INTRODUCTION

In the development of multicellular organisms, growth anddifferentiation are tightly coordinated. Although significantprogress has been made in the understanding of these twoprocesses individually, we know relatively little about howdifferentiation is integrated with cellular growth. The first steptowards differentiation, the specification of cell fate or tissuetypes, frequently depends on cellular interactions mediated byconserved signaling pathways. One of the best understoodintracellular signaling pathways involves the activation of thesmall G-protein Ras in response to the stimulation of growthfactor receptors including the receptor tyrosine kinase (RTK)family (for review see Boguski and McCormick, 1993;Rommel and Hafen, 1998; Frame and Balmain, 2000).Activation of Ras triggers a highly conserved cascade ofkinases including Raf, MEK and MAP kinase (MAPK). In theDrosophila eye, Ras signaling is required for cell growth,survival of postmitotic cells (Bergmann et al., 1998;Dominguez et al., 1998, Kurada and White, 1998), and thespecification of ommatidial cell fates (reviewed by Freeman,1997). Studies in the Drosophilawing have shown that loss ofRas function reduces cell size (Diaz-Benjumea and Hafen,1994). In the absence of Ras, cell cycle length is prolonged andcells accumulate at the G1/S transition (Prober and Edgar,2000).

The role of Ras in growth control is shared with the PI3-Kinase/Protein Kinase B (PI3K/PKB) pathway (Oldhamet al., 2000a; Weinkove and Leevers, 2000). Cells lackingcomponents of the PI3K/PKB pathway grow more slowly and

divide at a smaller size. Neither partial nor complete loss-of-function mutations in genes encoding components of thispathway affect the differentiation or patterning of adultstructures. In mammalian cells, the Ras/MAPK andPI3K/PKB pathways cooperate in the control of cell cycleprogression. Transition from G1 into S phase of the cell cyclerequires the activation of both pathways (Gille andDownward, 1999; Jones et al., 1999). Little is known,however, about where and how these two pathways convergeto regulate growth. In this paper, we present evidence that theproduct of the lilliputian(lilli) gene cooperates with both theRas/MAPK and the PI3K/PKB pathway in Drosophila in thecontrol of cell fate and cell size.

Mutations in lilli were identified in a screen for dominantsuppressors of the rough eye phenotype caused by constitutiveactivation of Raf during eye development (Dickson et al.,1996). Here we show that Lilli plays a partially redundantfunction downstream of Raf in cell fate specification in thedeveloping eye. In addition, complete loss of Lilli functionleads to a reduction in cell and organ size. The lilligeneencodes a nuclear protein related to the AF4/FMR2 family. Inhumans, mutations affecting genes of this family are associatedwith specific diseases. Loss ofFMR2gene transcription causesmental retardation (Chakrabarti et al., 1996; Gecz et al., 1996;Gu et al., 1996). Translocations between MLL (a humantrithorax-related gene) and AF4or AF5q31are involved inacute lymphoblastic leukemia (Domer et al., 1993; Nakamuraet al., 1993; Taki et al., 1999, and references therein). Ourresults provide the first insight into the function of this class oftranscription factors during Drosophiladevelopment.

791Development 128, 791-800 (2001)Printed in Great Britain © The Company of Biologists Limited 2001DEV5448

Members of the AF4/FMR2 family of nuclear proteins areinvolved in human diseases such as acute lymphoblasticleukemia and mental retardation. Here we report theidentification and characterization of the Drosophilalilliputian (lilli) gene, which encodes a nuclear proteinrelated to mammalian AF4 and FMR2. Mutations in lillisuppress excessive neuronal differentiation in response to aconstitutively active form of Raf in the eye. In the wild type,Lilli has a partially redundant function in the Ras/MAPKpathway in differentiation but it is essential for normal

growth. Loss of Lilli function causes an autonomousreduction in cell size and partially suppresses the increasedgrowth associated with loss of PTEN function. These resultssuggest that Lilli acts in parallel with the Ras/MAPK andthe PI3K/PKB pathways in the control of cell identity andcellular growth.

Key words: AF4/FMR2, Cell size, Differentiation, Drosophilamelanogaster, PTEN, lilliputian(lilli)

SUMMARY

Lilliputian: an AF4/FMR2-related protein that controls cell identity and cell

growth

Franz Wittwer, Alexandra van der Straten, Krystyna Keleman ‡, Barry J. Dickson ‡ and Ernst Hafen*

Zoologisches Institut, Universität Zürich, Winterthurerstrasse 190, 8057 Zürich, Switzerland‡Present address: Research Institute of Molecular Pathology (IMP), Dr Bohr-Gasse 7, A-1030 Vienna, Austria*Author for correspondence (e-mail: hafen@zool.unizh.ch)

Accepted 13 December 2000; published on WWW 7 February 2001

792

MATERIALS AND METHODS

Genetics13 EMS-induced alleles and three P-element alleles (lilliP1=k05431,lilli P2=k07920, lilli P4=k16124; Torok et al., 1993) of theSu(Raf)2A/lillicomplementation group were isolated (Dickson et al.,1996). lilli was mapped to the cytological interval 23C1-2 byrecombination mapping (Dickson et al., 1996) and by non-complementation with the deficiencies Df(2L)C144and Df(2L)JS17.The chico1 allele and the PTEN117.35 allele have been describedpreviously (Boehni et al., 1999; Oldham et al., 2000b). The ey-FLPsystem has been described by Newsome (Newsome et al., 2000).

Clonal analysisDifferent lilli alleles were recombined onto FRT40 chromosomes andclonal analysis was performed using the FLP/FRT system describedby Xu and Rubin (1993). For the generation of clones in the adult eye,larvae of the genotype y w hs-Flp; lilli FRT40/ P(w+) FRT40weresubjected to a heat shock 24-48 hours after egg deposition (AED) for1 hour at 37°C to induce mitotic recombination. Histological sectionsof the eyes were done as described previously (Basler and Hafen,1988). To generate clones in the imaginal discs, larvae of the genotypey w hs-Flp/y w; lilli FRT40/P(arm-lacZ w+) FRT40were subjected toa heat shock 48-72 hours AED for 25 minutes at 33°C to inducemitotic recombination at a low frequency. Larvae at late third instarstage were dissected. Discs were fixed and permeabilized and stainedwith antibodies. Antibodies were: mouse anti-β-gal (1/2000,Promega), rabbit anti-phosphohistone H3 (1:2000, Upstatebiotechnology) and FITC-secondary antibodies (1/200, JacksonImmunoResearch). Nuclei were stained using the TO-PRO3 DNAstain (T3605, Molecular Probes). Pictures were taken on a Leica TCS-NT confocal laser scanning microscope and clone size was measuredby using the NIH image program. The TUNEL assay was performedas described by Boehni (Boehni et al., 1999).

