Transvection in the Drosophila Abd-B Domain: Extensive Upstream Sequences Are Involved in Anchoring Distant cis-Regulatory Regions to the Promoter

RESULTS

In adult wild-type males the program specifying the development of the A7 abdominal segment does not produce a visible tergite or sternite (Figure 1a). This developmental program depends on the normal functioning of the iab-7 cis-regulatory domain that is responsible for generating the appropriate level of Abd-B RNA/protein expression in PS12/A7. When iab-7 is deleted on both homologues, Abd-B expression in PS12/A7 comes under the control of the iab-6 cis-regulatory domain, resulting in the transformation of PS12/A7 into a copy of PS11/A6 (Galloniet al. 1993). Because of the haplo-insufficiency of the Abd-B homeotic gene, disruptions in the normal developmental program of PS12/A7 are evident in animals heterozygous for an iab-7 mutation, such as iab-7^Sz, which deletes the iab-7 region (Galloniet al. 1993). In this case, A7 assumes an identity in between that of the normal A6 and A7. The mixed identity of A7 is manifested as a rudimentary tergite-like structure posterior to A6 with a characteristic shape and size (Karchet al. 1985). This phenotypic transformation is shown in Figure 1b and summarized diagrammatically in Figure 2. A similar effect on A7 development is also observed in animals heterozygous for an Abd-B mutation; however, in this case the disruptions in development are not restricted to A7, and the morphology of the adjacent segments A6 and A5 are altered as well.

A phenotype exactly the opposite of that produced by iab-7 mutations is observed for the gain-of-function mutation Fab-7; PS11/A6 is transformed into PS12/A7. This is due to the ectopic activation of iab-7 in PS11/A6. As a consequence, the number of visible segments in Fab-7 homozygous males is reduced from six to five (Gyurkovicset al. 1990). Although disruptions in A6 development are also found in Fab-7/+ males, the extent of the transformation is reduced compared to that observed in homozygous mutant animals. Instead of the PS12/A7 identity seen in homozygotes, PS11/A6 assumes a mixed PS11/A6-PS12/A7 identity in animals heterozygous for the Fab-7 mutation. In fact, the mixed identity phenotype of PS11/A6 observed in Fab-7/+ males closely resembles the mixed identity phenotype of PS12/A7 observed in iab-7⁻/+ (or Abd-B⁻/+) male flies (see Figure 1, c and b).

The incomplete transformation of PS11/A6 in Fab-7/+ males could be attributed to a haplo-insufficiency for iab-7. In Fab-7/+ males, the iab-7 cis-regulatory domain on the Fab-7-containing homologue is active in PS11/A6, whereas the iab-7 cis-regulatory domain on the wild-type homologue is not. If regulation only occurs in cis, Abd-B expression in PS11/A6 should be driven by interactions between the iab-7 cis-regulatory domain and the Abd-B gene only on the Fab-7 homologue. Such interactions should not occur on the wild-type homologue in PS11/A6 because iab-7 is inactive. This interpretation is supported by the phenotype of Fab-7/iab-7^Sz mutant animals. In this genotype, only a single copy of the iab-7 cis-regulatory domain should be available to drive Abd-B expression, not only in A6 and but also in A7. In males carrying this mutant combination, both PS11/A6 and PS12/A7 would be expected to assume an identity in between that of PS11/A6 and PS12/A7. This is the case. Moreover, as illustrated in Figure 1d and Figure 2, segments A6 and A7 in the Fab-7/iab-7^Sz males closely resemble the A7 segment found in iab-7⁻/+ (Figure 1b) males.

Taken together, these results would seem to suggest that, if the rest of the Abd-B domain remains unchanged, the phenotype/identity of PS12/A7 (or PS11/A6 in the Fab-7 mutant) directly reflects the number of active iab-7 cis-regulatory domains.

Fab-7 is suppressed by Abd-B point mutations in trans: We next examined the effects of combining Fab-7 with an Abd-B point mutation, Abd-B^D16. If our hypothesis that the identity of PS11/A6 and PS12/A7 depends upon the number of iab-7 domains that are active in these two parasegments is correct, then Fab-7/Abd-B^D16 males would be expected to have the same intermediate phenotype in A6 as was observed in Fab-7/iab-7^Sz males. However, as illustrated in Figure 1e and Figure 2, this is not the case. Although A7 has the same phenotype in the two mutant combinations, the phenotype of segment A6 in Fab-7/Abd-B^D16 males differs from that in Fab-7/iab-7^Sz males. The A6 tergite in the Fab-7/Abd-B ^D16 genotype is enlarged relative to the A7 tergite, and more closely resembles the wild-type A6 segment. This unexpected finding suggests that, unlike iab-7^Sz, the Abd-B^D16 mutation partially suppresses the gain-of-function phenotype of Fab-7 in A6. This partial suppression is not due to some unusual properties of the Abd-B^D16; precisely the same A6 phenotype was observed when other Abd-B point mutations (Abd-B^S1, Abd-B^S4; Tionget al. 1985) were combined with Fab-7 (data not shown).

These observations could mean that the activity of Abd-B is “more haplo-insufficient” in A6 than in A7. If this were true, we would expect to observe a similar suppression of the Fab-7 phenotype in A6 by the deletion Df(3R)P9, which removes the entire BX-C, including the Abd-B gene. However, as was observed for the Fab-7/iab-7^Sz combination, segments A6 and A7 in Fab-7/Df(3R)P9 males show the same intermediate A6-A7 phenotype (see Figure 1f). Similarly, the combination of Fab-7 with a deletion, Df(3R)R59, which removes both iab-6 and iab-7 but not the Abd-B gene, has the same phenotype as Fab-7/iab-7^Sz (data not shown), indicating that the functioning of iab-6 in trans is irrelevant for the phenotype of Fab-7 in A6.

Chromosomal rearrangements eliminate the effect of the Abd-B point mutations: The suppression of the Fab-7 gain-of-function phenotype by Abd-B point mutations in trans implies that our original assumption, namely, that regulatory interactions only occur in cis, is incorrect. Instead, it suggests that some type of trans-regulatory interaction or “transvection” must also occur. In particular, a trans-regulatory interaction could potentially explain why the Fab-7 gain-of-function phenotype of Fab-7/Abd-B⁺ animals is stronger than that of Fab-7/Abd-B⁻ animals; in the former case, the Abd-B genes in cis and in trans would produce functional products, whereas in the latter case, functional products would only be produced by the Abd-B in cis.

