hairy encodes a bHLH repressor that regulates several developmental processes in Drosophila, including embryonic segmentation and neurogenesis. Segmentation repressors such as Krüppel and knirps have been shown to function over short distances, less than 50–100 bp, to inhibit or quench closely linked upstream activators. This mode of repression permits multiple enhancers to work independently of one another within a modular promoter. Here, we employ a transgenic embryo assay to present evidence that hairy acts as a dominant repressor, which can function over long distances to block multiple enhancers. hairy is shown to repress a heterologous enhancer, the rhomboid NEE, when bound 1 kb from the nearest upstream activator. Moreover, the binding of hairy to a modified NEE leads to the repression of both the NEE and a distantly linked mesoderm‐specific enhancer within a synthetic modular promoter. Additional evidence that hairy is distinct from previously characterized embryonic repressors stems from the analysis of the gypsy insulator DNA. This insulator selectively blocks the hairy repressor, but not the linked activators, within a modified NEE. We compare hairy with previously characterized repressors and discuss the consequences of short‐range and long‐range repression in development.
hairy (h) regulates several developmental processes in Drosophila. It is expressed in a periodic pattern in the early embryo, and helps define the seven‐stripe pattern of fushi tarazu (ftz) expression (Nüsslein‐Volhard and Weischaus, 1980; Ish‐Horowicz et al., 1985; Carroll and Scott, 1986; Howard and Ingham, 1986; Ish‐Horowicz and Pinchin, 1987). Later, h restricts sensory bristle formation by repressing the proneural gene achaete (ac) (Falk, 1963; Botas et al., 1982; Moscoso del Prado and García‐Bellido, 1984; Orenic et al., 1993; Ohsako et al., 1994; Van Doren et al., 1994). h is also expressed in the developing eye, where it functions as an inhibitor of morphogenetic furrow progression (Carroll and Whyte, 1989; Brown et al., 1995).
The h protein belongs to the hairy‐related class of basic helix‐loop‐helix (bHLH) transcription factors, which includes deadpan and members of the Enhancer‐of‐split complex [E(spl)‐C] (Rushlow et al., 1989; Akazawa et al., 1992; Bier et al., 1992; Sasai et al., 1992; Feder et al., 1993; Ishibashi et al., 1993). Many of these proteins have been shown to act as transcriptional repressors (Akazawa et al., 1992; Sasai et al., 1992; Ishibashi et al., 1993; Ohsako et al., 1994; Van Doren et al., 1994; Dawson et al., 1995; Fisher et al., 1996). The hairy‐related proteins share several regions of homology, including a basic DNA‐binding region containing a signature proline residue, a helix‐loop‐helix dimerization domain, a hydrophobic domain of unknown function and the C‐terminal tetrapeptide sequence, WRPW, which interacts with the groucho co‐repressor (Knust et al., 1992; Wainwright and Ish‐Horowicz, 1992; Paroush et al., 1994; Dawson et al., 1995; Fisher et al., 1996; Grbavec and Stifani, 1996). These repressors bind DNA sequences (‘class C sites’) that are distinct from the E‐box motifs recognized by most bHLH transcription factors (Ohsako et al., 1994; Van Doren et al., 1994).
Transcriptional repression is essential for establishing localized patterns of gene expression during embryogenesis (Small et al., 1992; Studer et al., 1994; Kirchhamer and Davidson, 1996). In Drosophila, most of the spatially localized regulatory proteins present in the early embryo function as repressors. Four modes of transcriptional repression have been proposed (reviewed by Levine and Manley, 1989; Johnson, 1995). First, non‐DNA‐binding proteins can antagonize the function of transcriptional activators by preventing them from binding DNA. Members of the emc/Id class of HLH proteins, which lack a DNA‐binding domain, dimerize with bHLH activators to form inactive complexes (Benezra et al., 1990; Van Doren et al., 1991; Cabrera et al., 1994). Second, repressors can prevent activators from binding to DNA by occupying their binding sites (‘competition’). This type of repression is seen for the chicken ovalbumin regulatory factor, COUP‐TF, which inhibits the binding of retinoic acid and retinoid X receptors (Tran et al., 1992; Liu and Chiu, 1994). Homeodomain‐containing proteins, which as a group have relatively poor DNA‐binding sequence specificity, have been proposed to mediate repression by competing for ‘generic’ homeodomain recognition sequences that are also bound by homeodomain activators (e.g. Han et al., 1989).
