Graded requirement for the zygotic terminal gene, tailless, in the brain and tail region of the Drosophila embryo

The establishment of the segmentation pattern in the Drosophila embryo requires both maternal and zygotic gene products during early embryogenesis. Maternally active genes establish the overall anterior-posterior axis of the embryo and large regions of the pattern along this axis (Niisslein-Volhard, 1977, 1979; Schüpbach & Wieschaus, 1986; Nüsslein-Volhard et al. 1987). Zygotic genes are required for the formation of more localized regions of the segmentation pattern. The zygotic tailless (tll) gene (Jürgens et al. 1984; Strecker et al. 1986) is required for the formation of the anterior and posterior ectodermal regions that border the aggregate domains affected by the zygotic gap mutations (Wieschaus et al. 1984; Lehmann & Nüssslein-Volhard, 1987; Bender et al. 1987; Petschek et al. 1987).

Defects observed in the mature embryonic cuticle of tll embryos can be traced to defects in the 9h embryo (Strecker et al. 1986). Anteriorly, an abnormal dorsal cephalopharyngeal skeleton can be traced to a reduced clypeolabrum and procephalic lobe. Posteriorly, the absence of the tail region [term used by Jürgens (1987) to refer to the larval epidermis behind the posterior boundary of the seventh abdominal segment] can be traced to an absence of the most posterior abdominal segments. Despite the absence of the tail region, tailless embryos are of normal body length due to an expansion of the remaining abdominal segments (Strecker et al. 1986). This alteration in the remaining positional values has been shown to occur by the cellular blastoderm stage (Mahoney et al. 1986; Mahoney & Lengyel, 1987), consistent with the hypothesis that the tailless gene has a fundamental role in establishing and/or maintaining positional information along the anterior-posterior axis at the time of cell determination.

We present here a detailed genetic and phenotypic investigation of the tll gene and its relationship to the maternai terminal gene family (defined by Nüsslein-Volhard et al. 1987). The isolation of simple, overlapping deficiencies of tll facilitated the characterization of null and hypomorphic genotypes. We have determined the elements of the embryo most sensitive to tll gene activity and have correlated the degree of anterior and posterior phenotypic severity in the tll allelic series. Using anti-horseradish peroxidase staining, we tested the effect of a lack of tll gene activity on the brain. Characterization of double mutants establishes the relationship between tll, maternal terminal and zygotic gap genes.

The original tll allele (tll¹ = L1O.22) was isolated on a st e marked chromosome in an EMS mutagenesis screen (Jürgens et al. 1984). Gene and chromosome symbols are described by Lindsley & Grell (1968) and Lindsley & Zimm (1985, 1986, 1987). All stocks were maintained on standard Drosophila media (Ashbumer & Thompson, 1978).

The deficiency and duplication chromosomes used to determine the location of tll are shown in Fig. 1. The terminal 3R duplications, Dp(3;l)s 34,152P, 93,150P, 79, 1A and 48, were provided by Dr R. MacIntyre and have been previously described (Frisardi & MacIntyre, 1984; Kongsuwan et al. 1986; Strecker, 1987). The terminal 3R deficiencies, Df(3R)s J55, R97, A113 and L129, have also been previously described (Lindsley et al. 1972; Kongsuwan et al. 1986; Strecker, 1987).

2- to 5-day-old males homozygous for the recessive marker ca (Lindsley & Grell, 1968) were irradiated with 3600rad and crossed to untreated H^DIcd ca/In(3R)C, Tb Sb cd ca virgin females at 25°C (50 males:200 females per bottle). After 2–3 days, males were transferred to fresh bottles and given additional untreated H^DIcd ca/In(3R)C, Tb Sb cd ca virgin females. Males were discarded 5 days after irradiation. The heterozygous (ca)*/In(3R)C F₁ progeny were then screened for induced tll mutations as follows. Single (ca)*/In(3R)C F₁ males were crossed to three to five tll¹/In(3R)C, Tb Sb cd ca virgin females. Each test mating was kept in a vial for 12 to 15 days at 25°C after which it was screened for the absence of the (ca)*/tll¹ F₂ progeny class. Since only the latter will have normally shaped pupal cases, while the remaining progeny carry the Tb (Tubby) pupal marker on the In(3R)C balancer chromosome, the absence of wildtype pupae was used as a selection criterion for lines containing putative tll alleles. In addition, virgin (ca)*/In(3R)C F₁ females were mated en masse to tll¹/ In(3R)C males for two days, after which fertilized females were placed individually in vials and kept at 25°C for 12 to 15 days. Each vial was then screened as described above. From matings that exhibited an absence of wildtype pupae, (ca)*/In(3R)C males and virgin females were crossed inter se to establish a balanced stock.