Cloning of lilliGenomic DNA fragments flanking each of the three P-elementinsertions were cloned by plasmid rescue (Mlodzik et al., 1990) andsequenced. This allowed us to locate the lilli gene on contigAC007765 of the Berkeley Drosophila Genome Project (BDGP)which is consistent with the previous cytological localization of lillito the interval 23C1-2. Overlapping partial lilli cDNAs were used toreconstruct the full length cDNA sequence. This sequence data hasbeen submitted to the GenBank database under accession numberAF293971. cDNA clones were isolated from a λgt10 third instar eye-antennal disc library and by searching the BDGP database. Thelongest cDNAs identified were LD15159, GH23691 and LD18077.LD15159 contains the sequence of exons 1, 2 and the first part of exon3, while GH23691 and LD18077 contain the cDNA sequence of the3′ part of lilli starting within exon 7. LD18077 had a much longer 3′UTR than GH23691. A polyadenylation signal was identified only inthe 3′ UTR of LD18077, but not of GH23691, suggesting that thelatter cDNA corresponds to a truncated message.

The Genescan program (Burge and Karlin, 1997) predicted the 3′end of exon 3 and the 5′ end of exon 7 and in addition exons 4-6 thatwere not present in the isolated cDNAs. The missing part of the lillicDNA sequence was isolated by reverse transcriptase-PCR (RT-PCR).RNA from Drosophilaembryos and larvae was isolated and used tosynthesize first strand cDNA with random or oligo-dT primers. PCRwas performed by using different combinations of primers, thatmatched the sequences of cDNAs LD15159 and GH23691 or of thepredicted exons 4-6. We were able to amplify and sequence partialoverlapping cDNAs spanning exons 2 and 3 (with primers LD55′-TACATATACAACATCGGCCG-3′ and FW6 5′-GTAGTTGATC-TAGTTGTTGCTG-3′), exon 3 to exon 5 (with primers oFW225′-CCTCGACCTTAACCCTTTGG-3′and oFW24 5′-GTGACTCC-GGAAGCTGCTTTG-3′), and exon 5 to exon 7 (with primers oFW23

5′-AGGCAAGCCAAATGCCTCTG 3′ and oFW26 5′-CCGCTCA-CGTTCTCGTTCC-3′).

In situ hybridization to whole-mount embryos was performed asdescribed by Tautz and Pfeifle (1989). Antisense RNA probes weregenerated with the DIG RNA labeling kit (Boehringer Mannheim)using cDNA LD18077 as a template. Sense probes generated inparallel with the same template were used as negative controls.

Molecular characterization of lilli allelesGenomic DNA from flies heterozygous for lilli was isolated. DNAfragments corresponding to exons 3-12 were amplified by PCR andeither sequenced directly or subcloned, using the pCRII-Topo cloningkit (Invitrogen), before sequencing. PCR-induced mutations wereexcluded by sequencing several independent PCR products for eachallele.

Generation of flies expressing lilli transgenesTo generate the UAS-lillirescue construct a 4.4 kb genomic fragmentcontaining the coding part of exon 3 and ending at the NheI site withinexon 7, was fused upstream of the lilli cDNA GH23691 (GH23691fragment starting with NheI site in exon 7 and ending with exon 12;Fig. 4A). This fusion gene was subcloned into the pUAST Drosophilatransformation vector (Brand and Perrimon, 1993) to generate UAS-lilli (construct pFW13). The HA-tag was introduced at the 3′ end ofthe open reading frame in UAS-lilliusing pMZ55 (a gift from M.Zecca and D. Nellen) to generate UAS-lilli-HA (construct pFW9). TheUAS-lilli construct was expressed by using the UAS-GAL4system(Brand and Perrimon, 1993). The driver lines were hs-GAL4(obtainedfrom the Bloomington stock center) and sE-sP-GAL4, whichexpresses GAL4 under control of the sev-enhancer-sev-promoterregulatory elements in R7 precursor cells (P. Maier and E. H.,unpublished results).

Subcellular localization of Lilli-HA was analyzed in third instarlarvae of the genotype y w; UAS-lilli-HA/+, hs-GAL4/+. Salivaryglands were dissected 90 minutes after heat shocking for 1 hour at37°C. They were fixed, permeabilized and stained with the mouseanti-HA antibodies (1/1500, Boehringer Mannheim) and FITCsecondary antibodies (1/60). Pictures were taken using a Leica TCS-NT confocal laser scanning microscope. For generating marked clonesoverexpressing the UAS-lilli rescue construct transgenic flies wereused containing the GMR regulatory sequence (Hay et al., 1997) andthe GAL4 gene separated by a flip-out cassette containing the w+

minigene as a marker (F. Rintelen and E. H., unpublished results). Inheat shocked larvae of the genotype y w hs-Flp/y w; GMR>FRTwhite+ FRT>GAL4/UAS-lilli the w+ cassette is removed in clones thuspermitting expression of UAS-lilliunder the control of the GMR-GAL4 driver.

RESULTS

Lilli has a redundant function in photoreceptor cellfate specificationMutations in lilli were identified in two independent geneticscreens aimed at identifying novel components in theRas/MAPK pathway that regulates R7 photoreceptordevelopment. Dickson et al. (Dickson et al., 1996) recoveredmultiple alleles of a complementation group called Su(Raf)2Ain a screen for mutations that dominantly suppress themultiple-R7 phenotype resulting from constitutive activation ofthe Raf kinase. Neufeld et al. (Neufeld et al., 1998) recoveredalleles of a complementation group called SS2-1in a screen formutations that suppress the rough eye phenotype caused byoverexpression of Sina. Both Su(Raf)2Aand SS2-1weremapped to the same chromosomal location (23C1-2) and

F. Wittwer and others

793Lilli regulates cell growth and cell differentiation

belong to a single complementation group, which we renamedlilliputian (lilli) based on the small size of homozygous lillimutant cells (see below). lilli mutant embryos show terminaland segmentation defects (Dickson et al., 1996) and lillimutations are allelic with l(2)00632,which was identified in ascreen for zygotic lethal mutations with maternal effectphenotypes (Perrimon et al., 1996).

In contrast to its absolute requirement during embryogenesiswe have shown that lilli is dispensable for the specification ofthe endogenous R7 (Dickson et al., 1996). However, itis essential for the formation of ectopic R7 cells inducedby the activation of either Sev, Ras, Raf or MAPK inthe cone cell precursors (Dickson et al., 1996 and Fig.1A,B). Interestingly, while the ectopic R7 cells causedby constitutive activation of Raf were suppressed byremoval of one copy of lilli, the multiple-R7 phenotypecaused by constitutive activation of Sev, Ras, or MAPKwas only suppressed by complete removal of Lillifunction. To test whether Lilli is required only for theformation of ectopic R7 cells or whether it alsoperforms a partially redundant function in theendogenous R7 cell, we generated clones of lilli mutantcells in a background containing the hypomorphicrafHM7allele. The rafHM7 mutation causes a reduction ofraf transcript levels (Melnick et al., 1993) and leads tothe loss of R7, as well as some R1-R6 cells, in a subsetof ommatidia (Dickson et al., 1992, and Fig. 1C).Indeed, the number of both R7 and R1-R6 cells inrafHM7 mutants was further reduced significantly withinthe lilli mutant clones (Fig. 1D). This result suggeststhat Lilli functions not only in the formation of ectopicR7 cells but also plays a partially redundant roledownstream of Raf in the specification of theendogenous photoreceptor cells.

