Effects of ectopic expression of caudal during Drosophila development

The genetic analysis of development is facilitated by the ability to observe the effects of loss and gain of function of gene products, traditionally through the study of mutations. Recently, the genetic analysis of development has been supplemented by the technique of controlled ectopic expression of cloned genes. For example, the ubiquitous expression of the segmentation gene fushi (ftz) during early development has been used to study the establishment of metameric stability (Struhl, 1985; Ish-Horowicz et al. 1989), while we have provided a detailed study of the homeotic transformations caused by ectopic overexpression of the homeotic gene Antennapedia (Gibson and Gehring, 1988). Here, we report the use of heat-shock-induced ectopic expression of the caudal gene to address the problem of the role of a molecular gradient in the establishment of positional information during early Drosophila embryogenesis.

The caudal gene (Mlodzik et al. 1985; Hoey et al. 1986; MacDonald and Struhl, 1986) was first isolated by cross-hybridisation to the homeobox sequences present in a number of genes of the Antennapedia and Bithorax Complexes of D. melanogaster (McGinnis et al. 1984). These genes are thought to control segmental identity by virtue of their function as trans -acting regulators of transcription (Gehring, 1987; Krasnow et al. 1989; Scott et al. 1989). Maternal and zygotic cad messenger RNAs derive from two different promoters, but share a common coding sequence which encodes a single protein species (Mlodzik and Gehring, 1987a).

The earliest detectable expression of CAD protein (CAD) is after the fifth or sixth nuclear division (MacDonald and Struhl, 1986). During the syncitial blastoderm stage, CAD comes to be expressed in a gradient receding from the posterior pole. The formation of this gradient is dependent on the presence of the bicoid gene product, which is thought to repress cad translation in the anterior of the embryo (Mlodzik and Gehring, 1987b). Protein derived from maternal transcripts persists in the pole cells throughout gastrulation (Mlodzik and Gehring, 1987a), and zygotic cad message is detected in the third instar larval gonads. Thus, cad is also expressed in the germ line of Drosophila, the only homeobox gene known to be expressed there so far.

As the early gradient fades away, zygotic transcription appears to be initiated in a single stripe of 4-5 cells between 13% and 19% egg length, measured from the posterior pole. Protein continues to be expressed in these cells throughout germ band extension. Later on, expression is also found in the genital imaginai disc (Mlodzik and Gehring, 1987a). Such expression patterns, by comparison with those of the homeobox genes of the Antennapedia and Bithorax Complexes, are consistent with a role for cad in specifying segmental identity of the most posterior structures of the epidermis (A10), including the anal pads and the telson, as well as portions of the posterior gut and the malpighian tubules.

The generation of cad mutants by MacDonald and Struhl (1986) confirmed that cad plays a role in the promotion of posterior segmental development, although rather than leading to unequivocal homeotic transformations, loss of zygotic cad appears to result in deletion of particular posterior cuticular structures. The maternal function can be partially rescued by paternally supplied gene products. Nevertheless, cad is involved in promotion of the segmentation process, since in the absence of all cad activity in embryos, most of the abdominal segments are missing or fused. The phenotype of cad and its effects on the embryonic fate map suggest that cad probably does not encode a morphogen responsible for determination and positional information along the anterior-posterior axis. In order to further examine the role of caudal, we have studied the effects of ectopic overexpression of the gene at all stages of Drosophila development.

The pHT cad plasmid was constructed as a derivative of pHT4 with the rosy gene as a visible eye-colour marker (Schneuwly et al. 1987). Genomic cad DNA from the Clal restriction site 21 nucleotides upstream of the maternal transcription start site was fused to downstream sequences of the cDNA clone pSC33F (Mlodzik and Gehring, 1987a) at the Xhol site in the first exon. The insert fragment from the resulting plasmid, p33FG, was excised with Clal and EcoRl, and blunt-ended with T4 DNA polymerase. After addition of Kpnl linker oligonucleotides, this fragment was inserted into the Kpnl site of pHT4 to yield pHTcad. The pSP65 plasmid was used as the vector for all intermediate subcloning steps. The flies were injected with CsCl gradient purified pHTcad DNA in the presence of pjr25.1wc as the source of transposase enzyme (Karess and Rubin, 1984). Homozygous or balanced lines were generated by crossing to appropriate balancer chromosomes.

A construct lacking most of the sequences of the leader of the maternal cad mRNA was constructed using a similar strategy as described above. Sequences upstream of the XhoI site were derived from a deletion clone generated for sequencing that was lacking most the leader (88 nucleotides upstream of the translation start codon were still present). However, injections of this construct (pHT cadA) never gave rise to a transformant line even though a larger than usual number of embryos were injected with two different preparations of plasmid DNA.