Transvection is classically tested by generating chromosomal rearrangements that disturb homologous pairing. For this purpose, we first screened for X-ray-induced mutations on the Fab-7 chromosome, which reduced or eliminated the suppression of the Fab-7 gain-of-function phenotype by Abd-B^D16, that is, for Fab-7*/Abd-B^D16 males in which the A6 tergite more closely resembled the A7 tergite. [We used Abd-B^D16 as a representative allele in this and subsequent experiments because it is a molecularly characterized mutation that does not code for a detectable protein (Bouletet al. 1991).] Out of ~3000 mutagenized chromosomes, we recovered 12 independent Fab-7*/Abd-B ^D16 males in which the A6 tergite was as thin or nearly as thin as the A7 tergite. For further analysis, we established stocks of these 12 Fab-7* chromosomes.

Chromosomal rearrangements that interfere with the postulated transvection between Fab-7 and Abd-B might also be expected to disrupt pairing-dependent interactions elsewhere in BX-C. A well-known example of a pairing-dependent regulatory interaction in BX-C is the wing transformation induced by the Cbx¹Ubx¹ chromosome when it is paired with a wild-type copy of BX-C (Lewis 1955, 1985). The 12 Fab-7* chromosomes were tested for their ability to interact with the Cbx¹Ubx¹ chromosome, and we found that all but one significantly suppressed the wing phenotype induced by the Cbx¹ mutation. Moreover, the strength of the phenotypic effects of these 11 Fab-7* chromosomes closely paralleled that found in combination with Abd-B^D16: mutations that exerted a strong effect on the Fab-7/Abd-B^D16 phenotype also strongly suppressed the wing phenotype (not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Abdominal cuticles of adult males. (a) +/+. In wild type, A7 is represented only by a pair of tracheal openings (arrowhead). (b) iab-7^Sz/+. The haplo-insuffiency of the iab-7 cis-regulatory region results in the addition of a rudimentary, seventh tergite. (c) Fab-7/+. Ectopic activation of iab-7 in A6 reduces the sixth tergite to the size of the seventh in iab-7^Sz/+. (d) Fab-7/iab-7^Sz. Because only a single iab-7 is active in A6 and A7 (that of the Fab-7 chromosome), both sixth and seventh tergites are reduced to the size of the seventh in iab-7^Sz/+. (e) Fab-7/Abd-B^D16. A7 is similar to that of the previous genotype, but the A6 tergite is enlarged. (f) Fab-7/Df(3R)P9. If, in addition to iab-7, the trans copy of Abd-B is also deleted, both A6 and A7 tergites are again similar to those of the genotype Fab-7/iab-7^Sz. (Additionally, due to the partial transformation of A5 into A4, lack of dark pigmentation in patches of A5 is also evident.) (g) Fab-7/Mcp^B116iab-7^SzAbd-B^D16. The size of the sixth tergite is similar to that of the sixth tergite in Fab-7/Abd-B^D16 males e, indicating that the presence or the absence of an inactive copy of iab-7 in trans is indifferent for the suppression of the Fab-7 phenotype. However, A7 is enlarged to the size of A6 in Fab-7/Mcp^B116iab-7^SzAbd-B^D16 trans-heterozygotes. As shown in h, if Fab-7 is replaced with a wild-type chromosome in this genotype (Mcp^B116iab-7^SzAbd-B^D16/+), A6 becomes similar to wild type, but A7 remains about the same as in Fab-7/Mcp^B116iab-7^SzAbd-B^D16, indicating that the Fab-7 deletion only ectopically activates the wild-type iab-7 in PS11, but does not alter its ability to transregulate Abd-B.

As expected from the suppression of the Cbx¹ phenotype, cytological examination of these 11 Fab-7* chromosomes revealed that they all had rearrangements affecting the right arm of the third chromosome. The position of the breakpoints for the 11 Fab-7* chromosomes that suppress the Cbx¹ phenotype (TSR-1^Sz,Fab-7 to TSR-11^Sz,Fab-7) are listed in Table 1. All 11 of them have one breakpoint between BX-C and the centromere, and a second breakpoint outside of this region. This arrangement of breakpoints follows the rules established by Lewis (1955) for chromosomal aberrations that disrupt transvection in BX-C. The one exception, TSR-12^Sz,Fab-7, proved to be an iab-4 mutation, and it has an inversion breakpoint in BX-C (see Table 1). Although this rearrangement should disrupt pairing in the Abd-B region of BX-C, it would not be expected to affect pairing in the Ubx region of BX-C.

These results argue that the suppression of Fab-7 gain-of-function phenotype by Abd-B point mutations is likely to be due to a nonproductive trans-regulation. When a wild-type gene is present on both homologues (Fab-7/+), both cis- and trans-regulatory interactions can contribute to the level of ABD-B protein that is ultimately expressed in PS11/A6. Although these same cis- and trans-interactions would also occur in Fab-7/Abd-B^D16 flies, functional Abd-B protein would only be produced by the Abd-B gene on the Fab-7 chromosome. It should be noted that this model requires that the regulatory capacity of the ectopically activated iab-7 domain in PS11/A6 is limiting; otherwise, the presence of a competing Abd-B point mutant in trans to the Fab-7 chromosome would have no phenotypic consequences. This supposition—that the regulatory capacity of iab-7 is limiting—seems to be correct, because the phenotype of PS11/A6, when the Abd-B gene in trans is either removed entirely (as in Fab-7/Df(3R)P9) or no longer available for pairing (as in TSR,Fab-7/Abd-B^D16), is indistinguishable from Fab-7/+.

Competition between the iab-7 cis-regulatory domains: In the experiments described in the previous section, nonproductive trans-interaction between the iab-7 regulatory domain on the Fab-7 homologue and the Abd-B point mutation on the other homologue suppressed the gain-of-function phenotype in A6. We speculated that the trans-interactions between iab-7 and the mutant Abd-B gene were permitted in A6 because the iab-7 regulatory domain on the homologue containing the mutant Abd-B gene is not active in this segment (see Galloniet al. 1993). In contrast, such trans-interactions might not be as strong in A7 because cis-interactions between the Abd-B gene and the iab-7 regulatory domain on the same homologue would reduce the ability of the Abd-B promoter to interact in trans. This model could potentially explain the phenotypic difference between A6 and A7 in Fab-7/Abd-B^D16 males (see Figure 2).

