Homeotic (or Hox) genes are key determinants in specifying the anteroposterior axis of most animals. The temporal and spatial expression of these genes requires the presence of large and complex cis-regulatory regions. The Abdominal-B Hox gene of the bithorax complex ofDrosophila is regulated by several infraabdominal domains,which determine Abdominal-B expression in abdominal segments A5 to A9(parasegments 10 to 14). Some of the infraabdominal domains have been characterized, including an infraabdominal-8 domain, which has been located 3′ to the Abdominal-B transcription unit. We have analyzed the expression and mutant phenotype of a P-lacZ element inserted close to the Abdominal-B m origin of transcription and of derivatives of this transposon. Some of these derivatives represent a particular class of mutations in the bithorax complex, because they transform the eighth and ninth abdominal segments without affecting more anterior metameres. The analysis of these mutations and of transformants carrying sequences upstream the Abdominal-B m transcription unit has allowed us to define a new infraabdominal-8 regulatory region, located 5′ to the Abdominal-B transcription unit, and has helped to characterize better the complex regulation of the Abdominal-Bgene.
Homeotic (Hox) genes determine body structures along the anteroposterior axis in most organisms. In Drosophila, the homeotic genes are located in two clusters, the bithorax complex (BX-C) and the Antennapedia complex(ANT-C) (Kaufmann et al., 1990; Lewis,1978; Sánchez-Herrero et al., 1985), which generate segmental diversity in the embryo and the adult. This segmental specification is achieved by the particular temporal and spatial distribution of Hox proteins. Such a fine regulation requires the presence of large cis- regulatory regions(Bender et al., 1983;Karch et al., 1985; Kaufmann et al., 1990).
The Abdominal-B (Abd-B) Hox gene of the BX-C specifies abdominal segments A5 to A9 [parasegments (PS) 10 to 14](Casanova et al., 1986;Sánchez-Herrero et al.,1985; Tiong et al.,1985). It codes for two proteins that share the C-terminal amino acid sequence, including the homeodomain, but differ in the N-terminal region:Abd-B M (or Abd-B I) protein, which is expressed in PS10-13; and Abd-B R (or Abd-B II) protein, which is present in PS14(Boulet et al., 1991;Celniker et al., 1989;DeLorenzi and Bienz, 1990;Zavortink and Sakonju, 1989)(see Fig. 1A). Mutations that affect the Abd-B M protein complement those affecting the Abd-B R protein, and both types of mutations fail to complement Abd-B mutations that eliminate the two products (Casanova et al.,1986; Celniker et al.,1990; DeLorenzi and Bienz,1990). Abd-B M expression starts in PS13 and is later observed,successively, in PS12, PS11 and PS10, with increasing levels in more posterior parasegments (Celniker et al.,1989; DeLorenzi and Bienz,1990). This indicates that there are regulatory sequences controlling Abd-B expression in these parasegments.
The Abd-B m regulatory region extends over more than 50 kb 3′ to the Abd-B transcription unit, and is genetically subdivided into several infraabdominal (iab) domains, fromiab-5 to iab-8 (Barges et al., 2000; Galloni et al.,1993; Gyurkovics et al.,1990; Karch et al.,1985; Lewis, 1978;Zhou et al., 1999). Deletions of these domains cause transformations of the corresponding parasegment into the next, more anterior one (Galloni et al., 1993; Mihaly et al.,1997). DNA elements within these domains, when fused to thelacZ gene, direct β-galactosidase expression in the embryo with precise parasegmental anterior limits of expression(Barges et al., 2000;Busturia and Bienz, 1993;Zhou et al., 1999). Similarly,P-lacZ elements inserted in a specific iab region showβ-galactosidase signal restricted to particular parasegments(Barges et al., 2000;Bender and Hudson, 2000;Galloni et al., 1993;McCall et al., 1994;Zhou and Levine, 1999). Therefore, the iab domains seem to represent autonomous units that control Abd-B M expression from A5/PS10 to A8/PS13, thus specifying these metameres (Boulet et al., 1991;Celniker et al., 1990;Galloni et al., 1993;Gyurkovics et al., 1990;Mihaly et al., 1997;Sánchez-Herrero and Akam,1989; Sánchez-Herrero,1991).