Mutations in lilli affect growth and cell sizeAlthough clones homozygous for lilli alleles in the adulteye were wild-type with respect to photoreceptor celldifferentiation and arrangement, we noted that in clonesof strong alleles (i.e. lilli4U5 and lilli15D1) the size of thephotoreceptor cells was reduced in comparison to theheterozygous cells adjacent to the clone (Fig. 2A).Analysis of mosaic ommatidia showed that thisphenotype is cell autonomous. A reduction of cell sizein mutant tissue is typical for components of thePI3K/PKB and Ras/MAPK pathways (Oldham et al.,2000a; Weinkove and Leevers, 2000).

In addition to reducing cell size, loss of Lilli functionmay also decrease cell number. Clones mutant for chico,which display similar cell size defects as lilli, growmore slowly and cannot compete with faster growingwild-type cells (Boehni et al., 1999). In contrast, twoexperiments indicate that growth of lillimutant tissue isnot impaired during early larval development: first,when we generated sister clones, one homozygous forlilli, the other homozygous for chico, we found that lillimutant cells competed successfully with heterozygoustissue while chicomutant cells were outcompeted (Fig.2B). Second, the size of lilli clones was similar to thatof their wild-type sister clones in the wing imaginal discand the eye imaginal disc anterior to the furrow.

Posterior of the furrow however, lillimutant clones wheresignificantly smaller than their wild-type sister clones (Fig.2C,D). Cells posterior to the morphogenetic furrow undergo afinal round of cell division and are then integrated into theommatidial clusters (Freeman, 1997). The reduced clone sizecould thus be the result of a failure of some lillimutant cellsto undergo this last division or to a slight increase in apoptosisin lilli clones posterior to the morphogenetic furrow. UsingTUNEL staining to detect cells undergoing apoptosis, and anti-

Fig. 1.Lilli is required in the R7 cell and acts downstream or in parallel ofrl/MAPK. (A) Histological sections through the sE-rlDN mutant eyescontaining ectopic R7 cells. sE-rlDN is a transgene encoding an activatedversion of the MAPK expressed under control of the sevenhancer (Brunneret al., 1994). (B) The formation of multiple R7 cells is suppressed in a lilli 3E8

clone (marked by the absence of red pigment, clone border indicated by a redline) in a sE-rlDN mutant background. One ommatidium containingadditional R7 cells and one ommatidium with a single R7 cell is indicated byan arrowhead and arrow, respectively. On average, ommatidia outside thelilli 3E8 clone have 2.7 R7 cells (n=192) while ommatidia within a clone haveonly 1.1 R7 cells (n=135). The difference in the number of R7 cells perommatidium was significant for all 4 eyes analyzed (t-test: P<0.005. Foreach eye examined at least 24 ommatidia inside the clone where comparedwith at least 27 ommatidia outside the clone). (C) A histological sectionthrough an eye of a rafHM7/Y fly is shown. The rafHM7 mutation results in areduction in the number of photoreceptors per ommatidium. (D) In ahomozygous lilli3E8 clone (marked by the absence of red pigment) in arafHM7/Y background the number of photoreceptor cells is further reduced.Outside the lilli3E8 clone, 26% of the ommatidia contain a normalcomplement of photoreceptors. This number is reduced to 9% inside theclone. This reduction is significant (t-test: P<0.01; percentages of completeommatidia in clones of 7 eyes were compared with percentages of completeommatidia in areas outside the clones in the same 7 eyes. The analyzed areascontained between 17 and 48 ommatidia each). Genotypes: (A)y w; 2xsE-rlSem(DN)/+; (B) y w hs-Flp/y w; lilli3E8 FRT40/P(w+) FRT40; 2xsE-rlSem(DN)/+; (C) y w rafHM7/Y; (D) y w rafHM7 hs-Flp/Y; lilli3E8 FRT40/P(w+) FRT40;

794

phosphohistone H3 staining to detect mitotic cells, we wereunable to observe a significant difference between mutant andcontrol clones. The relatively small difference in clone sizemay make it difficult to observe significant differences in celldeath or mitosis with methods that detect cells undergoing cell

death or cell division, respectively only at the time of theexperiment. Therefore, at present we do not know whether thereduction in clone size posterior to the morphogenetic furrowis due to an increase in apoptosis or an inhibition of the cellcycle or a combination of both.

F. Wittwer and others

Fig. 2. Lilli function is also required for cell and organ size but not for cell proliferation. (A) Tangential section through an eye containing alilli 4U5 clone (marked by the absence of red pigment). Within the clone, all photoreceptor cells are reduced in size compared to the neighboringwild-type and heterozygous photoreceptor cells. At the border of the clone, mosaic ommatidia composed of phenotypically wild-typephotoreceptors (marked by the faint yellowish pigment at the base of their rhabdomeres) and small mutant photoreceptor cells are visible,indicating that lillicontrols cell size autonomously. Some additional small lilli 4U5 clones consisting of only few cells are visible dispersed in thearea of wild-type cells. (B) Cell competition assay. lilliP1 homozygous mutant cells are marked by the dark red pigment because they containtwo copies of the w+ marker. Cells heterozygous for lilliP1 and chico1 are orange (w+/w). Cells homozygous for chico1 are marked by theabsence of pigment (w/w) and are not visible in the adult eye because they have been out-competed by the faster growing heterozygous andhomozygous lilli cells. (C) Lilli is not required for cell proliferation but lilliclones posterior to the morphogenetic furrow are reduced in sizewhen compared to their sister clones. Confocal image of lilli 4U5 mitotic clones that were generated by heat shock 48-72 hours after eggdeposition and analyzed in late third instar larvae. Clones of lilli 4U5 cells are marked by the absence of anti-β-gal staining (not green). Theircorresponding sister clones are marked by the more intense green staining (two copies of lacZ). Nuclei are stained by TO-PRO (red). For twoclone/sister clone pairs the border is indicated. The gray bar at the top of the picture indicates the position of the morphogenetic furrow.Anterior is to the left and dorsal is up in A-C. (D) Analysis of growth properties of lilli 4U5 clones in comparison with their wild-type sisterclones in imaginal discs. The areas of clone/sister clone pairs in the pouch region of the wing imaginal disc, and in the eye imaginal disc eitheranterior or posterior to the morphogenetic furrow are shown. Graphs show the sizes (clone areas in pixels) of individual pairs of lilli4U5 clones(black bars) and +/+ sister clones (white bars). In addition, the area of the lilli 4U5 clone relative to its sister clone was calculated for each pair.The means of these values are (as a percentage of sister clone ± standard deviation): for wing pouch 102±19%, for the eye anterior to themorphogenetic furrow: 83±33%, for the eye posterior to the morphogenetic furrow: 56±21%. Similar results were obtained for the lilli15D1

allele (data not shown).The ratio for clones in the eye posterior to the furrow was significantly smaller than the ratio for clones anterior to thefurrow (t-test: P<0.025; 9 ratios of clones anterior of the morphogenetic furrow were compared with 9 ratios of clones posterior to the furrow).(E) Selective removal of Lilli function from the eye imaginal discs using the ey-Flpsystem (Newsome et al., 2000) results in heads that arereduced in size (left fly) when compared to wild-type flies (right). Genotypes: (A) y w hs-Flp/y w; lilli4U5 FRT40/P(w+) FRT40;(B) y w ey-Flp/Y; lilli P1 FRT40/chico1FRT40; (C) y w hs-Flp/y w; lilli4U5FRT40 /P(arm-lacZ w+) FRT40. (E, left side) y w ey-Flp/y w; lilli4U5

FRT40/P(w+) l(2)2L-3.1 FRT40; (E, right side) y w ey-Flp/y w; lilli4U5 FRT40/CyO.

