A Gain-of-Function Screen for Genes That Affect the Development of the Drosophila Adult External Sensory Organ

Corresponding author: Yuh Nung Jan, Howard Hughes Medical Institute, Departments of Physiology and Biochemistry, Third and Parnassus Ave., University of California, San Francisco, CA 94143-0725., ynjan{at}itsa.ucsf.edu (E-mail)

Communicating editor: T. SCHÜPBACH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The Drosophila adult external sensory organ, comprising a neuron and its support cells, is derived from a single precursor cell via several asymmetric cell divisions. To identify molecules involved in sensory organ development, we conducted a tissue-specific gain-of-function screen. We screened 2293 independent P-element lines established by P. Rørth and identified 105 lines, carrying insertions at 78 distinct loci, that produced misexpression phenotypes with changes in number, fate, or morphology of cells of the adult external sensory organ. On the basis of the gain-of-function phenotypes of both internal and external support cells, we subdivided the candidate lines into three classes. The first class (52 lines, 40 loci) exhibits partial or complete loss of adult external sensory organs. The second class (38 lines, 28 loci) is associated with increased numbers of entire adult external sensory organs or subsets of sensory organ cells. The third class (15 lines, 10 loci) results in potential cell fate transformations. Genetic and molecular characterization of these candidate lines reveals that some loci identified in this screen correspond to genes known to function in the formation of the peripheral nervous system, such as big brain, extra macrochaetae, and numb. Also emerging from the screen are a large group of previously uncharacterized genes and several known genes that have not yet been implicated in the development of the peripheral nervous system.

THE development of the Drosophila adult external sensory (es) organ, a mechanosensory bristle, involves lateral inhibition and asymmetric division, two mechanisms that underlie numerous developmental processes (POSAKONY 1994 ; JAN and JAN 1995 ; CAMPOS-ORTEGA 1996 ). First, a single sensory organ precursor (SOP) cell is selected from a proneural cluster, a group of cells that are competent to become neuronal precursors, via lateral inhibition. Genes within the achaete-scute complex (AS-C) and the daughterless (da) gene are required to confer neuronal potential to these cells (GHYSEN and DAMBLY-CHAUDIERE 1989 ). After the SOP cell is singled out, it divides asymmetrically to produce two different secondary precursor cells, IIA and IIB. IIA gives rise to two external cells: one shaft cell (trichogen) and one socket cell (tormogen). IIB gives rise to the internal cells: one neuron, one sheath cell, and, for at least one class of es organs, an additional glial cell (HARTENSTEIN and POSAKONY 1989 ; GHO et al. 1999 ).

The Notch (N) signaling pathway mediates the cell-cell interactions that occur during lateral inhibition. The transmembrane protein Notch is a receptor and its principal ligand during lateral inhibition is Delta (reviewed in ARTAVANIS-TSAKONAS et al. 1999 ). Within the proneural cluster, Notch signaling is mediated through the transcription factor Suppressor of Hairless [Su(H)] and results in the activation of target genes at the Enhancer of split [E(spl)] locus (SCHWEISGUTH and POSAKONY 1992 ; FORTINI and ARTAVANIS-TSAKONAS 1994 ; BAILEY and POSAKONY 1995 ; JARRIAULT et al. 1995 ; LECOURTOIS and SCHWEISGUTH 1995 ). Hairless (H) is believed to act as an antagonist of Notch through physical interaction with Su(H) (BROU et al. 1994 ; BANG et al. 1995 ).

Both Notch-mediated cell-cell interactions and asymmetric segregation of the cell-intrinsic determinant Numb operate during divisions of the SOP lineage (POSAKONY 1994 ; RHYU et al. 1994 ). During divisions of the SOP cell and its progeny, Numb protein is unequally segregated to one of the two resulting daughter cells. In that cell, Numb inhibits the activity of N, which receives signals from two redundant ligands, Delta and Serrate (RHYU et al. 1994 ; FRISE et al. 1996 ; GUO et al. 1996 ; ZENG et al. 1998A ). The pathways downstream of Notch are different for the asymmetric divisions of IIA and IIB cell lineages. Su(H) acts as a transducer of Notch signaling only within IIA and her daughter cells; the downstream molecules that mediate Notch signaling in the IIB cell lineage are unknown (WANG et al. 1997 ). A potential downstream target of Su(H) in IIA is tramtrack (ttk), a gene that does not appear to have a function during lateral inhibition (GUO et al. 1995 , GUO et al. 1996 ). Another gene that affects lineage events and might be a component of the Notch signaling pathway is sanpodo (DYE et al. 1998 ; SKEATH and DOE 1998 ).

The Notch signaling cascade in the SOP cell lineage differs from that involved in lateral inhibition. Additional components involved in N signaling during asymmetric divisions of the SOP lineage remain to be identified (e.g., ones that are specific for the IIB cell lineage).

Many genes with a function in lateral inhibition or asymmetric divisions of the adult es organ lineage, such as N, Delta, numb, prospero (pros), and ttk, were initially identified due to embryonic loss-of-function (lof) phenotypes (LEHMANN et al. 1981 , LEHMANN et al. 1983 ; UEMURA et al. 1989 ; DOE et al. 1991 ; VAESSIN et al. 1991 ; XIONG and MONTELL 1991 ; SALZBERG et al. 1994 ). However, pleiotropy or redundancy of gene function may hamper the identification of other genes important for the formation of the adult es organ. One strategy to identify such genes is to look for gain-of-function (gof) phenotypes.

For this purpose, we screened 2293 independent Drosophila lines with the modular P-element-based EP (enhancer/promoter) misexpression element devised by P. Rørth (RORTH 1996 ; RORTH et al. 1998 ). This misexpression element contains upstream activating sequence (UAS) sites that are recognized by the transcriptional activator Gal4 (BRAND and PERRIMON 1993 ). Tissue-specific overexpression of genes that lie near the EP element can be achieved by using a line that expresses Gal4 in specific cells. In cells that both express Gal4 and carry the EP element, Gal4 binds to the UAS sites and causes misexpression of the adjacent gene.

