The pineapple eye Gene Is Required for Survival of Drosophila Imaginal Disc Cells

Corresponding author: Nicholas E. Baker, Albert Einstein College of Medicine, 1300 Morris Park Ave., Bronx, NY 10461., baker{at}aecom.yu.edu (E-mail)

Communicating editor: R. S. HAWLEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Each ommatidium of the Drosophila eye is constructed by precisely 19 specified precursor cells, generated in part during a second mitotic wave of cell divisions that overlaps early stages of ommatidial cell specification. Homozygotes for the pineapple eye mutation lack sufficient precursor cells due to apoptosis during the period of fate specification. In addition development is delayed by apoptosis during earlier imaginal disc growth. Null alleles are recessive lethal and allelic to l(2)31Ek; heteroallelic combinations can show developmental delay, abnormal eye development, and reduced fertility. Mosaic clones autonomously show extensive cell death. The pineapple eye gene was identified and predicted to encode a novel 582-amino-acid protein. The protein contains a novel, cysteine-rich domain of 270 amino acids also found in predicted proteins of unknown function from other animals.

IN Drosophila as in other organisms, development of the adult from the egg is associated with both specification of diverse cell types and increased cell number and body size. Differentiation and patterning of many body regions have been intensively studied. Mechanisms of growth and proliferation are a more recent focus (OLDHAM et al. 2000 ; EDGAR et al. 2001 ; JOHNSTON and GALLANT 2002 ). Many adult fly structures derive from so-called imaginal discs, embryonic cells that are set aside to proliferate and grow during larval life without differentiating until metamorphosis, when they replace the larva with adult structures within the pupa. Larval life occurs largely independently of imaginal discs, so that defects in imaginal disc growth or patterning can be detected from their later effects on adult structures. The compound eye is an adult structure for which cell number is particularly crucial, because the very regular eye structure depends on precise numbers and arrangements of multiple specialized retinal cell types. For this reason defects in compound eye morphology can be an indication of altered cell proliferation or survival.

The compound eye of Drosophila is composed of hundreds of nearly identical ommatidia or unit eyes (WOLFF and READY 1993 ). Both the number and the type of cells in ommatidia are invariant. Each ommatidium is constructed by 19 progenitor cells; 18 are specified postmitotically and 1 continues dividing to form a bristle organ. Ommatidial progenitor cells are specified by cell-cell interactions in the eye imaginal disc, the epithelium that gives rise to the eye and head. Cell fate specification begins at the posterior margin of the eye imaginal disc, progressing anteriorly as a "morphogenetic furrow" moves across the eye disc epithelium. Anterior to the morphogenetic furrow, cells in the eye imaginal disc proliferate. The cell cycle arrests in the morphogenetic furrow, and the first 5 cells of each ommatidium are specified. The remaining cells reenter the cell cycle in the "second mitotic wave" posterior to the morphogenetic furrow and later give rise to the other 14 precursor cell types. The second mitotic wave (SMW) makes an important contribution to the number of cells per ommatidium. The number of ommatidia is already fixed by prior specification of the 5-cell preclusters, so that if the SMW is prevented by targeted expression of the cell cycle inhibitor p21^WAF1/CIP1 then defective ommatidia are generated from the depauperate pool of progenitor cells. By contrast, changes in growth and proliferation anterior to the morphogenetic furrow are likely to lead to changes in the number of ommatidial preclusters and so to larger or smaller eyes containing more or fewer ommatidia, each constructed from the normal complement of precursor cells (DE NOOIJ and HARIHARAN 1995 ; BAKER 2001 ; NEUFELD and HARIHARAN 2002 ).

Ommatidia that assemble in the absence of sufficient precursor cells lack some cells and are unable to stack into the crystalline lattice typical of the normal eye. Such eyes have a roughened, irregular eye surface (DE NOOIJ and HARIHARAN 1995 ). The phenotype is associated with variable ommatidial defects. Often later ommatidial cell fates cannot be specified, such as pigment cells that normally function to separate ommatidia into individual facets. In this case multiple ommatidia, each perhaps deficient in particular cell types, share oversized facets (DE NOOIJ and HARIHARAN 1995 ).

This phenotype of GMRp21^WAF1/CIP1 resembles that of mutations in an endogenous Drosophila gene, pineapple eye. Ommatidia from pineapple eye (pie) mutants initiate development normally, but become increasingly defective, and cell types that are specified later in the cascade such as cone cells, pigment cells, and some photoreceptor cells are often missing (BAKER et al. 1992 ). The pie adults have roughened and irregular eyes with large facets like those described for GMRp21^WAF1/CIP1. We have characterized the pineapple eye gene and its mutants further to understand how it contributes to proper eye cell number.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly strains:
Isolation of the pie^EB3 mutations was described by BAKER et al. 1992 . A similar allele, pie^CB6, was obtained in the same screen but is no longer extant. Other pie alleles were described by CLEGG et al. 1993 . GMRp35 was described by HAY et al. 1994 ; GMRp21 was described by DE NOOIJ and HARIHARAN 1995 ; hid^WRX+1 and hid⁰⁵⁰¹⁴ were described by GRETHER et al. 1995 ; Df(2L)J2, Df(2L)J27, Df(2L)J39, Df(2L)J77, Df(2L)J106, Df(2L)J16, and Df(2L)J17 were described by CLEGG et al. 1993 ; l(2)54 was described by SANDLER 1977 ; and mat(2)QM47 was described by SCHUPBACH and WEISCHAUS 1986 .

