Phenotypic comparison between maternal and zygotic genes controlling the segmental pattern of the Drosophila embryo

The longitudinal pattern of the Drosophila embryo is controlled by the concerted activity of gene products provided during oogenesis (maternally active genes) and embryogenesis (zygotically active genes). An initially relatively coarse system of positional information laid down by maternal gene products becomes successively refined towards the repeating pattern of segments first by the division into domains by the products of the zygotic gap genes and subsequently by the action of pair-rule and segment-polarity genes (Nüsslein-Volhard & Wieschaus, 1980; Ingham & Martinez-Arias, 1986; Ingham, 1988; for review see Akam, 1987). Maternal genes affecting anteroposterior pattern have been classified into three groups according to their phenotype: the terminal group, the anterior group and the posterior group (Nüsslein-Volhard et al. 1987). Together the three maternal gene groups control the establishment of the entire segmental pattern. Embryos that lack all maternal information show no anteroposterior pattern (Nüsslein-Volhard et al. 1987, R.L. unpublished data).

The information provided by the maternal genes is interpreted by zygotic genes. The best candidates for genes that may directly respond to the maternal signals are the zygotic gap genes (Table 1; Nüsslein-Volhard & Wieschaus, 1980). The number of genes with a ‘gap’ phenotype is small and each gap gene has a distinct phenotype. Similar to mutations in the maternal genes, gap mutations cause large continuous deletions including several consecutive segments while the remaining structures are relatively normal. In this article, I would like to summarize and discuss some of the results concerning the establishment of positional information in the egg cell and its interpretation by differential activation of zygotic genes. The first part deals with the properties of the posterior group genes as they have been characterized genetically as well as by fate-map analysis. This description should provide some idea about the methods used to characterize phenotypic groups. In the second part, these results will be compared to similar studies on the anterior and terminal group. The goal of this article is to point to the relative roles maternal anteroposterior genes and zygotic gap genes play in the generation of the segmented pattern.

Mutations in eight diferent genes affect the segmentation of the embryonic abdomen: knirps (Nüsslein-Volhard & Wieschaus, 1980; Jürgens et al. 1984), tudor (Boswell & Mahowald, 1985), vasa, valois, staufen (Schüpbach & Wieschaus, 1986), oskar, pumilio and nanos (Lehmann & Nüsslein-Volhard, 1986, 1987a, unpublished data). To test whether the products of all eight genes are involved in the same developmental pathway, I studied the lethal phenotype and its origin for each locus.

Fig. 1 shows the strongest phenotypes produced by mutations in one of the maternal posterior group genes, nanos, and the zygotic gene knirps (for the wild-type pattern refer to Fig. 1). Embryos derived from nanos females lack all abdominal segments while the regions anterior and posterior to the abdomen, the head-thorax region and the telson, respectively, appear normal. However, in embryos mutant for a strong kni allele, two abdominal segments are formed. The first abdominal segment is enlarged and more rows of denticles are formed than in a wild-type abdominal segment (14 – 16 instead of 6 – 7 in the wild type). Morphologically this segment resembles a first abdominal segment, but by genetic criteria (double mutant combinations with various mutants of the Bithorax-Complex) it seems that this segment is of mixed segmental identity (A1 – A5). The entire field acquires thoracic morphology only in kni embryos which lack Ubx and abdA (Sanchez-Herrero et al. 1985) but not in double mutants between kni and Ubx^c1 (Casanova et al. 1988). The second abdominal segment formed by a knirps embryo corresponds morphologically as well as genetically (complete transformation only in double-mutants with Df(3R) P9 (Lewis, 1978)) to a normal eighth abdominal segment.