On the basis of overexpression studies with genes previously shown to be involved in adult es organ formation, we expected certain phenotypes from such a gof screen. Overexpression of genes such as numb, ttk, Su(H), H, and N give phenotypes opposite to the respective lof phenotypes (BANG and POSAKONY 1992 ; LIEBER et al. 1993 ; RHYU et al. 1994 ; SCHWEISGUTH and POSAKONY 1994 ; GUO et al. 1995 ; DOHERTY et al. 1997 ; WANG et al. 1997 ). Overexpression of N or its transducer Su(H) during lateral inhibition results in loss of entire es organs due to suppression of SOP formation. At later stages, during asymmetric division, overexpression of these two genes produces up to four external cells, all socket-like, due to IIB-to-IIA cell and/or shaft-to-socket cell transformations (LIEBER et al. 1993 ; SCHWEISGUTH and POSAKONY 1994 ; WANG et al. 1997 ; DOHERTY et al. 1997 ; Fig 1). Conversely, misexpression of H, which antagonizes Notch signaling, results in increased numbers of SOPs, IIA-to-IIB, and socket-to-shaft transformations (BANG and POSAKONY 1992 ).

View larger version (16K): In this window In a new window Download PPT slide   Figure 1. Potential cell fate transformations in the IIA sublineage. (A) In wild-type, IIA divides asymmetrically to give rise to shaft (sh) and socket (so) cells. (B) Reduction of N signaling results in socket-to-shaft transformations. (C) Conversely, increased N signaling (e.g., in Hairless mutants) results in shaft-to-socket transformations. Genetic interactions were assayed on the basis of the effects of the EP misexpression on heterozygous mutant N or H phenotypes and vice versa.

In our screen, we first identified lines that produced visible misexpression phenotypes in the external cells of the es organ, i.e., the daughters of IIA. Next, we analyzed the effect of misexpression on the sheath cell, a daughter of IIB. Finally, we examined the effect of reducing N or H function on the gof phenotype. These analyses, combined with preliminary molecular characterizations, have led to the identification of genes previously shown to be important for es organ development, as well as other genes that may be involved in this process.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Drosophila stocks:
The collection of 2293 EP target element lines was a generous gift of P. Rørth through the Berkeley Drosophila Genome Project. For tissue-specific analysis of the misexpression effects, the individual EP lines were crossed to sca-Gal4, a P{Gal4} line with an insertion at the scabrous locus (NAKAO and CAMPOS-ORTEGA 1996 ). The sca-Gal4 line expresses Gal4 in SOP and surrounding cells and later in the lineage of the es organ. To test the effects of different levels of expression, parents from initial crosses were serially transferred and progeny from individual crosses were raised at 18, 25, and 29° during larval and pupal stages. The phenotypes at 29° were generally stronger and more penetrant. All subsequent crosses were maintained at 29°.

The A101 line carries an insertion of P{lacZ,ry⁺} at the neuralized locus (USUI and KIMURA 1993 ). It expresses nuclear ß-galactosidase in the SOP cell and the es organ lineage. On the notum, lacZ expression is strongest in the nuclei of the two external support cells. The pros-lacZ enhancer trap line P{lacZ,ry⁺} expresses ß-galactosidase in the sheath cell. We visualized ß-galactosidase expression by X-gal staining of pharate adults.

Genetic interactions:
To test genetic interactions with N, males from individual EP lines were crossed to w^aN^55E11/FM6; sca-Gal4/CyO females and the phenotypes of w^aN^55E11/+; sca-Gal4/+ flies carrying one copy of the EP element were compared to those of FM6/+; sca-Gal4/+ flies carrying one copy of the EP element and to those of w^aN^55E11/+; sca-Gal4/+ flies without the EP element. Most lines that showed a positive interaction were retested using a reciprocal crossing scheme with w^aN^55E11 /w·Y; sca-Gal4/CyO males (w·Y is a partial duplication of the first chromosome including the N locus). Genetic interactions with H were tested by crossing males from individual EP lines with y w; sca-Gal4/CyO; FRT H^E21/TM3 females. Phenotypes of y w; sca-Gal4/+; FRT H^E21/+ flies with one copy of the EP element were compared to those of y w; sca-Gal4/+; TM3/+ flies carrying one copy of the EP element and to those of sca-Gal4/+; FRT H^E21/+ flies without the EP element. For most crosses, parents were serially transferred and progeny from individual crosses were maintained at 18, 25, and 29° during larval and pupal stages. This genetic interaction scheme allowed us to evaluate changes of the EP misexpression phenotypes as an enhancement or suppression. In addition, enhancement or suppression of the H mutant phenotype was evaluated. Since N/+ flies lack a bristle phenotype, only the enhancement of N haploinsufficiency could be detected.

Molecular analysis:
Genomic sequences flanking the 3'end of the EP misexpression element were isolated by plasmid rescue using EcoRI or SacII (PIROTTA 1986 ). Sizes of three independent clones for each plasmid rescue were compared to determine the number of insertions per line. In total, there were 7 lines with two insertions (7/105 = 6.7%). Genomic sequences adjacent to the EP element were sequenced.

Flanking sequences were analyzed by searching the Berkeley Drosophila Genome Project (BDGP) and National Center for Biotechnology Information databases. Expressed sequence tags (EST) within a 3-kb distance from EP element insertion sites were tested for sequence similarities using "blastx" searches. Sequenced genomic regions within a 3-kb distance from EP element insertions for which no candidate transcripts had been identified were tested using open reading frame finders. Only significant sequence similarities were reported (see Table 1).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Using the modular misexpression system (RORTH 1996 ; RORTH et al. 1998 ), we misexpressed genes in the SOP cell and its neighbors and examined the effects on the development of the adult external sensory organ. The sca-Gal4 line was chosen as driver because it is expressed in clusters of cells surrounding the presumptive macro- and microchaetae on the notum and head (Fig 2). Expression persists in the SOP lineage. All misexpression phenotypes described in this paper are produced by sca-Gal4 in conjunction with an EP insertion. We then examined the effects of reducing N or H function on the gof phenotype. The enhancer trap lines A101 and pros-lacZ were used to assist our characterization of misexpression phenotypes. A101-lacZ expresses ß-galactosidase strongly in the nuclei of the two external support cells, while pros-lacZ expresses ß-galactosidase specifically in the sheath cell, one of the internal cells.