For the mutagenesis, adult males of the genotype w; l(2)k08229/l(2)k10307 were exposed to -radiation (4000 rads) and mated with w;pie^EB3/In(2LR)Gla females. F₁ flies with rough eyes or lacking eye pigmentation were bred where possible to establish stocks putatively mutant for pie or deleted for nearby genes. (2)k08229 and l(2)k10307 correspond to P-element insertions carrying the [w⁺] gene inserted in chromosome bands 31F1-3 or 31F4-5, respectively (TOROK et al. 1993 ).

Histology:
Immunochemistry using ELAV, CM1, anti-cyclin B, and anti-cut antibodies was performed as described (GAUL et al. 1992 ; FU and NOLL 1997 ; LEE et al. 2000 ; BAKER and YU 2001 ; YU et al. 2002 ). Sectioned material was prepared as described by BAKER et al. 1992 .

Molecular biology:
Growth and selection of bacterial plasmids, cosmids, and bacteriophage were performed according to standard methods (SAMBROOK et al. 1989 ). For the sequencing of mutant DNA, genomic DNA was prepared from larvae of homozygous pie^EB3 adults or from pie^E1-16/Df(2L)J77 third instar larvae selected by absence of the dominant marker Tubby present in pie^E1-16/T(2;3)SM5TM6B and Df(2L)J77/T(2;3)SM5TM6B siblings.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The pie mutant affects imaginal disc cell survival:
As described previously, flies homozygous for the pie^EB3 mutation had rough eyes. Facet size varied from smaller than normal to enlarged, apparently fused facets (Fig 1A and Fig B). Sections confirmed absence of pigment, cone, and photoreceptor cells (Fig 1C and Fig D). The pattern of missing cells varied, many ommatidia being normal and no specific cell type being exclusively affected (BAKER et al. 1992 ). The adult sections of pie^EB3 homozygotes were essentially indistinguishable from those described for GMRp21, in which arrest of the SMW leads to shortfall of unspecified precursor cells and frequent inability to specify later cell fates (DE NOOIJ and HARIHARAN 1995 ).

View larger version (141K): [in this window] [in a new window] [Download PPT slide]   Figure 1. The pie mutant phenotype. (A) Scanning electron microscopy shows the regular structure of the wild-type fly eye (anterior to the left). (B) Eyes from pieEB3 homozygotes are rough and irregular with variable facet size. Some facets are larger than normal. (C) Section through wild-type eye. The section extends from more distal levels above to more basal levels below. Each ommatidium is similar and surrounded by reddish pigment cells. (D) Section through pieEB3 homozygote showing the variable composition of ommatidia. Many ommatidia have too few photoreceptor cells missing. Missing pigment cells result in clusters of photoreceptor cells sharing single enlarged facets. Some ommatidia appear abnormally rotated.

If pie^EB3 caused arrest of the SMW, we would expect that cyclin B would not accumulate and that mitotic figures would be absent posterior to the morphogenetic furrow, as reported for GMRp21 (Fig 2, A–F). To test this notion, eye discs from pie^EB3 homozygotes were labeled with anti-cyclin B or with basic fuchsin, a stain that reveals mitotic figures. Cyclin B protein accumulates in cells that have progressed through G₁ but not through mitosis. In wild type, 80% of SMW cells divide, mostly in the column 3–5 region posterior to the morphogenetic furrow (BAKER and YU 2001 ). These cells degrade their cyclin B at mitosis (EVANS et al. 1983 ; KNOBLICH and LEHNER 1993 ). Contrary to the prediction that pie^EB3 arrested the SMW, cyclin B protein was seen to accumulate in a second mitotic wave posterior to the morphogenetic furrow, and mitotic figures were observed in fuchsin-stained preparations (Fig 2, A–F). Fuchsin-stained preparations did reveal a difference in cell death between pie^EB3 and wild-type eye discs, however. Although cell death is rare in wild-type eye discs (WOLFF and READY 1991 ), abundant apoptotic bodies were evident in pie^EB3 eye discs, both anterior and posterior to the morphogenetic furrow (Fig 2B and Fig F). Since the rough eye phenotype seen in GMRp21 is thought to be due to a shortfall in the number of cells produced by the SMW, it seemed possible that a similar outcome could result from elevated cell death.

View larger version (111K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Cell cycle and survival in pie. A–F show portions of eye imaginal discs (anterior to the left), labeled either for differentiating photoreceptor cells (ELAV antigen in magenta; A, C, and E) and cyclin B (green; A, C, and E) or with basic fuchsin (B, D, and E). Vertical arrowheads in A–F show the position of column 0. A and B show wild type. Cyclin B labels cells that reenter the cell cycle in the second mitotic wave posterior to the morphogenetic furrow (A). Basic fuchsin labels mitotic figures in the second mitotic wave. (C, D, and F) pieEB3 homozygotes. Cyclin B (C) and basic fuchsin (D) identify a second mitotic wave similar to that of wild type. Patterning of ommatidia becomes progressively more abnormal posterior to the morphogenetic furrow, as described previously (BAKER et al. 1992 ). (E) GMRp21. Cyclin B reveals normal proliferation anterior to the morphogenetic furrow, but no second mitotic wave posterior to the morphogenetic furrow, quite unlike either wild type or pieEB3 (A and C). (F) A more basal focal plane of the same fuchsin-stained pieEB3 preparation shown in D reveals many apoptotic bodies (arrows). By contrast dying cells are almost undetectable in wild-type eye discs (not shown). (G) Scanning electron microscopy shows that the adult eye of pieEB3 GMRp35 is much less abnormal than that of pieEB3 homozygote (compare Fig 1B). (H) Eyes from GMRp35 strains are also slightly rough.