A third proposed form of repression is ‘quenching’, whereby a repressor works over short distances, usually <100 bp, to inhibit closely linked activators. Repressors and activators are thought to co‐occupy nearby sites, but the repressor prevents the bound activator from interacting with the transcription complex. The Drosophila proteins snail (sna), Krüppel (Kr), giant (gt) and knirps (kni) were first shown to bind DNA elements that overlap activator sites in native promoters, prompting the suggestion that they repress transcription via competition (Small et al., 1991; Hoch et al., 1992; Ip et al., 1992). However, in more recent studies the repressor sites have been uncoupled from activator sites, and repression is observed even when they bind 50–100 bp from upstream activators (Gray et al., 1994; Arnosti et al., 1996a,b; Gray and Levine, 1996a). Direct protein–protein interactions between repressor and linked activator have not been demonstrated. An alternative model invokes transient inhibitory interactions between the repressor and one or more components of the transcription complex (see Gray and Levine, 1996b). Regardless of mechanism, this form of repression is ‘local’, since the repressors function only within the vicinity of their binding sites.
A fourth model for repression, silencing, differs from competition and local repression with respect to range of action. The Drosophila gradient morphogen, dorsal (dl), can function as a long‐range silencer. dl is inherently an activator, but can repress heterologous enhancers and promoters over distances of several kilobases when bound near appropriate ‘co‐repressors’ (Doyle et al., 1989; Lehming et al., 1994; Huang et al., 1995; Cai et al., 1996). Silencers may interact directly with the transcription complex or recruit heterochromatin to the promoter region, thus blocking access of basal transcription factors (see Herschbach and Johnson, 1993a).
In order to determine how h functions as a repressor, we analyzed a variety of fusion genes containing synthetic h binding sites in transgenic embryos. These studies suggest that h is a silencer, which can repress upstream activators over distances of at least 1 kb. h mediates dominant repression and can silence multiple enhancers in a modular promoter. These results suggest that h may repress transcription through a mechanism that is distinct from the local mode of repression employed by most other repressors present in the early Drosophila embryo. Further support for this view stems from the analysis of fusion promoters containing the gypsy insulator DNA. The insulator selectively blocks h, but not closely linked activators, suggesting that h might directly interact with one or more components of the basal transcription complex. We discuss the implications of dominant repression in development.
Synthetic h binding sites were inserted in the rhomboid neuroectodermal enhancer (rho NEE). This enhancer is 700 bp in length and directs reporter gene expression in lateral stripes within the presumptive neuroectoderm of the early embryo (Ip et al., 1992). The NEE is activated by dorsal (dl) and bHLH proteins in ventral and lateral regions, but is repressed by sna in the ventral mesoderm (Ip et al., 1992). Many of the experiments involved the use of a modified rho NEE, whereby the sna repressor sites were eliminated, resulting in expression in both ventral and lateral regions (e.g. Figure 1B; Ip et al., 1992). Transgenic embryos were hybridized with either a lacZ or white digoxigenin‐labeled antisense RNA probe to visualize reporter gene expression (see Materials and methods).
h mediates transcriptional repression
The rho NEE contains four high‐affinity dl binding sites that are clustered within a central 300 bp region of the enhancer. There are also five bHLH activator sites (E boxes) that are interspersed among the dl sites. Only the four dl sites are depicted in the diagrams accompanying the figures (Figure 1), but both dl and bHLH binding sites are essential for robust expression (Ip et al., 1992). We inserted two high‐affinity h binding sites in the rho NEE; these are located 50 bp from the central cluster of dl binding sites (see diagrams in Figure 1C and E). This modified NEE directs a segmental pattern of expression (Figure 1C). Sites of interstripe repression appear to coincide with regions of h expression (data not shown).
Interstripe repression persists when the h binding sites are moved 150 bp from the nearest dl sites (Figure 1D). The ability of h to repress transcription over this distance distinguishes it from sna, Kr and kni, which must map within 50–100 bp of the dl activators (Gray et al., 1994; Arnosti et al., 1996b; Gray and Levine, 1996a). We also assayed expression of the divergently transcribed white reporter gene. This was done to investigate the possibility that the downstream h site (see Figure 1D diagram) might block basal transcription factors within the lacZ promoter. The white transcription start site is over 300 bp from the nearest h site, beyond the range of ‘basal quenching’ (see Gray and Levine, 1996a). The white expression pattern is similar to the lacZ pattern, suggesting that h can repress NEE activators over a distance of at least 150 bp (Figure 1E and F).
