The hairy (h) segmentation gene of Drosophila regulates segmental patterning of the early embryo, and is expressed in a set of anteroposterior stripes during the blastoderm stage. We have used a set of h gene deletions to study the h promoter and the developmental requirements for individual h stripes. The results confirm upstream regulation of h striping but indicate that expression in the anterodorsal head domain depends on sequences downstream of the two transcription initiation sites. Surprisingly, the two anterior-most h domains appear to be dispensable for head development and embryonic viability. One partial promoter deletion expresses ectopic h, leading to misexpression of other segmentation genes and embryonic pattern defects. We demonstrate that h affects patterning outside its apparent stripe domains, supporting a model in which primary pair-rule genes act as concentration-dependent transcriptional regulators, i.e. as local morphogens.

The Drosophila embryonic body-plan arises from a cascade of transcriptional regulation that subdivides the blastoderm embryo into successively more precise spatial domains (reviewed by Ingham, 1988; Howard, 1990; Small and Levine, 1991). Early maternal positional cues lead to regionalised transcription of the gap genes whose overlapping protein domains direct striped expression of the pairrule genes. Both these classes of zygotic genes encode known or putative transcription factors that directly regulate the transcriptional domains within the syncytial blastoderm embryo.

The transduction of early regionalised signals into an accurate metameric (reiterated) pattern can be considered to arise in three stages. First, gap genes define serially repeated domains of pair-rule transcription, each stripe responding to different combinations of overlapping gap proteins (Howard et al., 1988; Goto et al., 1989; Harding et al., 1989; Rushlow et al., 1989; Stanojevic et al., 1989; Howard and Struhl, 1990; Pankratz et al., 1990; Riddihough and Ish-Horowicz, 1991; Small et al., 1991; Stanojevic et al., 1991; Small et al., 1992). Three pair-rule genes, hairy (h), even-skipped (eve) and runt have been termed the ‘primary’ pair-rule genes because they respond directly to gap gene cues, not to other pair-rule genes (Carroll and Scott, 1986; Howard and Ingham, 1986; Ingham and Gergen, 1988). This view derives initially from studies of h chromosomal translocations that disrupt upstream h sequences and cause loss of specific h stripes (Howard et al., 1988). Further studies show that striped expression of h and eve requires about 14 kb and more than 8 kb of upstream DNA respectively, within which lie independent cis-acting transcriptional regulatory elements (‘stripe elements’) that control individual stripes (Howard et al., 1988; Goto et al., 1989; Harding et al., 1989; Rushlow et al., 1989; see below).

Second, the reiterated signals provided by primary pairrule genes direct striped expression of ‘secondary’ pair-rule genes. Thus, a small regulatory region, the 0.6 kb ‘zebra element’ immediately upstream of the site of fushi tarazu (ftz) transcriptional initiation, is sufficient to regulate striped expression of this pair-rule gene (Hiromi et al., 1985; Hiromi and Gehring, 1987; Dearolf et al., 1989; Ueda et al., 1990). Primary pair-rule gene expression appears largely independent of other, ‘secondary’ pair-rule genes.

Finally, primary and secondary pair-rule genes together define the expression patterns of segment polarity genes, such as the adjacent one cell-wide stripes of wingless (wg) and engrailed (en) that establish and maintain the boundaries between the parasegmental metameres within which subsequent pattern is refined (DiNardo and O’Farrell, 1987; Ingham et al., 1988).

More definitive evidence for relatively independent stripe elements within the h promoter comes from recent experiments using lacZ reporter genes. For most h stripes, independent though overlapping transcriptional regulatory elements can be defined that sense regionalised signals and direct lacZ transcription in positions roughly corresponding to h stripe domains (Howard and Struhl, 1990; Pankratz et al., 1990; Riddihough and Ish-Horowicz, 1991). Such gene fusions are sensitive to gap gene mutations, consistent with gap proteins acting directly to regulate h striping (Riddihough and Ish-Horowicz, 1991). eve striping appears similarly controlled, and an eve stripe 2 element has been demonstrated to include multiple binding sites for candidate gap protein regulators (Small et al., 1991; Stanojevic et al., 1991).

The above studies demonstrate the effects of upstream stripe elements on a heterologous promoter, usually that of a heat-shock gene, but are unable to examine aspects of pair-rule striping that depend on the endogenous h promoter or on downstream sequences. Indeed, none of the h-lacZ constructs express in the anterodorsal (AD) h domain in the head, and several lacZ reporter constructs show imprecise striping indicative of missing regulatory elements (Howard and Struhl, 1990; Riddihough and Ish-Horowicz, 1991). Such reporter-gene constructs are also unable to reveal in vivo consequences of altered h patterning. h embryos suffer severe pattern deletions that probably result from h’s role in regulating secondary pair-rule genes. Sequential loss of h stripes leads to successively more extensive cuticular pattern deletions (Howard et al., 1988), but the function of individual pair-rule stripes has not yet been examined in detail.

In this paper, we analyse the in vivo activity of a set of h deletion mutations constructed in vitro. These cause loss of individual h stripes, consistent with striped expression being regulated by an array of independently acting upstream elements within the h promoter. The results also demonstrate that the AD domain of h expression is dependent on sequences downstream of the h promoter, and that the two anterior domains of h expression are not required for cephalic development. By analysing the requirements of individual h stripes for cuticular patterning and segmentation gene expression, we show that h affects embryonic patterning outside apparent stripe domains. We interpret these results in terms of a model where primary pair-rule genes act as local morphogens that define accurate boundaries of their target genes.

