The promoter targeting sequence facilitates and restricts a distant enhancer to a single promoter in the Drosophila embryo

Large developmentally regulated gene complexes usually contain enhancers that activate their promoters from great distances. However, many enhancer-interacting transcription factors can only stimulate transcription over very short ranges when assayed in yeast(Dorsett, 1999). It is conceivable that more complex organisms adopt additional mechanisms to facilitate long distance enhancer-promoter communications. One such mechanism is suggested by transacting factors Chip and Nipped-B, which by themselves do not activate transcription but facilitate the action of a remote enhancer in the Drosophila cut locus (Rollins et al., 1999; Torigoi et al.,2000). A different type of mechanism is provided by specializedcis-regulatory elements. The locus control regions (LCRs) found in the β-globin locus, the immunoglobulin locus, and several other large loci define a class of gene and tissue-specific long range acting elements (Bulger and Groudine,1999; Fernandez et al.,1998; Grosveld et al.,1993; Li et al.,2001). These elements usually contain cell type-specific transcription regulator interacting sites and confer to transgenes high levels of insertion site-independent gene expression. However, a genericcis-regulatory element that does not possess enhancer activity but can facilitate long-range enhancer-promoter communications has not been found.

Recent genetic studies in the homeotic selector gene Abdominal-Bfrom the Drosophila bithorax complex (BX-C) have identified acis-regulatory region, termed transvection mediating region(tmr), that may facilitate long-range enhancer-promoter interactions(Hendrickson and Sakonju,1995; Hopmann et al.,1995; Sipos et al.,1998). Abd-B contains an extended 3′ regulatory region that is functionally subdivided into distinct enhancer domains calledinfra-abdominal (iab)-5, iab-6, iab-7 andiab-8 by insulators, or boundary elements such asFrontabdominal (Fab)-7 and Fab-8(Barges et al., 2000;Hagstrom et al., 1996;Karch et al., 1985;Mihaly et al., 1998;Mihaly et al., 1997;Zhou et al., 1999;Zhou et al., 1996)(Fig. 1A,B). Enhancer elements from the 3′ regulatory region can activate the Abd-B promoter over the intervening insulators and long distances. These enhancers continue to activate Abd-B when the 3′ regulatory region is translocated to different chromosomal locations, or even to a different chromosome (from the third to the Y chromosome)(Hendrickson and Sakonju,1995; Hopmann et al.,1995; Sipos et al.,1998). This strong regulatory interaction depends on the 9.5 kbtmr (Hopmann et al.,1995). These observations suggest that an enhancer-facilitating mechanism exists in Abd-B, possibly within the tmr. Consequently, a novel cis element, the promoter targeting sequence(PTS) (Zhou and Levine, 1999)has been identified from the tmr. The PTS has a distinctive anti-insulator activity, allowing an enhancer to activate its promoter over the intervening insulator in transgenic embryos. In the presence of an insulator, the PTS also appears to have a promoter targeting activity,allowing an enhancer to activate only one promoter when two are present in the transgene. We report that the PTS facilitates long-range enhancer-promoter interactions in transgenic embryos, and that it mediates the promoter targeting function by restricting enhancer activities to a single promoter. We also show that the PTS functions only when itself and an insulator are located between an enhancer and a promoter.

To generate P-transgene shown inFig. 2A,B, the 1.6 kb IAB8 enhancer was inserted either at a BamHI site upstream of or aPstI site downstream of the Transponsase(Tp)-lacZ gene of the C4PLZ vector(Zhou et al., 1996). Construct#17 and #28 were generated by inserting the 5.3 kbBamHI-HindIII fragment of the tmr region at theBglII site, 3′ of lacZ. The forward orientation (IAB8 proximal to lacZ gave rise to #17, whereas the reverse orientation(IAB8 distal to lacZ) generated #28. Transgene W14 was made similar to the construct in Fig. 5A of Zhou and Levine (Zhou and Levine,1999), except a 290 bp 5′ DNA sequence was used instead of the 625 bp PTS. To build construct W32, a 2 kb HindIII andPstI fragment (Fig. 1B) from the tmr containing 290 bp PTS, 590 bpFab-8 and additional 1.2 kb 5′ DNA was cloned into theBamHI site between two FRT sites previously inserted into theBamHI site of a pBluescript SK+ that lacks theBglII site. In parallel, a 2.7 kb EcoRI fragment(Fig. 1B) from the tmrthat contains the 1.6 kb IAB8 enhancer was inserted into the PstI site of the C4PLZ vector. The FRT flanked PstI-HindIII fragment was then excised as a BglII fragment and inserted into theBglII site of the IAB8-containing C4PLZ vector. To make W78 and W79,the 1 kb IAB5 enhancer is first inserted into the eve-lacZ 3′located PstI site of -42 eve Casper(Zhou et al., 1996). Theeve promoter was then replaced by Tp promoter. AXbaI site located between PstI and the 3′ oflacZ was converted into a NotI site creatingTpCasperNIAB5 (TpCasperN that lacks IAB5 was also made, see later). In the meantime, the 340 bp SpeI-BamHI suHwinsulator and the BglII fragment of FRT flanked 625 bp PTS were sequentially inserted into SpeI-BamHI and BamHI sites of a modified pBluescript that contains an additional NotI site converted from the KpnI site. The NotI fragment containing FRT sites, PTS, and suHw was inserted into the NotI site ofTpCasperNIAB5. W81 and W82 were generated similarly, except IAB5 was included in the NotI fragment and inserted into theTpCasperN vector.

