Experimental investigations of eukaryotic enhancers suggest that multiple binding sites and trans-acting regulatory factors are often required for wild-type enhancer function. Genetic analysis of the stripe 2 enhancer of even-skipped (eve), an important developmental gene in Drosophila, provides support for this view. Given the importance of even-skipped expression in early Drosophila development, it might be predicted that many structural features of the stripe 2 enhancer will be evolutionarily conserved, including the DNA sequences of protein binding sites and the spacing between them.

To test this hypothesis, we compared sequences of the stripe 2 enhancer between four species of Drosophila: D. melanogaster, D. yakuba, D. erecta and D. pseudoobscura. Our analysis revealed a large number of nucleotide substitutions in regulatory protein binding sites for bicoid, hunchback, Kruppel and giant, as well as a systematic change in the size of the enhancer. Some of the binding sites in D. melanogaster are either absent or modified in other species. One functionally important bicoid-binding site in D. melanogaster appears to be recently evolved.

We, therefore, investigated possible functional consequences of sequence differences among these stripe 2 enhancers by P-element-mediated transformation. This analysis revealed that the eve stripe 2 enhancer from each of the four species drove reporter gene expression at the identical time and location in D. melanogaster embryos. Double staining of native eve protein and transgene mRNA in early embryos showed that the reporter gene mimicked native eve expression and, in every case, produced sharply defined stripes at the blastoderm stage that were coincident with eve stripe 2 protein.

We argue that stripe 2 eve expression in Drosophila evolution can be viewed as being under constant stabilizing selection with respect to the location of the anterior and posterior borders of the stripe. We further hypothesize that the stripe 2 enhancer is functionally robust, so that its evolution may be governed by the fixation of both slightly deleterious and adaptive mutations in regulatory protein binding sites as well as in the spacing between binding sites. This view allows for a slow but continual turnover of functionally important changes in the stripe 2 enhancer.

Differential gene expression during development can now be described in terms of the interactions between cis-acting elements and trans-acting factors (Arnone and Davidson, 1997). Recent success in correlating the expression of key regulatory factors in development with major evolutionary features of arthropod and vertebrate body plans is predicated on the parallel assumption that evolutionary differences in gene expression will also be understood in terms of specific changes the same cis- and trans-acting factors. Since many of the trans-acting regulatory factors are highly conserved proteins, it has been suggested that regulatory evolution is brought about primarily by substitutions in cis regulatory sequences rather than in the proteins themselves (Averof et al., 1996).

A detailed experimental dissection of a regulatory sequence has been carried out for the stripe 2 enhancer governing the transcriptional regulation of a pair-rule gene, even-skipped (eve) in Drosophila (Stanojevic et al., 1991; Small et al., 1992; Arnosti et al., 1996). eve encodes a homeodomain-containing protein that is expressed during embryonic development in both arthropods and vertebrates (Kenyon, 1994; Patel et al., 1994). The spatial and temporal patterns of embryonic expression of eve are tightly regulated in D. melanogaster, but are different among diverse insect species, such as the fruitfly, grasshopper and beetle (Patel et al., 1994). In Schistocerca, for example, eve does not exhibit a pair-rule pattern of expression (Patel et al., 1992), whereas in Tribolium it does show pair-rule expression, but with the formation of segments proceeding sequentially rather than simultaneously, as in Drosophila (Patel et al., 1994). In Schistocerca, vertebrates and C. elegans, eve homologs are expressed in the posterior regions of embryos, suggesting a role in the specification of posterior identities rather than defining segmental boundaries throughout the embryo (Patel et al., 1992, 1994; Ruiz i Altaba and Melton, 1989; Bastian and Gruss, 1990; Ahringer, 1996). Indeed, mutations in C.elegans vab-7 (the eve ortholog) lead to the deletion of posterior structures (Ahringer, 1996).

eve plays a key role in the segmentation process of D. melanogaster (Nusslein-Volhard et al., 1985). Prior to the completion of cellularization, eve is expressed in series of seven transverse stripes in the blastoderm (Frasch and Levine, 1987). The transcriptional regulation of these stripes is complex. The eve cis-regulatory region contains a series of separate modular enhancers that control the expression of individual stripes (Goto et al., 1989; Harding et al., 1989; Small et al., 1992, 1996). Stripes 2 and 3 are controlled by nonoverlapping enhancers which are separated by approximately 1.5 kb of ‘spacer’ sequence (Small et al., 1993, 1996). The stripe 3 enhancer sequence also regulates the expression of stripe 7 (Small et al., 1996). Short-range repression permits these enhancers to direct transcription from a common promoter independently (Small et al., 1993; Gray et al., 1994).

Extensive in vitro mutagenesis of trans-acting factor-binding sites in a 480 bp Minimal Stripe 2 Element (MSE) has allowed the refinement of a mechanistic model for stripe 2 activation and repression (Small et al., 1992; Arnosti et al., 1996). The D. melanogaster MSE contains 12 transcription factor-binding sites, including six activator and six repressor sites. According to the model, binding of bicoid and hunchback in the MSE is required for the activation of eve transcription in the anterior half of the embryo. The stripe borders are determined by the activities of two repressors, giant in anterior region and Kruppel in posterior region (Arnosti et al., 1996). The fact that four of six bicoid and hunchback activator sites directly overlap giant or Kruppel repressor sites led to the early suggestion that these repressors might define the stripe borders through competition for shared binding sites (Stanojevic et al., 1991; Small et al., 1992). However, the overlap of giant-binding sites with bicoid and hunchback sites has recently been shown not to be essential for giant function in vivo (Arnosti et al., 1996). The authors interpreted this result as indicating that giant and Kruppel repress eve transcription at stripe boundaries by a mechanism involving short-range quenching.

In an attempt to understand the relationship between sequence evolution and enhancer structure-function, we previously compared the stripe 2 enhancer region in population samples of the sibling species, D. melanogaster and D. simulans (Ludwig and Kreitman, 1995), in another closely related species, D. erecta (Kreitman and Ludwig, 1996), and in a distantly related species, D. picticornis (Sackerson, 1995). Interspecific comparisons revealed that most, but not all, of the D. melanogaster homologs of the bicoid, hunchback, Kruppel and giant protein-binding sites were present in the other species. One binding site for bicoid, bcd-3, was not present in either D. erecta or D. picticornis, indicating that it was a recently evolved site in D. melanogaster (Kreitman and Ludwig, 1996). Surprisingly, this site has been experimentally shown to be functionally important in the D. melanogaster MSE (Small et al., 1992). Nucleotide substitutions were present in the majority of binding sites; only three binding sites were completely conserved in all of the species. The spacer regions separating adjacent binding sites also differed in sequence as well as in length. Similar mutational changes were found segregating as polymorphisms within species (Ludwig and Kreitman, 1995).

