Eukaryotic enhancers act over very long distances, yet still show remarkable specificity for their own promoter. To better understand mechanisms underlying this enhancer-promoter specificity, we used transvection to analyze enhancer choice between two promoters, one located in cis to the enhancer and the other in trans to the enhancer, at the yellow gene of Drosophila melanogaster. Previously, we demonstrated that enhancers at yellow prefer to act on the cis-linked promoter, but that mutation of core promoter elements in the cis-linked promoter releases enhancers to act in trans. Here, we address the mechanism by which these elements affect enhancer choice. We consider and explicitly test three models that are based on promoter competency, promoter pairing, and promoter identity. Through targeted gene replacement of the endogenous yellow gene, we show that competency of the cis-linked promoter is a key parameter in the cis-trans choice of an enhancer. In fact, complete replacement of the yellow promoter with both TATA-containing and TATA-less heterologous promoters maintains enhancer action in cis.
EUKARYOTIC enhancers are able to act over long distances, sometimes interacting with promoters hundreds of kilobase pairs away. At the same time, enhancers can show a high degree of specificity, finding and interacting with their own promoter but not with other nontarget promoters. One mechanism for how this specificity is achieved suggests that intrinsic properties of enhancers and promoters limit an enhancer to a particular promoter (Li and Noll 1994; Hansen and Tjian 1995; Kapoun and Kaufman 1995; Merli et al. 1996; Ohtsuki et al. 1998; Sharpe et al. 1998; Butler and Kadonaga 2001; Cai et al. 2001; Conte et al. 2002). Here, we explore this mechanism using the phenomenon of transvection.
Transvection is a process by which a gene can affect the expression of its homologous gene on a separate chromosome in a pairing-dependent manner. It has been observed in Drosophila, where homologous chromosomes are aligned and paired in somatic cells. In addition, transvection and related processes have been observed in fungi, plants, and mammals, where extensive somatic homolog pairing has not been demonstrated, suggesting that even transient pairing interactions might have long-term consequences on gene expression (reviewed in Pirrotta 1999; Wu and Morris 1999; Burgess 2002; Duncan 2002; Kennison and Southworth 2002; most recently Bean et al. 2004).
Our studies explore transvection at the yellow gene of Drosophila. The yellow gene is required for dark pigmentation of the cuticle, including the wings, body, and bristles, and is expressed under the control of tissue-specific enhancers located in the 5′ upstream region and intron of the gene (Geyer and Corces 1987; Martin et al. 1989; Wittkopp et al. 2002). Mutations at yellow reduce pigmentation, sometimes to a fully mutant yellow color. Interestingly, some combinations of mutant yellow alleles that reduce pigmentation in the same tissues show intragenic complementation, such that flies heterozygous for two mutant alleles show nearly wild-type pigmentation (Geyer et al. 1990; Morris et al. 1999a). This intragenic complementation depends on the somatic pairing of yellow genes (Geyer et al. 1990; Chen et al. 2002; Savitsky et al. 2003; S. Ou, J. R. Morris and C.-t. Wu, unpublished results). One mechanism of transvection at yellow involves the action of the wing and body enhancers of one gene in trans on the promoter of its homolog on a separate chromosome (Geyer et al. 1990; Morris et al. 1998; Chen et al. 2002). As these enhancers usually prefer to act on their own promoter in cis, that is, show cis preference (Geyer et al. 1990; Morris et al. 1999a,b), transvection at yellow provides a system where enhancer choice is changed from a state of cis preference to one that also includes trans action.
The goal of our studies is to clarify how an enhancer chooses between a cis-linked promoter and one located in trans. It has been shown that enhancer choice can be modulated by the state of the cis-linked promoter (Geyer et al. 1990; Martínez-Laborda et al. 1992; Hendrickson and Sakonju 1995; Casares et al. 1997; Sipos et al. 1998; Morris et al. 1999a,b). Previously, we found that mutations introduced by targeted gene replacement (Gloor et al. 1991; Keeler et al. 1996) of the yellow TATA box or initiator (Inr), two core promoter elements (reviewed in Smale and Kadonaga 2003), released the upstream wing and body enhancers to act in trans (Morris et al. 1999b). This model can be illustrated by contrasting the abilities of two alleles, y1 and ytata-1, to contribute enhancer activity in trans to y82f29, which lacks both the wing and the body enhancers (Morris et al. 1998). As shown in Figure 1, the wing and body enhancers of y1, which carries an intact promoter, are restricted to cis action while those of ytata-1, in which the TATA box has been altered, can act in trans as well as in cis (Morris et al. 1999b).