The iab domains are separated from each other by boundary elements that isolate them, thus preventing abnormal activation of Abd-B in more anterior segments (Barges et al.,2000; Gyurkovics et al.,1990; Hagstrom et al.,1996; Hagstrom et al.,1997; Mihaly et al.,1997; Zhou et al.,1996; Zhou et al.,1999; Zhou and Levine,1999). Deletions of these elements cause posteriorward transformations (Barges et al.,2000; Gyurkovics et al.,1990; Mihaly et al.,1997), owing to the inappropriate expression of Abd-B in a certain parasegment with the levels and distribution of a more posterior one(Celniker et al., 1990;Galloni et al., 1993;Sánchez-Herrero, 1991). In some cases, it has been shown that boundary regions are physically close to sequences that respond to trans-regulators of the Polycombtype (Polycomb response elements, or PREs), which restrict homeotic gene expression throughout development(Brock and van Lohuizen, 2001). The combined activity of PRE and boundary elements maintains the particularAbd-B expression of each parasegment.
Of all the abdominal segments determined by the Abd-B M protein, the A8/PS13 is clearly different from the rest in the set of structures that derive from it, both in the embryo or in the adult, such as the posterior spiracles or the female genitalia (Campos Ortega and Hartenstein, 1985;Nöthiger et al., 1977). An iab-8 regulatory region, which directs PS13 specific expression in the embryo, has been identified 3′ to the Abd-B transcription unit (Barges et al., 2000;Zhou et al., 1999)(Fig. 1A). The iab-8domain is separated from the iab-7 domain, which is located more proximally, by a Fab-8 boundary that also contains a PRE sequence adjacent to it (Barges et al.,2000; Zhou et al.,1999; Zhou and Levine,1999). However, there are no known mutations affecting just the A8, and rearrangements that affect the iab-8 domain do not transform this segment (Gyurkovics et al.,1990; Hendrickson and Sakonju,1995; Hopmann et al.,1995; Tiong et al.,1987).
We have made use of a P-element insertion in the 5′ region of theAbd-B gene to isolate novel mutations that affect the A8 segment and we have characterized a putative iab-8 region in the 5′ region of Abd-B. Our results indicate that iab-8 sequences 5′and 3′ to the Abd-B m transcription unit contribute to the regulation of Abd-B in the A8 segment.
MATERIALS AND METHODS
The mutations used are described elsewhere(FlyBase, 1999). TheAbd-B mutations employed are: Abd-B null mutations;Abd-BM1, Abd-BD18; Abd-B m mutations:Abd-BM5, Abd-BM3, Abd-BD14; Abd-B r(iab-9) mutation: Abd-Bx23-1; mutations iniab-5 to iab-7 regulatory sequences: iab-7Sz,iab-7MX2; Fab-8 mutations: Fab-864,Fab-8416. We have renamed the iab-948 andiab91065 (also known as iab-765)mutations (Celniker et al.,1990; Karch et al.,1985; Kuhn et al.,1992) as Abd-B48 andAbd-B65 mutations, respectively, as they affect the wholeAbd-B domain (although very mildly in the case ofAbd-B65). DfP9 is a deficiency for the whole bithorax complex and Pc3 a null mutation at thePolycomb (Pc) locus(Lewis, 1978). ThePc3 Abd-Blac1 double mutant was made by standard recombination.
Constructs and transformation of embryos
The pRCm1 construct (Busturia and Bienz,1993) was used in the transformation experiments that gave rise to the Abd-Blac1 mutation. This vector carries therosy gene as a transformation marker, a fragment from theAbd-B gene and the lacZ gene. The Abd-B fragment is a 5.6 kb Pst-HindIII fragment from 50,504 to 44,912(Martin et al., 1995). This fragment includes the Abd-B m promoter region, the initiation site of the Abd-B m (or A) transcript, the leader sequence and the sequence encoding for the first five amino acids of the Abd-B m protein fused to the bacterial lacZ gene. This Pst-HindIII fragment is referred to in the text as the Abd-B proximal promoter, Abd-Bpp(Busturia and Bienz, 1993). The P-element giving rise to the Abd-Blac1 mutation is the construct pRCm 70.2, and contains, in addition, a HindIII fragment of the Abd-B 3′ regulatory region (from 64,580 to 59,073) inserted into KpnI-XbaI sites of pRCm1 (seeFig. 1A).
In the analysis of the 5′ Abd-B regulatory region, the following constructs were used to transform embryos.
BEAbd-BppX2, which contains the EcoRI fragment from 47,992 to 40,968 (X2 fragment) cloned into the Abd-Bppwhite vector (A. Busturia,unpublished). The Abd-Bpp white vector contains the Abd-Bpp sequence(Busturia and Bienz, 1993)cloned into the pwZ50PL vector, kindly provided by T. Gutjahr.
