Probing gene activity in Drosophila embryos

The segmentation pattern of the Drosophila wild-type embryo is characterized by a number of easily identifiable cuticular structures. They include skeletal elements of the involuted head and ventral denticle belts that define by size, pattern and orientation the anterior part of the three thoracic and eight abdominal segments. Further landmarks such as sensory organs and the posterior tracheal endings (‘Filzkorper’), in combination with the denticle belts, allow one to un-equivocally determine the polarity and quality of each segment in preparations of the larval cuticle (see Fig. 1D).

The segmentation pattern of Drosophila is established at about blastoderm stage and it requires both maternally and zygotically active genes. Genetic analysis has identified a number of genes with zygotic activity that regulate key steps during pattern formation. Mutations in these genes cause specific defects in the segmental pattern of the embryo that allow the definition of classes of segmentation genes required for the subdivision of the embryo into segmental units (Nüsslein-Volhard & Wieschaus, 1980).

Krüppel (Kr) is a member of the gap class of segmentation genes that are characterized by a deletion of adjacent body segments in the mutant embryo. Embryos homozygous for Kr mutations die before hatching and show a unique phenotype. A total of twenty-eight alleles can be ordered into a phenotypic series. In amorphic alleles, all three thoracic and five out of eight anterior abdominal segments are deleted. Deleted segments are partially replaced by a mirror-image duplication of parts of the normal posterior abdomen (compare Fig. 1A and D) often including the dorsally located Filzkorper. Some intermediate alleles have all thoracic and four abdominal segments deleted but no duplication except that ectopic Filzkorper develop frequently close to the head region (Fig. 1B). In weaker alleles, progressively fewer segments are deleted and the prothorax is always developed (Fig. 1C-G). The weakest detectable phenotype is observed in heterozygous Kr embryos showing small defects in the denticle bands of thoracic and anterior abdominal segments. Such embryos may hatch and survive to become adults. The common motif of all alleles so far analysed is the defect in the thorax region and as the alleles get stronger, a deletion of progressively larger regions in the segment pattern up to eight segments in the strongest amorphic alleles. The interpretation of this phenotypic series is a lack of Kr function in strong alleles, increasing residual Kr⁺ activity in intermediate and weak alleles and half the normal Kr⁺ activity in heterozygous Kr embryos, which are almost normal. Aside from the fact that the Kr gene, its requirement and possible interaction with other genes for normal segmentation is interesting in its own right, it appeared sensible to use the Kr mutant embryos as a biological assay system for Kr* activity provided by injected material, and to use changes along the phenotypic Kr series as an indicator for inhibition of Kr⁺ activity in wild-type embryos being injected with gene-specific probes. This experimental design is especially promising in viewing the accessibility of Drosophila eggs and embryos for injection studies (see Anderson & Nüsslein-Volhard, 1984 for details).

Injection of wild-type cytoplasm provides phenotypic rescue in mutant Kr embryos. Embryos from a Kr SMI mating (SMI is a balancer chromosome to maintain Kr stocks) were injected. To distinguish homozygous Kr embryos from their siblings, the mutant Kr¹ chromosome was marked in all experiments with a dopadecarboxylase mutation (Dde) as it renders the cuticle and mouth parts of homozygous Kr¹ larvae unpigmented. Such embryos express the strong Kr phenol-type (Fig. 1B) which is always associated with a duplication of the sixth abdominal segment in reversed polarity (Fig. 2A). Injection of cytoplasm taken from whole wild-type embryos up to the late blastoderm stage into stages younger than late blastoderm stage had no effect on this phenotype, independent of where it was injected into the Kr mutant embryos. By contrast, when cytoplasm was taken from a middle region of blastoderm-stage wild-type embryos and transferred into a middle region of Kr embryos, up to 40 % of these developed segments with normal polarity anterior to the sixth abdominal segment (‘phenotypic rescue’, Fig. 2B). This indicates weak but significant Kr⁺ activity in Kr mutant embryos provided by the transferred cytoplasm. The weak phenotypic rescue encouraged us to analyse, under standardized conditions, the developmental profile of Kr⁺ activity in wild-type cytoplasm, its spatial distribution and the region responding to rescue in Kr mutant embryos.