If this model is correct, cis-competition in segment A7 should be eliminated by deleting the iab-7 regulatory domain on the Abd-B^D16-containing chromosome. The lack of competition from the iab-7 domain in cis should result in an increase in the frequency or strength of trans-interactions between the iab-7 domain on the Fab-7 chromosome and the mutant Abd-B gene on the other homologue.

To test this prediction, we recombined an iab-7 deletion mutant, iab-7^Sz, onto the Abd-B^D16 chromosome. The phenotype of +/iab-7^SzAbd-B^D16 and Fab-7/iab-7^SzAbd-B^D16 males is shown in Figure 1h and Figure 1g, respectively. Two findings are of interest.

First, the gain-of-function phenotype of Fab-7 in A6 is still suppressed in Fab-7/iab-7^SzAbd-B^D16 males (Figure 1g), and this segment closely resembles the A6 segment found in Fab-7/Abd-B ^D16 males (Figure 1e). The similarity between these two mutant combinations would argue that the iab-7 cis-regulatory domain on the Abd-B^D16 chromosome does not contribute to the suppression of the Fab-7 phenotype in A6 and would be consistent with the idea that the iab-7 cis-regulatory domain is normally inactive in this segment (Galloniet al. 1993). Second, segment A7 in both +/iab-7 ^SzAbd-B^D16 (Figure 1h) and Fab-7/iab-7^SzAbd-B^D16 (Figure 1g) males is even more strongly transformed toward an A6 identity than it is in +/Abd-B^D16 or in Fab-7/Abd-B^D16 males. In fact, A6 and A7 are indistinguishable in Fab-7/iab-7^SzAbd-B^D16 males and closely resemble the appearance of A6 in Fab-7/Abd-B^D16 and A7 in +/iab-7^SzAbd-B^D16. It should be emphasized that this transformation in A7 identity cannot be attributed solely to the lack of iab-7 and Abd-B function on the iab-7^SzAbd-B^D16 chromosome. Thus, the transformation of A7 toward A6 in Fab-7/iab-7^SzAbd-B^D16 males is much more extreme than it is when Fab-7 is combined with deficiencies e.g., Df(3R)P9, Figure 1f, or Df(3R)C4, not shown, which completely remove both iab-7 and Abd-B (compare Figure 1g and f). As expected, if the Fab-7 chromosome is replaced by its rearranged version in Fab-7/iab-7^SzAbd-B^D16, i.e., in TSR,Fab-7/iab-7^SzAbd-B^D16 (not shown), the phenotype of A6 and A7 becomes indistinguishable from that of TSR,Fab-7/Abd-B^D16 or Fab-7/Df(3R)P9 (shown in Figure 2 and Figure 1f, respectively).

These findings indicate that trans-regulatory interactions in segment A7 have no phenotypic consequences if an iab-7 domain is present on the chromosome containing the mutant Abd-B gene. To demonstrate that this domain must not only be present but also active, we recombined Fab-7 with the Abd-B point mutation Abd-B^D16 and then examined the phenotype of Fab-7/Fab-7Abd-B^D16 males. In these animals, both copies of iab-7 will be ectopically activated in A6. If we are correct in assuming that the regulatory domain in cis must be active in order to reduce trans-regulatory interactions with the mutant Abd-B gene, then the phenotype of A6 and A7 in these animals should be identical and resemble the A7 segment in Fab-7/Abd-B ^D16 animals. Indeed, this is the case (see Figure 2).

Two different (but not mutually exclusive) hypotheses could explain the phenotypic difference between the segments in which one vs. two iab-7 regions are active, e.g., the A7 of Fab/iab-7^Sz Abd-B^D16 and Fab-7/Abd-B^D16. The first postulates that the iab-7 regulatory domains engage not only in cis-but also in trans-regulatory interactions (indicated by dotted arrows in Figure 2). In this case, the resulting phenotype depends upon the sum of the interactions contributed by either one or two active iab-7 domains. In the second, cis-interactions between the iab-7 regulatory domain and the Abd-B gene on the same homologue would, because of competition, tend to reduce or suppress trans-interactions between this Abd-B gene and the iab-7 regulatory domain on the other homologue. When there is no competing active regulatory domain in cis, the Abd-B gene will be more readily available to engage in trans-interactions. It is not possible to distinguish between these two models with the genetic tools presently available; however, we favor the second model, as our analysis of the effects of promoter deletions (see below) suggests that the balance between cis- and trans-interactions can be shifted by competition.

Mutations that delete sequences near the 5′ end of the Abd-B gene enhance trans-regulation: The ability of the iab-7 regulatory domain to interact with the Abd-B genes in cis and in trans is not equal. If the cis- and trans-interactions were symmetrical, the phenotype of Fab-7+/+Abd-B and Fab-7Abd-B/++ would be identical. However, this is clearly not the case. In contrast to the modest but detectable transformation of A6 toward A7 in Fab-7/Abd-B^D16 flies, we see no trace of an A6 to A7 transformation in Fab-7Abd-B^D16/+ flies. Similarly, Abd-B^D16 does not complement iab-7^Sz to any detectable degree in A7 (Figure 6a).

We reasoned that it might be possible to strengthen the interaction of iab-7 with the trans copy of Abd-B by weakening its interactions with the Abd-B gene in cis. Because the regulatory interactions of iab-7 with the Abd-B gene are likely to be mediated, at least in part, by sequences in the vicinity of the Abd-B promoter, we thought that it might be possible to weaken the cis-interactions by substituting Abd-B point mutants with an Abd-B 5′ deletion mutant, Abd-B^D14 (m⁻; Karchet al. 1985), which lacks sequences around the proximal promoter (Bouletet al. 1991). As anticipated, we found that Abd-B^D14 weakly complements the iab-7 loss-of-function mutant iab-7^Sz in A7 (Figure 6b). Moreover, unlike in Fab-7Abd-B^D16/+, we were able to detect a weak transformation of A6 into A7 in Fab-7Abd-B^D14/+ adults. This phenotype of weak trans-interaction is best seen in the sixth sternites of females (see Figure 3a).

