We have used a polyclonal antiserum specific for the Drosophila segmentation gene, hairy (h), to analyse its expression during embryogenesis. The pattern of wildtype expression resembles that of h transcription, being expressed in stripes at the blastoderm stage, h is also expressed later in the stomodæum, proctodæum, tracheal pits and mesoderm. We demonstrate that h protein stripes show consistent phase relationships to those of the even-skipped (eve) pair-rule gene. We examine h protein patterns in embryos mutant for other segmentation genes, including h itself. We show that lack of h activity appears not to affect h striping, arguing that h expression is not under autoregulatory control. We also show that h activity is not needed for tracheal invagination. Mutations that are rearranged upstream of the h gene cause the loss of specific stripes, indicating that the h promoter includes activating elements that respond to specific spatial cues. Our observations suggest that pairrule striping may be under redundant control, and we discuss possible implications for hierarchical models of pair-rule gene action.

The basic body plan of the Drosophila embryo is under the control of the segmentation genes, whose interactions subdivide the Drosophila embryo into increasingly refined anteroposterior spatial domains (reviewed in Akam, 1987; Ingham, 1988). The blastoderm-stage embryo is subdivided into parasegmental metameric units through the action of the pair-rule genes, the first segmentation genes to show overt periodicity. Each parasegment (PS) is marked at the anterior by the expression of a single-cell-wide stripe of the segmentpolarity gene engrailed (en), and at the posterior by a similar stripe of wingless (wg) expression (Fig. 1; Martinez-Arias and Lawrence, 1985; Ingham et al. 1988). Thereafter, each parasegment behaves as a secondary developmental field within which more elaborate pattern is refined.

Fig. 1.

Diagram of segmentation gene expression at the blastoderm stage, indicating pair-rule domains, en and wg domains, and parasegmental organisation. The putative phasing of h domains is based on the results of Ingham et al. (1985a), Carroll et al. (1988) and this paper.

Fig. 1.

Diagram of segmentation gene expression at the blastoderm stage, indicating pair-rule domains, en and wg domains, and parasegmental organisation. The putative phasing of h domains is based on the results of Ingham et al. (1985a), Carroll et al. (1988) and this paper.

The pair-rule genes act in combination to define the domains of segment-polarity gene expression. Thus, en activation requires the combination of even-skipped (eve) and paired (prd), or otfushi tarazu (ftz) and odd-paired (opa) (Ingham et al. 1988). Such pair-rule genes encode nuclear proteins whose sites of direct action can only be short-range (see above reviews). Their stripes must be established with great precision if they are to define one-cell-wide spatial domains. Current data are consistent with accurate overlapping phasing of pairrule genes, but have mainly been restricted to localising transcripts by in situ hybridisation, a technique that is not accurate to the single cell (see below).

The mechanism whereby stripes are established is not yet clear but involves both positive and negative regulation of transcription (Hiromi et al. 1985; Edgar et al. 1986; Hiromi and Gehring, 1987; Howard et al. 1988). Most pair-rule genes are initially expressed over very broad domains that resolve into stripes during the final syncitial interphase. For each pair-rule gene, transcription is enhanced within its stripe domains and suppressed between them. Laying down signals for differential expression appears to require the gap genes, such as Krtippel (Kr), hunchback (hb) and knirps (kni). Gap gene expression precedes that of the pair-rule genes, and arises in wide overlapping domains (Knipple et al. 1985; Jackie et al. 1986; Tautz, 1988). It is unclear how these broad regions give rise to precise pair-rule striping, as the gap gene protein domains do not have sharp margins that would serve as obvious boundaries.

Further refinement of the pattern may arise from interactions between pair-rule genes. Three genes hairy (h), runt (run) and even-skipped (eve) regulate each others expression and are therefore implicated in the generation of periodicity. These ‘primary’ pair-rule genes are also required for ftz striping (Carroll and Scott, 1986; Howard and Ingham, 1986; Harding et al. 1986; Ingham and Gergen, 1988), suggesting that ftz striping may be a response to other pair-rule genes. In contrast, ftz appears not to regulate the striping of primary pair-rule genes, as h, run and eve pattern is normal in ftz embryos.

The best evidence for a direct interaction between pair-rule genes comes from experiments showing that h regulates ftz transcription, ftz stripes are broadened in h mutant embryos, suggesting that h behaves as a negative regulator of ftz striping (Howard and Ingham, 1986; Carroll and Scott, 1986). Moreover, the induction of ectopic h expression from a heat-shock promoter represses ftz transcription (Ish-Horowicz and Pinchin, 1987). The time course of the response argues that h is acting directly on ftz expression.

Such regulation predicts that h and ftz domains should be exclusive. However, there are indications that h and ftz transcriptional domains are not exactly reciprocal. In situ hybridisation with mixed h and ftz probes shows that one cell per double-metamere expresses neither h nor ftz at a time when the h and ftz stripes are each one metamere wide (Ish-Horowicz et al. 1985; Ingham et al. 1985a). This has been confirmed using mixed h and ftz antibodies (Carroll et al. 1988; Hooper, 1989), and suggests that h and ftz domains might overlap by one cell (Fig. 1). However, analysing serial sections with different probes is insufficiently precise to define such a single cell overlap.

Carroll et al. (1988) stained embryos with mixed antibodies and suggested that occasional strongly stained cells are due to h/ftz overlap. However, such cells were seen in very few embryos, leaving open whether there is a consistent overlapping phase relationship. Moreover, ftz expression diminishes towards the posterior of each domain (see Lawrence et al. 1987), so that the extent and degree of overlap can be somewhat uncertain.

In this paper, we describe the immunohistochemical detection of the h protein using a polyclonal A-specific antiserum and relate the protein domains to those of other pair-rule genes. We have previously determined the DNA sequence of ∼10kb from the chromosomal h locus and of several h cDNAs (Rushlow et al. 1989). The h transcript encodes a putative protein of 337 amino acids, that has been shown to be active in pattern formation: its ectopic expression induces pair-rule pattern defects at blastoderin (Ish-Horowicz and Pinchin, 1987).

h also acts during late larval/early pupal development in the establishment of adult bristle pattern (Ingham et al. 19856). Previous work has suggested that it may operate through the Achaete-Scute Complex (AS-C), repressing ectopic expression of the achaete gene (Falk, 1963; Botas et al. 1982; Mosocoso del Prado and Garcia-Bellido, 1984). Several lines of evidence indicate that a single h protein mediates both functions, arguing that it operates similarly in bristle patterning and in segmentation (Rushlow et al. 1989).

We show here that the h protein is nuclear, consistent with a role in transcriptional regulation. We describe the pattern of h expression in wild-type embryos and present direct evidence that pair-rule domains subdivide the embryo into overlapping metameric phasings, as expected if pair-rule genes act in combination to regulate segment-polarity domains. We also analyse h expression in h mutant embryos and in embryos mutant for other segmentation genes, and discuss how models of h striping might account for the patterns observed.

h mutant stocks were isolated as described in Table 1. Other mutations were the gifts of Drs C. Nüsslein-Volhard and R. Lehmann.

Table 1.

Classification of h alleles acording to protein expression. Adapted from Ingham et al. (1985b) 

Classification of h alleles acording to protein expression. Adapted from Ingham et al. (1985b)
Classification of h alleles acording to protein expression. Adapted from Ingham et al. (1985b)

Making the anti-h antibody

Fusion proteins

Full-length h protein-coding sequence was excised as a 1.25 kb BviElI(blunted)-SalI fragment. (The former enzyme cuts immediately after the initiator ATG - Rushlow et al. 1989). The h sequences were inserted between the BamHI (blunted) and SalI sites of pUR292 (Riither and Miiller-Hill, 1983). The resulting plasmid, βgh, encodes an IPTG-inducible lacZ-h fusion protein (Fig. 2A). T7h was a gift from Walter Gehring’s laboratory, and includes the same 6-coding sequences cloned into pAR3038 (Rosenberg et al. 1987).

Fig. 2.

Expression of bacterial h fusion protein. (A) Protein gel of h fusion proteins stained with Coomassie Blue. (B) Western blot of same gel, stained with affinity purified anti-h serum. Samples are whole-cell extracts from: (1) pUR292 (no insert); (2) βgh; (3) pAR3038 (no insert); (4) T7-h; (5) host bacteria (no plasmid); M protein markers.

Fig. 2.