795Lilli regulates cell growth and cell differentiation

Consistent with the reduced clone size in the eye discposterior to the morphogenetic furrow we found that theselective removal of Lilli function in eye and head precursorcells by the ey-Flpsystem resulted in flies with reduced eyeand head size (Fig. 2E). The reduction in eye size was causedby a reduction in both the number and size of ommatidia (24%and 18%, respectively, for lilli4U5; Fig. 3I,K). The degree ofhead size reduction allowed us to define a class of strong alleles(e.g. lilli 4U5, lilli 15D1) and a class of weak alleles (e.g. lilli 3E8,lilli P1, lilli P4). The small head phenotype of lilli mutant tissueis similar to that observed with components of the PI3K/PKBpathway (Boehni et al., 1999; Montagne et al., 1999).

Lilli partially suppresses the overgrowth phenotypecaused by the loss of PTEN functionSince loss of Lilli function affects cell size and head size, wetested for genetic interactions between lilliand componentsof the PI3K/PKB pathway. PTEN acts as a negative regulator

in the PI3K/PKB pathway by dephosphorylating thesecond messenger phosphatidylinositol 3,4,5-trisphosphate(PtdIns(3,4,5)P3; PIP3, for reviews see Rameh and Cantley,1999; Di Cristofano and Pandolfi, 2000). Tissues mutant forPTEN show hyperplastic and hypertrophic growth: PTENmutant cells are larger and proliferate at a higher rate than wild-type cells (Goberdhan et al., 1999; Huang et al., 1999; Gao etal., 2000).

Removal of PTENfunction from the eye imaginal disc tissueusing the ey-Flpsystem (Newsome et al., 2000) resulted in anincrease in eye and head size (Fig. 3A; Oldham et al., 2000b).The increase in eye size was due to an increase in cell numberand cell size as indicated by the increase in number and sizeof ommatidia (Fig. 3I,K). To test whether the PTEN large-headphenotype was modified by the removal of Lilli function, wegenerated PTEN, lilli double mutant eyes. Indeed, eyes doublemutant for PTENand lilli were considerably smaller thanPTENmutant eyes due to a reduction in cell number and cell

Fig. 3. lilli cooperates withPI3K/PKB signaling.(A-H) Scanning electronmicrographs of eyes in whichthe function of PTEN or Lilli orboth has been removed frommost of the cells by the ey-Flpsystem. Images in the same row(i.e. A-D or E-H) were taken atthe same magnification. Scalebars 100 µm (A-D) and 20 µm(E-H). For A-D and L, anterioris to the left and dorsal is up.(A,E) Removal of PTENleads toincreased size of the headcapsule and the eyes due tohyperplastic growth. In addition,the ommatidia are enlarged andadditional bristles are formed.(B,F) Heads double mutant forboth PTENand lilli are smallerthan those of PTENsinglemutants or wild type.Interommatidial bristles are lostin lilli, PTEN double mutantheads. (C,G) lillimutant headsare smaller than wild-type headsand have ommatidia of reducedsize. (D,H) Control flies inwhich the ey-Flpsystem is usedin combination with an FRT40chromosome alone (i.e. withouta mutation). The size of the headis nearly wild type. (I) Numberof ommatidia in eyes of thesame genotype as shown in A-H.Scanning electron micrographs of female flies of the indicated genotypes were taken and used to count ommatidia. The standard deviation isindicated by the error bar. The differences between the mean values is significant (t-test: P<0.025, n=4 eyes for all genotypes). (K) Size ofommatidia in eyes of the same genotype as shown in A-H. Scanning electron micrographs of female flies of the genotypes indicated were usedto determine the average area covered by an ommatidium, based on measurements of 7 neighboring ommatidia. The mean values aresignificantly different except for the difference between ‘PTEN+lilli’ and ‘control’ (t-test: P<0.025, n=4 eyes for each genotype). (L) hs-Flp-induced clone in the eye lacking both Lilli and PTEN function. The clone is marked by the absence of red (w+) pigment. One mutantphotoreceptor (no red pigment in rhabdomere) and a heterozygous photoreceptor (pigment present in rhabdomere) are marked by a black or awhite arrow, respectively. Genotypes: (A,E: ‘PTEN’): y w ey-Flp/y w; PTEN117.35FRT40/P(w+) l(2)2L-3.1 FRT40; (B,F: ‘PTEN+lilli’)y w ey-Flp/y w; lilli 4U5 PTEN117.35FRT40/ P(w+) l(2)2L-3.1 FRT40; (B,F: ‘lilli’)y w ey-Flp; lilli4U5 FRT40/P(w+) l(2)2L-3.1 FRT40; (D,H: ‘control’)y w ey-Flp/y w; FRT40/P(w+) l(2)2L-3.1 FRT40; (L)y w, hs-Flp/y w; lilli4U5 PTEN117.35FRT40/P(w+) FRT40.

796 F. Wittwer and others

797Lilli regulates cell growth and cell differentiation

size (Fig. 3B,C,I,K). Loss of Lilli function, however, did notcompletely suppress the PTEN phenotype. Although theseresults suggest that Lilli and PTEN cooperate in the control ofcell and organ growth, the absence of a clear-cut epistasisbetween the two mutants indicates that Lilli does not actdownstream of PTEN in a simple linear pathway but it ratheracts in a parallel pathway required for growth.

Molecular characterization of the lilli gene To identify the lilli gene, DNA flanking the lilli P-elementinsertions (see Materials and Methods) was cloned and used to

initiate a genomic walk of the 23C1-2 region. A search forcDNA clones in this region revealed several ESTs thatconsisted of exons located near the insertion site of the Pelements. Sequencing of all available cDNAs defined atranscription unit spanning a region of 67 kb and consisting of12 exons (Fig. 4A). Since one P element is located in the firstintron and two other P elements are inserted in the second 54kb intron, this transcription unit was a good candidate for thelilli gene. Two further findings demonstrate that this candidategene is indeed responsible for Lilli function. First, missense ornonsense mutations were identified in the predicted lillitranscription unit in 8 out of 11 EMS-induced lilli allelesexamined (Fig. 4B, and see below). Second, low levelubiquitous expression of a minigene rescue construct (UAS-lilliin Fig. 4A) using a heat-shock-GAL4driver in the absence ofheat shock rescued lilli homozygous mutant flies to adulthood.The rescued flies appeared normal – except that the size of thewing and the head was slightly reduced in comparison to wild-type flies (data not shown). Furthermore, the dominantsuppression of the sE-RaftorY9-induced rough eye phenotype byremoval of one copy of lilli was reverted by expressing theUAS-lilli minigene under the control of the sevE-sevP-GAL4driver (data not shown). These results establish the identity ofthe lilli gene.

lilli encodes a novel nuclear protein related to themammalian AF4/FMR2 family of transcription factors The predicted lilliopen reading frame encodes a protein of1673 amino acids which has a C-terminal domain (amino acids1418-1668, C-terminal homology domain, CHD, defined byNilson et al., 1997) homologous to members of the AF4/FMR2family of mammalian proteins (Fig. 4B,F). In this region, Lilliis 31-37% identical to the corresponding sequences of humanAF4, FMR2, AF5q13 and LAF4. Interestingly, the position ofthe intron/exon boundaries within the region encoding theCHD domain are conserved between lilli and the AF4familymembers suggesting that those exons have a commonevolutionary origin.