If the pie^EB3 phenotype was caused by excess cell death posterior to the morphogenetic furrow, we predicted that preventing cell death posterior to the morphogenetic furrow would suppress the pieEB3 phenotype. The GMRp35 transgene was used to suppress cell death. GMRp35 expresses the caspase inhibitor protein baculovirus p35 posterior to the morphogenetic furrow (HAY et al. 1994 ). Basic fuchsin staining confirmed the absence of apoptotic bodies posterior to the morphogenetic furrow in pie^EB3 GMRp35 eye discs, although apoptosis was abundant anterior to the morphogenetic furrow where no p35 protein was expressed (not shown). Adult pie^EB3 GMRp35 flies showed more normal eye morphology than did pie^EB3 homozygotes, although the eyes were still slightly rough and contained some facets of abnormal size (Fig 2G and Fig H).

These results indicate that the pie^EB3 mutation causes cell death in the eye imaginal disc and that cell death posterior to the morphogenetic furrow contributes to the shortfall of ommatidial cells and to the roughened eye. Because GMRp35 did not suppress the rough eye phenotype completely, it cannot be excluded that pie^EB3 may affect other processes in addition to cell survival. It is also possible, however, that the residual eye roughness is due to cell death triggered within the morphogenetic furrow prior to GMR-driven p35 expression. In addition it should be noted that GMRp35 causes mild eye roughening by itself, due to activity of the p35-insensitive caspase Dronc (YU et al. 2002 ). An attempt was made to assess the effect of earlier blockade of all caspases through mutation of head involution defective (hid), a gene required for all caspase activity and apoptosis during eye development (KURADA and WHITE 1998 ; YU et al. 2002 ). Since even trans-heterozygous pie^EB3/+; hid/+ flies proved inviable, however, the pie hid double mutant phenotype could not be examined.

Other imaginal discs were examined to see whether pie was required only in eye discs. Basic fuchsin staining identified abnormal excess cell death in wing and leg imaginal discs (not shown but see also Fig 4). In addition, it was noted that pie^EB3 homozygotes were delayed developmentally (Fig 3). The average (mean) pie^EB3 homozygote emerged after 13.5 days, 2.5 days later than the average for pie^EB3/+ controls. In addition, fewer pie^EB3 homozygotes than predicted were obtained from Mendelian ratios. The pie^EB3 homozygotes had normal bristle size and morphology (not shown), unlike Minute flies that are developmentally delayed due to reduced translation (MORATA and RIPOLL 1975 ; LAMBERTSSON 1998 ) or diminutive flies that show reduced cellular growth (JOHNSTON et al. 1999 ). This difference suggests that the delay in pie^EB3 homozygotes might not be due to deficient translation or growth.

View larger version (19K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Developmental delay in pie. Adult emergence of pieEB3 homozygotes and heterozygous pieEB3/+ siblings from the same cultures is shown. Flies were reared under uncrowded conditions on rich yeast-glucose food medium to minimize competition for food and resources. The adult pieEB3 homozygotes typically emerged 3 days later than controls and comprised 23% of the adult population instead of the expected 50%.

View larger version (106K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Cell death in pie mutant clones. (A) Head containing pieEB3 homozygous clones (unpigmented) generated by eyFlp-mediated mitotic recombination. Homozygous mutant tissue occupies as much of the eye as the darkly pigmented twin spots and shows almost normal morphology, unlike the rough appearance of entirely mutant pieEB3 eyes (compare Fig 1B). (B) Head containing pieE1-16 homozygous clones (unpigmented) generated by eyFlp-mediated mitotic recombination. Homozygous clones occupy little of the adult eye, which has a roughened appearance. C–F show differentiation in eye discs containing clones of cells homozygous for pieEB3 (C and E) or pieE1-16 (D and F) mutations. ELAV labeling of differentiating photoreceptor cells in C and D shows neighboring ommatidial clusters touching due to lack of intervening nonneural cells (arrows). The defect is less extreme in pieEB3 clones than in pieEB3 homozygotes, however (compare C with Fig 2C). Cone cells detected by cut expression are also abnormally arranged within mutant clones (E, pieEB3; F, pieE1-16). G–J show imaginal discs labeled for apoptotic cells containing activated caspases. G and H show eye imaginal discs containing clones of cells homozygous for pieEB3 (G) or pieE1-16 (H). I and J show wing imaginal discs containing clones of cells homozygous for pieEB3 (I) or pieE1-16 (J). Apoptotic cells are seen within mutant clones both anterior and posterior to the morphogenetic furrow in the eye disc and in wing discs. Cell death is less abundant in pieEB3 clones than in pieE1-16 clones. pieEB3 clones are also larger than pieE1-16 clones. Note that cell death occurs fairly evenly through mutant clones, does not seem concentrated in particular disc regions, and is neither more nor less prevalent near clonal boundaries with wild-type cells.

pie null alleles are homozygous lethal:
Deficiency chromosomes were used to map the pie locus precisely. In a previous study pie had been mapped to chromosome interval 32A (BAKER et al. 1992 ), but this proved to be an error. Instead we found that pie^EB3 failed to complement deficiencies in the 31E region. Specifically, pie^EB3 failed to complement Df(2L)J2, Df(2L)J27, Df(2L)J39, Df(2L)J77, and Df(2L)J106, but complemented Df(2L)J16 and Df(2L)J17, consistent with location of pie in 31E [see CLEGG et al. 1993 for details of these deficiencies].