Expression of bacterial h fusion protein. (A) Protein gel of h fusion proteins stained with Coomassie Blue. (B) Western blot of same gel, stained with affinity purified anti-h serum. Samples are whole-cell extracts from: (1) pUR292 (no insert); (2) βgh; (3) pAR3038 (no insert); (4) T7-h; (5) host bacteria (no plasmid); M protein markers.

Antibodies

The fusion protein was purified by standard methods (Carroll and Laughon, 1987; Harlow and Lane, 1989), and innoculated into rabbits. In summary, the rabbits were boosted at 3weekly intervals, and the antisera tested for activity by Westerns blots. Binding was detected immunohistochemically using HRP-coupled secondary antibody. The serum was first purified by absorption against lacZ-expressing bacterial extracts coupled to CNBr-activated Sepharose (Sigma). The antibody was affinity purified on a lacZ-h column and further absorbed against 10 –15h embryos at its final dilution of 1:5. Mouse anti-A antibody was generated against the T7-h fusion protein (S.M. Pinchin, unpublished results).

Immunohistochemistry on embryos

Embryos were collected and handled according to standard methods (Wieschaus and Niisslein-Volhard, 1986). Immunohistochemical detection used biotinylated secondary antibodies and peroxidase-ABC complexes (Vectastain), essentially as described by MacDonald and Struhl (1986), except that the final staining with diamino-benzidine (DAB) was performed at pH 5.5 (Frasch, personal communication). Embryos were dehydrated in 95 % EtOH and mounted in JB-4 methacrylate (Polysciences), and viewed under Nomarski illumination. Staining patterns were analysed by comparing the positions and dimensions of the stripes with the wild-type pattern.

The h/eve double-labelling used mouse anti-li (gift of S.M. Pinchin) and rabbit anti-eve (Frasch et al. 1987). Embryos were stained with a mixture of the two antisera. Thereafter, the proteins were visualised using a mixture of enzyme-linked secondary antibodies (Jackson Labs), alkaline phosphatase for one protein and peroxidase for the other. Embryos were stained for the former with bromo-chloro-indolyl phosphate and nitro-blue tétrazolium substrates (BRL), rinsed well, and then stained with DAB. Embryos were mounted in PBS containing 0.1% Tween 20, and viewed under Nomarski illumination.

Antibodies to h fusion proteins

DNA sequencing of h cDNAs has defined a single long translational reading-frame encoding a putative h protein (Rushlow et al. 1989). We placed these proteincoding sequences downstream of a β-galactosidase (lacZ) gene and expressed the fusion protein in E. coli under the control of an IPTG-inducible promoter (Fig. 2A; see Materials and methods). This construct (βgh) directs the high-level expression of a major ∼160 ×103Mr protein, the size expected for a lacZ-h fusion (Fig. 2B). Gel-purified fusion protein was used to generate a rabbit antiserum that reacts specifically against h protein (see Materials and Methods). Western blots show that the adsorbed serum recognises the lacZ fusion protein, as well as several other bacterial h fusion proteins. Fig. 2C shows that the antiserum is A-specific, staining both lacZ-h and T7-h, a fusion protein in which h-coding sequences are fused to 13 aminoacids of the T7 gene 10 protein.

We confirmed that the antiserum recognises h protein in fixed Drosophila embryos by staining late blastoderm embryos in which h is being actively transcribed. Immunohistochemical staining shows a pattern of 7 stripes plus an anterodorsal head-patch (‘stripe’ 0) (Fig. 3A). Embryos homozygous for Df(3L)hi22, a deficiency that lacks the h coding region (Ish-Horowicz et al. 1985), do not stain, demonstrating that the staining is specific for h protein (Fig. 3B). 25% (30/ 117) of late blastoderm embryos in a Df(3L)hi22/TM3 stock fail to stain for h.

Fig. 3.

Specific h protein staining in Drosophila embryos. (A) Blastoderm embryo stained for h protein expression, h domains (0 –7). (B) Lack of staining in a Df(3L)hi22 embryo. (C) Lack of ventral staining in stripe 1. (D) h protein localises in the nucleus. Anterior dorsal is top left.

Fig. 3.

Specific h protein staining in Drosophila embryos. (A) Blastoderm embryo stained for h protein expression, h domains (0 –7). (B) Lack of staining in a Df(3L)hi22 embryo. (C) Lack of ventral staining in stripe 1. (D) h protein localises in the nucleus. Anterior dorsal is top left.

The pattern of h protein accumulation resembles that of h transcripts (Ingham et al. 1985a); the only departure is the lack of h staining in the ventralmost 8 to 10 cells of stripe 1 (Fig. 3C). Such incomplete stripes would be difficult to detect by in situ hybridisation of sectioned material. Carroll et al. (1988) suggested that only 10 % of embryos lack h staining in ventral stripe 1. We find that all embryos lack this ventral staining, but that it is only evident in embryos orientated such that the ventral surface is exposed (Fig. 3C). Such ventral repression resembles the dorsal restriction of h expression in the head-patch, and indicates that dorsoventral control of h expression extends more posteriorly than previously expected. Indeed, eve appears to be subject to a similar repression (see below).

Fig. 3C,D shows that h staining is predominantly nuclear. This localisation is consistent with h′s proposed role as a transcriptional regulator.

Wild-type h expression

Blastoderm and extending germ band

h protein stripes are first seen early in cleavage cycle 14, at about the time that the nuclei begin to elongate. We do not detect uniform h protein at earlier stages when h transcripts accumulate uniformly throughout the embryo (Ingham et al. 1985a), although such low-level h expression might be difficult to visualise. Alternatively, there may be translational control of h expression prior to stage 14.

The first h stripes to appear are 1,2,3 and 7. Stripes 4 and 6 are fused to 3 and 7 respectively when first detectable (Fig. 4A). By mid-cleavage stage 14 (Foe and Alberts, 1983), the striped h pattern is well-established (Fig. 4B). Stripe 4 remains weaker than the others, corresponding to the domain most sensitive to loss of h activity. Embryos heterozygous for amorphic h alleles often show segment defects at A4/A5 (Ingham et al. 19856).

Fig. 4.

Expression of h protein during blastoderm and germ band extension. (A) Early interphase 14; (B) midinterphase 14; (C) early gastrula (stage 6 - Campos-Ortega and Hartenstein, 1985), cephalic furrow (cf); (D) early stage 8 - end of germ band extension, vestiges of h stripes remain in the mesoderm (arrow).

Fig. 4.

Expression of h protein during blastoderm and germ band extension. (A) Early interphase 14; (B) midinterphase 14; (C) early gastrula (stage 6 - Campos-Ortega and Hartenstein, 1985), cephalic furrow (cf); (D) early stage 8 - end of germ band extension, vestiges of h stripes remain in the mesoderm (arrow).

The h stripes are not of equal widths (Figs4B, 9). Weaker h expression in stripe 4 is reflected in its being the narrowest stripe. Conversely, the anterior two stripes are consistently 1 –2 cells broader than stripes 3, 5, 6 and 7, which are each approximately 3 –4 cells wide, i.e. about the width of a segment primordium. Once mature, the h stripes are surprisingly stable. There are no dramatic differences in stripe widths or staining intensities within a stripe, nor do their widths change significantly during the blastoderm stage. This suggests that their regulation differs significantly from that of ftz or eve whose patterns of expression are dynamic (Carroll and Scott, 1985; Frasch et al. 1987).

The h stripes decay at the onset of gastrulation and germ band extension (Fig, 4C). Within 30 –40min, ectodermal h stripes are no longer detectable, although faint periodic staining in the mesoderm persists until the end of germ band extension (Fig. 4D; 60 –90 min postgastrulation; stages 9/10, Campos-Ortega and Hartenstein, 1985). The disappearance of h protein parallels the decay of h transcripts and indicates that both are very unstable, thereby resembling the ftz gene products (Edgar et al. 1986).

The anterodorsal h head-patch lies between about 80 and 95% egg length (EL), and is separated from the most anterior stripe by 6 –8 unstained cells (Fig. 4B). It is about 12 cells long and 15 cells wide. Unlike the other h stripes, h staining in the head persists during germ band elongation, lengthening to 19 cells while narrowing to 2 –4 cells wide (stages 7/8 Campos-Ortega and Hartenstein, 1985). By stage 9, the patch has migrated to the clypeolabrum, including the primordium of the labral tooth, a structure deleted in strong h mutant embryos (Fig. 5A; Ingham et al. 1985b).

Fig. 5.

