In a screen for mutations affecting Drosophila eye devel-opment, we have identified a gene called fat facets (faf) which is required for cell interactions that prevent par-ticular cells in the developing eye from becoming pho-toreceptors. Analysis of eyes mosaic for faf+ and faf cells shows that faf is required in cells near to, but out-side, normal developing photoreceptors and also outside of the ectopic photoreceptors in mutant facets. faf is also essential during oogenesis, and we show that a faf-lacZ hybrid protein is localized via the first 392 amino acids of faf to the posterior pole of oocytes. Posterior local-ization of faf-lacZ depends on oskar. oskar encodes a key organizer of the pole plasm, a specialized cytoplasm at the posterior pole of embryos. The pole plasm is required for germ cell formation and contains the deter-minant of posterior polarity, encoded by nanos. Although other pole plasm components are required for localization of nanos RNA or for nanos protein func-tion, faf is not. We have cloned the faf gene, and have shown that it encodes two similar large (∼300 103Mr) proteins that are unique with respect to other known proteins.

Short-range cell interactions guide the assembly of the Drosophila compound eye in the eye imaginal disc, as cells sequentially join growing clusters of unit eyes, or facets (recently reviewed by Banerjee and Zipursky, 1990; Rubin, 1991). Many genes have been identified that are involved in particular steps of the assembly pathway, for example, scabrous, retina aberrant in pattern, rough, Star, Bar, seven-up, sevenless, seven in absentia, bride-of-sevenless, and glass (Baker et al., 1990; Mlodzik et al., 1990a; Karpi-low et al., 1989; Tomlinson et al., 1988; Saint et al., 1988; Heberlein and Rubin, 1991; Higashijima et al., 1992; Mlodzik et al., 1990b; Hafen et al., 1987; Carthew and Rubin, 1990; Reinke and Zipursky, 1988; Moses et al., 1989). The products of all of these genes are required in specific cells within the growing facets, where they either participate in cell-cell interactions or control cell determi-nation from within the cell. The large number of such genes supports the model that facet assembly is controlled by cells within the facets (Tomlinson and Ready, 1987a).

In recent studies of the argos (Freeman et al., 1992) and groucho (gro; Fischer-Vize et al., 1992) genes, it was shown that cells outside of the developing facets also influence facet assembly. Mutations in both genes have a sim-ilar eye phenotype; a few cells are inappropriately deter-mined as photoreceptors. argos encodes a diffusible factor (Freeman et al., 1992) and gro encodes a nuclear protein (Delidakis et al., 1991). Nothing more is certain about the mechanism of either gene product’s function in the eye, or their relationship to each other.

Here we describe a gene called fat facets (faf), that also influences facet assembly from outside the facet. faf mutant eyes contain one or more ectopic photoreceptor cells in nearly every facet. By analyzing facets mosaic for faf+ and faf photoreceptors, we show that faf is required in cells near to, but outside the eight photoreceptors in wild-type facets, and outside the extra photoreceptors in mutant facets. Unlike argos, faf does not appear to be diffusible. Unlike the pleiotropic roles of gro, the essential role of faf in repression of neural cell determination is limited to the eye disc.

faf is also required during oogenesis. Embryos from faf mutant mothers never cellularize, except for the formation of a few primordial germ cells, called pole cells, at the pos-terior of the embryo. Moreover, we find that a faf-lacZ hybrid protein is localized to the posterior pole of oocytes.

The localization of particular mRNAs and proteins to opposite poles of the oocyte is the mechanism by which anterior-posterior polarity of the embryo is established (reviewed by St. Johnston and Nüsslein-Volhard, 1992). Posterior polarity depends on ten posterior group genes, nine of which have been organized into a functional hier-archy: cappuccino (capu) and spire (spir), staufen (stau), oskar (osk), vasa (vas), tudor (tud), valois (vls), nanos (nos) and pumilio (pum) (Boswell et al., 1991; Manseau and Schüpbach, 1989; St. Johnston et al., 1991; Ephrussi et al., 1991; Kim-Ha et al., 1991; Ephrussi and Lehmann, 1992; Hay et al., 1990; Lasko and Ashburner, 1990; Lehmann and Nüsslein-Volhard, 1991, 1987; Macdonald, 1992; D. Barker, J. Moore, L.K. Dickinson and R.L., unpublished data). The posterior group genes function through the nos gene product, the posterior determinant, by localizing nos RNA to the posterior of the embryo, or modulating the func-tion of nos protein. Except for nos and pum, the posterior group genes are also required for pole cell formation (reviewed by Lehmann, 1992). stau and vas proteins, and osk RNA and protein are known to be localized to the pos-terior of oocytes prior to nos RNA, and the localization of each depends on the genes upstream in the hierarchy (St. Johnston et al., 1991; Ephrussi et al., 1991; A. Ephrussi and R.L., unpublished data; Hay et al., 1988b; Lasko and Ash-burner, 1990). Nothing is known about the mechanism of protein localization to the posterior of the oocyte.

Here we show that a faf-lacZ hybrid protein containing only the first 392 amino acids of faf is posteriorly localized as a protein, and that its localization depends on osk. Unlike the posterior group genes, faf is not required for nos RNA localization or function. We have cloned the faf gene, and have shown that it encodes two large (∼300×103Mr) pro-teins with different carboxy termini, unique with respect to other known proteins.

Drosophila genetics Fly lines

Fly lines

The enhancer trap line AE127, marked with white+, has a P ele-ment in the svp gene (M. Mlodzik, J. Heilig and G.M.R., unpub-lished data). The posterior group mutant alleles used were capuRK and spirRP (Manseau and Schüpbach, 1989), stauD3, osk150, osk346, vasD1, vas011, nosL7 (Lehmann and Nüsslein-Volhard, 1991), stauHL54, tudWC8, vlsRB71 (Schüpbach and Wieschaus, 1986) and pum680 (Lehmann and Nüsslein-Volhard, 1987). All other mutant markers are described in Lindsley and Zimm (1992). Flies were grown on standard food at 25°C.

Generation of faf alleles

Six EMS-induced faf alleles were isolated by crossing muta-genized bw; st males to bw; st fafFO8/TM6B virgins, and screen-ing approximately 27,000 bw; *st/st fafFO8 male progeny for rough eyes. Thirteen additional faf alleles were induced with X-rays as follows. st males were exposed to X-rays (4000 rads) and then crossed to bw; marbCD4st cu sr e fafFO8/TM6B virgins. (marbCD4 is another rough eye mutation; J. A. F.-V. and G. M. R., unpublished data). Approximately 30,000 bw; *st/fafFO8 male progeny were screened for rough eyes. One additional faf allele (fafBP) was isolated in a P-M hybrid dysgenesis screen.

Birm2; st males were crossed to (2-3)99B virgins at 16-18°C. (The Birm2 and (2-3)99B chromosomes are described by Robertson et al., 1988). The male progeny (Birm2/+; (2-3)/st) were crossed to bw; marbCD4st cu sr e fafFO8/TM6B virgins at 22oC, and >100,000 st/fafFO8 male progeny were screened for rough eyes.

Meiotic mapping of faf

First, bw; st fafFO8/th st cu sr e ca virgins were crossed with bw; th st cu sr e ca males. Many progeny of all genotypic classes were individually mated with bw; st fafFO8/TM6B virgins to determine whether or not the recombinant chromosome contained fafFO8. By this process, fafFO8 was positioned near the tip of chromosome 3R. Next, fafFO8 was found to be distal to a white+ P element (P[w+]) insertion in polytene chromosome region 99A (a gift of the Rubin laboratory enhancer trappers) by crossing w; st e fafFO8/P[w+] virgins to w; st e fafFO8 males and scoring many progeny for the e, w+ and faf phenotypes.

Other fly crosses

Standard genetic crosses were used to combine fafFO8 with AE127, the posterior group genes with a faf/lacZ P element on the X chro-mosome, and fafI, fafII and fafIII P elements on chromosomes X and II with fafBX4.

Generation of fafeye clones

Clones of fafFO8/fafFO8 cells marked by the absence of the white gene product (no pigment granules associated with photoreceptor or pigment cells) were generated by crossing w1118; st fafFO8/TM6B males with w1118; P[w+]90E virgins and X-irradiat-ing (1000 rads) their progeny as first instar larvae. P[w+]90E is a white+ P element inserted in polytene region 90E (P[(w, ry)D]3;Levis et al., 1985). w1118; fafFO8/fafFO8 eye clones were observed in ∼1/30 flies.

Isolation of faf genomic and cDNA

Unless otherwise noted, all molecular biology was carried out using standard procedures (Sambrook et al., 1989). The starting point for an 83 kb chromosomal walk (see Pirrotta, 1986) through the faf locus was a 5.2 kb Drosophila genomic DNA fragment adjacent to a P element in polytene region 100D (a gift of the Rubin laboratory enhancer trappers) isolated by ‘genomic rescue’ (Pirrotta, 1986). We walked within a non-amplified Sau3A-partial Drosophila genomic DNA library in bacteriophage λgem12 (Promega), constructed from genomic DNA of the isogenic st strain in which the faf alleles were originally induced. The walk was monitored by using the most distal recombinant phage iso-lated from each step as a probe for in situ hybridization to fafBX4/+ polytene chromosomes. After a phage hybridized to both ends of the fafBX4 inversion, we walked another two steps and then attempted to locate, within the walk, DNA lesions associated with each of the homozygous viable faf alleles. Blots of genomic DNA prepared from flies homozygous for each faf allele and restricted with either EcoRI or BamHI were sequentially hybridized with recombinant phage spanning the entire 83 kb walk, and lesions associated with fafBX4 and fafFBB12 were identified.

Seven different partial faf cDNAs were isolated from an eye disc cDNA library (constructed by Alan Cowman) using a 4.5 kb EcoRI fragment of genomic DNA as a hybridization probe. All of the cDNAs were subcloned into the Bluescript plasmid (Strat-agene) as EcoRI fragments. The relationships between the seven cDNAs were determined by hybridizing DNA blots of cDNA restriction fragments with probes prepared from each cDNA. The cDNAs were mapped onto the genomic DNA by using cDNAs as hybridization probes to blots of restricted recombinant phage clones spanning the walk.

DNA sequencing and analysis

The DNA sequence of cDNAs 3-2 (∼3.1 kb) and 7-3 (∼5.8 kb), subcloned into the Bluescript plasmid (Stratagene), were deter-mined on one strand from a plasmid primer by constructing nested sets of deletions with exonuclease III and S1 nuclease (Erase-A-Base Kit, Promega) and using Sequenase (USB). The sequence of the 3′-most 0.5 kb of cDNA 6-5 was determined starting with a plasmid primer and continually synthesizing 17-mer oligonu-cleotide primers corresponding to the end of each sequence. The opposite strand of the genomic DNA sequence was determined using 17-mer oligonucleotide primers synthesized based on the cDNA sequence. Additional primers were synthesized to sequence through introns. DNA sequences were compiled and analyzed using the MacVector programs. Database searches for nucleic acid and protein sequence similarities were performed at the NCBI using the BLAST network service (PIR, SwissProt, Genpept, and GPUpdate databases).

Plasmid constructions

DNA manipulations were carried out using standard procedures (Sambrook et al., 1989). The three genomic DNA plasmids (fafI, fafII, fafIII) were constructed by piecing together subcloned EcoRI and BamHI fragments of various recombinant λ phage clones, and then cloning them as NotI fragments into the P element transfor-mation vector pW8 (Klemenz et al., 1987) in the same orienta-tion as the white gene. To construct faf/lacZ, faf coding sequences ending at the BglII site after the first 392 amino acids were joined to a BamHI site at the start of the lacZ coding region. A 4.2 kb XbaI-BglII fragment of fafIII (the 5′-most 4.2 kb with an engi-neered XbaI site at the EcoRI site) was ligated with a 4.0 kb BamHI-EcoRI fragment of pC4βgal (Thummel et al., 1988; con-taining lacZ coding sequences and poly(A) addition sequences from SV40), into a plasmid vector restricted with XbaI and EcoRI. The resulting 8.2 kb faf/lacZ fragment was then removed as a NotI fragment and cloned into pW8. fafnuc/lacZ was generated from faf/lacZ by cloning sequences encoding a nuclear localization signal (KKKRKV; Kalderon et al., 1984) in frame into the XhoI site, 54 amino acids downstream of the ATG. First, two 26-mer oligonucleotides were annealed (5′-TCGAGGGCTAGCGGT-TAACAGATCTC and 5′-TCGAGAGATCTGTTAACCGCTAGCCC) and cloned into the XhoI site as an adaptor for another set of oligonucleotides containing the nuclear localization signal. The plasmid was then restricted within the adaptor with NheI and BglII, and annealed 44-mer oligonucleotides with one XbaI and one BamHI end (5′-CTAGAAACATGACCCCCCCCAAGAA-GAAGCGCAAGGTGGAGGAC and 5′-GATCGTCCTCCAC-CTTGCCGCTTCTTCTTGGGGGGGGTCATGTTT) were ligated into the adaptor. In total, 19 amino acids were added.