The cuticle of unhatched mature embryos was prepared for examination by the procedure of van der Meer (1977), modified as described previously by Strecker et al. (1986). Anteriorly, structures of the cephalopharyngeal skeleton (described by Jürgens et al. 1986) were examined in detail, namely the dorsal arms, dorsal bridge, vertical plates, ventral arms, median tooth (labrum) and dorsal sac (Fig. 3A). (Note that the region termed the pharyngeal ridge in our original description of the tll mutant phenotype [Strecker et al. 1986] is what we now refer to as the dorsal cephalopharyngeal skeleton.) Posteriorly, tll embryos were examined for the presence of cuticular structures of the tail region as defined by Jürgens (1987). In addition to the ventral denticle belts and dorsal hairs of the seventh and eighth abdominal segments, the presence or absence of the dorsal straight spinules (or Fell), Filzkörper, posterior spiracles, anal tuft and anal pads was noted (Fig. 3I; Lohs-Schardin et al. 1979; Sato & Denell, 1986; Whittle et al. 1986; Jürgens, 1987).

Staged embryos were collected and prepared for SEM according to the protocol of Mahowald & Turner (1978) modified as described by Strecker et al. (1986). Anteriorly, 9 h tll embryos were examined for the degree of reduction of the clypeolabrum, procephalic lobe and optic plaques (Fig. 4A; Turner & Mahowald, 1979; Schoeller, 1964). (Note that in our original description of the tll mutant phenotype [Strecker et al. 1986] the optic plaques were incorrectly termed optic lobes.)

The supra- and suboesophageal ganglia were visualized by staining with anti-horseradish peroxidase antibody (Jan & Jan, 1982; Hartenstein & Campos-Ortega, 1986) using modifications suggested by J. Campos-Ortega (personal communication). Embryos were collected and aged to 16–18 h at 22°C at which time they were dechorionated with 50% bleach and transferred to heptane:4% paraformalde-hyde/PBS (phosphate-buffered saline: 130 mm-NaCl, 7 mm-Na₂HPO₄, 3mm-NaH₂PO₄) pH 7·2 (1:1) and rotated for 20 min. The fixative was removed and embryos were devitellinized by shaking vigorously in 100% methanol for 1 min. Then embryos were washed in PBS and postfixed for 40 min in PBS containing 4% paraformaldehyde. Embryos were washed 20 min each in PBS, PBS containing 0·2% BSA (PBS/BSA) and PBS containing 0·2% BSA and 0·2 % Triton X-100 (PBT) and then incubated for 20 min in 10% goat serum (Sigma) in PBT. Following addition of rabbit anti-horseradish peroxidase (HRP; Cappel) to a final dilution of 1:4000, the embryos were incubated overnight at room temperature with gentle agitation. Embryos were then washed in PBT, preblocked by a 30 min incubation in 10 % goat serum in a volume appropriate for diluting the second antibody, then incubated for 3 h at room temperature with goat anti-rabbit IgG conjugated to HRP (Sigma) added to a final dilution of 1:50. Embryos were washed 20 min each in PBT, PBS/BSA and PBS, and were then stained for HRP activity using diaminobenzidine (1mg ml^-1, 0 ·01 % H₂O₂; Polysciences). The reaction was monitored under a dissecting microscope and stopped by diluting with PBS. Stained embryos were stored in PBS containing 0 ·02 % NaN₃ and were mounted in glycerol for microscopic examination and photography using differential interference contrast optics. Embryos were scored for the presence of the following brain structures; supra- and suboesophageal ganglia, anterior and posterior trunks of the supraoesophageal neuropile and the frontal commissure (Campos-Ortega & Hartenstein, 1985; Fig. 3E).

