Analysis of gene expression during development in the homeotic mutant Contrabithorax of Drosophila melanogaster

Mutants belonging to a pseudoallelic series, the ‘bithorax’ system, produce several kinds of homeotic transformations involving the mesothoracic, metathoracic and first abdominal segments of Drosophila melanogaster (Lewis, 1963, 1964). Amongst them, the mutant Contrabithorax (Cbx) leads to the differentiation in the mesothorax (wing) of structures typical of the metathorax (haltere).

At the phenotypic level the main characteristic of Cbx is that it presents partial expression, that is, only a fraction of the cells in the wing segment show mutant phenotype, the rest of them being wild type. Since the cells are genetically identical and develop in the same developmental system, we are faced with the problem of what mechanisms are responsible for this differential behaviour. Moreover, assuming that the development of a haltere or a wing requires the function of not one but several specific genes, the expression in the Cbx mesothoracic cells of the mutant (haltere) or wild-type (wing) phenotype requires the co-ordinate induction and/or repression of specific sets of genes.

One possible explanation is that phenotypic mosaicism might stem from a determinative event somewhere in the development of the mutant disc which produces separate lineages of cells which either differentiate the mutant or the wild-type phenotype (see Postlethwait & Schneiderman, 1969, 1971).

Taking advantage of the previous genetic information about the bithorax system (Lewis, 1963,1964,1967) and how it controls the development of various segments (Morata & Garcia-Bellido, in preparation), I have made a genetic and clonal analysis of the mutant Cbx in an attempt to understand the processes that lead to the expression of this phenotypic diversity.

The main results suggest that: (1) the partial expression of the Cbx genotype is due to partial expression of the bithorax system in the mutant wing; (2) those cells which express the mutant phenotype are not related to each other: small clones generated by X-irradiation late in development often include both transformed and untransformed cells.

The following mutants were used in this work: bx³,pbx, Cbx, Ubx¹and Ubx¹³⁰. All map at 3 –58 ·8. Several alleles are known of bx and Ubx but only one for pbx and Cbx. The mutant Ubx¹³⁰ is associated with In (3LR) Ubx¹³⁰ (see Lindsley & Grell, 1968).

The translocation T(13)P115 was also used. It is an insertional translocation, discovered by E. B. Lewis, carrying a fragment of the third chromosome (including all bithorax genes) to the first chromosome proximal to forked. The aneuploid segregant Df(3)P115/+ is viable but sterile and shows the same phenotype as Ubx/+.

The mutant multiple wing hairs (mwh) was used. It is located at 3 –0 ·0 and affects trichome differentiation producing several processes per cell instead of one. It can be identified even in very small clones in both wing and haltere (Garcia-Bellido & Merriam, 1971 a; Morata & Garcia-Bellido, in preparation).

In mitotic recombination experiments, larvae were irradiated at 1000 R (R = 2-58 × 10⁻⁴ C/kg) using an X-ray machine Phillips 151 Be at standard working conditions of 150 kV at 15 mA with a 2 mm aluminium filter.

Larvae of all ages were irradiated and their ages at irradiation calculated according to the time between irradiation and pupariation.

Hatched adults were treated with hot KOH 10 % for 10 min to digest soft parts, washed in alcohol and mounted in Euparal for microscopical examination.

Almost all the information known today about the bithorax system comes from the studies of Lewis (1963, 1964, 1967) and the reader is strongly advised to consult his articles for a full description of the different mutants.

The bithorax system contains at least five loci, three of which produce recessive mutations (bithorax, bithoraxoid and postbithorax) and two others (Ultrabithorax and Contrabithorax) produce dominant ones. (Another two mutants, Contrabithoraxoid and Ultraabdominal, have been described by Lewis (1968) but as yet little information about them is available.) Although the homeotic transformations produced by these mutants affect both dorsal and ventral structures, only dorsal transformations will be considered in this report.

The mutants in the bithorax (bx) locus produce a transformation of the anterior haltere into anterior wing, the mutant postbithorax (pbx) transforms the posterior haltere into posterior wing. Thus flies homozygous for bithorax and postbithorax have four wings and no halteres. The mutant bithoraxoid (bxd) transforms the first abdominal segment into metathorax and it presents also postbithorax transformation because of its effect on the wild allele of postbithorax located to its right (Lewis, 1963, 1964).