These observations suggest three broad classes of models for enhancer choice in cis and in trans. One proposes that enhancer choice is sensitive to some aspect of competency involving the cis-linked promoter, such as transcription rate, the binding of specific transcription factors to the promoter, or a particular chromatin state. This model is in line with studies of enhancers and promoters in cis-linked configurations (Sharpe et al. 1998; Cai et al. 2001; Conte et al. 2002). Another model proposes that enhancer choice is influenced by the effect of promoter mutations on the state of pairing between two yellow alleles. That is, sequence heterology resulting from mutations in the cis-linked promoter or flanking regions may lead to promoter unpairing, and this unpairing may then render the intact promoter in trans to an enhancer more attractive, perhaps through conformational changes or by making it more accessible to transcription factors (Sipos et al. 1998; Morris et al. 1999a). A third model suggests that mutations in core promoter elements, by altering key promoter sequences, exert their effects on enhancer action by changing the identity or signature of the promoter. Such changes might cause an enhancer to seek a native promoter on a separate chromosome rather than interact with a cis-linked promoter that is of the wrong identity. Studies of cis-linked enhancers and promoters have revealed that enhancers can discriminate among different promoters on the basis of the presence or absence of core promoter motifs (Ohtsuki et al. 1998; Butler and Kadonaga 2001; Conte et al. 2002).
Here, we use targeted gene replacement at yellow to further examine these three models. In particular, we ask whether a mutation of a yellow core promoter element, the TATA box, releases enhancers to act in trans due to its effects on promoter competency, pairing, or identity. Surprisingly, we find that while a 6-bp mutation of a core promoter element releases enhancers, complete replacement of 193 bp of the yellow promoter with any of three heterologous promoters fails to release enhancers to trans action. We argue that the roles of promoter pairing and identity in the cis-trans decision are likely to be at most secondary to that of promoter competency. In addition, we consider our data in light of the possibility that competency of a promoter may reflect features of the promoter in addition to transcription strength.
MATERIALS AND METHODS
Scoring of pigmentation:
Pigmentation was scored as previously described (Morris et al. 1998). We scored 1- to 3-day-old females using a five-point scale, where 1 represents the null or nearly null state and 5 represents the wild-type or nearly wild-type state. For consistency, flies were scored by comparison to flies whose wing and body pigmentation was previously reported, including y1/y1 (1, 1), y2/y2 (1, 1–2), ytata-1/y82f29 (2, 1–2), ytata/ytata (3, 2), y1#8/y82f29 (3, 3), y1#8/y2 (4, 4), and y+/y+ (5, 5). However, because pigmentation represents a continuum, there is some variation of pigmentation for flies scored at the same value. Complementation between two alleles required that wing or body pigmentation was one point darker on the pigmentation scale than that of corresponding tissues in females homozygous for either allele. Flies were cultured at 25° ± 1° as previously described (Morris et al. 1998). Crosses were done with three to four females mated to three or more males in vials that were brooded daily. Both temperature and crowding were carefully controlled.
The templates used in the targeted gene replacement experiments were full-length 7.8-kbp yellow genes (GenBank accession nos. X06481 and X04427) with the designated sequence changes (Figure 2) cloned into a modified Bluescript plasmid (Stratagene, La Jolla, CA) in which the KpnI site is replaced by an XbaI site (pBSX). pUC8ySB is a pUC8 plasmid containing the wild-type 3.1-kbp SalI-BamHI yellow fragment; pUC8y1SB contains the A-to-C mutation of y1; pUC8ySBtata contains the 6-bp sequence change in the TATA box present in ytata; pUC8ySBtata-1 contains the 6-bp TATA mutation and the A-to-C mutation of y1; pBSXyBG is pBSX containing the 4.7-kbp BamHI-BglII yellow fragment (Morris et al. 1999b). All sequence changes were first incorporated into pUC8ySB or one of its derivatives noted above, which was then digested with SalI and BamHI and cloned into pBSXyBG to generate full-length templates in pBSX. All changes were confirmed by sequencing.