A total of 66,500 flies derived from X-irradiated germ cells (4000 rads) were screened for failure to complement pie^EB3 to isolate new alleles. Several individuals appearing to carry newly induced pie mutations were identified, but in no case did such individuals breed successfully, and the putative new mutations could not be recovered. In addition the pie^EB3/deficiency phenotype was semilethal. Those adults that do survive are extremely sickly; the females were invariably sterile and the males bred very poorly. These findings suggested that the pie^EB3 mutation was hypomorphic and that the pie null phenotype might include lethality and/or sterility. To explore this, complementation was tested between pie^EB3 and representatives of eight lethal complementation groups in the 31E region that, like pie^EB3, complemented Df(2L)J16 but not Df(2L)J77 (CLEGG et al. 1993 ). All of these complemented pie^EB3 except l(2)31Ek, indicating that pie^EB3 was a viable allele of this locus.

Five alleles of l(2)31Ek were obtained from existing strains, and all trans combinations of these alleles with one another, with pie^EB3, and with Df(2L)J77 were examined to identify putative null alleles of the pie locus. As noted previously, combinations of l(2)31Ek mutations show complex complementation and diverse phenotypes (CLEGG et al. 1993 ). The results are summarized in Table 1. pie^EB3 is the only homozygous viable allele with an abnormal eye. Multiple trans-allelic combinations between lethal alleles are viable with similar rough eyes, however, and the homozygous viable, female sterile allele pie^QM47 has a rough eye in trans to certain other pie alleles. Several genotypes, including pie^G2-4/l(2)54 and pie^E1-16/l(2)54, also exhibit held-up wings and loss of wing margin material.

The complex complementation pattern reported in Table 1 precluded a simple allelic series for pie alleles. Two of the alleles behaved most like deficiencies and were studied further. These were pie^E1-16, induced by EMS mutagenesis, and pie^G2-4, induced by gamma-irradiation.

pie null mutations affect cell survival autonomously:
Since both pie^E1-16 and pie^G2-4 were homozygous lethal, mosaics were studied to determine the effects of pie loss of function on imaginal development. Clones of pie^E1-16 homozygous cells appeared rougher in adult eyes than clones of pie^EB3 did (Fig 4A and Fig B). Imaginal disc clones were examined by confocal microscopy after labeling with antibodies against Elav, to detect differentiating photoreceptor cells, or against Cut, to detect nonneuronal cone cells (Fig 4, C–F). Whereas clones of pie^EB3 developed almost normally (Fig 4C and Fig E), pie^E1-16 clones contained smaller ommatidial clusters showing a variable shortfall in photoreceptor and cone cell differentiation (Fig 4D and Fig F).

The development of pie^E1-16 clones resembled that of pie^EB3 homozygotes, suggesting that pie^E1-16 also caused imaginal disc cell death. To test this, pie^E1-16 clones were labeled with an antibody that recognizes activated caspases. Caspases were activated in many pie^E1-16 homozygous cells in eye disc and wing disc clones and the morphology of the labeled cells supported the hypothesis that they were apoptotic (Fig 4H and Fig J). Not all homozygous pie^E1-16 cells label for activated caspase at any given time. Apoptotic cells were evenly distributed through the clones, not obviously correlated with distance to wild-type cells outside the clone, consistent with an autonomous effect on cell survival (Fig 4H and Fig J).

The development of pie^EB3 clones was less abnormal than that of pie^EB3 homozygotes, suggesting that the pie^EB3 mutation might act nonautonomously. To test this, pie^EB3 clones were labeled with an antibody that recognizes activated caspases. pie^EB3 clones showed abundant cell death, similar to pie^EB3 homozygotes, suggesting that pie^EB3 affects cell survival autonomously (Fig 4G and Fig I). As for pie^E1-16, apoptotic cells were distributed evenly through pie^EB3 clones. We suggest that in mosaics, normal cells from outside pie^EB3 clones are recruited to cell fates in place of pie^EB3 homozygous cells that have died, permitting more normal development of pie^EB3 clones than is possible for pie^EB3 homozygotes.

Molecular identification of the pie gene:
As a first step toward locating the pie gene, genomic cosmid clones from the 31E region were obtained (SIDEN-KIAMOS et al. 1990 ) and restriction mapping and Southern blotting were used to prepare a map of the 31E region (Fig 5A). PCR primers were based on the sequence of the da gene (which maps left of pie genetically) within clone 100A9. Reduced Southern blot signals in heterozygotes located the proximal break of Df(2L)J77 within clone 192F1. Df(2L)J77 is null for pie, placing the pie gene within a genomic region of 70 kb (Fig 5A).

View larger version (44K): [in this window] [in a new window] [Download PPT slide]   Figure 5. Genomic organization of the pie region. (A) Three cosmid clones from the 31E region were determined by restriction mapping and cross-hybridization to cover 90 kb of genomic DNA including the daughterless gene and the proximal breakpoint of Df(2L)J77. The position of the inversion breakpoint associated with pieG2-4 is shown and assigned as the origin on the arbitrary scale. (B) More detailed map of the pieG2-4 breakpoint region including genomic organization of three transcription units identified by screening an imaginal disc cDNA library. The breakpoint maps within 300 bp close to the 3' end of the pie transcript. Coding regions are indicated by solid shading of the exon bars. (C) Amino acid sequence predicted by the pie open reading frame. The allele pieE1-16 is frameshifted at Asp203 (boldface, underlined). The allele pieEB3 replaces Gln391 with a stop codon (boldface, underlined). A consensus defined for molybdopterin binding, [GA]-X(3)-[KRNQHT]-X(11,14)-[LIVMFYWS]-X(8)-[LIVMF]-X-C-X(2)-[DEN]-R-X(2)-[DE] (WOOTON et al. 1991), is matched by amino acids Gly156, Arg160, Leu174, Ile183, Cys185, Glu188, Arg189, and Asp192 of the predicted PIE protein.