The mutants in the Ultrabithorax (Ubx) locus are homozygous lethal, showing in heterozygous condition a slight enlargement of the halteres plus a variable and small (1 –6) number of wing-type bristles. All of them show mutant phenotype when in trans with any of the recessive mutants. It has been shown (Lewis, 1963; Morata & Garcia-Bellido, in preparation) by mosaic analysis that the phenotype of Ubx homozygous clones is that of the triple combination for the three recessive mutants, suggesting that mutants in Ubx locus produce an inactivation of the whole bithorax system. This interpretation is supported by the fact that the heterozygous deficiency for the whole bithorax system, DfP115/ + shows Ubx phenotype.

Some of the Ubx alleles, like Ubx¹³⁰, produce very extreme mutant phenotypes in trans with bx, pbx and bxd which are not distinguishable from those produced by the same mutants in trans with their deficiency. Thus bx³/Ubx¹³⁰ has the same extreme phenotype as bx³/DfP115, pbx/Ubx¹³⁰ has the same as pbx/DJP115, etc.,… (Morata, 1973). This result suggests that the mutant Ubx¹³⁰ produces a complete lack of function of the bithorax system equivalent to the physical deficiency.

All of the mutants described so far affect only the haltere or the first abdominal segment. Moreover the apparent lack of function of the bithorax system in genetic combinations like Ubx¹³⁰/DfP115 (which is lethal in individuals but viable in somatic spots) only affects the differentiation of the cells in the haltere and in the first abdominal segment and does not affect the differentiation of cells of any other segment including the wing (Morata, 1973). This fact suggests that the activity of the bithorax system is only required in the haltere and first abdominal segment.

In spite of this, there is a mutant of the series, Contrabithorax (Cbx), which affects the wing partially transforming it into haltere. The phenotype of Cbx can be explained (Lewis, 1963, 1964) by supposing that the mutant produces a derepression in the wing of the wild-type alleles of bithorax and postbithorax. As the normal function of these genes can be defined as preventing wing formation and directing the disc to develop as a haltere, their activity in the wing would lead to haltere formation there, producing Cbx phenotype.

This view is consistent with Cbx being dominant (due to the activity of the dominant alleles bx⁺ and pbx⁺) and also with some other characteristics of the mutation (Lewis, 1963).

The normal wing and haltere have been described by several authors (see, for example, Garcia-Bellido & Merriam (1971a) for the wing and Loosli (1959) andOuweneel & van der Meer (1973) for the haltere). In those cases of extreme expression, homozygous Cbx flies produce an almost complete transformation of wing into haltere (Fig. 1 A) but it is a rare event and in the majority of flies this transformation is not complete, the wing blade being a mosaic of wing and haltere differentiations (Fig. 1 B). Mature Cbx wing discs also show characteristics of both wing and haltere discs.

In the majority of cases, each cell differentiates wing or haltere structures and intermediates do not appear (Fig. 2 A). However, trichomes can be detected which are an intermediate size between those of wing and haltere (Fig. 2B). Those trichomes might be secreted by ‘hybrid’ wing-haltere cells.

Usually the mutant wing is composed of large patches of wing and haltere tissue as those shown in Fig. 2 A but together with these, other small haltere patches can frequently be detected (Fig. 2C). The minimum size of these small patches is about 30 cells.

It is interesting to note that in the case of extreme expression, the only wing structures seen are wing bristles that are completely surrounded by haltere trichomes (Fig. 2D).

The expression of the mutant genotype is variable in the different regions of the mutant wing. Fig. 3 shows the percentage of transformation in the major regions of the wing. In the posterior part (defined by the line between veins III and IV) the expression is 100 % (a complete transformation). In the anterior regions there is variable degree of transformation ranging from 90 % in C to 11 % in A’. The degree of transformation is weaker the more anterior are the regions in the wing blade.

The fact that most of the mutant wings are mosaics of wing and haltere structures allows us to make correlation between the type of haltere cells and their location in the wing. This is expressed in Fig. 4 and shows a correspondence between homologous regions along the proximo-distal axis so that the proximal haltere regions (metanotum) appear in the proximal wing regions (notum) the intermediate haltere regions (scabellum and pedicellum) appear in the intermediate wing regions (proximal and medial costa) and the distal haltere (capi-tellum) appear in distal wing (distal costa and distal wing blade). This type of homologous correspondence has also been described for other homeotic mutants (Gehring, 1966; Gloor & Kobel, 1966; Postlethwait & Schneiderman, 1971) and also appears in the opposite transformations by bithorax and postbithorax mutants (Ouweneel, 1973; Morata & Garcia-Bellido, in preparation).