Later h expression. (A) Stage 9 - staining of the clypeolabrum (cl) and proctodæum (p); (B and C) stage 10 - staining of the stomodæum (s) and the tracheal pit (tp) primordia (although present, the stomodæal staining is out of the plane of focus in B); (D) stage 12 - mesodermal (m) staining; (E) stage 15 - staining of hind gut (hg) and anal pads (ap).

Fig. 5.

Later h expression. (A) Stage 9 - staining of the clypeolabrum (cl) and proctodæum (p); (B and C) stage 10 - staining of the stomodæum (s) and the tracheal pit (tp) primordia (although present, the stomodæal staining is out of the plane of focus in B); (D) stage 12 - mesodermal (m) staining; (E) stage 15 - staining of hind gut (hg) and anal pads (ap).

Stages 9 to 16

h protein expression in the proctodæum, the posterior part of the posterior midgut invagination, commences at late stage 9 (Fig. 5A) and continues until about stage 16 (about 16 h). It is unclear whether this h expression has any functional significance.

After the completion of germ band extension (4.5 –5 h; stage 10), h expression arises in the stomo-dæum and the tracheal pit primordia (Fig. 5B,C). The latter sites appear as a series of 10 segmentally repeated clusters of about 4 to 5 h-staining cells, before cells begin to invaginate. Again, the staining is nuclear. The number of stained cells initially increases to 12 (Fig. 5C), but decreases to ∼8 as invagination proceeds. We do not see any internal stained cells under Nomarski illumination, suggesting that internalised cells may cease expressing h. Tracheal h expression ceases by the onset of germ band retraction, before the pits are closed, and before generalised mesodermal staining commences. There is no staining of the 11th invagination that is thought not to contribute to the tracheal system (Campos-Ortega and Hartenstein, 1985).

Germ band retraction to hatching

During germ band retraction (7.5 h, stage 11), h expression reappears throughout the mesoderm, extending throughout both visceral and somatic layers (Fig. 5D). This staining persists until about stage 14 (10h 20min). At about this time, the hind-gut and the anal plates express h (Fig. 5E) and continue to do so until stage 16 (16 h), the latest at which phase-partition fixation is effective. We cannot exclude weak h expression in other tissues at these late stages as the multiple cell-layers cause increased background that would mask weak signals, although no such staining is evident using Nomarski optics.

In contrast to many other segmentation genes, h expression is not detected in the embryonic central nervous system. However, the later h bristle phenotype shows that h is involved later in the adult peripheral nervous system. The above patterns are consistent with the patterns of reporter gene expression driven by the h promoter. A β-galactosidase gene inserted into the h locus shows similar patterns of expression to those described above (Fasano et al. 1988).

h expression in h mutant alleles

Only one h allele, Df(3L)hi22, lacks the protein-coding sequences (Ish-Horowicz et al. 1985). We have examined h expression in the other alleles and have grouped them into 4 classes according to the degree and pattern of protein production (Table 1).

Class A - no protein

Apart from Df(3L)hi22, only two alleles, hIL79K (Fig. 6A) and h7H9K (not shown; Carroll et al. 1988) accumulate no detectable h protein. Both are EMS-induced mutations that show no obvious DNA lesions (Ish-Horowicz et al. 1985; Howard, 1986). Nevertheless, 25% of the embryos from the hIL79K and h7H94 stocks fail to stain for h at any stage, further confirming the specificity of the antiserum, and indicating that these alleles do not make antigenic h protein. The transcription pattern in hIL79K embryos appears normal (Howard and Ingham, personal communication), suggesting that this mutation blocks or truncates h translation.

Fig. 6.

Patterns of h protein expression in h mutant alleles. (A) hIL79K- no staining. Reduced levels of expression in stripes 2 to 7 in h1 embryos (B) and h5HO7 (C). (D) h8K11S showing fused bands 3 and 4.

Fig. 6.

Patterns of h protein expression in h mutant alleles. (A) hIL79K- no staining. Reduced levels of expression in stripes 2 to 7 in h1 embryos (B) and h5HO7 (C). (D) h8K11S showing fused bands 3 and 4.

Class B - reduced levels of expression

One h allele is distinguished by affecting the levels but not the patterns of some stripes, h1 is a homozygous-viable allele showing only a minimal segmentation phenotype that is associated with the insertion of a 7 kb gypsy transposable element about 5 kb upstream of the h promoter (Holmgren, 1984; M. Lardelli, personal communication). We find that h protein expression is qualitatively normal in h1 embryos, but the levels of h protein are reduced in stripes 2 to 7 (Fig. 6B). This pattern is not due to enhanced h expression in stripe 1 and the head-patch, as shown by staining mixtures of h1 and + embryos (not shown). Reduced levels of h expression are consistent with h1 being semilethal and showing considerable pattern defects when heterozygous with strong h alleles (Ingham et al. 1985b).

Class C - inactive protein

The majority of EMS-induced h alleles show apparently normal staining patterns such that we are unable to distinguish the homozygous embryos. These alleles show strong pair-rule cuticular phenotypes, indicating that they encode nonfunctional h protein (Table 1). Indeed, we have shown that the h5H07 allele is unable to repress ftz expression, leading to broadened ftz stripes similar to those of Df(3L)ht27and hK1 (not shown).

We have confirmed that such amorphic alleles show a normal pattern of h expression by analysing alleles that were induced on a h1 containing chromosome, e.g. h5H07 (Table 1). Despite the lack of functional h protein in h5H07 embryos, h pattern is indistinguishable from that in h1 embryos, showing a normal head-patch and stripe 1, but weaker stripes 2 to 7 (Fig. 6B,C). This indicates that normal h patterning occurs in the absence of active h protein. In particular, levels of h expression are independent of h function, i.e. h is not subject to autoregulation.

This is also supported by analysis of h8K115, which encodes a chromosomal deletion of about 250 bp that includes the carboxy-terminal protein-coding region (Ish-Horowicz et al. 1985; S.M. Wainwright, unpublished observations). The strong pair-rule cuticular phenotype indicates that the deletion inactivates the h product (Table 1). h8K115 leads to ectopic h expression between stripes 3 and 4, i.e. homozygous h8K115 embryos show fused stripes 3 and 4 (Fig. 6D). Individual stripes are resolved by the onset of gastrulation, indicating that their resolution is retarded, not prevented, h expression appears otherwise normal, despite lack of h function.

Class D - loss of stripes

The final class of mutation affects the expression of specific h stripes and comprises hm13, hm7, hm8, and hK1. These alleles are all rearranged upstream of the h promoter and can be ranked according to the severity of their pattern defects. In hm3, stripes 3 and 4 are missing (Fig. 7A). In hm7, stripes 6 and 7 are also lacking, and stripe 2 reduced in intensity (Fig. 7B). hm8 retains the head-patch and stripes 1 and 5, and hK1 expresses only the head-patch (Fig. 7C,D). These patterns of protein expression agree with previous analysis of h transcription in these alleles (Howard et al. 1988). The local effects on h pattern are paralleled by the cuticle defects in homozygous embryos. Thus, hm3 shows defects between T1 and A2, corresponding to the missing stripes 3 and 4, and other alleles show correspondingly more severe defects (Ingham et al. 1985b; Howard et al. 1988).

Fig. 7.

h staining in mutations that cause loss of specific stripes. (A) lack of stripes 3 and 4 in hm3 ; (B) stripes 0, 1, 2 and 5 in hm7 (C) stripes 0, 1 and 5 in hm8; (D) headpatch staining in hKl .

Fig. 7.

h staining in mutations that cause loss of specific stripes. (A) lack of stripes 3 and 4 in hm3 ; (B) stripes 0, 1, 2 and 5 in hm7 (C) stripes 0, 1 and 5 in hm8; (D) headpatch staining in hKl .

Significantly, the length of retained h upstream sequences correlates with the severity of phenotype. More extensive h deletions cause progressive loss of stripes, suggesting that the promoter region includes m-controlling elements associated with expression of individual or specific groups of stripes (Howard et al. 1988). hm3 is broken ∼10 kb upstream of the h promoter and loses stripes 3 and 4 (Howard et al. 1988; M. Lardelli, personal communication), suggesting that this upstream region includes sequences specifically required for these stripes. hm7 and hm8 delete a further 0.5 –4.5 kb and also affect stripes 2, 6 and 7. Finally, hKI retains only ∼3kb of 5 ′ sequence and loses all h domains except the head-patch. These results indicate that different h stripes are responding to different spatial cues.

h is not required for tracheal pit invagination

Adthough several tissues stain for h after the blastoderm stage, it is unclear whether expression in such regions is functional. The precise h expression in tracheal primordia led us to enquire whether tracheal pit invagination requires h function. The extreme pattern disruptions caused by the strongest h alleles make it difficult to identify tracheal primordia, so we examined embryos homozygous for h8K115, an allele that makes detectable but non-functional h protein (see above). Such embryos should display only 5 pit primordia due to pair-rule deletions halving the number of metameres.