All maternal posterior group genes show the strong phenotype described for nanos. The difference in phenotypic strength between the zygotic and maternal mutants may be due to maternally derived kni product. To test this idea I compared the phenotype of homozygous kni embryos derived from a germ line homozygous mutant for kni with those derived under normal conditions from a heterozygous germ line. Germ line precursor cells from the progeny of heterozygous kni flies were transplanted into sterile hosts (for method used refer to Lehmann & Nüsslein-Volhard, 1987b). The germ lines of four females were homozygous and those of twelve were heterozygous for kni (total number of fertile females = 34). No difference in the mutant phenotype could be detected between the mutant progeny. The kni gene seems thus to be expressed exclusively by the embryo itself.

Weak alleles have been identified for kni (Jürgens et al. 1984) and all seven maternal loci (Boswell & Mahowald, 1985; Schüpbach & Wieschaus, 1986; Lehmann & Nüsslein-Volhard, 1986, 1987a, unpublished data). It is thus possible to compare the effect residual gene activity of any given locus has on the final mutant pattern. The hypomorphic series of all seven maternal genes is basically identical and has been described earlier for the oskar allele osk³⁰¹. With increasing phenotypic strength segments are lost from the middle region of the abdomen (A4-A6) while the first and the eighth abdominal segment are most insensitive to variations in gene activity (Fig. 2). The phenotype of the strong, intermediate and weak kni alleles are found as intermediates of the maternal series (compare Fig. 2A-D,B-E,C-F). The strong phenotypic similarities between the maternal genes and kni suggest a common role these genes play for the development of the embryonic abdomen.

Since the final lethal phenotype is the consequence of an early developmental misrouting it is necessary to study the origin of the pattern abnormalities observed in the cuticle pattern. Fate-map changes in the abdominal region are difficult to detect during early development of mutant embryos because no morphological markers can be used. The anterior and posterior dorsal folds, for example, are formed in strong mutant embryos and the head fold is at its normal position in all mutants with the exception of stau where it forms more anteriorly (Schüpbach & Wieschaus, 1986). Shortly after the onset of gastrulation mutant embryos deviate from wild type as they do not fully extend the germ band to the dorsal side. This effect is more pronounced in the strong maternal mutants than in kni embryos. Later during development localized cell death occurs in the abdominal region of all maternal mutants and kni.

At the blastoderm stage and thus prior to any morphological deviation from wild-type development, fate map changes can be detected in mutant embryos probed with polyclonal antibodies directed against the product of the segmentation gene fushi tarazu (ftz) (Carroll & Scott, 1985, 1986; Carroll et al. 1986). In wild-type embryos at this stage, ftz protein is expressed in seven transverse stripes separated by stripes of non-expressing nuclei (Fig. 3A). Each stripe is about 3 – 4 nuclei wide, while the seventhftz expressing stripe is 5 – 6 nuclei wide. The repeating pattern of ftz expression can be used to map the segmental primordia. The primordium of the abdomen spans from the middle of the third stripe (parasegment 6, Martinez-Arias & Lawrence, 1985) to the anterior border of the last stripe (parasegment 14) corresponding to a region between 50 % and 20 % egg length (0% egg length corresponds to the posterior pole).

For each locus, ftz expression was monitored in embryos of different phenotypic strength (Fig. 4). When we compare the pattern of ftz expression in embryos that would have developed the same late cuticle phenotype, the phenotypic series of all maternal genes is similar. A series of fate map changes leads from the wild-type fate map with about even spacing of seven ftz expressing stripes in the region between 65 % and 10 % egg length to a dramatically changed fate map in strong mutant embryos (Fig. 3B, Carroll et al. 1986). In these only four ftz stripes can be detected. The anterior border of the first stripe is at its normal position but the first two ftz stripes and the interstripe are expanded such that each is five to six cells wide instead of three to four in wild type. The third stripe is less intensely stained with the antibody and is only found on the dorsal side. The fourth stripe resembles from its position and size the expression pattern of the last, seventh, wild-type stripe (Fig. 3B). The stripes four to six are missing. The interpretation of the strong mutant pattern is facilitated by the pattern of expression in embryos of intermediate phenotype (Fig. 3D,F,H). In the abdominal region, the size of the ftz expressing and nonexpressing regions becomes reduced to one to two cells in embryos that would show single segment deletions (Fig. 3H). In stronger mutant embryos, the three stripes are fused into one, or less frequently two, broad regions of expression (Fig. 3D,F) and finally in the strongest phenotype they disappear. In the abdominal region, we thus observe a fusion of metameric primordia into enlarged units while in the thoracic region (first three stripes) segmental primordia seem to expand harmoniously towards the posterior with decreasing in gene activity (Fig. 5).