The four insertional promoter templates, tT, tT-1, TT, and TT-1, were made by introducing a double-stranded linker (see below) at the EagI site of pUC8ySB, pUC8y1SB, pUC8ySBtata, or pUC8ySBtata-1, respectively.
The eight promoter scan (ps) templates were made using a PCR strategy with one of the two primers carrying the desired sequence changes. For ps1 and ps1-1, primers 1 and m13rev (primer sequences shown below) were used to generate PCR products with pUC8ySB or pUC8y1SB as a template, respectively. The resulting PCR products were digested with EagI and BamHI and cloned into pUC8ySB. For ps2 and ps2-1, primers 2 and m13rev were used in a similar strategy. For ps3 and ps3-1, primers 3 and 4 were used with pUC8ySB as a template, and the resulting PCR products were digested with KpnI and EagI and cloned into pUC8ySB or pUC8y1SB, respectively. For ps4 and ps4-1, primers 3 and 5 were used in a similar strategy.
The six heterologous promoter templates were made by first cloning the KpnI-BamHI yellow fragment from pUC8ySBps4 into pBS to make pBSyKBps4. Primer pairs 6/7, 8/9, and 10/11 were then used to PCR amplify a 193-bp fragment from the even skipped (eve), heat-shock protein 70 (hsp70), and white (w) promoter regions of wild-type Canton-S genomic DNA, respectively. These were TA cloned (Invitrogen, San Diego), confirmed by sequence analysis, cut with ClaI and AflII, and cloned into pBSyKBps4. The resulting plasmid was digested with KpnI and BamHI and cloned into pUC8ySB.
Constructs for P-element-mediated transformation were made by digesting the three pBSX plasmids containing the heterologous promoters with HindIII and NotI, religating the resulting 2.9-kbp pBSX fragment and the 5.5-kbp 3′ yellow fragment, digesting the resulting plasmid with XbaI, and cloning this fragment into the XbaI site of pCaSper3.
Targeted gene replacement and P-element-mediated germ-line transformation:
Targeted gene replacement was carried out as previously described using yh12w+ as a target allele on a chromosome that also carried w1118 and plasmid sources of templates and transposase at concentrations of 1.0 and 0.25 mg/ml, respectively (Keeler et al. 1996; Morris et al. 1999b). On average, 500 embryos were injected per construct, with a larval survival rate of 40% and a conversion rate of 1% (of total embryos injected). Candidate lines were screened by single-fly PCR using primers 12 and 13, which give an 874-bp PCR product. Those lines giving wild-type-sized products were confirmed by digesting the PCR product with restriction enzymes diagnostic of the desired changes and sequence analysis. Southern analysis of genomic DNA digested separately with HindIII/BamHI and PstI and hybridized to a full-length yellow probe was performed to confirm the integrity of the yellow gene (data not shown). P-element-mediated transformation was done as previously described (Rubin and Spradling 1982; Morris et al. 1998) using 0.5 mg/ml construct, 0.1 mg/ml helper plasmid, and the Df(1)y− ac− w1118 stock as the host.
Flies were allowed to lay eggs for 24 hr in bottles at 25° and pupae were collected 8 days later. Total RNA was collected [Eppendorf (Madison, WI) Eukaryotic Perfect RNA] and poly(A)+ RNA was isolated [Promega (Madison, WI) PolyATtract] and electrophoresed (5 μg/lane) on a 1% agarose/formaldehyde gel followed by hybridization with a 32P-labeled EcoRI-BglII yellow fragment from the second exon and rp49 as a loading control. Quantitation was based on band intensity relative to the average of the band intensities for the four data points from two independent y+ lines generated by targeted gene replacement. The average level of steady-state mRNA in our wild-type Canton-S line was 91% (over three trials) of the level in our y+ control lines (data not shown). Preliminary experiments using Canton-S pupae collected daily from days 6–10 indicated that day 8 showed maximal steady-state yellow mRNA expression (data not shown). On the basis of these results, day 8 pupae were used in subsequent experiments. However, because time courses were not done for each of the alleles, it is possible that maximum expression is shifted earlier or later compared to that of Canton-S pupae.
Primers and linkers:
Primer and linker sequences used in this study are indicated below in 5′ to 3′ orientation with mutated nucleotides in boldface type.
