We report the embryonic and adult phenotypes of a number of mutations of the abd-A gene of the bithorax complex. Some of them result in loss of abd-A function in the whole abd-A domain and are usually lethal. These probably eliminate or inactivate abd-A protein products. Other mutations affect only part of the abd-A domain. These are viable, appear to map outside the abd-A transcription unit, and presumably alter the normal spatial regulation of abd-A products.

We propose a model of abd-A structure based on a protein-coding region and two cis-regulatory regions. Regulatory region 1, 3′ to the transcription unit, contains positive and negative regulatory elements. Regulatory region 2, 5′ to the transcription unit, establishes the correct level of abd-A activity in the abdominal metameres.

The bithorax complex (BX-C) specifies the identity of the portion of the body of Drosophila between and including parasegments 5 and 13 (Lewis, 1978; Sánchez-Herrero et al. 1985). The BX-C consists of three genes (Sánchez-Herrero et al. 1985; Tiong et al. 1985) termed Ültrabithorax (Ubx), abdominal-A (abd-A) and Abdominal-B (Abd-B) each one containing a homeobox (McGinnis étal. 1984; Scott and Wiener, 1984; Regulski et al. 1985). Of these, abd-A is responsible for the specific identity of parasegments 7 to 9 and is also necessary for the correct development of parasegments 10 to 13. Like the rest of the BX-C, abd-A has been cloned and there is a preliminary molecular characterization of its structure (Karch et al. 1985; Regulski et al. 1985; Rowe and Akam, 1988). A number of mutations of abd-A have been reported (Sánchez-Herrero et al. 1985; Tiong étal. 1985; Karch et al. 1985) but the genetic and phenotypic description of many of these mutations is still incomplete. There is not a genetic dissection of abd-A comparable to those of Ubx and Abd-B (Casanova et al. 1985a; Peifer and Bender, 1986; Casanova et al. 1986).

In this paper, we describe 19 mutations of abd-A, some of which have been defined molecularly by Karch et al. 1985. Our principal aim is to understand the functional components of abd-A. We find that some abd-A mutations eliminate or severely reduce abd-A functions and map at or close to the homeobox-containing unit. Other abd-A mutations map outside the transcription unit and alter the local distribution of abd-A products. The latter define regulatory regions of the gene.

abd-A mutations used and their names

The majority of the abd-A mutations used in this work (Table 1) have already been reported, but not described, in papers by Lewis, 1978; Kuhn et al. 1981; Sánchez-Herrero et al. 1985; Tiong et al. 1985 and Karch et al. 1985. Appropriate references are given in the main text. Some mutations have been induced in our laboratory while others come from other sources, especially from Welcome Bender and Francois Karch. In naming the mutations in this paper, we have followed the suggestion of Duncan (1987) and have distinguished between mutations affecting the whole domain, labelled abd-A, and those only affecting part of the domain, which are named infraabdominal (iab) followed by a number indicating the abdominal segment where they have the most prominent effect. This nomenclature, while sensible, is not devoid of problems; mutations of the second category are often difficult to classify because they affect similarly more than one metamere. Moreover, many of them exhibit dominant gain-of-function as well as recessive loss-of-function phenotypes. In renaming some of the mutants, we have considered only the recessive phenotypes.

Table 1.

abd-A mutations

abd-A mutations
abd-A mutations

We also describe here some new mutations: abd-AMX3 and iab-4MXA, which have been X-ray-induced in our laboratory, and iab-2M1, that appeared spontaneously in a Cy/Pm; Sb/TM2 stock.

Except in the cases described below, all these mutations complement those of Ubx and Abd-B, hence only the abd-A gene is altered. The exceptions are abd-AC51 which contains a bx mutation, zah-2Uabl which has a breakpoint in bxd and another in the r region of Abd-B (Karch et al. 1985; Casanova et al. 1986) and finally iab-4,5DB which is associated with a deletion removing parts of both abd-A and Abd-B. The presence of these mutations does not affect the abd-A phenotypes reported in this paper as the tester chromosomes we have used (see Results) have normal doses of Ubx and Abd-B.

In many instances in which it was necessary to discriminate the dominant from the recessive phenotype of the abd-A mutations, we used the T(1;3)P115. In this rearrangement (see Casanova et al. 1985b for description), a chromosomal portion, (DpP115) carrying a normal set of BX-C genes has been translocated to the first chromosome. It can be segregated to produce flies with supernumerary copies of the BX-C.

Measure of the homeotic transformations

The assessment of homeosis in the abdominal segments is difficult because they have similar patterns. We find it, for example, very hard to distinguish the third (A3) from the fourth (A4) abdominal segment, therefore we may have overlooked transformations of one into the other. The segment Al is clearly different from the rest because it contains in the dorsal side (tergite) bristles that are finer and shorter than those of the other segments. It also contains a characteristic cuticular arch. The posterior compartment of Al (Kornberg, 1981) is the only posterior abdominal compartment that can be identified morphologically in the adult. On the ventral side (stemitc), the Al segment has no bristles whereas the other segments have between 15 and 20. Thus any transformation of Al into a more posterior segment or vice versa can readily be distinguished.

Transformations involving segments A2–A4 are the most difficult to see because their tergites are very similar. Ventrally the sternite of A2 has a distinctive bristle pattern and contains a Wheeler’s organ. Also the pair of sensillae that appear in the anterior region are further apart in A2 than in other stemites. The A5 and A6 segments can be distinguished in male flies on the basis of the pigmentation of the tergite and the number of the bristles of the sternite of A6.