Plasmid constructions

The techniques employed for the construction of the plasmids used in this study are documented in (Sambrook et al., 1989). The neo* vector is a derivative of pUChsneo (Steller and Pirrotta, 1985) that was modified as follows (S. M. Parkhurst, personal communication). The XbaI site in the hsp70 promoter of the neo gene was destroyed by filling-in with Klenow polymerase and religation. A polylinker (see below) was then inserted between the SalI and BamHI sites. The SalI site was mutated during this process and can no longer be restricted by this enzyme, so the polylinker thus has the sites (in order); ApaI, NcoI, SfiI, XhoI, NotI, XbaI, BamHI. The polylinker sequence is:
formula

To construct Δh-14.0, the 14.3 kb XhoI-XbaI fragment from the genomic clone λh6 (Howard, 1986) was subcloned between the XhoI and XbaI sites of the neo* polylinker to form h5X. 21.4 kb of the h locus was reconstructed by subcloning the 7.1 kb XbaI fragment from λh3 containing the h protein-coding sequences (Howard, 1986) into the XbaI site of h5X.

The 5′ deletion series was then generated from Δh-14.0. Constructs are named according to the length of DNA they retain upstream of the upstream initiation site of h transcription, 492 bp from the start of h translation (Fig. 1; Rushlow et al., 1989). The plasmid was linearised with XhoI and partially restricted with either HpaI (for Δh-8.2, and Δh-4.9), MluI (for Δh-11.1), KpnI (for Δh-6.9 and Δh-0.2), or SalI (for Δh-3.4 and Δh-1.8). The plasmid ends were blunted with Klenow or T4 DNA polymerase, as appropriate, XhoI linkers (12-mer, New England Biolabs) were attached, and the plasmids recircularised. To construct Δh-9.2, we isolated the upstream 3.8 kb EcoRI-MluI fragment, attached an XhoI linker to the filled-in EcoRI end, and then ligated it into XhoI and MluI restricted Δh-14.0.

MluINV and MluIΔ were constructed by inverting or deleting DNA between the two MluI sites in Δh-14.0. KpnINV and KpnIΔ were generated by inversion or deletion of DNA between the two KpnI sites.

Analysis of h transgenes

We analysed the embryonic activity of X and 2nd chromosome transgenes in the absence of the endogenous h gene by crossing to Df(3L)hi22/TM3 flies and self-crossing the transgene/+; hi22/+ progeny. Where possible, h expression patterns were examined from at least two lines of each construct, additional lines being analysed if expression from the first two lines differed. Transgenes on the 3rd chromosome were recombined with hi22. To analyse the h expression pattern produced by cosh we examined embryos from cosh; hIL79/hi22 flies. hi22 is a deletion of the h coding regions (Ish-Horowicz et al., 1985); hIL79 is an extremely hypomorphic allele (Ingham et al., 1985b; Wainwright and Ish-Horowicz, 1992). We demonstrated that Δh-14.0 rescues h embryonic lethality by mating Δh-14.0 hIL79/TM3 and Δh-14.0 hi22/TM3 flies and scoring for Δh-14.0 hIL79h-14.0 hi22 progeny. These are identified by the strong h bristle phenotype that is not rescued by our transposons (Rushlow et al., 1989). Rescue by Δh-6.9 in combination with KpnIΔ, and by KpnIΔ alone, was demonstrated by generating Δh-6.9; KpnIΔ hi22/KpnIΔ hIL79 and KpnIΔ hi22/KpnIΔ hIL79 flies respectively. Rescue by KpnINV was demonstrated by generating KpnINV/+; KpnINV hi22/hc1 flies (the two independent KpnINV inserts used are each lethal when homozygous) where hc1 is a null allele (Holmgren, 1984; Ingham et al., 1985b; Wainwright and Ish-Horowicz, 1992).

Analysis of embryos

Embryos were collected, handled and analysed according to standard methods (Wieschaus and Nüsslein-Volhard, 1986). Immunohistochemical staining of proteins used second antibodies directly coupled to alkaline phosphatase (Jackson ImmunoResearch Laboratories inc.) and Substrate Kit II (Vector Laboratories). Embryos were mounted in JB-4 methacrylate (Polysciences) and photographed using Nomarski optics. runt expression in MluIΔ hi22 embryos was visualised by whole-mount in situ hybridization performed essentially as described by Tautz and Pfeifle (1989).

PCR mapping of P-element insertions into the h locus

Genomic DNA was purified from adult flies by standard techniques (Ashburner, 1989). A pair of PCR reactions were performed for each insertion, each including a h primer directed upstream toward the upstream h cap site and either a left ‘PL’ or right ‘PR’ P-element primer directed out of the transposon (Fig. 4A). PCR was performed in 50 μl as previously described (Erlich, 1989): 50 cycles of 94°C/1 minute; 58°C/2 minutes; 72°C/3 minutes. Only one of the P element primers yields an amplified fragment, defining the orientation of the insertion; fragment length maps the site of transposon insertion. Initial mapping used h#3568 (corresponding to DNA between positions +717 and +696 downstream of the upstream cap site; Rushlow et al., 1989) in combination with either the PL or PR primers. Mapping was refined by using closer primers and then testing whether amplified fragments included diagnostic restriction sites (see legend to Fig. 4A).

Germline transformation

Germline transformants were generated using a mixture of 500 μg/ml transgene plasmid and 100 μg/ml pП27.1 injected into bw; st embryos, selected on G418-containing food (Spradling, 1986; Parkhurst et al., 1990). Inserts were mapped and balanced as previously described (Riddihough and Ish-Horowicz, 1991).

