Abstract
We have developed a specific polyclonal antibody that recognizes the protein products of the abdominal-A (abdA) gene, a member of the bithorax complex of Drosophila. The normal expression domain extends from parasegments 7 to 13, in good correspondence with previous genetic and molecular results. However, while the anterior border of expression is precisely demarcated by a parasegmental boundary, the posterior border does not coincide with a lineage boundary. Within the normal domain, the expression of abd-A shows intrametameric modulation; the amount of product is higher in posterior compartments and in the most anterior cells of the anterior compartments and then gradually decreases. We have examined the effect on abd-A expression of a number of mutations, some mapping within and others outside the abd-A transcription unit. Those mapping to the transcription unit eliminate or severely reduce the amount of abd-A antigen, while those mapping outside produce an abnormal distribution of abd-A protein. Finally, we show that the abd-A gene is down-regulated in part of the Abdominal-B (Abd-B) domain, precisely in those regions where the Abd-B gene is expressed at high levels.
Introduction
The homeotic genes of Drosophila, which are clustered in the Antennapedia (ANT-C) and bithorax (BX-C) complexes (Lewis, 1978; Sánchez-Herrero et al. 1985; Mahaffey and Kaufman, 1988), establish the characteristic development (identity) of the body segments or parasegments (Martinez-Arias and Lawrence, 1985). The BX-C is composed of three genes, Ultrabithorax (Ubx), abdominal-A (abd-A) and Abdominal-B (Abd-13), and specifies the identities of parasegments 5 to 14 (Sánchez-Herrero et al. 1985; Casanova et al. 1987), which include part of the thorax and the entire abdomen. The gene abd-A is necessary for the normal identities of most abdominal segments.
One fundamental feature of the homeotic genes is that their areas of activity are defined by compartment boundaries (see Morata and Lawrence, 1977). This correlation between domains of activity and of cell lineage is of importance for it suggests that homeotic gene activities, which perdure for most of the development (Morata and Garcia-Bellido, 1976), are maintained by cell lineage. In several well-studied cases, Ubx, Antennapedia (Antp), Deformed and Sex combs reduced, (Akam and Martinez-Arias, 1985; White and Wilcox, 1985; Wirz et al. 1986; Jack et al. 1988; Mahaffey and Kaufman, 1987; LeMotte et al. 1989), the anterior border of expression coincides with an anteroposterior compartment boundary. The posterior borders are not so well defined and, in Ubx and Antp (Struhl and White, 1985; Wirz et al. 1986), they do not correspond to a known compartment line. Furthermore, the levels of expression in the posterior part of the domain seem to depend on interactions with other homeotics (Hafen et al. 1984; Struhl and White, 1985; Wirz et al. 1986).
Within their own domains, the homeotic genes present local variations of expression. Ubx, for example, is less expressed in parasegment 5 than in 6 (Akam and Martinez-Arias, 1985; White and Wilcox, 1985) and within the individual metameres, the Ubx product is less abundant in the posterior compartments (Martinez-Arias and White, 1988). In the Abd-B domain, there are quantitative and also temporal differences in different metameres (Kuziora and McGinnis, 1988; Sánchez-Herrero and Crosby, 1988; Celniker et al. 1989; DeLorenzi and Bienz, 1990).
It is clear that there must be a number of factors responsible for these complex spatial and temporal regulations. Trans acting products like those of Polycomb (Lewis, 1978) or of segmentation genes (White and Lehmann, 1986; Martinez-Arias and White, 1988) as well as regulatory elements of the homeotic genes themselves (Casanova et al. 1985, 1987; Peifer and Bender, 1986; DeLorenzi and Bienz, 1990) are involved in the process.
The expression of abd-A is not well characterized. Genetic results (Sánchez-Herrero et al. 1985; Casanova et al. 1987), in situ hybridization to tissue sections (Harding et al. 1985; Rowe and Akam, 1988) and the expression of the gene in the visceral mesoderm (Tremml and Bienz, 1989) indicate that the limits of its expression are defined by parasegmental borders, but a precise definition of its embryonic domain is still needed.
The genetic and molecular structure of abd-A is known is some detail (Karch et al. 1985; Busturia et al. 1989). It comprises around 60 kb of DNA and appears to have just one homeobox-containing transcription unit (Karch et al. 1985; F. Karch, personal communication). Some mutations map within and others outside the transcription unit, suggesting that they affect distinct functions of the gene. From genetic studies (Busturia et al. 1989), there is evidence of the existence of at least two regulatory regions, one 5’, the other 3’ to the transcription unit.
Further characterization of the different regulatory regions of abd-A is of importance in connection with the problem of the spatial regulation of the expression of the BX-C genes. Current models (Lewis, 1978; Peifer et al. 1987; Casanova et al. 1987) propose the existence of parasegment-specific regulatory elements similar to those found for the Ubx gene (Casanova et al. 1985; White and Wilcox, 1985; Beachy et al. 1985; Peifer and Bender, 1986), but presently there is little evidence for their existence in the case of abd-A (Busturia et al. 1989) and Abd-B.