Mutant alleles of various genes were placed in combination with tll. The alleles used and their sources were trunk [trk^RA41, Schupbach & Wieschaus, 1986] from T. Schüp-bach and E. Wieschaus; fs(1)Nasrat [fs(1)N²¹¹, Degelmann et al. 1986] from A. P. Mahowald; Krüppel [Kr¹, Wieschaus et al. 1984] and hunchback [hb^14F, Lehmann & Nüsslein-Volhard, 1987; Bender et al. 1987] from Bowling Green Stock Center; extra sex combs [esc² and esc⁵, Struhl, 1981] from G. Struhl. Double-mutant embryos were obtained utilizing standard genetic crosses; the mature cuticle of these embryos was examined as described above.

In the EMS mutagenesis screen in which it was identified (Jürgens et al. 1984), only one tll allele was recovered. We wished to obtain additional tll alleles to determine the null phenotype, the regions of the pattern most sensitive to loss of tll gene function and whether anterior and posterior regions of the embryo are affected to a similar degree by different tll alleles. From 10 100 X-ray mutagenized chromosomes, we identified six additional tll alleles. Two of the alleles, tll^a and tll^c, are cytologically normal. The cytology of the three recovered deletions, tll^e, tll^f and tll^g, places the tll locus in the chromosomal region 100Al,2-100B4,5 (Fig. 1). The localization of tll was further refined with the recovery of the tll² allele, an inversion with a distal breakpoint between bands 100A5,6 and 100B1,2. Additional confirmation of this localization was obtained from the observation that the combination of the proximal part of T(Y;3)A113 with Dp(3;1)150P, which creates a synthetic deletion of the region between 100A5,6 and 100Bl,2 (Kongsuwan et al. 1986), uncovers the tll mutant phenotype (Fig. 1). Embryos carrying this synthetic deficiency, and tll¹, tll², tll^a or Df(3R)tll^e, die and exhibit the embryonic tll phenotype. All the evidence is thus consistent with the localization of the tll gene within 100A5,6-100Bl,2.

The tll alleles were ordered into a graded phenotypic series on the basis of the following description of the development of the anterior and posterior regions of the embryo. Anteriorly, the major elements of the dorsal cephalopharyngeal skeleton are the dorsal bridge, dorsal arms and vertical plates (Fig. 3A); fate mapping traces the origin of these structures to the procephalic lobe of the early segmented embryo, and earlier, to the acronal anlage of the cellular blastoderm (Jürgens et al. 1986). Posteriorly, fate mapping indicates that the cuticular elements of the tail region arise from the prospective abdominal segments in the cellular blastoderm as follows: anal pads (All and A10), anal tuft (telson, which is dorsal to A9/A10), dorsal straight spinules or Fell (anterodorsal A8), spiracles and Filzkórper (posterodorsal A8) and the eighth abdominal denticle belt (anteroventral A8) (Fig. 3I; Jürgens, 1987; Sato & Denell, 1986; Whittle et al. 1986).

Recovery of the Df(3R)tll^e and Df(3R) tll^g overlapping deficiencies (Fig. 1) permitted the construction of trans-heterozygotes which contain a small deficiency of the region including the tll gene and which give rise to the tll null phenotype (Fig. 2E). Anteriorly, tll^e/ tll^g (=tll null) embryos have normal derivatives of the gnathal (mandibular, maxillary and labial) segments, i.e. antennomaxillary sense organs, lateralgräten, ventral arms, H-piece, mouthhooks, cirri and labial sense organs. The most obvious anterior cuticle defect is an abnormal dorsal cephalo-pharyngeal apparatus; the contralateral homologues of the dorsal bridge fail to fuse and the dorsal arms are reduced in length (Fig. 3C). Although the median tooth (labrum) is formed, it is located in a more posterior position along the dorsal sac; additionally, scleritized material lies within the dorsal sac just posterior to the labrum (‘dorsal pouch’ syndrome described by Jürgens et al. 1986). Posteriorly, these embryos lack all derivatives of the tail region, as well as the posteroventral cuticle of A7. In the most extreme embryos (a small fraction of the total), the denticle belt and dorsal hairs of A7 are also deleted (Fig. 3K). Although all cuticular structures of the tail region are missing in tll null embryos, the anal opening and proctodeum are always present.