BEAbd-BppL14, which contains the EcoRI-PstI fragment from 40,968 to 33,634 (L14 fragment) cloned into the same vector.
BEUbxppB4, which contains the HindIII-EcoRI fragment from 44,912 to 40,968 (B4 fragment) inserted upstream of the Ubx-lacZreporter gene in the CaSpeR Ubx-lacZ vector(Qian et al., 1991). This B4 fragment was also cloned into the pwHZ50PL vector, which includes the promoter of the hsp70 gene, to obtain construct BEpwHZ50PLB4.
pwHZ50PLPS, which contains the BamHI-HindIII fragment from 48,864 to 44,912 (PS fragment) inserted into the pwHZ50PL vector.
Transformation of embryos was done as described previously(Busturia and Bienz, 1993).
Isolation of P-element derivatives
Derivatives of the P-element insertion in Abd-B(Abd-Blac1) were obtained by several methods. In one of them, P [ry+Abd-Blac1]/TM3, Δ2-3 dysgenic males were crossed to ry506 females andry- males obtained from this cross were individually isolated and balanced. A different set of experiments, intended to isolate derivatives independently of their having (or not) the ry gene, was carried out as follows: individual dysgenic males of the genotype st red sbd2 Abd-Blac1/TM3, Δ 2-3 were crossed to Abd-BM1/TM6B females. The progeny was checked for either the presence of an Abd-B phenotype in theAbd-Blac1/Abd-BM1 progeny, or for the loss of the haploinsufficient Abd-B phenotype of theAbd-Blac1 insertion (small seventh tergite and a few bristles in the sixth sternite) in the Abd-Blac1/TM6Bmales. In the appropriate cases, individual males were isolated, and stocks were established and further characterized by crosses with differentAbd-B alleles. To ascertain if these derivatives werery+ or ry-, they were recombined withry506 and crossed to ry506 females. The class II (iab-8,9) mutants described in this work were isolated by this last method.
In situ hybridization
It was carried out as described by(Casares and Sánchez-Herrero,1995). The probe for the m transcript is a genomicBamHI fragment from 50,702 to 48,864. The probe to detect all theAbd-B transcripts, referred to as `common probe', is a genomicHindIII-SalI fragment from 56,556 to 52,989 that includes the homeobox.
It was performed as described previously(Casares and Sánchez-Herrero,1995). The rabbit anti-β-galactosidase antibody (Cappel) was used at a concentration of 1:2000. The Abd-B antibody, used at 1:20 concentration, is monoclonal antibody 1A2E9 (kindly provided by S. Celniker),and recognizes both Abd-B M and R proteins(Celniker et al., 1989). The anti-En monoclonal antibody (Patel et al.,1989) was used at 1:10 concentration, and was kindly provided by T. Kornberg. Staging of the embryos was determined as described previously(Campos-Ortega and Hartenstein,1985).
Embryonic cuticle analysis
It was carried out as previously described(Wieschaus and Nüsslein-Volhard,1986).
The molecular study of the Abd-Blac1 mutation and of the different derivatives was carried out by Southern blot analysis using the following probes: (1) Abd-B DNA not included in the pRCm 70.2 construct; (2) Abd-B DNA included in the construct; and (3) DNA from the rosy and lacZ genes. The lacZ probe was obtained from the Bluescript vector. The rosy probe is a 7.2 kbHindIII fragment from the HZ50PL vector(Hiromi and Gehring, 1987). DNA extractions and methods of Southern hybridization were carried out as described previously (Sullivan et al.,2000) with slight modifications.
Inverse PCR was done as described in FlyBase(http://www.fruitfly.org/methods).
The Abd-Blac1 insertion
Within a study of the Abd-B regulation, a genomic fragment including the iab-8 regulatory region and the Fab-8 boundary was subcloned into a vector carrying a 5.6 kb fragment from the Abd-Bgene (Abd-Bpp) and the lacZ gene(Busturia and Bienz, 1993) (see Materials and Methods and Fig. 1A). Transformant lines carrying this construct were isolated and embryos from each line stained with anti-β-galactosidase antibody. Four of the lines showed the basic staining pattern of the Abd-Bpp reported previously (Busturia and Bienz,1993), and shown in Fig. 1B-C.