Cuticle preparations

Embryos were prepared for microscopic inspection following the standard method of Van der Meer (1977). Adult cuticle was digested with hot KOH, dehydrated with propanol and mounted in Euparal.

β-gal staining of the adult cuticle

Abdomens were cut from freshly eclosed flies fixed in 0.1 % glutaraldehyde (Fluka) and washed with PBS. They were then transferred to X-gal solution (0.2 % 5-bromo-4chloro-3-indoly-β-D-galactopyranoside in 10 mM-sodium pyrophosphate, 0.5M-NaCl, 1.0mM-MgC12, 5mM-potassium ferricyanide, 5 mM-potassium ferrocyanide) and stained for several hours at 37 °C. Stained abdomens were then dehydrated through a series of alcohols and mounted in Canada Balsam.

(1) Classification of the abd-A mutations used

All the mutations at the abd-A locus used in this work are listed in Table 1. They have been characterized with reference to the combination DpP10 Df109 (Morata et al. 1983), which we consider as a complete lack of abd-A function. This is lethal and gives a very strong abd-A phenotype. In this genotype, most of abd-A DNA, from +33–36 to +86–93 (Karch et al. 1985), is deleted. Thus no abd-A protein product can originate from this combination. It complements all Abd-B alleles, so the Abd-B gene appears to be intact. It also contains a normal Ubx gene in the DpPIO fragment.

We have divided the mutations in two groups: (1) those that have loss-of-function phenotypes similar to that of DpP10 Df109 (heretofore referred to as Df), which affect the entire abd-A domain and (2) those whose phenotypes affect only part of the abd-A domain and/or which are associated with gain of function of abd-A outside the normal domain. Those of the first group will probably eliminate or inactivate executive abd-A gene products, whereas the second group will include mutations altering regulatory elements of the gene.

(2) Mutations eliminating or reducing abd-A function in the entire abd-A domain

Except those involving abd-AM3, all the homozygotic or trans-heterozygotic combinations of these mutations were zygotic lethal. We have examined the hetnizygous embryonic phenotype of all them. Those of Df and abd-AP1° (formerly T(2;3P10)) have already been described (Morata et al. 1983).

Except in the cases of abd-AP10, abd-AC26 and abd-AM3, the embryonic phenotypes were very similar and like that of Df; abdominal parasegments 7–9 are completely transformed into parasegment 6 and parasegments 10–13 also resemble parasegment 6 but the transformation is not complete (Sánchez-Herrero et al. 1985). Of the exceptions, abd-AP10 and abd-AC26 show less extreme phenotypes, indicating that abd-A is not wholly inactive. This is supported by the finding that epidermal cell clones hemizygous for abd-Ap10 show partial adult transformations whereas in the cases of Df or abd-AM1adult transformations are more extreme (Morata et al. 1983; Sánchez-Herrero et al. 1985). Nevertheless, embryos hemizygous for either abd-Ap10 or abd-AC26 have all abdominal segments (except Al) affected and this comprehensive phenotype is probably responsible for the lethality.

The hemizygous viable mutation abd-AM3 shows a very weak embryonic phenotype; it can only be detected as some reduction in the width of the denticle belt of A2–A4 segments. This may be an indication of transformation towards Al.

The adult phenotype of abd-AM3 in combinations with the other mutations of this class is characterized by a general effect on segments A2–A6; the number of bristles in the stemites of all these segments is reduced, although the effect on A2 is more marked in the tergites where we observe the presence of some fine, short bristles and often a lateral cuticular arch. These are characteristic pattern elements of Al. In addition, we have observed in the naked territory between tergites some tubular structures which resemble the posterior compartment of T3p. By en-/3gal staining (Busturia and Morata, 1988), we have shown that these structures belong to the posterior compartment. These results suggest that in abd-AM3/Df flies there is a weak transformation of parasegments 7, 8 and 9 towards parasegment 6 (T3p-Ala).

The combinations of abd-AM3 with abd-Ap10abd-AC26, abd-AMX1, abd-AMX2 and abd-AMX3 present a weaker phenotype than the rest. This was expected for abd-Ap10 and abd-AC26 but not for abd-AMX1abd-AMX2and abd-AMX3 which exhibit strong embryonic phenotypes. It suggests that the latter mutations may not completely inactivate abd-A.

Some of these mutations have been localized on the DNA by Karch et al. 1985 (see also Duncan, 1987) as shown in Table 1. They define a DNA interval from +35 kb to +57 kb, which includes the homeobox-containing transcription unit (Karch et al. 1985; Rowe and Akam, 1988). Therefore, these mutations eliminate or inactivate abd-A protein products. The cases of abd-Ap10, abd-AC26 and abd-AMX2do not conform to this rule, for their moderate phenotypes indicate that there is abd-A product left. However, the first two mutations map around +35 kb, very close to the 3’end of the transcription unit, whereas abd-AMX2 is located around +57 kb, very close to the 5’ end. In these cases, we suspect the lesions may lie just outside the coding unit or, if inside, there still is some functional protein produced.

(3) Mutations affecting part of the abd-A domain

As we indicated above these mutations are likely to alter regulatory elements of the gene. We describe them in the order they are listed in Table 1 according to the adult abdominal segment where they have the most prominent effect, starting from A1-A2.