View larger version (82K): In this window In a new window Download PPT slide   Figure 2. Macro- and microchaetae are arranged in stereotyped patterns on the notum of Drosophila (for recent review on es organ pattern formation, see SIMPSON et al. 1999 ). (A) Four dorsocentral (dc) and four scutellar (sc) macrochaetae decorate the adult notum. (B) sca-Gal4 expresses Gal4 (in green, driving UAS-GFP) within the four cut-expressing cells of the es organ (red) and surrounding cells (BLOCHLINGER et al. 1993 ). On the scutellum and between the dorsocentral macrochaetae, sca-Gal4 is expressed not only in the developing sensory organs but also in surrounding domains.

In total, 4.6% of the lines (105/2293) produced phenotypes affecting the number or fate of outer cells of the es organ. These phenotypes fall into three major classes:

1. class I: loss of external support cells (sockets and shafts)

2. class II: supernumerary es organs or support cells

3. class III: potential cell fate transformations, with increases in one cell type associated with loss of another cell type.

Table 1 and Table 2 summarize the molecular, phenotypic, and genetic interaction data presented in this study. Many EP lines resulted in phenotypes with characteristics of more than one class. To simplify the classification, all EP lines with potential lineage transformation phenotypes were grouped into class III independently of other phenotypes. Similarly, among the remaining EP lines, those with phenotypes that include supernumerary es organs or subsets of support cells were grouped into class II independently of other phenotypes. Many lines in all three classes also exhibited an altered morphology of shaft or socket cells.

Loss of external cells:
We identified 52 lines representing 40 loci that produced loss of some or all of the external and internal support cells. Loss of both external and internal support cells could arise from loss of the entire es organ. Alternatively, the support cells could have been transformed into neurons. Genes responsible for such phenotypes could interfere with lateral inhibition and function in lineage decisions, prevent cell cycle progression, or result in cell lethality.

This is the largest class of EP lines and includes P-element insertions into genes known to have important functions in asymmetric cell division, lateral inhibition, and other aspects of development. For example, misexpression of extra macrochaetae (emc) by EP(2)0415 caused a loss of macro- and microchaetae (Fig 3A) that resembles the phenotype of a dominant emc mutation (emc^D; CRAYMER 1980 ). emc acts as a repressor that blocks the activity of achaete and scute gene function during sensory organ neurogenesis (ELLIS et al. 1990 ; GARRELL and MODOLELL 1990 ; SKEATH and CARROLL 1991 ; VAN DOREN et al. 1991 ) and its misexpression is predicted to block SOP formation.

View larger version (84K): In this window In a new window Download PPT slide   Figure 3. Examples of class I misexpression phenotypes. (A) Misexpression of EP(3)0415 at the extra macrochaetae locus resulted in the loss of scutellar and dorsocentral macro- and microchaetae. (B) Several insertions targeting escargot, including EP(2)0684, resulted in the loss of almost the entire population of macro- and microchaetae. (C) Misexpression of EP(2)2584 at the dacapo locus resulted in the loss of external cells of scutellar and dorsocentral macrochaetae. The shaft cell morphology of many macrochaetae was abnormal. The arrowhead indicates an abnormal shaft cell morphology.

Another example is the misexpression of escargot (esg) [by EP(2)0683, EP(2)0684, EP(2)2009, EP(2)2159, and EP(2)2408], which caused the most severe loss of es organs observed in this screen. In EP(2)0684 and EP(2)2009, there was an almost complete loss of es organs on the notum (Fig 3B). esg encodes a zinc finger protein that acts as a repressor of Scute/Daughterless-dependent transcription in vitro (WHITELEY et al. 1992 ; FUSE et al. 1994 ). It also acts as negative regulator of endoreplication in imaginal tissues (HAYASHI et al. 1993 ; HAYASHI 1996 ).

We also identified several genes known to be required for correct cell cycle progression. dacapo [EP(2)2584] is a cyclin-dependent kinase inhibitor that is required during embryogenesis for a timely exit from the cell cycle (LANE et al. 1996 ; DE NOOIJ et al. 1996 ). Misexpression of dacapo produced a loss of external cells of scutellar and dorsocentral macrochaetae (Fig 3C). In some cases, there was a single prospero-positive cell that was no longer accompanied by shaft and socket cells. Another gene, divisions abnormally delayed (dally), encodes a proteoglycan that is required for normal cell cycle progression (NAKATO et al. 1995 ) and might act as coreceptor for Wingless (LIN and PERRIMON 1999 ; TSUDA et al. 1999 ). Misexpression of this gene by EP(3)3168 resulted in the occasional loss of scutellar or dorsocentral macrochaetae. Misexpression of these genes could interfere with SOP lineage events by blocking cell cycle progression (e.g., by forcing the SOP cell to exit mitosis) or, in the case of dally, by affecting Wingless signaling, which is involved in the patterning of es organs (PHILLIPS and WHITTLE 1993 ).

A large number of P-element insertions targeted genes that are known to have essential functions during development but have not previously been implicated in sensory organ development. One line, carrying an insertion at the inscuteable (insc) locus [EP(2)2010], exhibited a loss of external structures of scutellar macrochaetae without a concurrent loss of the prospero-positive sheath cell. Whether this phenotype is entirely due to altered expression of insc, which serves an essential function in asymmetric divisions of delaminating neuroblasts and embryonic muscle progenitor cell divisions (KRAUT et al. 1996 ; CARMENA et al. 1998 ), requires further study. One potential complication is the presence of the gene skittles, which encodes the phosphatidylinositol 4-phosphate 5-kinase, in the first intron of insc. Misexpression of skittles has been shown to generate ectopic es organs (HASSAN et al. 1998 ). It is not clear whether misexpression of insc, skittles, or both is driven by EP(2)2010.