Embryonic heat shocks were performed essentially as described in Gibson and Gehring (1988), except that the embryos were staged at the syncitial blastoderm stage (just as the periphery of the embryos started to clear) rather than the early cellular blastoderm stage (when the membranes start to form between nuclei). Such embryos are no more than two and one half hours old (Campos-Ortega and Hartenstein, 1985). Larval heat shocks were also administered as previously described, except that the heat shocks were delivered by submerging the glass fly tubes in up to three centimetres of water (such that the top of the food was clearly below the surrounding water level) at 37°C for 45 min. Typically, between three and five successive heat shocks were given at 4h intervals (recovery at 25°C), starting from 68h after egg laying. The cuticular structures were examined after mounting of embryos or adult heads in Hoyer’s or Faure’s medium respectively, also as described in Gibson and Gehring (1988).

Rescue of the hypomorphic cad¹ allele (MacDonald and Struhl, 1986) was achieved with the following crosses (cad² and cad³ could not be rescued in this assay):

The rf10 chromosome contains the markers _fiz^9H34, ry⁵⁰⁶, e^s. The final stock was maintained at 18°C: the lethal b pr cad¹(caudal mutation) and HTcad2 (heat-shock caudal ) chromosomes are balanced over the CyO and TM3 chromosomes, respectively. When grown at 25°C, we observed b pr Sb progeny of this cross, indicating that about 10 percent of homozygous cad¹ embryos had been rescued by the HTcad2 chromosome, by comparison with 1 percent homozygous b pr cad¹ survivors in an otherwise wild type background.

All embryo fixations and antibody stainings were performed as described previously (Patel et al. 1987), with the following modification: the biotin-avidin HRP system was replaced with secondary HRP-conjugated goat anti-mouse or goat antirabbit antibodies (BioRad). The primary antibodies used were as follows: rabbit anti-Kr (Gaul el al. 1987), rabbit anti-hb (Tautz, 1988), rabbit anti-ftz (Mlodzik et al. 1987; Krause et al. 1988), monoclonal anti-en (Patel et al. 1989), and rabbit anti-Dfd (Mahaffey et al. 1989).

Using the same approach to that used to analyse the effects of ectopic expression of Antp, cad sequences derived from the zygotic cDNA clone 33F and genomic DNA (Mlodzik and Gehring, 1987a: see Methods) were placed under the transcriptional control of the Drosophila hsp70 heat-shock promoter in the pHT4 vector described by Schneuwly et al. (1987). The resulting plasmid, pHTcad (Fig. 1 A), was then used to transform flies by P-element-mediated transformation (Rubin and Spradling, 1982; Spradling and Rubin, 1982), and three independent lines were established. Most of the experiments described below were performed with the HTcad3 line, which carries a homozygous viable insertion on the third chromosome, although similar results were obtained with the other two lines. We were unable to obtain transformants with a second construct, in which most of the untranslated leader of the maternal RNA was deleted (see Methods). As shown in Fig. 1 B, shortly after a 20min heat shock at the syncitial blastoderm stage of development, the nuclei of all cells of HTcad3 embryos contain relatively high levels of CAD detected by whole-mount immunohistochemical staining. Since the protein appears to be evenly distributed, it is likely that enough mRNA has been produced to overcome the block of translation at the anterior end of the embryo. We confirmed that the construct produces functional CAD by crossing the HTcad2 chromosome into flies homozygous for the hypomorphic cad’ allele (MacDonald and Struhl, 1986), and observing 10% rescue after growth at 25°C (see Methods for details).

The protein gradient was disrupted by ectopic overexpression of HT cad as early as possible in development. Zygotic transcription is not detectable in Drosophila until two hours after fertilization, at the syncitia) blastoderm stage. Embryos of this age can be staged by dechorionating a collection of two hour old embryos (maintained at 25°C), submerging them in water, and selecting those in which the periphery of the embryo just commences to clear when observed under a dissecting microscope (see Campos-Ortega and Hartenstein, 1986 for a detailed description of early embryogenesis). If syncitial blastoderm HTcad3 embryos are subjected to a 20 min heat shock at 37°C, and allowed to complete embryogenesis at 25 °C, two effects on development are apparent: a partial failure of head involution, and a significant disruption of segmentation (Fig. 2).

By contrast with the ectopic expression of Antenna-pedia protein (Gibson and Gehring, 1988), the failure of head involution after ectopic expression of Caudal protein is not accompanied by any obvious signs of homeotic transformation of segmental identity. The defect involves a malformation of the embryonic head skeleton as well as a displacement of the mouthparts relative to one another, but no supernumerary denticles or structures characteristic of more posterior body parts have been detected, even after multiple heat shocks delivered during the first seven hours of embryogenesis. Nor have we been able to detect any consistent or significant malformation of the posterior embryonic structures (see Jiirgens, 1987) in response to ectopic expression of CAD. Although occasional thickening of the telson is observed, such an effect can also be induced by ectopic expression of a variety of other constructs (unpublished data).