Infraabdominal-2K (iab-2K)

It is caused by a gypsy transposon inserted at +27.5 kb (Karch et al. 1985). The adult phenotype of iab-2K was first described by Kuhn et al. 1981 and consists of a partial transformation of the second abdominal segment (A2) towards Al. The strongest phenotype appears in hemizygous combinations or in combinations with some of the lethal abd-A alleles described above. Of the abd-A domain, only the anterior compartment of A2 is affected by the mutation; the bristles of the dorsal side (tergite) become smaller and finer, resembling those of Al (Fig. IB). The number of bristles on the ventral side (stemite) is reduced as the A2 sternite is transformed towards the Al, which has no bristles. For example, in the A2 stemite, there are 15±2 bristles in the wild type and 2±1 in iab-2K/ abd-AM1. In addition the Wheeler’s organ, a characteristic structure of A2, is nearly always absent. The homozygous iab-2K phenotype is similar but weaker.

However, we observed that iab-2K/+ and iab-2K/ iab-2K flies frequently present in the Al tergite some bristles that are longer and thicker than those of Al, suggesting a partial transformation of Al towards A2 (or other more posterior tergite). Since the A2 pattern is established by abd-A activity, which is not normally active in Al (part of the Ubx domain), this observation suggests that in iab-2K, together with a loss of function in A2, there is some ectopic activity of abd-A in Al.

This gain-of-function phenotype of iab-2K flies is dramatically enhanced if they are also made homozygous for su(Hw)2, which suppresses most of the mutations caused by gypsy transposons (Modolell et al. 1983). In the doubly mutant flies, the tergite of A2 is entirely normal but that of Al is strongly transformed towards a more posterior segment (Fig. 1C). This transformation affects preferentially the tergite. In addition, we observed the presence of abdominal tissue anterior to Al. We have tested the segmental provenance of this tissue by generating en-βgal/+; su(Hw)2iab-2K/ su(Hw)2iab-2K flies in which the adult posterior compartments can be labelled by β-gal activity (Busturia and Morata, 1988). We find that the abdominal tissue anterior to Al is located anterior to the T3p stripe and therefore is originated in the anterior compartment of the third thoracic segment.

Fig. 1.

Dorsal aspect of the first three abdominal segments in mutants affecting Al and A2 development. (A) WT. Note that the bristles in Al are shorter and finer than those of A2 and A3. There is also a characteristic cuticular arch (arrow). The posterior compartment of Al (Kornberg, 1981) is a depigmented region (dotted line) devoid of bristles and is connected smoothly with A2a. In contrast, the posterior compartments of all other abdominal segments are not normally visible as they form a fold located under the anterior compartment. (B) iab-2K/Df. The A2 segment is partially transformed into Al as indicated by the small size of the bristles. The transformation appears to be compartmental as Alp and A2p are not affected. Note in Ala (arrows) some bristles are longer than normal indicating a slight transformation towards a more posterior segment. (C) su(Hw)2iab-2K/su(Hw)2iab-2K. Almost complete transformation of Al into A2. The transformation affects Ala as indicated by the size of the bristles and the disappearance of the cuticular arch. Alp is also transformed, generating a discontinuity between Al and A2. (D) su(Hw)2iab-2K/su(Hw)2. This transformation is similar to but weaker than that of C. Both Ala and Alp are affected. (E) iab-2Uabl/abd-AM1. Reciprocal partial transformation of Al towards A2 and of A2 towards Al. As a result Al and A2 look quite similar (compare with WT). The transformation affects both Ala and Alp. Note the presence of the characteristic cuticular arch (arrows) in the two segments. (F) iab-2M1/Df. Partial transformation of A2a towards Ala. The existence of the discontinuity between A2 and A3 suggests that A2p is not transformed. Alp also remains normal.

Fig. 1.

Dorsal aspect of the first three abdominal segments in mutants affecting Al and A2 development. (A) WT. Note that the bristles in Al are shorter and finer than those of A2 and A3. There is also a characteristic cuticular arch (arrow). The posterior compartment of Al (Kornberg, 1981) is a depigmented region (dotted line) devoid of bristles and is connected smoothly with A2a. In contrast, the posterior compartments of all other abdominal segments are not normally visible as they form a fold located under the anterior compartment. (B) iab-2K/Df. The A2 segment is partially transformed into Al as indicated by the small size of the bristles. The transformation appears to be compartmental as Alp and A2p are not affected. Note in Ala (arrows) some bristles are longer than normal indicating a slight transformation towards a more posterior segment. (C) su(Hw)2iab-2K/su(Hw)2iab-2K. Almost complete transformation of Al into A2. The transformation affects Ala as indicated by the size of the bristles and the disappearance of the cuticular arch. Alp is also transformed, generating a discontinuity between Al and A2. (D) su(Hw)2iab-2K/su(Hw)2. This transformation is similar to but weaker than that of C. Both Ala and Alp are affected. (E) iab-2Uabl/abd-AM1. Reciprocal partial transformation of Al towards A2 and of A2 towards Al. As a result Al and A2 look quite similar (compare with WT). The transformation affects both Ala and Alp. Note the presence of the characteristic cuticular arch (arrows) in the two segments. (F) iab-2M1/Df. Partial transformation of A2a towards Ala. The existence of the discontinuity between A2 and A3 suggests that A2p is not transformed. Alp also remains normal.