To rule out the possibility that this weak transformation is due to a “leakiness” of the Abd-B^D14 mutation, we examined the phenotype of Fab-7Abd-B^D14/Df(3R)P9 hemizygotes and Fab-7Abd-B^D14/Fab-7Abd-B^D14 homozygotes. In neither case did we detect transformation of A6 into A7 (not shown). In fact, it appears that the weak transformation observed in Fab-7Abd-B^D14/+ animals requires that the wild-type copy of the Abd-B gene in trans can pair. Thus, transformation is not detectable when pairing is efficiently disrupted (Figure 3b). Conversely, trans-interaction can be detected in both Fab-7Abd-B^D14/iab-7^Sz and Fab-7Abd-B^D14/Df(3R)R59 flies (not shown), where the trans copy of the iab-7 regulatory domain is deleted but the Abd-B gene is intact.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Diagrammatic recapitulation of the most relevant results (presented in part in Figure 1) and their interpretation. Continuous line represents the active form of iab-6 and iab-7; shaded rectangle, the inactive (closed) form of iab-7; small black dots, the Fab-7 and the putative Fab-8 boundaries; vertical lines, the endpoints of Fab-7 and iab-7^Sz deletions; filled circle, the proximal Abd-B promoter; bold arrow, the Abd-B class A transcription unit; and X, the presence of a point mutation. The regulatory interaction between Abd-B and iab-7 is represented by curved arrows. Dotted arrows and question marks symbolize the possibility of mutual trans-regulation when iab-7 cis-regulatory regions are intact and active on both homologues. The width of black oblongs in column 2 roughly corresponds to the size of the tergites of the different genotypes.

Other lines of evidence also suggest that the Abd-B^D14 promoter deletion mutant is a weaker competitor for interactions with iab-7, either in cis or in trans, than a structurally wild-type Abd-B gene. First, the trans-suppression of the Fab-7 gain-of-function phenotype in segment A6 of Abd-B^D14/Fab-7 males is significantly weaker (but still detectable) than in the combination of Fab-7 with any of the Abd-B point mutants (compare Figure 4, b and a). Second, the haplo-insufficient phenotype of Abd-B^D14/+ in A7 is clearly weaker than that of Abd-B point mutations (compare Figure 4, b and a), which suggests a hyperactivation of the wild-type Abd-B gene in trans. These differences are pairing dependent, because the phenotypes of Abd-B^D14/TSR,Fab-7 (Figure 4c) and Abd-B^D16/TSR,Fab-7 flies are indistinguishable (not shown).

Isolation of new Abd-B mutations which enhance trans-regulation by iab-7: The observation that the small Abd-B^D14 deletion enhances iab-7 trans-regulation points to sequences in the vicinity of the proximal Abd-B promoter as potential targets for cis- (and trans-) regulatory interactions. However, the finding that Abd-B^D14 still suppresses Fab-7 in trans to some extent (compare Figure 4, b and c) suggests that additional sequences may also be relevant for trans-regulation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Adult cuticles of (a) Fab-7Abd-B^D14/+, (b) TSR-10^Sz,Fab-7Abd-B^D14/In(3LR)TM3, Sb Ser females. When pairing of the homologues is normal a, A6 is weakly transformed toward A7 as evidenced by the presence of characteristics that are normally found on the wild-type A7, such as hairs pointing toward the middle of the sixth sternite and the absence of trichomes around the sternite (indicated by arrowheads). When pairing is efficiently disrupted (by the rearrangements TSR-10^Sz and In(3RLR)TM3), this mild Fab-7 phenotype is no longer detectable b. In addition, the haplo-insufficient phenotype in A7 is clearly stronger (the size of the seventh sternite is enlarged) than in a, indicating that the hyperactivation of the Abd-B gene is also eliminated in the absence of somatic pairing.

To further characterize sequences from the Abd-B gene that contribute to these regulatory interactions we took advantage of a hypomorphic Abd-B mutation, UC21-10,1-d, which has a P-transposon at position +159 on the molecular map (McCallet al. 1994) just upstream of the transcription start site of the Abd-B class A transcript. By itself, this insertional mutation, like the promoter deletion in Abd-B^D14, weakly enhances iab-7 trans-regulation. Thus, Fab-7UC21-10,1-d in trans to a wild-type Abd-B gene transforms A6 toward A7. That this transformation of A6 to A7 is not due simply to the residual Abd-B activity of the UC21-10,1-d allele, but rather arises from trans-activation of the wild-type Abd-B gene, is suggested by the finding that the A6-A7 transformation is strongly reduced when Fab-7UC21-10,1-d is combined with the point mutant Abd-B^D16 (lethality covered by Dp(3;1)bxd¹¹¹).

To identify sequences that contribute to cis- (and trans-) regulatory interactions, we mobilized the UC21-10,1-d transposon on the Fab-7UC21-10,1-d chromosome and then selected for excision events that eliminated the remaining Abd-B m activity (see materials and methods). We then tested these new Abd-B alleles for their ability to enhance trans-regulation of a wild-type Abd-B. Out of nine new Abd-B alleles, we identified two mutations (Abd-B^PSz1 and Abd-B^PSz2) that enhanced the Fab-7 gain-of-function phenotype in A6 when trans to a wild-type copy of BX-C more strongly than Abd-B^D14. Genomic Southern blotting revealed that Abd-B^PSz1 has a 10.8-kb deletion between map position +155.6–156.8 and +165.6–166.6, and Abd-B^PSz2 carries a smaller deletion of about 5.5-kb from+155.6–156.8 to+160.2–161.3 (Figure 5).