Analysing interactions between gap genes and MluIΔ

MluIΔ1/TM3 virgin females (an MluIΔ line showing a high frequency of cuticular defects) were mated to balanced males carrying the segmentation gene mutation of interest. Flies were classified into three categories according to the extent to which tergites T2 and T3 were fused: complete, partial or normal. Flies were assessed as ‘normal’ if no defects in the 2nd or 3rd tergites could be observed. ‘Complete’ fusions of the 2nd and 3rd tergites showed continuous pigmentation at the posterior margin of the fused tergite, and no other 2nd/3rd tergite remnants. All other flies were declared ‘partial’ fusions.

Sequence analysis

This analysis was performed in collaboration with Tom Kidd. Sequences extending 150 bp upstream and downstream of each h transcriptional initiation site were compared using the Intelligenetics ALIGN program. The most significant homologies are shown in Fig. 4E.

Upstream h regulatory elements

A 28 kb DNA fragment from the h gene (cosh) rescues h segmentation mutations and therefore includes the regulatory sequences necessary to drive expression in stripes at the blastoderm stage (Fig. 1; Rushlow et al., 1989). We defined upstream sequence requirements for h expression by analysing a series of 5′ deletion mutations derived from Δh-14.0 which contains the same 14 kb of upstream DNA as cosh, but lacks 6.9 kb of downstream sequence (Fig. 1). Our results are generally in accord with previous studies (Howard et al., 1988; Howard and Struhl, 1990; Pankratz et al., 1990; Riddihough and Ish-Horowicz, 1991) showing that more extensive 5′ deletions cause the sequential loss of specific h stripes (Figs 1, 2).

Two internal deletions, KpnIΔ and MluIΔ, and two upstream inversions, KpnINV and MluINV, also lack specific h stripes (Fig. 2E,F). In all cases, the patterns of missing stripes are in accord with previously mapped stripe-controlling elements (Figs 1, 2C-F, 3A). With the exception of MluIΔ (see below), expression from repositioned or reorientated stripe elements remains accurate, as judged by the normal cuticular patterning directed by the deletion and inversion constructs (not shown; Lardelli, 1991).

Stripe 2 appears to be regulated by multiple dispersed elements, because it is expressed in two almost complementary constructs, Δh-6.9 and KpnIΔ (Fig. 2C,D). This finding is consistent with the progressive reduction in stripe 2 expression in hm7, Δh-8.2, Δh-6.9 and hm8 (Howard et al., 1988), and with the reporter gene analysis.

Ectopic h expression in the MluIΔ construct is associated with pattern deletions

In embryos carrying the MluIΔ construct, stripe 5 is broadened anteriorly such that h protein extends into the region between h stripes 4 and 5 (‘interstripe’ 4/5; Fig. 3A-C). This ectopic expression leads to dominant pattern defects; all homozygous MluIΔ embryos are affected in the most strongly affected transformant line. MluIΔ larvae and adults are viable but show fusion of the A2 and A3 segments, correlating with the domain of h misexpression (Fig. 3G-I).

Expression of other segmentation genes is altered in MluIΔ embryos. The misexpression of h in interstripe 4/5 leads to complete loss of ftz stripe 4, and weakening of runt stripe 4; in contrast, eve stripe 4 is broadened (Fig. 3D-F). These interactions are consistent with previous studies of pair-rule interactions showing that h regulates ftz and runt negatively, and eve positively (Carroll and Scott, 1986; Howard and Ingham, 1986; Ish-Horowicz and Pinchin, 1987; Ingham and Gergen, 1988). Nevertheless, we note that runt and h are coexpressed in the broadened h stripe (see Discussion).

h expression from the MluIΔ construct is not uniform, being normal within the stripe 5 domain and weaker in the interstripe (Fig. 3C). As h is expressed at wild-type levels within the stripe, the control elements that drive stripe 5 expression are still functional. Thus, h expression in the interstripe region appears to be due to novel activating sequences being transposed adjacent to the stripe 5 element, not to the disruption of repressor sequences. MluINV and hm8 share similar breakpoints to MluIΔ but do not expand stripe 5 (Figs 1, 2F; Howard et al., 1988).

The distal MluIΔ breakpoint lies within sequences required for expression of h stripes 3 and 4 (Howard and Struhl, 1990), so translocating them adjacent to the stripe 5 element. Such a fusion of two elements could lead to an inappropriate response to gap gene cues that regulate patterning in this region of the embryo, in particular, to knirps (kni) and Krüppel (Kr). We tested this idea by studying the effects of varying gap-gene dosage on the MluIΔ segmentation phenotype. The results imply that h misexpression in MluIΔ is activated by Kr+ and suppressed by kni+, features that are consistent with the expected regulation of h stripes 3 and 4. The frequency and severity of A2/3 segment fusions in MluIΔ/+ flies is strongly enhanced by heterozygosity for kni, and suppressed by heterozygosity for Kr (Table 1). In contrast, heterozygosity for hunchback(hb) or giant (gt) does not affect MluIΔ penetrance (Table 1).

h transposon insertion mutations argue for downstream regulatory elements

Previous studies of upstream h sequences have failed to define upstream sequences that drive AD expression (Howard and Struhl, 1990; Riddihough and Ish-Horowicz, 1991). Our h constructs indicate that downstream sequences are required. Like cosh, Δh-14.0 activates h in all 7 stripes, and rescues h mutations. However, it fails to drive h expression in the anterodorsal AD domain (Fig. 2A,B; Materials and Methods), presumably due to the absence of 6.9 kb of downstream sequence.