Cytoplasm from the 45–55 % egg region (0 % is the posterior pole) was taken from wild-type embryos at stages between egg deposition and late blastoderm, and transferred into the same region of Kr embryos at pole cell to migration stages. As shown in Fig. 3A, phenotypic rescue was obtained with cytoplasm from blastoderm stage donors, but not with cytoplasm from younger embryos. Furthermore, the rescue response was increased by use of older cytoplasm indicating the Kr⁺ activity accumulates during blastoderm stage. However, when the highly active cytoplasm was injected into Kr embryos at late blastoderm stage, no rescue was observed. This indicates that Kr⁺ activity was injected after the phenocritical period and/or that the molecules ultimately providing Kr⁺ activity require some time to accumulate the minimum level of activity that is necessary for phenotypic rescue. In the light of (i) a correlation of Kr⁺ activity in the cytoplasm and Kr⁺ mRNA accumulation at the respective stages (see Knipple et al. 1985 for details), (ii) first indirect evidence for the Kr product being a DNA-binding protein which should be present in nuclei (see Rosenberg et al. 1986 for details) and (iii) the protein product is excluded from the injection experiments, we favour the view that Kr mRNA is transferred with cytoplasm into the Kr embryo and it requires there amplification of the Kr gene products by translation before it provides Kr rescuing activity.

Kr mutant embryos were injected, at pole cell stage, with cytoplasm taken from different regions of wild-type embryos. Cytoplasm from any region of embryos younger than syncytial blastoderm stage was ineffective (Fig. 4A), while cyto-plasm taken from the middle region but not from 0–30 % or 75–100 % of egg length of blastoderm-stage embryos, showed phenotypic rescue. This indicates that Kr⁺ activity is at least enriched, if not exclusively present, in the middle of the wild-type embryos (Fig. 4B), and possibly available at about blastoderm stage.

Cytoplasm was taken from the middle region of blastoderm-stage wild-type embryos and then transferred into different regions of Kr embryos. Phenotypic rescue response was only seen between 30–70 % of the egg length (Fig. 4C) which is the region where Kr⁺ activity accumulates in wild-type embryos. This demon-strates that both Kr⁺ activity in wild-type embryos and the rescue responsive region in Kr embryos coincide within the limits of resolution.

The position of both Kr⁺ activity accumulation and Kr⁺ requirement in the middle region of the embryo correlates with the region affected in weak Kr mutant embryos (Fig. 1), the blastoderm fate map position of thoracic and anterior abdominal anlagen (see Fig. 4D), and the region where Kr⁺ transcripts accumulate during blastoderm stage (see Knipple et al. 1985).

The finding that Kr⁺ activity can be transferred into Kr mutant embryos (its effect being to weaken the strong Kr phenotype) clearly demonstrates that low Kr⁺ activity provokes biological resonance. Considering that only about 2 % of the total egg volume was transferred and the Kr gene product is required in more than 20 % of the wild-type eggs, the effective dilution of Kr⁺ activity is by more than one order of magnitude. This means that the weak phenotypic rescue observed is within the expected range of biological response, provided that 50 % gene activity in Kr heterozygous embryos already express a weak Kr phenotype. Based on this, we felt encouraged to use Kr mutant embryos as a diagnostic tool to identify the Kr coding gene sequences on cloned genomic DNA which should cover the Kr region.

Genetics and deletion mapping placed the Kr locus in polytene chromosome band 60 F3 at the tip of the right arm of chromosome two. Clones obtained from microdissected DNA of the corresponding band facilitated the isolation of some 50 kilobases (kb) of genomic DNA in a series of overlapping clones (Fig. 5; Preiss et al. 1985 for details). The use of cytologically mapped deletions enabled us to identify DNA sequences within a 4 kb interval as those required for Kr⁺ gene function. Both molecular analysis and subsequent transcript mapping were consistent with the localization of the Kr locus in or close to the region 0 to +10 shown in Fig. 5. To identify the Kr⁺ gene function, we injected cloned DNA into mutant embryos and scored them for possible phenotypic rescue effects resulting from Kr⁺ activity provided by the injected DNA.

DNAs from various clones for most of the cloned region were injected into various regions of pole cell stage embryos. The segment anterior to the sixth abdominal segment showed normal polarity in about 40 % of the homozygous Kr embryos only after injection of clone ER3 (which contains an 18 kb segment of genomic Drosophila DNA) into the middle of the embryo. These embryos more closely resembled the intermediate rather than the strong Kr phenotype of uninjected Kr embryos. A small fraction of injected Kr embryos (about 5 %) developed more than two additional segments with normal polarity (see also Fig. 2), indicative of a substantial weakening of the amorphic phenotype.

In our most successful experiments, we injected about 300 pl of DNA (130/zg ml⁻¹) which corresponds to about 10⁶ molecules. This number is about 100 times the number of Kr⁺ gene copies of normal blastoderm-stage embryos, or 500 times the number of Kr⁺ gene copies being expressed (see Preiss et al. 1985). Earlier experiments demonstrated that about 80 % of the injected DNA is rapidly degraded and that transient transcription from the remaining DNA is more than one order below the efficiency of endogenous gene transcription.