By using a specific anti-nbd-A antibody, we address some of the problems about the expression and regulation of abd-A. We describe the embryonic domain of abd-A expression and how it is affected by different classes of abd-A and Abd-B mutations.
Materials and methods
(1) Mutant stocks and crosses
All mutations in the abd-A and Abd-B genes used in this work have been described in Sánchez-Herrero et al. 1985; Karch et al. 1985; Casanova et al. 1986 and Busturia et al. 1989.
To facilitate the identification of mutant embryos for abdA, we have used a TM3 hb-β-gal balancer chromosome recently made by Dr Gary Struhl. This offers the advantage that any homozygous mutant embryo in a stock balanced with TM3 hb-β-gal can be readily recognized after double staining with (β-gal and abd-A antibodies because it does not contain β-gal antigen in the hb domain. Further details on this chromsome are to be provided by Dr Struhl (in prep.).
(2) Production of the abd-A antibody
The pabexl clone transformed into K38 cells was obtained from Jeff Simon (Harvard Medical School). This clone contains a 1.1 kb Nhei-SmaX fragment from an abd-A cDNA inserted into pT7–7. A protein from this cDNA was overproduced as described in Tabor and Richardson, 1985. After induction, cells were centrifuged, washed and resuspended in 10% sucrose 20 mM Tris pH 8 and ImM EDTA. After incubation for 45 min with lysozyme, the suspension was frozen in liquid nitrogen and thawed twice, sonicated and, after adding NaCl to 0.5 M, it was centrifuged for 30 min at 10 000 rev min-1. The pellet was used to immunize two rats. After two boosts, an antiserum was obtained and tested for immunostaining. As described in the main text, this antiserum specifically recognizes abd-A protein.
(3) Antibody staining
For immunodetection, embryos were dechorionated in bleach, washed and fixed in PEM (Pipes 0.1M pH6.9; ImM EGTA; 2 HIM MgSO4), 4% formaldehyde/heptane 1:3 for 20 min with continuous shaking. Then, the vitelline membrane was removed in heptane/methanol and the embryos collected and stored in methanol. For labelling, the embryos were hydrated and permeabilized in BBT (BBS pH 7+0.1% BSA+0.1 % Triton) for 30 min and treated with the primary antibody overnight at 4°C. After several washings in BNT (BBT+2% goat serum), they were incubated with biotinylated secondary antibody (Amersham) at a dilution 1:300 for 2h at room temperature and washed in PBT (PBS+0.05% Tween 20). The embryos were reacted with the Vectastain ABC Kit (Vectorlabs) and stained with DAB (0.5mgml-1)/PBT plus 2 μl of 3% H2O2. In the case of double-labelling experiments (Lawrence et al. 1987), the same process was done twice, with the difference that the second staining included 0.02% nickel sulphate and 0.02% cobalt chloride. The embryos were dehydrated in alcohol series, embedded in Araldite/acetone and mounted in Araldite. The working dilutions for the primary antibodies were anti-abd-A 1:500, anti-invected 1:500, anti-β-gal (Cappel) 1:2000 and anti-Ubx 1:1. The anti-invected antibody was obtained from Dr T. Kornberg and the anti-Ubx from Dr R. White.
Results
(1) Expression of the abd-A protein in wild-type embryos
For the analysis of the expression of abd-A protein in different embryonic periods, we have produced a polyclonal rat anti-abd-A antibody (see Materials and methods). The specificity of the antibody is revealed by the absence of staining in the abdominal region of embryos homozygous for deficiencies of abd-A. However, in those embryos after germ band retraction, the antibody marks some cells near the brain. Although this label is clearly unrelated to abd-A, it is a useful internal control of the staining procedure in the case of abd-A mutants.
Precise limits of expression were determined by double-labelling experiments (Lawrence et al. 1987) with anti-flW-A and either anti-engrailed (Patel et al. 1989) or anti-Ubx (White and Wilcox, 1984) antibodies. This permits us to locate the abd-A domain with respect to those of en and Ubx.
The first sign of abd-A protein appears during the stages 9–10 of development (Campos-Ortega and Hartenstein, 1985), at the extended germ band period (Fig. 1A). The label is nuclear and appears simultaneously in parasegments 7–13, precisely the parasegments where the lack of function abd-A mutations produce a phenotype (Sánchez-Herrero et al. 1985; Busturia et al. 1989). Cells in the posterior compartments and those around the tracheal pits are the most intensely stained (Fig. IB). The anterior border of abd-A expression appears to coincide with parasegment 7, for, as far as we can judge, the coincidence of en and abd-A expression in Alp is complete (Fig. 1A,C). Thus, the anterior limit of abd-A expression is strictly parasegmental. Not so the posterior limit, which is segmental laterally and does not define a compartment border on the ventral side (Fig. 1C,D); some ventral cells in A8a are the most posteriorly stained cells with abd-A. Since the abd-A mutations transform the denticle belt in A8a (Sánchez-Herrero et al. 1985), we presume that these cells include the precursors of the A8a larval epidermis. The mesoderm layer also contains abd-A protein, which is out of register with respect to the ectoderm by approximately one parasegment; it extends from parasegment 8 to 12 (Fig. 1G).