Modified NEEs were expressed in various h mutants; an example is shown in Figure 2. This embryo is homozygous for the hm8 mutation, which contains a deletion in the h promoter region that eliminates all of the stripe‐specific enhancers, except stripes 1 and 5 (Howard et al., 1988). The modified rho NEE is repressed in just two domains, corresponding to h stripes 1 and 5 (Figure 2B). No repression is observed in embryos homozygous for hIL79K, a point mutation which introduces a stop codon after the bHLH motif of h (data not shown).
h is a dominant repressor
We tested the ability of h to repress transcription in an ‘enhancer‐autonomous’ fashion, whereby a repressor selectively inhibits only the enhancer to which it is bound (reviewed by Gray and Levine, 1996b). h binding sites were inserted in a modular promoter containing the rho NEE, as well as two tandem copies of the proximal enhancer (2xPE) from the twist (twi) promoter region, which mediates expression in the presumptive mesoderm (Jiang et al., 1991; Pan et al., 1991). The NEE used here contains the native sna repressor sites, which exclude expression from the ventral mesoderm and restricts the pattern to lateral stripes in the neuroectoderm (see Figure 3A). The NEE–2xPE fusion promoter directs an additive pattern of expression that includes lateral stripes (mediated by the NEE) and a band of staining in the presumptive mesoderm (mediated by 2xPE).
The NEE–2xPE fusion promoter directs a very different pattern of expression when two h sites are placed within the NEE (Figure 3B). The modified NEE mediates lateral stripes that are repressed in a pair‐rule pattern. Interstripe repression is also observed for the 2xPE enhancer, even though the closest h repressor site maps 290 bp from the distal‐most dl activator site within the PE (see diagram in Figure 3B). h repressor sites within the modified NEE continue to repress both the rho lateral stripes and the 2xPE pattern when spacer sequences separate the two enhancers by either 630 bp (Figure 3C) or 1370 bp (Figure 3D). In the latter configuration, the nearest h repressor site maps ∼2 kb away from the lacZ transcription start site. This long‐range action contrasts with the local repression mediated by the four native sna sites contained within the NEE. In this case, rho expression is excluded from the presumptive mesoderm, but the neighboring 2xPE is unaffected. These experiments suggest that h is a dominant repressor, while sna functions in a local fashion (see Gray and Levine, 1996a).
h is a long‐range repressor
The preceding experiments suggest that h can repress two enhancers even when bound only within the NEE. The next series of experiments addresses the possibility that this dominant repression depends on close linkage of the h sites with NEE activators. A single h repressor site was placed within a defective NEE lacking sna repressor sites (Figure 4).
A single h site placed 50 bp upstream of the nearest dl activator provides significant repression of the rho NEE (Figure 4B; compare with 4A). Repression is still seen when this site is placed 150 bp upstream of dl (Figure 4C). However, the single h site has little effect on the activity of the enhancer when placed 250 bp upstream of dl (Figure 4D). These findings raise the possibility that h must bind near upstream activators in order to mediate efficient repression. However, the preceding experiments represent a rather stringent test of the repressor since only a single h binding site was used. Additional experiments were done with multiple h sites (Figure 5).
Predictably, a single h site has no effect on NEE activity at a distance of 1 kb upstream of the nearest dl activator (Figure 5A). However, efficient repression is observed when two tandem h sites are used in this experiment (Figure 5B). This result provides additional evidence that h is distinct from previously characterized local repressors. For example, four clustered sna binding sites are unable to repress the even‐skipped (eve) stripe 2 or stripe 3 enhancers over a distance of just 150 bp (Gray and Levine, 1996a).
The h repressor is selectively blocked by an insulator DNA
The preceding results suggest that h functions as a long‐range, dominant repressor. Previous studies have shown that the gypsy insulator DNA can block a variety of enhancers, but fails to inhibit the dl–corepressor complex within the zerknüllt (zen) silencer element (VRE; Cai and Levine, 1995). Additional experiments were done to determine whether the gypsy insulator can block h.
The 340 bp gypsy insulator DNA contains 12 closely linked binding sites for the zinc finger protein, suppressor of Hairy wing [su(Hw); Spana et al., 1988]. The insulator selectively blocks distal, not proximal, enhancers in transgenic embryos. A variety of enhancers have been tested, including the eve stripe 2 and stripe 3 enhancers, the hairy H1 enhancer and the rho NEE (Cai and Levine, 1995, 1997). Among these enhancers, the NEE is relatively refractory to the gypsy insulator, as shown in Figure 6.