Here we investigate whether the evolutionary changes in the stripe 2 enhancers of four Drosophila species have any discernible effects on the timing or spatial localization of stripe 2 expression. A functional analysis of the eve stripe 2 enhancers from D. melanogaster, D. yakuba, D. erecta and D. pseudoobscura was carried out using a reporter gene in P-element-mediated transformants. We were able to spatially and temporally localize stripe 2-driven lacZ in early embryos by using a double staining technique for native eve protein and lacZ mRNA, and by including the melanogaster stripe 3+7 eve enhancer as an internal standard in our constructs. The goal of this work is to relate patterns of natural sequence variation in the eve stripe 2 enhancer to our experimentally derived understanding of its function.

Drosophila stocks

D. erecta (stock number 1013) was obtained from the Drosophila Species Stock Center in Bowling Green. D. yakuba, D. pseudoobscura (Est-51.00) and D. melanogaster Oregon RC were obtained from Michael Ashburner, R. Richmond and C. I. Wu, respectively. Transgenic lines of D. melanogaster containing a stripe 2,3+7-lacZ gene fusion construct was kindly provided by S. Small (line 51 and 55 transformed with construct 2; see Small et al., 1993).

Cloning, amplification and sequencing

Genomic DNA was prepared from single adult male flies as previously described (Ludwig and Kreitman, 1995). The eve regulatory region of D. erecta, containing part of the coding region, the proximal promoter and part of the stripe 2 regulatory element, was cloned as a PCR fragment of approximately 1600 bp length. Primers for the amplification of this fragment (5′catcttctgcgggcgtttgt3′ and 5′ctgccgttcaaggagttatc3′) were designed from conserved sequences in D. melanogaster and D. simulans. The 5′ portion of the D. erecta stripe 2 element was obtained by inverse PCR (Ochman et al., 1989). Two sets of universal primers for the amplification of the stripe 2 enhancer region were then designed from aligned sequence comparisons of D. melanogaster (Canton-S GenBank reference X78903), D. picticornis (Sackerson, 1995), and our sequences of eve from different alleles of D. melanogaster and D. simulans (Ludwig and Kreitman, 1995), and D. erecta. One set of primers (containing 5′ EcoRI sites) –

U1+: 5′aaaagaattcatttgctgcggtnagtcg3′ and

U2−: 5′aaaagaattctgrtgtctytccatrttrta3′, or

U3−: 5′aaaagaattcmtgccrttcarsgarttrtc3′ – was used to amplify a region that spanned the autoregulatory region to the coding region of eve. This fragment was then reamplified with nested primers (containing 5′ BamHI or EcoRI sites) –

U4+: 5′aaaaggatccgagatcggcgctttgtgag3′ and

U2−: 5′aaaagaattctgrtgrctytccatrttrta3′, or

U3−: 5′aaaagaattcmtgccrttcarsgarttrtc3′ – so that it extended from the spacer between the elements for stripes 3+7 and 2 to the eve coding region. All PCR fragments were cloned into the Stratagene vector, pBluescript II SK+. Sequencing templates were prepared from amplification of the cloned inserts with the universal M13 primers (−20) and Reverse. All sequences were determined for both strands using template-specific primers. To eliminate PCR and cloning errors, each sequence was confirmed by sequencing PCR templates obtained directly from the amplification of genomic DNA. Sequencing was carried out on an Applied Biosystems Model 373a automated sequencer using TAG DyeDeoxytm terminator cycle sequencing chemistry, as described in Ballard and Kreitman (1994).

Stripe 2 elements

The region containing the stripe 2 element for each species was also obtained by PCR. A 2.5 kb fragment containing the stripe 2 element was amplified from DNA prepared from a single fly using primers U4+ and U3−. Nested primers based on conserved sequences, Kr6+: 5′aaaaggtaccaatataacccaataattt3′ and U5−: 5′aaaagaattcaaacatttattatgatgatataatca3′, were then used to obtain the stripe 2 element from the previously amplified fragment. The stripe 2 enhancer sequences from the four species used in the transformation experiments, therefore, are identical in that they begin and end at completely conserved sequences flanking the enhancer, and they contain all the DNA between these conserved sites.

The primers, Kr6+ and U5− contained the restriction sites for Asp718 and EcoRI, respectively, at their 5′ ends. Following digestion with these enzymes, the PCR fragments were cloned into Asp718 and EcoRI sites of the plasmid. Inserts with correct sequences were identified for further use by sequencing independent clones.

P-element transformation vector

The organization of the transformation vector, CaSpeR eve 3, lacZ, is shown in Fig. 1. This construct contains a minimal eve stripe 3+7 enhancer (∼500 bp), a 3′ spacer (∼300 bp), and the eve proximal promoter linked to lacZ. The stripe 3+7 enhancer and 3′ spacer of D. melanogaster was obtained from the plasmid pE5.2 lacZ (Small et al., 1993) as an 800 bp BamHI-AflII fragment. pE5.2 lacZ contains a 5.2 kb PstI fragment from the eve promoter inserted into the PstI site at −42 of pELI (Lawrence et al., 1987). pELI lacZ contains the basal eve promoter (from −42), the intact 100 bp untranslated leader and the coding sequences for the first 22 amino acids of the eve protein fused to amino acid number 5 of the lacZ coding sequence. A fragment containing the minimal stripe 3+7 enhancer, the spacer and the eve proximal promoter-lacZ fusion was cloned into the P-element transformation vector CaSpeR (Thummel et al., 1988) using unique BamHI and XbaI sites. A unique restriction site, Asp718 (KpnI), was inserted near the PstI site located downstream of the spacer region, so that stripe 2 enhancer elements from different species could be cloned into this vector in the proper orientation. To accommodate this cloning strategy, unique restriction sites in the polylinker, PstI and EcoRI, were eliminated. The stripe 2 and stripe 3+7 enhancers are separated in the final construct by the native 300 bp spacer to ensure autonomous regulation of the lacZ reporter (Small et al., 1993).

Fig. 1.

Physical organization of the stripe 3+7 and stripe 2 enhancers in relation to the eve proximal promoter and lacZ reporter gene in the P-element vector, CaSpeR.

Fig. 1.

Physical organization of the stripe 3+7 and stripe 2 enhancers in relation to the eve proximal promoter and lacZ reporter gene in the P-element vector, CaSpeR.