We describe below the phenotypes of several tll alleles, which exhibit weaker phenotypes than that of tll null embryos. Each of these alleles, when in trans to Df(3R) tll^e, give rise to embryos which exhibit a slightly more extreme distribution of mutant phenotypes than that observed when the allele is homozygous; this was determined by scoring the phenotypic range of 50 embryos of each genotype. Based on these observations, tll¹, tll^a and tll² are considered hypomorphs.

Embryos that are tll¹/tll¹ or tll¹/Df(3R) tll^e (Fig. 2D) are indistinguishable from tll null embryos in the anterior, but show a weaker phenotype in the posterior: while the ventral denticle belt of A8, dorsal straight spinules, Filzkôrper, posterior spiracles, anal tuft and anal pads are missing, a small tuft of A8 dorsal hairs and all cuticular structures of A7 are present (Fig. 3J). In a few less extreme cases, tll¹/tll¹ embryos display a cluster of denticles, probably a remnant of the A8 denticle belt, in addition to dorsal hairs at the posterior end of the embryo.

Homozygous tll^a or tll^a/ Df(3R) tll^e embryos show a weaker phenotype (Fig. 2C) than tll null and tll¹ embryos. Anteriorly, tll^a/ tll^a embryos have a continuous dorsal bridge but reduced dorsal arms (Fig. 3B); the median tooth is positioned just posterior to the mouthhooks as in wildtype embryos. There is little, if any, scleritized material in the dorsal sac. Posteriorly, tll^a/ tll^a embryos are indistinguishable from tll¹ embryos (Fig. 3J).

Homozygous tll² embryos exhibit the weakest mutant phenotype. Approximately half of these embryos hatch from the chorion and die as first instar larvae, showing the same phenotype as the lethal embryos (Fig. 2B). Anteriorly, the cephalopharyngeal skeleton is normal in tll² homozygous or tll²/ Df(3R) tll^e embryos. Posteriorly, the majority of tll²/tll² and tll²/Df(3R) tll^e embryos and larvae have a slightly reduced eighth abdominal denticle belt and partial Filzkôrper. These embryos frequently have only one posterior spiracle and an anal tuft in an abnormal position, adjacent to the dorsal straight spinules. The most extreme tll²/Df(3R) tll^e embryos have a reduced A8 ventral denticle belt and lack the dorsal straight spinules in addition to the remaining elements of the tail region, while the least extreme tll²/tll² embryos exhibit all tail derivatives except the anal pads (Fig. 2B).

Structures of the cephalopharyngeal skeleton that are reduced or missing in mature tll embryos (dorsal bridge and arms) have been fate mapped to a position in the blastoderm embryo adjacent to the presumptive brain (supraoesophageal ganglion; Jürgens et al. 1986). Furthermore, during germband extension, elements of the cephalopharyngeal skeleton map to the surface of the procephalic lobe while the supraoesophageal ganglion is formed from the procephalic and optic lobes (Fig. 4B; Jürgens et al. 1986; Campos-Ortega & Hartenstein, 1985).

To test the hypothesis that tll also results in the reduction of the adjacent supraoesophageal ganglion anlage, we used antibody to horseradish peroxidase, which stains the central nervous system (Jan & Jan, 1982). In tll null embryos, the central nervous system is truncated abruptly at the anterior end of the suboesophageal ganglion (Fig. 3G). Both the anterior trunk of the supraoesophageal neuropile (at) and the posterior trunk of the supraoesophageal commissure (pt) are missing. Consistent with this observation, the posterior half of the procephalic lobe (p) and the adjacent optic plaque (o) [described in Calliphora by Schoeller (1964)] are deleted in tll null embryos at the completion of germ band retraction (Fig. 4C, arrow). The optic plaque borders the optic lobe invagination and has been proposed to be the primordium of Bolwig’s organ, the larval visual organ (Bolwig, 1946; Schoeller, 1964). The hypomorphic tll^a/ tll^b embryos exhibit a consistently less extreme reduction of the brain region (Fig. 3F), lacking only the portions posterior to the posterior trunk of the supraoeso-phageal ganglion (pt), consistent with the hypomorphic effect of the tll^a allele on the cephalopharyngeal skeleton (see above). Furthermore, the more hypomorphic tll²/Df(3R) tll^e embryos have a normally formed supraoesophageal ganglion, consistent with the normal cephalopharyngeal skeleton in tll² embryos.