Since radiation-induced mutations are often associated with DNA lesions detectable by Southern blotting, DNA from the gamma-induced mutation pie^G2-4 was compared with control DNA across the critical region. A single polymorphism was detected 29 kb to the right of da. Southern blots with multiple enzymes mapped a breakpoint that was not present in the unmutagenized progenitor strain to a 300-bp segment defined by EcoRI and BamHI restriction sites (Fig 5A and Fig B). Since no deletion or duplication was indicated, the Southern analysis predicted that pie^G2-4 was associated with an inversion. Polytene chromosomes were examined to test this. The pie^G2-4 chromosome was found to contain a cytologically visible inversion between chromosome regions 31E and 32A (not shown). These findings were consistent with the model that gamma-irradiation induced a chromosome inversion breaking within the pie gene to generate the pie^G2-4 allele.

To identify genes affected by the pie^G2-4 inversion breakpoint in 31E, genomic BamHI fragments around the breakpoint (8.3 kb in total) were used to probe a cDNA library derived from imaginal disc RNA (gift of A. Cowman and G. M. Rubin). Inserts from positive clones were characterized by restriction map and cross-hybridization and fell into three discrete classes, indicating three transcription units in the pie^G2-4 region (Fig 5B). The largest clone was sequenced for each of the three cDNA classes, and corresponding genomic DNA was also sequenced to establish the intron-exon structure. The leftmost clone contained a 3242-bp cDNA predicted to encode a novel kinesin-like protein (GenBank accession no. AF247500). The second clone contained an 1148-bp cDNA predicted to encode a replication factor C protein (GenBank accession no. AF247499). The most centromeric clone contained an 1852-bp cDNA encoding a novel protein (GenBank accession no. AF247501). The G2-4 breakpoint mapped toward the 3' end of this transcript (Fig 5B).

We predicted that if the centromeric 1852-bp cDNA corresponded to the pie gene, the open reading frame might be altered in point mutants. Genomic DNA was sequenced from the EMS-induced pie^EB3 and pie^E1-16 alleles and compared with the unmutagenized control chromosomes to assess this. Whereas neither the kinesin-like gene nor the RFC gene was altered in these mutations, the third gene was altered in both. The pie^EB3 chromosome contained a C-to-T transition at position 1213 compared to the cDNA sequence, substituting a TAG stop codon for the CAG codon for Gln391 of the predicted protein (Fig 5C). The pie^E1-16 chromosome contained a 13-bp deletion corresponding to nucleotides 644–656 of the cDNA, predicting the substitution of a novel sequence of 16 amino acids followed by a termination codon for Asp203 (Fig 5C). Taken together with the rearranged transcription unit in the pie^G2-4 allele, these results confirm the identity of this novel open reading frame with the pie gene. Truncation of the pie^E1-16 product earlier than that of pie^EB3 may explain why pie^EB3 is hypomorphic compared to pie^E1-16 and pie^G2-4.

The pie gene sequence predicts a protein of 582 amino acids that lacks apparent secretory signal sequences, potential transmembrane domains, or recognized conserved domains (Fig 5C). Since it also lacks apparent nuclear localization or mitochondrial import sequences, it is possible that pie encodes a cytoplasmic protein. Inspection of the sequence suggests that the PIE protein can be viewed as containing two domains (Fig 5C). Cysteine is unusually common among amino acids 9–281 (27 of these 273 codons encoded cysteine). Proline is unusually common among amino acids 291–505 (31 of these 215 codons encoded proline).

Database searches identified several predicted genes of unknown function from human, mice, and mosquitoes that contained regions highly similar to the Cys-rich region of PIE. Fig 6 shows an alignment of seven related sequences and the Cys-rich domain from PIE. All but one of these domains show 30–35% amino acid identity with PIE and 50% similarity. This very high degree of similarity speaks to a highly conserved structure and molecular function. No one of these seven sequences appears significantly more closely related to PIE, and which if any of the vertebrate sequences might be a pie ortholog is not apparent. Although the arrangement of cysteines is strongly conserved, there are several examples of nonconserved cysteines, and Cys118 from PIE is replaced by histidine in all the other sequences. This observation, along with the lack of apparent transmembrane or signal sequences and the presence of several perfectly conserved histidines, suggests the PIE protein might be involved in metal binding. In addition, the PIE sequence contains the consensus for molybdopterin binding (Fig 5C; Interpro IPR000572; Prosite PS00559). Molybdopterin is the molybdenum cofactor for all molybdenum-containing enzymes except one (WOOTTON et al. 1991 ). None of the other proteins match the molybdopterin-binding consensus, and we therefore suspect that the consensus in PIE may be fortuitous.

