Genes that control dorsoventral polarity affect gene expression along the anteroposterior axis of the Drosophila embryo

Pattern formation in the Drosophila embryo requires the activities of several groups of genes that function in a region-specific manner during early development. The genes appear to act along either the shorter dorsoventral axis or the longer anteroposterior axis of the embryo. The absence of single gene products results in distinct alterations of cell fates and abnormal pattern formation. Among the genes that function along the dorsoventral axis, 11 loci have been identified thus far that are maternally expressed (Anderson & Nüsslein-Volhard, 1984; Anderson, Jürgens & Nüsslein-Volhard, 1984; T. Schüpbach, personal communication). The loss of any of ten of these gene products through mutation leads to ‘dorsalization’ of the embryo: cells at ventral and lateral positions assume dorsal fates and ventral structures fail to develop. The dorsoventral regulatory genes are thought to comprise an interacting network that establishes a ‘gradient’ of positional information along the dorsoventral axis (Nüsslein-Volhard, 1979; Anderson & Nüsslein-Volhard, 1984) .

Anteroposterior pattern formation in the Drosophila embryo is best understood in terms of the genes that control the number and identity of body segments (Nüsslein-Volhard & Wieschaus, 1980). Segmentation and homeotic loci appear to function in determining the fates of groups of cells along this axis. For those segmentation genes whose pattern of expression is known, such as the fushitarazu {ftz) and engrailed {en) genes, RNA transcripts (or proteins) are localized to distinct bands of cells encircling the embryo (Hafen, Kuroiwa & Gehring, 1984; Kornberg, Siden, O’Farrell & Simon, 1985; Carroll & Scott, 1985; DiNardo, Kuner, Theis & O’Farrell, 1985) . Genes such as ftz and en are members of a hierarchy of interacting genes that control the formation of body segments and pattern formation within the segments (Howard & Ingham, 1986; Carroll & Scott, 1986; Carroll, Winslow, Schüpbach & Scott, 1986) .

Genetic studies have suggested that pattern formation along each axis is controlled independently by two different sets of genes, since defects observed in single mutants appear to be restricted to one axis. The independence of the two sets of genes is difficult to assess definitively since abnormalities in embryonic development often lead to distorted structures that obscure normal pattern elements. For example, in dorsalized mutants that fail to gastrulate, segmentation is difficult to observe because the larva assumes a twisted form and most structures used to score anteroposterior development are ventral structures that are lacking in these mutants (Nüsslein-Volhard, 1979; Nüsslein-Volhard, Lohs-Schardin, Sander & Cremer, 1980). We have been able to more directly observe the influence of mutations by using antibodies to the ftz segmentation protein (Carroll & Scott, 1985) as a positional marker in the early embryo. We observe that disruption of dorsoventral polarity by mutations in dorsoventral regulatory loci results in positional shifts in ftz expression. We conclude that the axes of the embryo are not entirely independent. Certain aspects of anteroposterior gene expression respond to dorsoventral positional information.

Embryos were collected, fixed, stained with anti-ftz antibody, staged and photographed as described in Carroll & Scott (1985).

Mutant embryos were obtained from females either homozygous or transheterozygous for maternal-effect mutations, or from crosses of heterozygotes in the cases of the zen, twi, and sna loci. Mutant stocks were kindly provided by Drs K. Anderson, C. Nüsslein-Volhard, T. Schüpbach and P. Simpson and by the Bowling Green Drosophila Stock Center.

We have previously described the details of ftz protein expression in the early Drosophila embryo (Carroll & Scott, 1985). In wild-type animals, the ftz protein is expressed in seven broad stripes of nuclei encircling the embryo. Initially the stripes and ‘interstripes’ are about the same width (Fig. 1A), but at the initiation of gastrulation the stripes are clearly narrower than the unstained regions that lack ftz protein (Fig. 1B). This transition represents a decrease in the number of cells expressing the ftz protein. From a lateral view (Fig. 1B) the spaces between the stripes appear to be even, but it is clear that the number of nuclei stained is fewer and the spacing of the stripes more compressed at the dorsal surface of the embryo (Fig. 1B). The dorsoventral difference in the expression of ftz in wild-type embryos is important to our interpretation of the dorsal class of mutants described below and merits further description.

During the early stages of development, there are more cells on the ventral surface of the embryo than on the dorsal surface. Between the anterior border of the first ftz stripe and the posterior border of the seventh stripe there are approximately 42-44 nuclei dorsally and 52-54 nuclei ventrally. This difference is not accommodated by a symmetrical reduction of each stripe and interstripe dorsally. Close inspection reveals that the periodicity of ftz stripes on the dorsal surface is uneven, with the first, second and seventh stripes being wider than the third through sixth (Fig. 1C). The second through fourth interstripes are reduced in width. In contrast, the stripes of the ventral surface of the embryo are spaced more evenly, approximately five nuclear diameters apart (Fig. 1D). Thus, some mechanism causes cells to activate ftz at different intervals depending upon dorsoventral position. The likely candidate is the group of genes that acts during oogenesis to control dorsoventral polarity.