P-transformation and whole-mount in situ hybridization

P-element-mediated germline transformation was carried out according to Robertson et al. (1988). For each construct, at least one insertion in each of the three major chromosomes of D. melanogaster was generated to control for the influence of position effect on lacZ expression (Ludwig et al., 1993). Between 6 and 10 independent stable transformed lines were generated for each construct and at least three independent lines were examined for lacZ expression. Embryos were doubly stained to allow simultaneous detection of eve protein and lacZ mRNA by in situ hybridization. eve protein was detected using an anti-eve monoclonal antibody (mAb 2B8; Patel et al., 1994), HRP-conjugated secondary antibody and DAB histochemistry, according to a rapid staining protocol with an addition of RNAse inhibitor (Patel, 1994, 1996). lacZ mRNA was detected by in situ hybridization using a DIG-labeled antisense lacZ RNA probe. The preparation of the probes and the whole-mount in situ hybridization was carried out according to a protocol provided by S. Small (Jiang et al., 1991a).

Analysis of enhancer stripe 2 expression using a reporter gene

We evaluated the position of lacZ expression relative to native eve protein expression at a fixed timepoint in embryonic development. The expression of lacZ mRNA driven by the stripe 3+7 enhancer provided an internal control for the timing of development. The position of the experimental stripe 2-lacZ mRNA was evaluated when the lacZ mRNA stripes 3+7 precisely coincided with the corresponding stripes of the native eve protein. The experimental justification for this procedure is given in the Results section and in Fig. 2. As an additional way to assess the stage of development, we also analyzed the extent of cellularization in each embryo. In all cases, our comparisons of stripe 2 expression occurred at mid-cellularization.

Fig. 2.

eve protein (brown) and lacZ mRNA (purple) simultaneously detected in a transgenic line of D. melanogaster (provided by S. Small) transformed with the D. melanogaster stripe 2 (MSE), 3+7-lacZ gene fusion. (A) Stage 5(2)-5(3) blastoderm; (B) high magnification view of stripes 2 and 3, from A. (C) Early germ band extension (stage 6); (D) high magnification view of stripe 2 and 3, from C. (A,B) Embryos at the mid-cellularization stage, when the native eve expression and lacZ expression coincide along stripes 2, 3 and 7. (C,D) The discordance of native eve stripes relative to persistent lacZ mRNA stripes (see text for details).

Fig. 2.

eve protein (brown) and lacZ mRNA (purple) simultaneously detected in a transgenic line of D. melanogaster (provided by S. Small) transformed with the D. melanogaster stripe 2 (MSE), 3+7-lacZ gene fusion. (A) Stage 5(2)-5(3) blastoderm; (B) high magnification view of stripes 2 and 3, from A. (C) Early germ band extension (stage 6); (D) high magnification view of stripe 2 and 3, from C. (A,B) Embryos at the mid-cellularization stage, when the native eve expression and lacZ expression coincide along stripes 2, 3 and 7. (C,D) The discordance of native eve stripes relative to persistent lacZ mRNA stripes (see text for details).

Alignment of DNA sequences

The eve stripe 2 regions from Drosophila species were initially aligned with ClustalV with gap penalty = 8 and length gap penalty = 2. Manual adjustment of this alignment was necessary to improve local alignments of some of the binding sites. The GenBank references for the sequences are AF042712 (D. pseudoobscura), AF042711 (D. erecta), AF042710 (D. yakuba) and AF042709 (D. melanogaster).

Characterization of eve stripe 2-binding site evolution

Inspection of the aligned sequences, shown in Fig. 3, allowed the identification of homologous sequences corresponding to many, but not all, of the D. melanogaster DNA-binding sites (Fig. 4). The spacers between the conserved binding sites vary in length and are not well conserved as a rule. Fig. 4 contains binding site sequences from D. melanogaster, which were identified by DNA footprinting (Stanojevic et al., 1991), and the homologous sequences from five other species. Included in this table are the sequences from the three species that are the subject of this study as well as the sequences from two other species, D. simulans (Ludwig and Kreitman, 1995) and D. picticornis (Sackerson, 1995).

Fig. 3.

Alignment of eve stripe 2 enhancer regions from four species of Drosophila. Gaps in aligned sequences are indicated by dashes. The binding sites in D. melanogaster for the trans-acting factors, bicoid (BC), hunchback (HB), Kruppel (KR) and giant (GT), are shown above the sequence. Blocks 1 and 2 are conserved sequences (see text for detail). mel, D. melanogaster; yak, D. yakuba; ere, D. erecta; pse, D. pseudoobscura.

Fig. 3.

Alignment of eve stripe 2 enhancer regions from four species of Drosophila. Gaps in aligned sequences are indicated by dashes. The binding sites in D. melanogaster for the trans-acting factors, bicoid (BC), hunchback (HB), Kruppel (KR) and giant (GT), are shown above the sequence. Blocks 1 and 2 are conserved sequences (see text for detail). mel, D. melanogaster; yak, D. yakuba; ere, D. erecta; pse, D. pseudoobscura.

Fig. 4.

Stripe 2 enhancer binding sites in D. melanogaster and homologous sequences from five other Drosophila species. N/A, no homologous sequence identified. mel, D. melanogaster; sim, D. simulans; yak, D. yakuba; ere, D. erecta; pse, D. pseudoobscura; pic, D. picticornis.

Fig. 4.

Stripe 2 enhancer binding sites in D. melanogaster and homologous sequences from five other Drosophila species. N/A, no homologous sequence identified. mel, D. melanogaster; sim, D. simulans; yak, D. yakuba; ere, D. erecta; pse, D. pseudoobscura; pic, D. picticornis.

Two binding sites do not have obvious functional homologs in one or more species (Fig. 4). A hb-1 site is entirely absent in D. erecta and no corresponding sequence can be found in D. pseudoobscura. Only D. melanogaster has a viable bcd-3 site, TATAATCGC, including the required central motif, TAAT. The homologous sequences of the two related species, D. yakuba and D. erecta, TGCACTCGG and TATGTATCGC, respectively, probably cannot be functional as bicoid-binding sites. No homologous sequence can be identified in D. pseudoobscura. The presence of the bcd-3-binding site only in D. melanogaster and D. simulans indicates that this is a relatively new site that evolved in the ancestral lineage leading to these two species.

Of the 17 known binding sites in D. melanogaster (see Figs 3, 4), only three (kr-6, kr-5 and bcd-5) are completely conserved among all six species. Many of the binding sites, however, differ by only one or two base changes, indicating that these sequences must be functionally constrained. Another indication of functional constraint is the fact that the vast majority of mutational changes in these binding sites occurred only once in the phylogeny of the species. For example, of 17 variable positions in the six Kruppel-binding sites (Fig. 4), only two of them have more than one mutational change. One of them, position 8 in Kr-1, has mutated (at least) twice to three nucleotides (G, C and A). The other doubly mutated site, position 4 in kr-3, is one in which non-sister species share a mutational change, G → C. The best available phylogeny of the D. melanogaster species subgroup (Jeffs et al., 1994), places D. yakuba as a sister-species of D. melanogaster and D. simulans, and D. erecta as the outgroup of these species. If this phylogeny is correct, then position 4 in kr-3 mutated once from G →C prior to the split of D. erecta from the other subgroup species, and mutated again back to G following the split of D. yakuba from its sister species. Thus, this site has a convergent mutation, G→ C→ G.