Flies homozygous for su(Hw)2 but heterozygous for iab-2K also show a clear, though milder, transformation of Al (Fig. 1D), indicating that in the absence of the su(Hw) product, iab-2K behaves as a dominant, gain-of-function mutation. Flies homozygous for su(Hw)2 alone do not show this effect. All the genotypes generated for the study of iab-2K and two other iab-2 mutations are presented in Table 2. These observations indicate that the recessive (A2 towards Al) transformation of iab-2K flies is mostly or entirely due to the activity of the gypsy transposon, which requires the normal products of su(Hw) (Parkhust and Corees, 1986; Mazo et al. 1989). This gypsy activity probably interferes with the abd-A transcription unit in A2. Once the gypsy interference is removed by eliminating the su(Hw)+, the iab-2K breakpoint reveals its genuine, dominant gain-of-function phenotype.

Table 2.

Transformations caused by iab-2 mutations in Al and A2 segments

Transformations caused by iab-2 mutations in Al and A2 segments
Transformations caused by iab-2 mutations in Al and A2 segments

We have studied the embryonic phenotypes of iab-2K/Df and of su(Hw)2iab-2K/su(Hw)2iab-2K and found in both cases a normal pattern in the larval epidermis. Thus the phenotypes associated with iab-2K appear to be restricted to imaginal cells.

Infraabdominal-2S3 (iab-2S3)

It is associated with a breakpoint in the position +26 kb (Karch et al. 1985), very close to that of the gypsy insert of iab-2K (+27.5 kb). It also shows a transformation of A2 towards Al, but is weaker than that of iab-2K. A mild transformation can be observed in hemizygous flies as some reduction in the number of bristles in A2 stemite.

Given the proximity to the iab-2K lesion, we checked for a possible gain-of-function phenotype in Al and found that indeed the Al tergite shows some large bristles indicative of transformation towards a more posterior segment. This phenotype can be observed in heterozygous iab-2S3/+ flies but it is enhanced in homozygotes. It can also be observed in flies carrying supernumerary doses of the BX-C, indicating that the transformation of Al is a genuine gain-of-function phenotype.

Infraabdominal-2Uab1 (iab-2Uab1)

This mutation shows a dominant and a recessive phenotype. Hemizygous iab-2Uabl/Df adults show partial transformations of Al into A2 (described by Davis and Kiger, 1977 and Lewis, 1978) and of A2 into Al. As a result segments Al and A2 look very much alike (Fig. IE); other abdominal segments are not affected. The same phenotype appears in iab-2Uabl/ abd-AM1. In wh-2Uabl/+ or DpP115-, iab-2Uabl/Df adults, which have one normal abd-A gene, the transformation of Al remains unaltered whereas that of A2 is completely suppressed. The transformation of Al is also unaffected by supernumerary copies of abd-A.

All these observations indicate that iah-2Uabl produces adventitious activity of abd-A in Al, responsible for the dominant phenotype, and a reduction of the normal abd-A function in A2.

The dominant transformation of Al seen in adults is not detected in nzfi-2Uabl/+ larvae. Even homozygous iub-2Uabl embryos containing an extra dose of the BX-C (necessary to cover the recessive bxd, abd-A and Abd-B phenotypes) display a normal segment pattern.

lnfraabdominal-2M1 (wb-2M1)

It appeared spontaneously in our laboratory. In hemizygous adults, the A2 segment is transformed towards Al with an expressivity similar to that of iab-2K/ Df (Fig. IF). However, unlike iub-2K, it does not have any effect on the Al segment.

Trans combinations with lethal abd-A alleles like abd-AM1, abd-AM2, abd-A024 and abd-AC51 present similar phenotypes. There is no effect on A3 or any other segment, thus this allele has the features of a genuine iab-2 mutation of the type predicted by Lewis, 1978.

We have examined the embryonic phenotype in hemizygous condition and found it like the wild type.

Infraabdominal-3271 (iab-3277)

This mutation has been described by Karch et al. 1985. It is an X-ray-induced partial revertant of Mcp and contains a breakpoint in +64 kb.

Adults of genotype iab-3277/Df present a partial transformation of Al towards A2 and of A3-A5 towards A2. These transformations are visualized by the appearance of Wheeler’s organs, more or less defective, in all these segments. Although transformations of posterior compartments of abdominal segments are normally difficult to assess given the lack of morphological landmarks, in good preparations we have observed that the posterior compartments of A2-A5 of these flies are transformed into Alp (Fig. 2A). Thus the anterior abdomen of iab-3277/Df adults appear to be formed by Alp-A2a units, corresponding to parasegment 7. The same set of transformations is observed in iab-3277/ abd-AM1and in homozygous iab-3277adult flies. The iab-3277mutation is therefore associated with a gain of function of abd-A activity in Al and a relative loss in A3-A5. We have discriminated these two phenotypic components by generating adults with one or more copies of the normal abd-A gene; the genotypes iab-3277/ +, Dpll5’, iab-3277/ + and DpP115;, iab-3211/iab-3277 show suppression of the transformations of A3-A5 but Al remains partially transformed towards A2.

Fig. 2.