Further evidence that AD expression requires downstream elements comes from several h mutations in which upstream P-element insertions weaken striped but not AD expression. hP(Lac,ry+)L43a (Fasano et al., 1988), h58.2c (A. Martinez Arías, personal communication), and h2715.1, (Hiromi et al., 1986) are transposon insertions into the h gene (Fasano et al., 1988; Riddihough and Ish-Horowicz, 1991). h58.2c and h2715.1 embryos express h normally in the AD domain but only weakly in the stripes, as do some but not all P(Lac,ry+)L43a embryos (Fig. 4B-D).

We determined the precise location of the transposons using the polymerase chain reaction (PCR; see Materials and Methods). Our results show that they are inserted very near the two h transcription start sites: h2715.1 and h58.2c lie 105±15 bp and <12 bp, respectively, upstream of the distal initiation site; P(Lac,ry+)L43a is inserted <12 bp upstream of the proximal initiation site, in the 5′-leader of the longer h transcript (Fig. 4A). Such transposon insertions would selectively affect upstream stripe elements, rather than downstream AD regulatory elements, consistent with the observed expression patterns.

h58.2c and P(Lac,ry+)L43a retain weak h expression despite the adjacent transposon insertions, suggesting that both promoters depend on downstream sequences. Computer analysis reveals extensive DNA sequence similarities extending 89 bp downstream of each cap site (Fig. 4E; Materials and Methods). The first 49 bp of each transcript are 61% identical, suggesting that these downstream sequences may be functionally important, presumably recognising similar factors needed for h transcriptional initiation.

Expression of h in the anterodorsal patch and first stripe is not required for embryonic and adult viability

Despite its lack of h expression in the AD domain, Δh-14.0 rescues h mutant embryos. Δh-14.0 h embryos are viable and have normal head structures (Fig. 5B; see Materials and Methods). The AD domain corresponds to the labral primordium, so we re-examined head morphology in hi22 embryos. Contrary to a previous report (Ingham et al., 1985b), we found that the labrum is not absent (Fig. 5E), but deformed according to the ‘dorsal pouch syndrome’ described by (Jürgens et al., 1986). Thus, labral development and embryonic viability are not dependent on h AD expression.

We also examined embryonic viability in the absence of h stripe 1. Surprisingly, it also appears dispensable; indeed, embryos lacking both the AD and stripe 1 h domains are viable, albeit at a low frequency. 2% of h embryos containing two copies of the KpnIΔ construct hatch and survive to adulthood (6/492; see Materials and Methods), and 34% (17/50) have normal head structures. The adult flies display the h bristle phenotype that is not rescued by h transgenes (Rushlow et al., 1989; Materials and Methods), and lack one abdominal segment due to the missing stripe 5. Viability is low because most KpnIΔ h embryos express insufficient stripes 2, 3, or 4, disrupting essential head and thoracic and/or abdominal segments (Fig. 5D).

The dispensability of stripe 1 for head development is confirmed by the examination of hi22 embryos. Despite the grossly disorganised heads, they retain all appropriate cuticular structures derived from the mandibular and maxillary segmental anlagen that lie within h stripe 1 (Fig. 5E-H).

h mutations affect patterning outside the h stripe domains

Analysis of the cuticular pattern in embryos carrying our h deletions has allowed us to relate different pattern deletions associated with specific stripes of h expression (Table 2). Some h stripes affect only a single metameric boundary; e.g. loss of stripe 7 removes the PS12/13 parasegmental boundary (compare MluINV; hi22 and Δh-9.2; hi22 embryos; Fig. 6B,C; Table 2). However, certain h stripes are required for more than one boundary, including primordia that lie outside the stripe domain. Lack of stripe 5 (KpnIΔ hi22 embryos) removes both the PS8/9 and PS9/10 parasegmental borders (fusion of segments A3-A5; Fig. 6D). hm3 embryos display fusions of the T2/A1 or A1/A3 denticle bands (PS5/6 and PS 7/8; Ingham et al., 1985b; Howard et al., 1988), whose primordia also lie outside h stripe domains.

These pattern defects are preceded by the inappropriate expression of other pair-rule genes. ftz expression in KpnIΔ hi22 embryos is expanded anteriorly into the h interstripe 4/5 domain, often fusing with stripe 4 (Fig. 6E). eve stripe 5 is almost completely repressed (Fig. 6F). The disruptions of pair-rule expression lead to loss of en stripes 9 and 10, the latter lying within h interstripe 5/6 (Fig. 6G). Thus, h is required to establish parasegmental boundaries that lie outside its apparent stripe domains.

Our results are most simply explained if nuclei in the interstripes domains are exposed to low levels of h protein which are physiologically significant but immunohistochemically undetectable. Target genes would be differentially regulated according to their sensitivity to h protein concentration. Such a concentration-dependent model has been proposed to explain various patterning defects induced by ectopic runt and eve expression (Gergen and Wieschaus, 1986; Manoukian and Krause, 1992; see Discussion).

Additional evidence that h acts in a concentration-dependent manner comes from the displacement of en stripes when h expression is altered. h stripes 2, 3 and 4 are severely weakened in KpnIΔ hi22 embryos (Fig. 2D). This leads to shifting and pairing of alternate en stripes (Fig. 6G), resembling that caused by reduced eve activity (Frasch et al., 1988). Such displacement of expression domains is a characteristic response to changes in morphogen concentration (Driever and Nüsslein-Volhard, 1988; see Discussion). The differential responses of ftz and runt to interstripe h expression in MluIΔ embryos (Fig. 3D,E) are also consistent with concentration-dependent h action.

h transcription requires both upstream and downstream control elements

Our analysis of h deletions complements the mapping of stripe elements in the h promoter, and confirms that sequences regulating blastoderm h transcription are spread over almost 30 kb (Fig. 1; Howard and Struhl, 1990; Pankratz et al., 1990; Riddihough and Ish-Horowicz, 1991). Upstream h elements confer similar expression domains on both the hsp70 and h promoters, although perhaps not with the same precision (see below).