P-element transformation vectors containing lacZ andwhite reporter genes were introduced into the Drosophilagermline by injecting yw⁶⁷ embryos as described previously(Rubin and Spradling, 1982). Between 30 and 60 independent transformants were obtained for each of the recombinant P-element shown. In situ hybridization was performed essentially as described in previous reports (Tautz and Pfeifle, 1989; Zhou et al., 1999).

Transgenic flies expressing the Flip recombinase were kindly provided by Gary Struhl and Steve Small (Wu et al.,1998). To recombine different FRT-flanked DNA element away from the transgene, females carrying the transgene were mated with males that express the Flp recombinase under the control of a sperm-specifictubulin promoter (Wu et al.,1998). In F₁ males, the recombinase binds the FRT sites and deletes the intervening DNA. These male flies were collected and mated toyw virgin females to establish stocks that are subsequently analyzed by RNA in situ hybridization.

Embryos from different transgenic strains were collected and stained with anti white or lacZ RNA probes in parallel. Enhancer strength was quantified by measuring the differential absorption of transmitted light(ΔEV) between spot a (unstained region) and b (stained region) in the embryos in Fig. 5A using a digital spot light meter (Sekonic L608). The reading reflects the relative intensity (2^ΔEV-1) of the staining that is linear with staining reaction time during a 1 hour incubation (seeFig. 5B). It is assumed that alkaline phosphatase (AP) activity represents enhancer strength, which can be expressed as: enhancer strength=constant ×(2^ΔEV-1)/time. To compare enhancer strength, heterozygous embryos from different samples are fixed and stained in parallel for 45 minutes. The ΔEV values of approximately 30 embryos are measured. Enhancer strengths are plotted as bar graphs inFig. 5C,D.

Previous studies have showed that the Fab-8 insulator could block enhancers (including IAB8) from activating a promoter when it is interposed between the enhancer and its promoter(Zhou et al., 1999;Zhou and Levine, 1999;Barges et al., 2000). A recent study has also demonstrated that the 625 bp PTS element is able to overcome the enhancer blocking effect of the Fab-8 insulator(Zhou and Levine, 1999). We tested whether the minimal 290 bp 5′ DNA from the 625 bp PTS could overcome the Fab-8 insulator, and, in addition, facilitate the IAB8 enhancer activity. We studied a 5.3 kb BamHI-HindIII fragment from the tmr that contains the minimal 290 bp of the PTS,the 580 bp Fab-8 insulator and the 1.6 kb IAB8 enhancer(Zhou et al., 1999;Zhou and Levine, 1999) (seeFig. 1B).

The IAB8 enhancer directs a narrow band of transcription in the posterior region of the embryo (see arrow in Fig. 2A). Similar to other early Drosophila enhancers that have been tested (Ohtsuki et al.,1998; Zhou et al.,1996), its activity attenuates as its distance from the promoter increases (compare Fig. 2A with 2B). This 5.3 kb tmr was placed in both the forward(5′→3′ when IAB8 is between the promoters andFab-8/PTS, see construct #17 inFig. 2) and the reverse orientation (construct #28 in Fig. 2). To monitor transcriptional activity of the IAB8 enhancer,embryos from individual transgenic strains were collected and subjected to whole-mount RNA in situ hybridization for the white (w) orlacZ genes.