Other 5′ deficiencies also enhance trans-regulation: When we compared the transformation of A6 into A7 in Fab-7 Abd-B⁻/+ flies, it appeared that the strength of the trans-activation of the wild-type Abd-B gene is greatest in the largest promoter deletion, that is, Fab-7Abd-B^PSz1/+ > Fab-7Abd-B^PSz2/+ > Fab-7Abd-B^D14/+ > Fab-7Abd-B^D16/+. Interestingly, precisely the same relationship is observed for the complementation of the iab-7 deletion, iab-7^Sz, in A7. Thus, the strongest complementation is observed in Fab-7Abd-B^PSz1/iab-7^Sz flies, which is followed by Fab-7Abd-B^PSz2/iab-7^Sz, Fab-7Abd-B^D14/iab-7^Sz, and finally, Fab-7Abd-B^D16/iab-7^Sz. The similarity between these two assays suggested that we could measure the ability of any Abd-B mutation to promote trans-regulation by examining its complementation of iab-7^Sz in A7. Using this assay, we tested the trans-regulating ability of three previously isolated deficiencies, Df(3R)C4 (Karchet al. 1985), Df(3R)U110 and Abd-B^RD18 (Hopmannet al. 1995) that remove the distal part of BX-C. As expected, Df(3R)C4, which removes iab-7 and part of iab-6 in addition to Abd-B, does not show any complementation. However, both Df(3R)U110 and Abd-B^RD18, which leave iab-7 intact, complement iab-7^Sz to a high degree (Figure 6, d and c, respectively). As expected, these two deletions do not suppress the Fab-7 phenotype in trans, and the phenotype of A6 in Fab-7/Df(3R)U110 and Fab-7/Abd-B ^RD18 (not shown) is similar to Fab-7/+ (see Figure 1c).

The approximate limits of the complementing deletions are shown in Figure 5. These deletions can be ordered in increasing strength of iab-7^Sz complementation as follows: Abd-B^D14 < Abd-B^PSz2 < Abd-B^RD18 < Abd-B^PSz1. From this data it would appear that trans-regulation is strongest in the case of deletions that remove the largest region of Abd-B DNA on the 5′ side of the Abd-B gene (compare Figure 6, b and c). It is also clear that deficiencies that strongly enhance trans-regulation extend well beyond the region thought to correspond to the proximal Abd-B promoter (Busturia and Bienz 1993).

We note that unlike Abd-B “point mutations,” the Abd-B^PSz2, Abd-B^RD18 or Abd-B^PSz1 deletions do not show any sign of an A7 to A6 transformation when they are trans to a wild-type BX-C. This difference presumably reflects the ability of these large promoter deletions to enhance trans-regulatory interactions with the wild-type Abd-B gene. In fact, in the case of these larger deletions, the hyperactivation of the wild-type Abd-B gene in trans is even stronger than that observed in the case of the smaller promoter deletion, Abd-B^D14, because a weak A7 to A6 transformation (the presence of at least a single bristle in tergite A7; Figure 4b) is observed in Abd-B^D14/+ males. However, these promoter deletions do show the expected haplo-insufficient phenotype when they are combined with a chromosome like TSR, Fab-7 that disrupts pairing. In this case, their phenotype is indistinguishable from that of TSR,Fab-7/Abd-B^D14 (see Figure 4c).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Abdominal cuticles of adult males of the genotypes (a) Fab-7/Abd-B^D16, (b) Fab-7/Abd-B^D14 and (c) TSR-1^Sz,Fab-7/Abd-B^D14. Suppression of the Fab-7 phenotype by the deletion mutant Abd-B^D14 is significantly weaker (but still detectable) than in the combination with the point mutation Abd-B^D16 (compare the size of the sixth tergites in a, b and c). Due to the hyperactivation of the wild-type Abd-B gene, the haplo-insufficient phenotype of A7 is suppressed in b (A7 tergite is missing and only a single hair at the tracheal opening, indicated by arrowhead, shows that it is not completely wild type). When pairing is efficiently disrupted c, both hyperactivation and trans-suppression are eliminated, and A6 and A7 tergites are equally thin. The thickness of arrows in the drawings under each of the abdomens indicates the relative strength of cis- and trans-regulation.

Df(3R)U110 is an unusual case. It is the largest of the Abd-B deficiencies that still retains iab-7, and it extends from near the 3′ end of Abd-B into the 90A region. As expected from the large size of this deficiency, it exhibits by far the strongest enhancement of trans-regulation when in trans to iab-7^Sz. However, unlike any of the other deficiencies, the extent of complementation by the trans copy of iab-7 varies from hemitergite to hemitergite and sometimes appears to be clonal even within a hemitergite (Figure 6d). Df(3R)U110 also differs from the other Abd-B deficiencies in that it fails to complement the loss-of-function iab-7 phenotype of Df(3R)R59, a deficiency that extends proximally from the 3′ end of the Abd-B gene to the bxd region. The only sign of some complementation in Df(3R)U110/Df(3R)R59 males is the presence of black pigmentation in tergite A6 and A7, a phenotype corresponding to the result of the pairing-insensitive, long-range interaction described by Hopmann et al. (1995) and Hendrickson and Sakonju (1995). All of the other Abd-B deletion mutations complement the R59 deficiency in A7 to a degree roughly similar to that observed for iab-7^Sz (data not shown). The lack of this complementation between Df(3R)U110 and Df(3R)R59 does not seem to be due to the fact that these two deficiencies overlap in the region just 3′ to the Abd-B gene. Like Df(3R)U110, the deletion in Abd-B^RD18 overlaps Df(3R)R59 in this 3′ region, yet Abd-B^RD18 shows the same degree of complementation of Df(3R) R59 as that observed for the Abd-B^PSz1, which does not overlap.

The most likely explanation for the unusual behavior of Df(3R)U110 is illustrated in Figure 7. In the case of nonoverlapping deficiency combinations such as Abd-B^RD18/iab-7^Sz (Figure 7a), the iab-7 and Abd-B DNA segments on the two homologues loop out, but remain in relatively close proximity. This also holds for the overlapping deficiency combination Abd-B^RD18/Df(3R)R59 (see Figure 7b). In contrast, the remaining iab-7 and Abd-B DNA segments in the Df(3R)U110/Df(3R)R59 are located on opposite sides of the large unpaired deficiency loops (see Figure 7c), and this distance (which to a first approximation corresponds to the size of the smaller deficiency) may prevent effective trans-regulation. In the case of the Df(3R)U110/iab-7^Sz, the iab-7 deficiency is quite small, and the remaining iab-7 and Abd-B DNA segments would still be in relatively close proximity. (The hypothetical chromosome structure of Df(3R)U110/iab-7^Sz is essentially the same as that of Abd-B^RD18/iab-7^Sz, shown in Figure 7a.) Thus, the difference between the phenotype of Df(3R)U110/iab-7^Sz and that of Df(3R)U110/Df(3R)R59 would suggest that proximity has a strong influence in the restoration of A7 identity by trans-regulation.