Expression of abd-A protein at the extended germ band period. Embryos of pictures A – F are doubly stained for engrailed (blue) and abd-A (ochre) (A) Stage 9 – 10 embryo showing the general domain of abd-A expression from parasegments 7 to 13 (arrows). × 200. (B) Lateral view of parasegments 10 –13 in an embryo of stage 10. The expression of abd-A is restricted to posterior compartments and to cells around the tracheal pits (p). Note that there is no label in the tracheal pit of A8a. × 400. (C) Lateral view of an embryo in the process of germ band retraction. The posterior border of abd-A antigen coincides with that of en (arrow), indicating that unlike the anterior one, it is a segmental boundary. (D) Ventral view of parasegments 11, 12 and 13 of an embryo of the same age as the previous one. Some, but not all, the cells of A8a are labelled for abd-A. The pattern of expression in 13 is clearly different from 11 and 12. (E) Embryo retracting the germ band showing in the epidermal cells (arrows) the gradient of parasegmental expression. (F) Differential expression of abd-A antibody in parasegments 7 and 8. There is less amount of product in 7 than in 8. (G) Simple staining for abd-A to show the differences of abd-A expression in the ectodermal and mesodermal layers. The staining in the mesoderm extends from parasegment 8 to 12 (arrows). (H) Embryo doubly stained for Ubx (ochre) and abd-A (blue) to show the reciprocal levels of expression of these two genes. Some parasegments are indicated.
Expression of abd-A protein at the extended germ band period. Embryos of pictures A – F are doubly stained for engrailed (blue) and abd-A (ochre) (A) Stage 9 – 10 embryo showing the general domain of abd-A expression from parasegments 7 to 13 (arrows). × 200. (B) Lateral view of parasegments 10 –13 in an embryo of stage 10. The expression of abd-A is restricted to posterior compartments and to cells around the tracheal pits (p). Note that there is no label in the tracheal pit of A8a. × 400. (C) Lateral view of an embryo in the process of germ band retraction. The posterior border of abd-A antigen coincides with that of en (arrow), indicating that unlike the anterior one, it is a segmental boundary. (D) Ventral view of parasegments 11, 12 and 13 of an embryo of the same age as the previous one. Some, but not all, the cells of A8a are labelled for abd-A. The pattern of expression in 13 is clearly different from 11 and 12. (E) Embryo retracting the germ band showing in the epidermal cells (arrows) the gradient of parasegmental expression. (F) Differential expression of abd-A antibody in parasegments 7 and 8. There is less amount of product in 7 than in 8. (G) Simple staining for abd-A to show the differences of abd-A expression in the ectodermal and mesodermal layers. The staining in the mesoderm extends from parasegment 8 to 12 (arrows). (H) Embryo doubly stained for Ubx (ochre) and abd-A (blue) to show the reciprocal levels of expression of these two genes. Some parasegments are indicated.
The expression of abd-A appears to be modulated in a metameric fashion. This can be observed in embryos of stage 11, when abd-A expression is well established; the cells of the posterior compartments are strongly labelled and in the anterior compartment there is a gradient of intensity diminishing posteriorward (Fig. 1E,F). The cells just anterior to the en stripe have almost no label. With the exception of parasegment 13, this pattern is reiterated in every metamere, although parasegment 7 (and possibly 8) contains less abd-A protein than the rest. This modulation appears to be complementary to that described for Ubx (Struhl and White, 1985; Martinez-Arias and White, 1988). In Fig. 1H we illustrate this by a double staining for Ubx and abd-A’, the regions with a high level of abd-A antigen show little Ubx and vice versa.
After germ band retraction (Fig. 2A), the segmental grooves are visible and abd-A protein is observed on both sides of the grooves, from A1/A2 to A7/A8; the expression in Alp coincides with that of engrailed. In A8, however, the abd-A label only appears in some cells in the anterior compartment, as in earlier stages. In the intervening segments, we observe a repeated pattern of expression reminiscent of that seen at the extended germ band period; the strongest label appears in the posterior compartment cells and in the most anterior cells of the anterior compartment. Then it decreases gradually in the posterior direction, although there is always label above background (Fig. 2B). It is of interest that the amount of abd-A protein in A4, A5, A6 and A7 is the same or slightly higher than in A2 and A3, even though in A5, A6 and A7 there is also Abd-B protein (Celniker et al. 1989; DeLorenzi and Bienz, 1990). This suggests that there is no down-regulation of abd-A by Abd-B in these segments.