RNA blot analysis

Total nucleic acid was prepared from 5 ovaries or 20 eye disc complexes by dissecting the tissues in PBS and immediately trans-ferring them to a microfuge tube containing 200 μl of homo-genization buffer (50 mM Tris pH 7.5, 50 mM NaCl, 10 mM EDTA, 0.5% SDS, 2 mg/ml proteinase K). The tissues were homogenized with a small teflon homogenizer, incubated at 37°C for 45 minutes and extracted with phenol/chloroform, which was then back-extracted with homogenization buffer. The pooled supernatants were ethanol precipitated, the pellets washed with 100% ethanol and resuspended in formaldehyde gel loading buffer (Sambrook et al., 1989). The total nucleic acid preparations were electrophoresed on formaldehyde gels, blotted and hybridized with 32P-labeled, double-stranded DNA probes according to standard procedures (Sambrook et al., 1989).

P element transformation

w1118 embryos were injected according to standard procedures (Spradling, 1986) with each of the faf genomic DNA plasmids and faf-lacZ hybrid plasmids at 300 μg/ml, and the helper plas-mid pπ25.7Δ2-3 at 100 μg/ml. Plasmids for injections were puri-fied on CsCl-EtBr equilibrium density gradients (Sambrook et al., 1989). Transgenic flies were identified by their red eye color, mapped to a chromosome and balanced using standard genetic crosses.

Sections and scanning electron microscopy of adult eyes

Eyes were fixed, embedded in plastic and sectioned essentially as described in Tomlinson and Ready, 1987b. Heads were bisected to ensure penetration of fix. Sections were 1 μm. Scanning electron microscopy was performed as described in Kimmel et al., 1990.

β-galactosidase histochemistry

The procedures followed for β-galactosidase histochemical stain-ing are essentially those of Y. Hiromi and M. Mlodzik (personal communication). Ovaries and eye discs were dissected in PBS, fixed in a depression slide for 15 minutes at room temperature in 1% glutaraldehyde in PBS, and then washed with PBS at least twice for 10 minutes. The wash was replaced by prewarmed (65°C for at least 15 minutes) staining solution (10 mM NaPO4 (pH 7.2), 150 mM NaCl, 1 mM MgCl2, 3 mM K4[Fe(CN)6], 3 mM K3[Fe(CN)6]) and then by prewarmed staining solution to which 1/30 volume of 8% X-Gal in DMSO was added. Ovaries were incubated for 5 hours or less and eye discs overnight at room tem-perature. The staining reaction was stopped by several rinses in PBS and the tissues were then transferred to 80% glycerol in PBS, and mounted in this solution. Ovaries and eye discs from wild-type (non-transgenic) flies were used as negative controls.

Embryos were dechorionated with 50% bleach, washed with H2O and fixed in a depression slide for 15 minutes in 0.5 ml hep-tane saturated with 25% glutaraldehyde in PBS. Embryos were transferred to a glass slide and excess heptane blotted. After the heptane evaporated, embryos were transferred to double-sided tape on another glass slide by gently sandwiching the embryos between two slides. Embryos were then covered with PBS, devitellinized with a tungsten needle and transferred to a depression slide. After removing the PBS, embryos were incubated at room temperature in staining solution (above) without X-Gal for 5 minutes, and then in staining solution with X-Gal. faf/lacZ embryos began to stain almost immediately. Stained embryos were quickly rinsed in 70% and then 100% ethanol and mounted in 80% glycerol in PBS.

Immunostaining eye discs

Third instar larval eye disc complexes, attached only to the mouth hooks, were dissected in PBS. All antibody incubations and washes between them were performed in 96-well microtiter dishes, and eye discs were transferred from well to well with tungsten hooks (A. Tomlinson, personal communication). For mAb22C10, discs were treated exactly as described in Tomlinson and Ready, 1987a. mAbro staining was performed by a modification of the procedure in Kimmel et al., 1990. Discs were fixed for 45-55 min-utes on ice (0.1 M Pipes (pH 7.0), 2.0 mM EGTA, 1.0 mM MgSO4, 4% paraformaldehyde), washed for at least 15 minutes on ice in PBS+0.1% Triton X-100 (PBST), and then incubated in mAbro diluted in PBST for at least 1 hour at 4°C. After three 5 minute washes in PBST, discs were transferred to biotin-conju-gated goat anti-mouse secondary antibody (Jackson) diluted in PBST and incubated for at least 1 hour at 4°C. Following three 5 minute washes in PBST, discs were transferred to ABC com-plex (Vectastain Elite Kit, Vector) diluted in PBST for exactly 1 hour, washed three times for 5 minutes in PBST and put on ice. Discs were incubated in HRP-staining solution (1× PBS, 0.1% Triton X-100, 0.5 mg/ml DAB, 0.02% CoCl2, 0.02% NiCl2, 0.003% H2O2) for ∼1 minute while monitoring the level of stain-ing with a dissecting microscope, and then transferred to 1× PBS. Discs were transferred to 30% ethanol, the peripodial membranes were removed and then the discs were dehydrated (50, 70, 90, 2× 100% ethanol) and mounted in DPX (Fluka). faf/lacZ and fafnuc/lacZ eye discs were stained with a monoclonal antibody to β-galactosidase (Promega) by the same procedure used for mAbro staining, except discs were dissected in 0.1 M sodium phosphate (pH 7.2), PLP fix was used (Tomlinson and Ready, 1987a) and 0.1 M sodium phosphate +0.2% saponin was substituted for PBST in all subsequent steps. Convincing staining of faf-lacZ protein behind the morphogenetic furrow required the biotin intermediate amplification step. AE127 eye discs were similarly stained except that the biotin amplification step was omitted and instead, an HRP-conjugated goat anti-mouse secondary antibody was used (Biorad).

Immunostaining embryos

Embryos were dechorionated in 50% bleach, washed in H2O and fixed by gentle shaking for 20 minutes in 1 part fix (0.1 M Pipes (pH 7.0), 2 mM MgSO4, 1 mM EGTA, 4% formaldehyde) : 3 parts heptane. The fix was removed and embryos devitellinized by adding an equal volume of methanol and shaking hard. Devitellinized embryos were rinsed twice with methanol, then washed three times for 30 minutes with BBT (1× PBS, 0.1% BSA, 0.1% Tween20) and incubated at 4°C overnight with preadsorbed primary antibody (anti-vas polyclonal rabbit sera or anti-hb poly-clonal rat sera) diluted in BBT. Embryos were then washed four times for 30 minutes with BNT (PBS + 0.2% goat serum) and incubated for 2 hours in preadsorbed biotin-conjugated goat anti-rabbit (for anti-vas) or biotin-conjugated goat anti-rat (for anti-hb) secondary antibody (Jackson) diluted in BBT. After at least three 30 minute washes in PBT (PBS + 0.1% Tween 20) embryos were incubated for exactly 1 hour in ABC complex (Vectastain Elite Kit, Vector), washed three times in PBT, then put on ice. Embryos were incubated for a few minutes in HRP staining solution (0.5 mg/ml DAB +0.006% H2O2 in PBT) and the reaction stopped by washing with PBT. Embryos were dehydrated with ethanol (30, 50, 70, 90, 95, 2× 100%), rinsed for 30 seconds in acetone and mounted in Araldite (Ladd).

Hoechst and phalloidin staining embryos

Embryos were double-stained with Hoechst 33258 (Sigma) and rhodamine-conjugated phalloidin (Molecular Probes) exactly as described in Wieschaus and Nüsslein-Volhard (1986).

RNA in situ hybridization

RNA was detected in whole embryos and ovaries using digoxi-genin-labeled DNA probes exactly as described in Ephrussi et al. (1991), except that the methanol/DMSO step was omitted for the embryos, the levamisole was omitted from the developing buffer, and the hybridization buffer was 50% formamide, 5× SSC, 100 μg/ml sonicated salmon sperm DNA, 50 μg heparin, 0.1% Tween 20. Digoxigenin-labeled double-stranded DNA probes were pre-pared from 200 ng of DNA as described in Ephrussi et al. (1991). The probes were a 2.1 kb EcoRI fragment of faf cDNA 3-2, a 1.9 kb fragment of nos cDNA from pN5 (Wang and Lehmann, 1991), a 0.8 kb fragment of a cyclinB cDNA and linear Bluescript plas-mid (Stratagene) as a negative control. Embryos and ovaries were dehydrated and mounted using exactly the same procedure as for the immunostained embryos.

Identification of faf alleles

The two original faf alleles (fafFO8 and fafFBB12) were iden-tified as mutations causing rough eyes in an extensive screen of the Drosophila third chromosome for recessive viable mutations with abnormal eye morphology (J.A.F.-V., R.W. Carthew and G.M.R., unpublished data; see Fischer-Vize et al., 1992). Subsequently, 20 additional alle-les, generated chemically, with X-rays, or by hybrid dysgenesis (Table 1), were identified as rough-eyed flies car-rying the mutagenized chromosome in trans to fafFO8 (Materials and methods). Based on the severity of eye roughness of homozygotes, the faf alleles were categorized as strong or weak (Table 1 and see below).

Table 1.

fat facets alleles

fat facets alleles
fat facets alleles

Meiotic mapping positioned faf very close to the tip of chromosome 3R (Materials and methods). faf was specifi-cally localized to polytene region 100E by analyzing the polytene chromosomes of several faf alleles (Table 1) and finding that fafBX4 is an inversion with breakpoints in poly-tene regions 98D and 100E, and fafBP is deficient for the entire 100EF region (data not shown).

The faf mutant phenotype

Eye phenotype

The eyes of faf mutant flies appear rough due to irregular-ities in the normally precise hexagonal array of facets (Fig. 1A,B). In order to investigate the faf mutant phenotype beneath the eye surface, the retinas of faf flies were exam-ined in tangential sections. The wild-type retina is a hexag-onal lattice of identical facets, or ommatidia (Fig. 1C). An ommatidium has eight photoreceptor cells (R-cells), each uniquely positioned within a trapezoidal array. There are six outer photoreceptor cells, R1-R6, with large light-gath-ering organelles, called rhabdomeres, and two inner R-cells, R7 and R8, with small rhabdomeres. The trapezoids are ori-ented in one direction and are symmetrical with respect to an equator. Many defects are present in strong faf mutant retinas, the most striking of which is the appearance of one, two or occasionally three extra outer photoreceptor cells in most facets (Fig. 1D and legend).

Fig. 1.

faf mutant eye phenotype. (A,B) Scanning electron micrographs (Materials and methods) of wild-type and fafFO8 eyes, respectively, show that the hexagonal facet array of faf mutant eyes is irregular and that bristles are sometimes absent or multiplied. (C,D) Tangential sections (Materials and methods) through wild-type and fafFO8 eyes, respectively. Photoreceptor cells may be identified by their rhabdomeres, which appear as darkly staining circles. Photoreceptor cell R7 is visible in this plane of section. The most obvious defect in the faf mutant retina (D) is the appearance of one or two extra outer photoreceptor cells in most facets (∼88%). There are also facets with too few photoreceptors (∼2%), facet fusions, rhabdomere malformations and wild-type facets (∼10%). In addition, the orientations of the facets are non-uniform. The bar (in D) is 20 μm in C,D, and 100 μm in A,B.

Fig. 1.

faf mutant eye phenotype. (A,B) Scanning electron micrographs (Materials and methods) of wild-type and fafFO8 eyes, respectively, show that the hexagonal facet array of faf mutant eyes is irregular and that bristles are sometimes absent or multiplied. (C,D) Tangential sections (Materials and methods) through wild-type and fafFO8 eyes, respectively. Photoreceptor cells may be identified by their rhabdomeres, which appear as darkly staining circles. Photoreceptor cell R7 is visible in this plane of section. The most obvious defect in the faf mutant retina (D) is the appearance of one or two extra outer photoreceptor cells in most facets (∼88%). There are also facets with too few photoreceptors (∼2%), facet fusions, rhabdomere malformations and wild-type facets (∼10%). In addition, the orientations of the facets are non-uniform. The bar (in D) is 20 μm in C,D, and 100 μm in A,B.