D. picticornis differs from D. pseudoobscura and the four D. melanogaster subgroup species at 11 nucleotide positions in Kruppel-binding sites. These sites are well conserved, however, in the five Sophophoran species: only one mutational difference can be found at these positions. Given the cumulative evolutionary time separating the five Sophophoran species, many of these sites must be functionally constrained in the Sophophoran species. It would appear, therefore, that essentially every individual nucleotide position in the six Kruppel sites is functionally constrained. This raises the interesting question, addressed in the discussion, as to whether the observed changes in these binding sites can be selectively neutral ones, given the strong indication that none of the sites at which these changes occur are free to evolve.

Length changes

The stripe 2 enhancer is bordered on the 3′ and 5′ sides by completely conserved blocks of 18 bp and 26 bp, respectively (marked as blocks A and B in Fig. 3). The D. pseudoobscura stripe 2 enhancer region, at 1027 bp, is 28% larger than the corresponding 798 bp region in D. melanogaster. The D. erecta and D. yakuba stripe 2 regions, 849 bp and 844 bp, respectively, are intermediate in length. We investigated the distribution of length changes across the stripe 2 enhancer by identifying all of the intervals separating conserved binding sites. We refer to these intervals as ‘spacers’. Since bcd-3 and hb-1 sites could not be identified in all of the species, we substituted conserved blocks located adjacent to these sites (identified as blocks 1 and 2 in Fig. 3). Spacer lengths in the four species differ in 11 of the 12 intervals (Table 1). The one invariant spacer interval is a conserved 9 bp sequence located between kr-5 and gt-3. Its invariant length, therefore, is likely to be due to specific constraints on that sequence. D. melanogaster and D. pseudoobscura have the smallest and largest spacer lengths, respectively, differing by an average of 37.7%. Interestingly, the least variable spacer, excluding the invariant kr-5 to gt-3 spacer, as measured by the coefficient of variation, is the one located in the middle of the stripe 2 element between kr-4 and bcd-2. Previously, we showed that the middle of the enhancer, from gt-3 to gt-1, is more variable in terms of length than two clusters of overlapping binding sites that flank this region (Kreitman and Ludwig, 1996). The present data indicate, however, that the length variability in the middle region is not uniformly distributed.

Table 1.

Distances between putative binding sites or conserved motifs

Distances between putative binding sites or conserved motifs
Distances between putative binding sites or conserved motifs

Although 11 of the 12 spacers vary in length among the species, their relative lengths do not change appreciably (Table 1). Thus, for example, the rank order correlation of spacer lengths between D. melanogaster and D. pseudoobscura is r=0.79, and is nearly identical to that between the much more closely related species, D. melanogaster and D. yakuba (r=0.81). This suggests that selection may limit the range of permissible lengths for each spacer.

Pattern of native stripe 2 and transgene expression during embryonic development

We investigated the dynamics of transgene expression relative to native eve expression by assembling a developmental series of stained embryos. A representative series is shown in Fig. 5 for the construct containing the D. erecta stripe 2 element. In each of the experimental constructs from the four species, the spatial localization of lacZ mRNA stripes changes during embryogenesis relative to native eve protein expression. As expected, a broad anterior band of lacZ transcript is initially present in early embryos, which then resolves into distinct stripes at around the time of cell wall formation in the syncytial blastoderm. At the first appearance of well-resolved native eve stripes, the eve protein and the lacZ transcript coincide on a cell-for-cell basis along the full length of the stripes. Subsequently, however, the native eve stripes are positioned slightly forward in the embryo relative to the lacZ stripes, and this is apparent for both stripe 2 and for stripes 3+7. We have also observed the same phenomenon in two transgenic lines containing a different construct of D. melanogaster eve stripe 2 and 3+7 enhancers, provided by S. Small (Fig. 2). After cellularization, native eve stripes undergo the process of refinement, in which they narrow from about four cells to about two cells by loss of expression in the posterior region (Warrior and Levine, 1990). The control of this process requires an upstream cis-acting element, which is located approximately 5 kb from the transcriptional start site. This element has been shown to respond to early stripe eve expression and to regulatory inputs from other pair-rule genes (Goto et al., 1989; Harding et al., 1989; Jiang et al., 1991b; Fujioka et al., 1995). The fact that our constructs and the construct provided by S. Small do not contain the autoregulatory element, probably explains the misalignment of the endogenous eve protein stripes and the persistent of lacZ-mRNA in the transgenic embryos at stages following the completion of cellularization.

Fig. 5.

Developmental series of D. melanogaster embryos transformed with the D. erecta stripe 2 enhancer and D. melanogaster stripe 3+7 enhancer-lacZ gene fusion. (A-H) eve protein (brown) and lacZ mRNA (purple) simultaneously detected in the embryos at precellularization stage to beginning of gastrulation stage. (A-C) Sagittal focus; (D-H) superficial focus. (A-C) The process of the activation of D. erecta stripe 2 lacZ and D. melanogaster stripe 3+7 lacZ expression. (D) The beginning of eve protein stripe maturation; the anterior border of the eve stripe 1 is defined. (D-F) Early eve protein and lacZ mRNA stripe formation. (F) Mid-cellularization stage, when native eve expression and lacZ expression coincide along stripes 2, 3 and 7. A decrease in stripe 2 lacZ expression relative to stripe 3 lacZ expression can be seen at this stage. (G,H) The discordance of native eve stripes relative to persisting lacZ mRNA stripes is apparent in panels (see text for details).

Fig. 5.

Developmental series of D. melanogaster embryos transformed with the D. erecta stripe 2 enhancer and D. melanogaster stripe 3+7 enhancer-lacZ gene fusion. (A-H) eve protein (brown) and lacZ mRNA (purple) simultaneously detected in the embryos at precellularization stage to beginning of gastrulation stage. (A-C) Sagittal focus; (D-H) superficial focus. (A-C) The process of the activation of D. erecta stripe 2 lacZ and D. melanogaster stripe 3+7 lacZ expression. (D) The beginning of eve protein stripe maturation; the anterior border of the eve stripe 1 is defined. (D-F) Early eve protein and lacZ mRNA stripe formation. (F) Mid-cellularization stage, when native eve expression and lacZ expression coincide along stripes 2, 3 and 7. A decrease in stripe 2 lacZ expression relative to stripe 3 lacZ expression can be seen at this stage. (G,H) The discordance of native eve stripes relative to persisting lacZ mRNA stripes is apparent in panels (see text for details).