Other known developmental regulators found in this screen include gliotactin [EP(2)2306], which encodes a transmembrane protein that functions in peripheral glia to establish the blood-nerve barrier (AULD et al. 1995 ); fat facets [EP(3)0381], which encodes a deubiqui-tination enzyme required for correct eye development (FISCHER-VIZE et al. 1992 ; HUANG et al. 1995 ); apontic [EP(2)2373], a gene involved in multiple processes, including head patterning (GELLON et al. 1997 ) and heart morphogenesis (SU et al. 1999 ); Drosophila lim-domains only [EP(X)1306, EP(X)1383, and EP(X)1394], a gene with a role in wing patterning (MILAN et al. 1998 ; SHORESH et al. 1998 ; ZENG et al. 1998B ), longitudinals lacking (lola) [EP(2)0343], which is required for correct axonal projection (GINIGER et al. 1994 ); and hnRNP 27C [EP(2)0748], which encodes a heterogeneous nuclear RNA-associated protein. Previous studies suggest that different heterogeneous nuclear RNA-associated proteins may play a role in the development of the es organ (HAMMOND et al. 1997 ; ZUR LAGE et al. 1997).

This class includes insertions at 15 previously uncharacterized genes. Four of these insertions showed genetic interactions with N or H (see Table 2), indicating that they affect genes that are potentially in the N signaling pathway. These genes are therefore good candidates for future analyses.

Supernumerary es organs or support cells:
Thirty-eight lines, carrying insertions at 28 loci, caused misexpression phenotypes with increased numbers of internal and external cell types. We further subdivided these lines into two subclasses. One subclass of lines produced ectopic (i.e., spatially separate) es organs; these might arise from defective lateral inhibition or ectopic proneural activity. The other subclass of lines exhibited supernumerary support cells that were clustered together. This phenotype could be due to either increased cell numbers within an es organ or formation of several tightly associated es organs. Such phenotypes could result from defects in lateral inhibition or cell cycle regulation.

In this class, there are 16 previously uncharacterized genes (Table 1). To distinguish lines that affect lateral inhibition from those that affect other functions, we tested a subset of these lines for genetic interactions with N and H. Eight lines representing eight independent loci displayed significant genetic interactions (see Table 2).

Ectopic supernumerary es organs: This subclass includes big brain [EP(2)2278], a gene involved in lateral inhibition that encodes a channel-like transmembrane protein (RAO et al. 1990 ). Also in this subclass are two genes with a known function in eye development: yan [EP(2)0598 and EP(2)2500], which encodes an ETS domain nuclear protein that has an essential function in photoreceptor cell development (LAI and RUBIN 1992 ; O'NEILL et al. 1994 ); and hedgehog [EP(3)3521], which is involved in multiple developmental processes including eye furrow progression (HEBERLEIN et al. 1993 ; MA et al. 1993 ). hedgehog has also been implicated in the correct patterning of es organs on the adult notum (GOMEZ-SKARMETA and MODOLELL 1996 ; MULLOR et al. 1997 ). Another gene, split ends (spen) [EP(2)2583], resulted in a misexpression phenotype with increased numbers of scutellar and dorsocentral macrochaetae (Fig 4A). spen has multiple developmental functions including correct axon formation (KOLODZIEJ et al. 1995 ) and control of correct segment identity (WIELLETTE et al. 1999 ). Two insertions near nuclear fallout [EP(3)3324 and EP(3)3339] resulted in additional scutellar macrochaetae and in one-socket/two-shaft phenotypes. This gene encodes a coiled-coil protein with a function in cortical actin organization and cytokinesis (ROTHWELL et al. 1998 ).

View larger version (68K): In this window In a new window Download PPT slide   Figure 4. Examples of class II misexpression phenotypes. (A) Misexpression of EP(2)2583 at the split ends locus resulted in ectopic additional scutellar and dorsocentral macrochaetae (arrowheads). (B) Insertion EP(3)3622 resulted in tufting, a phenotype with clustered shafts, and ectopic scutellar and dorsocentral macrochaetae. (C) Misexpression of EP(3)3559, which targets a new regulatory subunit of protein phosphatase2A, resulted in increased numbers of support cells. (D) Similarly, misexpression of EP(2)2437 resulted in increased numbers of internal and external cell types. EP(2)2437 is an insertion in antisense orientation within EST SD02913 and may cause lof effects. Arrows indicate ectopic macrochaetae.

Several previously uncharacterized genes targeted by the EP element displayed genetic interactions with N and H. For example, EP(3)3622 produced a misexpression phenotype with additional es organs and tufts (i.e., a large number of clustered shafts; Fig 4B). The misexpression phenotype produced by EP(3)3622 is enhanced by removing one copy of N and suppressed by removing one copy of H (Table 2).

Increased numbers of internal and external support cells: Supernumerary internal and external support cells could arise from ectopic cell divisions caused by altered cell cycle regulation. A previously uncharacterized gene targeted by EP(3)3559 has sequence similarities with human regulatory subunits of protein phosphatase 2A (PP2A). Genes coding for the regulatory subunit B of PP2A (abnormal anaphase, twins) are involved in both cell cycle progression and cell fate determination (GOMES et al. 1993 ; SHIOMI et al. 1994 ). EP(3)3559 shows increased numbers of support cells in each es organ (Fig 4C). This misexpression phenotype mimics the phenotype observed in twins, a mutation in the regulatory B subunit of PP2A (UEMURA et al. 1993 ). Regulatory subunits that are under temporal or tissue-specific control in turn regulate the activity of PP2A. It will be of interest to test how the newly identified regulatory subunit regulates the function of PP2A.

Three insertions at a novel locus, EP(2)0639, EP(2)2148, and EP(2)2437, produce supernumerary support cells in the es organ (Fig 4D). The orientation of the EP elements at this locus is such that they presumably generate a partial antisense transcript. Therefore, the phenotypes could be caused by lof or neomorphic effects.