Transvection in the Abd-B domain is not zeste-dependent: Because most of the previously documented cases of transvection are dependent on the presence of a wild-type copy of the X-linked gene zeste (Kaufmanet al. 1973; Micol and García-Bellido 1988), we tested whether zeste is also required for trans-regulatory interactions involving the Abd-B gene. Surprisingly, we found that mutations in the zeste locus (z^a, z^op6, z¹, z^v77h) had no apparent affect on any of the trans-regulatory interactions in the Abd-B domain described above. Hendrickson and Sakonju (1995) and Hopmann et al. (1995) have also found that zeste function is not required for the long-range pairing-insensitive trans-interaction that they have uncovered in the Abd-B domain.

Is transvection tissue or stage specific? We wondered whether iab-7 is able to trans-regulate the Abd-B gene in tissues besides those giving rise to the adult cuticle. To investigate this question, we used immunostaining to examine the Abd-B expression in embryos of different genetic backgrounds.

The various control embryos gave the expected patterns of Abd-B protein expression described by others (Celniker et al. 1989, 1990; DeLorenzi and Bienz 1990). In wild type, ABD-B protein increases in a stepwise fashion between PS10 and PS14 (Celnikeret al. 1989; Sánchez-Herrero 1991; see also Figure 8a). In Fab-7 homozygous embryos, the wild-type pattern of ABD-B protein is observed in all parasegments except for PS11, where Abd-B is expressed in a PS12-like pattern (not shown; see Galloniet al. 1993). As anticipated, no protein can be detected when the Fab-7 embryos are also homozygous for the Abd-B point mutation, Abd-B^D16 (not shown; Bouletet al. 1991). In the case of Fab-7Abd-B^D14 homozygotes, protein is observed in the parasegment PS14, but not in PS10–PS13 (Figure 8b). This is expected because the Abd-B^D14 deletion removes the transcription start site of the class A Abd-B transcripts, which are specifically expressed in PS10–PS13. Finally, there is no ABD-B protein in PS10–PS12 of embryos homozygous for the deletion Df(3R)R59, whereas the normal pattern of protein staining is observed in PS13 (where the expression of Abd-B is under the control of the iab-8 cis-regulatory domain) and in PS14 (Figure 8c).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Structure of Abd-B deletions used in complementation tests with the deletions iab-7^Sz and Df(3R)R59, shown on the molecular map. Continuous bold line indicates the regions covered by the deletions (Karchet al. 1985; Gyurkovicset al. 1990; Galloniet al. 1993; Hopmannet al. 1995); dashed line, uncertainty in the locations of the endpoints; triangle, the position of the insertion UC21-10, l-d (McCallet al. 1994). Transcriptional unit of class A and part of class B, C and γ are indicated under the molecular map (Zavortnik and Sakonju 1989).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Abdominal cuticles of adult males of the genotype: (a) Sb Abd-B^D16/iab-7 ^Sz, (b) Abd-B^D14/iab-7^Sz, (c) Abd-B^RD18/iab-7 ^Sz and (d) Df(3R)U110/iab-7^Sz. The point mutation Abd-B^D16 does not complement iab-7^Sz to any detectable degree, as A7 appears to be completely transformed into A6. However, the size of the A7 tergite and sterniteis progressively reduced in b–d as compared to a, indicating that deletions that remove increasingly larger regions upstream of the Abd-B class A transcription unit complement the iab-7^Sz mutation to increasingly higher degrees. In the right column, the diagrammatic interpretation of the phenotype for each genotype is presented. The thickness of curved arrows corresponds to the increasing strength of trans-interaction (cis-interaction is not shown).

We next examined the Abd-B expression pattern in Fab-7Abd-B^D16/Df(3R)R59 and Fab-7Abd-B^D14/Df(3R)R59 embryos. On the basis of regulatory interactions inferred from the phenotype of the adult cuticle, we anticipated that the Abd-B gene on the R59 homologue would be trans-activated in the embryos carrying the Abd-B^D14 deletion mutant, but not in the embryos carrying the Abd-B^D16 point mutation. For Fab-7Abd-B^D14/Df(3R)R59 embryos, we found the same levels of ABD-B protein in the CNS of PS11 and PS12 (see Figure 8d). This is the result expected if the iab-7 regulatory domain on the Fab-7 chromosome is able to trans-regulate the Abd-B gene on the R59 homologue in PS11 and PS12. Thus, for this mutant combination, trans-regulation is the same in the embryonic CNS as it is in the adult. Surprisingly, although we were unable to detect trans-regulation of the Abd-B gene on the R59 homologue in adult Fab-7Abd-B^D16/Df(3R)R59 animals, trans-regulation could be detected in the embryonic CNS. As in Fab-7Abd-B^D14/Df(3R)R59 embryos, we found the same levels of ABD-B protein in the CNS in PS11 and PS12 of Fab-7Abd-B^D16/Df(3R)R59 embryos (not shown).

Although we found clear evidence for trans-regulatory interactions between the Fab-7Abd-B⁻ and R59 homologues in the embryonic CNS, such interactions were not evident in other embryonic tissues. Thus, in animals of the same two genotypes, we were unable to detect ABD-B protein in either the ectoderm or mesoderm of PS11 and PS12. This was true even at the beginning of germ-band retraction, when Abd-B expression is most readily evident in wild-type embryos in these parasegments (not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

The position of the iab-7 regulatory region and the Abd-B class A transcription unit (marked with bold line and arrow, respectively) in different pair-wise combinations of deletions. Continuous line represents DNA; parentheses indicate the breakpoints of deletions. (A) nonoverlapping; (B, C) overlapping combinations of deletions. Note that in B the regions marked with bold are relatively close to each other, and in C they are localized at the far ends of the unpaired region.