View larger version (98K): [in this window] [in a new window] [Download PPT slide]   Figure 6. Comparison of PIE to other genes. BLAST searches detect a number of other predicted genes sharing cysteine-rich domains to the amino-terminal portion of the predicted PIE protein (AF247501). Shown are alignments between amino acids 1–281 from PIE and portions of predicted proteins from Anopheles gambiae (EAA07617 and EAA00707), D. melanogaster (NM1345030 corresponding to CG9576), Mus musculus (XM_126910 and NM027949), and Homo sapiens (AB037554, BC02202, and T46480). Amino acids identical with the corresponding PIE amino acid are shown in magenta; conservative substitutions are shown in green. AB037554 and XM_126910 are likely human and mouse orthologs of one another, as are BC02202 and NM027949, but evolutionary relationships among the other sequences are less clear (not shown).

Many Pro-rich protein sequences were found by similarity searches with the Pro-rich domain of PIE but none appeared to resemble PIE specifically or also to contain the Cys-rich PIE domain. However, the second Drosophila gene identified as sharing the PIE Cys-rich domain (Fig 6), which corresponds to predicted gene CG9576 (ADAMS et al. 2000 ), resembles PIE in also containing a carboxyl region rich in Pro, Ser, and acidic residues, although these regions from PIE and CG9576 cannot be satisfactorily aligned (not shown).

Attempts to express portions of the PIE protein as bacterial fusion proteins were largely unsuccessful; only a fusion of the carboxy-terminal 67 amino acids fused to the carboxyl terminus of glutathione S-transferase has been obtained. Immunization with this protein produced mouse antisera that were specific for the PIE-specific portion of the fusion protein on Western blots, but could not detect endogenous PIE protein products by Western blotting or immunostaining of Drosophila tissues or cells. The pie gene must be expressed in imaginal disc cells, however, since it was required there cell autonomously.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In this article we used the adult eye phenotype associated with defective cell number to identify a candidate gene for cell number regulation and have shown that this pineapple eye gene encodes a putative cytoplasmic protein required for proper cell survival. Unlike some cell lethal mutations, pie mutations are not absolutely inviable, but instead predispose cells in imaginal discs to a high rate of apoptosis. Apoptosis is identifiable by activation of endogenous caspases and preventable by retinal expression of the caspase inhibitor p35. Within imaginal discs of pie mutants, apoptosis occurred indiscriminately at many locations, and no obvious spatial pattern of sensitivity was observed. We focused on eye and wing imaginal discs but noticed cell death in other imaginal discs also (data not shown). It is intriguing that the main function of pie should seem to be reducing the rate of apoptosis, but as yet we have no clue about the molecular or biochemical function of the protein product. Nevertheless, the pie gene contains a domain of 270 amino acids with striking homology throughout the animal kingdom. So far all these genes are of unknown biochemical function, although we suspect that this may be a metal-binding domain.

The pie mutant phenotype illustrates the distinct consequences of cell death at different developmental stages. Retinal cell death posterior to the morphogenetic furrow leads to a shortfall in retinal precursor cells and so to defects in the ommatidial structure of the mature retina. These defects resemble those seen when the second mitotic wave is blocked. This emphasizes the importance of adequate cell number for retinal development and confirms that the second mitotic wave is important for providing adequate precursor cells (DE NOOIJ and HARIHARAN 1995 ). The pie phenotype further indicates that retinal cells cannot be replaced after the second mitotic wave. Even though pie mutants have seemingly normal cell proliferation, cells lost posterior to the morphogenetic furrow are not replaced by compensatory cell divisions. An indication of this fact also came from previous findings that X-ray damage incidental to induction of mitotic recombination was associated with defects in ommatidial structure among cells posterior to the morphogenetic furrow at the time of irradiation, but irradiation earlier had little effect on retinal structure (BECKER 1957 ).

Remarkably, the retina is the exception in exhibiting morphological defects as a consequence of high rates of cell death (along with the wing margin, which is also abnormal in certain allelic combinations). Developmental delay seems to be the main effect of cell death in other tissues, without obvious morphological consequences. Previous studies of imaginal disc damage indicate that imaginal discs need to attain a critical size to trigger metamorphosis (RUSSELL 1974 ; SIMPSON and SCHNEIDERMAN 1975 ). More accurately, since removing entire imaginal discs has no effect on developmental timing, presence of immature or growing imaginal discs must inhibit metamorphosis (SIMPSON et al. 1980 ; SZABAD and BRYANT 1982 ). The pie phenotype suggests that the critical threshold beyond which imaginal discs cease preventing metamorphosis must be cell number or a property dependent on cell number, such as tissue mass or range of pattern.

The lack of morphological consequences of cell death in the pie mutant also contrasts with two other phenomena associated with cell death, namely pattern duplication and cell competition. Pattern anomalies such as leg duplications and triplications have not been seen in pie mutants although they are commonplace when clones of conditionally lethal cells die [reviewed in MEINHARDT 1983 ]. Such duplications are thought to result from inappropriate apposition of distinct cell populations on death of intervening cells, if the distinct cell populations interact to induce a new organizing region in the imaginal disc (MEINHARDT 1983 ). One possibility is that apoptotic cell death in pie removes cells in a different way from the uncharacterized death mechanisms of conditional-lethal cells. Alternatively, we speculate that stochastic apoptosis in pie mutants occurs in a salt-and-pepper fashion so that sufficient cells always survive to buffer spatially distinct cell populations.