Expression of abd-A after germ band retraction. All embryos of this panel are doubly stained for en (blue) and abd-A (ochre). (A) General domain of abd-A function in parasegments 7 – 13. The gradient of parasegmental expression, reminiscent of that seen at the extended germ band period, is also visible in the epidermal cells. (B) Modulation of the expression on the ventral side of the metameres of the abd-A domain. There is intense staining in the posterior compartment and in the most anterior cell of the anterior compartment, then it decreases gradually in posterior direction. (C) Early stages in the formation of the tracheal trunk. Cells from around the tracheal pits (arrows) migrate internally and anteriorly and later fuse to form the tube. All these cells originate from anterior compartments as indicated by the lack of en staining in them. As a result of their migration, the abd-A component of the tracheal trunk is located anterior to the epidermis of parasegment 7. (D) Expression in the visceral mesoderm from parasegment 8 to 12 (indicated by arrows). Note the lack of engrailed label. (E) Expression in the embryonic central nervous system. There is not a complete coincidence in the expressions of en and abd-A; notice the en-stained neuron (arrow), which is not labelled by abd-A.
Expression of abd-A after germ band retraction. All embryos of this panel are doubly stained for en (blue) and abd-A (ochre). (A) General domain of abd-A function in parasegments 7 – 13. The gradient of parasegmental expression, reminiscent of that seen at the extended germ band period, is also visible in the epidermal cells. (B) Modulation of the expression on the ventral side of the metameres of the abd-A domain. There is intense staining in the posterior compartment and in the most anterior cell of the anterior compartment, then it decreases gradually in posterior direction. (C) Early stages in the formation of the tracheal trunk. Cells from around the tracheal pits (arrows) migrate internally and anteriorly and later fuse to form the tube. All these cells originate from anterior compartments as indicated by the lack of en staining in them. As a result of their migration, the abd-A component of the tracheal trunk is located anterior to the epidermis of parasegment 7. (D) Expression in the visceral mesoderm from parasegment 8 to 12 (indicated by arrows). Note the lack of engrailed label. (E) Expression in the embryonic central nervous system. There is not a complete coincidence in the expressions of en and abd-A; notice the en-stained neuron (arrow), which is not labelled by abd-A.
After germ band retraction, cells from around the tracheal pits migrate internally and anteriorly to form the tracheal tree (see Campos-Ortega and Hartenstein, 1985). These cells stain for abd-A but not for en indicating that they all originate from anterior compartments. As a consequence of their migration, the anterior limit of abd-A expression in the tracheal tree has moved anteriorly with respect to the epidermis (Fig. 2C). By stage 14 the ventral cord matures and shows intense staining for abd-A. From the double staining with en, it can be seen that the anterior limit of abd-A in the central nervous system (CNS) corresponds with the border of parasegment 7. However, we observe that the coincidence of the en and abd-A labels is not always cell by cell; there are two neurons (Fig. 2E) on each side, marked by en, that do not contain detectable levels of abd-A.
The expression in the visceral mesoderm can be clearly determined after germ band retraction (Fig. 2D). It follows the same rule described for the mesoderm layer; its anterior limit is offset one parasegment with respect to the epidermis, starting at parasegment 8, as has been noted before (Tremml and Bienz, 1989), and ending at parasegment 12. We find no en label in the mesoderm indicating that in this layer there is no distinction between anterior and posterior compartments, in agreement with previous observations (Lawrence, 1982).
After the germ band shortening, the staining of the amnioserosa cells can be readily studied. As these cells are not organized in segments, it is not possible to ascertain their segmental origin. We find abd-A protein in amnioserosa cells from the level corresponding to A3-A4 segments in the epidermis down to the end of the abdomen, although the most posterior cells are less strongly labelled.
(2) Different classes of abd-A mutations affect abd-A expression differently
In many of the experiments aimed to study the expression of the gene product in abd-A mutations, we have used stocks with a balancer chromosome carrying a β-gal insert (see Materials and methods). Thus embryos homozygous for the mutation can be recognized for the lack of anti-β-gal staining.
We have studied three types of abd-A mutations: (1) deletions of the gene like Df109 and DfP9’, (2) mutations mapping within the abd-A transcription unit: the alleles Ml, MX1, MX2, C26, P10 and Cl and (3) mutations mapping outside, either 5’ (iab-3277, iab-3MX47, iab-3ub4, iab-4MX4 and iab-4302) or 3’ (iab-2kiab-2MI and iab-2UabI) to the transcription unit.