Searching the Drosophilagenome database did not revealany sequences in addition to lilli that share homologies withAF4/FMR2 family members. Thus lilliappears to be the onlymember of this family in Drosophila. FMR2, AF4 and LAF4are nuclear proteins and these have properties of transcriptionfactors. They possess a putative transactivation domain andhave the ability to bind DNA in vitro (Ma and Staudt, 1996;Gu et al., 1996; Nilson et al., 1997; Gecz et al., 1997; Milleret al., 2000). Like other family members, Lilli is rich in prolineand serine residues (9.0% and 12.7% of all amino acids,respectively). For human AF4 and LAF4, the serine-rich regionhas been shown to be situated in a part of the protein that hastransactivation potential (Ma and Staudt, 1996). Serines areconcentrated in Lilli at the same relative position as inAF4/FMR2 family members (Ser domain, amino acid 860-965of Lilli; Fig. 4B). However, domains within the AF4/FMR2family members that are located N-terminal of the serine-richregion (i.e. NHD domain and ALF domain, defined by Nilsonet al., 1997) are not present in Lilli.

In addition to the serine-rich region, two other features ofLilli indicate a possible function in activating transcription:first, four glutamine-rich stretches in the N-terminal region ofLilli (Q1-Q4: amino acids 3-14, 32-48, 172-180 and 500-521)

Fig. 4. Molecular organization of the lillilocus and characterizationof mutant alleles. (A) Genomic structure of the lillilocus at 23C1-2.The position of the centromere is to the right. Transcription units ofthe β-subunit of the geranylgeranyl transferase (β-ggt), the NTPase,and the RNA-binding protein 9 (rbp9) are indicated. For the lillitranscript individual exons are shown. White boxes represent non-coding sequences and black boxes the lilli coding region. Theapproximate locations of P-element insertions in lilli(P1:lilli P1=k05431, P2: lilliP2=k07920, P4: lilliP4=k16124) are indicatedby inverted triangles. Below, the lillirescue construct is shown (seetext for details). (B) Schematic representation of the Lilli proteinindicating the predicted domains and sequence similarities to otherproteins (for amino acid positions see text). Lilli shares the serine-rich (Ser) and the C-terminal homology domain (CHD) with proteinsof the AF4/FMR2 family but does not contain the NHD and ALFdomain. Q1-Q4: glutamine-rich stretches. Homology with POU:region homologous to part of the transactivation domain of POUclass III transcription factor from chick. AT-hook: AT-hook motif.NLS1 and NLS2: putative bipartite nuclear localization signalspredicted by the pfscan program (Swiss Institute for ExperimentalCancer Research, Epalinges, Switzerland). Locations of the mutantlesions are indicated above the protein structure. Molecularly, themutations consist of the following changes (in brackets: first aminoacid affected by the mutation):lilli 3E8: CAG (Q1192) → TAG (stop);lilli 4P5: CGA (R803) → TGA (stop); lilli4U5: 20 bp deletion startingwithin the codon for T758 which results in the replacement of theoriginal amino acid sequence with a shorter sequence with shiftedreading frame; lilli5X5: CAG (Q1100) → TAG (stop); lilli16F1: TAC(Y1459) → TTC (F1459) and GAG (E1461) → GTG (V1461); lilli 16Q1:101 bp deletion, starting within the codon for L1534which results inthe replacement of the original amino acid sequence with a shortersequence with shifted reading frame; lilli17G1: CGA (R1632) → TGA(stop); lilli22E1: CAA (Q500) → TAA (stop). (C) A short amino acidstretch in Lilli is homologous to part of the POU class IIItranscription factor from chick. The black boxes indicate amino acididentity while the gray boxes indicate amino acid similarity. Dotsindicate gaps in the sequence that were introduced to optimize thealignment. (D) Alignment of the AT-hook of Lilli with that of humanHMG-I(Y) (Homo sapiens, amino acid 7-34; GenBank accessionnumber L46353) and rat HMG-I(Y) (Rattus norvegicus, amino acid7-34; GenBank accession number U89695). Gaps and identical orconserved positions are indicated as in C. (E) The HA-tagged versionof Lilli (Lilli-HA) is localized in the nucleus when overexpressed insalivary glands. A confocal image of salivary glands stained withanti-HA is shown. Lilli-HA expression was induced by heat shock inlarvae of the following genotype: y w; UAS-lilli-HA/+, hs-GAL4/+.(F) Clustal W alignment of the CHD of Lilli and human AF5q31,FMR2, AF4 and LAF4. Black boxes indicates amino acid identity,gray boxes amino acid similarity in at least four of the five proteins.Amino acids bordering splice sites are underlined for Lilli, AF4 andFMR2. Note that most splice sites are conserved (arrows). Theconserved Tyr1459that is replaced by alanine in allele lilli16F1 ismarked by an asterix. Gaps are indicated as in C.

798

and, second, a domain (amino acid 245-300) with homologyto part of a putative transactivation domain from the POU classIII transcription factor (Fig. 4B,C; Levavasseur et al., 1998).Lilli has two putative bipartite nuclear localization signals(amino acid 889-906 and amino acid 1286-1303, respectively,Fig. 4B). In addition, Lilli has a sequence (amino acid 836-859; Fig. 4B,D) which matches the consensus sequence for theHMG-I(Y) DNA binding motif, termed the AT-hook (Reevesand Nissen, 1990). This DNA binding domain is not found inother members of the AF4/FMR2 family. The AT-hook motifbinds to the minor groove of the DNA double helix at A/T-richsequences (Reeves and Nissen, 1990).

Analysis of the EMS-induced mutations provided furtherinsight into putative functional domains of the Lilli protein(Fig. 4B and figure legend). lilli 4U5 contains a short deletionresulting in a frameshift and is probably a functional null allele,since the encoded protein lacks most of the predicted domainsincluding the AT-hook, the serine-rich domain, and the nuclearlocalization signals. Thelilli 3E8 mutant protein is predictedto have a C-terminal truncation but since it still retains theputative DNA-binding and Ser domain it is likely to haveresidual activity. Interestingly, although lilli3E8 was able tosuppress theRaftorY9 induced rough eye to the same extent asthe other EMS alleles or deficiencies for the locus, it showeda less severe reduction in head size in ey-FLP mosaics thanlilli 4U5 and lilli 15D1 (data not shown). This suggests that thevery C-terminal domain of Lilli is essential for its function asa transducer of the Raf signal but is less critical with regard toits growth promoting role. In the protein encoded bylilli 16F1,Tyr1459 is replaced by alanine. Intriguingly, Tyr1459 isconserved throughout the AF4/FMR2 family (Fig. 4F; thephenotype of lilli16F1 in ey-Flpmosaics was weak and similarto the Lilli3E8 phenotype).

