The PTEN protein family. The predicted DPTEN protein was compared with human PTEN (hPTEN; Steck et al. 1997; Li et al. 1997a), C. elegans PTEN (DAF-18; the carboxy-terminal portion of this 965-amino-acid protein is omitted; Ogg and Ruvkun 1998) and the yeast PTEN homolog TEP1 (yTEP1 or YNL128W; Li et al. 1997b). Identical amino acids are shaded in black and conservative changes in gray. The extent of the tensin domain is demarcated by arrows. Three regions within this domain that are required for phosphatase activity are underscored, including the highly conserved phosphatase signature motif (bold underline; Myers et al. 1997). The positions of missense mutations identified in cancer or in dominant genetic disorders are marked by asterisks (Maehama and Dixon 1999). All but one of the affected residues are identical in DPTEN. Crosses above the sequence highlight PTEN-specific amino acids that are not found in other known members of either the phosphatase or tensin families.

DPTEN controls cell size and architecture. (A–C) Cross sections through eyes containing clonal tissue homozygous for a w DPTEN¹ chromosome (A,C) and a w⁺ DPTEN³ chromosome (B). In A and C, w mutant photoreceptors can be distinguished from their phenotypically wild-type pigmented neighbors (cf. Fig. 3A). In contrast, w⁺ mutant pigment cells in B carry two copies of the the w⁺ gene and are therefore more pigmented than their phenotypically wild-type neighbors. In A and B, mutant (black arrow) and normal heterozygous (white arrow) tissues are indicated. Although the polarity of mutant ommatidia is largely unaffected, w photoreceptor cell bodies (A; only low levels of pigment are seen in the heterozygous tissue in this sample) and w⁺ pigment cells (B) lacking DPTEN are increased in size. Also note the elliptical structure of mutant rhabdomeres. Rhabdomeres from mutant and phenotypically wild-type photoreceptors are marked with black and white arrowheads respectively in C. Multiple ommatidia of mixed genotype in C reveal that the effects on size and rhabdomere morphology are cell-autonomous; the inset shows a single ommatidium (black arrow) in which the two mutant photoreceptors present are marked (black arrowheads) together with two phenotypically wild-type photoreceptors (white arrowheads). (D,E) Wings containing clonal tissue homozygous for Df(2L)170B (D; note that this fly also carried the DJNK genomic rescue construct pWZ; Riesgo-Escovar et al. 1996) and for DPTEN¹ (E). Clones were marked with the forked mutation (unshaded regions). Mutant cells are enlarged, and frequently cross-veins fail to extend from wild-type tissue (white arrowheads) into the mutant region (black arrowheads), although occasionally additional cross-veins are observed (data not shown). A minority of mutant cells contains duplicated wing hairs (black arrows). The genotypes of the flies were y, w, hsFLP1; DPTEN¹, P[ry⁺; hs–neo; FRT]40A/P[ry⁺; w⁺]30C, P[ry⁺; hs–neo; FRT]40A (A,C), y, w, hsFLP1; P[ry⁺; w⁺]30C, DPTEN³, P[ry⁺; hs–neo; FRT]40A/P[ry⁺; hs–neo; FRT]40A (B), y, w, hsFLP1, f^36a; Df(2L)170B, P[ry⁺; hs–neo; FRT]40A/P [f⁺]30B, P[ry⁺; hs–neo; FRT]40A; pWZ/+ (D) and y, w, hsFLP1, f^36a; DPTEN¹, P[ry⁺; hs–neo; FRT]40A/P[f⁺]30B, P[ry⁺; hs–neo; FRT]40A (E).

DPTEN completely suppresses the growth-promoting activity of overexpressed Dp110 in the wing. Wings from flies carrying the en–GAL4 driver alone (A), or overexpressing Dp110 (B), DPTEN (C), or both Dp110 and DPTEN (D) under en–GAL4/UAS control in the posterior compartment of the wing. The compartment boundary lies just anterior to LIV; the posterior compartment therefore includes wing veins LIV and LV (labeled in A and marked with arrowheads in the other panels). Although the wing produced by overexpression of Dp110 is crumpled and distorted, note that its posterior compartment is enlarged compared with the anterior compartment, a phenotype most clearly seen in the region between LIV and LV. In contrast, this area is reduced in flies overexpressing DPTEN and overexpressing the combination of DPTEN and Dp110. The genotypes of flies were w; P[w⁺; en–GAL4]/+ (A), w; P[w⁺; UAS–Dp110]/P[w⁺; en–GAL4] (B), w; P[w⁺; UAS–DPTEN]/P[w⁺; en–GAL4] (C), and w; P[w⁺; UAS–Dp110], P[w⁺; UAS–DPTEN]/P[w⁺; en–GAL4] (D).