Embryos homozygous for Df109 of DfP9 only show staining in the cells near the brain and nowhere else. Embryos carrying mutations in the transcription unit produce a reduction or elimination of abd-A product. The alleles M1 and MX1 show no abd-A antigen (Fig 3C), whereas in MX2 and P10 there is some, the amount of product in P10 being greater. In the case of P10, the diminution is not uniform; there is a gradient of product from almost none in A2 to higher levels in A6 and A7 segments (Fig. 3B). These observations fit very precisely with genetic results (Morata et al. 1983; Busturia et al. 1989) that showed that there is some abdA function in MX2 and P10. The latter is particularly illuminating because it was shown (Morata et al. 1983) that P10 has a segment-dependent graded phenotype, in accordance with the expression of abd-A protein. The mutations MX2 and P10 are breakpoints mapping to the extremes of the transcription unit (Karch et al. 1985), suggesting that they do not affect the coding region but remove some regulatory sequences close to it (see Discussion).
abd-A expression in (A) wild-type, (B) abd-Ap10° and (C) abd-AM1 embryos. In B, there is a gradient in the amount of antigen, which is particularly clear in the nerve cord, from low in parasegment 7 (arrow) to much higher in 12 (arrow). In C, due to the null nature of the mutation, there is no abd-A protein in the domain, only the cells near the brain are labelled.
abd-A expression in (A) wild-type, (B) abd-Ap10° and (C) abd-AM1 embryos. In B, there is a gradient in the amount of antigen, which is particularly clear in the nerve cord, from low in parasegment 7 (arrow) to much higher in 12 (arrow). In C, due to the null nature of the mutation, there is no abd-A protein in the domain, only the cells near the brain are labelled.
The mutation Cl is special in that it is a deletion with breakpoints in introns of Ubx and abd-A producing a hybrid abd-A-Ubx product (Rowe and Akam, 1988; Casanova et al. 1988) in which Ubx provides the carboxyl terminus of the protein and the homeobox. Because the hybrid product is now in part under the control of regulatory elements of Ubx (Rowe and Akam, 1988; Casanova et al. 1988), the abd-A portion of the protein should be expressed in the Ubx domain. However, we find that our antibody does not react with the Cl product, unlike the other extant anti-ahrf-A antibody (F. Karch, personal communication). Our antibody only recognizes epitopes in the carboxyl portion of the protein.
Mutations mapping outside the transcription unit cause a misregulation of abd-A function (Busturia et al. 1989). Our results support this view; all the mutations tested with the anti-aM-A antibody show presence of abd-A protein; however, the pattern of expression is abnormal in some of them.
The iab-2 mutations studied, iab-2k, iab-2MI and iab-2UabI, may produce a partial loss of abd-A product, but this is hard to ascertain. In iab-3277 and iab-3MX47, we also observe abd-A staining in the normal domain, but the amount of antigen in A3-A7 appears slightly reduced with respect to the wild-type. In the two iab-4 mutations, iab-4KlX4 and iab-4302, we do not find any alteration in abd-A expression. Unfortunately, the intensity of the abd-A stain within the A3–A7 region is quite uniform, so it is not possible to establish whether a given mutation changes, say, the A4 pattern into A3. For this reason, metamere-specific transformations may have been overlooked.
In addition to the effect in the normal domain, the mutations of the infraabdominal type frequently produce ectopic expression of the abd-A protein, in accordance with genetic data (Busturia et al. 1989). We have observed this in iab-2k, iab-2Uabl, iab-327, iab-3Uab4 and the combinations like iab-2S3/iab-3277. In general, this is observed in embryos after the germ band has retracted, but in the cases of iab-2UabI and iab-3Uab4 it occurs during germ band elongation (Fig. 4C). In iab-2k the ectopic expression is more marked and it affects the dorsolateral region of the thoracic segments (Fig. 4A) but it is observed only if the embryos are also homozygous for the su(Hw)2. This mutation is known to suppress virtually all the gypsy induced mutant alleles (Modolell et al. 1983). This suggests that the relative loss of function of iab-2k in the presence of the wild-type allele of su(Hw) is due to the presence of the su(Hw) product interfering with transcriptional control elements of abd-A, (Parkhust and Corees, 1986) and that the genuine effect of the breakpoint of iab-2k is an ectopic expression of abd-A product (Busturia et al. 1989).
Ectopic expression of abd-A protein in infraabdominal mutations. (A) Stage 11 iab-2k homozygous embryo also defective in su(Hw) function. Note the scattered cells anterior to the normal abd-A domain showing presence of the abd-A antigen (arrows). (B) Embryo of similar age of genotype iab-3Uah1 showing cells of parasegment 6 with abd-A product (arrow). (C) iab-3Uab4 embryo at the extended germ band period expressing abd-A protein in parasegment 6 (arrows).