The fifth line, however, showed a restricted β-galactosidase expression, confined to the posterior abdominal segments and resembling theAbd-B pattern of expression. Males heterozygous for this insert presented a typical Abd-B haploinsufficient phenotype (some bristles in the sixth sternite and a small seventh tergite). Therefore, we decided to characterize this line further. The insertion point of the P-element was located by inverse PCR at position 48,957(Martin et al., 1995), 242 bp upstream the Abd-B m transcription start site(Zavortink and Sakonju, 1989). Genetic tests with Abd-B m and Abd-B r mutations indicated that the insertion is an amorph for the Abd-B r function and a strong hypomorph for Abd-B m function. Staining of homozygous embryos with an anti-Abd-B antibody showed almost absence of expression in PS10-14. These observations indicate that the P-element has generated a mutation inAbd-B that we have called Abd-Blac1.
The lacZ expression of Abd-Blac1 embryos differs from that of Abd-B in wild-type embryos in several respects(Fig. 2). InAbd-Blac1 embryos, the lacZ expression starts simultaneously in PS12-14 at stage 9, and this signal is maintained at stage 10 (Fig. 2A) and thereafter(Fig. 2C). Expression in PS10-11 is low or absent and, at late stages, there is a strong PS12 signal in the ventral cord (Fig. 2E). By contrast, Abd-B products are observed only in PS13 and PS14 in stage 10 wild-type embryos (Fig. 2B). Later on, Abd-B expression decreases from PS14 to PS10 both in the epidermis and the ventral cord (Fig. 2D,F). The particular strong lacZ expression in the ventral cord of Abd-Blac1 embryos disappears inPc3 Abd-Blac1 double mutant embryos(Fig. 2G).
We hypothesize that the lacZ expression inAbd-Blac1 embryos may have some characteristics derived from the place of insertion and others due to the Abd-B regulatory region included within the P-element. To delimit each contribution, and to obtain mutations in the region of the Abd-B promoter, we mobilized the P element (see Materials and Methods).
Some Abd-Blac1 derivatives constitute a distinct class of mutations in Abd-B
We obtained different types of derivatives when mobilizingAbd-Blac1, but most of them could be classified into two different groups, defined by their phenotype, Abd-B expression and molecular structure.
Derivatives of the first group (class I; nine alleles isolated) arery- but retain the lacZ gene. They present in heterozygosis an Abd-B haploinsufficient phenotype and complementAbd-B r (iab-9) mutations. In hemizygosis they show a transformation of segments A5-A8 (PS10 to PS13) into A4 (PS9)(Fig. 3A), characteristic ofAbd-B m mutations (Casanova et al., 1986). Accordingly, in class I homozygous embryos Abd-B protein or RNA (`common' probe) expression is detected normally in PS14 but very weakly in PS10-13 (Fig. 3B). β-galactosidase staining of these embryos shows a strong signal in PS13 (Fig. 3B). The phenotype and staining pattern of these mutations resemble the derivatives of a P-lacZ element, iab-8Plac(+159), inserted very close to Abd-Blac1(McCall et al., 1994).
Derivatives of the second group (class II; six alleles isolated) arery+, have lost the lacZ gene, and show no haploinsufficient phenotype. Homozygous adults are not recovered because of a second-site lethal mutation present in the originalAbd-Blac1 chromosome. In hemizygous condition, these mutants show a distinct phenotype. In females, the dorsal A8 (eighth tergite)is covered with bristles and trichomes characteristic of an A6/A7 segment(Fig. 4B, compare withDfP9/+ females in Fig. 4A) and the genitalia is variably transformed into a sternite(Fig. 4F; compare with wild-type females in Fig. 4E). The males show an eighth tergite bearing trichomes(Fig. 4D, compare withDfP9/+ males in Fig. 4C) and a small eighth sternite with one to six bristles(Fig. 4H, compare with wild-type males in Fig. 4G). These phenotypes indicate there is a partial transformation of A8 to A6/A7. In both sexes, the A5-A7 segments of class II hemizygous adults only show the haploinsufficient phenotype that is due to the DfP9. These mutants lack genitalia and analia, and show occasionally a small ninth segment(Fig. 4I). Class II mutations do not affect the dominant phenotype of the Fab-864 andFab-8416 mutations, which delete the Fab-8boundary and transform PS12 into PS13(Barges et al., 2000). They also complement mutations like iab-7Sz andiab-7MX2, which affect A7 and A5-A7 segments, respectively(Galloni et al., 1983;Sánchez-Herrero et al.,1985), but fail to complement Abd-B r (iab-9)mutations. In hemizygous embryos, the A8 segment is transformed into a more anterior one (Fig. 4K, compare with the wild type in Fig. 4J). In the view of the embryonic and adult phenotypes, we have named these derivatives iab-8,9 mutations. This is a distinct class of BX-C mutations that transform the A8 and the A9 but not more anterior segments. TheTabR96 mutation(Celniker and Lewis, 1987)also affects both segments, although the effect in the A8 is very weak.