The majority of stripe elements behave discretely, with particular deletions abolishing expression of specific stripes. However, the stripe 2 element appears to be partially degenerate and dispersed over a region of several kb. A series of deletions broken in the region between −9.2 kb and −4.9 kb weaken, but do not abolish stripe 2 (Howard et al., 1988; this paper), and two essentially complementary deletions (KpnIΔ and Δh-6.9) each express stripe 2 weakly. This explains why small upstream fragments do not drive stripe 2 reporter gene expression (Howard and Struhl, 1990). Such degeneracy may not be exclusive to stripe 2 and may be inherent in defining the regulatory thresholds that appear to define stripe boundaries (e.g. Small et al., 1991; Stanojevic et al., 1991; Small et al., 1992). The molecular mechanisms of such transcriptional thresholds are likely to depend on multiple binding sites, as has been defined for transcriptional activation by bicoid of the hunchback (hb) promoter (Driever and Nüsslein-Volhard, 1989; Struhl et al., 1989). While upstream elements regulate the expression of h in stripes, expression in the anterodorsal (AD) head-domain requires downstream sequences. Not only do constructs lacking downstream sequences fail to activate AD h expression, but transposon insertions adjacent to the transcription start sites (h58.2c, L43a) selectively reduce h expression in the stripes (Fig. 4C,D). Such insertions displace upstream sequences away from the h cap site but leave downstream regions unaffected, suggesting that AD-specific expression is regulated by the latter. We have not yet succeeded in defining discrete elements that drive reporter gene expression in an AD domain (M. L., unpublished observations).

Basal h promoter activity probably also depends on downstream sequences. Computer analysis reveals extensive sequence conservation downstream of the two h cap sites, suggesting that they are recognised by similar regulatory factors (Fig. 4E). Both promoter sites lack TATA consensus sequences (Rushlow et al., 1989), and 3 bp downstream of each initiation site lies a TCAG/TTC sequence resembling the Inr motif described for mammalian TdT and various TATA-independent Drosophila transcripts (Smale and Baltimore, 1989; Arkhipova and Ilyin, 1991). A requirement for downstream sequences may explain why upstream h elements confer slightly imprecise expression domains on an hsp70 promoter (Riddihough and IshHorowicz, 1991). h expression from our deletion constructs must be accurate because they rescue cuticular patterning (Lardelli, 1991).

Inessential domains of h expression in the head

Surprisingly, the AD and stripe 1 domains of h expression appear dispensable for embryonic and adult viability. KpnIΔ h embryos lack these domains, yet can form normal heads and, occasionally, viable adults. This finding led us to reexamine h mutant embryos, and to show that all head structures, although grossly disorganised, appear to be retained (Fig. 5E-H). Either h expression in the AD domain and stripe 1 is not required for essential developmental processes, or pattern formation in these anterior regions is under redundant control. Genetic and evolutionary considerations have suggested that patterning mechanisms differ between the Drosophila head and body-trunk (Cohen and Jürgens, 1990, 1991; Finkelstein and Perrimon, 1991). Head-specific patterning might supplement or override requirements for anterior h expression. For example, anterior ftz expression is subject to ‘polar repression’ by the terminal co-ordinate system, preventing ftz expression anterior of 65%EL (Hiromi et al., 1985; Edgar et al., 1986). Thus, h stripe 1 would not be needed to repress ftz and, indeed, ftz stripe 1 expression does not expand in KpnIΔ hi22 embryos that lack h stripe 1 (Fig. 6E).

Transposed regulatory elements misactivate h expression in MluIΔ

The MluIΔ construct expresses h ectopically in the 4/5 interstripe region and causes deletion of the A3 denticle belt in embryos (Fig. 3G). The pattern deletions result from altered pair-rule expression – reduced ftz and runt expression, expanded eve expression – that lead to loss of the PS7/8 boundary. These effects are consistent with previously demonstrated interactions between h and other pair-rule genes (Carroll and Scott, 1986; Howard and Ingham, 1986; Ish-Horowicz and Pinchin, 1987; Ingham and Gergen, 1988). The interstripe h expression probably arises from the juxtaposition of sequences required for stripe 3 and 4 close to the stripe 5 element (see Results), an interpretation that is supported by the sensitivity of MluIΔ to Kr and kni genedosage (Table 1). Thus, interstripe repression can be overcome by novel adjacent activating sequences, arguing that stripe boundaries are defined by combinations of activation and repression rather than exclusively the latter (Carroll, 1990).

The viability of MluIΔ flies indicates that restricted pattern deletions in the mid-abdomen can be tolerated, whereas equivalent deletions would be lethal in the head, thorax, or posterior abdomen. This is confirmed by our recovery of KpnIΔ h flies that lack h stripe 5 and A4 structures. Fused abdominal defects caused by reduced h expression are also not lethal (Ingham et al., 1985b). Ectopic segmentation gene expression in the mid-abdomen, e.g. driven by the h stripe 5 element, may allow analysis of segmentation gene interactions without associated dominant lethality.