The retinas of all strong faf mutants appear similar (data not shown). The retinas of all weak faf alleles are also sim-ilar to each other and generally have as few as 1% mutant facets (data not shown; Table 1 legend). The eye defects of the strong alleles are likely to represent the complete loss-of-function (null) eye phenotype of faf, as the strong mutants (including the viable inversion fafBX4 that breaks within the faf transcript (see below)) show the same pheno-type when homozygous or in trans to the fafBP deficiency (data not shown).

Maternal effect phenotype

All of the strong faf alleles and three of the weak alleles cause female sterility (Table 1 and legend). Females homozygous for any of these alleles have apparently normal ovaries, but lay eggs that never form cuticle and never hatch. In order to examine the mutant embryos in more detail, embryos of wild-type and homozygous fafFO8 moth-ers were stained consecutively with the DNA stain Hoechst to reveal nuclei, and with rhodamine-conjugated phalloidin, which stains f-actin and thus allows visualization of cell membranes (Materials and methods). No normal embryos from fafFO8 mothers were ever observed. After the fertil-ized egg is laid, the nucleus normally undergoes 14 syn-chronous cycles of division prior to cellularization (see Foe and Alberts, 1983). By division cycle 10, the nuclei have migrated to the egg periphery to form the ‘syncytial blas-toderm’ (Fig. 2A) and the primordial germ cells, called pole cells, located at the posterior pole of the embryo, form (Fig. 2B). The rest of the nuclei continue synchronous divisions until cycle 14, when cellularization produces the ‘cellular blastoderm’. In wild-type embryos, f-actin is associated with each nucleus at syncytial blastoderm (Fig. 2B), and then with cell membranes at cellular blastoderm (Fig. 2F; Karr and Alberts, 1986). In embryos from faf mutant moth-ers (hereafter called faf mutant embryos), no normal syn-cytial blastoderm embryos are observed. The embryo shown in Fig. 2C,D appears to be at least in cycle 10 because pole cells are present (Fig. 2D), but most of the nuclei have not migrated to the periphery (Fig. 2C). Also commonly observed were embryos in which patches of asynchronously dividing nuclei had migrated to the periphery (Fig. 2G). Except for the pole cells (which are fewer in number and spread out instead of grouped together as in wild-type -see Fig. 11D, below), no cellularization of faf mutant embryos was ever observed (Fig. 2H).

Fig. 2.

The phenotype of embryos from faf mutant mothers. Embryos, collected after 4 hours of egg-laying, from wild-type (A,B,E,F) and fafFO8 mutant (C,D,G,H) mothers were labeled consecutively with the DNA stain Hoechst and rhodamine-conjugated phalloidin, which stains f-actin, and photographed using fluorescent light with either fluorescein (A,C,E,G) or rhodamine (B,D,F,H) filters to show nuclei and f-actin filaments, respectively (Materials and methods). The same embryo is shown in each of the horizontal pairs of panels. (A,B) A wild-type embryo fixed during the mitosis that concludes cycle 11. Nuclei are at the periphery (A), and as cellularization does not occur until cycle 14 (below), f-actin is seen associated with each nucleus (B; see Karr and Alberts, 1986). The posterior of the embryo at a deeper focal plane is shown to the right of B, where the pole cells (pc), which form at cycle 10, are visible. (C,D) A faf mutant embryo, in which few nuclei have migrated to the periphery (C) although pole cells have formed (D), indicating that cycle 10 has already been reached. (E,F) A wild-type embryo at cycle 14 (cellular blastoderm). Cellularization is apparent as f-actin outlines the cell membranes (F). (G,H) The most mature-appearing faf mutant embryos seen. Patches of asynchronously dividing nuclei have migrated to the periphery (G), but no cellularization is apparent (H).

Fig. 2.

The phenotype of embryos from faf mutant mothers. Embryos, collected after 4 hours of egg-laying, from wild-type (A,B,E,F) and fafFO8 mutant (C,D,G,H) mothers were labeled consecutively with the DNA stain Hoechst and rhodamine-conjugated phalloidin, which stains f-actin, and photographed using fluorescent light with either fluorescein (A,C,E,G) or rhodamine (B,D,F,H) filters to show nuclei and f-actin filaments, respectively (Materials and methods). The same embryo is shown in each of the horizontal pairs of panels. (A,B) A wild-type embryo fixed during the mitosis that concludes cycle 11. Nuclei are at the periphery (A), and as cellularization does not occur until cycle 14 (below), f-actin is seen associated with each nucleus (B; see Karr and Alberts, 1986). The posterior of the embryo at a deeper focal plane is shown to the right of B, where the pole cells (pc), which form at cycle 10, are visible. (C,D) A faf mutant embryo, in which few nuclei have migrated to the periphery (C) although pole cells have formed (D), indicating that cycle 10 has already been reached. (E,F) A wild-type embryo at cycle 14 (cellular blastoderm). Cellularization is apparent as f-actin outlines the cell membranes (F). (G,H) The most mature-appearing faf mutant embryos seen. Patches of asynchronously dividing nuclei have migrated to the periphery (G), but no cellularization is apparent (H).

In summary, the most obvious defects in faf null mutants are in the eye and in the embryo. In the eye, the most striking abnormality is the appearance of ectopic outer pho-toreceptors. faf mutant embryos never reach normal syncy-tial blastoderm and except for some pole cells, never cel-lularize.

Developmental defects in faf mutant larval eye discs

To determine if defects in early stages of ommatidial assem-bly contribute to the faf eye phenotype, wild-type and strong faf mutant larval eye discs were stained with the neural specific antibody mAb22C10 (Fujita et al., 1982). mAb22C10 reveals the sequence of photoreceptor cell assembly (R8, R2/5, R3/4, R1/6, R7) as each R-cell expresses the mAb22C10 antigen when it acquires neural identity (Fig. 3A; Tomlinson and Ready, 1987a). As omma-tidial development proceeds in a posterior-to-anterior wave in the eye disc, facets at all stages of photoreceptor cell assembly are observed in one larval disc. The most striking defect observed in stained faf mutant larval eye discs is the appearance of one or two ectopic photoreceptor cells in most developing facets at the time when R3/4 normally begin to stain with mAb22C10 (Fig.3B,C). These cells may be the mystery cells, which are normally positioned between R3 and R4 in a precluster that emerges from the morphogenetic furrow, but then disappear into the surrounding pool of undifferentiated cells without expressing neural antigens (Tomlinson and Ready, 1987a; Wolff and Ready, 1991a).

Fig. 3.

The mystery cells become ectopic photoreceptors in faf mutant eye discs. Wild-type (A) and fafFO8 (B,C) larval eye discs were stained with mAb22C10 (Materials and methods) to reveal the sequence of photoreceptor cell assembly. mAb22C10 stains photoreceptor cell cytoplasm and the discs are photographed to show the apical tips of the photoreceptor cells near the morphogenetic furrow. The arrows point along the direction of the morphogenetic furrow. Five-cell clusters containing R8,2,5,3,4 are apparent in wild-type (A) discs. At a similar distance from the furrow in faf mutant discs (B), clusters containing 6 or 7 cells are seen. The cell(s) which appear to be the extra photoreceptors (see below), situated between R3 and R4, are indicated by asterisks (*). Improperly spaced clusters (C) also appear occasionally in faf mutant discs. (D) faf mutant larval eye discs stained with an antibody to the ro protein (Materials and methods), which is expressed in the nuclei of R2,5,3,4. One or two ectopic nuclei (*), apparently located between R3 and R4, express ro. (E) faf mutant larval eye discs carrying one copy of the enhancer trap P element AE127, which is inserted in the svp gene and expresses β-galactosidase in the nuclei of R3,4,1,6, were stained with an antibody to β-galactosidase (Materials and methods). A cluster is shown containing an ectopic nucleus (*), apparently between R3 and R4. As the mystery cells are normally located between the R3/4 pair, the ectopic photoreceptor cells are likely to be improperly determined mystery cells. ro and svp appear to be otherwise appropriately expressed, suggesting that many of the photoreceptor cells in faf mutant discs acquire their normal identities. The bar (in C) is 10 μm in all panels.

Fig. 3.

The mystery cells become ectopic photoreceptors in faf mutant eye discs. Wild-type (A) and fafFO8 (B,C) larval eye discs were stained with mAb22C10 (Materials and methods) to reveal the sequence of photoreceptor cell assembly. mAb22C10 stains photoreceptor cell cytoplasm and the discs are photographed to show the apical tips of the photoreceptor cells near the morphogenetic furrow. The arrows point along the direction of the morphogenetic furrow. Five-cell clusters containing R8,2,5,3,4 are apparent in wild-type (A) discs. At a similar distance from the furrow in faf mutant discs (B), clusters containing 6 or 7 cells are seen. The cell(s) which appear to be the extra photoreceptors (see below), situated between R3 and R4, are indicated by asterisks (*). Improperly spaced clusters (C) also appear occasionally in faf mutant discs. (D) faf mutant larval eye discs stained with an antibody to the ro protein (Materials and methods), which is expressed in the nuclei of R2,5,3,4. One or two ectopic nuclei (*), apparently located between R3 and R4, express ro. (E) faf mutant larval eye discs carrying one copy of the enhancer trap P element AE127, which is inserted in the svp gene and expresses β-galactosidase in the nuclei of R3,4,1,6, were stained with an antibody to β-galactosidase (Materials and methods). A cluster is shown containing an ectopic nucleus (*), apparently between R3 and R4. As the mystery cells are normally located between the R3/4 pair, the ectopic photoreceptor cells are likely to be improperly determined mystery cells. ro and svp appear to be otherwise appropriately expressed, suggesting that many of the photoreceptor cells in faf mutant discs acquire their normal identities. The bar (in C) is 10 μm in all panels.

In order to determine if the ectopic photoreceptors are indeed positioned between R3 and R4, we examined the expression of the rough (ro;Tomlinson et al., 1988; Saint et al., 1988) and seven-up (svp;Mlodzik et al., 1990b) genes in faf larval eye discs. Both ro and svp are normally expressed in R3 and R4 (Kimmel et al., 1990; Mlodzik et al., 1990b; Fig.3 legend). In faf mutant eye discs, both genes are also expressed in one or two cells adjacent to R3 and R4. Thus, the ectopic R-cells in faf mutant larval eyes are likely to be the mystery cells.

faf+function is required outside normal and ectopic photoreceptors

In order to determine which cells require faf+ function to exclude the ectopic R-cells from the precluster, marked (w ) clones of fafFO8 mutant cells in wild-type eyes were generated using X-ray induced somatic recombination (Materials and methods). Within the patch of w faf cells, the retina shows the faf mutant phenotype, and outside the w clone, the retina appears wild type (Fig.4A). As there are no lineage restrictions on cells within particular facets (Ready et al., 1976; Lawrence and Green, 1979; Wolff and Ready, 1991b), there are mosaic ommatidia at the clone borders, containing both w+faf+ and w faf photoreceptor cells (Fig. 4A). Some of the mosaic facets are phenotypi-cally wild type (that is, there are two inner and six outer R-cells) and some are typical faf mutant facets, with more than six outer R-cells (Fig.4A).

Fig. 4.

faf+ function is required outside of the photoreceptor cells. (A) A tangential section through a typical w faf clone in a w+faf+ eye, generated by X-ray-induced somatic recombination (Materials and methods). The w clone is recognizable by the absence of pigment granules in the pigment cells forming the hexagonal lattice, and also near each photoreceptor (black dots). The large arrow points to the w+ ectopic cell in a phenotypically mutant mosaic facet at the clone border. The small arrow indicates a phenotypically wild-type facet at the clone border. (B) A tabulation of data obtained by examining both phenotypically wild-type and phenotypically mutant mosaic facets in many different clones. The R-cell genotypes in 209 wild-type mosaic facets in 20 separate clones were scored. (For technical reasons, a smaller number of R8 cells was scored.) As there are normally ∼10% wild-type facets in an faf mutant eye, if there were no requirement for faf in particular R-cells, those R-cells would be expected to be w+faf+ at a frequency of ∼55% [50%+(1/2)(10%)]. In contrast, if faf were required in particular R-cells in order to form a wild-type facet, those R-cells would be expected to be faf+ at a frequency of ∼95% [100%-(1/2)(10%)=95%]. The data indicate that all R-cells are faf+ at a similar frequency, averaging at 65%. This slightly higher than random (55%) frequency suggests that faf is required in cells closely related by lineage to (near to) the R-cells (see Tomlinson et al., 1988; Reinke and Zipursky, 1988; Carthew and Rubin, 1990; Mlodzik et al., 1990b.) Phenotypically mutant facets were also examined, and the tabulation of the frequency of w+faf+ outer R-cells in 78 mosaic facets at the borders of 18 different clones is shown. In contrast to the results obtained with the wild-type mosaic facets, the frequency of faf+ outer R-cells in the mutant mosaic facets (35%) is lower than random [50%-(1/2)(10%)=45%], suggesting that either R-cells within these facets or cells related by lineage to them must be faf in order to produce a phenotypically mutant facet. (C-I) Phenotypically mutant mosaic facets in which one or two extra outer photoreceptors are neatly added. In these facets, knowing that the extra cells are the mystery cells, located between R3 and R4, it is possible to identify all of the outer R-cells. We assume that no gross rearrangement of the R-cells occurs in the eye disc after the assembly stages that we have observed. The ectopic cell (labeled ‘m’ for mystery cell) and also sometimes either R3, R4 or both, are w+faf+. Only the w+ R3,4, and m-cells are labeled. The facets in E and F also contain w+ R8’s (not visible in this plane of section). The facet in H has two ectopic R-cells, only one of which is w+. These facets show that faf expression in the ectopic cells or in the cells they contact (R8,3,4) is not sufficient to exclude the mystery cells from the facet. We observed a total of five facets like those shown in C and D, six additional facets in which the ectopic cell was w+, and four facets where R8 could be scored definitively as w+. No phenotypically mutant facets were observed in which all R-cells were w+. We interpret this as a reflection of the proximity of the faf-requiring cells to the mystery cells, and thus the low probability of the cell(s) requiring faf+ function being w while the mystery cells and all of the other R-cells in a mutant facet are w+.