Due to the internalization of regions adjacent to the anterior domain (acron) affected by the tll mutation, an alteration in the anterior segmentation pattern of the mature mutant embryo is difficult to assess. However, with the complete deletion of acronal derivatives in tll null embryos there is a substantial increase in segment length, not only posteriorly in segments A2 through A6, but also anteriorly in T1 (Fig. 5). Consistent with the increase of T1 length in these embryos, there is an increase in the size of the dorsal ridge, the dorsal component of the labial segment (Fig. 4C; Campos-Ortega & Hartenstein, 1985). Thus, concomitant with the deletion of the acron and tail regions in tll embryos, there is an expansion of the remaining segmentation pattern toward the opposite ends of the embryo, consistent with previous observations (Strecker et al. 1986; Mahoney & Lengyel, 1987). The increase in segment length is correlated with the severity of the mutant phenotypes, such that tll null embryos exhibit the greatest increase in length of their anterior and posterior body segments, while hypomorphic tll embryos (tll¹/Df(3R) tll^e, tll^a/Df(3R) tll^e and tll²/ Df(3R) tll^e) exhibit proportionately smaller increases in segment length (data not shown; Strecker, 1987).

What is the position of tll in the hierarchy of maternal and zygotic loci that determine the body pattern in the Drosophila embryo? The phenotypes of two classes of pattern mutants, i.e. the maternal terminal (torso, trunk, fs(1)Nasraf, originally referred to as ‘torso-like’ genes; Sch ü pbach & Wieschaus, 1987; Degelmann et al. 1987; Nüsslein-Volhard et al. 1987) and the zygotic gap genes (Nüsslein-Volhard & Wieschaus, 1980; Kriippel, Wieschaus et al. 1984; hunchback, Lehmann & Nüsslein-Volhard, 1987 and Bender et al. 1987) suggest that they may be involved in functions related to that of the tll gene. The relationship between tll and these pattern mutants was addressed through double mutant and phenotypic analysis. To determine if the tll gene acts independently of the segmental identity of the subregion it affects, tll was combined with the homeotic mutant, extra sex combs (esc).

The segmental domains deleted by the tll mutation overlap with those regions deleted by the maternal effect terminal mutations (Fig. 7). A phenotypic comparison of the anterior and posterior cuticle defects in trk and tll embryos reveals the common pattern elements deleted by these mutations. In the anterior, tll embryos lack only acronal derivatives (dorsal arms, dorsal bridge), while trk embryos lack both these and the labral derivative: the median tooth (Fig. 3D; Schüpbach & Wieschaus, 1986; Jürgens et al. 1986). In addition, like tll embryos, trk embryos exhibit a significant reduction in the size of the supraoesophageal ganglion (Fig. 3H). In the posterior, tll affects only the ectodermal tail region and A7, while trk embryos lack both this ectodermal region and endodermal structures as well, i.e. the posterior midgut and proctodeum (Fig. 3L; Schüp-bach & Wieschaus, 1986). Finally, as we have described above for tll, the remaining positional values in maternal terminal embryos are expanded to replace those which are deleted (Schüpbach & Wieschaus, 1986; Degelmann et al. 1986).