Lilli protein is localized in the nucleuslilli expression in the embryo was analyzed by in situhybridization. We found high levels of lilli mRNA in theunfertilized egg, which is consistent with the maternalrequirement for Lilli function (Dickson et al., 1996; Perrimonet al., 1996). In the early embryo and until gastrulation, lilli isexpressed uniformly. During later stages of embryogenesis, wefound a slightly elevated expression in the nervous system(data not shown). In order to determine the subcellularlocalization of Lilli, we generated an HA-tagged minigeneconstruct (UAS-lilli-HA, see Material and Methods). UAS-lilli-HA rescued lilli lethality suggesting that this construct encodesa functional Lilli protein (data not shown). Staining salivaryglands of larvae in which expression of the HA-tagged Lilli(Lilli-HA) was induced by heat shock, revealed that Lilli-HAis mainly localized in the nucleus (Fig. 4E). Nuclearlocalization of Lilli-HA was also observed in the eye imaginaldisc (data not shown). Therefore, we believe that, like itsmammalian homologs, Lilli also normally functions in thenucleus (Nilson et al., 1997; Miller et al., 2000).

DISCUSSION

FMR2 and AF4 are the founding members of a family ofputative transcriptional activators in humans (Domer et al.,1993; Nakamura et al., 1993; Chakrabarti et al., 1996; Gecz et

al., 1996; Gu et al., 1996). Little is known about their normalcellular function, although mutations in the correspondinggenes are associated with human diseases such as mentalretardation and acute lymphoblastic leukemia. Here wedescribe the characterization of the first invertebrate memberof this family. Lilli cooperates with the Ras/MAPK and thePI3K/AKT pathway in the control of cell growth and celldifferentiation.

Lilli has a partially redundant function downstreamof Ras/MAPK signaling in cell fate specificationLoss-of-function mutations in lilli where identified asdominant suppressors of the specification of supernumerary R7photoreceptor cells in response to constitutive activation of Rafin the developing eye (Dickson et al., 1996). However, withoutconstitutive activity of the Ras/MAPK pathway, the normalnumber of photoreceptor cells is specified in each ommatidiumin the complete absence of Lilli function. Given the fact thatlilli encodes a putative transcription factor, a possibleexplanation for why lilli mutants specifically suppress thedifferentiation of extra R7 photoreceptor cells is that lillimaybe required for the expression of the sE-RaftorY99 transgene.Indeed, we and others have noted that lilli mutations affecttranscript levels of transgenes containing a hsp70 basalpromoter (data not shown and Tang et al., 2001). Although theeffect of lilli on hsp70transcription may explain some of thegenetic interactions (Greaves et al., 1999; Rebay et al., 2000),it is not sufficient to explain the specific interaction with theRaf/MAPK pathway for the following reasons. (1) In theoriginal screen, we used two different RaftorY9 transgenes. Onecontained the sev enhancer fused to the hsp70 promoter (sE),the other the sevenhancer fused to the sevpromoter (sEsP;Dickson et al., 1996). lilli mutations suppressed both. (2) lillimutations failed to suppress the rough eye phenotype causedby the ectopic expression of rough under the control of the sevenhancer and heat shock promoter. (3) Heterozygosity for lillidoes not suppress the multiple R7 phenotype of SevS11, atransgene encoding a truncated version of the Sev receptor alsoexpressed under the control of the sev enhancer and the hsp70promoter (sE; Basler et al., 1991). The formation of multipleR7 cells caused by sE-SevS11 or sEsP-RasV12 (Fortini et al.,1992) is only suppressed in homozygous lilli clones (Dicksonet al 1996). If lilli controlled expression of the sevenhancer,we would expect a dominant suppression of all sevenhancerdriven transgenes. (4) Finally, as we show here, completeloss of Lilli function in clones decreased the number ofphotoreceptor cells recruited in the background of ahypomorphic mutation of raf, rafHM7. Therefore, we concludethat Lilli has a specific function in regulating the efficiencyof signal transduction downstream of Raf. As a putativetranscription factor, Lilli may regulate the expression levels ofone or multiple components of the Ras/MAPK signalingpathway that become rate-limiting when Ras/MAPK signalingis too high, in cells where it is normally low (as in the case ofectopic activation of Raf in the eye), or when signaling isreduced.

Lilli function in growth controlGrowth is controlled by many different signals and duringdevelopment it is tightly linked to pattern formation (Day andLawrence, 2000). In Drosophila, the insulin receptor controls

F. Wittwer and others

799Lilli regulates cell growth and cell differentiation

growth via the PI3K/PKB pathway without affecting patternformation (reviewed by Oldham et al., 2000b; Weinkove andLeevers, 2000). The EGF receptor also controls growth mostlikely via the Ras/MAPK pathway (Diaz-Benjumea and Hafen,1994; Prober and Edgar, 2000). Little is known about thenuclear factors that integrate the signals from these twopathways. Could the nuclear factor Lilli be a point ofconvergence of these two pathways for growth control?Although the cell size phenotype observed in lilli mutants issimilar to that of mutants in either of the two pathways,mutations in lillialso display some unexpected phenotypes thatare difficult to reconcile with Lilli functioning exclusively as atarget of the two pathways. (1) lillimutant cells do not have agrowth disadvantage during early imaginal disc development.(2) The partial suppression of the overgrowth phenotype ofPTEN mutants by lilli mutants suggests a complex epistaticrelationship between lilliand the PI3K/PKB pathway. (3) BothLilli overexpression and lilliloss-of-function mutations reducecell size (Fig. 2A, and data not shown). This is in contrast tocomponents of the PI3K/PKB and Ras/MAPK signalingpathways. Overexpression of PI3K or PKB increases cell size,whereas loss-of-function mutations decrease cell size (Leeverset al., 1996; Verdu et al., 1999; Weinkove et al., 1999).Similarly, loss- and gain-of-function mutations of Ras haveopposite effects on cell size (Prober and Edgar, 2000). Lillimay activate the transcription of genes involved in cell growthand differentiation.

Lilli and AF4/FMR2 phenotypesHow does the function of Lilli in Drosophilacorrelate with thephenotypes observed in disease conditions associated withmutations in the human homologsAF4, AF5q31and FMR2? Inthe case of acute lymphoblastic leukemia two related genes,AF4 and AF5q31, are involved in translocations to the MLLgene (also known as ALL-1, htrx), which encodes a humanhomolog of the Drosophilaprotein Trithorax (Domer et al.,1993; Nakamura et al., 1993; Taki et al., 1999, and referencestherein). We show that Lilli function is essential for increasedacitivity in the Ras/MAPK pathway. Activating mutations inRas are frequently found in lymphomas but normally absent inleukemias associated with MLLrearrangements (Mahgoub etal., 1998, and references therein). It is possible that expressionof the MLL/AF4 fusion gene circumvents the need for Rasactivation in transformation. Recently, the targeted inactivationof the Af4gene in mice has been reported (Isnard et al., 2000).Af4 mutant mice display an altered development of lymphoidcells and, on a mixed 129/Balb/c background, a subset show asignificant reduction in body weight at birth. The reduction inweight may suggest that AF4 and Lilli control cellular growthin mice and Drosophila, respectively. Thus, further analysis ofLilli function may provide insights into the function of this classof nuclear proteins both in vertebrate development and disease.