Pattern of stripe 2 lacZ expression from different species

We compared the spatial patterns of stripe 2 transgene expression from four different species to ask whether the evolutionary differences in the stripe 2 sequences had detectable effects on either the anterior-posterior localization or the width of the stripes. To ensure that each stained embryo being compared was at the same timepoint in development, we took advantage of the fact that when both native and transgene stripes are first sharply resolved at the mid-cellularization stage, the 3+7 lacZ stripes coincide nearly perfectly with their corresponding native eve stripes. Since all of our constructs carried a common 3+7 element, we restricted our analysis to those embryos in which the native and transgene 3+7 stripes coincided. Those embryos were assumed to be at a nearly identical timepoint in development, at least with respect to the progression of the morphogenetic gradients affecting the spatiotemporal expression of stripes 3+7.

Representative double-stained embryos of transformed lines bearing stripe 2 enhancers from D. melanogaster, D. yakuba, D. erecta and D. pseudoobscura are shown in Fig. 6. The eve stripe 2 enhancer regions from all four species produce a pattern of lacZ expression that is coincident with the D. melanogaster native eve stripe 2. After examining double-stained embryos from each of the replicate transformed lines, we did not detect any consistent shift or expansion of stripe 2 transgene expression. We conclude that the evolutionary differences in the eve stripe 2 enhancers have little or no effect on the function of the enhancer in terms of the spatial localization of early stripe 2 expression.

Fig. 6.

Comparison of lacZ mRNA expression driven by stripe 2 enhancers from D. melanogaster, D. yakuba, D. erecta and D. pseudoobscura. eve protein (brown) and lacZ mRNA (purple) are simultaneously detected in the embryos transformed with the stripe 2 enhancers from each of the four species and D. melanogaster stripe 3+7 enhancer-lacZ gene fusion. Embryos were selected to be at the same timepoint in development by choosing ones in which the native eve protein stripes 3 and 7, and the lacZ stripes coincided, and by observing the extent of cellularization. (A,B) Stripe 2 enhancer from D. melanogaster; (C,D) stripe 2 enhancer from D. yakuba; (E,F) stripe 2 enhancer from D. erecta; (G,H) stripe 2 enhancer from D. pseudoobscura. (A,C,E,G) Sagittal focus; (B,D,F,H) higher magnification in superficial focus of stripes 2 and 3, from the embryos in A,C,E,G. The eve stripe 2 enhancer regions from all four species produce a pattern of lacZ expression that is coincident with the D. melanogaster native eve stripe 2.

Fig. 6.

Comparison of lacZ mRNA expression driven by stripe 2 enhancers from D. melanogaster, D. yakuba, D. erecta and D. pseudoobscura. eve protein (brown) and lacZ mRNA (purple) are simultaneously detected in the embryos transformed with the stripe 2 enhancers from each of the four species and D. melanogaster stripe 3+7 enhancer-lacZ gene fusion. Embryos were selected to be at the same timepoint in development by choosing ones in which the native eve protein stripes 3 and 7, and the lacZ stripes coincided, and by observing the extent of cellularization. (A,B) Stripe 2 enhancer from D. melanogaster; (C,D) stripe 2 enhancer from D. yakuba; (E,F) stripe 2 enhancer from D. erecta; (G,H) stripe 2 enhancer from D. pseudoobscura. (A,C,E,G) Sagittal focus; (B,D,F,H) higher magnification in superficial focus of stripes 2 and 3, from the embryos in A,C,E,G. The eve stripe 2 enhancer regions from all four species produce a pattern of lacZ expression that is coincident with the D. melanogaster native eve stripe 2.

Level of stripe 2 lacZ expression from different species

The staining technique used in this study did not permit accurate quantification of lacZ expression. However, we did notice that the lacZ stripes produced by non-melanogaster stripe 2 enhancers were noticeably less intense when compared to the D. melanogaster stripe 3+7 control, at the time of their coincidence with native eve stripe 2. The lacZ staining intensity was lowest for the D. erecta stripe 2 construct. These qualitative differences in staining intensity could be seen in independent transformants of each construct. We also noticed that these stripes were not as uniform along the dorsal-ventral axis. All of these constructs contained the homologous region of stripe 2 enhancer DNA (see Materials and methods), so any difference in expression must be due to mutational differences in the enhancers. Possible explanations for the reduced lacZ expression by the non-melanogaster stripe 2 enhancers are presented below.

Conservation of the stripe 2 enhancer expression

The experimental results presented here show that the evolutionary divergence of the four eve stripe 2 enhancers has no discernible effect on either the timing or spatial localization of stripe 2 expression. Each of the four stripe 2 enhancers directs lacZ expression to the same set of cells that are expressing native eve protein in D. melanogaster, and they do so at identical timepoints in embryonic development (within the time resolution of our analysis). eve stripe 2 expression, therefore, is functionally conserved to a remarkable degree in these species. This functional conservation, we hypothesize, must be the consequence of stabilizing selection maintaining a single narrow band of eve expression in the early embryo.

The lack of evidence for functional evolution of the stripe 2 enhancer implies, by logical extension, that there has also been no species-specific coevolution of this enhancer with the morphogens to which they are responding. We hypothesize that the spatial and temporal expression of these morphogens must be nearly the same in each of the species in order that the stripe 2 enhancers from each of them respond identically to the regulatory signals of D. melanogaster. More specifically, features of the trans-acting factors – bicoid, hunchback, Kruppel and giant – responsible for the enhancer’s activity must also be functionally conserved in each of the four species. This argument is consistent with the observation that the domains of expression of many segmentation genes are largely conserved within the Diptera (Sommer and Tautz, 1991). The experimental test of this prediction, however, awaits the reciprocal transformation of the stripe 2 reporter constructs into non-melanogaster species.

Evolutionary changes in the stripe 2 enhancer

In contrast to the functional conservation of the stripe 2 enhancer expression, we found that two binding sites, bcd-3 and the hb-1, do not have obvious homologs in the other species. D. pseudoobscura, D. erecta and D. yakuba do not have a bcd-3 site, and D. pseudoobscura and D. erecta do not have a hb-1 site. (It is also possible that D. yakuba’s hb-1 site is nonfunctional, even though a mutated version of it can be identified.) Of the two sites not present in these species, one of them, the bcd-3 site, has previously been shown to be required for ‘normal’ MSE expression (Small et al., 1992). It is possible that the reduced transgene expression observed for the non-melanogaster constructs is the result of the smaller number of activator binding sites present in their stripe 2 enhancers. If true, the evolutionary gain of the bcd-3 and hb-1 sites in D. melanogaster and D. simulans may be an adaptive response to a reduction in the level of bicoid and/or hunchback proteins in these species. At present, there is no empirical evidence bearing on this hypothesis. An alternative evolutionary hypothesis for the gain of these sites in the lineage leading to D. melanogaster is presented below.