The iab-8,9 phenotype correlates with the Abd-B protein distribution in iab-8,9 homozygous embryos: Abd-B expression is normal in PS10 and 11, reduced in PS13 and almost absent in PS14(Fig. 5B,D, compare with wild-type expression in Fig. 5A,C). In situ hybridization with either a probe specific for them transcript or with the `common' probe also shows reduced signal in PS13. In some of these embryos, especially in the germband retracted stage,Abd-B expression in PS12 ectoderm is slightly increased compared with wild-type embryos (Fig. 5D,compare with Fig. 5C). These data suggest that iab-8,9 mutations affect the expression ofAbd-B r transcripts in A9/PS14 and of the Abd-B m RNA mainly in the A8/PS13.
We have characterized the molecular structure of 3 class I and 2 class II derivatives by Southern blot analysis (see Materials and Methods). The data indicate that both types of mutations are the result of recombination events between the `exogenous' promoter (the Abd-Bpp included in the P element) and the corresponding endogenous sequences. In class I derivatives, the ry,Fab-8 and iab-8 sequences of the P element are lost as a result of this recombination. In iab-8,9 alleles, the lacZ gene is lost (see Fig. 6). We have also defined, by inverse PCR, the location of the defective P-element of theiab-8,9T1J mutation. This location is precisely in the same position as in the Abd-Blac1 mutation.
Characterization of iab-8 regulatory sequences in the 5′ region of the Abd-B transcription unit
The genetic and molecular characterization of iab-8,9 mutations suggested that an iab-8 regulatory region could be located upstream the Abd-B m transcription unit (seeFig. 6). To test this prediction, we fused either a 7 kb genomic fragment from 47,992 to 40,968 (X2 fragment) or a 7.3 kb fragment from 40,968 to 33,634 (L14 fragment) to the basal construct containing the Abd-Bpp, obtaining the BEAbd-BppX2 and BEAbd-BppL14 constructs, respectively (Fig. 7A and Materials and Methods). We stained embryos of five independent lines transformed with each construct with an anti-β-galactosidase antibody. The BEAbd-BppX2 construct shows, in four out of five lines, an increase in staining in lateral epidermal cells of PS13(A8) compared with embryos transformed with the basal construct(Fig. 7C; compare withFig. 7B), suggesting the presence of a PS13 regulatory region in the construct. In late stages of embryogenesis the PS13 signal is not observed, but the embryos show a strong expression in posterior and anterior spiracles(Fig. 7F). The BEAbd-BppL14 construct does not give a particular pattern in any segment of the transformed embryos (not shown).
The Abd-Bpp shows, by itself, a strong basal staining that could obscure some particular signal coming from the tested fragments. To overcome this problem, and to better define the putative PS13 regulatory region, we fused the HindIII-EcoRI fragment from 44,912 to 40,968 (B4 fragment) upstream two different basal promoters: The Ubx minimal promoter (Qian et al., 1991),generating the construct BEUbxppB4, and the hsp70 minimal promoter(Hiromi and Gehring, 1987),obtaining the construct BEpwHZ50PLB4 (Fig. 7A). Neither the hsp70 minimal promoter nor theUbx minimal promoter on their own gives any segment-specific staining pattern in embryos (Hiromi and Gehring,1987; Qian et al.,1991). We stained six lines of embryos transformed with the BEUbxppB4 construct and seven of those transformed with the BEpwHZ50PLB4 one. None of the two constructs presents a stronger expression in PS13, although both show a clear signal in both anterior and posterior spiracles in late embryonic stages (Fig. 7D,G and data not shown). This suggests that the putative iab-8 sequences require the presence of an intact Abd-B promoter to direct a PS13 specific expression.
Regulation of expression in the spiracles
The construct containing only the Abd-Bpp shows a strong expression in anterior and posterior spiracles (Busturia and Bienz, 1993) (Fig. 1C, Fig. 7E). This suggests that the Abd-B promoter is specifically activated in both anterior and posterior spiracles, although gap genes repress Abd-Bexpression in anterior spiracles (Harding and Levine, 1988; Reinitz and Levine, 1990). The same expression in the posterior and anterior spiracles is observed in embryos transformed with a construct carrying aBamHI-HindIIII fragment (PS) included within the Abd-Bpp(see Fig. 7). We note, however,that the B4 fragment also directs expression in these organs(Fig. 7G). This suggests that there is more than a single Abd-B 5′ region that could direct expression in the spiracles.