Using double-mutant constructs, we investigated how the embryo responds to an alteration in anterior-posterior positional values caused by both the maternal effect terminal and zygotic tailless mutations. The larval cuticle of 50 embryos from a cross of trk¹/ trk¹; tll^a/In(3R)C females to + /+;Df(3R) tll^e/In(3R)C males was scored (Fig. 6A). Although 25 % (12 embryos) of these should have been tll^a/Df and thus lacked tll⁺ gene activity, all 50 embryos appeared identical to trk embryos and exhibited the most extreme trk phenotype (Fig. 6A). [In contrast, 18 % (9/50) of the tll⁺ embryos from trk/trk mothers exhibited weaker posterior phenotypes, namely a complete seventh abdominal segment and reduced eighth abdominal denticle belt.] Similar results were obtained in a double-mutant combination of fs(1)N²¹¹ and tll^a (Fig. 6B). Thus the loss of both tll and trk (or fs(1)N) does not result in a more extreme mutant phenotype than for trk (or fs(1)N) alone, but does result more frequently in embryos exhibiting the most extreme trk (or fs(1)N) phenotype. These results are consistent with the idea that trk, fs(1)N and tll are involved in a common process.

Like tll, the gap mutations also affect large, contiguous, aperiodic regions of the segmentation pattern. We placed tll in combination with the gap mutants Kr and hb to determine whether these zygotic genes interact. The tll¹,hb^14F double mutant produced a cuticular phenotype which is the sum of the tll and hb mutant phenotypes (data not shown; Strecker, 1987). As expected, the tll¹,Kr¹embryo lacks T1 through A5 as well as the telson and most of A8; A6 and A7 are present. The tll,Kr double-mutant phenotype is consistently more extreme, however, than the predicted sum of the two phenotypes: there are additional denticles in the region between A6 and A7, which lack normal polarity and point toward the ventral midline of the embryo. Furthermore, there is a partial fusion between all the ventral denticle belts (Fig. 6D). This is similar to the ‘lawn’ phenotype observed in hb,kni and h.ftz double-mutant embryos, where there is a fusion and loss in polarity of the ventral denticle belts, which is interpreted as a failure of the embryo to form segments (Nüsslein-Volhard et al. 1985). A common characteristic among these three double-mutant combinations is that the mutant loci that were combined have complementary phenotypes with respect to one another (Nüsslein-Volhard et al. 1985; Strecker & Merriam, 1986). Presumably the deletion of large, complementary regions of the segmentation pattern, found for the tll,Kr, hb,kn and ftz,h double mutants, results in the remaining segment boundaries becoming fused or failing to form.

The maternal effect gene, esc, is required during development for the correct specification of segment identity; the body segments in embryos from esc/esc females mated to esc/esc males develop like A8 (Struhl, 1981). To determine whether tll acts independently of the segment identity of the region it is required to specify, embryos from the cross of est²/esc⁵;tll¹/ + females to esc²/esc⁵;tll¹/ + males were examined (Fig. 6F). Approximately one quarter of these embryos, while exhibiting the characteristic A8 transformation of the gnathal, thoracic and abdominal segments, also lacked the telson and had one fewer A8 denticle belt. These embryos were presumed to be doubly mutant for esc and tll gene functions. Although these embryos lacked derivatives of the three most-posterior abdominal segments, they were not shorter in length, due to an increase in the width of the remaining body segments. The transformation of most of the body segments to A8 by the esc mutation did not alter the effect of the tll mutation. This is perhaps not surprising, given that esc appears to act toward the end, and tll toward the beginning, of nuclear cycle 14 (Struhl & Brower, 1982; Mahoney & Lengyel, 1987). We conclude that the tll gene acts independently of segment identity.

As the zygotic tll and maternal terminal genes appear to be involved in the same process (see above), we tested for maternal tll⁺ expression by asking whether extra maternal copies of the tll⁺ gene result in a less extreme tll mutant phenotype. This approach has been shown to be consistent with results obtained using germline clones (Wieschaus & Noell, 1986).