Tergites of Al and A2 in flies carrying iab mutations mapping 5’ to the transcription unit. (A) iab-3277/ Df. Ala is partially transformed into A2a as suggested by the presence of large bristles (compare with WT in Fig. 1A). Alp remains unaltered but A2p is transformed into Alp. Both Alp and A2p are marked by dotted lines. (B) iab-4302/iab-4302. A partial transformation of Ala towards a more posterior segment is indicated by a group of long bristles (arrow). Alp is also transformed; the depigmented area smoothly continuous between Al and A2 is replaced by cuticular folding characteristic of the posterior compartments of A2-A6 segments.

Fig. 2.

Tergites of Al and A2 in flies carrying iab mutations mapping 5’ to the transcription unit. (A) iab-3277/ Df. Ala is partially transformed into A2a as suggested by the presence of large bristles (compare with WT in Fig. 1A). Alp remains unaltered but A2p is transformed into Alp. Both Alp and A2p are marked by dotted lines. (B) iab-4302/iab-4302. A partial transformation of Ala towards a more posterior segment is indicated by a group of long bristles (arrow). Alp is also transformed; the depigmented area smoothly continuous between Al and A2 is replaced by cuticular folding characteristic of the posterior compartments of A2-A6 segments.

Infraabdominal-3Uab4 (iab-3Uab4)

This mutation maps at +63 (Karch et al. 1985) and also has a dominant (Lewis, 1978) and a recessive phenotype. iab-3Uab4/Df adults show a phenotype similar to that of iab-3277/Df-, Al is transformed towards A2 and A3-A5 become like A2. The posterior compartments are also transformed into Alp, although not as clearly as in iab-3277/Df. The transformation of Al is stronger than in iab-321 and in addition there is frequently a group of abdominal-like bristles located anterior to Al. By homology with the su/(Hw)2iab-2K/su(Hw)2iab-2K phenotype, we interpret these bristles as originated in T3a, thus suggesting that the iuh-3Uab4 phenotype extends to the thorax. We have distinguished the dominant and recessive phenotypic components by adding extra copies of abd-A. DpP115;, iab-3Uab4/Df, iab-3Uab4 and DpP115;, iab-3Uab4/+ have A3-A5 segments wild type but the transformations of T3 and Al remain as in iab-3Uab4/Df.

Infraabdominal-4302 (iab-4302)

Karch et al. (1985), described the hemizygous phenotype and also the molecular lesion, a breakpoint at +86 kb. The segment A4 is transformed into A3 and A2 is transformed A3 as revealed by the loss or abnormal development of the Wheeler’s organ and the alteration of the bristle pattern of the A2 stemite. We have confirmed their observations for iab-4302/Df adults although the pigmentation pattern of A4 is variable and its transformation sometimes difficult to assess. We find the same phenotype in iab-4302/abd-AM1 and in iab-4302/ abd-AM2 flies. Homozygous iab-4302 adults also show a similar phenotype but, in addition, the Al segment, particularly the posterior compartment, is transformed towards a more posterior segment. The same phenotype is seen in Al of DpP115;, iab-4302/ iab- 4302 flies (Fig. 2B).

Infraabdominal-4MX4 (iab-4MX4)

It has recently been X-ray induced in our laboratory. It has some features in common with iab-4302 so we have labelled it as iab-4. iab-4MX4/Df adults show a transformation of Al, especially Alp, towards a more posterior segment. The sternite of A2 frequently has the Wheeler’s organ abnormal or absent. The pigment pattern of the A4 tergite is modified and resembles A3. The hemitergites do not fuse well and leave a cleft in the medial region.

The effects seen in Al and A2 are not recessive as they appear in heterozygous iab-fMX4/ + and in Dpll5-, iab-4MX4/Df adults. This indicates that iab-4MX4 produces some gain of function of abd-A in Al and A2, together with a relative loss in A4. Unfortunately, the abd-AMX4 chromosome is homozygous lethal, most probably due to a second site mutation, so we did not study the effect of two doses of the gain-of-function phenotype of the mutation.

Infraabdominal-4,5DB> (iah-4,5DB> )

This mutation is caused by a DNA deletion from +83 to + 113 (Karch et al. 1985). The deletion extends into Abd-B as it removes a piece of DNA where several Abd-B mutations map (Karch et al. 1985). However, the tester abd-A mutant chromosomes we have used contain Abd-B+ so the Abd-B phenotypical component should have been eliminated.

Flies of the genotype iab-4,5DB> /Df or iab-4,5DB> / abd- AM1 show Al slightly modified towards a more posterior segment; also the Wheeler’s organ is sometimes abnormal suggesting a transformation of A2. In males, the A5 tergite is depigmented resembling that of A4. However, iab-4,5DB> / + males also present a partially depigmented A5 tergite suggesting that the A5 depigmentation may be in part due the haplo insufficiency of the Abd-B fragment missing in the iab-4,5DB> chromosome. The transformations in Al and A2 are stronger in homozygous flies but are not affected by supernumerary doses of abd-A+ indicating that the effects seen in Al and A2 are due to gain of function of abd-A.

(4) Complementation studies

The abd-A locus is defined as a single complementation group (Sánchez-Herrero et al. 1985). The abd-A mutations do not complement among themselves and also fail to complement the iab alleles.

Complementation analyses among the iab alleles are complicated by the fact that, as we show above, many of them exhibit a combination of gain- and loss-of-function phenotypes. For complementation we have only considered the latter.

Trans-heterozygous combinations of iab-2 alleles show the characteristic transformation of A2 towards Al, except in iab-2S3/iab-2Uabl flies that show no detectable effect in A2 (Al is strongly transformed to A2).