As expected, there are only 5 groups of A-staining cells in h8K115 embryos at the elongated germ band stage (Fig. 8D). Each cluster includes a pit invagination that can be seen under Nomarski illumination (not shown) and by scanning electron microscopy (SEM) (Fig. 8C).

Fig. 8.

Tracheal invaginations in h mutations. (A-C) SEMs of wild-type (A), hIL79K (B), and h8K115 (C) embryos. (D) h stained pit primordia in h88115.

Fig. 8.

Tracheal invaginations in h mutations. (A-C) SEMs of wild-type (A), hIL79K (B), and h8K115 (C) embryos. (D) h stained pit primordia in h88115.

This is not peculiar to h8KIIS, because other strong h alleles including hI4H107, have 5 stained primordia that are able to form invaginations (not shown). Thus, h function appears not to be required for this aspect of tracheal development.

The above is consistent with tracheal pit formation in h,IL79K embryos, which make no h protein (see above). Such embryos are usually severely disrupted, but some show a regular pair-rule pattern of tracheal primordia (Ingham et al. 1985a). In such embryos, 5 pits can be clearly seen (see also Fig. 8B), confirming that tracheal invagination does not require h activity.

h stripes overlap but lie anterior to the eve stripes

Combinatorial function of pair-rule genes requires that different stripes show overlapping registers (see Introduction). We therefore related the stripe domains of h to those of even-skipped (eve), a pair-rule gene whose domains are reciprocal to those of ftz (Harding et al. 1986; MacDonald et al. 1986; Frasch and Levine, 1987; Frasch et al. 1988). Overlap of h and ftz domains would predict that the h domains should lie anterior to the eve stripes (see Fig. 1). This should be readily visible, as eve anterior boundaries remain sharply defined throughout late blastoderm (Lawrence et al. 1987).

Fig. 9 shows that the h domains consistently lie anterior to the eve domains. We used immunohistochemical staining with antibodies against h and eve (Frasch et al. 1987) to distinguish cells expressing h, eve and h +eve (see Materials and methods). The embryos in Fig. 9A and 9B display h-expressing cells in blue, eve cells in brown, and h +eve cells as dark grey. The embryo in Fig. 9C is stained reciprocally so that h cells are brown and eve cells blue. Each h stripe overlaps and extends anterior to an eve stripe, indicating that the anterior h and eve boundaries are displaced throughout the embryo. Unexpectedly, eve staining of stripe 1 is incomplete, like that of h; ventral eve stripe 1 is largely lacking, although a few weakly-staining posterior cells remain (not shown; M. Frasch, personal communication). Thus, both h and eve expression must be under dorso ventral control.

Fig. 9.

Double-label visualisation of h and eve domains. h/eve double staining using mouse anti-h and rabbit anti-eve. Dorsal (A) and lateral (B) views of stripes 1 to 7: A visualised with alkaline phosphatase (blue), eve with peroxidase (brown); h +eve cells stain browny-grey. (C) lateral view: h cells stained with peroxidase (brown), and eve cells with alkaline phosphatase (blue). Not all stripes are completely in focus due to the curved nature of the embryo.

Fig. 9.

Double-label visualisation of h and eve domains. h/eve double staining using mouse anti-h and rabbit anti-eve. Dorsal (A) and lateral (B) views of stripes 1 to 7: A visualised with alkaline phosphatase (blue), eve with peroxidase (brown); h +eve cells stain browny-grey. (C) lateral view: h cells stained with peroxidase (brown), and eve cells with alkaline phosphatase (blue). Not all stripes are completely in focus due to the curved nature of the embryo.

The h/eve relationships at the level of individual cells are most easily analysed in embryos that show the strongest h staining, towards the end of cellularisation. h stripes 3 to 7 lie one cell anterior to the corresponding eve domains. This is most readily seen for stripe 3 where the staggered line of h-expressing cells is generally one cell wide (Fig. 9A,B), but a similar relationship also holds for the more posterior stripes. The line of h cells can be incomplete where h staining is weaker, especially for stripes 4 and 5. Such variable staining is most likely due to reduced h expression in these regions although we cannot exclude the possibility of a less rigorous h/eve phase relationship.

As expected, the eve domains extend posterior to those of h, although the boundary is less distinct because of the lower levels of eve expression at the posterior of each stripe (Fig. 9). Nevertheless, most stripes show a posterior line of cells expressing eve but not h. In general, the eve domains extend one cell posterior to those of h.

In contrast, the terminal h and eve stripes are differently organised, reflecting broader h and eve bands. Thus, the two anterior h stripes are about 1 cell broader, resulting in 2 anterior h-expressing cells that do not express eve (Fig. 9). eve stripe 7 is also broader, causing it to extend 2 cells posterior to the seventh h stripe (Fig. 9A,C). Thus, all h/eve stripes show similar h +eve overlaps, perhaps suggestive of some functional significance in subsequent development. However, no segment-polarity gene that is specifically expressed in these cells is currently known.

h expression and other segmentation genes

We have analysed the effects of maternal and zygotic segmentation genes on h expression. In general, the results conform to the expectations of current hierarchical models: h striping is affected in domains corresponding to those in which cuticular pattern is affected. Nevertheless, it is impossible to define whether the pattern alterations reflect direct or indirect action on h expression. Moreover, the numbering of mutant stripes is necessarily ambiguous. Abnormal stripes cannot be assigned a definitive identity as we cannot ascertain the regulatory gene products to which they are responding. Hence we shall designate such stripes using a ‘prime’ numbering (e.g. 7 ′).

Maternal genes

h expression requires the action of the maternal segmentation genes: h transcription is altered in embryos derived from bicaudal and dorsal mothers (Ingham et al. 1985a). We have extended these results by analysing embryos from mothers mutant for maternal genes that are specifically required in the establishment of anteroposterior polarity (Nüsslein-Volhard et al. 1987). [For simplicity, we refer to the defective embryos according to the genotypes of the mutant mothers from which they derive.]

bicoid

bicoid (bed) acts to establish the anterior embryonic fate-map; bed embryos lack head and anterior thoracic structures (Frohnhôfer and Nüsslein-Volhard, 1986; Driever and Nüsslein-Volhard, 1988). Staining such embryos with h antibody shows that only 4 h stripes are retained (Fig. 10A). As posterior structures arise normally in bed embryos, the posterior 3 stripes probably reflect normal stripes 5 to 7 that are shifted somewhat anteriorly. However, the anterior-most stripe is stronger than would be expected for stripe 4. More likely, it represents a duplicated stripe 7 (stripe 7 ′) corresponding to the duplicated posterior structures that develop at the anterior of bed embryos (Frohnhôfer and Nüsslein-Volhard, 1986). Duplication of the posterior ftz stripe also occur in bed embryos (Frohnhôfer and Nüsslein-Volhard, 1987).

Fig. 10.

h protein patterns in segmentation gene mutant embryos. (A-C) Maternal mutations: embryos from mothers mutant for bcdEl (A); osk166 (B); torWK (C) (the altered spacing between stripes 3 and 4 is not typical in such embryos). (D-K) Zygotic mutations: (D) hbP ×T15; (E) Kr1;(F)KniIID48;(G)tllL10; (H)kni,L26he7M48, (I)kniIL26Phb7M48tllL10; (J) run112 and (K) eve1 27 embryos.