Flies that are Dp(3;1)93/X;Df(3R)A113/TM6 have two functional copies of the tll⁺ gene, while those that are tll¹/TM3 have only one. If these two genotypes are reciprocally crossed to each other, a proportion of the F₁ progeny that are tll¹/Df(3R)A113 will not inherit Dp(3;l)93 and hence will have no functional tll gene (Fig. 1). If the tll⁺ gene is expressed maternally as well as zygotically, a difference in phenotypic severity should appear among the resultant tll¹/Df(3R)A113 progeny from these reciprocal crosses. 50 embryos from each reciprocal cross were scored with respect to the mature anterior and posterior cuticular phenotypes and were found to be indistinguishable from one another. Furthermore, embryos from reciprocal crosses using Dp(3;1)34 and Dp(3;1)150P (Fig. 1) were also observed to be phenotypically indistinguishable. These results indicate that wildtype maternal levels of the tll⁺ gene do not lessen the tll mutant phenotype. Although gene dosage experiments do not provide as conclusive evidence as pole cell transplantation, the results reported above are consistent with the proposal that tll⁺ is not maternally expressed.

We have shown that in tll embryos the supraoeso-phageal ganglion (brain) is deleted, the procephalic lobe is significantly reduced and, in the cephalophar-yngeal skeleton of the mature cuticle, the dorsal arms and dorsal bridge are missing. The anlagen for these structures have been fate mapped to the acron in the cellular blastoderm (see fig. 9 in Jürgens et al. 1986). In addition, tll embryos lack derivatives of the seventh abdominal segment and tail region, which has been defined as comprised of segments A8-A11 and an unsegmented telson (see fig. 7 in Jürgens, 1987). Based on these observations, we conclude that the tll gene is required for the formation of the acron and tail region in the cellular blastoderm. The anlagen for the acron and tail occupy approximately the same area at opposite ends of the blastoderm fate map and are equidistant from the midpoint (50 % egg length) along the anterior-posterior axis (Hartenstein et al. 1985; Jürgens et al. 1986; Jürgens, 1987). The requirement for the tll gene by these two distant regions of the early embryo suggests that there is an underlying developmental relationship between the acron and tail regions in Drosophila.

The tll alleles were ordered in a continuous graded series ranging from the most extreme to the weakest allele: Df(3R)tll^e,Df(3R)tll^g> tll¹> tll^a> tll² (data summarized in Table 1). There is a reasonably good correlation between the severity of the anterior and posterior mutant phenotypes, i.e. the weakest posterior phenotype is observed in tll² embryos which have a normal cephalopharyngeal skeleton, while the most severe posterior reduction in tll null embryos is correlated with the most extreme deletions of structures in the dorsal cephalopharyngeal apparatus. This correlation between anterior and posterior phenotypic severities suggests that the anterior and posterior functions of the tll gene are not separable.

The allelic series further revealed the differential requirements of elements of the segmentation pattern for tll gene activity (Table 1). In the posterior, the most hypomorphic tll embryos lack only the anal pads, which have been fate mapped to the presumptive A10 and All region in the cellular blastoderm and which constitute the most posterior structure of the tail region (Jürgens, 1987). With increasing phenotypic severity, we observed deletion of the anal tuft, which arises from a more anterior blastoderm position, the telson region, dorsal to the presumptive A9/A10 boundary. Less sensitive structures are the posterior spiracles (posterodorsal A8), followed by the Filzkôrper (posterodorsal A8) and the eighth abdominal denticle belt (anteroventral A8). Finally, in the most extreme tll phenotypes, the anterior half of A8 and posterior half of A7 are deleted. These observations suggest that there is a graded, posterior- to-anterior, requirement for the tll gene product in the posterior ectodermal region of the embryo; the peak of this requirement is in the presumptive All region (∼ 12% egg length).

The effect of different tll alleles on the formation of the supraoesophageal ganglion suggests that there is a similar graded requirement for the tll gene product in the anterior (Table 1). Hypomorphic tll^a embryos lack the most posterior region of the supraoesophageal ganglion; this region includes, but may not be limited to, the optic lobe (Campos-Ortega & Hartenstein, 1985). An absence of the optic lobe in hypomorphic tll embryos is consistent with the abnormal rounded shape of the optic plaque (Strecker et al. 1986), which is presumably related to the failure of the optic plaque to become internalized just prior to the advance of the dorsal fold (Strecker, 1987). tll null embryos are missing not only the posterior, optic lobe-containing portion of the brain, but also the anterior portion of the brain (Fig. 3, at, pt) and, at an earlier stage, the optic plaques. These observations suggest that there is a graded requirement for the tll gene product in the primordia of the supraoeso-phageal ganglion; the peak of this requirement is probably in the presumptive optic lobe.