In the forward orientation, IAB8 weakly stimulated both the divergently transcribed w and Tp-lacZ genes(Fig. 2C;Table 1). Its activity on theTp promoter is very similar to that of the 1.6 kb IAB8 alone at the same location (Fig. 2B),suggesting that the distally located PTS and Fab-8 do not affect the communication between IAB8 and the transgenic promoters. In the reverse orientation, however, IAB8 activates the transgenic promoters despite the intervening Fab-8 insulator. In any particular strain, IAB8 exhibits the selective activation of either the Tp-lacZ(Fig. 2D) or the w(Fig. 2E) gene. This effect is seen in about 50% of the transgenic lines(Table 1). In the remaining strains, IAB8 does not activate any of the transgenic promoters due to the enhancer blocking effect of the Fab-8 insulator (data not shown). These results are similar to that of the previous study(Zhou and Levine, 1999),suggesting that the 290 bp PTS exhibits the anti-insulator and promoter targeting activities. This is confirmed by transgene W14(Table 1), where PTS could overcome a heterologous suHw insulator and target IAB8 to thew or Tp promoters. It should be noted that in the forward orientation (Fig. 2C), IAB8 is located 5.5 kb away from the w promoter and 4.9 kb away from thelacZ promoter, whereas in the reverse orientation(Fig. 2D,E), it is 8.2 kb away from w and 7.5 kb away from lacZ. Rather than reducing the activity, the greater distance between IAB8 and the promoters caused by inserting Fab-8 and PTS resulted in an increase of IAB8 enhancer activity. This is seen by comparing the activities of IAB8 on Tp-lacZin Fig. 2B versus 2D, orw in Fig. 2C versus 2E. In either case, the enhancer activity is much stronger when both PTS and Fab-8 are located between IAB8 and the promoters. These results indicate that PTS may facilitate the long distance interactions between IAB8 and the w or the Tp promoter.

To confirm the enhancer facilitating activity and to eliminate position effects due to differential chromosomal insertion sites of the transgene, we used the Flp-FRT system, which causes the removal of FRT-flanked DNA sequences from the transgene after introducing the Flp recombinase by genetic cross(Golic and Lindquist, 1989). This technique permits the analysis of the transgene in the same chromosomal context before and after the test DNA is removed. We flanked a 2.0 kbPstI-HindIII fragment from the tmr region (seeFig. 1B) that contains both the minimal 290 bp PTS and the 580 bp Fab-8 insulator with the FRT sites and placed this group of elements between the 3′ end of lacZand 2.7 kb EcoRI fragment from the same region(Fig. 1B) that contains the 1.6 kb IAB8 enhancer (W32 in Fig. 3). When transgenic embryos were analyzed, results similar to those without FRT sites were obtained. In about half of the transgenic strains IAB8 selectively activates one of the two divergently transcribed wand the Tp-lacZ genes (Table 1). In the remaining strains, IAB8 does not activate any of the transgenic promoters. An example of a PTS-mediated IAB8-w interaction is shown in Fig. 3A,B. Here,IAB8 strongly activates w but not the closely positioned Tppromoter (compare Fig. 3A with 3B). However, after the removal of PTS and Fab-8 by Flp-mediated recombination, the intense and selective IAB8-winteraction disappeared. Instead, IAB8 activates both w andTp-lacZ, but with greatly reduced activity, despite the fact that the enhancer is now 2.0 kb closer to the promoters(Fig. 3C,D). Similar result was obtained when we place the Fab-8 and PTS elements between the 3′ end of lacZ and the heterologous Neural Ectoderm Enhancer(NEE) from the rhomboid gene (Ip et al., 1992) (data not shown). These results strongly suggest that the PTS, in combination of the Fab-8 insulator, facilitates enhancer-promoter interaction, and that it restricts the enhancer activity to a single promoter.

To confirm that the enhancer-facilitating and single promoter activating effects were due to the PTS but not a possible synergy between DNA sequences located within the tmr, we tested the PTS in the absence ofFab-8 and other tmr sequences. The PTS, the heterologoussuHw insulator (Cai and Levine,1995; Dorsett,1993; Geyer and Corces,1992), and the IAB5 enhancer(Busturia and Bienz, 1993) were placed at the 3′ of lacZ in the order given (see construct W81 under Fig. 3E,F). PTS andsuHw were flanked by a direct repeat of FRT sites so that both elements could be removed by recombination. In about one third of all transgenic strains, IAB5 selectively activated only one of the two (wor Tp) promoters (see Fig. 3E,F for selective IAB5-lacZ interaction), suggesting that the PTS mediates promoter targeting in these transgenic lines(Table 1). The activities of IAB5 on the targeted promoters are consistently strong, with slight variations among different strains (Fig. 5D). In the remaining lines, IAB5 does not activate either of the promoters. Presumably, PTS does not function in these lines, and the IAB5 enhancer is blocked by the intervening suHw insulator. The enhancer-facilitating activity was confirmed by Flp-FRT analysis(Fig. 3G,H). Similar toFig. 3A-D, the simultaneous removal of both PTS and suHw dramatically reduced IAB5-lacZinteraction (over 10-fold reduction, seeFig. 5D). By contrast,IAB5-w interaction, which was undetectable before the recombination,could now be detected (compare w activity inFig. 3G with 3E).