All of the remaining 14 binding sites identified in D. melanogaster are conserved at the sequence level, but only three of them completely so. A striking feature of the large majority of the nucleotide substitutions in the binding site sequences is that each substitution is present only once in the phylogeny. In other words, most of the substitutions in the binding sites occur at otherwise conserved, and presumably functionally important, positions. This suggests that the observed changes at these sites must not be selectively neutral. Rather, we speculate that they are likely to be either adaptive substitutions or slightly deleterious mutations fixed by genetic drift. A site in which a slightly deleterious mutation has been fixed is a good candidate for a subsequent convergent substitution by the adaptively favored mutation. Two convergent mutations can be identified in all of the binding sites, the aforementioned position 4 in kr-3 (also bcd-1) and position 2 in gt-3. The latter site, however, has mutated to three of four possible nucleotides, and may be one of the few exceptional binding site positions that is not functionally constrained.

Nearly all of the spacers that separate adjacent binding sites in the stripe 2 enhancer are variable in length. For example, in D. melanogaster, the gt-2-binding site is 53 bp from bcd-4, its closest activator site, but it is 135 bp from the bcd-4 site in D. pseudoobscura. A rough proportionality of spacer lengths, however, is maintained in each species, indicating that there may be evolutionary constraints on the magnitude of acceptable changes. Nevertheless, the differences in spacer lengths among species raises the possibility that the dynamics of quenching of specific activators by nearby repressors may vary.

Evidence supporting spacing requirements for transcription activation and repression shows that insertions and deletions in an enhancer, even small ones, have the potential to be subject to natural selection. Previous studies in Drosophila and yeast have shown that the spacing between interacting bicoid-binding sites is critical for activation of transcription, although the spacing is different in the two species (Hanes et al., 1994). Mechanisms of short-range transcriptional repression, such as local quenching and dominant repression, require close linkage (<100 bp) of the repressor with upstream activators (Gray et al., 1994; Gray and Levine, 1996).

Evolutionary explanations for functional conservation and structural change

The stripe 2 enhancers of the four Drosophila species contain numerous differences, including the number of binding sites, the sequences of the binding sites and the spacing between them. Despite these differences, they all drive lacZ expression at the appropriate time and location in D. melanogaster early embryos. We now consider two evolutionary mechanisms to account for this functional conservation of the stripe 2 enhancer in the face of the observed structural differences. First, all of the sequence changes might be selectively neutral. It is reasonable to speculate that base substitutions and small length changes in the spacer regions have no functional effect on eve stripe 2 expression. If so, then these substitutions will have been selectively neutral. However, at least one change – the gain of a bcd-3 site in D. melanogaster – has been shown to be functionally important, and this change is not likely to have been selectively neutral. Otherwise, one would have to argue that the results from transgene analysis of the MSE do not apply to the in vivo expression of the native stripe 2 element. We think this is unlikely, due to the inherently greater sensitivity of natural selection in these species to detect mutations of extremely small functional effect, including ones that cannot possibly be measured experimentally. Codon preference in Drosophila is a good example of selection acting on synonymous mutations that have extremely subtle effects on the expression level of a gene (Akashi, 1995).

We have previously argued that the evolution of multiple binding sites for bicoid, hunchback, Kruppel and giant in the stripe 2 enhancer can be understood in terms of selection for functionally robust localization of eve expression, possibly owing to selection for the canalization of pair-rule gene expression. A mechanistic explanation for multiple binding sites provides support for this argument. Ma et al. (1996) recently showed that multiple binding sites for bicoid promote cooperative binding of this protein in an enhancer element of the hunchback gene, and that this cooperativity is necessary to achieve a sharp on/off switch of gene expression. We propose that many of the substitutions in the binding sites of the stripe 2 enhancer and some of the length changes in the spacers have functional effects on stripe 2 expression, but that the magnitude of these effects are ameliorated by functional robustness of the enhancer. In addition, flexibility in the structural design of the enhancer allows for rapid compensatory evolutionary changes, leading to overall functional conservation.

This proposition is compatible with the view that stabilizing selection acting on the timing and spatial localization of stripe 2 expression is the main evolutionary force governing this enhancer’s evolution. Mutations of small effect, including slightly deleterious ones, can become fixed under stabilizing selection when there is functional ‘redundancy’ and epistasis, or when a large number of segregating mutations are contributing to a quantitative character (Kimura, 1981). Both are characteristic features of the stripe 2 enhancer architecture. Adaptive compensatory changes would be required in order to re-establish optimal regulatory performance after the fixation of deleterious mutation by genetic drift. The emergence of the bcd-3 site in D. melanogaster and D. simulans, according to this argument, may be a specific evolutionary response to the flux of weakly functional substitutions occurring at other activator sites.

eve stripe 2 expression is influenced by a number of factors extrinsic to the enhancer itself, each of which can also be subject to stabilizing selection to maintain the optimal regulation of eve stripe 2 expression. These extrinsic factors include the distance between the enhancer and the basal promoter, the basal transcriptional complex through which it acts, and the structure of the eve autoregulatory element. However, the fact that the stripe 2 enhancers from the four species produce indistinguishable temporal and spatial patterns of expression in our experiments, indicates that these extrinsic factors are not coevolving with the stripe 2 enhancers to regulate these aspects of gene expression.

Levels of gene expression driven by the stripe 2 enhancers from different species

The constancy seen in the spatial and temporal patterns of the enhancer-driven lacZ expression is not observed for levels of stripe 2 lacZ expression, which appear to be lower when driven by non-melanogaster sequences, especially that of D. erecta. One possible explanation for this observation is that non-melanogaster sequences lack the bcd-3 and hb-1 activator binding sites. As has previously been shown by Small et al. (1992), mutations in the low-affinity bcd-3 activator site caused a reduction, but not loss, of stripe 2 expression. The significance of the hb-1 activator binding site for stripe 2 expression has not been determined experimentally, but its absence will decrease the number of activator molecules that can be bound to the enhancer. This decrease might weaken the overall activation ability of the enhancer (Arnosti et al., 1996). Alternatively, it is possible that the absence of hb-1 in D. erecta and D. pseudoobscura, and its possible absence in D. yakuba, allows Kruppel protein bound to the kr-1 site to repress transcription by directly interacting with the proximal promoter in our construct. Similarly, if there are direct interactions between any of the activators and the proximal promoter in our construct, then length differences in the enhancers may be influencing the strength of those interactions. Finally, it is possible to hypothesize the presence of unidentified repressor binding site(s) in these species’ enhancers that are not present in the D. melanogaster stripe 2 enhancer.