We are grateful to A. H. Tang and G. M. Rubin (University ofCalifornia, Berkeley) for exchanging information prior to publication.We thank C. Hugentobler and P. Faller for technical assistance, T.Gutjahr for scanning electron microscopy and computer assistance, T.Halfar for the TUNEL staining, P. Zipperlen, J. Berger and M. Spoerrifor DNA sequencing, D. Nellen for antibodies, F. Rintelen for the flip-out overexpression system, S. Oldham for the PTENmutant flies andthe Bloomington Stock Center and P. Maier for fly stocks. We alsothank R. Boehni, D. Bopp, P. Gallant, T. Gutjahr, L. K. MacDougall,

K. Nairz, S. Oldham, and H. Stocker for critical reading of themanuscript and for helpful discussions. F. W. was supported by a grantfrom the EU project ‘signaling in development and disease’.

REFERENCES

Basler, K., Christen, B. and Hafen, E.(1991). Ligand-independent activationof the Sevenless receptor tyrosine kinase changes the fate of cells in thedeveloping Drosophilaeye. Cell 64, 1069-1081.

Basler, K. and Hafen, E. (1988). Control of photoreceptor cell fate by theSevenless protein requires a functional tyrosine kinase domain. Cell 54, 299-311.

Bergmann, A., Agapite, J., McCall, K. and Steller, H. (1998). TheDrosophilagene hid is a direct molecular target of Ras-dependent survivalsignaling. Cell 95, 331-341.

Boguski, M. S. and McCormick, F. (1993). Proteins regulating Ras and itsrelatives. Nature, 366, 643-654.

Boehni, R., Riesgo-Escovar, J., Oldham, S., Brogiolo, W., Stocker, H.,Andruss, B. F., Beckingham, K. and Hafen, E. (1999). Autonomouscontrol of cell and organ size by CHICO, a Drosophila homolog ofvertebrate IRS1-4. Cell 97, 865-875.

Brand, A. H. and Perrimon, N. (1993). Targeted gene expression as a meansof altering cell fates and generating dominant phenotypes. Development 118,401-415.

Brunner, D., Oellers, N., Szabad, J., Biggs, W. H., III, Zipursky, S. L. andHafen, E. (1994). A gain-of-function mutation in DrosophilaMAP kinaseactivates multiple receptor tyrosine kinase signaling pathways. Cell 76, 875-888.

Burge, C. and Karlin, S. (1997). Prediction of complete gene structures inhuman genomic DNA. J. Mol. Biol. 268, 78-94.

Chakrabarti, L., Knight, S. J., Flannery, A. V. and Davies, K. E. (1996).A candidate gene for mild mental handicap at the FRAXEfragile site. Hum.Mol. Genet. 5, 275-282.

Day, S. J. and Lawrence, P. A. (2000). Measuring dimensions, the regulationof size and shape. Development 127, 2977-2987.

Di Cristofano, A. and Pandolfi, P. P. (2000). The multiple roles of PTEN intumor suppression. Cell 100, 387-390.

Diaz-Benjumea, F. J. and Hafen, E. (1994). The Sevenless signaling cassettemediates DrosophilaEGF receptor function during epidermal development.Development 120, 569-578.

Dickson, B., Sprenger, F., Morrison, D. and Hafen, E.(1992). Raf functionsdownstream of Ras1 in the Sevenless signal transduction pathway. Nature360, 600-603.

Dickson, B. J., van der Straten, A., Dominguez, M. and Hafen, E.(1996).Mutations modulating Raf signaling in Drosophila eye development.Genetics 142, 163-171.

Domer, P. H., Fakharzadeh, S. S., Chen, C. S., Jockel, J., Johansen, L.,Silverman, G. A., Kersey, J. H. and Korsmeyer, S. J.(1993). Acutemixed-lineage leukemia t(4;11)(q21;q23) generates an MLL-AF4fusionproduct. Proc. Natl. Acad. Sci. USA 90, 7884-7888.

Dominguez, M., Wasserman, J. D. and Freeman, M.(1998). Multiplefunctions of the EGF receptor in Drosophilaeye development. Curr. Biol.8, 1039-1048.

Fortini, M. E., Simon, M. A. and Rubin, G. M. (1992). Signalling by thesevenless protein tyrosine kinase is mimicked by Ras1 activation. Nature355, 559-567.

Frame, S. and Balmain, A. (2000). Integration of positive and negativegrowth signals during ras pathway activation in vivo. Curr. Opin. Genet.Dev. 10, 106-113.

Freeman, M. (1997). Cell determination strategies in the Drosophila eye.Development 124, 261-270.

Gao, X., Neufeld, T. P. and Pan, D.(2000). DrosophilaPTEN regulates cellgrowth and proliferation through PI3K-dependent and -independentpathways. Dev. Biol.221, 404-418.

Gecz, J., Gedeon, A. K., Sutherland, G. R. and Mulley, J. C. (1996).Identification of the gene FMR2, associated with FRAXEmental retardation.Nat. Genet. 13, 105-108.

Gecz, J., Bielby, S., Sutherland, G. M. and Mulley, J. C.(1997). Genestructure and subcellular localization of FMR2, a member of a new familyof putative transcription activators. Genomics 44, 201-213.

Gille, H. and Downward, J. (1999). Multiple Ras effector pathwayscontribute to G1 cell cycle progression. J. Biol. Chem. 274, 22033-22040.

800

Goberdhan, D. C., Paricio, N., Goodman, E. C., Mlodzik, M. and Wilson,C. (1999). Drosophilatumor suppressor PTEN controls cell size and numberby antagonizing the Chico/PI3-kinase signaling pathway. Genes Dev. 13,3244-3258.

Greaves, S., Sanson, B., White, P. and Vincent, J. P. (1999). A screen foridentifying genes interacting with armadillo, the Drosophila homolog ofbeta-catenin. Genetics 153, 1753-1766.

Gu, Y., Shen, Y., Gibbs, R. A. and Nelson, D. L.(1996). Identification ofFMR2, a novel gene associated with the FRAXE CCG repeat and CpGisland. Nat. Genet., 13, 109-113.

Hay, B. A., Maile, R. and Rubin, G. M. (1997). P element insertion-dependent gene activation in the Drosophila eye. Proc. Natl. Acad. Sci. USA94, 5195-200.

Huang, H., Potter, C. J., Tao, W., Li, D. M., Brogiolo, W., Hafen, E., Sun,H. and Xu, T. (1999). PTEN affects cell size, cell proliferation andapoptosis during Drosophila eye development. Development 126, 5365-5372.

Isnard, P., Core, N., Naquet, P. and Djabali, M. (2000). Altered lymphoiddevelopment in mice deficient for the mAF4 proto-oncogene. Blood, 96,705-710.

Jones, S. M., Klinghoffer, R., Prestwich, G. D., Toker, A. and Kazlauskas,A. (1999). PDGF induces an early and a late wave of PI 3-kinase activity,and only the late wave is required for progression through G1. Curr. Biol.9, 512-521.