Although tll embryos lack acronal derivatives and the tail region, they are of normal body length due to the increase in the size of the remaining body segments, particularly those segments immediately adjacent to the deleted regions. This alteration of the remaining segmentation pattern correlates with the severity of the anterior and posterior mutant phenotypes observed in tailless embryos. With the deletion of a larger fraction of the anterior and posterior domains, there is a greater expansion of the remaining segmentation pattern. This can be traced back to an altered fate map in the blastoderm embryo, detected from altered expression patterns of fushi taraza (ftz) and hairy (h) genes (Mahoney & Lengyel, 1987). These observations, coupled with the smooth posterior-anterior progression of deleted abdominal structures in the tll allelic series, support our previous suggestion that the tailless gene is required for the establishment and/or maintenance of graded positional information along the anterior-posterior axis (Mahoney & Lengyel, 1987).

Among the existing mutations known to affect the body pattern of the embryo, those in the maternal terminal family, i.e. torso, trunk and fs(1)Nasrat (Schüpbach & Wieschaus, 1986; Degelmann et al. 1986), result in a phenotype most closely resembling that of tll. The terminal phenotype consists of the deletion of derivatives of the labrum and acron, anteriorly, and the tail region and endodermal derivatives, posteriorly. The pattern elements deleted by the tll mutation, the acron and tail, are a subset of those affected by mutants in the terminal family (Fig. 7).

In addition to deleting structures from the same anterior and posterior domains, both tll and the terminal mutants result in an expansion of the remaining pattern, rather than the death of improperly patterned cells. This constitutes a fundamental distinction between tll and the gap gene mutants. Anteriorly, the pattern expansion is observed as an increase in the segment width of T1 and the dorsal ridge in tll embryos (Fig. 4C), and an anterior shift of the cephalic furrow and anterior midgut invagination in torso and trunk embryos (Schüpbach & Wieschaus, 1986) . Posteriorly, the pattern expansion is observed as an increase in abdominal segment width (A5, A6) in tll embryos, and a posteriorward shift of subterminal embryonic structures in torso and trunk embryos (Schüpbach & Wieschaus, 1986). The posteriorward shift in pattern can be traced to changes in the blastoderm fate map. As assessed by h and ftz gene expression, similar shifts are seen in fs(1)N, tll, tor and trk embryos (Degelmann et al. 1986; Mahoney et al. 1986; Mahoney & Lengyel, 1987; Mlodzik et al. 1987) . These results suggest that the zygotic tailless gene is required to maintain the positional information initially established by the action of the maternal terminal genes.

This idea is supported by the results of doublemutant analysis. The same structures are missing from embryos deficient for tll and trk (or fs(1)N) as from embryos deficient only for trk (or fs(1)N); furthermore, trk,tll double-mutant embryos more frequently exhibit a more extreme trk phenotype than do trk,tll⁺ embryos. These results are consistent with those obtained with tor,trk double mutants (Schüpbach & Wieschaus, 1986). [In contrast to tll, the gap mutant giant does not recognize the alteration in positional values resulting from the fs(1)N²¹¹mutation (Petschek et al. 1987).] We interpret our double-mutant results as suggesting that trk, fs(1)N, tor and tll are involved in a common pathway required to establish and maintain positional information at the anterior and posterior ends of the embryo. Specifically, we propose that tll mediates the establishment of the acron and tail regions within the larger domains affected by the maternal terminal genes.

We wish to thank Jose Campos-Ortega and Volker Hartenstein for helpful discussions regarding the central nervous system, and Kathryn Anderson and Howard Lipshitz for critically reading the manuscript. We also thank Trudi Schüpbach, Eric Wieschaus, and Gary Struhl for sending us fly strains. This research was supported by NSF Grant DCB 84-03539 to J.A.L. and J.R.M., and NIH Grant GM 38432 to J.R.M. T.R.S. was supported by USPHS National Service Award GM7104 and a predoctoral fellowship from the American Association of University Women.