To confirm that the PTS could indeed help IAB5 overcome the suHwinsulator, we also constructed transgenes W78 and W79 that are similar to W81,but contain an FRT-flanked PTS (Table 1). In the former, PTS was placed between the 3′ end oflacZ and the suHw insulator, while in the latter, PTS was interposed between suHw and IAB5. Transgenic embryos carrying either of these constructs exhibit a selective activation of w orlacZ in any given strain (see Fig. 4A,B for selective activation of lacZ by IAB5 in transgene W78). Removal of the PTS by Flp-mediated recombination caused the loss of IAB5 activated transcription, as IAB5 became blocked by thesuHw insulator (compare Fig. 4B with 4D). These experiments indicate that the PTS exhibits the anti-insulator and promoter targeting activities when it is either upstream or downstream of an insulator.

In summary, these results clearly demonstrate that the enhancer facilitating function depends on the PTS element, not other unknown elements located within the tmr. This result also indicates that the promoter targeting activity is not a result of random positional effect, or preferential insertion of the transgene into specific chromosomal locations that may silence one of the promoters present in the transgenic vector. It is due to a PTS-dependent, active restriction of the enhancer activity to only one of the two available promoters (Fig. 3G,H). The enhancer-facilitating and single promoter activating effects are genetically stable in that they are memorized in up to 60 generations without a loss or change of promoter targeting.

We also conducted semi-quantitative analyses of PTS-mediated enhancer facilitating and single promoter activating activities by quantifying RNA in situ hybridization. The staining intensity (as measured by optical absorption of stained Drosophila embryos,Fig. 5A) shows linear relationship with staining time 60 minutes after the addition of substrates for alkaline phosphatase (AP) (Fig. 5B). Assuming that the AP activity directly reflects enhancer strength and that the AP activity remains constant during an 1 hour incubation, we can compare the activity of the same enhancer in different transgenes if the embryos are fixed and stained in parallel. Using this method, we found that IAB8 is two- to fivefold stronger when it is located less than 100 bp from the Tp promoter than when located 5.5 kb downstream of Tp-lacZ (Fig. 2A,B, and Fig. 5C). We then compared IAB8 and IAB5 activities with or without facilitation by the PTS. We found that PTS facilitates IAB8 about eightfold and IAB5 ten- to 17-fold, respectively (Fig. 5C,D). In transgenic lines (W81) showing specific IAB5-lacZ interaction, no w expression could be detected,but after the Flp-mediated removal of PTS and suHw, w became activated by IAB5 (Fig. 5D). In two of the four lines (W81-2 and W81-4) w expression was detectable within 45 minutes, whereas in the remaining two (W81-1 and W81-3), wexpression could be detected only after extended incubation (∼2 hours,data not shown). The strength of IAB5-w or IAB5-lacZinteractions after the removal of suHw/PTS is similar to that of IAB5 alone originally cloned at the same location (data not shown). These results provided quantitative evidence for the enhancer-facilitating and single promoter-activating activities of the PTS.

The results shown in Fig. 2also suggest that PTS must be interposed between an enhancer and a promoter. To test whether this is true, the PTS and suHw DNA were placed distal to the IAB5 enhancer, downstream of lacZ (W82 inTable 1). From over 40 transgenic strains isolated and examined, none displayed enhancer facilitating and the distinctive single promoter-activating effects(Table 1). In these lines, IAB5 activated both the w and Tp promoters, with activities similar to that of IAB5 alone in the same location (data not shown). This result, together with the data from Fig. 2 suggests that PTS activity is location dependent, in that it only functions when itself (and an insulator) is located between the enhancer and the promoter.