Fig. 4.

faf+ function is required outside of the photoreceptor cells. (A) A tangential section through a typical w faf clone in a w+faf+ eye, generated by X-ray-induced somatic recombination (Materials and methods). The w clone is recognizable by the absence of pigment granules in the pigment cells forming the hexagonal lattice, and also near each photoreceptor (black dots). The large arrow points to the w+ ectopic cell in a phenotypically mutant mosaic facet at the clone border. The small arrow indicates a phenotypically wild-type facet at the clone border. (B) A tabulation of data obtained by examining both phenotypically wild-type and phenotypically mutant mosaic facets in many different clones. The R-cell genotypes in 209 wild-type mosaic facets in 20 separate clones were scored. (For technical reasons, a smaller number of R8 cells was scored.) As there are normally ∼10% wild-type facets in an faf mutant eye, if there were no requirement for faf in particular R-cells, those R-cells would be expected to be w+faf+ at a frequency of ∼55% [50%+(1/2)(10%)]. In contrast, if faf were required in particular R-cells in order to form a wild-type facet, those R-cells would be expected to be faf+ at a frequency of ∼95% [100%-(1/2)(10%)=95%]. The data indicate that all R-cells are faf+ at a similar frequency, averaging at 65%. This slightly higher than random (55%) frequency suggests that faf is required in cells closely related by lineage to (near to) the R-cells (see Tomlinson et al., 1988; Reinke and Zipursky, 1988; Carthew and Rubin, 1990; Mlodzik et al., 1990b.) Phenotypically mutant facets were also examined, and the tabulation of the frequency of w+faf+ outer R-cells in 78 mosaic facets at the borders of 18 different clones is shown. In contrast to the results obtained with the wild-type mosaic facets, the frequency of faf+ outer R-cells in the mutant mosaic facets (35%) is lower than random [50%-(1/2)(10%)=45%], suggesting that either R-cells within these facets or cells related by lineage to them must be faf in order to produce a phenotypically mutant facet. (C-I) Phenotypically mutant mosaic facets in which one or two extra outer photoreceptors are neatly added. In these facets, knowing that the extra cells are the mystery cells, located between R3 and R4, it is possible to identify all of the outer R-cells. We assume that no gross rearrangement of the R-cells occurs in the eye disc after the assembly stages that we have observed. The ectopic cell (labeled ‘m’ for mystery cell) and also sometimes either R3, R4 or both, are w+faf+. Only the w+ R3,4, and m-cells are labeled. The facets in E and F also contain w+ R8’s (not visible in this plane of section). The facet in H has two ectopic R-cells, only one of which is w+. These facets show that faf expression in the ectopic cells or in the cells they contact (R8,3,4) is not sufficient to exclude the mystery cells from the facet. We observed a total of five facets like those shown in C and D, six additional facets in which the ectopic cell was w+, and four facets where R8 could be scored definitively as w+. No phenotypically mutant facets were observed in which all R-cells were w+. We interpret this as a reflection of the proximity of the faf-requiring cells to the mystery cells, and thus the low probability of the cell(s) requiring faf+ function being w while the mystery cells and all of the other R-cells in a mutant facet are w+.

Examination of both the phenotypically wild-type and mutant mosaic facets shows that faf+ function is not required in photoreceptor cells. First, the genotypes of the different photoreceptors in phenotypically wild-type mosaic ommatidia were scored to determine if there is a tendency for particular photoreceptors to be faf+. None of the eight photoreceptors was faf+ nearly as frequently as expected if faf+ function is required in a particular photoreceptor cell (Fig. 4B and legend). Second, we find that in the pheno-typically mutant facets, the ectopic photoreceptor cells are not always faf , and the R-cells that they contact (R3, R4 and R8) are also not always faf , consistent with the analy-sis of the wild-type mosaics (above). Thus, it is not the absence of faf+ function in the ectopic cells, or the cells they contact, that results in their misdetermination as pho-toreceptors.

The analysis of phenotypically wild-type and mutant facets suggests that faf is required in cells closely related by lineage to (that is, near to) the photoreceptors. Each of the photoreceptors in the phenotypically wild-type mosaic ommatidia are faf+ at a higher than random frequency (Fig. 4B and legend). In contrast, the outer photoreceptors within the mutant mosaic ommatidia are faf+ at a lower than random frequency (Fig. 4B and legend).

In summary, we interpret these observations as indicat-ing that cells near to, but outside the normal or ectopic pho-toreceptors in a particular facet must be faf+ in order to pre-vent the neuralization of the mystery cells.

Cloning the faf gene

Having localized faf to polytene region 100E (above), we isolated Drosophila genomic DNA adjacent to a P element inserted in 100D (Materials and methods) and used this DNA as a starting point for an 84 kb chromosomal walk within a bacteriophage lambda Drosophila genomic DNA library, towards the telomere of 3R (Materials and meth-ods, Fig. 5A). Progress in the walk was monitored by hybridizing, in situ, phage clone probes from each step to polytene chromosomes of the homozygous viable inversion fafBX4 (Table 1, Materials and methods). A phage clone hybridized to both ends of the inversion about 55 kb into the walk (Fig. 5A; data not shown). An additional 29 kb of genomic DNA was isolated distal to the inversion break-point. The fafBX4 breakpoint was localized to a 1.8 kb EcoRI-BamHI fragment of genomic DNA and a small dele-tion in a neighboring 1.5 kb BamHI fragment was found to be associated with fafFBB12 (Fig. 5A; data not shown -see Materials and methods). A 4.5 kb EcoRI fragment of genomic DNA containing the region of both mutant lesions was used to isolate many cDNAs from an eye imaginal disc cDNA library (Materials and methods). All of the cDNAs are either identical to or contained within the three shown in Fig. 5A (Materials and methods). DNA sequencing (below) of the three cDNAs as indicated in Fig. 5A revealed that the cDNAs represent two transcripts of at least ∼8500 and ∼8900 nt, which differ in their 3′ ends. (We do not know whether the 5′ end of the cDNA 3-2 represents the actual start of the faf transcript.) This result correlates with an analysis of RNA blots (Fig. 5B) hybridized with the 4.5 kb EcoRI fragment, which shows two transcripts of slightly different size in the appropriate range. Conceptual transla-tion of the two transcripts reveals long open reading frames differing at their 3′ termini (see below).

Fig. 5.

Cloning the faf gene. (A) The thick black line is a restriction map (B, BamHI; R, EcoRI) of the fat facets locus within approximately 83 kb of cloned Drosophila genomic DNA, isolated by walking within a bacteriophage genomic DNA library from a P element inserted in polytene region 100D (Materials and methods). The numbers above the line represent the approximate sizes of the restriction fragments indicated. The hatched and stippled bars just beneath the line indicate the restriction fragments containing one fafBX4 inversion breakpoint and a small deletion associated with fafFBB12, respectively, determined by hybridizing recombinant phage probes to DNA blots of restricted genomic DNA from flies carrying these mutant alleles (Materials and methods). The line labeled ‘probe’ indicates the 4.5 kb EcoRI fragment of genomic DNA used to screen an eye disc cDNA library by hybridization (Materials and methods). Three of the seven different cDNAs isolated are shown as shaded bars extending approximately beneath the genomic regions to which they correspond. The lengths of the cDNAs are ∼3.1 kb (3-2), ∼5.8 kb (7-3) and ∼6.2 kb (6-5). The other four cDNAs isolated are partial products of those shown, as determined by DNA blot hybridization experiments (Materials and methods). The complete DNA sequences of 3-2 and 7-3 were determined, as well as ∼0.5 kb of the 3′ end and ∼0.2 kb at the 5′ end of 6-5, revealing a long open reading frame and two differentially spliced 3′ ends (Fig. 6; Materials and methods). The cDNA sequences were confirmed by sequencing the opposite strand of genomic DNA, which revealed seventeen exons, as shown at the top of the figure (Fig 6; Materials and methods). Due to alternate splicing, the sixteenth exon is shorter in 6-5 than in 7-3, and the seventeenth exon is contained only in 6-5. ATG and TAG indicate start and stop codons, respectively. The long hatched bars indicate the approximate extents of three different genomic DNA fragments (fafI, fafII and fafIII are ∼17, 14 and 13 kb, respectively) introduced into the Drosophila genome by P element transformation (Materials and methods). fafI extends from the indicated EcoRI site through 4.6 kb downstream of the indicated BamHI site, fafII extends between the two BamHI sites indicated, and fafIII extends between the indicated EcoRI and BamHI sites. All three genomic DNA fragments (2 lines of fafI, 3 lines of fafII and 1 line of fafIII) complement the rough eye and female infertility phenotypes of fafBX4 (Materials and methods). Although only one fafIII line obtained could be tested, we are confident that the ∼13 kb fafIII fragment contains the entire faf gene because its 5′ end is identical to that of faf I and its 3′ end is identical to that fafII. (B) An autoradiograph of an RNA blot revealing the two faf transcripts. Each lane contains total RNA prepared from 20 third instar larval eye imaginal disc complexes or 5 ovaries of each of the three Drosophila lines indicated (Materials and methods). The autoradiograph shown was following hybridization with a 32P-labelled double-stranded DNA probe corresponding to the 4.5 kb EcoRI fragment shown in (A). The bottom panel is a lighter exposure of the first four lanes. Identical results were obtained when the blot was stripped of old probe and hybridized with the upstream 3.5 kb EcoRI fragment or the downstream 6.1 kb BamHI fragment of DNA (see A). In order to approximate the sizes of the transcripts shown, the blot was also probed with labelled DNA corresponding to a spectrin gene (∼13 kb) and a non-muscle myosin gene (∼7 kb; Materials and methods). Both faf transcripts are expressed in wild-type and fafFO8 eye discs and ovaries. As neither the size nor the level of faf mRNA appears to be affected in fafFO8, this allele is likely to be a point mutation within the coding sequences. As expected (see A), only one truncated transcript is detected in fafBX4 RNA.

Fig. 5.