Genetic interactions with N and H were found with EP(2)0647, an insertion at a gene that has sequence similarities with BTB-domain-containing proteins such as Pipsqueak. Misexpression of this gene resulted in, among other phenotypes, increased numbers of support cells associated with es organs.

Potential cell fate transformations:
We expected to identify P-element insertions that target genes that function in the asymmetric divisions of the stereotyped es organ lineage. In total, 15 lines representing 10 loci resulted in apparent cell fate transformations. These lines fall into three subclasses. The first two subclasses are transformations within the IIA cell sublineage: (a) a socket-to-shaft cell transformation, which would result in a two-shaft/no-socket phenotype (twinned phenotype); and (b) a shaft-to-socket cell transformation, which would result in a no-shaft/two-socket phenotype. The third subclass is transformations from IIA to IIB, which would result in loss of external support cells (balding). However, mechanisms other than transformations may cause these phenotypes as well (e.g., ectopic cell division of one type of support cell combined with the elimination of another type of support cell).

Potential transformations of socket cell to shaft cell: The misexpression of numb by EP(2)2542 resulted in socket-to-shaft transformations similar to the numb overexpression phenotype (Fig 5A; RHYU et al. 1994 ). The misexpression phenotype of EP(2)2542 also included the loss of external structures of macrochaetae. This phenotype might be the result of IIA-to-IIB transformations.

View larger version (77K): In this window In a new window Download PPT slide   Figure 5. Examples of class III misexpression phenotypes. (A) Misexpression of EP(2)2542 at the numb locus resulted in apparent socket-to-shaft transformations. (B) EP(2)0587 at the grapes locus caused apparent shaft-to-socket transformations on the abdomen. Double sockets are indicated by the presence of two large A101 lacZ-positive nuclear stains. (C) Misexpression of EP(2)0386 produced apparent shaft-to-socket transformations on the abdomen, as indicated by the presence of two large A101 lacZ-positive nuclear stains. (D) The abdominal misexpression phenotypes of EP(3)0596 were apparent shaft-to-socket transformations (asterisk) and branching of shaft cells (arrowhead). (E) Misexpression of EP(2)2478 resulted in apparent IIA-to-IIB or neuron-to-sheath transformations. In the absence of external support cells, two proslacZ- positive sheath cells were tightly associated (asterisk). (F) Similarly, misexpression of EP(3)3390 resulted in apparent IIA-to-IIB or neuron-to-sheath transformations. Two associated proslacZ-positive sheath cells were commonly scored in the absence of differentiated external structures (asterisk). However, abnormal cuticular structures were visible (arrowheads). Potential transformation phenotypes are indicated with an asterisk.

Each of the two insertions [EP(X)1149 and EP(X)1179] that target the same unknown gene produced both socket-to-shaft and reciprocal shaft-to-socket transformations (Fig 6C). Both lines also caused a loss of external support cells on the notum.

View larger version (119K): In this window In a new window Download PPT slide   Figure 6. A group of 41 EP lines carry insertions near genes that when misexpressed, produced an abnormal es organ morphology. Examples are as follows: (A) EP(2)2356 caused branching of shafts into multiple tips (arrowheads). (B) Flattened and enlarged socket cells were commonly scored with EP(3)3463. (C) EP(X)1149 resulted in potential shaft-to-socket transformations. Socket cells frequently displayed protruding shaft-like tips. (D) EP(2)2317 resulted in the severe reduction of shaft cells into shortened or dot-like structures. Arrowheads indicate abnormal cell morphology. Potential transformation phenotypes are indicated with an asterisk.

Potential transformations of shaft cell to socket cell: This subclass includes string, twine, and grapes, three genes with a function in mitotic or meiotic cell cycle regulation (EDGAR and O'FARRELL 1989 ; ALPHEY et al. 1992 ; COURTOT et al. 1992 ; FOGARTY et al. 1994 , FOGARTY et al. 1997 ). We identified four independent insertions at or near the string locus [EP(3)1213, EP(3)3261, EP(3)3426, and EP(3)3432]. With the exception of EP(3)1213, which carries an insertion ~1.5 kb upstream of the normal transcript, the other three insertions lie close to the transcription initiation site (see Table 1). However, only EP(3)1213 resulted in possible shaft-to-socket transformations, raising the question whether a gene other than string is affected in this line. The misexpression by EP(3)3261 produced increased numbers of internal and external support cells. X-gal staining with enhancer trap lines A101 lacZ and prospero lacZ, which mark the external and the sheath cells, respectively, showed an approximate doubling of the cell number in many es organs (not shown).

Insertions near grapes [EP(2)0587] and twine [EP(2)0613] resulted in potential shaft-to-socket transformations on the abdomen and notum, respectively (Fig 5B). Mutations in grapes, a protein kinase with homologies to Saccharomyces cerevisiae CHK1, have been shown to interfere with the DNA replication checkpoint control of the cell cycle (FOGARTY et al. 1997 ). In addition, embryos mutant in grapes exhibit cortical cytoskeletal defects during syncytial divisions (SULLIVAN et al. 1993 ). Misexpression of twine caused, in addition to possible shaft-to-socket transformations, a four-socket phenotype. twine, a cdc25 homolog, has a function during male and female meiotic divisions and participates in some aspects of mitotic control at the syncytial embryo stage (ALPHEY et al. 1992 ; COURTOT et al. 1992 ; EDGAR and DATAR 1996 ).

The most prominent phenotype found with two other lines, [EP(2)0386 and EP(2)0988], was apparent shaft-to-socket cell transformations on the abdomen. X-gal staining with enhancer trap line A101 lacZ, which predominantly marks two large nuclei of the two external cells of the es organ, confirmed the presence of two socket cells (Fig 5C). A third line, EP(3)0596, produced a similar misexpression phenotype (Fig 5D).