Transcriptional regulation by h appears to be concentration dependent

Embryos completely lacking h activity suffer extensive pattern deletions, more severe than the prototypic pair-rule deletion of alternate metameres (Ingham et al., 1985b). We have shown that the pattern defects are due to direct effects of h on interstripe primordia. For example, lack of h stripe 5 (KpnIΔ h) causes loss of the PS9/10 boundary whose primordium lies within interstripe 4/5 (Fig. 6H). The mispatterning of interstripe primordia is not due to secondary effects on metameric stability because the pattern defects are associated with altered early patterns of segmentation gene expression (ftz, eve, en; Fig. 6E-G). Corresponding regional effects on ftz, eve and en patterns are found in h embryos (ML., unpublished observations).

The apparent nonautonomous action of h is presumably due to exposure of nuclei in the interstripes to low but physiologically significant levels of h protein. The effects of increased runt+ gene-dosage on patterning in interstripe runt domains can be explained similarly (Gergen and Wieschaus, 1986). Before they are expressed in stripes, h and runt are both initially expressed at low levels throughout most of the embryo (Ingham et al., 1985a; Gergen and Butler, 1988). Nevertheless, we consider it unlikely that this early interstripe expression is important for patterning. Ectopic h expression at these stages appears not to cause pattern defects (Ish-Horowicz and Pinchin, 1987).

Rather, we suggest that primary pair-rule protein stripes extend into and regulate gene expression within ‘interstripe’ domains. It has recently been demonstrated that ectopic eve expression shows concentration-dependent repression of target genes, raising the possibility that primary pair-rule genes act as local morphogens in defining target gene domains (Manoukian and Krause, 1992). The boundaries of h and runt protein expression are somewhat diffuse (see Kania et al., 1990), so could lay down concentration gradients that would pattern locally. Concentration-dependent repression by h is consistent with the effects of h misexpression in MluIΔ embryos, in which ectopic h completely represses ftz, but runt expression is only affected near the margins of endogenous h expression where h concentrations are highest (Fig. 3D,E). Also, ftz expression is inhibited by immunologically undetectable levels of h protein in heat shocked hsp70-h embryos that do not affect runt expression (Ish-Horowicz and Pinchin, 1987; Gergen, personal communication). Further evidence comes from the sensitivity of en stripe-spacing to reduced h activity (Results).

Another attractive feature of this view is that it explains how the accurate phasing of secondary pair-rule genes is achieved. Subtle differences in stripe expression domains should be due either to direct interpretation of gap protein gradients, or to interactions between pair-rule genes. The former model appears unlikely because it demands that gap proteins directly establish relative positional distinctions that are accurate to the single-cell. In contrast, local gradients of primary pair-rule proteins would readily define the precisely phased secondary pair-rule stripes that specify segment-polarity domains, and thereby establish sharp metameric boundaries.

The authors wish to thank Susan Parkhurst, Guy Riddihough, Phil Ingham and Sheena Pinchin for much valuable help, advice and discussion, to Phil Ingham, Susan Parkhurst, Zeev Paroush and Guy Riddihough for critical reading of the manuscript, and to Armen Manoukian for a preprint of the hs-eve paper. We are also grateful to Susan Parkhurst for providing the neo* plasmid and to Tom Kidd for performing the DNA sequence comparisons. We should like to thank Alfonso Martinez-Arías, Steve Kerridge and Yash Hiromi for the h enhancer trap lines. Mouse polyclonal anti-h antibodies were provided by Sheena Pinchin; rabbit antiftz, rat anti-en, and anti-eve antibodies were the kind gifts of Henry Krause, Anita Taylor, and Manfred Frasch respectively. M. L. thanks the Imperial Cancer Research Fund for their excellent support in the form of a student bursary.