Eleven maternal-effect loci have been identified that affect the dorsoventral polarity of the developing embryo. Mutation of any one of ten of these genes leads to dorsalization of the embryo (Anderson & Nüsslein-Volhard, 1984). The dorsoventral polarity mutants form a normal cellular blastoderm and do not alter the egg shape (Nüsslein-Volhard, 1979). Therefore, gene expression can be examined in mutant embryos until the time when morphogenetic movements become severely defective at the initiation of gastrulation. Inspection of the ftz expression pattern in the early cellular blastoderm or just prior to the initiation of gastrulation in dorsoventral mutants reveals that dorsoventral polarity loci influence the pattern of ftz expression (Fig. 2). From the moment when the ftz stripes are first detectable, during cellularization of the blastoderm, the stripes encircling the embryo have an abnormal periodicity (Fig. 2A) and are formed in much more perpendicular orientation to the long axis of the animal in a dorsal embryo than those in the wild-type embryo. At the onset of gastrulation, the first two stripes oí ftz protein are clearly wider than the next four stripes, and the second through fourth interstripes consist of fewer nuclei as well (Fig. 2B,C). This pattern is uniform around the entire circumference of the embryo and resembles the ftz pattern formed on the dorsal surface of the wild-type embryo (compare Figs 1C and 2C). The abnormal pattern is not the result of specific cell movements. Rather, the ftz stripes have formed over a shorter interval of the egg length. Thus, in a dorsal mutant embryo, all cells assume a dorsal fate and the ftz gene is activated at a dorsal-type interval around the entire dorsoventral axis. Once gastrulation begins, the pattern of ftz expression becomes obscured by the abnormal furrows formed around the entire circumference of the embryo (Fig. 2D).

We have eliminated the possibility that genetic background effects of the dl¹ chromosome causes pattern alterations by demonstrating that the dl¹/Df(2L)TW119progeny exhibit the same defect, as do all progeny of the other dorsalizing maternaleffect mutations. We have examined abnormal ftz gene expression caused by ten maternal-effect mutants for the dorsoventral polarity system. The loci and the alleles examined are listed in Table 1. All ten maternally active dorsalizing mutations affect the periodicity oí ftz expression and cause the stripes to form at a dorsal-type periodicity over an interval of 42-44 cells. Additional examples of such mutant effects are shown iorspatzle (Fig. 2E) and gastrulation defective (Fig. 2F). The patterns of ftz expression are indistinguishable in mutants for all ten maternaleffect dorsal-ventral loci.

Not all loci affecting dorsoventral polarity have effects on the ftz protein pattern. Embryos from homozygous mutant cactus mothers, which appear to be partially ventralized (T. Schüpbach & E. Wieschaus, personal communication), exhibit a completely normal ftz pattern. In addition, a few zygotic loci have been identified that appear to be components of the dorsoventral pattern formation system (Table 1). Mutations of the zerknüllt (Wakimoto, Turner & Kaufman, 1984) twist (Simpson, 1983), and snail (Grau, Carteret & Simpson, 1984) genes affect the formation of structures characteristic of the dorsal (zen) or ventral (twi and sna) surface. Partially dorsalizing alleles of the twi and sna loci (twi^ey63 and sna^ryI) and strong zen mutants (zen^w36) do not have any effect on the periodicity offtz expression (data not shown).

It is improbable that any of the effects observed here represent a direct interaction between dorsoventral gene products and the ftz gene. It is more likely that the positions of the ftz stripes represent the outcome of a series of interactions that precede its expression in the early embryo. There are four principal reasons for this assertion. First, some of the maternally active loci have very early temperature-sensitive periods that end before ftz expression begins (Anderson & Nüsslein-Volhard, 1984). Second, the ftz patterns in all maternal-effect mutants are indistinguishable, suggesting that the maternal-effect genes act upon ftz through a common pathway an idea consistent with previous morphological studies of the dorsoventral genes (Anderson & Nüsslein-Volhard, 1984). Third, a large number of other maternally active and zygotically active genes have been shown to influence the initial pattern of ftz expression and some are probably affected by the dorsoventral genes and thus alter the ftz pattern (Carroll & Scott, 1986; Carroll et al. 1986). Fourth, the segmentation gene hairy which regulates the periodicity oí ftz expression (Howard & Ingham, 1986; Carroll & Scott, 1986) has also been shown to respond to dorsoventral position in that its dorsal anterior stripe is duplicated ventrally in the absence of the dl product (Ingham, Howard & Ish-Horowicz, 1985). We also note that the periodicity of the hairy striped pattern appears to be affected in dl embryos (see fig. 6 in Ingham et al. 1985). An alteration in hairy expression alone would be expected to lead to an alteration in ftz expression.

The dorsalizing series of mutants are believed to represent a group of genes that function to establish a graded continuum of positional information across the dorsoventral axis (Anderson & Nüsslein-Volhard, 1984) . Mutations that eliminate a single gene’s function in the system appear to reduce all information to a dorsal ground state. Since this dorsal ground state impacts upon the normal anteroposterior system, one must expect that there is a specific mechanistic link between components of the two patterning systems used in the two axes. The most likely explanation for the positional changes in anteroposterior gene expression observed here is that regulatory components (or signals) of the anteroposterior system are normally asymmetrically organized by action of the dorsoventral polarity system. We suspect that the anteroposterior regulatory components involved are those that regulate the pattern of segmentation gene activity, perhaps maternally expressed products such as those of the bicaudal-type loci (Mohler & Wieschaus, 1985) or the maternal segmentation loci (Schüpbach & Wieschaus, 1986).

We would like to thank Dr Kathryn Anderson for many stocks and helpful advice. We thank Drs C. Nüsslein-Volhard, P. Simpson, T. Schüpbach and E. Wieschaus for additional stocks; Drs Jim Kennison, John Tamkun, Joan Hooper and Allen Laughon for critical review of the manuscript; and Cathy Inouye for secretarial assistance. S.B.C. was supported by NIH Postdoctoral Fellowship GM-09756 and this work was supported by NIH Grant HD 18163 to M.P.S.