Potential transformations of IIA to IIB: Two insertions at two independent loci each produced potential IIA-to-IIB cell fate transformations, with two or more prospero-positive cells in the absence of external support cells. With EP(2)2478, both macro- and microchaetae exhibited a loss of external support cells as well as a duplication of presumptive sheath cells (Fig 5E). Similarly, the misexpression caused by EP(3)3390 resulted in a loss of external support cells of macro- and microchaetae as well as duplication of prospero-positive sheath cells (Fig 5F). In rare cases, up to four sheath cells were present.

Defective morphology of the es organ:
At least 41 lines, representing 38 loci, identified in this screen produced aberrant morphology of either the socket or the shaft cell. The following are examples of different morphology phenotypes observed.

Misexpression driven by EP(2)2356 produced an abnormal shaft cell morphology. Most prominently, the shaft cell was short and branched into many distal tips (Fig 6A). Branching of the shaft cell into two distal tips was observed in several lines [i.e., in EP(3)0596, Fig 5D].

Morphologically abnormal socket cells were produced by EP(3)3463. Among other phenotypes, the socket cells frequently were large and flattened (Fig 6B). EP(X)1149 (see also phenotype in class III) produced an abnormal socket cell morphology with a protruding tip similar to a short shaft (Fig 6C).

We observed a massive reduction in the size of shaft cells and morphologically abnormal socket cells with EP(2)2317, an insertion at elF-4A (Fig 6D). Similar phenotypes were seen with several other lines.

The sensitivity of cell morphology to the misexpression of candidate genes might yield an entry point to identify genetic components involved in differentiation and morphogenesis. Several of the phenotypes described here resemble phenotypes caused by mutations of genes that function in cytoskeletal assembly (CANT et al. 1994 ; TILNEY et al. 1995 , TILNEY et al. 1996 ).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Analyzing development of the es organ using a gain-of-function approach:
Traditionally, genetic screens have been based on the isolation of lof mutations. This approach has been invaluable in unraveling the mechanisms underlying many biological processes, including the formation of the peripheral nervous system (SALZBERG et al. 1994 ; KANIA et al. 1995 ; GO et al. 1998). However, lof screens have several limitations. Redundancy between genes that have overlapping functions might partially or completely mask gene function. In such cases, it is necessary to make double or multiple mutant combinations to produce a phenotype, an approach that is not generally applicable during lof screens. Moreover, early phenotypes caused by a mutation might prevent the detection of later phenotypes (MIKLOS and RUBIN 1996 ). Such limitations can be partially circumvented by screens that are based on analyzing the phenotypes of clones of mutant tissue generated by somatic recombination (XU and RUBIN 1993 ) or by screens for enhancers or suppressors of a particular mutant phenotype (SIMON et al. 1991 ). Nevertheless, many genes might have escaped detection by lof approaches.

The gof screening system devised by P. Rørth complements lof approaches. This system is based on the analysis of phenotypes generated by tissue-specific misexpression of genes using the UAS-Gal4 system. Any gene that produces a misexpression phenotype is detectable by the system in spite of possible functional redundancy and pleiotropy of gene function (RORTH 1996 ; RORTH et al. 1998 ). In addition, the tissue specificity of the UAS-Gal4 system allows the examination of misexpression phenotypes in the biological context of choice. In various screens, phenotypes that affected eye development, wing development, and follicle cell migration were analyzed (RORTH et al. 1998 ).

In this study, these 2293 randomly inserted P elements were each driven by a sensory-organ-specific Gal4 driver and any resulting misexpression phenotypes in the es organ were analyzed. Of these lines, 105 produced es organ phenotypes. Our preliminary phenotypic and molecular analyses suggest that we have identified genes that are involved in lateral inhibition, cell cycle control, cell fate specification, and cell differentiation. A subset of these genes is likely to play a role in es organ formation.

One potential drawback of gof screens is that misexpression of a gene may affect the development of tissues in which that gene is not normally expressed. In some cases, misexpression of a gene may ectopically effect a signaling pathway that functions in multiple developmental processes. Another concern is that phenotypes may be artificial. For example, the phenotype caused by misexpression of a gene at levels much higher than normal may interfere with development, even if that gene does not have a function in development.

To identify those genes that normally function in es organ development, it will be important to examine the lof phenotype, the expression pattern, and genetic interactions with genes known to be involved in es organ development.

The systematic misexpression screen identifies candidate genes that interfere with distinct developmental aspects of es organ formation:
Among the 105 lines (78 loci) identified in the screen, 49 lines (37 loci) correspond to previously characterized genes. A subset of these genes has been shown to have roles during es organ development. Some, such as emc and big brain, have a function in lateral inhibition (SKEATH and CARROLL 1991 ; RAO et al. 1992 ). Several are genes with a function in cell cycle regulation, including dacapo and string, and thus might be required during es organ cell division. Others, such as numb, are known to be involved in asymmetric cell division (RHYU et al. 1994 ). Moreover, a large group of genes with essential roles in other developmental processes were identified. Some of these genes, such as hedgehog and yan, have not been tested for their role in es organ development, but it is possible that they are involved in this developmental process as well. Since many of the known genes identified in this screen are likely to have normal functions in es organ development, the concern of the potentially artificial nature of the gof screen may be alleviated. It thus seems likely that at least a substantial subset of the new genes identified in our screen will turn out to be important for the formation of es organs, perhaps in some of the less understood aspects of es organ development, including the following:

Context-specific components of the N-signaling pathway: The transducers of N signaling in IIB and her daughters are currently not known (WANG et al. 1997 ). EP(2)2478 and EP(3)3390 target genes with possible functions in IIB and her daughters. Misexpression of those genes was sufficient to generate potential IIA-to-IIB or neuron-to-sheath transformations. One possible explanation for this phenotype is ectopic activation of IIB-specific target genes (e.g., by IIB or sheath-cell-specific N-signaling components).