Kurada, P. and White, K. (1998). Ras promotes cell survival in Drosophilaby downregulating hid expression. Cell 95, 319-329.

Leevers, S. J., Weinkove, D., MacDougall, L. K., Hafen, E. and Waterfield,M. D. (1996). The Drosophilaphosphoinositide 3-kinase Dp110 promotescell growth. EMBO J. 15, 6584-6594.

Levavasseur, F., Mandemakers, W., Visser, P., Broos, L., Grosveld, F.,Zivkovic, D. and Meijer, D. (1998). Comparison of sequence and functionof the Oct-6genes in zebrafish, chicken and mouse. Mech. Dev. 74, 89-98.

Ma, C. and Staudt, L. M. (1996). LAF-4encodes a lymphoid nuclear proteinwith transactivation potential that is homologous to AF-4, the gene fused toMLL in t(4;11) leukemias. Blood 87, 734-745.

Mahgoub, N., Parker, R. I., Hosler, M. R., Close, P., Winick, N. J.,Masterson, M., Shannon, K. M. and Felix, C. A. (1998). Rasmutationsin pediatric leukemias with MLLgene rearrangements. Genes ChromosomesCancer 21, 270-275.

Melnick, M. B., Perkins, L. A., Lee, M., Ambrosio, L. and Perrimon, N.(1993). Developmental and molecular characterization of mutations in theDrosophila-rafserine/threonine protein kinase. Development 118, 127-138.

Miller, W. J., Skinner, J. A., Foss, G. S. and Davies, K. E. (2000).Localization of the fragile X mental retardation 2 (FMR2) protein inmammalian brain. Eur. J. Neurosci. 12, 381-384.

Mlodzik, M., Baker, N. E. and Rubin, G. M. (1990). Isolation and expressionof scabrous, a gene regulating neurogenesis in Drosophila. Genes Dev. 4,1848-1861.

Montagne, J., Stewart, M. J., Stocker, H., Hafen, E., Kozma, S. C. andThomas, G.(1999). DrosophilaS6 kinase, a regulator of cell size. Science285, 2126-2129.

Nakamura, T., Alder, H., Gu, Y., Prasad, R., Canaani, O., Kamada, N.,Gale, R. P., Lange, B., Crist, W. M., Nowell, P. C. and et al. (1993). Geneson chromosomes 4, 9, and 19 involved in 11q23 abnormalities in acuteleukemia share sequence homology and/or common motifs. Proc. Natl.Acad. Sci. USA 90, 4631-4635.

Neufeld, T. P., Tang, A. H. and Rubin, G. M. (1998). A genetic screen toidentify components of the sina signaling pathway in Drosophila eyedevelopment. Genetics 148, 277-286.

Newsome, T. P., Asling, B. and Dickson, B. J. (2000). Analysis of Drosophilaphotoreceptor axon guidance in eye-specific mosaics. Development 127,851-860.

Nilson, I., Reichel, M., Ennas, M. G., Greim, R., Knorr, C., Siegler, G.,Greil, J., Fey, G. H. and Marschalek, R. (1997). Exon/intron structure ofthe human AF-4 gene, a member of the AF-4/LAF- 4/FMR-2 gene familycoding for a nuclear protein with structural alterations in acute leukaemia.Br. J. Haematol. 98, 157-169.

Oldham, S., Boehni, R., Stocker, H., Broggiolo, W. and Hafen, E.(2001a).Genetic control of size in Drosophila. Phil. Trans. R. Soc. Lond.355, 945-952.

Oldham, S., Montagne, J., Radimerski, T., Thomas, G. and Hafen, E.(2001b). Genetic and biochemical characterization of dTor, the Drosophilahomolog of the target of rapamycin. Genes Dev.14, (in press).

Perrimon, N., Lanjuin, A., Arnold, C. and Noll, E. (1996). Zygotic lethalmutations with maternal effect phenotypes in Drosophila melanogaster.II.Loci on the second and third chromosomes identified by P-element-inducedmutations. Genetics 144, 1681-1692.

Prober, D. A. and Edgar, B. A. (2000). Ras1 promotes cellular growth in theDrosophilawing. Cell 100, 435-446.

Rameh, L. E. and Cantley, L. C. (1999). The role of phosphoinositide 3-kinase lipid products in cell function. J. Biol. Chem. 274, 8347-8350.

Rebay, I., Chen, F., Hsiao, F., Kolodziej, P. A., Kuang, B. H., Laverty, T.,Suh, C., Voas, M., Williams, A. and Rubin, G. M. (2000). A geneticscreen for novel components of the Ras/Mitogen-Activated Protein Kinasesignaling pathway that interact with the yan gene of Drosophila identifiessplit ends, a new RNA recognition motif- containing protein. Genetics 154,695-712.

Reeves, R. and Nissen, M. S. (1990). The A.T-DNA-binding domain ofmammalian high mobility group I chromosomal proteins. A novel peptidemotif for recognizing DNA structure. J. Biol. Chem. 265, 8573-8582.

Rommel, C. and Hafen, E.(1998). Ras – a versatile cellular switch. Curr.Opin. Genet. Dev. 8, 412-418.

Taki, T., Kano, H., Taniwaki, M., Sako, M., Yanagisawa, M. and Hayashi,Y. (1999). AF5q31, a newly identified AF4-related gene, is fused to MLLininfant acute lymphoblastic leukemia with ins(5;11)(q31;q13q23). Proc.Natl. Acad. Sci. USA 96, 14535-14540.

Tang, A. H., Neufeld, T. P., Rubin, G. M. and Müller, H.-A. J. (2001).Transcriptional regulation of cytoskeletal functions and segmentation by anovel maternal pair-rule gene, lilliputian. Development 128, 801-813.

Tautz, D. and Pfeifle, C. (1989). A non-radioactive in situ hybridizationmethod for the localization of specific RNAs in Drosophilaembryos revealstranslational control of the segmentation gene hunchback. Chromosoma 98,81-85.

Torok, T., Tick, G., Alvarado, M. and Kiss, I. (1993). P-lacWinsertionalmutagenesis on the second chromosome of Drosophila melanogaster,isolation of lethals with different overgrowth phenotypes. Genetics 135, 71-80.

Verdu, J., Buratovich, M. A., Wilder, E. L. and Birnbaum, M. J. (1999).Cell-autonomous regulation of cell and organ growth in DrosophilabyAkt/PKB. Nat. Cell Biol. 1, 500-506.

Weinkove, D. and Leevers, S. J.(2000). The genetic control of organ growth,insights from Drosophila. Curr. Opin. Genet. Dev. 10, 75-80.

Weinkove, D., Neufeld, T. P., Twardzik, T., Waterfield, M. D. and Leevers,S. J. (1999). Regulation of imaginal disc cell size, cell number and organsize by Drosophilaclass I(A) phosphoinositide 3-kinase and its adaptor.Curr. Biol. 9, 1019-1029.

Xu, T. and Rubin, G. M. (1993). Analysis of genetic mosaics in developingand adult Drosophilatissues. Development, 117, 1223-1237.

F. Wittwer and others