Schematic diagram illustrating a model for the antagonistic roles of Dp110 and DPTEN in Drosophila growth control. In mammals, Drosophila, and C. elegans, PI3-kinase activity is regulated by signaling cascades that include the insulin receptor and IRS1-4 (Chen et al. 1996; Maehama and Dixon 1998; Ogg and Ruvkun 1998; Böhni et al. 1999). Disruption of this pathway in mice (Baker et al. 1993; Liu et al. 1993) and Drosophila (Böhni et al.1999) has shown that insulin receptor signaling controls tissue and body size. Our data indicate that DPTEN antagonizes this pathway in flies and decreases cell size and number. It seems likely that DPTEN's effects are largely mediated by reducing PIP3 levels and the activity of its putative downstream targets, including Dakt1. DPTEN's protein phosphatase activity, however, could also have a role (see Gu et al. 1999). DPTEN also appears to regulate organization of the actin cytoskeleton via currently uncharacterized mechanisms (dotted lines) that may involve either its protein or lipid phosphatase activity. DPTEN's effects on the cytoskeleton may influence its tumor suppressing activity by modulating processes such as cell spreading and invasiveness.

Molecular characterization of DPTEN. (A) Chromosomal location of DPTEN. The DPTEN gene resides at 31B/C on the second chromosome in a locus including the genes Dror (Wilson et al. 1993), bsk/DJNK (Riesgo-Escovar 1996; Sluss et al. 1996), and chico (Böhni et al. 1999). For our analysis, we employed three deficiencies generated by imprecise excision of a P element that disrupts chico [fs(2)4¹; see Sluss et al. 1996; Böhni et al. 1999]. Df(2L)41C deletes the 5′ half of the Dror gene; embryos homozygous for this deficiency fail to express Dror transcripts (data not shown), indicating that it is null for Dror. EMS-induced mutations in DPTEN map proximal to Dror, as they fail to complement the largest deficiency, Df(2L)170B, but complement Df(2L)41C and Df(2L)147F. Figure also shows the extent of a genomic rescue construct, vivI, used in this study. Restriction enzyme sites: BamHI (B); EcoRI (E); HindIII (H); SalI (S); XbaI (X). (B) Sequence of the DPTEN genomic locus. Comparison of multiple DPTEN cDNA clones indicates that the DPTEN gene encodes a transcript of at least 2.2 kb (uppercase nucleotides), containing an ORF of 509 amino acids (in bold below the nucleotide sequence; stop codons are represented by asterisks). The first nucleotide of the longest available cDNA is designated nucleotide one. The transcription unit is interrupted by 11 introns (lowercase nucleotides), all including the consensus sequences for 5′ donor and 3′ acceptor splice junctions (Mount 1982). Two methionine residues are found within the first eight residues of the ORF. The amino-terminal methionine, which is preceded by stop codons (underscored) in all three reading frames, is likely to be the translation initiator, because it matches the translation start consensus ([C/A]AA[C/A]ATG; Cavener 1987), except that an additional T has been inserted directly 5′ of the ATG, a feature also found in the eyeless gene (Quiring et al. 1994). The consensus polyadenylation signal site (position 4168; underscored) is located ∼30 bases upstream of the polyA tail (which in different cDNAs is located either at position 4197 or 4203; additional bases in extended transcript are in italics). Some cDNA and genomic sequences have a polymorphic deletion; they lack nine nucleotides from positions 3001 to 3009 (bold). All other polymorphisms do not affect the encoded DPTEN protein; they are also shown in bold (see also Materials and Methods). One cDNA (clone 6C) is derived from an alternatively spliced mRNA (dotted line in Fig. 1A; creating a splice fusion between nucleotides 3124 and 4016) that lacks exon 11 sequences (nucleotides 3397–3609) and therefore encodes a protein containing an additional 5 amino acids (italicized) at its carboxyl terminus.