All maternal mutants show the same fate map shifts with the exception of mutant staufen embryos (Figs 4 and 5). stau mutations cause an expansion of the thoracic region towards anterior and posterior. In the most extreme mutant phenotype, the region between the first and the third stripe (which correspond to the primordia of the posterior maxilla, the labium and the first, second and anterior third thoracic segments) extends from 75 % to 40 % egg length instead of 65 % to 48 % in wild type (Fig. 5). In the abdominal region, segments are compressed and finally lost in a pattern very similar to that observed in other maternal genes (Fig. 4). Genetic and molecular results suggest that the effect stau mutations have on the anterior fate map reflects the role of stau in the localization of bicoid product (Driever & Nüsslein-Volhard, 1988b; R.L. unpublished data).

The allelic series for knirps is rather similar to that described for the maternal mutants (Fig. 4) but, interestingly, changes in the fate map are restricted to the abdominal region while the thoracic region is not affected (Fig. 5). In extreme kni mutants, two ftz stripes of normal position are followed by an enlarged third stripe of normal intensity which encompasses the circumference of the embryo and a last stripe most likely resembling the normal seventh stripe (Fig. 3C) (see also Ingham & Gergen, this volume). In weak and intermediate phenotypes, segmental primordia in the abdominal region are compressed and lost in a sequence similar to that described for the maternal genes (compare Fig. 3C-D,E-F,G-H).

Our studies on the phenotype, fate map and development of mutant kni embryos and embryos derived from females mutant for each of the maternal posterior group genes indicate a common basis for the late mutant phenotype. In all mutants, we can detect similar fate-map changes in the abdominal region at the blastoderm stage (2 · 5h after egg deposition). These fate-map changes presumably lead to the generation of enlarged segmental primordia in which cell death occurs later during development. Thus we can visualize the final cuticle phenotype as the consequence of a primary defect in the blastoderm fate map and a later-occurring size-regulative process.

The wild-type function of all maternal posterior group genes is required for an abdomen-specific activity localized at the posterior pole (Lehmann & Nüsslein-Volhard, 1986, 1987a, unpublished data). Transplantation of posterior pole plasm from a wildtype embryo into the abdominal region rescues the abdominal phenotype of the maternal posterior group mutants. For osk mutants, a quantitative relationship between the activity found at the posterior pole and the degree to which abdominal segmentation is affected can be established. Strong mutant embryos contain no activity while weak alleles have residual activity (Lehmann & Nüsslein-Volhard, 1986). The transplantation experiments further indicate that the distribution of the signal from its source, the posterior pole to its target at the abdominal region is graded from posterior to anterior (Lehmann & Nüsslein-Volhard, 1987a). The harmonious expansion of the thoracic primordia in parallel to the enhancement of the mutant phenotype further suggests that the requirements for the signal are different along the anteroposterior axis. The fusion of segmental primordia in the abdominal region, on the other hand, does not follow a strict anterior-posterior pattern (see above) and may suggest different requirements of the posterior signal for the activation and/or repression of kni and neighbouring gap genes, such as Kr and giant (Petschek et al. 1987), which affects the development of the sixth through seventh abdominal segment (see legend of Table 1).