Cell cycle regulation of stereotyped lineage events: One likely link between cell cycle regulation and asymmetric cell division is the cell-cycle-dependent asymmetric localization of cell fate determinants and adaptor proteins (HIRATA et al. 1995 ; KNOBLICH et al. 1995 ; SPANA and DOE 1995 ; KRAUT et al. 1996 ; IKESHIMA-KATAOKA et al. 1997 ; SHEN et al. 1997 ; LU et al. 1998 , LU et al. 1999 ; SCHULDT et al. 1998 ). Untimely cell cycle progression or defective integration of cell cycle with the localization of Numb protein may create a phenotype reminiscent of numb lof, a phenotype that was observed with misexpression of the cell cycle regulatory genes grapes and twine.

In addition, cell cycle regulatory genes may serve additional functions that affect cell fate specification. grapes, for example, is essential for the normal formation of the cortical cytoskeleton during syncytial divisions (SULLIVAN et al. 1993 ). Given the importance of the cortical cytoskeleton during asymmetric division (BROADUS and DOE 1997 ; KNOBLICH et al. 1997 ), genes that regulate the dynamics of this structure may also turn out to be essential during cell fate decisions.

Highly stereotyped division patterns occur throughout Drosophila development (FOE 1989 ; GHO et al. 1999 ). Several cell cycle regulators, including dacapo, are required to control the cell division patterns in the neural lineages of the embryonic nervous system (CUI and DOE 1995 ; WEIGMANN and LEHNER 1995 ; DE NOOIJ et al. 1996 ; LANE et al. 1996 ; HASSAN and VAESSIN 1997 ). It is not known at this time whether dacapo normally functions during the development of the es organ to control precise cell division patterns.

Execution of morphogenesis: There are different types of genes that when misexpressed could give rise to morphology defects. These include genes that affect differentiation of a single cell type (e.g., shaft cell differentiation controlled by pax2; KAVALER et al. 1999 ) or that affect proper regulation of cytoskeletal dynamics. We found a large number of lines that, when misexpressed, resulted in aberrant morphogenesis of the socket or shaft cell. One phenotype observed was the branching of shafts. It has been suggested that mutations causing branched hairs are in genes that regulate the actin cytoskeleton (TURNER and ADLER 1998 ). Consistent with this prediction, mutations of genes with a function in actin bundle formation display similar branching phenotypes (CANT et al. 1994 ; TILNEY et al. 1995 , TILNEY et al. 1996 ). Several of the lines identified in this screen might provide additional components involved in executing shaft cell morphology or in regulating the actin cytoskeleton in other tissues. Less is known about the morphogenesis of socket cells. EP lines that affected predominantly socket cell morphology might provide clues to this process.

Genomic considerations and perspectives:
Genome sequencing by the European and Berkeley Drosophila Genome Projects (EDGP and BDGP) and the ease with which genomic sequences flanking the EP element can be cloned have greatly facilitated the identification of targeted genes. Of the insertion sites we sequenced, 49 (37 loci; 46.7% of all lines) matched known genes, 34 (28 loci; 32.4% of all lines) matched EST, and 22 (13 loci; 20.9% of all lines) matched sequenced genomic regions but still have no candidate transcripts.

Altogether, 105 lines or 4.5% of the lines tested gave rise to misexpression phenotypes. RORTH et al. 1998 reported comparable frequencies of misexpression phenotypes: 7% with ombGal4, 4% with dppGal4, 3% with slboGal4, and 2% with sevGal4. Among the few genes that were reported from those screens, we have isolated escargot, hedgehog, yan, scalloped, and big brain. It will be interesting to compare those screens to obtain an estimate of the overlap of the genes used in those different developmental processes.

In a separate database analysis, we searched for EP element insertions that target genes with a known function in neurogenesis and sensory organ development. Among seven EP element insertions that target six genes (extra macrochaetae, big brain, kuzbanian, neuralized, and Enhancer of split transcripts m2 and m7), only two insertions near two loci yielded misexpression phenotypes in our assay (extra macrochaetae, big brain). Five insertions near four loci did not cause obvious misexpression phenotypes (Table 3). Therefore, the misexpression screen was not fully efficient. Similarily, there may be other unknown genes with a function in es organ development that escaped detection even with an EP element inserted nearby.

Determining the exact insertion site and orientation of the EP element is essential to the interpretation of misexpression phenotypes. In the lines for which we identified a transcript, most of the EP transposons were inserted between -850 bp upstream and +800 bp downstream of the transcription start site (61/83 = 73.5%). Seven lines (8.4%) were identified with insertions at greater distances from the transcription start site of putative target genes. In these cases it is possible that additional transcripts that have not been identified might be located closer to the EP element. One example is EP(3)1213, which carries an insertion ~1.5 kb 5' of the transcriptional start site of string. The misexpression phenotype produced by this line was qualitatively different from other EP insertions closer to the string transcriptional start site. Whether these differences are attributable to different levels of expression or are caused by an unidentified transcript needs to be determined. Another 9 lines (10.8%) carried EP elements with an apparent antisense orientation and might generate partial antisense transcripts. How these antisense messages might cause phenotypes is not clear. In addition, there are several lines (6/83 = 7.2%) that carried insertions 3' of the CDS, or insertions within new transcripts for which the CDS is not known. In these cases, the phenotypes might be caused by truncated transcripts.

The EP transposon allows only the unidirectional transcription of potential target genes. Therefore, ~50% of the EP lines are expected to be in the correct orientation to drive misexpression of a sense transcript [only nine of the lines that gave rise to phenotypes with sca-Gal4 (8.6%) had an inverted or antisense orientation]. Thus, the total number of genes targeted for overexpression in the screen might be no more than 1150. The number of targeted genes is further reduced by multiple lines targeting the same gene (1.33 insertions/locus) and by insertions that lie too distantly to drive sufficient transcriptional activation.

The current estimate for the number of genes in the Drosophila genome by the BDGP is around 14,000 (based on MIKLOS and RUBIN 1996 ). Therefore, the EP collection targets ~10% of the genome. In an extrapolation, for a genome-wide saturation screen we would expect 800 different loci or ~5–6% of all genes to give rise to misexpression phenotypes. The future challenge will be to determine the biological significance of the genes identified during this screen.