Cloning the faf gene. (A) The thick black line is a restriction map (B, BamHI; R, EcoRI) of the fat facets locus within approximately 83 kb of cloned Drosophila genomic DNA, isolated by walking within a bacteriophage genomic DNA library from a P element inserted in polytene region 100D (Materials and methods). The numbers above the line represent the approximate sizes of the restriction fragments indicated. The hatched and stippled bars just beneath the line indicate the restriction fragments containing one fafBX4 inversion breakpoint and a small deletion associated with fafFBB12, respectively, determined by hybridizing recombinant phage probes to DNA blots of restricted genomic DNA from flies carrying these mutant alleles (Materials and methods). The line labeled ‘probe’ indicates the 4.5 kb EcoRI fragment of genomic DNA used to screen an eye disc cDNA library by hybridization (Materials and methods). Three of the seven different cDNAs isolated are shown as shaded bars extending approximately beneath the genomic regions to which they correspond. The lengths of the cDNAs are ∼3.1 kb (3-2), ∼5.8 kb (7-3) and ∼6.2 kb (6-5). The other four cDNAs isolated are partial products of those shown, as determined by DNA blot hybridization experiments (Materials and methods). The complete DNA sequences of 3-2 and 7-3 were determined, as well as ∼0.5 kb of the 3′ end and ∼0.2 kb at the 5′ end of 6-5, revealing a long open reading frame and two differentially spliced 3′ ends (Fig. 6; Materials and methods). The cDNA sequences were confirmed by sequencing the opposite strand of genomic DNA, which revealed seventeen exons, as shown at the top of the figure (Fig 6; Materials and methods). Due to alternate splicing, the sixteenth exon is shorter in 6-5 than in 7-3, and the seventeenth exon is contained only in 6-5. ATG and TAG indicate start and stop codons, respectively. The long hatched bars indicate the approximate extents of three different genomic DNA fragments (fafI, fafII and fafIII are ∼17, 14 and 13 kb, respectively) introduced into the Drosophila genome by P element transformation (Materials and methods). fafI extends from the indicated EcoRI site through 4.6 kb downstream of the indicated BamHI site, fafII extends between the two BamHI sites indicated, and fafIII extends between the indicated EcoRI and BamHI sites. All three genomic DNA fragments (2 lines of fafI, 3 lines of fafII and 1 line of fafIII) complement the rough eye and female infertility phenotypes of fafBX4 (Materials and methods). Although only one fafIII line obtained could be tested, we are confident that the ∼13 kb fafIII fragment contains the entire faf gene because its 5′ end is identical to that of faf I and its 3′ end is identical to that fafII. (B) An autoradiograph of an RNA blot revealing the two faf transcripts. Each lane contains total RNA prepared from 20 third instar larval eye imaginal disc complexes or 5 ovaries of each of the three Drosophila lines indicated (Materials and methods). The autoradiograph shown was following hybridization with a 32P-labelled double-stranded DNA probe corresponding to the 4.5 kb EcoRI fragment shown in (A). The bottom panel is a lighter exposure of the first four lanes. Identical results were obtained when the blot was stripped of old probe and hybridized with the upstream 3.5 kb EcoRI fragment or the downstream 6.1 kb BamHI fragment of DNA (see A). In order to approximate the sizes of the transcripts shown, the blot was also probed with labelled DNA corresponding to a spectrin gene (∼13 kb) and a non-muscle myosin gene (∼7 kb; Materials and methods). Both faf transcripts are expressed in wild-type and fafFO8 eye discs and ovaries. As neither the size nor the level of faf mRNA appears to be affected in fafFO8, this allele is likely to be a point mutation within the coding sequences. As expected (see A), only one truncated transcript is detected in fafBX4 RNA.

In order to show more conclusively that the cDNAs cor-respond to the faf gene, transgenic Drosophila lines were generated by P element transformation with the three dif-ferent fragments of genomic DNA shown in Fig. 5A (Mate-rials and methods). One copy of any of the three genomic DNA fragments complements completely both the eye and female sterile mutant phenotypes of fafBX4 (data not shown; see Materials and methods). In addition, blots of ovary and eye disc RNA probed with all regions of the smallest com-plementing genomic DNA fragment (fafIII) reveal only the two transcripts that hybridize to the 4.5 kb EcoRI fragment of genomic DNA (Fig. 5A). Thus, these are the only two RNAs transcribed within fafIII that are present in eye discs or ovaries.

Predicted faf protein sequences

The DNA sequences of three overlapping partial cDNAs were determined (Fig. 5A and Materials and methods). In order to confirm the cDNA sequences, the corresponding genomic DNA sequence was determined on the opposite strand A composite sequence is shown in Fig. 6. The genomic DNA sequence shows that the faf gene is com-posed of at least 17 exons. Conceptual translation of the two classes of cDNAs reveals proteins of 2711 and 2747 amino acids, differing slightly in their carboxy termini due to an alternate splice near the 3′ end of the primary tran-script. Apart from their large size, the faf proteins have few remarkable features. At amino acids 262-290, there is a potential leucine zipper structural domain, shown to be a dimerization site within transcription factors (Landschulz et al., 1988). As the leucine repeat is not adjacent to a par-ticularly basic (DNA-binding) domain, faf is unlikely to be a transcription factor. In addition, there is a likely PEST sequence (Rogers et al., 1986), often found in rapidly degraded proteins, at amino acids 18-34. No other signifi-cant similarity with another protein or DNA sequence in multiple databases has been found (Materials and methods).

Fig. 6.

DNA and predicted protein sequences of the faf gene. A composite genomic and cDNA sequence of 11,975 nt of the faf locus is shown, including the region containing all three cDNAs shown in Fig. 5A, and 360 nt upstream. The sequenced nucleotides are numbered in plain type at the left, starting with 360 nt upstream of the beginning of cDNA 3-2, indicated by the G at position 361. cDNA 3-2 has 4 additional bases (CACG) upstream of that G that are not contained in the genomic sequence, which could either be an artifact of cDNA cloning, or could indicate an intron/exon boundary. We have not determined whether the G at nt 361 is the actual transcription start site. The predicted protein sequence is shown using the single letter amino acid code, the letters placed beneath the middle nucleotide of each codon. Stop codons are indicated with asterisks (*). The numbers in bold at the left refer to the protein sequence. The ATG at position 605 is probably the start of translation as there are no other ATGs upstream in the 3-2 cDNA sequence, there are stop codons in all three reading frames within the 3-2 cDNA sequence upstream of nt 605, and the sequences upstream of the ATG are a good match with the Drosophila translation initiation consensus sequence (Cavener, 1987). The PEST sequence similarity at amino acids 18-34 (Rogers et al., 1986) and the leucine-zipper motif at amino acids 269-290 (Landschulz et al., 1988) are underlined. Introns are in lowercase. All of the intron/exon boundaries conform to the splice junction consensus sequence. Efforts to determine the sequence of the intron located between nt 7503 and 7504 were abandoned because plasmid subclones containing this region and of a suitable size for sequencing were unstable and produced various deletions with at least one end within the intron. The sequence of this region was determined on both strands of two independent cDNAs (7-3 and 6-5). The c at nt 10383 is the 3′ end of cDNA 7-3. As 7-3 is not polyadenylated, we do not know the exact location of the actual 3′ end of this spliced form of the faf mRNA. cDNA 6-5, however, is polyadenylated just after the final nt shown. The C at nt 11578 indicates the end of the genomic DNA sequence determined. There are eight single base differences between the genomic and cDNAs, which are indicated by underlined, bold letters. The sequence shown is the genomic sequence, and the corresponding changes in the cDNA are G-T (2638), C-T (4347), G-C (7605), C-T (8500), T-C (8563), C-T (8980), T-C (9754), C-G (10111) and A-T (11504). All the differences are third base codon substitutions except for position 11565, which is a first base substitution changing the amino acid from S to T.

Fig. 6.

DNA and predicted protein sequences of the faf gene. A composite genomic and cDNA sequence of 11,975 nt of the faf locus is shown, including the region containing all three cDNAs shown in Fig. 5A, and 360 nt upstream. The sequenced nucleotides are numbered in plain type at the left, starting with 360 nt upstream of the beginning of cDNA 3-2, indicated by the G at position 361. cDNA 3-2 has 4 additional bases (CACG) upstream of that G that are not contained in the genomic sequence, which could either be an artifact of cDNA cloning, or could indicate an intron/exon boundary. We have not determined whether the G at nt 361 is the actual transcription start site. The predicted protein sequence is shown using the single letter amino acid code, the letters placed beneath the middle nucleotide of each codon. Stop codons are indicated with asterisks (*). The numbers in bold at the left refer to the protein sequence. The ATG at position 605 is probably the start of translation as there are no other ATGs upstream in the 3-2 cDNA sequence, there are stop codons in all three reading frames within the 3-2 cDNA sequence upstream of nt 605, and the sequences upstream of the ATG are a good match with the Drosophila translation initiation consensus sequence (Cavener, 1987). The PEST sequence similarity at amino acids 18-34 (Rogers et al., 1986) and the leucine-zipper motif at amino acids 269-290 (Landschulz et al., 1988) are underlined. Introns are in lowercase. All of the intron/exon boundaries conform to the splice junction consensus sequence. Efforts to determine the sequence of the intron located between nt 7503 and 7504 were abandoned because plasmid subclones containing this region and of a suitable size for sequencing were unstable and produced various deletions with at least one end within the intron. The sequence of this region was determined on both strands of two independent cDNAs (7-3 and 6-5). The c at nt 10383 is the 3′ end of cDNA 7-3. As 7-3 is not polyadenylated, we do not know the exact location of the actual 3′ end of this spliced form of the faf mRNA. cDNA 6-5, however, is polyadenylated just after the final nt shown. The C at nt 11578 indicates the end of the genomic DNA sequence determined. There are eight single base differences between the genomic and cDNAs, which are indicated by underlined, bold letters. The sequence shown is the genomic sequence, and the corresponding changes in the cDNA are G-T (2638), C-T (4347), G-C (7605), C-T (8500), T-C (8563), C-T (8980), T-C (9754), C-G (10111) and A-T (11504). All the differences are third base codon substitutions except for position 11565, which is a first base substitution changing the amino acid from S to T.

Expression of faf-lacZ hybrid proteins in the eye disc

In order to determine which cells in the ovary and eye disc express faf, we ‘tagged’ the faf protein with the E. coli gene encoding β-galactosidase (lacZ) by constructing the two hybrid genes shown in Fig. 7A (Materials and methods). One hybrid, faf/lacZ, contains the faf promoter through the first 392 amino acids of the faf protein (Fig. 7A legend), fused in frame to sequences encoding lacZ. The other hybrid gene, fafnuc/lacZ, is identical to faf/lacZ, except that an oligonucleotide containing a nuclear localization signal was cloned in frame just after the first 54 amino acids encoded by faf (Materials and methods). Nuclear localiza-tion of an antigen greatly facilitates the identification of individual cells expressing it in the eye disc. Both constructs were introduced into the Drosophila genome by P element transformation (Materials and methods). Ten independent lines (five containing each construct) were analyzed as described below and showed similar results, except that fafnuc/lacZ lines showed nuclear expression.

Fig. 7.

Expression of faf-lacZ in eye discs and ovariesassayed by β-galactosidase activity staining. (A) The two faf-lacZ hybrid genes introduced into the Drosophila genome by P element transformation are shown (Materials and methods). faf/lacZ contains the first 4.4 kb of the fafIII genomic DNA fragment (Fig. 5A), including ∼2.1 kb upstream of the putative transcription start site (indicated by the arrow), the 5′-untranslated sequences and DNA encoding the first 392 amino acids of the faf protein, including the first two exons (black boxes), two introns and part of the third exon. The faf coding sequences are fused in frame after amino acid 392 to E. coli sequences encoding lacZ (stippled box), followed by termination sequences from SV40 (Materials and methods). fafnuc/lacZ is identical to faf/lacZ except that an oligonucleotide containing an 11 amino acid nuclear localization signal (hatched inverted triangle) was cloned in frame just after the first 54 amino acids (Materials and methods). (B) A histochemically stained (Materials and methods) third instar larval eye-antennal disc from a fafnuc/lacZ transgenic line shows β-galactosidase activity in the antennal portion (a) of the disc and in the eye portion (e), anterior to the morphogenetic furrow (mf). In C-J, posterior is to the right. (C-E) Portions of stained ovarioles from a faf/lacZ transgenic female are shown. β-galactosidase activity is detected in the germarium (g in panel C) and in the egg chambers at all subsequent stages of oogenesis. When the oocyte (o) becomes morphologically distinct from the nurse cells (n), the β-galactosidase activity is apparently restricted to the nurse cells, as the oocyte appears unstained at stage 8 (S8). (Stages are according to Mahowald and Kambysellis, 1980). (Some faf-lacZ protein must be present in stage-8 oocytes, however, because in fafnuc/lacZ lines, stain is seen concentrated in stage-8 oocyte nuclei). At stage10 (D), β-galactosidase activity is detected not only in the nurse cells, but also at the oocyte posterior pole. The posterior staining persists through the final stage of oogenesis (E-stage 14). The same results were obtained in vlsRB71/vlsRB71, tudWC8/tudWC8, pum680/pum680 and nosL7/nosL7 mutant females (Materials and methods). (F) A stained embryo from a faf/lacZ transgenic mother. Maternal β-galactosidase activity persists at the posterior (see text) and becomes incorporated into pole cells (pc). (G,H) Stage-10 (G) and stage-11 (H) oocytes from an osk150/osk346 mutant femalecontaining a faf/lacZ P element. β-galactosidase activity is present in the nurse cells, but there is none at the oocyte posterior pole. The same result was obtained in capuRK/capuRK, spirRP/spirRP and stauD3/stauHL54 females (Materials and methods). (I) A portion of an ovariole from a weak vas mutant female (vas011/vas011) carrying a faf/lacZ P element. Particulate β-galactosidase staining is seen throughout the ovariole, and posteriorly localized stain is present in stage 10 oocytes. Similar results were obtained with vasD1/vasD1, a strong allele. (J) A portion of an ovariole from a vasD1/vasD1 female carrying a faf/lacZ P element, showing that the particulate staining is outside of the nucleus. We do not yet understand the significance of the particulate staining.