In some experiments, we saw faint Abd-B specific staining of CNS in PS10 of both Fab-7Abd-B^D14/Df(3R)R59 and Fab-7Abd-B^D16/Df(3R)R59 embryos. This observation suggested that iab-5 and iab-6 may also be able to trans-regulate the Abd-B gene on the R59 homologue. To test this possibility, we stained Mcp^B116iab-7^SzAbd-B^D16/Df(3R)R59 embryos. The Mcp^B116 mutation is a deletion that removes a putative boundary between iab-4 and iab-5, ectopically activating the PS10-specific iab-5 in PS9 (Karchet al. 1994). In animals carrying a wild-type Abd-B gene in cis, the ectopic activation of iab-5 results in the appearance of ABD-B protein in PS9 of the embryonic CNS at a level normally found in PS10 (Sanchez-Herrero 1991). As iab-7 is also deleted on this Mcp^B116 chromosome, Abd-B expression in both PS11 and PS12 can only be driven by the iab-6 regulatory domain. Since the Abd-B gene in cis to Mcp^B116 and iab-7^Sz does not code for a detectable protein, ABD-B protein can only be expressed in these embryos by trans-regulation of the Abd-B gene on the Df(3R)R59 chromosome in PS9–PS12. As can be seen in Figure 8e, we can detect weak ABD-B expression in PS11–PS12, and an even weaker, but unmistakable expression is seen in PS9–PS10. This pattern of protein expression argues that the iab-5 and iab-6 cis-regulatory domains on Mcp^B116iab-7^SzAbd-B^D16 chromosome are capable of trans-regulation much like the iab-7 regulatory domain.

Long distance trans-regulatory interactions in the embryonic CNS: The results described in the previous section indicate that trans-regulatory interactions in the Abd-B domain may be significantly stronger in the embryonic CNS than in the adult epidermis. If this were true, it might be possible to detect trans-regulation in the CNS of embryos carrying chromosomal rearrangements that substantially perturb pairing in BX-C and eliminate transvection in the cells giving rise to the adult epidermis. To investigate this possibility we examined Abd-B expression in the CNS of embryos heterozygous for either Fab-7Abd-B^D14 or Fab-7Abd-B^D16 and chromosomes that have rearrangements that disrupt Abd-B and Ubx trans-regulation in the adult. These rearrangements include TSR-11A, TSR-59A, iab-7^MX2 (In(3LR)64A; 89A;89E; Karchet al. 1985), iab-7¹⁶⁴ (In(3LR)65;89E superimposed on In(3R)1000 with breaks in 81C and 90C; Celnikeret al. 1990).

Even though these rearrangements suppress trans-regulatory interactions in the cells giving rise to the adult epidermis, they do not prevent trans-regulatory interactions in the embryonic CNS, and we observe equal levels of ABD-B protein in PS11 and PS12 (as would be expected if the iab-7 regulatory domain on the Fab-7Abd-B⁻ chromosome is driving the expression of the Abd-B gene on the rearranged chromosome). The one exceptional case is the doubly rearranged iab-7¹⁶⁴ chromosome. In both Fab-7Abd-B^D16/iab-7¹⁶⁴ and Fab-7Abd-B^D14/iab-7¹⁶⁴ embryos, the Abd-B antibody staining is weak and irregular (Figure 8h), suggesting that the configuration of the rearrangement in this particular chromosome interferes with trans-regulation even in the CNS.

Long-distance regulatory interactions in the embryonic CNS are suppressed by the uninterrupted pairing of homologues: Our ability to detect long-distance trans-regulatory interactions in the CNS prompted us to investigate the parameters governing this phenomenon further. The chromosomal rearrangement In(3R) Fab-7iab-7^R5 was isolated as a revertant of Fab-7. It has a distal breakpoint at position +152 on the molecular map in the BX-C, very close to the 3′ end of the Abd-B m transcription unit. From its position, it seems likely that the break in BX-C may leave both the iab-7 regulatory domain and the Abd-B gene functionally intact. The proximal breakpoint is in the region 87C1-2,3.

Because iab-7 as well as iab-5 and iab-6 are separated by a large distance from the Abd-B gene in this inversion-bearing chromosome, these regulatory domains should be unable to contact the Abd-B gene, and the gene should not be expressed in PS10–PS12 in the CNS. This is the case in embryos homozygous for the In(3R)Fab-7iab-7^R5 inversion; we only see Abd-B staining in PS13–PS14 (not shown). Surprisingly, however, a different result is obtained when In(3R)Fab-7iab-7^R5 is combined with the deficiency chromosome Df(3R)P9 that removes the entire BX-C. In this combination, we detect a low but uniform level of Abd-B antibody staining in the CNS of PS11 and PS12 (Figure 8f).

From these results, it would appear that iab-7 is able to regulate a distantly located Abd-B gene on the same chromosome when the inversion is in trans to a BX-C deficiency, but not when it is homozygous. In the former case, two factors might facilitate this regulatory interaction. First, the loop formed by homologous pairing between the deficiency and the inversion chromosomes would bring the BX-C sequences on either side of the inversion breakpoints in close proximity. Second, there would be unpaired regions around the breakpoints of the inversion that at one end might expose the iab-7 regulatory domain and at the other end might expose the Abd-B gene and 5′ flanking regions. In the latter case, the iab regulatory domains and the Abd-B gene would not be brought in close proximity when the two inversion chromosomes pair. Instead, these sequences would be separated by the length of the inversion. Additionally, the pairing of the BX-C regions across the inversion breakpoint would be uninterrupted.

We wondered whether unpaired DNA segments could by themselves facilitate long-distance interactions. To investigate this possibility, we combined In(3R)Fab-7iab-7^R5 with another inversion, In(3R)Fab-7iab-7^R7, which has its breakpoints in 87D1,2 and 89E3,4 (Gyurkovicset al. 1990). Because these two inversions have similar but not precisely identical breakpoints, they will be paired over much of their length without forming the loop that is typically generated when an inversion chromosome pairs with a structurally wild-type homologue. However, at the proximal inversion breakpoint of one chromosome, there will be a locally unpaired DNA segment that is homologous to an unpaired DNA segment at the distal breakpoint of the other chromosome (see Figure 9). Thus, at the distal end of R5 and the proximal end of R7, there will a large unpaired DNA segment containing the 87C interval. Similarly, at the proximal end of R5 and at the distal end of R7, there will be a short ~10-kb unpaired DNA segment from BX-C that is located just downstream of the Abd-B transcription unit (between map position +141.5–143 and +152) (see Figure 9). If these unpaired DNA segments are able to facilitate long-distance regulatory interactions, Abd-B expression should not be restricted to PS13 and PS14 but instead should also be observed in PS11 and PS12. The results we obtained are consistent with this model. As illustrated in Figure 8g, Abd-B expression can be detected in PS11 and PS12 in the CNS of the R5/R7 embryos, but not in embryos homozygous for either the R5 or the R7 inversion. The expression pattern in the R5/R7 trans-heterozygous embryos is somewhat irregular, suggesting that trans-regulation occurs in some cells but not in others. (We suspect that the Abd-B expression in these embryos may be due to the activity of iab-7 on the R5 chromosome alone. This possibility is suggested by the observation that there is no detectable ABD-B protein in PS10–PS12 of In(3R)Fab-7iab-7^R7/Df(3R)P9 or In(3R)Fab-7iab-7^R7/Df(3R)R59 embryos. This observation would argue that iab-7 on the R7 chromosome is inactivated by the breakpoint in the BX-C.)