It has been observed by Lewis (1963) that Ultrabithorax mutants in trans with Cbx produce a slight bithorax and postbithorax phenotypes which is stronger than that produced by the same Ubx mutants in trans with a wild-type chromosome. This observation suggests that there is some hypofunction of bx⁺ and pbx⁺ genes of the Cbx chromosome.

In order to confirm this hypothesis flies of different genetic constitutions were synthesized and their phenotypes were analysed for homeotic transformation both in wings and halteres. As shown in Table 1, flies of the genotypes Cbx/bx³ and Cbx/pbx have a Cbx phenotype as expected and also show no sign of transformation in the halteres. However, flies of the genotypes Cbx/Ubx¹, Cbx/Ubx¹³⁰ and Cbx/DfP115 presented together with Cbx phenotype a slight but clear bithorax and postbithorax phenotype in the halteres, confirming the earlier observation of Lewis and inicating that there is some hypofunction of wild-type alleles of bx and pbx when in cis with Cbx. Thus, the evidence suggests that although Cbx flies have activity of bx⁺ and pbx⁺ alleles in the mesothorax, this activity is below normal. Concentrations of bx+ or pbx⁺ products that are near to threshold could explain the varied effect on mesothoracic cells. Thus partial phenotypic expression could reflect partial genetic activity.

The clonal analysis was performed in the mutant wing disc to study the lineage relationships of the wing and haltere structures, specially to investigate whether there are independent cell lineages for each type of structure.

Larvae of the genetic constitution Cbx/mwh were irradiated at different ages and wings of hatched adults were searched for mwh spots. Because mwh is much easier to detect in the wing blade, only clones in this region were considered.

The clones found were classified into three groups: (1) forming exclusively wing structures, (2) forming exclusively haltere structures, (3) forming both wing and haltere structures.

The number and average cell numbers in clones of the three types which arose after irradiation at different developmental ages are shown in Table 2. Clones embracing wing and haltere structures were found in all ages except in flies irradiated 0-24 h before pupariation (Fig. 5). This result suggests that there are not separate cell lineages for wing and haltere during most of development. The lack of this type of clone in the flies irradiated in the latest stage may be due to the small size (Table 2) of the clones so that the probability of crossing a border between wing and haltere is very low. As shown in Table 2 as the clones are larger the proportion of clones embracing wing and haltere increases.

The continuity of all of the clones embracing wing and haltere argues against the possibility of two independent events of mitotic recombination since the probability of conjunction of two independent clones to form a continuous pattern must be very low.

The supposed influence of division rate on homeotic expression has been emphasized several times (Ouweneel, 1969; see Postlethwait & Schneiderman, 1971, for discussion on this point). The results presented here argue against this hypothesis since the sizes of clones producing haltere or wing are more or less similar, suggesting that division rate does not influence the final differentiation. It also has been shown that the normal haltere and wing had very similar if not equal division rate (Morata & Garcia-Bellido, in preparation). Furthermore, the analysis of clones differentiating wing and haltere (Table 3) shows that the proportion of wing and haltere tissue in those clones is very variable, suggesting that the division rate of the cells within a given clone does not affect the differentiation.

As shown previously (p. 30) in Cbx flies the homeotic transformation of wing into haltere is likely due to the derepression in the mesothorax of the bithorax and postbithorax genes which are those responsible for the anterior and posterior transformation respectively. However, the genetic analysis shows that despite their derepression in the mesothorax, the activity of bx⁺ and pbx⁺ when in cis with Cbx is below the level found in wild-type flies. This hypofunction may explain the partial expression of Cbx genotype; that is, some cells might have enough bx⁺ or pbx⁺ product to differentiate mutant haltere phenotype while others would not and so produce normal wing cells.

In this context it is interesting to note that in many aspects the developmental pathways leading to wing and haltere formation are similar. Both discs grow during the same larval stages, have similar division rate and have an apparently homologous process of compartmentalization (Morata & Garcia-Bellido, in preparation; Garcia-Bellido, Ripoll & Morata, 1973). These results suggest that only the presence of bx⁺ and pbx⁺ gene products is enough to direct development into haltere while their absence produces wing structures.