Rules of enhancer evolution

The substitution pattern seen in this 5′ regulatory region of eve is similar to that found in other regulatory regions containing enhancer elements. Regulatory control regions of fushi tarazu (D. melanogaster vs. D. hydei [Maier et al., 1990]), hairy (D. melanogaster vs. D. virilis [Langland and Carroll, 1993]), and vestigial (D. melanogaster vs. D. virilis [Williams et al., 1994]) all exhibit nearly identical patterns of substitution: small blocks of strongly conserved sequences interspersed among stretches containing many base substitutions and length changes. All three genes play essential roles in developmental processes and, like eve, the regulatory control of expression in these genes is expected to be functionally conserved. Not surprisingly, in each of the three cases, P-element-mediated transformation of a non-melanogaster regulatory sequence directed the expression of a reporter gene in D. melanogaster in a spatially and temporally conserved manner. Given that these species diverged from their most recent common ancestor with D. melanogaster approximately 60 million years ago, the blocks of conserved sequences are almost certainly the result of natural selection (Hartl and Lozovskaya, 1994).

Only bicoid, hunchback, Kruppel and giant are known to be required for eve stripe 2 expression, and many of the conserved blocks in the stripe 2 enhancer are binding sites for these proteins (Fig. 3). Thus, factor binding appears to be the major selective constraint acting on the stripe 2 enhancer. There are conserved blocks in the stripe 2 enhancer, however, that are not known factor binding sites. These sites are not likely to be cryptic bicoid, hunchback or Kruppel sites since none of them have the required sequence motifs for these proteins. (The presence of cryptic giant-binding sites cannot be ruled out because no consensus is available for this protein.) It will be interesting to know whether these conserved blocks are binding sites for proteins, or whether they play other structural roles in enhancer function.

Almost all enhancers have a modular design and are characterized by having multiple binding sites for each of a small number of positive and negative regulators (Arnone and Davidson, 1997). The multiplicity of binding sites may be important in cooperative binding and in assuring robust performance. Quenching is also a general mechanism for negative regulation in enhancers. Thus, we expect that many enhancers will evolve in a manner similar to that of the eve stripe 2 enhancer. It should be of general interest, therefore, to determine how functional conservation of the stripe 2 expression pattern is achieved in evolution, given evidence that some changes are likely to have been selected.

We thank Steven Small for his helpful discussions, materials, and background information that has made this work possible. We would also like to thank Molly Duman-Scheel, Casey Bergman and Michael Wade for the discussion of this work. We thank Joy Hatzidakis for sequencing the region containing the stripe 2 element in D. erecta. This work was supported by NSF grant (MCB-9604477) to M. K. and M. Z. L. and NIH grant (GM47268) to N. H. P. who is also an HHMI investigator.