The fusion promoter used for these experiments contains a modified NEE that lacks sna repressor sites but contains two h sites. A defective eve stripe 2 enhancer was also included (see diagrams in Figure 6A and B), but it mediates sporadic expression that is not relevant to the analysis of NEE–insulator interactions. When a spacer sequence is placed between the modified NEE and lacZ promoter, staining is detected in ventral and lateral regions. The pattern is subdivided into pair‐rule repeats due to repression by h (Figure 6A), as seen previously (e.g. Figure 1C). The leftward white gene exhibits a similar, segmental staining pattern (Figure 6B). A distinct lacZ pattern is observed when the spacer sequence is replaced with the gypsy insulator DNA (Figure 6C). The NEE activators are not blocked, but instead, continue to drive lacZ expression in ventral and lateral regions. However, the staining pattern is continuous along the anteroposterior axis, and does not include pair‐rule repeats of interstripe repression (compare with Figure 6A). This observation suggests that the h repressor is selectively blocked, while the NEE activators are unaffected. As a control, the leftward white reporter gene continues to exhibit h–mediated repression since the insulator is not interposed between the enhancer and white promoter (see diagram in Figure 6C and D). We note that there is only a transient failure of the insulator to block NEE activators (with respect to lacZ). The embryos shown in Figure 6 are undergoing cellularization. By the completion of this process the insulator blocks the NEE, so that staining in ventral regions is essentially lost (data not shown).
We have presented evidence that h can repress heterologous enhancers in the early Drosophila embryo. Repression is observed even when h binding sites map far (1 kb or more) from both upstream activators and the target promoter. Moreover, h binding sites contained within a modified rho NEE also repress a second, distantly linked mesoderm‐specific enhancer (twi 2xPE) within modular promoters. This long‐range, dominant repression is distinct from the short‐range, local repression observed for previously characterized embryonic repressors such as sna. The analysis of fusion promoters containing the gypsy insulator DNA suggests that h interacts with one or more components of the basal transcription complex. We discuss the developmental implications of long‐range, dominant repression.
h is a long‐range, dominant repressor
h can repress the rho NEE even when bound 1 kb upstream of the closest dl activator sites (see Figure 5). In contrast, previously characterized embryonic repressors such as sna, Kr and kni, must bind within 50–100 bp of activators in order to inhibit transcription (Gray et al., 1994; Arnosti et al., 1996b; Gray and Levine, 1996a). Several different mechanisms can account for this long‐range repression. Perhaps h blocks distantly linked upstream activators. This type of mechanism has been invoked for the repression mediated by E2F–Rb complexes in mammalian cells (Weintraub et al., 1995). E2F is inherently an activator, but mediates repression by recruiting Rb, which in turn, can function over long distances (>1 kb) to inhibit specific upstream activators bound within the proximal promoter. In this particular example, the long‐range repressor exhibits regulatory specificity, and blocks just a subset of activators.
An alternative possibility is that h interacts directly with one or more components of the basal transcription complex. Previous studies suggest that the short‐range Kr repressor can interact with the β subunit of TFIIE (Sauer et al., 1995). However, this interaction must be weak and transient since Kr functions in a short‐range, local fashion and permits enhancer autonomy within the modular eve promoter (see Gray and Levine, 1996b). Perhaps h functions in a similar manner, but binds TFIIE with a higher affinity, thereby resulting in a general silencing of the promoter. Repressor–TFIIE interactions might impede procession of the pol II transcription complex.
Repression by h is not entirely unaffected by proximity to upstream activators. A single h binding site, which is probably recognized by a h homodimer (Ohsako et al., 1994; Van Doren et al., 1994), fails to mediate efficient repression when bound either 250 bp or 1 kb upstream of NEE activators (Figures 4 and 5). However, the use of two, tandemly linked h sites greatly extends the range of h‐mediated repression. Repression is seen even at a distance of 1 kb upstream of the NEE activators (see Figure 5). One interpretation of these results is that the occupancy of h binding sites is limiting. Efficient occupancy might depend on an ‘open’ chromatin state, which may be facilitated by the binding of nearby dl and bHLH activators to the NEE. When h sites are far from upstream activators, occupancy might depend on cooperative DNA binding interactions among h homodimers to linked sites.