Segmentation defects were manifested as fusions of one or more denticle belts in 45% of HTcad3 embryos heat shocked at the syncitial blastoderm stage (n=89, derived from four independent experiments, each with similar percentages of defects). A typical defect is shown in Fig. 2 B, while a strongly affected embryo in Fig. 2 C illustrates the point that all of the fusions observed involved fusions of T1/T2, T3/A1, A2/A3, A4/A5 and/or A6/A7. Minor defects usually involving A3 and A4 can be seen in up to 20% of wild-type embryos heat-shocked at the same stage of development (G.G., unpublished observations). Our interpretation of the segmental nature of the regions deleted in affected embryos is shown in Fig. 3. The defects are too variable and incomplete to allow a precise determination of which cells are deleted. Comparison with the findings of Struhl (1985) and Ish-Horowicz and Gyurko-vics (1988) suggests that ectopically expressed CAD causes similar defects to those caused by ectopic expression of the pair-rule segmentation gene fushi tar-azu. At the segmental level, it clearly does not have effects complementary to removal of the maternal cad contribution. However, globally the defects are most frequent at the opposite end of the embryo as those seen in loss-of-function mutant embryos. A similar range of segmentation defects has also been observed after ectopic expression of the segment polarity gene engrailed (Poole and Kornberg, 1988), although by contrast Engrailed apparently causes segmental fusions more frequently in the posterior than in the anterior portions of the embryo.

We have used antibodies against three proteins to monitor changes in the embryonic fate map in response to ectopic expression of CAD in syncitial blastoderm embryos. The wild-type protein staining patterns of fushi tarazu (FTZ) at the cellular blastoderm stage, and engrailed (EN) and Deformed (DFD) at germ band extension are shown in Fig. 4A, C, and E, respectively.

Segmentation defects seen in the mature embryonic cuticle are reflected in an altered distribution of FTZ staining within half an hour of induction of CAD. Fig. 4B shows the most frequently observed FTZ distribution: stripes 2, 3, and 4 are abnormally spaced and reduced (indicated with open arrowheads). Occasionally one of these stripes is completely missing (data not shown). In addition, in most heat-shocked HTcad3 embryos, the posterior three stripes appear to stain more intensely than the anterior ones. These results are in direct contrast to the effects of loss of maternal and zygotic cad function, which causes the anterior FTZ stripes to broaden and the posterior FTZ stripes to become severely reduced (MacDonald and Struhl, 1986).

The expression of EN and DFD protein is strongly altered at the germ band extension stage in 35-40% of HTcad3 embryos heat shocked at the syncitial blastoderm (Fig. 4D and F). The modification and reduction of staining of EN and DFD respectively provides strong evidence that the effects of CAD on the head are primarily due to disruption of head segmentation and the repression of head-determining gene activity, rather than to cad activity per se. In most embryos, the segmental repeat pattern of 14 stripes of EN-expressing cells (DiNardo et al. 1985) can still be seen, but some stripes are often reduced and irregularly spaced. Most often, the EN stripes in the head, thoracic and anterior abdominal anlagen appear changed in intensity and spacing (Fig. 4D). However, occasionally en expression is also affected in more posterior regions of the embryo (e.g. see most posterior reduced EN stripe in Fig. 4D). There is little sign of an increase in width of each EN stripe similar to that seen after ectopic expression of FTZ (Ish-Horowicz et al. 1989), suggesting that the segmental defects observed in the cuticle may arise as a result of a more general mechanism involving slight disruption of the expression of a number of segmentation genes. The extent of reduction of DFD staining is quite variable, but seems to be approximately equivalent in all cells given that DFD is normally expressed most strongly in the ventral cells of the maxillary and mandibular segments (Jack et al. 1988; Mahaffey et al. 1989). No alterations in the patterns of expression of the zygotic gap genes, Kriippel and hunchback, were seen in response to early induction of cad.

Since the known homeotic selector genes Antp, Sex combs reduced, Dfd, and Ultrabithorax (Gibson and Gehring, 1988; Gibson et al. 1990; Kuziora and McGinnis, 1988; Mann and Hogness, 1990) all produce their strongest homeotic transformations when ectopically expressed between five and seven hours after fertilization, the effects of ectopic CAD expression were also examined during later stages of development. However, no increased defects were observed. Indeed, after germ band extension, no effects on embryonic development were seen at all, and just prior to this stage, heat shocks caused only very minor disruptions of head involution in HTcad3 embryos (data not shown). Consequently, there does not appear to be a transformation of head parts to telson, as might have been expected if cad acting alone during a short period of development is capable of promoting posterior segmental identity.