The degree of rescue response, as in the case of cytoplasm transfer (see above) was in the expected range. These experiments therefore suggest that ER3 clone DNA contains Kr⁺ sequences and that our transient expression assay, based on phenotypic rescue, identifies the function of cloned DNA.

The above experiments used Kr mutant embryos as an indicator for Kr⁺ activity contained in the injected material which weakened the strong Kr phenotype. However, this way of identifying genes and their activity requires the prerequisite of genetical and cytological analysis in combination with a refined set of molecular techniques. This approach is, therefore, limited to a small number of biological systems where the combination of genetics, cytology and transfer of macromolecules and/or recombinant DNA transformation is accessible. To overcome this limitation and to establish a general tool for probing gene function in less fortunate systems, we designed experiments to inhibit a specific gene function. This assay involves the production of RNA containing the complementary sequences to the natural mRNA (‘anti-sense RNA’). Upon injection, both RNAs should form duplexes by hybridization in vivo and thus prevent the mRNA from being normally translated.

Several features make the Kr gene useful for assessing this sequence-specific inhibition of gene function. First, Kr gene function is defined by the number of phenotypes that can be aligned in an allelic series (see above). Second, the Kr transcripts accumulate in a defined region of the embryo. During the period of first expression, at syncytial blastoderm stage, nuclei and their surrounding cytoplasmic islands may be accessible to injected anti-sense RNA while individual cells that form during blastoderm stage may not. Third, the Kr⁺ gene is only transcribed between the syncytial blastoderm stage and the beginning of germband extension, producing a rare 2-5-kb poly(A)⁺ RNA transcript. Most of this transcript has been recovered in a 2-3 kb cDNA clone (Rosenberg et al. 1985). This cDNA was subcloned, in both orientations, into plasmid DNA containing the SP6 promoter which allows Kr sense and anti-sense RNA to be transcribed in vitro using SP6 RNA polymerase which specifically starts transcription at the SP6 promoter site (see Rosenberg et al. 1985, for a detailed description).

Wild-type embryos were injected with either sense or anti-sense RNA at the syncytial blastoderm stage. Sense RNA, which contained only part of the mRNA sequence, had no specific effect on the embryonic phenotype. By contrast, injected Kr anti-sense RNA had a dramatic effect on genetically wild-type embryos, i.e. they developed lethal Kr phenocopies (see Fig. 6) with up to 30 % frequency. Some of the phenocopies developed ectopic Filzkôrper close to the head region, as found in Kr mutants developing the intermediate phenotype. This new potential of the thoracic region to develop a structure from a dorsal-posterior position on the blastoderm fate map demonstrates unequivocally the production of Kr phenocopies in wild-type embryos. Extreme Kr phenocopies resembling the amorphic Kr phenotype (Fig. 1A), however, were not observed, indicating that Kr⁺ activity was not abolished completely (for details see Rosenberg et al. 1985).

The experiments with Kr embryos demonstrate the use of Drosophila mutant embryos for injection studies on the activity of the wild-type gene, localization of and local requirement for the gene product in vivo. Moreover, cloned DNA can be identified as functional sequences thus facilitating the delimitation of sequences being required for at least the coding region of a given gene. In this respect, a simple assay system allowed us to confirm predictions (Wieschaus, Nüsslein-Volhard & Kluding, 1984) made from genetic analysis: the middle region of the embryo accumulates Kr⁺ activity. This region coincides with the anlagen of thorax and anterior abdomen on the blastoderm fate map and appears to be most sensitive to the absence of the gene product, as reflected in the lack of thoracic and anterior abdominal segments in weak Kr alleles. The results of the injection experiments were the first indication of localized Kr⁺ gene product requirement during blastoderm stage which are possibly reflected in the localized expression of the Kr gene revealed by in situ hybridization of the molecular Kr probe to sections of embryos (see Knipple et al. 1985).

The reverse experiment, trying to inhibit a specific gene function by injection of anti-sense RNA offers a great potential not limited to Drosophila embryos. While in principle it is possible to physically isolate and clone almost any gene, it is often difficult or even impossible to ascribe a discrete function to a cloned DNA sequence, except for organisms where classical genetics has identified a gene function, and transformation of cells with foreign DNA is well established. Although far from being optimized, the potential of anti-sense RNA inhibition, especially in combination with transformation and amplification of ‘flipped gene constructs’ transcribing from strong promoters (for an example see Kim & Wold, 1985) is clear and needs no further explanation.