h does not function in a local fashion within the modified rho NEE. Instead, it works in a dominant manner and blocks both the NEE and a distantly linked mesoderm‐specific enhancer (the twi 2xPE). This repression is distinct from that mediated by short‐range repressors, such as sna. Indeed, the contrast between h and sna is highlighted in the experiments presented in Figure 3. The rho NEE used in these experiments contains four native sna repressor sites, which exclude expression from the ventral mesoderm and restrict the pattern to lateral stripes in the presumptive neuroectoderm (Ip et al., 1992). The sna repressor functions solely within the limits of the NEE and has no effect on the ventral expression mediated by the linked 2xPE enhancer. Thus, the NEE–2xPE pattern is strictly additive (Figure 3A) due to the local action of the sna repressor. In contrast, both h and a second long‐range repression element, the zen VRE, mediate dominant repression of the twi 2xPE (Jiang et al., 1992).
Targets of h‐mediated repression
As discussed above, it is possible that h interacts with either upstream activators or the basal transcription complex. The difference between the dominant repression mediated by h and the local repression exhibited by sna (and other ‘short‐range’ repressors) might correspond to the strength of the interactions between the repressors and target activators. Perhaps h makes stronger, more stable, contacts with these targets than does sna. A key issue regarding the mechanism of repression concerns the identities of the targets.
It is conceivable that h blocks upstream activators within the rho NEE and twi 2xPE. Both of these enhancers are thought to be activated, in part, by bHLH proteins, such as daughterless (da) and achaete‐scute (Jiang et al., 1991; Ip et al., 1992). It is conceivable that h bound to the modified rho NEE blocks bHLH activators located within both the NEE and 2xPE through specific protein–protein interactions (Dawson et al., 1995).
Dedicated interactions between h and bHLH activators are also consistent with the normal, endogenous rho and twi expression patterns seen during embryogenesis. We have treated h as a heterologous repressor, but in fact, both patterns are refined into a series of anteroposterior segmental repeats following cellularization (Jiang et al., 1991; Bier et al., 1992; Ip et al., 1992). It is conceivable that these refinements are mediated, in part, by the h repressor. Our analysis has been restricted to precellular embryos, prior to the time when the endogenous genes may be subject to h‐mediated repression. None the less, it is possible that both the rho NEE and twi 2xPE are ‘sensitized’ for repression by h.
Studies with the gypsy insulator (Figure 6) do not exclude this type of mechanism, but strongly suggest that h makes direct contact with one or more components of the transcription complex. The insulator selectively blocks h‐mediated repression of a modified rho NEE (Figure 6C), although the dl and bHLH activators are unaffected and continue to direct expression in ventral and lateral regions of early embryos. If h worked solely by blocking upstream bHLH activators, then the insulator should have no effect on interstripe repression. The simplest interpretation of this result is that h contacts the basal transcription complex independently of the dl and bHLH activators.
Mechanism of repression
h and hairy‐related bHLH repressors have been shown to interact with the co‐repressor protein groucho (gro) through the C‐terminal WRPW motif (Paroush et al., 1994; Fisher et al., 1996; Grbavec and Stifani, 1996). gro is not known to bind DNA, but fusions of gro with heterologous DNA binding domains have revealed that gro can act as a transcriptional repressor (Fisher et al., 1996). gro is required for proper neurogenesis, segmentation and sex determination, all of which involve hairy‐related bHLH repressors (Paroush et al., 1994). The gro protein and its mammalian homologs contain several repeats of a 40‐residue motif, termed the WD40 repeat, which is thought to mediate protein–protein interactions (Hartley et al., 1988; Stifani et al., 1992; for review, see van der Voorn and Ploegh, 1992). Tup1, a yeast co‐repressor protein that also contains WD40 repeats, is recruited to DNA by the α2 repressor in α‐type cells for the silencing of a‐specific genes (Keleher et al., 1992). Similarly, h and its relatives may recruit gro for silencing specific genes in the Drosophila embryo.
The yeast mating‐type repressors α2 and Tup1 have been reported to interact with histones. This observation raises the possibility that Tup1 mediates transcriptional silencing by influencing chromatin structure (Roth et al., 1992; Cooper et al., 1994; Edmondson et al., 1996). There is also evidence that Tup1 interacts with basal transcription factors (Herschbach and Johnson, 1993b). Perhaps h–gro and α2–Tup1 complexes mediate repression through similar mechanisms. Strong and stable interactions between these repressors and the basal transcription complex would be expected to cause dominant silencing of complex promoter regions.
