signal in the clypeolabral lobe moves more ventrally. At the caudal end of the embryo, wg transcripts are no longer distinguishable in parasegment 14 when the germband starts to retract, but transcripts are still found in the anal region and the hindgut even after the germband has shortened (Figs 2F, 3E,F,I). The wg transcription pattern now seems to remain constant during germband shortening and dorsal closure (Fig. 2H).

The wg transcription pattern is summarized in Fig. 4.

Detection of transcripts in whole Drosophila embryos is a useful complement to results obtained with sectioned material. Although whole mounts offer less resolution on a cell-by-cell basis, the whole pattern is better revealed in a single specimen. For example, the two small dorsal patches of wg transcription (Fig. 21) had previously been missed in sections. The method should be useful for comparing the accumulation of transcripts with the distribution of protein for gene products against which antibodies have been raised and analysing the expression of genes for which antibodies are not available. The technique is not limited to Drosophila embryos and has also been used for imaginal discs (Baker, 1988b, and unpublished data).

In the trunk region, the wg gene is required for the development of every segment and wg transcripts accumulate in a specific portion of each segment of the extended-germband-stage embryo (Nüsslein-Volhard & Wieschaus, 1980; Baker, 1987). This suggests that wg expression can be used as a molecular probe for segmentation, as has been the case with other segmentation genes (e.g. Kornberg et al. 1985; Ingham et al. 1985a,b;,Carroll & Scott, 1986). In situ hybridization has shown that by the end of germband extension all the embryo excepting the endoderm and germline (which are not thought to be segmented) is divided by 21 domains of wg transcription. These may identify segments in the head and caudal regions, as well as the trunk. Comparative studies suggest the head should contain intercalary, antennal and labral segments anterior to the three gnathal segments (Anderson, 1973; Rempel, 1975); the caudal region extends up to a 12th abdominal segment in some insects, but this number is often reduced by the loss of some and fusion of others (Matsuda, 1976). Below, these predictions are discussed in light of the wg transcription pattern (summarized in Fig. 4) and also with respect to the engrailed and gooseberry genes, which seem to have related patterns of expression. The picture that emerges, though not definitive, is a contribution towards understanding the terminalia of the Drosophila embryo.

The intercalary segment is thought to occupy a small ventrolateral region anterior to the mandibular segment (Schoeller, 1964; Turner & Mahowald, 1979; Jürgens et al. 1986). Transcripts from wg (Fig. 2D,J) and from en and gsb (Kornberg et al. 1985; DiNardo et al. 1985; Ingham et al. 1985b; Baumgartner et al. 1987) probably define parts of this segment (Fig. 4).

Fate-mapping studies suggest a dorsolateral position for the embryonic antennal segment (Anderson, 1973; Struhl, 1981, 1983; Jürgens, 1985; Jürgens et al. 1986). This is consistent with the position of a dorsolateral arm of wg transcription (Fig. 4). en and BSH9 are expressed in a similar region (DiNardo et al. 1985; Baumgartner et al. 1987). However, the putative antennal wg domain is larger than in other segments and also extends over much of the procephalic lobe. In Drosophila, the procephalic lobe, most of which contributes to the brain, has been described as unsegmented (Poulson, 1950), but it seems possible it actually belongs to a large antennal segment. Consistent with this, gynandromorph analysis showed the eye-antennal imaginal disc is recruited from a larger region of the embryo than are discs in other segments (Struhl, 1981). This region is anomalous in another respect, since it contains a further small patch of dorsolateral cells that express the en and gsb genes (DiNardo et al. 1985; Baumgartner et al. 1987).

The small paired dorsal regions of wg transcription correspond to the fate-map position of the labral segment (Jürgens, 1985; Jürgens et al. 1986). During germband extension, these two primordia join together and move anteriorly around the front of the embryo as the stomodeum invaginates until, by germband shortening, labral wg expression is on the ventral surface of the clypeolabral lobe (Technau & Campos-Ortega, 1985). Corresponding labral expression of gooseberry has been described (Baumgartner et al. 1987), but en expression seen in the clypeolabral lobe is more anterior and cannot lie within the labral segment (Kornberg et al. 1985; DiNardo et al. 1985; Ingham et al. 1985b; Figs 2B, 3D). A better candidate for labral en expression has now been detected in the dorsal head using a monoclonal antibody (T. Kornberg, personal communication).