Several iab domains in the Abd-B 3′ regulatory region control its expression in PS10-13. An iab-8 domain is located within this region, between 59,500 and 62,541(Barges et al., 2000;Zhou et al., 1999), and we refer to it as the 3′ iab-8 domain. However, mutants in which this domain is deleted (Fab-7R59 deletion) or separated from the Abd-B promoter (iab-7S10 translocation)show no effect in the A8 segment(Gyurkovics et al., 1990;Hendrickson and Sakonju, 1995;Hopmann et al., 1995;Tiong et al., 1987) (seeFig. 8), suggesting there are other iab-8 regulatory sequences in Abd-B. We have obtained genetic and molecular evidence indicating the presence of such sequences 5′ to the Abd-B m transcription unit.
This conclusion stems in part from the study of a P-element withAbd-B regulatory sequences inserted in the Abd-B gene(Abd-Blac1). This phenomenon, in which a P-element containing regulatory sequences of a gene is inserted in the same gene is called `homing' (Hama et al.,1990) and is particularly frequent in the BX-C(Bender and Hudson, 2000;Fitzgerald and Bender, 2001;McCall et al., 1994). The analysis of the Abd-Blac1 mutation and of some of its derivatives has helped to better understand Abd-B regulation.
Class I and Class II (iab-8,9) mutations
Some derivatives from the Abd-Blac1 insertion(iab-8,9 mutations) transform segments A8 and A9 but not more anterior segments. Curiously, the transformation of the A8 segment is not to A7 but, partially, to A6. This suggests that the iab-7 regulatory sequences are not fully active in the A8 segment, not even when iab-8activity is reduced. We explain below the effect on Abd-B expression that causes these transformations.
In iab-8,9 mutations, the defective P-element responsible for the phenotype retains the ry gene, the 3′ iab-8 region and the Fab-8 exogenous boundary, and is inserted in an intron of the B and C transcripts (see Fig. 1Aand Fig. 6). We propose that the ry gene included in the transposon interrupts this transcription,preventing Abd-B expression in A9/PS14. Such effect on transcription,which also occurs in Abd-Blac1, has been previously documented for other P elements inserted in introns of different genes(Casares et al., 1997;Horowitz and Berg, 1995;McCall et al., 1994). By contrast, in iab-8,9 embryos there is no general effect in the transcription of the Abd-B m transcript. The Abd-B mpromoter is normal, and, accordingly, Abd-B expression and function is wild type in PS10-12 (except for a slight increase in PS12 Abd-Bexpression). Therefore, the transformation of PS13 is due to the specific inactivation of iab-8 regulatory sequences. To explain this phenotype, we propose the existence of novel iab-8 regulatory sequences located upstream the Abd-B m transcription unit and named by us the 5′ iab-8 domain. These sequences would be inactive iniab-8,9 mutations because of the presence of the exogenousFab-8 boundary, placed between them and the Abd-B m promoter(see Figs 6,8).
The absence of Abd-B transforms female (A8) and male (A9)genitalia into leg tissue (Estrada and Sánchez-Herrero, 2001). In iab-8,9 mutations,however, the presence of low levels of Abd-B expression in these two metameres is sufficient to transform them, instead, to abdomen. This transformation prevents a normal development of the genital disc and indirectly produces the absence of analia, which derive from the A10 segment and do not require Abd-B function(Sánchez-Herrero et al.,1985; Tiong et al.,1985).
In class I derivatives, the exogenous Fab-8 and 3′iab-8 sequences, as well as the ry gene, are not present. The absence of the exogenous Fab-8 boundary allows for the activity of endogenous iab-8 sequences, and strong lacZ expression in PS13 results. Abd-B expression is reduced in PS10-PS13, as it is inAbd-Blac1 embryos, because the P-element interrupts theAbd-B promoter region. However, as the ry gene is absent,there is complete transcription of Abd-B r RNAs and normalAbd-B expression in PS14 (see Fig. 3B, Fig. 6).
The 5′ iab-8 regulatory domain of Abd-B
The existence of the 5′ iab-8 domain is supported by the analysis of transformants carrying fragments of this region. The PS13lacZ expression of embryos with constructs containing the X2 fragment is enhanced. Sequences from the distal part of this region (B4 fragment, not included in the Abd-Bpp), however, do not reproduce this particular expression when fused to heterologous minimal promoters (hsp70 or Ubxpromoters), suggesting that they require the Abd-B promoter for their proper activity. The need of enhancers to have their own promoter for a wild-type activity has been previously reported in Drosophila(Li and Noll, 1994;Merli et al., 1996;Ohtsuki et al., 1998;Schier and Gehring, 1992).