Cell competition is another phenomenon associated with cell death. Cell competition occurs when cell populations with different growth rates are apposed within the same compartment, such as occurs when clones of genetically wild-type cells are induced by mitotic recombination in a Minute heterozygous background (MORATA and RIPOLL 1975 ). Doubling of the faster growing cells is accelerated at the expense of the slower genotype, which can actually be eliminated from the compartment where it would have flourished had wild-type cells not been introduced. Recently several studies have indicated that apoptosis plays a significant role in eliminating the slower growing, competed cells (NEUFELD et al. 1998 ; MORENO et al. 2002 ). By contrast, clones of cells homozygous for pie mutations survive late into larval and even adult life, despite exhibiting much higher rates of apoptosis than surrounding cells. The apparent failure of pie homozygous clones to suffer from cell competition raises the possibility that apoptosis is not sufficient for cell competition to occur or that apoptosis in cell competition has particular spatial or biochemical properties that are not shared by apoptosis due to other causes.

In summary, we interpret two aspects of the pie phenotype to represent distinct consequences of the underlying cell death and to reflect changing importance of proper cell number during imaginal disc development. During the bulk of larval life the main role of increasing imaginal disc cell number is to provide adequate material for adult tissues. Once this is attained and adult differentiation begins, organs such as the adult retina must maintain precise cell numbers because almost every cell is specified for a particular fate and cannot be replaced if lost. In other organs, such as legs or wings, the most cells take relatively uniform epidermal fates and can substitute for one another with little consequence. These changing roles may in part justify temporal differences in cell cycle control, in which signals such as EGFR and Hh play essential cell cycle and survival roles in differentiating eye discs that differ from their roles during prior imaginal disc growth (BAKER and YU 2001 ; DUMAN-SCHEEL et al. 2002 ).

	FOOTNOTES

Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AF247499, AF247500, AF247501.

	ACKNOWLEDGMENTS

We thank our colleagues and I. Hariharan and K. Moses for reading the manuscript. We are grateful to R. Mottus, T. Grigliatti, and T. Schupbach for Drosophila strains; I. Siden-Kiamos for cosmids; G. Rubin for cDNA libraries; and P. O'Farrell, the Developmental Studies Hybridoma Bank at the University of Iowa, A. Srinivasan, and Idun Pharmaceuticals for antibodies. This work was supported by grants from the Howard Hughes Medical Institute Research Resources Program for Medical Schools, the U.S. Army Medical Research and Material Command, and the National Institutes of Health (GM-61230).

Manuscript received April 10, 2003; Accepted for publication August 12, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ADAMS, M. D., S. E. CELNIKER, R. A. HOLT, C. A. EVANS, and J. D. GOCAYNE et al., 2000  The genome sequence of Drosophila melanogaster.. Science 287:2185-2195.[Abstract/Free Full Text]

BAKER, N. E., 2001  Proliferation, survival and death in the Drosophila eye. Semin. Cell Dev. Biol. 12:499-507.[Medline]

BAKER, N. E. and S.-Y. YU, 2001  The EGF receptor defines domains of cell cycle progression and survival to regulate cell number in the developing Drosophila eye. Cell 104:699-708.[Medline]

BAKER, N. E., K. MOSES, D. NAKAHARA, M. C. ELLIS, and R. W. CARTHEW et al., 1992  Mutations on the second chromosome affecting the Drosophila eye. J. Neurogenet. 8:85-100.[Medline]

BECKER, H. J., 1957  Uber Rontgenmosaikflecken und Defektmutationen am Auge von Drosophila und die Entwicklungsphysiologue des Auges. Z. Indukt. Abstammungs. Vererbungsl. 88:333-373.

CLEGG, N. J., I. P. WHITEHEAD, J. K. BROCK, D. A. SINCLAIR, and R. MOTTUS et al., 1993  A cytogenetic analysis of chromosome region 31 of Drosophila melanogaster.. Genetics 134:221-230.[Abstract]

DE NOOIJ, J. C. and I. K. HARIHARAN, 1995  Uncoupling cell fate determination from patterned cell division in the Drosophila eye. Science 270:983-985.[Abstract/Free Full Text]

DUMAN-SCHEEL, M., L. WENG, S. XIN, and W. DU, 2002  Hedgehog regulates cell growth and proliferation by inducing Cyclin D and Cyclin E. Nature 417:299-304.[Medline]

EDGAR, B. A., J. BRITTON, A. F. DE LA CRUZ, L. A. JOHNSTON, and D. LEHMAN et al., 2001  Pattern- and growth-linked cell cycles in Drosophila development. Novartis Found. Symp. 237:36-42.

EVANS, T., E. T. ROSENTHAL, J. YOUNGBLOM, and T. HUNT, 1983  Cyclin: a protein specified by maternal mRNA in sea urchin eggs that is destroyed at each cleavage division. Cell 33:389-396.[Medline]

FU, W. and M. NOLL, 1997  The Pax2 homolog sparkling is required for development of cone and pigment cells in the Drosophila eye. Genes Dev. 11:2066-2078.[Abstract/Free Full Text]

GAUL, U., G. MARDON, and G. M. RUBIN, 1992  A putative Ras GTPase activating protein acts as a negative regulator of signaling by the sevenless receptor tyrosine kinase. Cell 68:1007-1019.[Medline]

GRETHER, M. E., J. M. ABRAMS, J. AGAPITE, K. WHITE, and H. STELLER, 1995  The head involution defective gene of Drosophila melanogaster functions in programmed cell death. Genes Dev. 9:1694-1708.[Abstract/Free Full Text]

HAY, B. A., T. WOLFF, and G. M. RUBIN, 1994  Expression of baculovirus P35 prevents cell death in Drosophila.. Development 120:2121-2129.[Abstract]