Ectopic expression of abd-A protein in infraabdominal mutations. (A) Stage 11 iab-2k homozygous embryo also defective in su(Hw) function. Note the scattered cells anterior to the normal abd-A domain showing presence of the abd-A antigen (arrows). (B) Embryo of similar age of genotype iab-3Uah1 showing cells of parasegment 6 with abd-A product (arrow). (C) iab-3Uab4 embryo at the extended germ band period expressing abd-A protein in parasegment 6 (arrows).
(3) Interactions of abd-A with Ubx and Abd-B
As we show above, there are reciprocal amounts of Ubx and abd-A antigens in the metameres of the abd-A domain. As Ubx is down-regulated by abd-A (Struhl and White, 1985), these results suggested that the metameric expression of Ubx may be the result of, or influenced by, the levels of abd-A. We tested this idea by looking at the expression of Ubx in the absence of abd-A function. In abd-A∼ embryos, most of the Ubx modulation is lost (Fig. 5B); the posterior compartments show high levels of Ubx activity. As there is normal expression of en in abd-A∼ embryos, this experiment suggests that, in the abd-A domain, en does not directly repress Ubx\ the low level of Ubx in posterior compartments in the wild type is due to downregulation by abd-A. However, not all the Ubx modulations disappear in abd-A∼ embryos; within the anterior compartments there are variations in the amount of product, which probably reflect a dependence of Ubx and abd-A genes on other upstream genes.
Expression of Ubx in the presence (A) and absence (B) of abd-A in embryos doubly stained for Ubx and β-gal after germ band retraction. The embryo in A is of genotype abd-AMI/TM3, hb-β-gal as indicated by the anterior hb label (arrow). The one in B lacks the hb label and is therefore homozygous for abd-A”1. The Ubx protein is expressed more uniformly in B and also there is more antigen in the posterior compartments. Some parasegments are indicated.
Expression of Ubx in the presence (A) and absence (B) of abd-A in embryos doubly stained for Ubx and β-gal after germ band retraction. The embryo in A is of genotype abd-AMI/TM3, hb-β-gal as indicated by the anterior hb label (arrow). The one in B lacks the hb label and is therefore homozygous for abd-A”1. The Ubx protein is expressed more uniformly in B and also there is more antigen in the posterior compartments. Some parasegments are indicated.
The study of the interactions of abd-A with Abd-B was particularly intriguing because, unlike other cases of interactions, for examples those of Antp and Ubx, and of Ubx and abd-A, (Hafen et al. 1984; Struhl and White, 1985), there is no indication of down-regulation of abd-A by Abd-B, at least as far posterior as parasegment 12. As we point out above, the level of abd-A protein in segments A5, A6 and A7 (part of the Abd-B domain) is, if anything, higher than in A2 and A3 (part of the abd-A domain). Only in A8, corresponding to the high levels of Abd-B protein (Celniker et al. 1989; De Lorenzi and Bienz, 1990), is there a marked reduction of protein with respect to more anterior segments.
In embryos totally deficient for Abd-B functions like those homozygous for Abd-BMI (Casanova et al. 1986), there is a clear alteration of the pattern of abd-A expression. There is ectopic presence of abd-A protein in parasegments 13, 14 and part of 15, precisely those that in the wild type contain high levels of Abd-B product (Kuziora and McGinnis, 1988; Sánchez-Herrero and Crosby, 1988; Celniker et al. 1989; DeLorenzi and Bienz, 1990). This derepression, however, does not occur simultaneously; at the extended germ band stage the ectopic expression of abd-A affects only cells of parasegment 13, stopping just short of A8p (Fig. 6B). After the germ band has retracted, it extends to parasegments 14 and 15 (Fig. 6C). These results suggest the existence of different mechanisms of control of abdA expression in the posterior regions (see Discussion). They also indicate that the down-regulation of abd-A by Abd-B requires high levels of Abd-B product; the lower amounts of Abd-B protein in parasegments 10, 11 and 12 (Celniker et al. 1989; DeLorenzi and Bienz, 1990) having no detectable effect on abd-A expression. Genetic and molecular studies have demonstrated that Abd-B contains two distinct functions, m and r, which are mediated by two different proteins sharing the carboxyl terminus, (Casanova et al. 1986; DeLorenzi et al. 1988; Sánchez-Herrero and Crosby, 1988; Kuziora and McGinnis, 1988). The m function is specific to parasegments 10, 11, 12 and 13, while the r function is restricted to 14 and 15.
Expression of abd-A in the absence of the Abd-B gene. (A) Ventral view of a wild-type embryo at the extended germ band period. Parasegments 10, 11, 12 and 13 are exposed. In 13 there is abd-A protein in some cells but not in those posterior to the tracheal pit (arrow). (B) Embryo of the same age as that in A but homozygous for the Abd-BM1 mutation. The abd-A label extends to the end of the A8a compartment. Note that the cells posterior to the tracheal pit are marked with the antibody. (C) Dorsal view of a Abd-BM1 homozygous embryo after germ band retraction. The abd-A mark (arrow) has now extended to parasegment 14 and some cells of A9p (parasegment 15). Note the label in the amnioserosa cells.