Fig. 10.

h protein patterns in segmentation gene mutant embryos. (A-C) Maternal mutations: embryos from mothers mutant for bcdEl (A); osk166 (B); torWK (C) (the altered spacing between stripes 3 and 4 is not typical in such embryos). (D-K) Zygotic mutations: (D) hbP ×T15; (E) Kr1;(F)KniIID48;(G)tllL10; (H)kni,L26he7M48, (I)kniIL26Phb7M48tllL10; (J) run112 and (K) eve1 27 embryos.

oskar

The oskar (osk) gene is needed for the establishment of posterior pattern, osk embryos lacking the active nanos (nos) component that defines abdominal segmentation (Lehmann and Nüsslein-Volhard, 1986; Lehmann, 1988; Sander and Lehmann, 1988). The nos product is needed to suppress abdominal hb expression and thereby permit that of kni (Hülskamp et al. 1989; Irish et al. 1989; Struhl, 1989). Fig. 10B shows that stripes 0 –3 are retained in osk embryos, although stripe 3 appears broadened and the interstripe 2/3 region is narrowed. Stripes 4 –7, corresponding to the affected domain, are replaced by a broadened stripe in the position of stripes 6 and 7.

torso

torso (tor) is required for the establishment of pattern in the embryo termini (Nüsslein-Volhard et al. 1987; Klingler et al. 1988). Its action is mediated by the action of the zygotic gap gene tailless (til) (see below). 7 domains of h expression are detectable in tor embryos, although their positions are substantially altered from wild-type (Fig. 10C). A patch of h expression at the extreme anterior tip may reflect domain 0′. The staining persists during gastrulation and early germ band extension (not shown). Similarly there is weak staining at the extreme egg posterior (except for the pole-cells), perhaps reflecting a shifted domain 7 ′ or 8 ′. Stripes 1 ′ –5 ′ correspond roughly to their wild-type equivalents, but 6 ′ is broadened and occasionally has the appearance of a double stripe, as though it represents a 6/7 fusion.

Gap genes

We have confirmed previous evidence that h patterning requires the action of the gap genes by examining protein pattern in gap mutant embryos (Ingham et al. 1986; Ingham and Gergen, 1988; Carroll et al. 1988). In all cases, we can interpret the altered pattern in terms of disrupted h striping within a region of the blastoderm corresponding the gap gene expression domains. However, the cross-regulatory interactions between gap genes (Jackie et al. 1986) prevent our distinguishing between direct and indirect interactions.

hunchback

hb is expressed zygotically in a broad anterior domain covering h stripes 1 to 2, and a narrower posterior stripe that corresponds roughly to h domain 7 (Tautz et al. 1987; Tautz, 1988). In embryos homozygous for hbPXTI5, stripe 0 ′ is retained, h stripe 1 ′ is narrower, stripe 2 is lacking, and a broad stripe covers the stripe 3 –4 ′ region (Fig. 10D). The posterior 3 stripes are broader than wild-type, but spaced normally. The anterior h stripes are not maintained so that only the 3 posterior stripes remain by the onset of germ band extension (not shown).

Krilppel

Kr is expressed in the central domain of the embryo, covering the region of h stripe 2 to 4 (Gaul et al. 1987). In Kr1 embryos, the lack of Kr protein results in a single broad h stripe covering approximately the stripes [2 –3 –4] ′ area, and another covering the stripe [5 –6] ′ area (Fig. 10E). The head patch and stripes 1 and 7 are unaffected. These results are consistent with the pattern of h transcription in Kr embryos (Ingham et al. 1986). Since Kr does not affect the terminal regions of the embryo, one may assume that the stripes seen in the positions of the head patch, stripes 1 and 7, represent h stripes 0, 1 and 7 respectively. However, there may be an aberrant response of h to an abnormal combination of regulatory gene products, caused by the lack of Kr expression. Alternatively, the broadened stripes such as that seen in the 2 –3 –4 ′ region of Kr mutant embryos may represent fusions of more than one stripe, that is, failure to repress h in the interstripes.

knirps

The kni cuticular phenotype shows that it is required in segments Al to A7 (Nüsslein-Volhard and Wieschaus, 1980; Lehmann, 1985). h stripes 0 to 3 appear unaffected in kniIID48 embryos, but the posterior pattern is severely disrupted. There is a loss of h expression in the 4 ′ region, an often weak stripe 5 ′ remains and a single, broad stripe in the 6 –7 ′ region (Fig. 10F). An alternative description would suggest that stripe 4 ′ is shifted, and stripes 5 –7 ′ are fused. Both interpretations are consistent with the kni cuticular phenotype and with the pattern of h transcription in kni mutant embryos (Ingham and Gergen, 1988), confirming that kni is necessary for h patterning in parasegments 7 to 10. The differing patterns between kni and osk embryos may be due to the effects of osk on hb expression (Hülskamp et al. 1989; Irish et al. 1989; Struhl, 1989).

tailless

The gap gene til has yet to be fully characterised but it affects structures derived from the acron and tail region (Jürgens et al. 1984; Strecker et al. 1986; Klingler et al. 1988). In embryos homozygous for the tllL,a mutation, the h pattern is only affected at the posterior end, showing a broad domain in the 6 –7 ′ region plus a faint posterior stripe outside the area where h protein is normally expressed (Fig. 10G). The latter may be due to derepression of a terminal 8th stripe that is occasionally seen by in situ hybridisation (Ingham, 1988), but whose expression we do not detect in wild-type embryos. Mahoney and Lengyel (1987) report a similar pattern of h transcripts although they did not detect the posterior stripe. They suggested that til embryos might lack h stripe 7. In this case, the terminal stripe might represent low-level expression from a shifted stripe 7 primordium. The h head patch is retained in til.embryos although its length may be slightly reduced.

Gap gene combinations

We also examined h expression in kni hb double mutant embryos, and kni hb til triple mutant embryos (Fig. 10H,I). Interpretation is difficult due to an inability to assign stripe identities. Both embryos retain a head patch and a narrow stripe 1 ′ and a very weak stripe 2 ′ (not visible in the particular kni hb til embryo of Fig. 101). Both genotypes show a weak 5 ′ stripe and an adjacent broad posterior stripe. In kni hb til embryos, the latter extends much more posteriorly than in the single- or double-mutant embryos, suggesting that h is subject to interacting gap gene control in this region.

Pair-rule genes

We have tested the effects on h patterning of run and eve the other pair-rule genes that affect ftz striping: (Carroll and Scott, 1986). h pattern is indeed affected in embryos homozygous for Df(l)run1112, but in a surprisingly mild manner. Stripe 1 appears somewhat broader than wild-type, and stripes 3/4 and 6/7 are only partially resolved (Fig. 10J). This indicates that the altered pattern results from partial failure to repress h expression in specific interstripes. The equivalent pattern of h transcripts shows fusion of stripes 3 and 4, and of 5, 6 and 7, although stripe 5 is ultimately refined (Ingham and Gergen, 1988). h expression may also be translationally controlled as we never see h protein expression between stripes 5 and 6.

eve is also required for h patterning. In embryos homozygous for Df(2R)eve127, h stripe 2 is greatly reduced or missing (Fig. 10K). This protein pattern thus resembles that of the h transcript, although we see no evidence of the premature decay of h expression described by Ingham and Gergen, 1988. The positions and intensities of the other stripes appear normal except that stripe 4 is of equal intensity to the other stripes and displaced towards stripe 3. Although eve may be involved in repressing stripe 4, we find that most non-mutant embryos in the eve127 stock also show a strong stripe 4 (not shown). Derepression of stripe 4 may be due to heterozygosity for eve or to an unlinked mutation on the eve chromosome.

Previous studies have implicated h in regulating the expression of other pair-rule segmentation genes (Howard and Ingham, 1986; Frasch and Levine, 1987; Ingham and Gergen, 1988). The nuclear localisation of h protein suggests that this role is exercised by regulating transcription, consistent with the repression of ftz transcription by ectopic h expression (Ish-Horowicz and Pinchin, 1987). We have recently shown that the h protein has protein homology to a region of the protooncogene N-myc that itself shows homology to transcriptional regulators (Rushlow et al. 1989). The h protein shares a sequence motif that may play a role in DNA binding and protein dimérisation (Murre et al. 1989). This domain is also found in other Drosophila proteins: daughterless, twist and the AS-C T4 and T5 transcripts (Villares and Cabrera, 1987; Caudy et al. 1988).

h affects the expression of other segmentation genes in addition to ftz. Thus, induction of ectopic h expression in early blastoderm stage embryos causes more severe pattern defects than pair-rule, presumably due to disturbance of other pair-rule genes (Ish-Horowicz and Pinchin, 1987). Moreover, run and eve expression are altered in h mutant embryos (Frasch and Levine, 1987; Ingham and Gergen, 1988), pattern alterations that are unlikely to be due to ftz.