These findings indicate that unpaired chromosomal DNA segments can facilitate long-distance interactions. One plausible hypothesis is that the two large unpaired 87C chromosomal DNA segments may be able to interact with each other (a form of “homologous pairing”), and it is this interaction that brings the inverted BX-C regulatory regions in close proximity to the Abd-B gene (see Figure 9). This hypothesis is supported by our finding that in ~50% of salivary gland polytene chromosomes, the In(3R)Fab-7iab-7^R5 and In(3R)Fab-7iab-7^R7 homologues form a “ring” or “circle,” apparently held together by pairing between the two 87C loops (not shown).

A prediction of this hypothesis is that long-distance regulatory interactions should only be observed when there are unpaired 87C chromosomal DNA segments at both the proximal and the distal breakpoints. To test this prediction, we recombined R5 and R7 to generate a chromosome, R5^LR7^R, which contains the proximal breakpoint of R5 and the distal one of R7. In the R5^LR7^R chromosome, the 87C region is deleted, and a ~10-kb BX-C segment 3′ to the Abd-B gene is present at both breakpoints in an inverted orientation. When the R5^LR7^R chromosome is combined with the R5 chromosome, there is only a single, unpaired 87C DNA segment located at the distal R5 breakpoint (see Figure 9). As anticipated, R5^LR7^R/R5 trans-heterozygotes differ from R5/R7 trans-heterozygotes in that they do not form a “circle” in salivary gland polytene chromosomes (data not shown). However, even though this “circle” cannot be detected, the Abd-B expression pattern in PS11 and PS12 of R5^LR7^R/R5 animals is indistinguishable from that observed in R5/R7 animals (data not shown). This is also true for the combination of R5^LR7^R/R7, which has only a single, unpaired 87C DNA segment, in this case at the proximal R7 breakpoint.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Immunohistochemical staining of ABD-B proteins in the ventral trunk region of the CNS dissected from Drosophila embryos. (a) wild type (+/+). Staining gradually increases from PS10 to PS14. (b) Fab-7Abd-B^D14/Fab-7Abd-B^D14. Staining is missing from PS10 to PS13. The intensity of staining in PS14 is similar to that in wild type. (c) Df(3R)R59/Df(3R)R59. Staining is missing from PS10 to PS12. (d) Fab-7Abd-B^D14/Df(3R)R59. A very low level of protein is detected in PS10 and stronger staining of equal intensity can be seen in PS11 and PS12, indicating trans-regulation of the wild-type Abd-B. (e) Mcp^B116iab-7^SzAbD-B^D16/Df(3R)R59. Weak staining of equal intensity is detected in both PS9 and PS10, and stronger staining is apparent in both PS11 and PS12. (f) In(3R)Fab-7iab7 ^R5/Df(3R)P9. Weak staining is observed in PS11 and PS12, indicating that, although the paracentric inversion In(3R)Fab-7iab7 ^R5 separates the cis-regulatory region iab-7 from the Abd-B promoter, they can interact via formation of a loop in hemizygous condition. (g) In(3R)Fab-7iab-7^R5/In(3R)Fab-7iab-7^R7. Weak staining with irregular pattern is detected in both PS11 and PS12, indicating trans-regulation in some cells. (h) In(3LR)iab-7¹⁶⁴/Fab-7Abd-B^D16. The doubly rearranged chromosome In(3LR)iab-7¹⁶⁴ efficiently disrupts pairing of homologous chromosomes, and staining in PS11 and PS12 is significantly decreased as compared to d. Additionally, the pattern of the remaining staining is irregular, indicating that trans-regulation occurs only in some of the cells.

Hence, contrary to our expectations, the regulation of Abd-B by a distant iab-7 regulatory domain does not appear to require interactions between unpaired 87C chromosomal DNA segments at the proximal breakpoint of one homologue and the distal breakpoint of the other. On the other hand, there is still a requirement for an unpaired DNA sequence because no long-distance regulation is observed when R5^LR7^R chromosome is homozygous (see Figure 9). We suspect that this unpaired sequence is most likely the short 10-kb BX-C DNA segment immediately downstream of the Abd-B gene. As mentioned above, in the R5/R7 combination there is an unpaired copy of this DNA sequence at the proximal breakpoint of one homologue that could potentially interact with a copy of the same sequence at the distal breakpoint of the other homologue (see Figure 9). In the R5^LR7^R chromosome, there are two copies of this sequence, one at each breakpoint. Consequently, when R5^LR7^R is combined with either R5 or R7 there are three copies of the sequence (see Figure 9). One of these is always unpaired and could potentially interact with the paired copies at the opposite breakpoint. (In this hypothesis, the presence of a single, unpaired 87C region at one end of the otherwise paired combinations of R5^LR7^R/R7 or R5^LR7^R/R5 may only facilitate the looping out of BX-C material and would not play a major role in long-distance regulation.)

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

The structure of chromosome pairs carrying different combinations of the inversions In(3R) Fab-7iab-7^R5, In(3R)Fab-7iab-7^R7 and In(3R)Fab-7R5^LR7^R. Some of these combinations exhibit trans-regulation (marked by +) in the embryonic CNS, and others do not (−). Continuous lines represent chromosomes, short vertical lines the breakpoints of the inversions, and bold arrows the Abd-B class A transcription unit. Thin arrows show the position and the orientation of cis-regulatory regions of Abd-B. Regions within dashed areas are supposed to play a crucial role in transvection. Note that in trans-regulating combinations (+) these regions are unpaired at least at one of the breakpoints.