Short‐range repression is a flexible form of gene regulation that permits enhancer autonomy within complex, modular promoters (see Gray et al., 1996b). In contrast, long‐range silencing represents a stringent form of gene control that appears to be employed by promoters which must be unequivocally on or off. An example is sex determination in Drosophila. The hairy‐related protein deadpan (dpn) represses the early promoter of the Sex‐lethal (Sxl) gene, thereby ensuring that Sxl is off in male embryos (Younger‐Shepherd et al., 1992; Barbash and Cline, 1995; Hoshijima et al., 1995).
Materials and methods
P‐element transformations and in situ hybridization
P‐elements were introduced into the Drosophila germline by injection of yw67 embryos as described by Small et al. (1992). In situ hybridizations were performed as described by Jiang et al. (1991), using digoxigenin‐UTP‐labeled antisense RNA probes to hairy, lacZ or white. At least three independent transgenic lines were generated and tested for each construct. To generate the embryos shown in Figure 2, transgenic flies were crossed into a hm8 background (Howard et al., 1988) and offspring carrying both the mutation and the transgene were mated with one another. Embryos were analyzed as described above.
Construction of transgenes
The 700 bp rho NEE (Ip et al., 1992; Gray et al., 1994) and the 520 bp twi 2xPE (Jiang and Levine, 1993) were inserted into the polylinker of the C4PLZ transformation vector (Wharton and Crews, 1993). Two versions of the rho NEE were used: the wild‐type enhancer (in Figure 3) and one with mutations in the four sna binding sites as described in Ip et al. (1992) (in Figures 1, 2, 4, 5 and 6). A 340 bp DraI fragment of the chloramphenicol acetyl transferase (CAT) gene coding sequence was used as a spacer in the constructs shown in Figures 3C and D, and 6A and B. A 750 bp fragment containing the coding region of the green fluorescent protein (GFP) was used as a spacer in the constructs shown in Figures 3D, and 5A and B. Neither the CAT nor GFP sequences were found to affect reporter gene transcription in embryos (data not shown). The su(Hw) element shown in Figure 6 is a 340‐bp fragment of the gypsy retrotransposon, which was isolated by PCR (Cai and Levine, 1995). The constructs shown in Figure 6 were made by inserting the 700 bp rho NEE, the CAT spacer and the su(Hw) element into a derivative of the CaSpeR‐AUG‐βgal transformation vector (Thummel et al., 1988) containing the eve basal promoter, starting at −42 bp and continuing through codon 22 of eve, fused to the lacZ gene (Small et al., 1992), and also containing a 480 bp eve stripe 2 enhancer, with deletions in three gt binding sites (described in Arnosti et al., 1996a).
Site‐directed mutagenesis of the rho NEE
The h binding site used in these experiments corresponds exactly to the optimal site determined by Van Doren et al. (1994) in their random binding site selection experiments: gcggCACGCGacat (capitals indicate strongly selected bases). Binding sites were added to the rho NEE by oligonucleotide‐directed mutagenesis using the Mutagene kit (Bio‐Rad, CA) as described in Small et al. (1992). h sites were placed 50 bp 5′ and 3′ of the d1 and d4 dl sites, respectively, in the constructs shown in Figures 1C and E, 3B–D and 6A–D. h sites were placed 150 bp 5′ and 3′ of the d1 and d4 dl sites, respectively, in the constructs shown in Figures 1D and F, and 2B. A single h site was placed 50, 150 and 250 bp upstream of the d1 dl site in the constructs shown in Figure 4B–D, respectively. The 750 bp GFP spacer was inserted into the construct shown in Figure 4D, between the h site and the rho NEE to create the construct shown in Figure 5A. A double‐stranded oligonucleotide containing a second h binding site was inserted into the previous construct, 5 bp downstream of the existing h site, to create the construct shown in Figure 5B.
We thank David Arnosti for valuable discussions, Haini Cai, Susan Gray, Keith Maggert, Paul Szymanski and Bob Zeller for providing DNAs, and Mark Van Doren and Jim Posakony for providing the hairy binding sequence prior to publication. We especially thank Jim Posakony for his generosity and continuing support. This work was supported by an NIH grant (GM 34431), and by a Markey predoctoral fellowship to S.B.
- Copyright © 1997 European Molecular Biology Organization