The differences between the phenotype of kni and the maternal genes indicate that the maternal genes do not exclusively act on the expression of kni. The effect of the maternal mutants on thorax development may well reflect the role that the maternal genes play in controlling the expression pattern of Krüppel. Kr affects the development of the thorax and the anterior abdomen (Fig. 1H; Wieschaus et al. 1984) and the Kr protein is expressed at early blastoderm in a region between 39 % and 55 % egg length (Gaul et al. 1987). The Kr protein domain is expanded quite prominently towards posterior in embryos lacking the maternal posterior gene products and only slightly enlarged in kni (Gaul & Jackie, 1987).

Studies on the anterior and terminal group of genes suggest that direct relationships similar to those between the posterior group genes can be established between maternal and zygotic gap genes on the basis of phenotypic analysis and fate mapping.

Five maternal genes, one gene expressed maternally as well as zygotically and one zygotically active gene belong to the terminal group: torso (tor), trunk (trk) (Schüpbach & Wieschaus, 1986); torsolike (tsl) (Nüsslein-Volhard et al. 1987; Frohnhöfer, 1987), fs(l)Nasrat (fs(l)N) (Degelmann et al. 1985), fs(lfpolehole (fs(l)ph) (Perimon et al. 1986); l(l)polehole (1(1)ph) (Perimon et al. 1985), tailless (til) (Strecker et al. 1986, 1988). As the common phenotype, the terminal structures normally formed at the anterior (acron) and posterior (telson) are missing in mutant larvae and the fate map is altered such that subterminal structures are formed at the ends. In extreme maternal mutants, all cuticular structures of the acron (labrum, dorsal bridge) and telson (anal pads, tuft, Filzkörper, spiracle) are missing (Fig. 1D). In addition, the posterior midgut rudiment and the hindgut are not formed posteriorly and the anterior midgut and the stomodeum are reduced to variable extent in different mutants. All maternal mutants also affect the cellularization at the posterior pole (pole hole effect) without interfering with the normal development of the pole cells. For mutants in three genes, more general defects in the cellularization of the entire embryo have been described (fs(l)N, fs(l)ph, l(l)ph).

The zygotic gene tailless (tll), on the other hand, specifically affects the terminal cuticular structures. In mutant tll embryos, the head skeleton is reduced and, posteriorly, the hindgut, the malpighian tubules and the telson are missing (Fig. 1E). Fate mapping experiments indicate that in tll embryos the anlagen of the remaining structures are expanded towards the poles (Mahoney & Lengyel, 1987) in a pattern similar to that observed for the maternal mutants (Degelmann et al. 1985; Mlodzik et al. 1987). tll embryos, in contrast to the maternal terminal group mutants, form a normal labrum and posterior midgut, cellularize normally and show no cell death during development (Strecker et al. 1986). The phenotypic differences between the maternal genes and tll suggest that the maternal genes may regulate the expression of additional zygotic genes required for the normal development of the most anterior (labrum, stomodeum, anterior midgut) and posterior (posterior midgut) structures and for cellularization.

Three maternal and one zygotic gene affect the anterior pattern in a similar way: bicoid (bed, Frohnhôfer & Nüsslein-Volhard, 1986), swallow (swa, Stephenson & Mahowald, 1987; Frohnhôfer & Nüsslein-Volhard, 1987), exuperantia (exit, Schüpbach & Wieschaus, 1987; Frohnhôfer & Nüsslein-Volhard, 1987) and hunchback (hb, Lehmann & Nüsslein-Volhard, 1987b; Bender et al. 1987). In contrast to the posterior and terminal group of genes, the genes in this class have distinctly different phenotypes, bicoid is the central gene in this class. In embryos derived from mutant bed females, head and thoracic structures are missing; at the anterior tip, the acron is replaced by a duplication of the telson, normally located posteriorly (Fig. 1F). The defects of mutants in the other two genes in this class, exu and swa, are less severe and resemble the phenotype of weak bed alleles (for description of phenotype and fate map refer to Frohnhôfer & Nüsslein-Volhard, 1987). From the zygotic genes identified so far, the phenotype of the gene hunchback shows most similarities with the bed phenotype, hb is expressed zygotically as well as maternally. Embryos derived from a germ line heterozygous for hb develop a normal acron and head and only the labium (the gnathal segment directly adjacent to the thorax) and the thorax are missing. Fate-mapping experiments carried out with mutant bed (Frohnhôfer & Nüsslein-Volhard. 1987) and hb embryos (Carroll & Scott, 1986; R.L. unpublished data) support the similarities in phenotype. In hb and bed mutant embryos, the two anterior stripes of ftz expression are deleted while the third stripe marking the border between thorax and abdomen is expanded towards anteriorly.