Moreover, it has been inferred from mosaic analysis (Morata & Garcia-Bellido,in preparation) that during most of the developmental period of the haltere disc the cells require the function of bithorax and postbithorax genes continually so that the removal of the wild-type alleles is expressed directly by a change in the state of determination of the affected cells. Only when the change is induced after 24 h before puparium formation (that is, in the late development of the disc) does the removal of the wild-type alleles fail to affect the final differentiation of the cells. This inability of the cells to respond to a change in the genetic constitution after a given moment has been called ‘perdurance’ by Garcia-Bellido & Merriam (1971 b). These results are in line with those obtained with other genes like hairy and Hairy wing (Hw) (Garcia-Bellido & Merriam, 1971 b) engrailed (Garcia-Bellido & Santamaria, 1972) and spineless-aristapedia (Postlethwait & Girton, 1974) and suggest that this continuous requirement is a general feature of the homeotic genes. One can speculate that all show perdurance in the late stages because of the persistence of the normal gene products during the remaining few divisions, rather than because of the attainment of an irreversible state of determination. This interpretation is supported by the observation of Garcia-Bellido (unpublished) that the perdurance of bithorax is lost after in vivo culture.

The fact that there is such a continuous dependence by the cells on the state of activity of their homeotic genes may help us to understand the development of the Cbx wing. Due to the hypofunction of bx⁺ and pbx⁺, the mutant cells may have an unstable state of determination that could change from one cell generation to another depending on the level of activity in each cell of bx⁺ andpbx⁺ products. What actually decides the final fate of the cells is the level of these products in the last divisions.

This instability of the state of determination is reflected by the clonal analysis since clones embracing wing and haltere structures were found in all except the last stage examined. The fact that these clones were not found in this last stage may be due either to the small size of the clones at this period (about four cells/clone, with little probability of crossing a border between haltere and wing) or to the perdurance of bx⁺ and pbx⁺ products in the last two-three divisions. If this latter possibility were true then the cells would be finally committed to differentiate wing or haltere at a very late stage of development, on average two-three divisions before the disc stops growing. In any case it is clear that during most of the development of Cbx wing disc the cells remain undetermined to produce wing or haltere.

In some respects these results are similar to those of Postlethwait & Schneiderman (1969, 1971) in their experiments with Antp^R. They were unable to find clones crossing the border between leg and antenna structures after 72 h of development, that is the third larval period. As they point out recently (Postlethwait & Schneiderman, 1974) the mutation Antp^R produces an unstable state of determination in the mutant antennal disc.

Despite this lack of stable determination in the Cbx wing disc some regions show a consistently higher degree of expressivity than others (see Fig. 4). The reasons that may lead to this region-specific expressivity are not fully understood, they may reflect different synthetic rates of bx⁺ and pbx⁺ products or different regional requirements.

Together with these regional differences some wing structures seem to be more easily transformed than others. In the cases of extreme expression it is often found (see Fig. 2D) that the only wing structures that can be observed are the bristles whereas all of the surrounding trichomes are typical of haltere. This differential behaviour of bristles versus trichomes in wing-haltere transformation has also been observed in experiments (Morata & Garcia-Bellido, in preparation) in which bx³/bx³ clones (marked with y mwh jv) were induced in normal (bx³/ + ) halteres. Clones were detected which presented wing bristles surrounded by marked (and therefore belonging to the same clone) haltere trichomes.

The causes of this difference between bristles and trichomes are not clear. The bristles appear consistently in the margin and it has been shown (Santamaria & Garcia-Bellido, in preparation) that bristle differentiation in the wing margin is dependent on interactions between cells in the dorsal and ventral surface. It is conceivable that those specific inductive stimuli demand a longer period of bithorax gene activity in the affected cells (it is known that presumptive bristle cells perform several extra divisions whereas presumptive epidermal cells do not (Lawrence, 1966), and this may uncover the hypofunction of the controlling genes in some cells. Other cells outside the margin do not receive the stimuli and remain haltere trichomes.

Another characteristic of the Cbx transformation is the proximo-distal correspondence between wing and haltere regions (see Fig. 4). This type of correspondence has been observed previously in other homeotic mutants (Gehring, 1966; Gloor & Kobel, 1966; Postlethwait & Schneiderman, 1969, 1971) as well as in bithorax mutants (Ouweneel, 1973; Morata & Garcia-Bellido, in preparation) and has been discussed in detail by Postlethwait & Schneiderman (1971). This correspondence probably indicates identical positional information (Wolpert, 1969) along at least the proximo-distal axis in the haltere and wing discs.

This identity in the positional information in homologous regions of wing and haltere probably produces the same developmental properties, whether the cells will differentiate haltere or wing depending exclusively on the activity of inactivity of the bithorax genes.

I thank Drs A. Garcia-Bellido, P. Santamaria and P. Ripoll for fruitful discussions. I am most grateful to Dr P. A. Lawrence for his comments and help during the preparation of the manuscript.