Fig. 7.

Expression of faf-lacZ in eye discs and ovariesassayed by β-galactosidase activity staining. (A) The two faf-lacZ hybrid genes introduced into the Drosophila genome by P element transformation are shown (Materials and methods). faf/lacZ contains the first 4.4 kb of the fafIII genomic DNA fragment (Fig. 5A), including ∼2.1 kb upstream of the putative transcription start site (indicated by the arrow), the 5′-untranslated sequences and DNA encoding the first 392 amino acids of the faf protein, including the first two exons (black boxes), two introns and part of the third exon. The faf coding sequences are fused in frame after amino acid 392 to E. coli sequences encoding lacZ (stippled box), followed by termination sequences from SV40 (Materials and methods). fafnuc/lacZ is identical to faf/lacZ except that an oligonucleotide containing an 11 amino acid nuclear localization signal (hatched inverted triangle) was cloned in frame just after the first 54 amino acids (Materials and methods). (B) A histochemically stained (Materials and methods) third instar larval eye-antennal disc from a fafnuc/lacZ transgenic line shows β-galactosidase activity in the antennal portion (a) of the disc and in the eye portion (e), anterior to the morphogenetic furrow (mf). In C-J, posterior is to the right. (C-E) Portions of stained ovarioles from a faf/lacZ transgenic female are shown. β-galactosidase activity is detected in the germarium (g in panel C) and in the egg chambers at all subsequent stages of oogenesis. When the oocyte (o) becomes morphologically distinct from the nurse cells (n), the β-galactosidase activity is apparently restricted to the nurse cells, as the oocyte appears unstained at stage 8 (S8). (Stages are according to Mahowald and Kambysellis, 1980). (Some faf-lacZ protein must be present in stage-8 oocytes, however, because in fafnuc/lacZ lines, stain is seen concentrated in stage-8 oocyte nuclei). At stage10 (D), β-galactosidase activity is detected not only in the nurse cells, but also at the oocyte posterior pole. The posterior staining persists through the final stage of oogenesis (E-stage 14). The same results were obtained in vlsRB71/vlsRB71, tudWC8/tudWC8, pum680/pum680 and nosL7/nosL7 mutant females (Materials and methods). (F) A stained embryo from a faf/lacZ transgenic mother. Maternal β-galactosidase activity persists at the posterior (see text) and becomes incorporated into pole cells (pc). (G,H) Stage-10 (G) and stage-11 (H) oocytes from an osk150/osk346 mutant femalecontaining a faf/lacZ P element. β-galactosidase activity is present in the nurse cells, but there is none at the oocyte posterior pole. The same result was obtained in capuRK/capuRK, spirRP/spirRP and stauD3/stauHL54 females (Materials and methods). (I) A portion of an ovariole from a weak vas mutant female (vas011/vas011) carrying a faf/lacZ P element. Particulate β-galactosidase staining is seen throughout the ovariole, and posteriorly localized stain is present in stage 10 oocytes. Similar results were obtained with vasD1/vasD1, a strong allele. (J) A portion of an ovariole from a vasD1/vasD1 female carrying a faf/lacZ P element, showing that the particulate staining is outside of the nucleus. We do not yet understand the significance of the particulate staining.

Eye discs of third instar transgenic larvae were histo-chemically stained for β-galactosidase activity (Materials and methods). Staining was seen within the entire region ahead of the morphogenetic furrow including the antennal disc (Fig. 7B), but only after prolonged incubation. Stain-ing eye discs with antibodies to β-galactosidase, which is more sensitive than histochemical staining, revealed an even lower level of faf-lacZ protein behind the furrow as well (Fig 8A-C; Materials and methods). Because the anti-body stain in fafnuc/lacZ lines is nuclear, it is possible to identify the cells that express faf-lacZ protein. fafnuc/lacZ eye discs stain with anti-β-galactosidase antibodies in all cells ahead of and behind the morphogenetic furrow, but not within the furrow (Fig. 8A-F). Therefore, faf-lacZ hybrid protein must be rapidly degraded in cells as they enter the furrow.

Fig. 8.

faf-lacZ expression in third instar larval eye-antennal discs. (A) An eye-antennal disc from a fafnuc/lacZ transgenic third instar larvae stained with an anti-β-galactosidase antibody (Materials and methods). Nuclear staining is seen throughout both the antennal portion of the disc (a), and the eye portion (e), except within the morphogenetic furrow, indicated by the arrow. (B) A wild-type eye-antennal disc stained with an antibody raised against a ubiquitous nuclear antigen (gro; see Materials and methods) is shown to demonstrate that nuclear staining present in the furrow (arrow) is readily visible. (C) A close-up of the eye portion of the disc in A, showing the nuclear staining within assembling facets and the absence of stained nuclei within the furrow (arrow). (D-F) Enlargements of progressively more posterior portions of the disc shown in A and C (the direction of the furrow is straight across the top of all panels), showing that fafnuc/lacZ is expressed in the nuclei of all photoreceptors (1-8) and cone cells (c). We could not specifically find stained mystery cell nuclei, probably because they are not usually apically located. However, stained basal nuclei of undetermined cells were observed (not shown). Also, peripodial membrane nuclei were stained. The bar (in F) is 10 μm in D-F, 40 μm in C and 120 μm in A and B.

Fig. 8.

faf-lacZ expression in third instar larval eye-antennal discs. (A) An eye-antennal disc from a fafnuc/lacZ transgenic third instar larvae stained with an anti-β-galactosidase antibody (Materials and methods). Nuclear staining is seen throughout both the antennal portion of the disc (a), and the eye portion (e), except within the morphogenetic furrow, indicated by the arrow. (B) A wild-type eye-antennal disc stained with an antibody raised against a ubiquitous nuclear antigen (gro; see Materials and methods) is shown to demonstrate that nuclear staining present in the furrow (arrow) is readily visible. (C) A close-up of the eye portion of the disc in A, showing the nuclear staining within assembling facets and the absence of stained nuclei within the furrow (arrow). (D-F) Enlargements of progressively more posterior portions of the disc shown in A and C (the direction of the furrow is straight across the top of all panels), showing that fafnuc/lacZ is expressed in the nuclei of all photoreceptors (1-8) and cone cells (c). We could not specifically find stained mystery cell nuclei, probably because they are not usually apically located. However, stained basal nuclei of undetermined cells were observed (not shown). Also, peripodial membrane nuclei were stained. The bar (in F) is 10 μm in D-F, 40 μm in C and 120 μm in A and B.

faf-lacZ protein was also consistently detected in other discs and larval and adult tissues (i.e. fat body, gut, larval ovary and testes, adult male sex organs; data not shown).

faf-lacZ protein, but not mRNA, is localized to the posterior pole of oocytes

Ovaries of transgenic females were also histochemically stained for β-galactosidase activity. Each Drosophila ovary is composed of about twenty ovarioles, which are strings of egg chambers in successive stages of maturity. Each egg chamber contains one oocyte and fifteen nurse cells, which provide the oocyte with RNAs and proteins, including pole plasm components which are transported to the oocyte pos-terior pole. β-galactosidase activity stain is seen through-out oogenesis within the nurse cell-oocyte cluster (Fig. 7C-E). When the oocyte becomes morphologically distinct from the nurse cells, cytoplasmic stain appears to be con-fined to the nurse cells (Fig. 7C). Later, high levels of stain are seen at the posterior pole of oocytes (Fig. 7D). The pos-terior staining persists throughout oogenesis (Fig. 7E), and after the egg is laid, faf-lacZ protein at the posterior becomes incorporated into the pole cells (Fig. 7F). The β-galactosidase activity in the pole cells is not due to zygotic transcription, as embryos from wild-type females mated with faf/lacZ males do not show pole cell staining (data not shown).

To determine whether faf mRNA, as well as faf-lacZ protein, is posteriorly localized in oocytes, we used a faf cDNA probe for whole mount in situ hybridization to ovaries of wild-type females (Materials and methods). The staining pattern observed is identical to that seen in ovaries of faf/lacZ transformants histochemically stained for β-galactosidase activity, except that no posteriorly localized mRNA is seen in oocytes (Fig. 9). Thus, faf-lacZ protein, not faf mRNA, is posteriorly localized, and the first 392 amino acids of faf protein is sufficient for localization.

Fig. 9.

faf transcripts in ovaries. faf transcripts were detected in wild-type ovaries by whole-mount in situ hybridization of a digoxigenin-labeled faf cDNA probe (Materials and methods). faf transcripts were detected in the germarium (G) and in the egg chambers in subsequent stages of oogenesis. At stage 8 (S8), the oocyte (o) is easily distinguishable from the nurse cells (n) and faf transcripts appear to be restricted to the nurse cells, through stage S10. At stage 10, the time that faf/lacZ protein first appears at the oocyte posterior, no posteriorly localized faf transcripts are detected.

Fig. 9.

faf transcripts in ovaries. faf transcripts were detected in wild-type ovaries by whole-mount in situ hybridization of a digoxigenin-labeled faf cDNA probe (Materials and methods). faf transcripts were detected in the germarium (G) and in the egg chambers in subsequent stages of oogenesis. At stage 8 (S8), the oocyte (o) is easily distinguishable from the nurse cells (n) and faf transcripts appear to be restricted to the nurse cells, through stage S10. At stage 10, the time that faf/lacZ protein first appears at the oocyte posterior, no posteriorly localized faf transcripts are detected.

Localization of faf-lacZ protein in oocytes depends on osk

We next asked whether posterior localization of faf-lacZ protein depends on the posterior group genes. Ovaries of females homozygous mutant for each of nine posterior group genes, that also carry a copy of the faf/lacZ P ele-ment, were stained for β-galactosidase activity (Materials and methods). β-galactosidase activity was detected at the oocyte posterior in vas and mutants downstream, but not in osk or mutants upstream in the hierarchy (Fig. 7G-I and legend). As stau protein and osk RNA and protein are pos-teriorly localized earlier than faf-lacZ protein, and vas pro-tein is localized at about the same time, faf appears to fit into the posterior group hierarchy.

Insights into the specific roles of some of the posterior group genes were recently obtained using transgenic flies containing an osk gene whose RNA is mislocalized to the anterior pole of oocytes (Ephrussi and Lehmann, 1992). Ephrussi and Lehmann have shown that osk RNA at the anterior is sufficient to localize nos RNA and also to recruit all of the pole plasm components required for the forma-tion of functional pole cells, which include vas and tud, but not capu, spir, stau or vls. Thus, the authors conclude that capu, spir and stau are required only to tether osk gene prod-ucts to the posterior pole, while osk, vas and tud are essen-tial for pole plasm formation and nos RNA localization. (The role of vls is uncertain.)

We find that faf-lacZ protein colocalizes with osk gene products at the anterior pole in embryos of transgenic moth-ers carrying one copy of faf/lacZ and one copy of anteri- orly localized osk (Fig. 10). As faf-lacZ localization in oocytes is independent of genes downstream of osk in the hierarchy, and genes upstream of osk appear to be required only to tie osk RNA to the posterior pole, we conclude that faf-lacZ protein localization is likely to depend only on osk gene products.

Fig. 10.

Posterior localization of faf-lacZ protein depends on osk. Pre-syncytial blastoderm embryos from mothers with one copy of the faf/lacZ P element (A), or from mothers with one copy of the faf/lacZ P element and one copy of a P element containing an anteriorly localizing osk gene (B) were stained with an anti-β-galactosidase antibody (Materials and methods). Posterior is to the right in both panels. faf-lacZ protein colocalizes with both wild-type and anterior osk RNA in B.

Fig. 10.

Posterior localization of faf-lacZ protein depends on osk. Pre-syncytial blastoderm embryos from mothers with one copy of the faf/lacZ P element (A), or from mothers with one copy of the faf/lacZ P element and one copy of a P element containing an anteriorly localizing osk gene (B) were stained with an anti-β-galactosidase antibody (Materials and methods). Posterior is to the right in both panels. faf-lacZ protein colocalizes with both wild-type and anterior osk RNA in B.

Pole plasm components are posteriorly localized in faf mutant embryos

The observation that faf-lacZ protein localization depends on osk suggested that posterior localization of downstream gene products (vas protein and nos RNA ) may depend on faf. cyclinB mRNA may also be considered downstream of faf as it is posteriorly localized, dependent on vas and genes upstream (Raff et al., 1990). Using nos and cyclinB cDNA probes for in situ hybridization, and an antibody to vas pro-tein, we find that all of these gene products are posteriorly localized in faf mutant embryos, and are incorporated into pole cells (Fig. 11).