Testing the role of transcription in the control of enhancer choice:
Previously, we demonstrated that a 6-bp mutation of the TATA box (ytata), the initiator (yinr), or the TATA box and initiator (ytata-inr) is sufficient to support transvection by releasing the wing and body enhancers to act in trans (Morris et al. 1999b; Table 1; Figure 2). Because these mutations were made in core promoter elements, the data suggested a model in which disruption of promoter competency, in particular its ability to support transcription, releases enhancers to act in trans. Here, we tested this model by determining the steady-state level of yellow expression through Northern analyses. We compared the level of yellow mRNA in flies carrying ytata, yinr, or ytata-inr to that in control flies carrying a wild-type y+ gene. Our y+ control flies were generated by the protocol used to generate ytata, yinr, and ytata-inr, that is, through targeted gene replacement (Gloor et al. 1991; Keeler et al. 1996) of the mutant yellow allele in our standard starting line, yh12w+ w1118 (see materials and methods; the w1118 null mutation of the white gene is present in all our convertant lines, but is not mentioned further). We found that steady-state levels of yellow mRNA are reduced in ytata, yinr, and ytata-inr to 1, 3, and 1%, respectively, of the level found in control y+ flies (Figure 3). These data are consistent with a model in which promoters that have been transcriptionally compromised release their enhancers to act in trans.
This interpretation further predicts that restoration of transcription to a transcriptionally compromised allele will recapture the upstream enhancers and prevent transvection. We tested this prediction by inserting a wild-type TATA box downstream of the mutated TATA box in ytata (Figure 2) and then testing the resulting allele for transcription and the ability of its enhancers to act in trans. Note that we chose to restore transcription to ytata instead of yinr because ytata is the more transcriptionally compromised of the two (Figure 3) and therefore provides a stronger test. Specifically, we used targeted gene replacement to insert the 27-bp region extending from position −44 to −18 of the wild-type yellow gene into the ytata allele at position −17/−18 (nucleotides numbered relative to the transcription start site of +1). The resulting allele carried a wild-type promoter 3′ of position −44. We called the new allele ytT, where t and T represent the mutated and wild-type TATA boxes, respectively, and their order represents their 5′ to 3′ order in vivo. Homozygous ytT/ytT and hemizygous ytT/Df (where Df represents the y− ac− w1118 chromosome that is deficient for all yellow sequences; Morris et al. 1998) flies showed fully wild-type pigmentation in all cuticular structures, including the wings, body, and bristles, similar to our y+ control flies (Table 1). However, by Northern analysis, steady-state yellow mRNA levels were only ∼11% of the level seen in y+ control flies (Figure 3).
Next we determined the ability of the wing and body enhancers of ytT to act in trans. To this end, we first generated a companion protein-null derivative of ytT such that translation of the transcripts made in the presence of the promoter alterations would not obscure our tests of the ability of the upstream enhancers to act on a promoter in trans. As was done for all subsequent companion alleles mentioned below, this protein-null derivative was generated by targeted gene replacement and bore, in addition to the changes in the promoter region, an A-to-C change in the ATG translation initiation codon. This change is identical to that found in the y1 protein-null allele (Geyer et al. 1990) and therefore, as expected, the companion allele for ytT, called ytT-1, gives a fully mutant phenotype (Table 1). We then determined whether ytT-1 could release its wing and body enhancers to act in trans by asking whether it complements y82f29 (Figure 1) and three other alleles, y2, y62a, and y2374, which, like y82f29, lack strong wing and/or body enhancer activity (Morris et al. 1999b). We found that ytT-1 does not complement each of these tester alleles. This finding suggests that restoration of transcriptional competency, as revealed by increased pigmentation and transcript levels in ytT compared to those of ytata, is sufficient to recapture the upstream enhancers and prevent transvection. Interestingly, yellow transcript levels in ytT flies reach only to the 11% level compared to the y+ control (Figure 3), indicating that transcription need not be restored to wild-type levels for the upstream enhancers to be restricted to cis action.
These data also argue against a model in which local promoter unpairing as a result of structural heterozygosity between two alleles in the promoter region is sufficient to release enhancers to trans action. Specifically, the 27-bp insertion present in ytT would be predicted to unpair the promoter region when paired with a tester allele, yet it does not support transvection. To test this interpretation, we made a second insertional yellow allele, called yTT, that differs from ytT only in that it carries a wild-type TATA box in place of the mutated TATA box (Figure 2). The yTT allele produced wild-type pigmentation in homozygous and hemizygous flies, and its protein-null companion allele, yTT-1, failed to complement each of the four tester alleles (Table 1). These findings provide additional support for the interpretation that local promoter unpairing as a result of sequence heterology between two alleles is not sufficient to release enhancers to act in trans.