In summary, we have shown that PTS facilitates long-range enhancer-promoter interactions in transgenic embryos. The enhancer facilitating activity depends on the anti-insulator and promoter targeting functions in that it only facilitates an enhancer when it is targeted to a promoter (compareFig. 2C with 2D,E). We have also provided evidence that the promoter targeting function is due to restricting the access of an enhancer to a single promoter, and not due to,for example, positional effects that might inactivate the other promoter present in the transgene. As shown in Fig. 3, the IAB5 or the IAB8 enhancer alone is capable of activating both the w and Tp promoters after the PTS/insulator fragment is removed by recombination. It is possible that the anti-insulator, promoter targeting, and enhancer-facilitating activities are inseparable and are possibly different aspects of the same activity. For example,enhancer-facilitating effect could be at least in part resulted from restricting the enhancer to a single promoter, which would prevent the enhancer from activating other promoters and consequently increase the probability of activating the `target' promoter.

In the Abd-B locus, the enhancer-facilitating property of the PTS could help distal enhancers such as IAB7 overcome the long distances and direct robust transcription activation of Abd-B. This notion is consistent with the genetic functions of the tmr and the loss-of-function phenotype of PTS mutants(Castrillon et al., 1993;Gyurkovics et al., 1990;Zhou and Levine, 1999). The single promoter-activating function could ensure that enhancers in the BX-C activate the cognate Abd-B promoter only. In this study, we have also shown that promoter targeting only occurred when PTS/insulator was placed between an enhancer and a promoter. This location-dependent characteristic of PTS suggests that strategic placement of PTS within Abd-B can facilitate specific enhancers. For example, in the Abd-B locus, PTS and Fab-8 are located between IAB7 (but not IAB8) and Abd-Bpromoter (Fig. 1A). In this arrangement, PTS may facilitate only the distal IAB7 but not the proximal IAB8 to the Abd-B promoter. In fact, this appears to be the case, as deletion of PTS causes a loss-of-function transformation of the seventh but not the eighth abdominal segment(Castrillon et al., 1993;Gyurkovics et al., 1990;Zhou and Levine, 1999). On this note, it is possible that multiple PTS-like elements exist in the BX-C to mediate long-distance regulatory interactions.

It can be seen from Fig. 3that the PTS-facilitated single enhancer-promoter interaction is stronger than the sum of enhancer-w and enhancer-Tp interactions when the IAB5 or IAB8 enhancers are placed alone at the 3′ of lacZ. These results suggest that the PTS-mediated enhancer facilitation is not just the consequence of restricting an enhancer to a single promoter, it must also actively promote long distance enhancer-promoter communications. It is possible that the PTS functions by establishing an insulator-insensitive,stable chromatin structure between the enhancer and a promoter, e.g. forming a`stable loop' and bringing the enhancer closer to the promoter(Fig. 6). Similar `loop'hypothesis has been proposed based on genetic analysis of the enhancer-promoter interactions in the Abd-B locus(Galloni et al., 1993). This model can not only explain the anti-insulator activity but can also account for the enhancer-facilitating and the single promoter activating activities. Such a stable association would ensure that enhancer-interacting activators are constantly present at the promoter, which would result in efficient promoter activation and, at the same time, prevent the enhancer from activating other promoters.

Our study also suggests that PTS functions as a generic element in transgenic embryos as it can target and facilitate a heterologous neuroectoderm enhancer NEE (Ip et al.,1992). It is possible that in other large genetic loci such as the odorant receptor gene complex (Mombaerts,1999) and the neural cadherin-like adhesion gene complex(Wu and Maniatis, 1999), where only one among several dozen promoters is activated in any given cell,PTS-like elements may contribute to the promoter-selective transcriptional activation.

In transgenic embryos, the PTS does not appear to exhibit promoter-specific activity as it can target either the w or the Tp promoter present in the transgene. It is not known what determines which of the two promoters to select. One possibility is that the decision is made by the interaction between the PTS and local chromatin structure. Alternatively, the selection could be a stochastic process. In the endogenous BX-C, however, the PTS must target the Abd-B promoter. Additional mechanism(s),therefore, must be in place to ensure enhancer-promoter specificity in theAbd-B locus.

This work is supported by ACS Cancer Center grant, the Pew Charitable Trust, Leukemia Research Foundation and WWW Smith Charitable Trust Fund. We thank Dr Qi Chen, Jian Zhou, Tiffany Helling and Ming Li for excellent technical assistance; and Guoping Da for helping with quantitative analysis. We also thank Drs Mike Levine, Brian Calvi, James Jaynes and Paul Lieberman for helpful comments on the manuscript. Finally, we would like to thank Dr Ellen Puré for the access of the Leica DIC microscope.