Fig. 11.

nos RNA is posteriorly localized in faf mutant embryos. nos transcripts were detected in 0-4 hour embryos from wild-type (A,C) and fafFO8 (B,D) mothers by in situ hybridization of whole embryos with a digoxigenin-labeled nos cDNA probe (Materials and methods). (A,B) Posterior localization of nos transcripts appears identical in wild-type (A) and faf mutant (B) presyncytial blastoderm embryos. nos RNA is incorporated into pole cells in wild-type embryos (C) and also in faf mutant embryos (D). The pole cells in faf mutant embryos are fewer in number and more spread out than in wild-type. The same results were obtained with all faf alleles that cause female sterility (see Table 1) and with fafBX4/fafBP, the allele combination most likely to be null. The same results were obtained when embryos from fafFO8 mutant mothers were hybridized with a cyclinB probe, or stained with anti-vas antibodies (Materials and methods).

Fig. 11.

nos RNA is posteriorly localized in faf mutant embryos. nos transcripts were detected in 0-4 hour embryos from wild-type (A,C) and fafFO8 (B,D) mothers by in situ hybridization of whole embryos with a digoxigenin-labeled nos cDNA probe (Materials and methods). (A,B) Posterior localization of nos transcripts appears identical in wild-type (A) and faf mutant (B) presyncytial blastoderm embryos. nos RNA is incorporated into pole cells in wild-type embryos (C) and also in faf mutant embryos (D). The pole cells in faf mutant embryos are fewer in number and more spread out than in wild-type. The same results were obtained with all faf alleles that cause female sterility (see Table 1) and with fafBX4/fafBP, the allele combination most likely to be null. The same results were obtained when embryos from fafFO8 mutant mothers were hybridized with a cyclinB probe, or stained with anti-vas antibodies (Materials and methods).

nos function is unimpaired in faf mutant embryos

nos protein facilitates the expression of genes that direct the abdominal segmentation pattern by preventing transla-tion of hunchback (hb) maternal mRNA in the posterior of the embryo (Tautz, 1988; Struhl, 1989; Irish et al., 1989; Hülskamp et al., 1989). Thus, hb protein normally accu-mulates only in the anterior half of the embryo. By stain-ing faf mutant embryos with an antibody to hb protein, we find that maternal hb protein is distributed normally in faf mutant embryos (data not shown; Materials and methods). Thus, faf is unlikely to be required for nos function.

The faf gene encodes two similar large proteins specifically required during two different developmental processes in Drosophila: compound eye assembly and oogenesis. In the larval eye disc, cells outside the assembling facets require faf for short-range cell interactions that prevent the mys-tery cells from becoming photoreceptors. Maternal faf pro-tein appears to be required for nuclear migration and cel-lularization in early embryogenesis. In addition, a faf-lacZ protein containing only the first 392 amino acids of faf is posteriorly localized in oocytes and its localization depends on osk.

faf and cell interactions in the eye disc

The faf gene appears to function in one of several negative regulatory mechanisms in place in the larval eye disc which ensure that each facet contains only eight photoreceptor cells. We have shown that faf is involved in a process whereby the mystery cells are prevented from becoming photoreceptors by cells near to but outside the photorecep-tors and mystery cells. argos, a diffusible protein and gro, a nuclear protein, appear to play roles in facet assembly similar to faf (Freeman et al., 1992; Delidakis et al., 1991; Fischer-Vize et al., 1992). Further experiments are required to determine if these three genes function in the same or different pathways. In contrast, analysis of svp mutants shows that photoreceptors R3 and R4 also repress neural-ization of the mystery cells (Mlodzik et al., 1990b).

Although our results suggest that faf is required in cells near the developing photoreceptors and mystery cells, we have not identified these cells. Presumably, when the mys-tery cells are influenced by the negative regulatory mech-anism involving faf, they are in contact with the cells in which faf is required. The precise time during eye devel-opment when faf function influences mystery cell fate is unknown. The first structure to emerge from the morpho-genetic furrow is a rosette in which 10-15 cells, including the R8,2,5,3,4 precursors and the mystery cells, form a ring around four or five core cells. The ring then opens and preclusters containing R8,2,5,3,4 precursors and the mys-tery cells are formed (Wolff and Ready, 1991a). Thus, the cells that require faf to prevent neurogenesis of the mys-tery cells could be the core cells or cells adjacent to the mystery cells within the ring. Alternatively, if the faf-dependent process occurs later, uncommitted epithelial cells surrounding the precluster could be involved. We also cannot rule out the possibility that faf acts anterior to the morphogenetic furrow.

The expression of ro and svp by the mystery cells in faf mutant eye discs is consistent with the hypothesis that they are R3/4 subtype photoreceptors. In gro mutant eye discs, it was shown more conclusively that the mystery cells become R3/4 cells (Fischer-Vize et al., 1992). This repeated observation is interesting because experiments with ro show that the normal R3 and R4 cells require inductive cues from R2 and R5 in order to join the facet (Tomlinson et al., 1988). The mystery cells are located between R3 and R4 and are not in contact with R2 and R5. Their determina-tion as R3/4 subtype photoreceptors suggests either that the inductive signal from R2/5 is permissive for photoreceptor cell determination and R3/4 subtype determination is a type of default state, or that R3/4 can send the same instructive signals as R2/5 (see Fischer-Vize et al., 1992).

The pattern of faf-lacZ expression in the larval eye disc

The expression pattern of the faf-lacZ hybrid protein in the larval eye disc is unusual; faf-lacZ is expressed in all cells anterior and posterior to the morphogenetic furrow. This pattern is consistent with the conclusion that faf is required in cells outside the early developing ommatidial clusters. It is possible that faf expression in cells within the assem-bling facets could be required for faf-dependent aspects of eye morphogenesis other than excluding the mystery cells, such as orientation and spacing of facets.

The absence of faf-lacZ protein within the morphogenetic furrow indicates that it must be rapidly degraded in cells as they enter the furrow and suggests two possibilities. First, normal faf protein in the eye disc may be unstable, and fac-tors that stabilize it or activate its transcription may be absent within the morphogenetic furrow. Alternatively, there may be a factor present in the furrow that actively degrades faf protein. faf-lacZ protein turnover may be medi-ated by the PEST sequence contained within the first 392 amino acids of faf. Some caution is warranted in interpret-ing lacZ hybrid protein expression patterns. However, as lacZ is generally more stable than endogenous Drosophila proteins in similar experiments (for example, see Mlodzik et al., 1990a), the normal faf protein is likely to be absent within the furrow.

Functions of faf in oogenesis

The localization of faf-lacZ protein to the posterior of oocytes and its dependence on osk suggests that faf has an important function at the posterior pole. (Although there may be stability differences between faf-lacZ and normal faf protein, we assume that the the normal faf protein is similarly localized.) As nos RNA is localized to and mater-nal hb protein is eliminated from the posterior of faf mutant embryos, a role for faf in abdominal pattern formation, which could have been obscured by the earlier role of faf in nuclear migration or cellularization, appears unlikely. Perhaps faf plays a role in pole plasm formation. Two pole plasm components essential for pole cell formation, osk and vas proteins, are also localized to the posterior pole of oocytes (Hay et al., 1988b; Ephrussi et al., 1991; Ephrussi and Lehmann, 1992). In addition, vas protein has been shown to be part of the polar granule, a particle composed of RNA and protein, associated with pole cells (Hay et al., 1988b). Although pole cells do form in faf mutant embryos, there are fewer than the wild-type number and they are irregularly positioned. Thus, faf could play a role in pole cell determination, development or function. Although the pole cell defects observed could be a secondary conse-quence of the failure of nuclear migration or cellularization in faf mutant embryos, this would not preclude a specific role for faf in pole cells as well. Further experiments are required to determine if faf mutant pole cells are functional, and to determine if faf protein is part of the polar granule.

The defects in faf mutant embryos indicates that faf plays a role in nuclear migration and possibly also in cellular-ization. It therefore seems puzzling that faf-lacZ is detected only at the posterior pole of oocytes and embryos. The absence of localized faf protein in faf mutant embryos does not cause the global embryogenesis defects as localization depends on osk and osk mutant embryos cellularize nor-mally. The defects in embryogenesis caused by faf muta-tions could be accounted for if some maternal faf protein is distributed throughout the oocyte or embryo. There is some evidence for non-localized faf-lacZ protein in early embryos stained with antibodies to β-galactosidase (data not shown). However, as the faf-lacZ protein may be more stable than endogenous faf, an anti-faf antibody would be useful for investigating this possibility. Interestingly, embryos from vas and vls mothers not only lack pole cells, but sometimes also display defects in cellularization (Schüpbach and Wieschaus, 1989; Hay et al., 1990; Lehmann and Nüsslein-Volhard, 1991).

J. A. F.-V. is extremely grateful to Doug Melton (Harvard Uni-versity) for his exceptional generosity in allowing me to work in his laboratory, where some of this work was carried out. We are indebted to Todd Laverty (UC Berkeley) for expert analysis of polytene chromosomes. We thank Seymour Benzer’s laboratory for mAb22C10, Ulrike Heberlein for mAbro, Anne Williamson for anti-vas antibodies, Robin Wharton and Gary Struhl for anti-hb antibodies, Christos Delidakis and Spyros Artavanis-Tsakonas for anti-gro antibodies, Sima Misra for the cyclinB plasmid, Dan Kiehart’s laboratory for high-molecular-weight spectrin and non-muscle myosin plasmids, the Rubin laboratory enhancer trappers for fly lines AEI27 and P[w+]100D and Anne Ephrussi for the anterior localizing osk flies. We also thank Bruce Kimmel for performing the scanning electron microscopy in Fig. 1B, Mike Simon for Fig. 1A, Bruce Kimmel and Doug Barker for their help with the computer analysis of the faf DNA sequence, and Erin O’Shea for examining the leucine-zipper motif. J. A. F.-V. is grateful to Chris Rongo, Liz Gavis, Francisco Pelegri, Doug Barker, Jen Mach and two anonymous reviewers for their com-ments on the manuscript, to everyone in the Lehmann lab for advice, discussion and instruction in how to look at ovaries and embryos, and to Sydney Vize for inspiring the name fat facets. J. A. F.-V. was supported by Helen Hay Whitney and Howard Hughes Medical Institute postdoctoral fellowships. Ruth Lehmann and Gerald M. Rubin are Howard Hughes Medical Institute Investigators.