The effect of ectopic expression of CAD on imaginai disc development has also been examined, since the cad gene is expressed in the genital disc during larval development. Although the delivery of three successive heat shocks at four hour intervals is lethal to up to 50% of HTcad3 early third instar larvae, no reproducible adult cuticular defects have been observed. Since the dead animals fail to progress beyond the white pre-pupal stage, we believe the lethality is due to defective larval development rather than imaginai disc transformations. It is noteworthy that such reduced survival is associated with the ectopic expression of almost all of the homeodomain-containing proteins we have examined and also for the eye-specific homeodomain protein rough (Kimmel et al. 1990). Surviving adults are also fertile, arguing that development of the genital disc is not seriously disrupted by over-expression of cad.

The observed partial rescue of the cad¹ mutation indicates that the heat-shock construct HTcad3 produces active protein which can be detected immunologically in all cells of the early embryo. However, none of the effects of ectopic CAD expression are indicative of homeotic transformations expected to arise as a result of CAD activity. Indeed, some caution must also be taken in the interpretation of the positive developmental defects caused by the protein, since most of the results reported here are also seen with derivatives of Antp which have deletions of large portions of the amino-terminus of that protein (Gibson et al. 1990). Nevertheless, we believe the results do allow us to address a number of questions regarding cad function.

The relatively mild consequences of destruction of the CAD gradient at the syncitial blastoderm by ubiquitous expression of the protein throughout the embryo argues against the importance of the early protein gradient in the establishment of positional information along the anterior-posterior axis. Rather, it is likely that the gradient is normally generated in order to prevent CAD from functioning in the head region during early development. Given that maternally supplied cad message is required early during development in the posterior region of the embryo (Mlodzik et al. 1985; MacDonald and Struhl, 1986), some mechanism has evolved to inhibit CAD production in the anterior region (Mlodzik and Gehring, 1987b). It is possible that this mechanism is translational repression by the bicoid protein (Mlodzik and Gehring, 1987b; W. Driever, pers. comm.), which is expressed in a gradient with the opposite orientation to CAD (Driever and NÜsslein-volhard, 1988). This situation is directly comparable to the proposed role for the posterior organizing gene nanos in repression of translation of the gap gene hunchback in posterior segments (HÜlskamp et al. 1989; Irish et al. 1989; Struhl, 1989). It is interesting to note that we were unable to obtain transformants with a construct lacking the leader of the maternal mRNA and thus some of the deleted sequences in this construct may be required for translational repression of cad RNA in the anterior regions of the embryo.

Our results are consistent with a role for CAD in the segmentation process. The loss of cad function in early embryos results in considerable segmentation defects in the posterior region of the embryo (MacDonald and Struhl, 1986) at least in part as a result of improper activation of the pair-rule segmentation gene ftz. Dearolf et al. (1989) have recently provided evidence that CAD indeed acts directly on the zebra element of the ftz promoter (Hiromi et al. 1985) to stimulate transcription in the posterior stripes. We have found that overexpression of CAD in posterior regions has only limited effects on segmentation, but that ectopic expression of CAD in anterior regions severely disrupts this process. Consequently, the CAD gradient is functionally necessary for the segmentation process, but almost certainly not in a strictly concentration-dependent manner.

The fact that no clear alterations of segmental identity were produced as a result of ectopic expression of CAD during early development provides a further example of the strict hierarchy of epistatic interactions which exists amongst the homeotic selector genes (Kuziora and McGinnis, 1989; Gibson and Gehring, 1988; Gibson et al. 1990; Gonzalez-Reyes et al. 1990; Mann and Hogness, 1990). Similarly, the finding that over-expression of CAD has almost no effect on imaginai disc development reflects that, particularly in the imaginai discs, states of determination are extremely stable. Remarkable as they are for their functional specificity and regulatory power, the homeotic selector genes are clearly restricted in their ability to act outside their normal domain of expression. If homeotic selector genes ultimately regulate the activity of so-called realizator genes, which are thought to include genes involved in cell growth, division, and cytodifferentiation (Garcia-Bellido, 1975), they do so in a highly regulated manner, depending on the cellular context within which they are expressed. The fact that homeotic selector proteins do not act universally strongly suggests that they require additional transcription factors to act. Two immediate challenges thus present themselves with respect to understanding how homeotic genes direct developmental determination: what is the nature of the additional factors, and what features of each homeodomain-containing protein allow them to interact differently.

We would like to thank Pam Jones and Ken Cadigan for their interest and comments on the manuscript, Martin Müller for Fig. 2 D, Erika Wenger-Marquardt and Alexander Schier for help in preparing the manuscript and all those mentioned in the text who kindly provided antibodies and unpublished results. This work was supported by grants from the Kantons Basel-Land and Basel-Stadt, and from the Swiss National Science Foundation.