Control of h striping/transcription

Evolution of the initially uniform pattern of h expression into stripes requires the enhancement of h expression within each stripe domain and the repression of interstripe h expression. Patterns of protein and transcript accumulation appear similar in both wildtype and mutant embryos, indicating that control is excercised predominantly at the level of transcription (this paper; Ingham et al. 1986; Carroll et al. 1988; Ingham and Gergen, 1988). h activation is not under autocatalytic control, unlike the ftz gene whose promoter includes a ftz-dependent transcriptional enhancer (Hiromi and Gehring, 1987). A wide range of h alleles encoding apparently amorphic proteins show normal levels of expression, and strong h alleles that were induced on a hf chromosome show /i1-like expression patterns (Fig. 6B,C). Moreover, the hIL79K allele that makes no h protein shows an approximately normal pattern of h transcription (Howard and Ingham, personal communication), and /kgalactosidase expression driven by the endogenous h promoter appears not to require h activity (Fasano et al. 1988). The above results exclude striping models such as those based on reaction-diffusion mechanisms requiring short-range auto-activation (Meinhardt, 1982).

h striping is not only /i-independent, but surprisingly independent of run and eve. eve is needed for maintenance of h stripe 2, but the spatial domain of this stripe must be defined by other factors, as some eve embryos still form a weak second h stripe, eve is not acting solely to define a h domain, because it is also required for run, ftz and eve stripes in this region (Frasch and Levine, 1987; Ingham and Gergen, 1988).

Lack of run product interferes with h repression between stripes 3 and 4, and 6 and 7. The stripes are partially resolved by the end of blastoderm, indicating that other factors play a role in repressing interstripe h expression. The somewhat different accumulation patterns of h protein and transcripts in run embryos may indicate translational control of h expression.

The effects of deleting upstream h sequences indicates that interstripe activation is mediated through cis- regulatory elements that respond to specific spatial cues (Fig. 7; Howard et al. 1988). These rearrangements do not reveal whether groups of stripes are co-regulated or if all stripes are independently controlled, but indirect evidence suggests that the latter may be the case. Although hm3 loses both stripes 3 and 4, only the latter is lost in kni embryos (Fig. 10F), as expected from the kni cuticular phenotype in which PS5 is unaffected. Moreover, stripes 3 and 4 show very different levels of expression in wild-type embryos throughout blastoderm (Fig. 4A,B). Similarly, the initial intensities of stripes 6 and 7 differ, indicating that they too are likely to be independent. Of course neighbouring stripes may still share common control factors.

The reduced expression of stripes 2 to 7 in h1 distinguishes elements controlling stripes 1 and 5, and suggests that the sequences controlling the former lie within 5 kb of the promoter, proximal to the site of gypsy insertion. Such a localisation is consistent with the retention of stripe 1 in hm8 . Although the altered h pattern may be caused by ‘transcriptional interference’ (Parkhurst and Corees, 1986) from the gypsy insertion (which is weakly transcribed at this stage), we prefer the explanation that the elements controlling stripes 2 to 7 become less efficient by being displaced 7 kb distally.

In contrast to the strong evidence for positively acting control elements, negative elements that suppress interstripe h expression are more elusive. Chromosomal rearrangements causing h overexpression should be dominant-lethal, due to ftz-misregulation. The only h allele that displays ectopic expression, h8K115s, contains a deletion in the 3 ′-end of the gene that encodes an altered transcript and an inactive protein (S.M.Wainright, unpublished observations), h transcription in h8k115 embryos is similarly altered indicating that transcript accumulation is affected (Howard and Ingham, personal communication). We cannot tell whether the delayed resolution of stripes 3 and 4 is due to the loss of a transcriptional control element or because of a more stable h gene product.

The most likely candidates for the regionalised factors specifying the h domains are the gap genes, either acting alone, or in combination with the maternal genes. However, no simple code for how gap gene products might define h domains emerges. The various gap genes interact in defining their domains, so the lack of an individual gap gene leads to altered patterns of the others (Jackie et al. 1986; Gaul et al. (1987)), and gap mutations also lead to altered patterns of other pairrule genes that may contribute to the mutant fi-pattern.

The only gene that unambiguously affects h expression in stripes 0 and 1 is bed. These stripes lie within the domain of hb, a gap gene that is under bed control (Driever and Nüsslein-Volhard, 1989). Nevertheless, neither stripe is lost in hb embryos, or in embryos mutant for the other gap genes tested. Both stripes persist in osk or tor embryos, although the head patch domain is shifted anteriorly in the latter. This failure to implicate any other gene suggests that bed might acts directly in specifying these h domains. Indeed, such a possibility is supported by our analysis of the h upstream sequences (Rushlow et al. 1989). The DNA sequences upstream of the h promoter includes a strong homology to the consensus bed binding-site defined for hb activation (Driever and Nüsslein-Volhard, 1989). It maps within a region that is implicated in controlling h stripe 1 expression (M. Pankratz, personal communication; G. Riddihough, personal communication). Of course, other genes would have to define stripe domains, presumably by modulating bed action.

Posterior h expression is enhanced in kni hb til embryos, suggesting that these genes may act negatively in regulating h. However, the detailed interactions will be complex, osk embryos are characterised by abdominal hb expression and lack of abdominal kni expression (Tautz, 1988; Nauber et al. 1988). The non-uniform abdominal h expression in osk embryos indicates that other genes must be involved in regulating posterior h expression. Similarly, the altered h pattern is tor embryos is not simply explained by lack of til expression. The ability to assign precise stripe identities in mutant embryos is required before a more detailed discussion of regulatory mechanisms is possible.

The cues that establish h pattern in the terminal regions of the embryo may be rather different from those in the segmented body region. Hence, the different phasing of the anterior margins of the first two h/eve stripes and the broadened posterior eve stripe (Fig. 9). Moreover, the cryptic eighth h stripe revealed in til embryos and the extensive posterior h expression in kni hb til embryos suggest that h expression is normally repressed posterior to the 7th stripe (Fig. 101). Such polar repression has previously been noted for ftz expression (Hiromi et al. 1985; Edgar et al. 1986).

Overlapping pair-rule phasings

In contrast to the evidence presented by Carroll et al. (1988), we do not find that the h pattern is dynamic once the striped pattern is resolved. Our double-labelling experiments demonstrate precisely overlapping pairrule phasings in the Drosophila blastoderm (Figs 1,9). It is clear that accurate one-cell-wide spatial distinctions are evident at the blastoderm stage. Carroll et al. (1988) have presented evidence via antibody mixing experiments that the anterior two h and eve stripes are displaced. Their alternative interpretation of more posterior phasing may stem from the weak signals they observed, and the indirect method they used to detect expression.

As eve and ftz domains are reciprocal, the displaced h/eve phasing shows that h and ftz domains must overlap consistently (Fig. 1). Presumably the transient overlap seen by Carroll et al. 1988 is due to variable ftz expression in posterior cells that overlap h. Our results also confirm that the narrowing at blastoderm of the ftz (and eve) domains occurs at the posterior of each stripe, consistent with the posterior narrowing during gastrulation and germ band extension (Lawrence et al. 1987).

Such overlapping domains are strongly predicted by combinatorial models of pair-rule gene control of segment-polarity striping (Gergen et al. 1986; Ingham et al. 1988). For example, the one-cell-wide odd-numbered en stripes correspond to overlapping eve and paired domains, at least within the resolution of in situ hybridisation. The stripes of honeveoff cells correspond to those that express wg at gastrulation (Fig. 1), although we do not assert that their h and eve states directly define cell identities (or even wg expression). There is no current evidence that h directly regulates segment-polarity genes, although it can do so by regulating other pair-rule genes. The odd-numbered wg domains appear to arise from contraction of the ftz stripes, perhaps induced by h (Ingham et al. 1988). Other components must contribute to the narrowing of ftz stripes as it continues throughout germ band extension in the absence of h.

The lack of interstripe ftz expression can be explained by h repression. However, how does ftz expression persist in the posterior cells of each ftz stripe if they also express h? One plausible explanation is that ftz repression is a two-stage process: eve efficiently repressing ftz in the odd-numbered parasegments; h acting more slowly in the most-posterior ftz cell. Previous evidence suggests that eve can repress ftz (Frasch et al. 1988; Ish-Horowicz et al. 1989); the delayed suppression of ftz expression in the h-expressing cell need only imply that h and eve repress ftz by different mechanisms.