Not all sequence changes in the promoter release enhancer to trans action:
These results indicate a strong correlation between promoter elements involved in transcription and those involved in enhancer choice. We next decided to test the strength of this correlation by introducing mutations in noncore promoter sequences. Will all elements that affect transcription also affect enhancer choice and vice versa, or are there elements dedicated to one process but not the other? Using targeted gene replacement, we made four ps mutations, yps1, yps2, yps3, and yps4 (Figure 2), each of which carried a 6-bp substitution in the promoter region, and four corresponding protein-null companion alleles, yps1-1, yps2-1, yps3-1, and yps4-1. Our design of these alleles was guided by the presence within the yellow promoter of an 80-bp region, extending from position −43 to +37, that shows 91% sequence identity with the homologous promoter region of the yellow gene of Drosophila subobscura, a species that diverged from D. melanogaster ∼30 million years ago (Munté et al. 1997). This sequence identity suggests that there might be important functional elements in this region, a prediction supported by sequence comparisons with other Drosophila species (Munté et al. 2000, 2001; Wittkopp et al. 2002). With this in mind, we placed two mutations (yps1 and yps2) within and two (yps3 and yps4) outside this region of conservation.
In spite of the different placements relative to the conserved region, however, all four mutations have similar pigmentation and transcriptional phenotypes. Homozygous or hemizygous yps flies show wild-type pigmentation (Table 1) and steady-state yellow mRNA levels in these mutants are between 16% (yps1) and 35% (yps3) of levels seen in our y+ control flies (Figure 3). The finding that mutations in noncore promoter elements nevertheless have a strong negative effect on steady-state mRNA levels is consistent with mutational analyses of noncore elements in other systems (for example, Dudley et al. 1999; Wray et al. 2003).
None of the companion alleles supported transvection when placed in trans to the four tester alleles (Table 1). These data suggest that not any 6-bp mutation in the promoter region allows transvection, even though all that we tested reduce transcript levels. In light of our observation that mutations of the yellow TATA box and Inr do release enhancers (Morris et al. 1999b), these data demonstrate the importance of core promoter elements in enhancer choice. Furthermore, as mutations in core elements reduce transcript levels more significantly than mutations in noncore sequences (Figure 3), one interpretation is that only mutations that most severely compromise transcription release enhancers to trans action.
The data also suggest a second way to look at the role of promoter competency in enhancer choice. Specifically, it may be that the primary feature of the promoter that guides enhancer choice is its integrity as defined, for example, by the array of core elements rather than by an absolute rate of transcription. According to this interpretation, the ps alleles maintain cis preference of enhancers because their core promoter elements are intact. Furthermore, this model is consistent with the ability of ytT and yTT to recapture enhancers; in spite of their low steady-state transcript levels, they nevertheless have an intact configuration of core promoter elements.
Enhancer action in cis can be maintained by heterologous promoters:
The data thus far support a key role for promoter competency in the cis-trans choice of an enhancer. Here we consider the third model in which promoter identity guides enhancer choice. In particular, we asked whether the 6-bp alteration in the TATA box changes the yellow promoter from a TATA-containing to a TATA-less promoter, thereby making it a poor target for the enhancers. Specifically, will the upstream enhancers of a yellow allele with a functional but heterologous promoter in place of the native promoter choose the heterologous promoter in cis rather than the native promoter in trans?
To answer this question, we used targeted gene replacement to substitute 193 bp of the yellow promoter from position −63 to +130, including the TATA box and Inr but not the translation initiation codon, with 193 bp of sequence from the promoter regions of the Drosophila eve, hsp70, and w genes (Figure 2). The resulting alleles were called yeve, yhsp70, and yw, and their protein-null companion alleles were called yeve-1, yhsp70-1, and yw-1. The eve promoter most closely resembles that of yellow in that both have a TATA box and Inr (MacDonald et al. 1986). The hsp70 promoter, like that of yellow and eve, has a TATA box and Inr but, in addition, contains multiple heat-shock elements (Perisic et al. 1989), one of which is included in the 193-bp sequence inserted into yellow. The white promoter is TATA-less, but does have an Inr and downstream promoter element sequences (Kadonaga 2002).