The combinations of iab-2 with iab-3 alleles show full complementation for the recessive phenotypes of both classes. Similarly trans combinations of iab-2 and iab-4 have no recessive phenotype. The combination of the two iab-3 alleles, iab-3277/ iab-3Uab4 produces the same mutant phenotype as homozygous iab-3277 flies. We find it very difficult to decide on the complementation of iab-3 and iab-4 alleles given the variability of the recessive phenotype but there is no transformation in the A3 segment, so they seem to complement as to this phenotype.

The trans combinations of iab alleles often produce a strong enhancement of the gain-of-function phenotypes in Al and A2 that many of them have. This is particularly clear in combinations involving iab-2K, iab-2s3, iab-3277, tab-4302 and iab-4MX4-, For example, 100 % of the iab-2S3/iab-3277 flies show transformation of Al into A2 and a similar result is found for iab-2K/iab-3277. In the genotype iab-4303/iab-4™K 4, Ala is transformed in about 50 % of the cases and Alp in the 100%. The Wheeler’s organ of A2 is always absent.

Mutations altering the normal function of the abd-A gene can be classified into two broad categories. (1) Those that affect the whole abd-A domain producing a general loss of abd-A function. We designate them as abd-A mutations. (2) Mutations that affect only part of the abd-A domain and presumably modify the spatial distribution of abd-A function. These are viable and we name them iab mutations.

The abd-A mutations are nearly always lethal, but we have detected phenotypic differences among them. There is a group formed by ahd-AM1, abd-AM2, abd-AD24 and abd-AC51 whose phenotypes are indistinguishable from that of the deletion of the gene; both in their embryonic phenotypes and in the adult phenotypes of their combinations with iab alleles. They therefore eliminate or severely inactivate abd-A gene products.

A second group is composed of the mutations abd-AP10abd-AC26, abd-AM3, abd-AMX1, abd-AMX2and abd-AMX3 that produce weaker segmental transformations either in embryonic or in adults. However, the whole abd-A domain is affected and probably this is, except in abd-AM3, incompatible with viability.

The localization of the abd-A mutations defines a region between +33 kb and +57 kb in the DNA map of Karch et al. (1985). This coincides extremely well with the region of the gene encoding the homeobox-containing transcription unit, which spans from +33 kb to +56.5−57kb (F. Karch, personal communication). Most probably abd-A mutations directly interfere with this unit. This in turn strongly suggests that all functions of the gene are mediated by the protein products encoded by it. If there are other transcription units, they should have a subsidiary (or no) role.

We note that four not-null abd-A alleles map at either one of the ends of the transcription unit; abd-Ap10 and abd-AC26 around +33, and ui>d-AMX1 and abd-AMX2 at the interval +55-58 kb. It might be that in these cases the protein coding unit is not directly affected or, if it is, it may produce a form of truncated protein that still retains some function. Our results predict that these chromosomes permit some abd-A function that should be detectable once an anti-ahd-A antibody is available.

The iab mutations are viable and can be subdivided into phenotypic classes, although the distinction is not always neat. They always fail to complement abd-A mutations indicating that the function(s) defective in iab alleles act in cis on the abd-A transcription unit, in the manner that for example bxd acts in cis on Ubx transcription (Hogness et al. 1985).

It also seems that iab-3 and iab-4 mutations complement as to the recessive transformation of A3 into A2, although the A4 to A3 transformation may remain. Thus a minimum of two distinct functions are defined, one located 3′ and the other 5′ the transcription unit.

Which is the nature of the functions defined by iab alleles? In our view, they have a regulatory role connected with the distribution of abd-A products in the different abdominal metameres. This is based on two lines of argument. First, in iab mutants, part of the abd-A domain develops normally while other parts show loss of abd-A function, i.e. abd-2M1 transforms A2 but A3 and A4 are normal; in iab-3277, A2 is normal but A3 and A4 are transformed. This is expected if the distribution of abd-A products, rather that the function of the abd-A transcription unit is abnormal. The second argument is that many of these mutations exhibit a gain-of-function phenotype in Al, outside the abd-A domain. This again suggests an abnormal distribution of abd-A products.

One important question is whether there exist within the iab regions individual elements responsible for abd-A expression in specific metameres, as proposed by recent models (Casanova et al. 1987 ; Peiferet al. 1987). i.e. whether there exist iab-2, iab-3, iab-4 etc. as distinct genetic elements. The recessive phenotype of iab-2 mutants suggests a metameric specific control (it is not clear whether compartmental, segmental or parasegmental). Also the observation that iab-4 mutations complement the recessive phenotype of iab-3 alleles suggests that they affect distinct regulatory regions, perhaps metameric specific, at the 5′ side of the gene. However, the iab-3 mutations show in parasegments 8 and 9 a similar transformation to parasegment 7. This would not be expected if there are distinct elements for parasegment 8 (iab-3) and 9 (iab-4). Besides we have shown that iab-3 and iab-4 mutations give rise to ectopic expression in Al and A2 (or parasegments 6 and 7) and this is not easily compatible with models based on metameric specific regulatory elements. The simplest view about the iab-3 and iab-4 mutations is that they cause a general misregulation of abd-A expression. This results in an abnormal distribution of abd-A products, which now may appear inappropriately in the Ubx domain or at abnormally high levels in parts of the abd-A domain (Alp and A2a). This phenomenon would produce a gain-of-function, usually dominant, phenotype. Other parts of the abd-A domain (A2p, A3 and A4) would have a reduced level of product and consequently a loss-of-function phenotype. This interpretation has the advantage that offers a single explanation for a complex mutant syndrome.