Ahringer
,
J.
(
1996
).
Posterior patterning by the Caenorhabditis elegans even-skipped homolog vab-7
.
Genes Dev.
10
,
1120
1130
.
Akashi
,
H.
(
1995
).
Inferring weak selection from patterns of polymorphism and divergence at ‘silent’ sites in Drosophila DNA
.
Genetics
139
,
1067
1076
.
Arnone
,
M. I.
and
Davidson
,
E. H.
(
1997
).
The hardwiring of development: organization and function of genomic regulatory systems
.
Development
124
,
1851
1864
.
Arnosti
,
D.
,
Barolo
,
S.
,
Levine
,
M.
and
Small
,
S.
(
1996
).
The eve stripe 2 enhancer employs multiple modes of transcriptional synergy
.
Development
122
,
205
214
.
Averof
,
M.
,
Dawes
,
R.
and
Ferrier
,
D.
(
1996
).
Diversification of arthropod Hox genes as paradigm for the evolution of gene functions
.
Seminars in Cell and Dev. Biol.
7
,
539
551
.
Ballard
,
J. W. O.
and
Kreitman
,
M.
(
1994
).
Unraveling selection in the mitochondrial genome of Drosophila
.
Genetics
138
,
757
772
.
Bastian
,
H.
and
Gruss
,
P.
(
1990
).
A murine even-skipped homologue, Evx1, is expressed during early embryogenesis and neurogenesis in a biphasic manner
.
EMBO J.
9
,
1839
1852
.
Frasch
,
M.
and
Levine
,
M.
(
1987
).
Complementary patterns of even-skipped and fushi tarazu expression involve their differential regulation by a common set of segmentation genes in Drosophila
.
Genes Dev.
1
,
981
995
.
Fujioka
,
M.
,
Jaynes
,
J. B.
and
Goto
,
T.
(
1995
).
Early even-skipped stripes act as morphogentic gradients at the single cell level to establish engrailed expression
.
Development
121
,
4371
4382
.
Goto
,
T.
,
Macdonald
,
P.
and
Maniatis
,
T.
(
1989
).
Early and late periodic patterns of even-skipped expression are controlled by distinct regulatory elements that respond to different spatial cues
.
Cell
57
,
413
422
.
Gray
,
S.
,
Szymanski
,
P.
and
Levine
,
M.
(
1994
).
Short-range repression permits multiple enhancers to function autonomously within a complex promoter
.
Genes Dev.
8
,
1829
1838
.
Gray
,
S.
and
Levine
,
M.
(
1996
).
Short-range transcriptional repressors mediate both quenching and direct repression within complex loci in Drosophila
.
Genes Dev.
10
,
700
710
.
Hanes
,
S. D.
,
Riddihough
,
G.
,
Ish-Horowicz
,
D.
and
Brent
,
R.
(
1994
).
Specific DNA recognition and intersite spacing are critical for action of the bicoid morphogen
.
Mol. Cell Biol.
14
,
3364
3375
.
Harding
,
K.
,
Hoey
,
T.
,
Warrior
,
R.
and
Levine
,
M.
(
1989
).
Autoregulatory and gap response elements of the even-skipped promoter of Drosophila
.
EMBO J.
8
,
1205
1212
.
Hartl
,
D. L.
and
Lozovskaya
,
E. R.
(
1994
). Genome evolution: Between the nucleosome and the chromosome. In
Molecular Ecology and Evolution: Approaches and Applications
(ed.
B.
Schierwater
,
B.
Streit
,
G. P.
Wagner
and
R.
DeSalle
), pp.
579
592
.
Verlag Basel, Switzerland
:
Birkhauser
.
Jeffs
,
P. S.
,
Holmes
,
E. C.
and
Ashburner
,
M.
(
1994
).
The molecular evolution of the alcohol dehydrogenase and alcohol dehydrogenase-related genes in the Drosophila melanogaster species subgroup
.
Mol. Biol. Evol.
11
,
287
304
.
Jiang
,
J.
,
Kosman
,
D.
,
Ip
,
Y. T.
and
Levine
,
M.
(
1991a
).
The dorsal morphogen gradient regulates the mesoderm determinant twist in early Drosophila embryos
.
Genes Dev.
5
,
1881
1891
.
Jiang
,
J.
,
Hoey
,
T.
and
Levine
,
M.
(
1991b
).
Autoregulation of a segmentation gene in Drosophila: Combinatorial interaction of the even-skipped homeo box protein with a distal enhancer element
.
Genes Dev.
5
,
265
277
.
Kenyon
,
C.
(
1994
)
If birds can fly, why can’t we? Homeotic genes and evolution
.
Cell
78
,
175
180
.
Kimura
,
M.
(
1981
).
Possibility of extensive neutral evolution under stabilizing selection with special reference to non-random usage of synonymous codons
.
Proc. Natl. Acad. Sci. USA
78
,
5773
5777
.
Kreitman
,
M.
and
Ludwig
,
M.
(
1996
).
Tempo and mode of even-skipped stripe 2 enhancer evolution in Drosophila
.
Seminars in Cell And Developmental Biology
7
,
583
592
.
Langland
,
J. A.
and
Carroll
,
S. B.
(
1993
).
Conservation of regulatory elements controlling hairy pair-rule stripe formation
.
Development
117
,
585
596
.
Lawrence
,
P. A.
,
Johnston
,
P.
,
Macdonald
,
P.
and
Struhl
,
G.
(
1987
).
Borders of parasegments in Drosophila embryos are delimited by the fushi tarazu and even-skipped genes
.
Nature
328
,
440
442
.
Ludwig
,
M Z.
and
Kreitman
,
M.
(
1995
).
Evolutionary dynamics of the enhancer region of even-skipped in Drosophila
.
Mol. Biol. Evol.
12
,
1002
1011
.
Ludwig
,
M. Z.
,
Tamarina
,
N. A.
and
Richmond
,
R. C.
(
1993
).
Localization of sequences controlling the spatial, temporal, and sex-specific expression of the esterase 6 locus in Drosophila melanogaster adults
.
Proc. Natl. Acad. Sci. USA
90
,
6233
6237
.
Ma
,
X.
,
Yuan
,
D.
,
Diepold
,
K.
,
Scarborough
,
T.
and
Ma
,
J.
(
1996
).
The Drosophila morphogenic protein Bicoid binds DNA cooperatively
.
Development
122
,
1195
1206
.
Maier
,
D.
,
Preiss
,
A.
and
Powell
,
J. R.
(
1990
).
Regulation of the segmentation gene fushi tarazu has been functionally conserved in Drosophila
.
EMBO J.
9
,
3957
3966
.
Nusslein-Volhard
,
C.
,
Kluding
,
H.
and
Jurgens
,
G.
(
1985
).
Genes affecting the segmental subdivision of the Drosophila embryo
.
Cold Spring Harbor Symp. Quant. Biol.
50
,
145
154
.
Ochman
,
H.
,
Ajioka
,
J. W.
,
Garza
,
D.
and
Hartl
,
D. L.
(
1989
). Inverse polymerase chain reaction. In
PCR Technology: Principles and Applications for DNA Amplification
(ed.
H. A.
Erlich
), pp.
105
111
.
New York: Stockton Press
.
Patel
,
N. H.
,
Ball
,
E. E.
and
Goodman
,
C. S.
(
1992
).
Changing role of even-skipped during the evolution of insect pattern formation
.
Nature
357
,
339
342
.
Patel
,
N. H.
,
Condron
,
B. G.
and
Zinn
,
K.
(
1994
).
Pair-rule expression patterns of even-skipped are found in both short- and long-germ beetles
.
Nature
367
,
429
434
.
Patel
,
N. H.
(
1994
). Imaging neuronal subsets and other cell types in whole-mount Drosophila embryos and larvae using antibody probes. In
Methods in Cell Biology
(ed.
L. S. B.
Goldstein
and
E. A.
Fyrberg
), pp.
445
487
.
San Diego
:
Academic Press
.
Patel
,
N. H.
(
1996
). In situ hybridization to whole-mount Drosophila embryos. In
A Laboratory Guide to RNA: Isolation, Analysis, and Synthesis
(ed.
P.
Krieg
), pp.
357
370
.
New York
:
Wiley-Liss
.
Robertson
,
H. M.
,
Preston
,
C. R.
,
Phillis
,
R. W.
,
Johnson-Schlitz
,
D. M.
,
Benz
,
W. K.
and
Engels
,
W. R.
(
1988
).
A stable genomic source of P element transposase in Drosophila melanogaster
.
Genetics
118
,
461
470
.
Ruiz i Altaba
,
A.
and
Melton
,
D. A.
(
1989
).
Involvement of the Xenopus homeobox gene Xho3 in pattern formation along the anterior-posterior axis
.
Cell
57
,
317
326
.
Sackerson
,
C.
(
1995
).
Patterns of conservation and divergence at the even-skipped locus of Drosophila
.
Mech. Dev.
51
,
199
215
.
Small
,
S.
,
Blair
,
A.
and
Levine
,
M.
(
1992
).
Regulation of even-skipped stripe 2 in the Drosophila embryo
.
EMBO J.
11
,
4047
4057
.
Small
,
S.
,
Arnosti
,
D. N.
and
Levine
,
M.
(
1993
).
Spacing ensures autonomous expression of different stripe enhancers in the even-skipped promoter
.
Development
119
,
767
772
.
Small
,
S.
,
Blair
,
A.
and
Levine
,
M.
(
1996
).
Regulation of two pair-rule stripes by a single enhancer in the Drosophila embryo
.
Dev. Biol.
175
,
314
324
.
Sommer
,
R.
and
Tautz
,
D.
(
1991
).
Segmentation gene expression in the housefly Musca domestica
.
Development
113
,
419
430
.
Stanojevic
,
D.
,
Small
,
S.
and
Levine
,
M.
(
1991
).
Regulation of a segmentation stripe by overlapping activators and repressors in the Drosophila embryo
.
Science
254
,
1385
1387
.
Thummel
,
C. S.
,
Boulet
,
A. M.
and
Lipshitz
,
H. D.
(
1988
).
Vectors for Drosophila P element-mediated transformation and tissue culture transfection
.
Gene
74
,
445
456
.
Williams
,
J. A.
,
Paddock
,
S. W.
,
Vorwerk
,
K.
and
Carroll
,
S. B.
(
1994
).
Organization of wing formation and induction of a wing-patterning gene at the dorsal/ventral compartment boundary
.
Nature
368
,
299
305
.
Warrior
,
R.
and
Levine
,
M.
(
1990
).
Dose-dependent regulation of pair-rule stripes by gap proteins and the initiation segment polarity
.
Development
110
,
759
767
.