All three heterologous promoters gave dark pigmentation in homozygous or hemizygous flies (Table 1), indicating that they can direct yellow transcription. Flies bearing the yeve allele were indistinguishable from wild-type flies, while flies bearing the yhsp70 allele were at least as dark, if not darker, than wild-type flies. In addition, yhsp70 flies had a dusky appearance with dark pigmentation even in abdominal interbands, which are the normally lightly pigmented regions between the dark abdominal bands. The yw flies were less dark than wild-type flies. Consistent with these observations, steady-state levels of yellow transcripts reached 80 and 71% of wild-type levels in yeve and yhsp70 flies, respectively, but only 12% in yw flies (Figure 3).
To verify that transcription from these heterologous promoters is enhancer driven and not constitutive, we made constructs that carried promoter replacements identical to those present in yeve, yhsp70, and yw, but that were deleted for the wing and body enhancers, and determined the pigmentation levels that they direct when integrated into the genome by P-element-mediated germ-line transformation. If transcription of yeve, yhsp70, and yw is enhancer dependent, then flies bearing these enhancerless transgenes should show mutant pigmentation. If expression of yeve, yhsp70, and yw is constitutive, then the transgenic flies should instead show dark pigmentation. Flies carrying the eve and white enhancerless transgenes showed fully mutant wing and body pigmentation in all five and three, respectively, independent transgenic lines, indicating that the wing and body pigmentation seen in yeve and yw flies is in fact the result of enhancer-dependent transcription. The eight independent lines bearing the hsp70 enhancerless transgene differed from each other, showing a range of wing and body pigmentation from fully mutant to nearly wild type. This observation may reflect position effects or might suggest that the dark pigmentation seen in yhsp70 flies is partially constitutive, but that to achieve consistently full pigmentation, input from an enhancer is required. Significantly, all of the transgenic lines carried an intact bristle enhancer and showed wild-type bristle pigmentation, indicating that the transgenes are capable of yellow expression in their ectopic locations.
We then tested the ability of yeve, yhsp70, and yw to release the wing and body enhancers to act in trans by placing each of their corresponding companion alleles in trans to the four tester alleles and assaying complementation. None of the companion alleles complemented the tester alleles (Table 1), suggesting that yellow enhancers prefer to act in cis, even on a foreign promoter, rather than act in trans on a wild-type yellow promoter. As these heterologous promoters maintain cis preference of the yellow enhancers in spite of their different sequences and identities relative to the native yellow promoter, our data indicate that promoter pairing and identity are unlikely to play primary roles in enhancer choice at yellow.
At the outset of our studies, we considered three mechanisms that may govern how an enhancer chooses between a cis-linked promoter and one located in trans: by some aspect of promoter competency, by pairing-mediated changes in gene topology, and by promoter identity. Although our data do not rule out any model, they argue that the latter two mechanisms do not, on their own and in their simplest form, dictate cis-trans decisions of the wing and body enhancers of yellow. In contrast, our data draw attention to the transcriptional process and the integrity of the promoter.
Significantly, consideration of the alleles from the point of view of transcription shows that those supporting transvection reduce transcript levels below ∼3% compared to levels in our y+ control flies, while those that do not support transvection maintain transcript levels above ∼11% (Figure 3). The difference in transcript levels between the two groups of alleles is significant (Mann-Whitney U-test, P = 0.005) and suggests that the trancriptional competency of a promoter may need to be compromised below a certain threshold, as assayed by steady-state transcript levels, before the cis-linked enhancers can be released to act in trans.
What this hypothetical transcription-based threshold represents in biological terms is not clear. It may reflect a particular rate of transcription, the binding of a specific transcription factor, the adoption of a particular chromatin state, or another aspect of promoters that can affect the attractiveness of a promoter to an enhancer (Li and Noll 1994; Hansen and Tjian 1995; Kapoun and Kaufman 1995; Merli et al. 1996; Ohtsuki et al. 1998; Sharpe et al. 1998; Butler and Kadonaga 2001; Cai et al. 2001; Conte et al. 2002). Importantly, as transvection requires many events before transcription, such as homology sensing and pairing, the cis-trans choice of a yellow enhancer may occur much earlier than transcription, possibly making a threshold defined by transcript levels a poor reporter of an earlier key step. Choice may also be a dynamic process, varying from cell to cell or tissue to tissue (Golic and Golic 1996; Gubb et al. 1997; reviewed in Wu and Morris 1999; Kennison and Southworth 2002). Indeed, we have not yet determined whether yellow enhancers can alternate from moment to moment between the cis and trans promoters, perhaps interacting simultaneously with both. Finally, as has been observed at the bithorax complex (Goldsborough and Kornberg 1996; Casares et al. 1997; Sipos et al. 1998), it is formally possible that yellow enhancers are released to trans action at a threshold higher than we have observed, even when the cis promoter is entirely wild type and fully functional. If true for any of the noncomplementing genotypes we have tested, such interactions are nonproductive, ineffective, or too infrequent to affect pigmentation.