One peculiar feature of many of these mutations is that the dominant transformations are restricted to adult structures; larval segments appear normal. This might be interpreted as indicating differential regulation for larval and adult cells. However, another possibility is that any dominant transformation affecting larval pattern may result in lethality. This phenomenon would result in the selection of mutations having little or no effect on larval patterns.

A model of abd-A structure

Our genetic results, together with the available information about the molecular mapping of abd-A mutations (Karch et al. 1985) suggest a model of abd-A structure outlined in Fig. 3. The model we propose is of necessity very general at this stage and will have to be refined in the light of subsequent molecular and genetic information.

Fig. 3.

Model of abd-A structure. The arrows point to the approximate positions of the lesions associated with the mutations regardless of whether they are breakpoints, inser-tions or deletions. The gene contains one transcription unit encoding all the executive products of the gene, and two regulatory regions connected with the spatial distribution of the gene products. Regulatory region 1 contains some control elements to prevent abd-A expression anterior to its normal domain and may also contain other elements to establish the appropriate level of abd-A product in A2. Regulatory region 2 appears to be connected with the general expression of abd-A in the abdomen since mutations in this region lead to a general alteration in the pattern of abd-A function (see text for details) affecting several segments, hb indicates the approximate position of the homeobox. The positions of the mutations abd-AS2 and iab-445, which have not been studied in this work, have been included as their phenotypes (Karch et al. 1985) match those of mutations in the same region.

Fig. 3.

Model of abd-A structure. The arrows point to the approximate positions of the lesions associated with the mutations regardless of whether they are breakpoints, inser-tions or deletions. The gene contains one transcription unit encoding all the executive products of the gene, and two regulatory regions connected with the spatial distribution of the gene products. Regulatory region 1 contains some control elements to prevent abd-A expression anterior to its normal domain and may also contain other elements to establish the appropriate level of abd-A product in A2. Regulatory region 2 appears to be connected with the general expression of abd-A in the abdomen since mutations in this region lead to a general alteration in the pattern of abd-A function (see text for details) affecting several segments, hb indicates the approximate position of the homeobox. The positions of the mutations abd-AS2 and iab-445, which have not been studied in this work, have been included as their phenotypes (Karch et al. 1985) match those of mutations in the same region.

All the protein products responsible for abd-A functions are encoded in the homeobox-containing transcription unit which is flanked by two cis-acting regulatory regions, 1 and 2.

The transcription unit is responsible for all the executive products of the gene. Therefore mutations that impede its function present a null phenotype equivalent to the deletion of the entire gene. Little is still known about the different components of this unit, number, size and disposition of exons and introns etc.

Regulatory region 1 is located 3′ to the transcription unit and is defined by the iab-2K and iab-2S3 breakpoints. This part of the gene is connected with the repression of abd-A activity anterior to its domain, that is, it acts as a negative regulatory element. Another BX-C gene, Ubx appears to have similar components localized in a homologous position, defined by the mutations Cbx3 and CbxTWT (Bender et al. 1983).

However, both iab-2K (in the presence of su(Hw)+) and iab-2S3 show a reduction of abd-A function in A2, suggesting they also interfere with another function necessary to promote abd-A activity in A2 and that should also be located 3’ to the transcription unit. The eventual localization of the DNA lesions of iab-2M1 and mh-2Uabl will help to further define these regions.

Regulatory region 2 extends for about 25 kb and is defined by the iab-3 and iab-4 mutations which provoke abnormal levels of abd-A function in A1-A4 segments. Unlike regulatory region 1, whose role appears to be restricted to the A2 segment and part of the Ubx domain, the role of regulatory region 2 appears to be more general. It is connected with the general regulation of abd-A products in the abdomen. It is still an open question whether there are distinct components within it. Further genetic and molecular analysis of this region is still required.

We thank Peter Lawrence for comments in the manuscript and the Agencia Nacional de Evaluación and the Ramón Areces Foundation for financial suppport.