The most-anterior domain of wg expression is in the primordium of the foregut. The foregut is part of the ectoderm and, like typical segments, contains imaginal primordia which give rise to part of the adult at metamorphosis (Anderson, 1972). The foregut could be considered the most anterior segmented part of the embryo. Possibly, en expression in the mostventral part of the clypeolabral lobe (see above) corresponds to this metamere.

Posterior to parasegment 14, both the anal region and the hindgut contain one domain each of wg and en expression at the extended germband stage (Fig. 3C,D). The straightforward interpretation is that there is one abdominal parasegment posterior to psl4, and that the hindgut should be considered a further metameric unit. Each of these also forms a pair of imaginal precursors which construct parts of the adult at metamorphosis (Anderson, 1972). However, the anal region seems to be a composite unit derived from a larger number of ancestral metameres (Matsuda, 1976; Dubendorfer & Nöthiger, 1982; Carroll & Scott, 1985). Vestiges of the primitive condition are seen in the larval cuticular structures (Jürgens, 1987) and in the distribution of BSH9 transcripts from the gsb locus (Baumgartner et al. 1987). During germband shortening, the anal region also seems to assimilate parasegment 14 (Figs 2F, 3E,F, 4; DiNardo et al. 1985).

In contrast to previous conclusions, whole mounts in situ show that the pattern of wg transcription does not remain uniform in the dorsal-ventral axis. Although, until the middle of the extended germband stage, wg transcription, like en, identifies cells at a particular anterior-posterior position in the segment, wg transcripts are subsequently lost from the lateral epidermis of the trunk segments and the shape of the remaining wg domain altered. This result raises doubts about the idea that segment polarity genes act only to label domains defined as rows of cells at the blastoderm stage. Apparently expression of the BHS9 transcript from gooseberry becomes modified similarly (Baumgartner et al. 1987).

Based on the timing and position of dorsoventral lineage restrictions in Drosophila and Oncopeltus, it does not seem likely that the lateral boundaries of wg transcription are compartment boundaries (Lawrence, 1973; Garcia-Bellido et al. 1976; Steiner, 1976; Wright & Lawrence, 1981). So far the phenotype of wg mutations has not suggested a difference in wg function laterally (Nüsslein-Volhard & Wieschaus, 1980; Nüsslein-Volhard et al. 1984; Perrimon & Mahowald, 1987; Baker, 1988a). One speculation is that the changes in wg transcription during the extended germband might reflect the formation of imaginal discs in which the segmental field is preserved whilst most cells become committed to the larval epidermis (Anderson, 1963, 1972; Meinhardt, 1983). This occurs at about the same time in embryogenesis (Geigy, 1931; Wieschaus & Gehring, 1976). The leg discs form near the ventral-lateral boundary of wg transcription (Turner & Mahowald, 1979). Mature leg discs contain wg transcripts only in ventral cells, as though they originate at the boundary region (Baker, 1988b). Indeed, wg mutations can affect the formation of Keilin’s organs (Baker, 1988a), which suggests they also affect disc formation (Keilin, 1915).

I thank Drs P. A. Lawrence and G. M. Rubin for space and materials, Drs A. Martinez-Arias and P. Ingham for teaching me about in situ hybridization and for encouragement, and Drs E. Wieschaus, V. Foe and, especially, M. Weir for interesting discussions and for communicating unpublished results. The emulsion-incubation step was initially suggested by Dr H. Steller. I thank Drs M. Weir for the T7(2)en plasmid, T. Kornberg for permission to quote unpublished work, G. Hess and R. Tjian for use of their Drosophila population cage, and M. Weir and P. Ingham for comments on the manuscript. Support was provided by a Training Fellowship from the Medical Research Council, postdoctoral fellowship DRG-901 from the Damon Runyon-Walter Winchell Cancer Fund and NIH grant GM 32795 to Dr G. M. Rubin.