Arkhipova
,
I. R.
and
Ilyin
,
Y. V.
(
1991
).
Properties of promoter regions of mdg1 Drosophila retrotransposon indicate that it belongs to a specific class of promoters
.
EMBO J
.
10
,
1169
-
1177
.
Ashburner
,
M.
(
1989
).
Drosophila: a Laboratory Manual
.
New York
:
Cold Spring Harbour Press
.
Carroll
,
S. B.
(
1990
).
Zebra patterns in fly embryos: activation of stripes or repression of interstripes?
Cell
60
,
9
-
16
.
Carroll
,
S. B.
and
Scott
,
M. P.
(
1986
).
Zygotically active genes that affect the spatial expression of the fushi tarazu segmentation gene during early Drosophila embryogenesis
.
Cell
45
,
113
-
126
.
Cohen
,
S. M.
and
Jürgens
,
G.
(
1990
).
Mediation of Drosophila head development by gap-like segmentation genes
.
Nature
346
,
482
-
484
.
Cohen
,
S.
and
Jürgens
,
G.
(
1991
).
Drosophila headlines
.
Trends Genet
.
7
,
267
-
272
.
Dearolf
,
C. R.
,
Topol
,
J.
and
Parker
,
C. S.
(
1989
).
Transcriptional control of Drosophila fushi tarazu zebra stripe expression
.
Genes Dev
.
3
,
384398
.
DiNardo
,
S.
and
O’Farrell
,
P. H.
(
1987
).
Establishment and refinement of segmental pattern in the Drosophila embryo: spatial control of engrailed expression by pair-rule genes
.
Genes Dev
.
1
,
1212
-
1225
.
Driever
,
W.
and
Nüsslein-Volhard
,
C.
(
1988
).
The bicoid protein determines position in the Drosophila embryo in a concentrationdependent manner
.
Cell
54
,
95
-
104
.
Driever
,
W.
and
Nüsslein-Volhard
,
C.
(
1989
).
The bicoid protein is a positive regulator of hunchback transcription in the early Drosophila embryo
.
Nature
337
,
138
-
143
.
Edgar
,
B. A.
,
Weir
,
M. P.
,
Schubiger
,
G.
and
Kornberg
,
T.
(
1986
).
Repression and turnover pattern of fushi tarazu RNA in the early Drosophila embryo
.
Cell
47
,
747
-
54
.
Erlich
,
H. A.
(
1989
).
PCR technology
.
New York
:
Stockton Press, New York
.
Fasano
,
L.
,
Coré
,
N.
and
Kerridge
,
S.
(
1988
).
Expression of a reporter gene resembles that of its neighbour: an insertion in the hairy gene of Drosophila
.
Roux’s Arch. Dev. Biol
.
197
,
507
-
512
.
Finkelstein
,
R.
and
Perrimon
,
N.
(
1991
).
The molecular-genetics of head development in Drosophila melanogaster
.
Development
112
,
899
-
912
.
Frasch
,
M.
,
Warrior
,
R.
,
Tugwood
,
J.
and
Levine
,
M.
(
1988
).
Molecular analysis of even-skipped mutants in Drosophila development
.
Genes Dev
.
2
,
1824
-
1838
.
Gergen
,
J. P.
and
Butler
,
B. A.
(
1988
).
Isolation of the Drosophila segmentation gene runt and analysis of its expression during embryogenesis
.
Genes Dev
.
2
,
1179
-
1193
.
Gergen
,
J. P.
and
Wieschaus
,
E.
(
1986
).
Dosage requirements for runt in the segmentation of Drosophila embryos
.
Cell
45
,
289
-
299
.
Goto
,
T.
,
Macdonald
,
P. M.
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
,
413422
.
Harding
,
K.
,
Hoey
,
T.
,
Warrior
,
R.
and
Levine
,
M.
(
1989
).
Autoregulatory and gap gene response elements of the even-skipped promoter of Drosophila
.
EMBO J
.
8
,
1205
-
1212
.
Hiromi
,
Y.
and
Gehring
,
W. J.
(
1987
).
Regulation and function of the Drosophila segmentation gene fushi tarazu
.
Cell
50
,
963
-
974
.
Hiromi
,
Y.
,
Kuroiwa
,
A.
and
Gehring
,
W. J.
(
1985
).
Control elements of the Drosophila segmentation gene fushi tarazu
.
Cell
43
,
603
-
613
.
Hiromi
,
Y.
,
Okamoto
,
H.
,
Gehring
,
W. J.
and
Hotta
,
Y.
(
1986
).
Germline transformation with Drosophila mutant actin genes induces constitutive expression of heat shock genes
.
Cell
44
,
293
-
301
.
Holmgren
,
R.
(
1984
).
Cloning sequences from the hairy gene of Drosophila
.
EMBO J
.
3
,
569
-
573
.
Howard
,
K.
and
Ingham
,
P.
(
1986
).
Regulatory interactions between the segmentation genes fushi tarazu, hairy and engrailed in the Drosophila blastoderm
.
Cell
44
,
949
-
957
.
Howard
,
K.
,
Ingham
,
P.
and
Rushlow
,
C.
(
1988
).
Region-specific alleles of the Drosophila segmentation gene hairy
.
Genes Dev
.
2
,
1037
-
1046
.
Howard
,
K. R.
(
1986
).
Molecular genetics of the hairy locus of Drosophila melanogaster. Ph.D. thesis, University of London
.
Howard
,
K. R.
(
1990
).
The blastoderm pattern
. In
Pattern formation in Drosophila
, (Ed.
P.W.
Ingham
), Seminars in Cell Biology,
1
,
pp161
-
172
, Philadelphia: W.B. Saunders Company.
Howard
,
K. R.
and
Struhl
,
G.
(
1990
).
Decoding positional information: regulation of the pair-rule gene hairy
.
Development
110
,
1223
-
1231
.
Ingham
,
P. W.
(
1988
).
The molecular genetics of embryonic pattern formation in Drosophila
.
Nature
335
,
25
-
34
.
Ingham
,
P. W.
,
Baker
,
N. E.
and
Martinez-Arías
,
A.
(
1988
).
Regulation of segment polarity genes in the Drosophila blastoderm by fushi tarazu and even-skipped
.
Nature
331
,
73
-
75
.
Ingham
,
P. W.
and
Gergen
,
J. P.
(
1988
).