Bender
,
W.
,
Akam
,
M.
,
Karch
,
F.
,
Beachy
,
P. A.
,
Peifer
,
M.
,
Spierer
,
P.
,
Lewis
,
E. B.
and
Hogness
,
D. S.
(
1983
).
Molecular genetics of the bithorax complex in Drosophila melanogaster
.
Science
221
,
23
29
.
Busturia
,
A.
and
Morata
,
G.
(
1988
).
Ectopic expression of homeotic genes caused by the elimination of the Polycomb gene in Drosophila imaginai epidermis
.
Development
104
,
713
720
.
Casanova
,
J.
,
Sánchez-Herrero
,
E.
,
Busturia
,
A.
and
Morata
,
G.
(
1987
).
Double and triple mutant combinations of the bithorax complex of Drosophila
.
EMBO J
.
6
,
3103
3109
.
Casanova
,
J.
,
Sánchez-Herrero
,
E.
and
Morata
,
G.
(
1985a
).
Prothoracic transformation and functional structure of the Ulirabithorax gene of Drosophila
.
Cell
42
,
663
669
.
Casanova
,
J.
,
Sánchez-Herrero
,
E.
and
Morata
,
G.
(
1985b
).
Contrabithorax and the control of spatial expression of the bithorax complex genes of Drosophila
.
J. Embryol. exp. Morph
.
90
,
179
196
.
Casanova
,
J.
,
Sánchez-Herrero
,
E.
and
Morata
,
G.
(
1986
).
Identification and characterization of a parasegment specific regulatory element of the Abdominal-B gene of Drosophila
.
Cell
47
,
627
636
.
Davis
,
R. C.
and
Kiger
,
J. A.
(
1977
).
A clonal analysis of tergite development in Drosophila of Ultraabdominal and Paradoxical Genotypes
.
Devi Biol
.
58
,
114
123
.
Duncan
,
I.
(
1987
).
The bithorax complex
.
A. Rev. Genet
.
21
,
285
319
.
Hogness
,
D. S.
,
Lipshitz
,
H. D.
and
Beachy
,
P. A.
(
1985
).
Regulation and products of the Ubx domain of the bithorax complex
.
Cold Spring Harbor Symp. Quant. Biol
.
50
,
181
194
.
Karch
,
F.
,
Weiffenbach
,
B.
,
Peifer
,
M.
,
Bender
,
W.
,
Duncan
,
I.
,
Celniker
,
S.
,
Crosby
,
M.
and
Lewis
,
E. B.
(
1985
).
The abdominal region of the bithorax complex
.
Cell
43
,
81
96
.
Kornberg
,
T.
(
1981
).
Compartments in the abdomen of Drosophila and the role of the engrailed locus
.
Devi Biol
.
86
,
363
372
.
Kuhn
,
D. T.
,
Woods
,
D. F.
and
Cook
,
J. L.
(
1981
).
Analysis of a new homeotic mutation (iab-2) within the bithorax complex in Drosophila melanogaster
.
Mol. gen. Genet
.
181
,
82
86
.
Lewis
,
E. B.
(
1978
).
A gene complex controlling segmentation in Drosophila
.
Nature, Lond
.
275
,
565
570
.
Mazo
,
A. M.
,
Mizrokhi
,
L. J.
,
Karavanov
,
A. A.
,
Sedkov
,
Y. A.
,
Krichevskaja
,
A. A.
and
Ilyin
,
Y. V.
(
1989
).
Suppression in Drosophila: su(Hw) and su(f) gene products interact with a region of gypsy (mdg4) regulating its transcriptional activity
.
EMBO J
.
8
,
903
911
.
McGinnis
,
W.
,
Levine
,
M. S.
,
Hafen
,
E.
,
Kuroiwa
,
A.
and
Gehring
,
W.
(
1984
).
A conserved DNA sequence in homeotic genes of Drosophila Antennapedia and bithorax complexes
.
Nature, Lond
.
308
,
428
433
.
Modolell
,
J.
,
Bender
,
W.
and
Meselson
,
M.
(
1983
).
Drosophila melanogaster mutations suppressible by the Suppressor of Hairy wing are insertions of a 7,3 kilobase mobile element
.
Proc. natn. Acad. Sci. U.S.A
.
80
,
1678
1682
.
Morata
,
G.
,
Botas
,
J.
,
Kerridge
,
S.
and
Struhl
,
G.
(
1983
).
Homeotic transformation of the abdominal segments of Drosophila caused by breaking or deleting a central portion of the bithorax complex
.
J. Embryol, exp. Morph
.
78
,
319
341
.
Parkhust
,
S. M.
and
Corces
,
V. G.
(
1986
).
Interactions among the gypsy transposable element and the yellow and suppressor of Hairy wing loci in Drosophila melanogaster
.
Molec. cell. Biol
.
6
,
47
53
.
Peifer
,
M.
and
Bender
,
W.
(
1986
).
The anterobithorax and bithorax mutations of the bithorax complex
.
EMBO J.
5
,
2293
2303
.
Peifer
,
M.
,
Karch
,
F.
and
Bender
,
W.
(
1987
).
The bithorax complex: control of segment identity
.
Genes and Development
1
,
891
898
.
Regulski
,
M.
,
Katherine
,
H.
,
Kostriken
,
R.
,
Karch
,
F.
,
Levine
,
M.
and
McGinnis
,
W.
(
1985
).
Horneo box genes of the Antennapedia and Bithorax Complex of Drosophila
.
Cell
43
,
71
80
.
Rowe
,
A.
and
Akam
,
M.
(
1988
).
The structure and expression of a hybrid homeotic gene
.
EMBO J
.
7
,
1107
1114
.
Sánchez-Herrero
,
E.
,
Vernos
,
L
,
Marco
,
R.
and
Morata
,
G.
(
1985
).
Genetic organization of the Drosophila Bithorax complex
.
Nature, Lond
.
313
,
108
113
.
Scott
,
M. P.
and
Weiner
,
A. J.
(
1984
).
Structural relationships among genes that control development: sequence homology between the Antennapedia, Ulirabithorax and fushi tarazu loci of Drosophila
.
Proc. natn. Acad. Sci., U.S.A
.
81
,
4115
4119
.
Tiong
,
S.
,
Bone
,
L. M.
and
Whittle
,
J. R. S.
(
1985
).
Recessive lethal mutations within the bithorax complex of Drosophila
.
Molec. gen. Genet
.
200
,
335
342
.
Van Der Meer
,
J. M.
(
1977
).
Optically clean and permanent whole mount preparation for phase constrast microscopy of cuticular structures of insect larvae
.
Drosophila Information Service
52
,
160
.