JOHNSTON, L. A. and P. GALLANT, 2002  Control of growth and organ size in Drosophila.. BioEssays 24:54-64.[Medline]

JOHNSTON, L. A., D. A. PROBER, B. A. EDGAR, R. N. EISENMAN, and P. GALLANT, 1999  Drosophila myc regulates cellular growth during development. Cell 98:779-790.[Medline]

KNOBLICH, J. A. and C. F. LEHNER, 1993  Synergistic action of Drosophila cyclin A and B during the G₂-M transition. EMBO J. 12:65-74.[Medline]

KURADA, P. and K. WHITE, 1998  Ras promotes cell survival in Drosophila by downregulating hid expression. Cell 95:319-329.[Medline]

LAMBERTSSON, A., 1998  The Minute genes in Drosophila and their molecular functions. Adv. Genet. 38:69-134.[Medline]

LEE, E.-C., S.-Y. YU, and N. E. BAKER, 2000  The SCABROUS protein can act as an extracellular antagonist of Notch signaling in the Drosophila wing. Curr. Biol. 10:931-934.[Medline]

MEINHARDT, H., 1983  Cell determination boundaries as organising regions for secondary embryonic fields. Dev. Biol. 96:375-385.[Medline]

MORATA, G. and P. RIPOLL, 1975  Minutes: mutants of Drosophila autonomously affecting cell division rate. Dev. Biol. 42:211-221.[Medline]

MORENO, E., K. BASLER, and G. MORATA, 2002  Cells compete for decapentaplegic survival factor to prevent apoptosis in Drosophila wing development. Nature 416:755-759.[Medline]

NEUFELD, T. P., and I. K. HARIHARAN, 2002 Regulation of growth and cell proliferation in eye development, pp. 107–133 in Drosophila Eye Development, edited by K. MOSES. Springer Verlag, New York.

NEUFELD, T. P., A. F. DE LA CRUZ, L. A. JOHNSTON, and B. A. EDGAR, 1998  Coordination of growth and cell division in the Drosophila wing. Cell 93:1183-1193.[Medline]

OLDHAM, S., R. BOHNI, H. STOCKER, W. BROGIOLO, and E. HAFEN, 2000  Genetic control of size in Drosophila. Philos. Trans. R. Soc. Lond. B 355:945-952.[Medline]

RUSSELL, M. A., 1974  Pattern formation in the imaginal discs of a temperature-sensitive cell-lethal mutant of Drosophila melanogaster.. Dev. Biol. 40:24-39.[Medline]

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual, Ed. 2. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SANDLER, L., 1977  Evidence for a set of closely linked autosomal genes that interact with sex chromosome heterochromatin in Drosophila melanogaster.. Genetics 86:567-582.[Abstract/Free Full Text]

SCHUPBACH, G. and E. WEISCHAUS, 1986  Maternal-effect mutations altering the anterior-posterior pattern of the Drosophila embryo. Roux's Arch. Dev. Biol. 195:302-317.

SIDEN-KIAMOS, I., R. D. SAUNDERS, L. SPANOS, T. MAJERUS, and J. TREANEAR et al., 1990  Towards a physical map of the Drosophila melanogaster genome: mapping of cosmid clones within defined genomic regions. Nucleic Acids Res. 18:6261-6270.[Abstract/Free Full Text]

SIMPSON, P. and H. A. SCHNEIDERMAN, 1975  Isolation of temperature-sensitive mutations blocking clone development in Drosophila melanogaster, and the effects of a temperature-sensitive cell lethal mutation on pattern formation in imaginal discs. Roux's Arch. Dev. Biol. 178:247-275.

SIMPSON, P., P. BERREUR, and J. BERREUR-BONNENFANT, 1980  The initiation of pupariation in Drosophila: dependence on growth of the imaginal discs. J. Embryol. Exp. Morphol. 57:155-165.[Medline]

SZABAD, J. and P. J. BRYANT, 1982  The mode of action of "discless" mutations in Drosophila melanogaster.. Dev. Biol. 93:240-256.[Medline]

TOROK, T., G. TICK, M. ALVARADO, and I. KISS, 1993  P-LacW insertional mutagenesis on the second chromosome of Drosophila melanogaster: isolation of lethals with different overgrowth phenotypes. Genetics 135:71-80.[Abstract]

WOLFF, T. and D. F. READY, 1991  Cell death in normal and rough eye mutants of Drosophila.. Development 113:825-839.[Abstract]

WOLFF, T., and D. F. READY, 1993 Pattern formation in the Drosophila retina, pp. 1277–1326 in The Development of Drosophila melanogaster, edited by M. BATE and A. MARTINEZ ARIAS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

WOOTTON, J. C., R. E. NICOLSON, J. M. COCK, D. E. WALTERS, and J. F. BURKE et al., 1991  Enzymes depending on the pterin molybdenum cofactor: sequence families, spectroscopic properties of molybdenum and possible cofactor-binding domains. Biochim. Biophys. Acta 1057:157-185.[Medline]

YU, S.-Y., S. J. YOO, L. YANG, C. ZAPATA, and A. SRINIVASAN et al., 2002  A pathway of signals regulating effector and initiator caspases in the developing Drosophila eye. Development 129:3269-3278.[Abstract/Free Full Text]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Shi, W. Articles by Baker, N. E. Search for Related Content PubMed PubMed Citation Articles by Shi, W. Articles by Baker, N. E.