The extreme hb phenotype is only produced by homozygous mutant embryos derived from a homozygous mutant germ line. Such embryos form a normal acron but almost all head structures and the thorax and anterior abdomen are deleted. The posterior abdomen is duplicated anteriorly with the plane of mirror-image duplication in the fourth segment (Fig. 1G). The lack of function hb phenotype can thus not be as easily described as a subpattern of the more extreme maternal phenotype. Mirror-image duplications within the abdomen occur rarely in bed embryos (Frohnhôfer, 1987). The occurrence of mirror-image duplications in hb may be due to novel zygotic interactions among the gap genes (Gaul & Jackie, 1987) caused by the lack of hb product, and may thus point to the important role the maternal and zygotic hb product has for controlling the expression of Kr and kni during normal development. In contrast to the zygotic expression of hb, which is under bed control, the distribution of the maternal hb product is under the influence of the maternal posterior group genes (Tautz, 1988).

The phenotypes of the maternal and zygotic mutants described in this article are summarized in Fig. 6. From the comparative analyses discussed we can conclude the following:

(1) There is a zygotically expressed gap gene with a phenotype similar to each of the maternal gene classes (Fig. 6). No maternal gene has so far been found with a phenotype similar to that of Krüppel. A zygotic gene affecting the most terminal regions of the embryo, the endodermal primordia and the acron has not yet been described.

(2) The hypomorphic series of maternal and zygotic genes follows similar principles. In the posterior and anterior group, segments are not lost in a strict anterior-posterior order (Frohnhôfer & Nüsslein-Volhard, 1987; this article), while an anterior-posterior order has been described for the terminal group, where with increase in phenotypic strength structures are lost towards the respective pole (Strecker et al. 1988).

(3) Mutant phenotypes within a group originate by similar principles (fate-map shifts, expanded segmental primordia, cell death) in the maternal and zygotic mutants (Carroll et al. 1986; Carroll & Scott, 1986; Mlodzik et al. 1987; Mahoney & Lengyel, 1987; Degelmann et al. 1986; Frohnhôfer & Nüsslein-Volhard, 1987; this article).

(4) The lack of function phenotype of each zygotic gap gene is less extreme than that of the respective maternal genes. Therefore, each maternal gene seems to control the expression pattern of more than one zygotic gene. This may be achieved by direct activation of other zygotic genes or, as shown for Krüppel, by suppression (Gaul & Jackie, 1987).

A number of maternal genes and almost all zygotic genes have been analysed molecularly (e.g. Stephenson et al. 1987; Frigerio et al. 1986; Berleth et al. 1988; Driever & Nüsslein-Volhard, 1988a,b; Preiss et al. 1984; Gaul et al. 1987; Tautz et al. 1987; Tautz, 1988). The role of specific maternal products for the activation of the zygotic counterparts has thus become amenable to direct molecular investigation. The phenotypic analyses summarized here indicate a complex pattern of interactions in order to provide the information required for the establishment of an integrated anteroposterior pattern.

I am especially thankful to Phil Ingham for his patience and commentaries on the manuscript. I thank H. Krause for the ftz antibody and A. Cron for help with the preparation of the manuscript.