The preceding data, combined with previous results from the literature,help to delimit the putative 5′ iab-8 domain. The mutationtuh-3 maps around 34,000 and affects the A9 but not the A8 segment(Casanova et al., 1986;Mack et al., 1997) (seeFig. 8), thus setting the distal limit of the putative iab-8 region. In the more proximal region, the Abd-B48 mutation, located at about 45,700,causes a strong reduction of Abd-B expression in PS10-PS13(Celniker et al., 1990;Karch et al., 1985;Kuhn et al., 1992) (seeFig. 8), indicating that this breakpoint interrupts the Abd-B promoter region required for correct PS10-13 expression (Sipos et al.,1998). By contrast, the Abd-B65 mutation,which breaks at about 43,700, only mildly affects PS10-PS12(Boulet et al., 1991;Celniker et al., 1990) (B. E.,F. C., A. B. and E. S.-H., unpublished), but affects a bit more strongly PS13(data not shown), suggesting the mutation breaks near the region that separates the Abd-B m upstream regulatory region common for PS10-PS13 from the 5′ iab-8 domain (seeFig. 8). All these results confine the 5′ iab-8 domain to the region indicated inFig. 8A.
Models to explain the PS12 lacZ expression inAbd-Blac1 embryos
The lacZ gene in Abd-Blac1 shows an early and strong signal in PS12 when compared with the Abd-B wild-type expression. The 3′ iab-8 sequences included in the P-element are unlikely to be responsible for this particular staining for three reasons:first, they are separated from the lacZ gene by the Fab-8boundary, which should prevent or attenuate the lacZ transcription directed by these sequences (Barges et al.,2000; Zhou et al.,1999; Zhou and Levine,1999) (see Figs 1,8). Second, the 3′iab-8 sequences direct expression in PS13 and not in PS12(Barges et al., 2000;Zhou et al., 1999). Finally,embryos carrying this transposon outside the Abd-B gene or in theUltrabithorax region of the BX-C do not show this PS12 expression(Fig. 1B,C and data not shown). Rather, we think that the Fab-8 and PRE exogenous sequences included in the P-element are the cause of this particular staining. We have observed that the strong PS12 β-galactosidase expression in the ventral cord ofAbd-Blac1 embryos disappears in a Pc mutant background. This suggests that PREs are instrumental in maintaining this PS12 signal.
A PRE is found distal to the Fab-8 boundary(Barges et al., 2000;Zhou et al., 1999), and therefore included in the Abd-B 3′ regulatory region of the P-element (see Fig. 8A). As PREs frequently pair (Fauvarque and Dura,1993; Kassis,1994; Muller et al.,1999; Sigrist and Pirrotta,1997), it is possible that the exogenous and endogenous PREs may pair in Abd-Blac1, creating a loop in the DNA. TheFab-8 boundaries may also contribute to this pairing(Hogga et al., 2001), as is the case with other types of boundary regions [such as the su(Hw) insulators (Cai and Shen,2001; Muravyova et al.,2001)]. Based on this pairing mechanism, we propose two models to explain the early and strong signal in PS12 of Abd-Blac1embryos.
In the first model (Fig. 8B), the pairing takes place between the endogenous and exogenousFab-8/PRE sequences (iab-8 sequences could also contribute to it). This would both approximate the iab-7 sequences (which direct expression in PS12) to the lacZ gene and stabilize the looping of these sequences, resulting in a strong expression in PS12. Supporting this model, P-lacZ elements inserted in the iab-7 domain show a similar strong β-galactosidase expression in PS12 in the ventral cord(Barges et al., 2000;Galloni et al., 1993). This suggests that the iab-7 sequences are driving lacZexpression in PS12 of Abd-Blac1 embryos. This model could also explain the early lacZ expression in PS12 for the following reason. In the wild type, the m transcript is detected consecutively and with decreasing amount of expression in PS13, PS12, PS11 and PS10(Boulet et al., 1991;Kuziora and McGinnis, 1988;Sánchez-Herrero and Crosby,1988). The outcome is that stage of development in which transcription is detected and the amount of expression are related(Crosby et al., 1993). The increased lacZ expression in PS12 of Abd-Blac1embryos may result in an `earlier' β-galactosidase expression in PS12. A similar case has been described for a P-lacZ insertion in theiab-7 domain: in this bluetail mutation, lacZexpression starts in PS12 at stage 8, before Abd-B expression is detected in PS12 of wild-type embryos(Galloni et al., 1993). The model requires the iab-7 sequences to bypass the Fab-8boundary, although this also takes place in the wild type.