Redundancy and striping models

What kinds of striping mechanisms can explain the precision of h/eve phasing? One class of model argues that the gap genes lay down a coarse striped prepattern that is refined by mutual interactions between pair-rule genes. Such models are based on reciprocal interactions between complementary pair-rule genes, in particular between h and run (Ingham and Gergen, 1988), and/or between ftz and eve (Frasch and Levine, 1987). However, h striping appears to be surprisingly independent of other pair-rule genes. On the other hand, models in which h and eve domains are established exclusively by the gap genes require that cri-controlling stripe elements can sense positional distinctions accurate to the single cell. This is mechanistically very difficult to envisage, especially as h stripes would not be sharpened by autoregulation.

Our preferred explanation for why h pattern is relatively refractory to pair-rule mutation is that the control mechanisms for stripe formation are redundant. Such redundancy is suggested by the partial resolution of eve and ftz stripes in embryos mutant for both h and run (Ingham and Gergen, 1988). It also explains how h striping is /i-independent although h/run mutual inhibition would have predicted that lack of h should affect run expression and thereby feed back on h.

Such redundancy could operate at the level of transcription and/or translation. We expect such transcriptional control to be exercised through independent activation of several pair-rule genes, even if they refine their domains through mutual interactions.

One corollary of such redundancy is that a genetic hierarchy of pair-rule genes may be more apparent than real. Hierarchical models arise from non-reciprocal interactions beween pair-rule genes: for example, h is required for ftz patterning, but h pattern is apparently normal in ftz embryos. However, we do not find that h protein striping significantly precedes that of ftz; by the time the h stripe pattern is fully resolved, so is that of ftz- Nor does eve striping obviously precede that of ftz (Frasch et al. 1988). Perhaps, ftz striping is not merely a passive response to h and eve. Ranking genes according to their effects on other segmentation genes could be misleading, their true role in patterning being masked by redundant control mechanisms.

We should like to thank Manfred Frasch for anti-eve antibodies, Sheena Pinchin for mouse anti-/i, Janni Niisslein-Volhard and Ruth Lehmann for their gifts of stocks, and Ruth for her invaluable advice and instruction in Drosophila embryology. We acknowledge Chris Rushlow and Kathy Howe, for the unpublished sequence data used to make the antibody; Phil Ingham, Ken Howard, Michael Lardelli and Mark Wainwright for unpublished observations; and Alfonso Martinez-Arias for first encouraging us to use immunohistochemistry. We should also like to thank our lab colleagues for much necessary advice and encouragement during the course of this work. We should like to acknowledge David Lane and David Glover for providing resources and expertise during the early stages of the project, and the members of their labs at Imperial College, London, in particular Will Whitfield, for practical advice on making and handling antibodies.