Interestingly, all nine alleles that maintain cis preference of enhancers also have an intact configuration of known core promoter elements. The insertional alleles have an intact configuration 3′ of position −44, the promoter scan alleles do not affect core elements, and the heterologous promoter alleles have foreign but otherwise intact promoter sequences. By contrast, the three alleles that do support transvection (ytata, yinr, and ytata-inr) all have mutations in core promoter elements (Morris et al. 1999b). This observation highlights the possibility that enhancer choice between two promoters may be influenced by the integrity of the promoter as determined by the configuration of core elements, the factors bound to them, and/or the chromatin state that they generate. That is, the nine alleles that maintain cis preference of their enhancers might do so because they each have an intact array of core promoter elements in spite of their differences in sequence, identity, and transcriptional profiles. This feature may be a molecular mark of a fully competent promoter in terms of transcription or, at the extreme, may reflect a promoter function independent of the transcriptional process.
A comparison of yinr with yw illustrates this point. The yinr allele has a 6-bp mutation in the initiator (Morris et al. 1999b), while yw has a white promoter substituted for the yellow promoter (Figure 2). Steady-state levels of yellow mRNA are 3% for yinr and 12% for yw as compared to that of our y+ controls and, as predicted by the threshold model, yinr supports transvection while yw does not. Interestingly, however, yw flies show lighter wing and body pigmentation as compared to yinr flies (Table 1) in spite of their higher transcript levels (Figure 3). It may be that white sequences in the 5′-untranslated region lead to the production of unstable transcripts, so that while transcript levels are relatively high, protein expression is not. If so, the transcription-based threshold would still apply. However, it may also be that yellow expression in tissues other than wing and body accounts for the discrepancy between phenotype and transcript levels of yw. If so, the transcription-based threshold would not apply in this case. Instead, it may be that the enhancers of yinr are released to act in trans because yinr carries a mutated core promoter element and therefore they perceive their promoter as inadequate, while the enhancers of yw are not released because they perceive the geography of their promoter as intact.
The ability of yw as well as yeve and yhsp70 to maintain cis preference also indicates that promoter pairing and identity are unlikely to be key determinants of cis-trans choice. Interestingly, at Ultrabithorax (Ubx), regulatory regions can act both in cis with a heterologous P-element promoter and in trans on their own promoter (Casares et al. 1997). This observation could reflect differences between yellow and Ubx in terms of the involvement of transposable elements, participating enhancers and promoters, or mechanisms of transvection.
In sum, our data suggest that the cis-trans choice of the yellow wing and body enhancers rests to a significant degree on the competency of the cis-linked promoter. Furthermore, our data call attention to two aspects of promoter competency with respect to the cis-trans choice: transcriptional competency and intactness of the promoter. A model of enhancer choice based on promoter function and integrity provides an explanation of why enhancers are able to detect small changes in key promoter elements but are apparently insensitive to complete replacement of the promoter.
The authors especially thank P. Geyer for years of close collaboration and insightful discussions, G. Gloor for invaluable advice on targeted gene replacement, J. Bateman, W. Bender, R. Emmons, C. Kaplan, J. Lokere, S. Ou, B. Williams, and F. Winston for discussions, and A. Moran for generous technical assistance. This work was supported by a National Institutes of Health grant (RO1 GM61936) to C.-t.W. and support from the William F. Milton Fund and the Harvard University Society of Fellows to J.R.M. and D.A.P.
↵ 2 Present address: Department of Biological Sciences, Stanford University, Stanford, CA 94305.
Communicating editor: S. Henikoff
- Received January 30, 2004.
- Accepted April 21, 2004.
- Genetics Society of America