The second model (Fig. 8C)proposes that Fab-7/PRE sequences pair with the exogenousFab-8/PRE DNA. Pairing between different Fab sequences have been previously proposed to occur in normal development(Hogga et al., 2001). We assume there is a competition between endogenous and exogenousFab-8/PRE sequences to pair with Fab-7/PRE and that the pairing with the exogenous ones is somehow preferred. As a result, the 3′ iab-8 sequences would be within the same domain as theiab-7 DNA and both would activate the lacZ gene in PS12. This model would explain the similar PS12 and PS13 expression inAbd-Blac1 embryos. Supporting this model, thelacZ signal in germ band extended Abd-Blac1embryos is similar to that of Fab-864 embryos. This mutation is a derivative of a P-lacZ element inserted in theiab-7 domain, which retains the lacZ gene; in this mutant,the Fab-8 boundary is deleted and, as a result, both theiab-7 and iab-8 regulatory sequences activate thelacZ gene (Barges et al.,2000). However, the model does not explain the absence of strong PS13 lacZ expression in the nervous system ofAbd-Blac1 embryos. It is possible that the pairing betweenFab/PRE sequences changes throughout development. If so, thelacZ pattern of Abd-Blac1 embryos could be explained by a combination of both models.
The two models predict an increase of Abd-B protein expression in PS12 ofAbd-Blac1 and iab-8,9 embryos. In the first model, iab-7 interaction with the Abd-B m promoter is reinforced. In the second one, 3′ iab-8 sequences could now be active in PS12. In Abd-Blac1 embryos, this effect could be difficult to detect, as Abd-B transcription is strongly reduced. However, it should be visible in iab-8,9 embryos, because the exogenous and endogenous Fab-8/PRE sequences could also pair and theAbd-B promoter is intact. We have observed a slight increase inAbd-B expression in PS12 of iab-8,9 homozygous embryos, but this effect is insufficient to cause a clear phenotype in eitheriab-8,9 homozygous embryos or iab-8,9 heterozygous females.
Abd-B regulation in PS13
Our data, and previous results (Barges et al., 2000; Zhou et al.,1999), support the existence of both 3′ and 5′iab-8 domains. The 3′ domain seems dispensable for a wild-type A8/PS13 development if the 5′ domain is intact, as in theFab-7R59 and iab-7S10 mutations, but it is not clear if the opposite is true. In iab-8,9 mutations, the 5′ iab-8 domain is inactive (determined by the presence of the exogenous Fab-8 boundary), but, according to the first model, the pairing of exogenous and endogenous 3′ iab-8 sequences may also reduce the activity of the 3′ iab-8 domain. Thus, in these mutations the activity of both 5′ and 3′ iab-8 sequences could be compromised. This interpretation suggests that either the 3′ or the 5′ iab-8 domains could determine Abd-B protein distribution in A8/PS13. Alternatively, both domains should be necessary for A8/PS13 development, but the 5′ domain alone could direct A8/PS13 development if the 3′ domain is absent. In any case, these results underline the complexity of Abd-B regulation.
We thank Ginés Morata for discussions and for comments on this manuscript; and Marianne Bienz and Shigeru Sakonju, in whose laboratories this work was initially carried out, for their help and encouragement. We specially thank an anonymous referee for insightful comments and for proposing model 2. We also thank Francois Karch for discussions and for solving many questions;Francisco A. Martín for help in the PCR experiments and in determining the Abd-Blac1 insertion point; Susan Celniker, Madeline Crosby, Francois Karch, Tom Kornberg, Shigeru Sakonju and Jumin Zhou for stocks, antibodies and vectors; and Rosa González for technical assistance. B. E. is recipient of a fellowship from the Comunidad Autónoma de Madrid. This work has been supported by grants from the Dirección General de Investigación Científica y Técnica (No PB98-0510 to E. S.-H. and BMC 2001-2178 to A. B.), the Comunidad Autónoma de Madrid (number 08.9/0003/98 to E. S.-H. and 08.9/0002/98 to A. B.) and an Institutional Grant from the Fundación Ramón Areces to the Centro de Biología Molecular Severo Ochoa.