Interactions between the pair-rule genes runt, hairy, even-skipped and fushi tarazu and the establishment of periodic pattern in the Drosophila embryo
. In
Mechanisms of Segmentation
, (Ed.
V.
French
,
P. W.
Ingham
,
J.
Cooke
and
J.
Smith
), Development,
104
Supplement,
51
-
60
.
Ingham
,
P. W.
,
Howard
,
K. R.
and
Ish-Horowicz
,
D.
(
1985a
).
Transcription pattern of the Drosophila segmentation gene hairy
.
Nature
318
,
439
-
445
.
Ingham
,
P. W.
,
Pinchin
,
S. M.
,
Howard
,
K. R.
and
Ish-Horowicz
,
D.
(
1985b
).
Genetic analysis of the hairy locus in Drosophila melanogaster
.
Genetics
111
,
463
-
486
.
Ish-Horowicz
,
D.
,
Howard
,
K. R.
,
Pinchin
,
S. M.
and
Ingham
,
P. W.
(
1985
).
Molecular and genetic analysis of the hairy locus in Drosophila
.
Cold Spring Harb. Symp. Quant. Biol
.
50
,
135
-
144
.
Ish-Horowicz
,
D.
and
Pinchin
,
S. M.
(
1987
).
Pattern abnormalities induced by ectopic expression of the Drosophila gene hairy are associated with repression of fushi tarazu transcription
.
Cell
51
,
405
-
415
.
Jürgens
,
G.
,
Lehmann
,
R.
,
Schardin
,
M.
and
Nüsslein-Volhard
,
C.
(
1986
).
Segmental organisation of the head in the embryo of Drosophila melanogaster
.
Roux’s Arch. Dev. Biol
195
,
359
-
377
.
Kania
,
M. A.
,
Bonner
,
A. S.
,
Duffy
,
J. B.
and
Gergen
,
J. P.
(
1990
).
The Drosophila segmentation gene runt encodes a novel nuclear regulatory protein that is also expressed in the developing nervous system
.
Genes Dev
.
4
,
1701
-
1713
.
Lardelli
,
M.
(
1991
).
Regulation and function of the pair-rule gene hairy in Drosophila melanogaster. PhD. thesis, Council for National Academic Awards
.
Manoukian
,
A. S.
and
Krause
,
H. M.
(
1992
).
Concentration-dependent regulatory activities of the even-skipped protein in Drosophila embryos
.
Genes Dev
.
6
,
1740
-
1751
.
Pankratz
,
M. J.
,
Seifert
,
E.
,
Gerwin
,
N.
,
Billi
,
B.
,
Nauber
,
U.
and
Jäckle
,
H.
(
1990
).
Gradients of Krüppel and knirps gene products direct pair-rule gene stripe patterning in the posterior region of the Drosophila embryo
.
Cell
61
,
309
-
317
.
Parkhurst
,
S. M.
,
Bopp
,
D.
and
Ish-Horowicz
,
D.
(
1990
).
X:A ratio, the primary sex determining signal in Drosophila, is transduced by helixloop-helix proteins
.
Cell
63
,
1179
-
1191
.
Riddihough
,
G.
and
Ish-Horowicz
,
D.
(
1991
).
Individual stripe regulatory elements in the Drosophila hairy promoter respond to maternal, gap and pair-rule genes
.
Genes Dev
.
5
,
840
-
854
.
Rushlow
,
C. A.
,
Hogan
,
A.
,
Pinchin
,
S. M.
,
Howe
,
K. R.
,
Lardelli
,
M. T.
and
Ish-Horowicz
,
D.
(
1989
).
The Drosophila hairy protein acts in both segmentation and bristle patterning and shows homology to N-myc
.
EMBO J
.
8
,
3095
-
3103
.
Sambrook
,
J.
,
Fritsch
,
E. F.
and
Maniatis
,
T.
(
1989
).
Molecular cloning: a laboratory manual
.
New York
:
Cold Spring Harbor Press
.
Smale
,
S. T.
and
Baltimore
,
D.
(
1989
).
The ‘initiator’ as a transcription control element
.
Cell
57
,
103
-
113
.
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.
,
Kraut
,
R.
,
Hoey
,
T.
,
Warrior
,
R.
and
Levine
,
M.
(
1991
).
Transcriptional regulation of a pair-rule stripe in Drosophila
.
Genes Dev
.
5
,
827
-
839
.
Small
,
S.
and
Levine
,
M.
(
1991
).
The initiation of pair-rule stripes in the Drosophila blastoderm
.
Curr. Opin. Genet. Dev
.
1
,
255
-
260
.
Spradling
,
A. C.
(
1986
).
P element-mediated transformation
. In
Drosophila: a practical approach
(Ed.
D.B.
Roberts
), pp
175
-
197
,
Oxford
:
IRL Press
.
Stanojevic
,
D.
,
Hoey
,
T.
and
Levine
,
M.
(
1989
).
Sequence specific DNAbinding activities of the gap proteins encoded by hunchback and Krüppel in Drosophila
.
Nature
341
,
331
-
335
.
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
.
Steller
,
H.
and
Pirrotta
,
V.
(
1985
).
A transposable P vector that confers G418 resistance to Drosophila larvae
.
EMBO J
.
4
,
167
-
171
.
Struhl
,
G.
,
Struhl
,
K.
and
Macdonald
,
P. M.
(
1989
).
The gradient morphogen bicoid is a concentration-dependent transcriptional activator
.
Cell
57
,
1259
-
1273
.
Tautz
,
D.
and
Pfeifle
,
C.
(
1989
).
A non-radioactive in situ hybridization method for the localization of specific RNAs in Drosophila embryos reveals translational control of the segmentation gene hunchback
.
Chromosoma
98
,
81
-
85
.
Ueda
,
H.
,
Sonoda
,
S.
,
Brown
,
J. L.
,
Scott
,
M. P.
and
Wu
,
C.
(
1990
).
A sequence-specific DNA-binding protein that activates fushi tarazu segmentation gene expression
.
Genes Dev
.
4
,
624
-
635
.
Wainwright
,
S. M.
and
Ish-Horowicz
,
D.
(
1992
).
Point mutations in the Drosophila hairy gene demonstrate in vivo requirements for basic, helixloop-helix, and WRPW domains
.
Mol. Cell Biol
.
12
,
2475
-
2483
.
Wieschaus
,
E.
and
Nüsslein-Volhard
,
C.
(
1986
).
Looking at embryos
. In
Drosophila, a practical approach
, (Ed.
D.B.
Roberts
), pp
199
-
226
,
Oxford
:
IRL Press
.