Baker
,
N. E.
,
Mlodzik
,
M.
and
Rubin
,
G. M.
(
1990
).
Spacing differentiation in the developing Drosophila eye: A fibrinogen-related lateral inhibitor encoded by scabrous
.
Science
250
,
1370
1377
.
Banerjee
,
U.
and
Zipursky
,
S. L.
(
1990
).
The role of cell-cell interactions in the development of the Drosophila visual system
.
Neuron
4
,
177
187
.
Boswell
,
R. E.
,
Prout
,
M. E.
and
Steichen
,
J. C.
(
1991
).
Mutations in a newly identified Drosophila melanogaster gene, mago nashi, disrupt germ cell formation of mirror-image symmetrical double abdomen embryos
.
Development
113
,
373
384
.
Carthew
,
R. W.
and
Rubin
,
G. M.
(
1990
).
seven in absentia, a gene required for specification of R7 cell fate in the Drosophila eye
.
Cell
63
,
561
577
.
Cavener
,
D.
(
1987
).
Comparison of the consensus sequences flanking translational start-sites in Drosophila and vertebrates
.
Nucl. Acids Res
.
15
,
1353
1361
.
Delidakis
,
C.
,
Preiss
,
A.
,
Hartley
,
D. A.
and
Artavanis-Tsakonas
,
S.
(
1991
).
Two genetically and molecularly distinct functions involved in early neurogenesis reside within the Enhancer of split locus of Drosophila melanogaster
.
Genetics
129
,
803
823
.
Ephrussi
,
A. E.
,
Dickinson
,
L. K.
and
Lehmann
,
R.
(
1991
).
oskar organizes the germ plasm and directs localization of the posterior determinant nanos
.
Cell
66
,
37
50
.
Ephrussi
,
A. E.
and
Lehmann
,
R.
(
1992
).
Induction of germ cell formation by oskar
.
Nature
358
,
387
392
.
Fischer-Vize
,
J. A.
,
Vize
,
P. D.
and
Rubin
,
G. M.
(
1992
).
A unique mutation in the Enhancer of split gene complex affects the fates of the mystery cells in the developing Drosophila eye
.
Development
115
,
89
101
.
Foe
,
V. E.
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
.
Freeman
,
M.
,
Klämbt
,
C.
,
Goodman
,
C. S.
and
Rubin
,
G. M.
(
1992
).
The argos gene encodes a diffusible factor that regulates cell fate decisions in the Drosophila eye
.
Cell
69
,
963
975
.
Fujita
,
S. C.
,
Zipursky
,
S. L.
,
Benzer
,
S.
,
Ferrus
,
A.
and
Shotwell
,
S. L.
(
1982
).
Monoclonal antibodies against the Drosophila nervous system
.
Proc. Natn. Acad. Sci. USA
79
,
7929
7933
.
Hay
,
B.
,
Jan
,
L. Y.
and
Jan
,
Y. N.
(
1988a
).
A protein component of Drosophila polar granules is encoded by vasa and has extensive sequence similarity to ATP-dependent helicases
.
Cell
55
,
577
587
.
Hay
,
B.
,
Ackerman
,
L.
,
Barbel
,
S.
,
Jan
,
L. Y.
and
Jan
,
Y. N.
(
1988b
).
Identification of a component of Drosophila polar granules
.
Development
103
,
625
640
.
Hay
,
B.
,
Jan
,
L. Y.
and
Jan
,
Y. N.
(
1990
).
Localization of vasa, a component of Drosophila polar granules, in maternal-effect mutants that alter embryonic anteroposterior polarity
.
Development
109
,
425
433
.
Hafen
,
E.
,
Basler
,
K.
,
Edstroem
,
J. E.
and
Rubin
,
G. M.
(
1987
).
sevenless, a cell-specific homeotic gene of Drosophila, encodes a putative transmembrane receptor with a tyrosine kinase domain
.
Science
236
,
55
63
.
Heberlein
,
U.
and
Rubin
,
G. M.
(
1991
).
Star is required in a subset of photoreceptor cells in the developing Drosophila retina and displays dosage-sensitive interaction with rough
.
Dev. Biol
.
144
,
353
361
.
Higashijima
,
S.
,
Kojima
,
T.
,
Michiue
,
T.
,
Ishimaru
,
S.
,
Emori
,
Y.
and
Saigo
,
K.
(
1992
).
Dual Bar homeo box genes of Drosophila required in two photoreceptor cells, R1 and R6, and primary pigment cells for normal eye development
.
Genes Dev
.
6
,
50
60
.
Hülskamp
,
M.
,
Schröder
,
C.
,
Pfeifle
,
C.
,
Jäckle
,
H.
and
Tautz
,
D.
(
1989
).
Posterior segmentation of the Drosophila embryo in the absence of a maternal posterior organizer gene
.
Nature
338
,
629
632
.
Irish
,
V.
,
Lehmann
,
R.
and
Akam
,
M.
(
1989
).
The Drosophila posterior-group gene nanos functions by repressing hunchback activity
.
Nature
338
,
646
648
.
Kalderon
,
D.
,
Roberts
,
B. L.
,
Richardson
,
W. D.
and
Smith
,
A. E.
(
1984
).
A short amino acid sequence able to specify nuclear location
.
Cell
39
,
499
509
.
Karpilow
,
J.
,
Kolodkin
,
A.
,
Bork
,
T.
and
Venkatesh
,
T.
(
1989
).
Neuronal development in the Drosophila compound eye: rap gene function is required in photoreceptor cell R8 for ommatidial pattern formation
.
Genes Dev
.
3
,
1834
1844
.
Karr
,
T. L.
and
Alberts
,
B. M.
(
1986
).
Organization of the cytoskeleton in early Drosophila embryos
.
J. Cell Biol
.
102
,
1494
1509
.
Kim-Ha
,
J.
,
Smith
,
J. L.
and
Macdonald
,
P. M.
(
1991
).
oskar mRNA is localized to the posterior pole of the Drosophila oocyte
.
Cell
66
,
23
35
.
Kimmel
,
B. E.
,
Heberlein
,
U.
and
Rubin
,
G. M.
(
1990
).
The homeodomain protein rough is expressed in a subset of cells in the developing Drosophila eye where it can specify photoreceptor cell subtype
.
Genes Dev
.
4
,
712
727
.
Klemenz
,
R.
,
Weber
,
U.
and
Gehring
,
W. J.
(
1987
).
The white gene as a marker in a new P-element vector for gene transfer in Drosophila
.
Nucl. Acids Res
.
15
,
3947
3959
.
Landschulz
,
W. H.
,
Johnson
,
P. F.
and
McKnight
,
S. L.
(
1988
).
The leucine zipper: A hypothetical structure common to a new class of DNA binding proteins
.
Science
240
,
1759
1764
.
Lasko
,
P. F.
and
Ashburner
,
M.
(
1988
).
The product of the Drosophila gene vasa is very similar to eukaryotic initiation factor-4A
.
Nature
335
,
611
617
.
Lasko
,
P. F.
and
Ashburner
,
M.
(
1990
).
Posterior localization of vasa protein correlates with, but is not sufficient for, pole cell development
.
Genes Dev
.
4
,
905
921
.
Lawrence
.
P. A.
and
Green
,
S. M.
(
1979
).
Cell lineage in the developing retina of Drosophila
.
Dev. Biol
.
71
,
142
152
.
Lehmann
,
R.
and
Nüsslein-Volhard
,
C.
(
1987
).
Involvement of the pumilio gene in the transport of an abdominal signal in the Drosophila embryo
.
Nature
329
,
167
170
.
Lehmann
,
R.
and
Nüsslein-Volhard
,
C.
(
1991
).
The maternal gene nanos has a central role in posterior pattern formation of the Drosophila embryo
.
Development
112
,
679
691
.
Lehmann
,
R.
(
1992
).
Germ plasm formation and germ cell determination in Drosophila
.
Curr. Opin. Genet. Devel., in press
.
Levis
,
R.
,
Hazelrigg
,
T.
and
Rubin
,
G. M.
(
1985
).
Separable cis-acting control elements for expression of the white gene of Drosophila
.
EMBO J
.
4
,
3489
3499
.
Lindsley
,
D. L.
and
Zimm
,
G. G.
(
1992
).
The Genome of Drosophila melanogaster
.
San Diego
:
Academic Press, Inc
.
Macdonald
,
P. M.
(
1992
).
The Drosophila pumilio gene: an unusually long transcription unit and an unusual protein
.
Development
114
,
221
232
.
Manseau
,
L. J.
and
Schüpbach
,
T.
(
1989
).
cappuccino and spire: two unique maternal-effect loci required for both the anteroposterior and dorsoventral patterns of the Drosophila embryo
.
Genes Dev
.
3
,
1437
1452
.
Mlodzik
,
M.
,
Baker
,
N. E.
and
Rubin
,
G. M.
(
1990a
).
Isolation and expression of scabrous, a gene regulating neurogenesis in Drosophila
.
Genes Dev
.
4
,
1848
1861
.
Mlodzik
,
M.
,
Hiromi
,
Y.
,
Weber
,
U.
,
Goodman
,
C.
and
Rubin
,
G. M.
(
1990b
).
The Drosophila seven-up gene, a member of the steroid receptor gene superfamily controls photoreceptor cell fates
.
Cell
60
,
211
224
.
Moses
,
K.
,
Ellis
,
M. C.
and
Rubin
,
G. M.
(
1989
).
The glass gene encodes a zinc-finger protein required by Drosophila photoreceptor cells
.
Nature
340
,
531
536
.
Pirrotta
,
V.
(
1986
).
Cloning Drosophila genes
.
In Drosophila: A Practical Approach
. (ed.
D. B.
Roberts
), pp.
83
110
.
Oxford
:
IRL Press
.
Raff
,
J. W.
,
Whitfield
,
W. G. F.
and
Glover
,
D. M.
(
1990
).
Two distinct mechanisms localise cyclin B transcripts in syncytial Drosophila embryos
.
Development
110
,
1249
1261
.
Ready
,
D. F.
,
Hanson
,
T. E.
and
Benzer
,
S.
(
1976
).
Development of the Drosophila retina, a neurocrystalline lattice
.
Dev. Biol
.
53
,
217
240
.
Reinke
,
R.
and
Zipursky
,
S. L.
(
1988
).
Cell-cell interaction in the Drosophila retina: the bride-of-sevenless gene is required in photoreceptor cell R8 for R7 cell development
.
Cell
55
,
321
330
.
Robertson
,
H. M.
,
Preston
,
C. R.
,
Phillis
,
R. W.
,
Johnson-Schlitz
,
D. M.
,
Benz
,
W. K.
and
Engels
,
W. R.
(
1988
).
A stable genomic source of P element transposase in Drosophila melanogaster
.
Genetics
118
,
461
470
.
Rogers
,
S.
,
Wells
,
R.
and
Rechsteiner
,
M.
(
1986
).
Amino acid sequences common to rapidly degraded proteins: the PEST hypothesis
.
Science
234
,
364
368
.
Rubin
,
G. M.
(
1991
).
Signal transduction and the fate of the R7 photoreceptor in Drosophila
.
Trends Genet
.
7
,
372
377
.
Saint
,
R.
,
Kalionis
,
B.
,
Lockett
,
T. J.
and
Elizur
,
A.
(
1988
).
Pattern formation in the developing eye of Drosophila melanogaster is regulated by the homeobox gene, rough
.
Nature
334
,
151
154
.
Sambrook
,
G.
,
Fritsch
,
E. F.
and
Maniatis
,
T.
(
1989
).
Molecular Cloning: A Laboratory Manual
.
Cold Spring Harbor
:
Cold Spring Harbor Laboratory Press
.
Schüpbach
,
T.
and
Wieschaus
,
E.
(
1986
).
Maternal-effect mutations altering the anterior-posterior pattern of the Drosophila embryo
.
Roux’s Arch. Dev. Biol
.
195
,
302
317
.
Schüpbach
,
T.
and
Wieschaus
,
E.
(
1989
).
Female sterile mutations on the second chromosome of Drosophila melanogaster. I. Maternal effect mutations
.
Genetics
121
,
101
117
.
Spradling
,
A. C.
(
1986
).
P-element mediated transformation
.
In Drosophila: A practical approach
. (ed.,
D. B.
Roberts
), pp.
175
197
.
Oxford
:
IRL Press
.
St. Johnston
,
D.
,
Beuchle
,
D.
and
Nüsslein-Volhard
,
C.
(
1991
).
staufen, a gene required to localize maternal RNAs in the Drosophila egg
.
Cell
66
,
51
63
.
St. Johnston
,
D.
and
Nüsslein-Volhard
,
C.
(
1992
).
The origin of pattern and polarity in the Drosophila embryo
.
Cell
68
,
201
219
.
Struhl
,
G.
(
1989
).
Differing strategies for organizing anterior and posterior body pattern in Drosophila embryos
.
Nature
338
,
741
744
.
Tautz
,
D.
(
1988
).
Regulation of the Drosophila segmentation gene hunchback by two maternal morphogenetic centers
.
Nature
332
,
281
284
.
Thummel
,
C. S.
,
Boulet
,
A. M.
and
Lipshitz
,
H. D.
(
1988
).
Vecotrs for Drosophila P element-mediated transformation and tissue culture transformation
.
Gene
74
,
445
456
.
Tomlinson
,
A.
and
Ready
,
D. F.
(
1987a
).
Neuronal differentiation in the Drosophila ommatidium
.
Dev. Biol
.
120
,
366
376
.
Tomlinson
,
A.
and
Ready
,
D. F.
(
1987b
).
Cell fate in the Drosophila ommatidium
.
Dev. Biol
.
123
,
264
275
.
Tomlinson
,
A.
,
Kimmel
,
B. E.
and
Rubin
,
G. M.
(
1988
).
rough, a Drosophila homeobox gene required in photoreceptors R2 and R5 for inductive interactions in the developing eye
.
Cell
55
,
771
784
.
Wang
,
C.
and
Lehmann
,
R.
(
1991
).
Nanos is the localized posterior determinant in Drosophila
.
Cell
66
,
637
648
.
Wieschaus
,
E.
and
Nüsslein-Volhard
,
C.
(
1986
).
Looking at embryos
.
In Drosophila: A Practical Approach
. (ed.
D. B.
Roberts
), pp.
199
226
.
Oxford
:
IRL Press
.
Wolff
,
T.
and
Ready
,
D. F.
(
1991a
).
The beginning of pattern formation in the Drosophila compound eye: the morphogenetic furrow and the second mitotic wave
.
Development
113
,
841
850
.
Wolff
,
T.
and
Ready
,
D. F.
(
1991b
).
Cell death in normal and rough eye mutants of Drosophila
.
Development
113
,
825
839
.

The GenBank Accession numbers for the sequences