Expression of abd-A in the absence of the Abd-B gene. (A) Ventral view of a wild-type embryo at the extended germ band period. Parasegments 10, 11, 12 and 13 are exposed. In 13 there is abd-A protein in some cells but not in those posterior to the tracheal pit (arrow). (B) Embryo of the same age as that in A but homozygous for the Abd-BM1 mutation. The abd-A label extends to the end of the A8a compartment. Note that the cells posterior to the tracheal pit are marked with the antibody. (C) Dorsal view of a Abd-BM1 homozygous embryo after germ band retraction. The abd-A mark (arrow) has now extended to parasegment 14 and some cells of A9p (parasegment 15). Note the label in the amnioserosa cells.
We have tested the effect on the down-regulation of abd-A of both the m and the r products. In the mutation Abd-BM5 the expression of abd-A is high in the whole of parasegment 13, including the region of A8a that in the wild type does not contain abd-A protein, and that coincides with the area of high expression of the m product. This effect is observed from the extended germ band stage. There is no expression of abd-A in parasegments 14 or 15.
In the mutants Abd-B r*12 * * * * * * * * * * 23’1 and Abd-B’uh’3, both m+r∼, there is adventitious abd-A expression in A8p, A9a and sometimes in A9p (parasegments 14 and 15, Fig. 7), but this is only observed after the germ band has retracted. This area of ectopic abd-A expression coincides with that of the wild-type function of the r element (DeLorenzi et al. 1988; Sánchez-Herrero and Crosby, 1988; Kuziora and McGinnis, 1988).
Effect of the absence of the r function of Abd-B on the expression of the abd-A gene at the shortened germ band period. Embryo of genotype Abd-Brx231(m+r−) showing expression in A8p and A9p (arrows). Notice the low level of expression in A8a compartment where abd-A is down-regulated by the m product of Abd-B.
Effect of the absence of the r function of Abd-B on the expression of the abd-A gene at the shortened germ band period. Embryo of genotype Abd-Brx231(m+r−) showing expression in A8p and A9p (arrows). Notice the low level of expression in A8a compartment where abd-A is down-regulated by the m product of Abd-B.
Discussion
Wild-type expression of abd-A and its regulation
Our results show that the embryonic domain of abd-A expression extends from parasegments 7 to 13, in good agreement with genetic data based on the phenotypes of abd-A mutations (Sánchez-Herrero et al. 1985; Busturia et al. 1989). The anterior limit of abd-A function is strictly parasegmental; double labelling for abd-A and en indicates a cell-by-cell correspondence in the Alp compartment.
The posterior limit is located within the A8a where some ventral cells express the protein, but this limit does not follow any compartment boundary. The expression of Ubx shows a similar phenomenon; only a subset of cells of the A8a compartment contain Ubx product (Fig. 1H). The posterior border of abd-A expression can be altered by removing gene products of the Abd-B gene. Under these conditions, the abd-A domain expands to parasegment 15, the last metamere of the body. Thus, in the absence of interactions with other homeotic genes, the genetic machinery of abd-A would make the gene active in parasegments 7 to 15. This is unlike Ubx which, in the absence of the trans interacting genes abd-A and Abd-B (Struhl and White, 1985; our own results), is only expressed down to parasegment 13, although the amount of product in 13 is higher than in the wild-type.
In the case of abd-A, it is remarkable that the expansion of the gene’s domain in the absence of Abd-B proceeds in a temporal order; by stage 10 (elongated germ band), it abuts the posterior limit of the m domain, and by stage 12 (after germ band shortening), it has spread to most or all of the r domain (figs 6C and 7). It suggests that there is an early activation of abd-A from parasegments 7 to 13, which is down-regulated by high levels of m protein in some cells of A8a. A second event of abd-A activation would occur later in parasegments 14 and 15, and is normally suppressed by the r protein.
Thus for both abd-A and Ubx, the limits of the posterior borders depend on interactions with other homeotic products. Since the results of these interactions may depend, like the case of abd-A and Abd-B, on threshold levels of product, it is not surprising that they do not follow lineage boundaries. The anterior borders of expression, however, are parasegmental because in those regions, for example, Ubx in parasegment 5 or abd-A in parasegment 7, these genes are not down-regulated. It will be interesting to see the original domains of activation of the homeotic genes in the absence of trans regulatory interactions. If they respond to genes (gap, pair-rule), whose main rule is to establish parasegments, these original domains should be defined by parasegmental boundaries at both ends.