DPTEN regulates adult eye development. Light micrograph (A) and scanning electron micrographs (B–F) of eye clones homozygous for DPTEN¹ (A,C–F) and Df(2L)170B (B). When compared with heterozygous tissue (white arrows in A–D), mutant regions (black arrows; in A, the mutant region is the nonpigmented region that does not express the w⁺ gene) contain enlarged ommatidial facets and aberrant clusters of up to four interommatidial bristles (seen at higher magnification in E). There are also more ommatidia in mutant clones than in adjacent wild-type twin spots (darker pigmented regions that express two copies of the w⁺ gene; marked with black arrowheads in A). Bristles are nonuniform in thickness, morphology, and length (D–F; shorter in E, several times longer in F). The arrangement of actin bundles within these bristles is abnormal (F). The genotypes of flies were y, w, hsFLP1; DPTEN¹, P[ry⁺; hs–neo; FRT]40A/P[ry⁺; w⁺]30C, P[ry⁺; hs–neo; FRT]40A (A,C–F) and y, w, hsFLP1; Df(2L)170B, P[ry⁺; hs–neo; FRT]40A/P[ry⁺; w⁺]30C, P[ry⁺; hs–neo; FRT]40A (B).

DPTEN acts antagonistically to the PI3-kinase Dp110 in the wing. (A–C) Wings from flies carrying the dpp–GAL4 driver alone (A), or expressing either a kinase-dead, dominant-negative form of Dp110, Dp110^D954A (B; Leevers et al. 1996), or wild-type DPTEN (C) under the control of dpp–GAL4 (Staehling-Hampton and Hoffmann 1994), which expresses GAL4 between wing veins LIII and LIV (arrowheads; wing veins LI to LV are marked in A). The area between these wing veins (called area I) is reduced by ∼25% in both B and C (see Table 1). (D–F) Wings from flies carrying the MS1096 driver alone (D) or expressing Dp110^D954A under the control of the MS1096 driver (E,F), which expresses GAL4 in the wing, particularly on the dorsal surface (Leevers et al. 1996; Capdevila and Guerrero 1994), thereby reducing wing area in E by >30%. In F, the fly is also heterozygous for DPTEN¹, partially suppressing this phenotype. The wing area increases by >8% compared with E, a 25% suppression of the reduced wing phenotype (see Materials and Methods). The genotypes of flies were w; Sp/+; P[w⁺; dpp–GAL4]/+ (A), w; P[w⁺; UAS–Dp110^D954A]/P[w⁺; dpp–GAL4] (B), w; P[w⁺; UAS–DPTEN (FF20.2)]/Sp; P[w⁺; dpp–GAL4]/+ (C), w, P[w⁺; MS1096–GAL4]/w (D), w, P[w⁺; MS1096–GAL4]/w; P[w⁺; UAS–Dp110^D954A]/+ (E), and w, P[w⁺; MS1096–GAL4]/w; DPTEN¹/+; P[w⁺; UAS–Dp110^D954A]/+ (F).

DPTEN affects subcellular localization of the actin cytoskeleton in the developing eye. Panels show confocal micrographs of third-instar larval eye imaginal disks containing DPTEN¹ clones. Anterior is left in all panels. The mutant area is marked by the absence of lacZ (A,B) or GFP (C,D) expression (lacZ- and GFP-positive tissue is shown in red), whereas actin elements are visualized by the actin binding molecule phalloidin (coupled to FITC in A and B and to rhodamine in C and D; green in all panels). The overlay is shown on the left side. White arrows in (A–C) point to the mutant area, which does not stain red. (A) Eye disk with mutant clone containing both uncommitted cells, found anterior to (left half of panel) and within the morphogenetic furrow (black arrowhead), and posterior cells within developing ommatidial preclusters. Actin elements within developing ommatidial clusters are shown in more detail in B. Both A and B show stacked confocal sections, which scan a depth of ∼5 μm in the apical half of the photoreceptor. Mutant cells have an altered distribution of actin microfilaments compared with their phenotypically wild-type neighbors, i.e., in the developing ommatidial clusters shown in B, there is a more intricate network of microfilaments in mutant cells. C and D show low and high magnification views, respectively, of a large DPTEN clone. Note in C, the reduced number of cells in the wild-type twin spot (bright red cells marked with white arrowhead) and the boxed area enlarged in D. D shows the altered organization of the apical cytoskeleton and neighboring microfilaments compared with wild-type tissue (apical projections marked with white arrows). In some instances, wild-type and mutant photoreceptors within the same cluster (marked with white arrowheads) appear to display normal and mutant phenotypes, respectively.