Akam
,
M.
(
1987
).
The molecular basis for metameric pattern in the Drosophila embryo
.
Development
101
,
1
22
.
Botas
,
J.
,
Moscoso Del Prado
,
J
and
Garcia-Beludo
,
A.
(
1982
).
Gene dose titration analysis in the search for trans-regulatory genes in Drosophila
.
EMBO J
.
1
,
307
310
.
Campos-Ortega
,
J. A.
and
Hartenstein
,
V.
(
1985
).
The embryonic development of Drosophila melanogaster
. (
Berlin
:
Springer Verlag
).
Carroll
,
S. B.
and
Laughon
,
A.
(
1987
).
Production and purification of polyclonal antibodies to the foreign segment of )3gal fusion proteins
.
In DNA Cloning: A Practical Approach
(ed.
D.
Glover
) Vol
3
,
89
111
,
IRL Press
,
Oxford
.
Carroll
,
S. B.
,
Laughon
,
A.
and
Thalley
,
B. S.
(
1988
).
Expression, function and regulation of the hairy segmentation protein in the Drosophila embryo
.
Genes Devi
.
2
,
883
890
.
Carroll
,
S. B.
and
Scott
,
M. P.
(
1985
).
Localisation of the fushi tarazu protein during Drosophila embryogenesis
.
Cell
43
,
47
57
.
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
226
.
Caudy
,
M.
,
Vássin
,
H.
,
Brand
,
M.
,
Tuma
,
R.
,
Jan
,
L. Y.
and
Jan
,
Y. N.
(
1988
).
daughterless, a Drosophila gene essential for both neurogenesis and sex determination, has sequence similarity to myc and the achaete-scute complex
.
Cell
55
,
1061
1067
.
Driever
,
W.
and
Nüsslein-Volhard
,
C.
(
1988
).
A gradient of bicoid protein in Drosophila embryos
.
Cell
54
,
83
93
.
Driever
,
W.
and
Nüsslein-Volhard
,
C.
(
1989
).
The bicoid protein is a positive regulator of hunchback transcription in the early Drosophila embryo
.
Nature, Lond
.
337
,
138
143
.
Edgar
,
B. A.
,
Weir
,
M. P.
,
Schubiger
,
G.
and
Kornberg
,
T.
(
1986
).
Repression and turnover pattern fushi tarazu RNA in the early Drosophila embryo
.
Cell
47
,
747
754
.
Falk
,
R.
(
1963
).
A search for a gene control system in Drosophila
.
Am. Nat.
91
,
129
132
.
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
.
W. Roux Arch, devl Biol
.
197
,
507
512
.
Foe
,
V. A.
and
Alberts
,
B. M.
(
1983
).
Studies of nuclear and cytoplasmic behaviour during the five mitotic cycles that precede gastrulation in Drosophila embryogenesis
.
J. Cell Sci
.
61
,
31
70
.
Frasch
,
M.
,
Hoey
,
T.
,
Rushlow
,
C.
,
Doyle
,
H.
and
Levine
,
M.
(
1987
).
Characterisation and localisation of the even-skipped protein of Drosophila
.
EMBO J
.
6
,
749
759
.
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 & Devi
.
1
,
981
995
.
Frasch
,
M.
,
Warrior
,
R.
,
Tugwood
,
J.
and
Levine
,
M.
(
1988
).
Molecular analysis of even-skipped mutants in Drosophila development
.
Genes & Devi
.
2
,
1824
1838
.
Frohnhofer
,
H. G.
and
Nüsslein-Volhard
,
C.
(
1986
).
Organization of anterior pattern in the Drosophila embryo by the maternal gene bicoid
.
Nature, Lond
.
324
,
120
125
.
Frohnhofer
,
H. G.
and
Nüsslein-Volhard
,
C.
(
1987
).
Maternal genes required for the anterior localization of bicoid activity in the embryos of Drosophila
.
Genes and Devi
.
1
,
880
890
.
Gaul
,
U.
,
Seifert
,
E.
,
Schuh
,
R.
and
Jàckle
,
H.
(
1987
).
Analysis of Krilppel protein distribution during early Drosophila development reveals post-transcriptional regulation
.
Cell
50
,
639
647
.
Gergen
,
J. P.
,
Coulter
,
D.
and
Wieschaus
,
E.
(
1986
).
Segmental pattern and blastoderm cell identities. In Gametogenesis and the Early Embryo
(ed.
S.
Subtleny
). New York Liss.
Symp. Soc. Devi Biol
.
43
,
195
220
.
Harding
,
K.
,
Rushlow
,
C.
,
Doyle
,
H. J.
,
Hoey
,
T.
and
Levine
,
M.
(
1986
).
Cross regulatory interactions among pair-rule genes in Drosophila
.
Science
233
,
953
959
.
Harlow
,
E.
and
Lane
,
D. P.
(
1988
).
Antibodies: a Laboratory Manual
.
Cold Spring Harbour Laboratory Press
.
Hiromi
,
Y.
and
Gehring
,
W. J.
(
1987
).
Regulation and function of the Drosophila segmentation gene fiishi 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
.
Holmgren
,
R.
(
1984
).
Cloning sequences from the hairy locus of Drosophila
.
EMBO J
.
3
,
569
573
.
Hooper
,
K. L.
(
1989
).
Spatial expression and function of the hairy protein in Drosophila melanogaster
.
D. Phil, thesis, University of Oxford, England
.
Howard
,
K. R.
(
1986
).
Molecular genetics of the hairy locus of Drosophila melanogaster
.
Ph.D. Thesis. University of London
.
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
).
Regionspecific alleles of the Drosophila segmentation gene hairy
.
Genes & Devi
.
2
,
1037
1046
.
Hülskamp
,
M.
,
Schroder
,
C.
,
Pfeifle
,
C.
,
Jâckle
,
H.
and
Tautz
,
D.
(
1989
)
Posterior segmentation of the Drosophila embryo in the absence of a maternal posterior organizer gene
.
Nature, Lond
.
338
,
629
632
.
Ingham
,
P. W.
(
1988
).
The molecular genetics of embryonic pattern formation in Drosophila
.
Nature, Lond
.
335
,
25
34
.
Ingham
,
P. W.
,
Baker
,
N. E.
and
Martinez-Arias
,
A.
(
1988
).
Regulation of segment polarity genes in the Drosophila blastoderm by fushi tarazu and even-skipped
.
Nature, Lond
.
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
.
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, Lond
318
,
439
445
.
Ingham
,
P. W.
,
Ish-Horowicz
,
D.
and
Howard
,
K. R.
(
1986
).
Correlative changes in homeotic and segmentation gene expression in Krüppel mutant embryos of Drosophila
.
EMBO J
.
5
,
1659
1665
.
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
.
Irish
,
V.
,
Lehmann
,
R.
and
Akam
,
M.
(
1989
).
The Drosophila posterior-group gene nanos functions by repressing hunchback activity
.
Nature, Lond
.
338
,
646
648
.
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
.
Ish-Horowicz
,
D.
,
Pinchin
,
S. M.
,
Ingham
,
P. W.
and
Gyurkovics
,
H. G.
(
1989
).
Autocatalytic ftz activation and metameric instability induced by ectopic ftz expression
.
Cell
57
(in press).
Jâckle
,
H.
,
Tautz
,
D.
,
Schuh
,
R.
,
Seifert
,
E.
and
Lehmann
,
R.
(
1986
).
Cross-regulatory interactions amoung the gap genes of Drosophila
.
Nature, Lond
.
324
,
668
670
.
Jürgens
,
G.
,
Wieschaus
,
E.
,
Nüsslein-Volhard
,
C.
and
Kluding
,
H.
(
1984
).
Mutations affecting the pattern of the larval cuticle in Drosophila melanogaster. II Zygotic Loci on the third chromosome
.
Wilhelm Roux Arch. devl. Biol
.
193
,
283
295
.
Klingler
,
M.
,
Erdélyi
,
M.
,
Szabad
,
J.
and
Nüsslein-Volhard
,
C.
(
1988
).
Function of torso in determining the terminal anlagen of the Drosophila embryo
.
Nature, Lond
.
335
,
275
277
.
Knipple
,
D. C.
,
Seifert
,
E.
,
Rosenberg
,
U. B.
,
Preiss
,
A.
and
Jâckle
,
H.
(
1985
).
Spatial and temporal patterns of Kruppel gene expression in early Drosophila embryos
.
Nature, Lond
.
317
,
40
43
.
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, Lond
.
328
,
440
442
.
Lehmann
,
R.
(
1985
).
Regionsspezifische Segmentierungsmutanten bei Drosophila melanogaster Meigen
.
Dissertation. Tübingen
.
Lehmann
,
R.
(
1988
).
Phenotypic comparison between maternal and zygotic genes controlling the segmental pattern of the Drosophila embryo
.
Development
104
Supplement
,
17
27
.
Lehmann
,
R.
and
Nüsslein-Volhard
,
C.
(
1986
).
Abdominal segmentation, pole cell formation and embryonic polarity require the localised activity of oskar, a maternal gene in Drosophila
.
Cell
47
,
141
152
.
MacDonald
,
P. M.
,
Ingham
,
P. W.
and
Struhl
,
G.
(
1986
).
Isolation, structure and expression of even-skipped: a second pair-rule gene of Drosophila containing a homeobox
.
Cell
47
,
721
734
.
MacDonald
,
P. M.
and
Struhl
,
G.
(
1986
).
A molecular gradient in early Drosophila embryos and its role in specifying body pattern
.
Nature, Lond
.
324
,
537
545
.
Mahoney
,
P. A.
and
Lengyel
,
J. A.
(
1987
).
The zygotic segmentation mutant tailless alters the blastoderm fate map of the Drosophila embryo
.
Devi Biol
.
122
,
464
470
.
Martinez-Arias
,
A.
and
Lawrence
,
P. A.
(
1985
).
Parasegments and Compartments in the Drosophila embryo
.
Nature, Lond
.
313
,
639
642
.
Meinhardt
,
H.
(
1982
).
Models of Biological Pattern Formation
. (
Academic Press
;
London
).
Moscoso Del Prado
,
J.
and
Garcia-Bellido
,
A.
(
1984
).
Cell interactions in the generation of chaetae pattern in Drosophila
.
Wilhelm Roux Arch, devl Biol
.
193
,
246
251
.
Murre
,
C.
,
McCaw
,
P. S.
and
Baltimore
(
1989
).
A new DNA Binding and dimérisation motif in immunoglobulin enhancer binding, daughterless, MyoD, and myc proteins
.
Cell
56
,
777
783
.
Nauber
,
U.
,
Pankratz
,
M. J.
,
Kjenlin
,
A.
,
Seifert
,
E.
,
Klemm
,
U.
and
Jâckle
,
H.
(
1988
).
Abdominal segmentation of the Drosophila embryo requires a hormone receptor-like protein encoded by the gap gene knirps
.
Nature
336
,
489
492
.
Nüsslein-Volhard
,
C.
,
Frohnhofer
,
H. G.
and
Lehmann
,
R.
(
1987
).
Determination of anteroposterior polarity in Drosophila
.
Science
238
,
1675
1681
.
Nüsslein-Volhard
,
C.
and
Wieschaus
,
E.
(
1980
).
Mutations affecting segment number and polarity in Drosophila
.
Nature, Lond
.
287
,
795
801
.
Parkhurst
,
S. M.
and
Corces
,
V. G.
(
1986
).
Interactions among the gypsy transposable element and the yellow and suppressor of Hairy-wing loci in Drosophila melanogaster
.
Mol. Cell Biol
.
6
,
47
53
.
Rosenberg
,
A. H.
,
Lade
,
B. N.
,
Chui
,
D. S.
,
Lin
,
F. W.
,
Dunn
,
J. J.
and
Studier
,
F. W.
(
1987
).
Vectors for selective expression of cloned DNA by T7 RNA polymerase
.
Gene
56
,
125
135
.
Rushlow
,
C. A.
,
Hogan
,
A.
,
Pinchin
,
S. M.
,
Howe
,
K. R.
,
Lardelu
,
M. T.
and
Ish-Horowicz
,
D.
(
1989
).
The Drosophila h protein acts in both segmentation and bristle patterning and shows homology to N-myc
.
EMBO J. (in press)
.
Rüther
,
U.
and
Müller-Hill
,
B.
(
1983
).
Easy identification of cDNA clones
.
EMBO J
.
2
,
1791
1794
.
Sander
,
K.
and
Lehmann
,
R.
(
1988
).
Drosophila nurse cells produce a posterior signal required for embryonic segmentation and polarity
.
Nature, Lond
.
335
,
68
70
.
Strecker
,
T.
,
Kongsuwan
,
K.
,
Lengyel
,
J.
and
Merriam
,
J.
(
1986
).
The zygotic mutant tailless affects the anterior and posterior of the Drosophila embryo
.
Devl Biol
.
113
,
64
76
.
Struhl
,
G.
(
1989
).
Differing strategies for organising anterior and posterior body patterns in Drosophila enbryos
.
Nature, Lond
.
338
,
741
744
.
Tautz
,
D.
(
1988
).
Regulation of the Drosophila segmentation gene hunchback by two maternal morphogenetic centres
.
Nature, Lond
.
332
,
81
284
.
Tautz
,
D.
,
Lehmann
,
R.
,
Schürch
,
H.
,
Schuh
,
R.
,
Seifert
,
E.
,
Kienun
,
A.
,
Jones
,
K.
and
Jàckle
,
H.
(
1987
).
Finger protein of novel structure encoded by hunchback, a second member of the gap class of Drosophila segmentation genes
.
Nature, Lond
.
327
,
383
389
.
Villares
,
R.
and
Cabrera
,
C. V.
(
1987
).
The achaete-scute gene complex of D. melanogaster: conserved domains in a subset of genes required for neurogenesis and their homology to myc
.
Cell
50
,
415
424
.
Wieschaus
,
E.
and
Nüsslein-Volhard
,
C.
(
1986
).
Looking at embryos
.
In Drosophila, a Practical Approach
(ed.
D. B.
Roberts
)
IRL Press
,
Oxford
, pp.
199
226
.