The origin of pattern duplications in segment polarity mutants of Drosophila melanogaster

The final body pattern of Drosophila depends, fundamentally, upon the activity of two classes of genes: the homoeotic genes, which control the pathway of differentiation followed by a given segment (Lewis 1978; Struhl 1982,1983) and the ‘segmentation genes’, which together define the number, position and primary pattern of each segment (Niisslein-Volhard & Wieschaus, 1980; Nüsslein-Volhard, Wieschaus & Jiirgens, 1982). Normal development requires a precise and integrated pattern of expression of all these genes, though how this is achieved is unclear.

Several of the segmentation genes belong to the ‘segment polarity’ class. Mutations at these loci result in a homologous pattern deletion in every segment and a duplication of the remaining wild-type pattern, the duplicated pattern having reversed polarity (Nüsslein-Volhard & Wieschaus, 1980). To investigate the compartmental origin of the duplicated structures, we have exploited two properties of the elements of the bithorax gene complex (BX-C): (i) the activation of different elements of the BX-C within the developing embryo follows geographical coordinates along its anteroposterior axis (Lewis, 1978); and (ii) the realms of activation of different elements of the complex are defined by compartment boundaries (Garcia-Bellido, Ripoil & Morata, 1973; Minana & Garcia-Bellido, 1982; Morata & Kerridge, 1982; Struhl, 1984; Hayes, Sato & Denell, 1984).

We have investigated the consequence of the absence of two BX-C elements, Ubx⁺ and abdA⁺, on the region of reversed polarity in some mutations of the segment polarity group with particular reference to the first abdominal segment (Al). In all the cases tested, the character of the region of reversed polarity of Al is found to depend upon Ubx⁺ but not adbA⁺, suggesting that it is derived from the anterior compartment.

The alleles of the three segment polarity loci studied were kindly provided by C. Nüsslein-Volhard and were as follows: gooseberry, Df(2R)IIX62 (zip, gsb) and Df(2R)SBl (gsb,Kr); hedgehog, hh^1J35, and cubitus interruptus dominant, ci^D. These alleles have been described previously (Nüsslein-Volhard et al. 1984; Jürgens Wieschaus, Nüsslein-Volhard & Kluding, 1984; Wieschaus, Nüsslein-Volhard & Jürgens, 1984).

The BX-C mutations employed were: Ubx¹; Df(3R)bxd¹⁰⁰, which lacks the Ubx⁺ and bxd⁺ functions; Df(3R)P9, a deficiency for the entire BX-C; and abd-A^S1 a lethal allele of the abdA complementation group to the right of Ubx defined by the distal breakpoint of Df(3R)P10 (Sanchez-Herrero, Vernos, Marco & Morata, 1985; Tiong & Whittle, unpublished). Descriptions of the BX-C mutations other than abd-A^S1 may be found in Lewis (1978).

Combinations of ciP and gsb with BX-C mutations were generated by intercrossing animals doubly heterozygous for both the segment polarity and homoeotic mutations. In the case of hh, recombinant chromosomes of the constitutions Ubx¹hh^1J35 and abd-A^S1hh^1J35 were constructed.

Eggs were collected from appropriate crosses over 12 h periods and incubated for a furthet 24 h at 25 °C. Pharate larvae were collected, dechorionated, fixed and cleared according to the procedure of van der Meer (1977). They were then mounted in a 1:1 mixture of lactic acid and Hoyers medium. Preparations were examined using dark-field and phase-contrast optics. Double mutants were identified on the basis of their novel phenotype and frequency.

The three segment polarity mutants analysed display a similar but distinguishable phenotype: the posterior region of every segment is deleted and replaced by a duplication of the remaining anterior region (see Fig. 1). The deletion is very large in hh and smaller in ci^D and gsb. Absence of the Ubx⁺ function results in the homoeotic transformation of various segments (Lewis 1978); the crucial observation for this study is that it causes the transformation of anterior Al to anterior T2, but has no effect on posterior Al in the first instar larva (Struhl, 1984 and Fig. 1). In contrast, in animals hemizygous for the abd-A^S1 mutation, anterior Al develops normally whilst posterior Al is transformed to posterior T3 (Tiong & Whittle, personal communication and Fig. 1).

In combinations of each of the three segment polarity mutations with Dfbxd¹⁰⁰ or Ubx¹, most of the denticle rows in Al, both with normal and altered polarity, appear thoracic in character (Fig. 2). This result suggests that the denticles with reversed polarity are derived, primarily, from the anterior compartment. If they belonged to the posterior compartment, they might be expected to remain abdominal in character. In fact, in a few cases with each mutant, we observed some abdominal denticles with reversed polarity amongst the transformed denticles (Fig. 2). These denticles clearly belonged to Al and appear separated from A2 by a region of naked cuticle.

In combinations of each of the three mutations with abd-A^S1, all the denticles of Al are abdominal in character (Fig. 2). This observation supports the notion that the region of reversed polarity is of anterior origin, since posterior Al is transformed to T3 in abd-A^S1 mutants. If this region were derived from the posterior compartment, it would be expected to become thoracic in the double mutant.

Our results suggest that the regions of reversed polarity in the segment polarity mutations are derived from the anterior compartment of each segment. An alternative interpretation is to suppose that each mutation causes the transformation of the posterior compartment to anterior. Since in Al the state of BX-C expression differs between compartments, (the anterior compartment requires Ubx⁺ while the posterior compartment does not), such a transformation could change the BX-C state of cells in posterior Al to that of anterior Al. If this were the case, complete absence of Ubx⁺ function should result in all the denticles in the reversed polarity region of Al being thoracic in character. Yet in each mutation studied we have observed cases of abdominal denticles in this region (for each genotype, a few deticles in one of ten double mutants). Alternatively, the level of BX-C expression in the embryo may be independent of the genetic specification of compartments. In this case, if the region of reversed polarity was derived from posterior Al it would differentiate an anterior pattern appropriate to its BX-C state (abdA⁺), namely that of anterior A2. In this case, absence of Ubx⁺ should have no effect on the reversed polarity region; however, we find that in the majority of the cases all the denticles of Al are transformed. The fact that in the absence of Ubx⁺ each mutant occasionally develops a few abdominal denticles suggests that some cells of posterior provenance can contribute to the region of reversed polarity and autonomously express their appropriate state of BX-C activity.

Since the majority of the reversed polarity region of Al requires Ubx⁺ but not abdA⁺ we conclude that it is derived from the anterior compartment of this segment. This conclusion can be extended to the other segments and is supported by the results of Garcia-Bellido & Santamaria (1972) which show that the activation of elements of the BX-C is not altered by a mutation like engrailed, which causes a partial posterior to anterior transformation. A corollary of our observations is that in the segment polarity mutants studied, the posterior compartment of each segment is either absent (hh) or greatly reduced in size (gsb and ci^D).

AMA is a recipient of a long term EMBO fellowship. We thank C. Nusslein-Volhard and R. Whittle for mutants and P. A. Lawrence, N. Baker and R. Holmgren for useful discussion and comments.