The expression of abd-A also shows an internal modulation (Figs IE,F and 2A,B), repeated in every metamere; there is high level of product in the posterior compartment and in the more anterior cells of the anterior compartment, then it decreases towards the anteroposterior compartment boundary. The establishment of these different levels of product depend on the function of polarity genes. It is already known (Martinez-Arias and White, 1988) that engrailed affects the metameric modulation of Ubx, although our results suggest that the effect of en on Ubx may be mediated by abd-A. It is not clear whether this modulation in the amount of abd-A is important for the metameric identities; at least for Ubx, the replacement of the normal metameric pattern of expression by the uniform level induced by an hsp70-Ubx gene (Gonzales-Reyes and Morata, 1990) does not produce a detectable alteration in the epidermal pattern.
One novel aspect of the abd-A expression is the observation that down-regulation may depend on quantitative levels and not. only on the nature of the product; the low levels of Abd-B product in segments A5, A6 and A7 (Celniker et al. 1989; DeLorenci and Bienz, 1990) does not apparently affect abd-A expression, but higher amounts in A8 eliminate abd-A activity in most cells. This kind of observation would tend to favour models of interactions based on competition for binding sites (Gibson and Gehring,1988).
Functional structure of abd-A
Our first conclusion concerning the functional structure of abd-A is that all the functions of the gene are executed by the protein encoded by the homeobox-containing transcription unit. The mutations located within it have a very strong phenotype (Busturia et al. 1989) and, as we show here, eliminate the abd-A protein. Only mutations located in the extreme ends of the transcription unit can still produce some protein, probably because they do not alter the coding region. The mutation P10 is illustrative in this respect; it is a breakpoint near the 3’ end and abd-A protein is synthesized, (Fig. 3B) but some regulatory sequences have probably been eliminated causing an abnormal gradient of expression. The P10 protein determines a near normal development in some abdominal segments (Morata et al. 1983).
The pattern of abd-A expression in the P10 is of interest in connection with the important problem of whether there are parasegment-specific regulatory elements within the gene. All the current models on functional organization of the BX-C (Lewis, 1978; Peifer et al. 1987; Casanova et al. 1987) suggest that the BX-C is deployed in a metamere-specific manner that is mediated by non-coding, cri-acting elements within the gene. Our results with the P10 deletion are difficult to interpret in the light of this class of model; in P10, the whole regulatory region located 3’ to the transcription unit (where the iab-2 element is located) has been removed, while all the upstream sequences where the iab-3 and iab-4 elements reside (Karch et al. 1985) are intact. Yet, in P10 embryos, we still observe some abdA expression in parasegment 7 (Fig. 3B), together with a drastic reduction in parasegments 8, 9 and 10. This indicates, first, that iab-2 is not the sole element responsible for abd-A expression in parasegment 7 and, second, that it is also required in more posterior parasegments where the presence of iab-3 and iab-4 is not sufficient to promote normal levels of expression. Furthermore, most of the mutations in the regulatory regions that we have examined, whether located 5’ or 3’ the transcription unit, give rise to a ectopic expression anterior to parasegment 7. Certainly, the regions of the gene where these mutations map are important for the regulation of the abd-A transcription unit, but how they act and which are their physical domains of function still remain obscure.
Role of abd-A in the acquisition of abdominal patterns
Previous genetic and developmental analyses (Sánchez-Herrero et al. 1985; Busturia et al. 1989) have established that the abd-A gene has an important role in determining the A2 – A8 (or parasegments 7 to 13) identities; a null abd-A mutation like abd-AMI strongly transforms all these segments towards Al (parasegment 6). The transformed region includes parasegments 10, 11, 12 and 13, even though they belong to the Abd-B domain. This observation is consistent with the high level of expression of abd-A in part of the Abd-B domain (parasegments 10–12 and in part of 13) that we have found, and with published reports (Sánchez-Herrero and Crosby, 1988; Celniker et al. 1989; DeLorenzi and Bienz, 1990) indicating a relatively low level of expression of Abd-B in the same region. It seems that abd-A is the principal homeotic product for the larval abdomen, while Abd-B plays a minor role in parasegments 10 – 12. As suggested by Akam et al. 1988, abd-A may be the original ‘abdominal’ gene of the BX-C, while Abd-B is mainly responsible for terminal structures (e.g. genitalia, analia) and its influence on abdominal development is a late evolutionary acquisition and mostly limited to adult structures. In fact, the larval patterns of A5, A6 and A7 (parasegments 10–12), which contain Abd-B protein, are barely distinguishable from those of A2–A4, which do not contain it. The constancy of the larval patterns in the whole region A2–A7 is also consistent with the similar patterns of abd-A expression found in all these embryonic segments.
ACKNOWLEDGEMENTS
We thank Jeff Simon for making the pabexl clone available to us and Pat Micelli and the Dr Struhl laboratory (Columbia University-HHMI) for help in the antibody work. Thanks also to Gary Struhl for sending the TM3 hb-β-gal balancer chromosome and Ernesto Sánchez-Herrero for many discussions and suggestions. This work has been financially supported by the Dirección General de Investigación Cientifica y Técnica and by the Fundación Ramon Areces.