	ACKNOWLEDGMENTS

We are most grateful to P. Rørth for the generous gift of EP lines. We thank Todd Laverty and G. Rubin for kindly providing us with the EP lines, and the lab of J. Campos-Ortega and the Bloomington Drosophila Stock Center for fly strains. Our sequence analysis was helped by the sequencing efforts of BDGP and EDGP. Thanks to B. Lu and S. Zhou for critical reading of the manuscript; to D. Doherty for providing us with Fig 2; and to other current and former members of the Jan lab for discussion, suggestions, and help. S.A.-S. was supported by a fellowship from the Deutsche Forschungsgemeinschaft. Y.-M. C. currently is supported by the Program in Biological Sciences Markey Grant and the Herb Boyer Fund. C.Z. is a postdoctoral associate, N.J.J. is a predoctoral associate, and L.Y.J. and Y.N.J. are investigators of the Howard Hughes Medical Institute.

Manuscript received November 16, 1999; Accepted for publication February 24, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALPHEY, L., J. JIMENEZ, H. WHITE-COOPER, I. DAWSON, and P. NURSE et al., 1992  twine, a cdc25 homolog that functions in the male and female germline of Drosophila.. Cell 69:977-988[Medline].

ARTAVANIS-TSAKONAS, S., M. D. RAND, and R. J. LAKE, 1999  Notch signaling: cell fate control and signal integration in development. Science 284:770-776[Abstract/Free Full Text].

AULD, V. J., R. D. FETTER, K. BROADIE, and C. S. GOODMAN, 1995  Gliotactin, a novel transmembrane protein on peripheral glia, is required to form the blood-nerve barrier in Drosophila.. Cell 81:757-767[Medline].

BAILEY, A. M. and J. W. POSAKONY, 1995  Suppressor of hairless directly activates transcription of enhancer of split complex genes in response to Notch receptor activity. Genes Dev. 9:2609-2622[Abstract/Free Full Text].

BANG, A. G. and J. W. POSAKONY, 1992  The Drosophila gene Hairless encodes a novel basic protein that controls alternative cell fates in adult sensory organ development. Genes Dev. 6:1752-1769[Abstract/Free Full Text].

BANG, A. G., A. M. BAILEY, and J. W. POSAKONY, 1995  Hairless promotes stable commitment to the sensory organ precursor cell fate by negatively regulating the activity of the Notch signaling pathway. Dev. Biol. 172:479-494[Medline].

BLOCHLINGER, K., L. Y. JAN, and Y. N. JAN, 1993  Postembryonic patterns of expression of cut, a locus regulating sensory organ identity in Drosophila.. Development 117:441-450[Abstract].

BRAND, A. H. and N. PERRIMON, 1993  Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development 118:401-415[Abstract].

BROADUS, J. and C. Q. DOE, 1997  Extrinsic cues, intrinsic cues and microfilaments regulate asymmetric protein localization in Drosophila neuroblasts. Curr. Biol. 7:827-835[Medline].

BROU, C., F. LOGEAT, M. LECOURTOIS, J. VANDEKERCKHOVE, and P. KOURILSKY et al., 1994  Inhibition of the DNA-binding activity of Drosophila suppressor of hairless and of its human homolog, KBF2/RBP-J kappa, by direct protein-protein interaction with Drosophila hairless. Genes Dev. 8:2491-2503[Abstract/Free Full Text].

CAMPOS-ORTEGA, J., 1996  Numb diverts notch pathway off the tramtrack. Neuron 17:1-4[Medline].

CANT, K., B. A. KNOWLES, M. S. MOOSEKER, and L. COOLEY, 1994  Drosophila singed, a fascin homolog, is required for actin bundle formation during oogenesis and bristle extension. J. Cell Biol. 125:369-380[Abstract/Free Full Text].

CARMENA, A., B. MURUGASU-OEI, D. MENON, F. JIMENEZ, and W. CHIA, 1998  Inscuteable and numb mediate asymmetric muscle progenitor cell divisions during Drosophila myogenesis. Genes Dev. 12:304-315[Abstract/Free Full Text].

COURTOT, C., C. FANKHAUSER, V. SIMANIS, and C. F. LEHNER, 1992  The Drosophila cdc25 homolog twine is required for meiosis. Development 116:405-416[Medline].

CRAYMER, L., 1980  New mutants report. Dros. Inf. Serv. 55:197-200.

CUI, X. and C. Q. DOE, 1995  The role of the cell cycle and cytokinesis in regulating neuroblast sublineage gene expression in the Drosophila CNS. Development 121:3233-3243[Abstract].

DE NOOIJ, J. C., M. A. LETENDRE, and I. K. HARIHARAN, 1996  A cyclin-dependent kinase inhibitor, Dacapo, is necessary for timely exit from the cell cycle during Drosophila embryogenesis. Cell 87:1237-1247[Medline].

DOE, C. Q., Q. CHU-LAGRAFF, D. M. WRIGHT, and M. P. SCOTT, 1991  The prospero gene specifies cell fates in the Drosophila central nervous system. Cell 65:451-464[Medline].

DOHERTY, D., L. Y. JAN, and Y. N. JAN, 1997  The Drosophila neurogenic gene big brain, which encodes a membrane-associated protein, acts cell autonomously and can act synergistically with Notch and Delta.. Development 124:3881-3893[Abstract].

DYE, C. A., J. LEE, R. C. ATKINSON, R. BREWSTER, and P. HAN et al., 1998  The Drosophila sanpodo gene controls sibling cell fate and encodes a Tropomodulin homolog, an actin/tropomyosin associated protein. Development 125:1845-1856[Abstract].

EDGAR, B. A. and P. H. O'FARRELL, 1989  Genetic control of cell division patterns in the Drosophila embryo. Cell 57:177-187[Medline].

EDGAR, B. A. and S. A. DATAR, 1996  Zygotic degradation of two maternal Cdc25 mRNAs terminates Drosophila's early cell cycle program. Genes Dev. 10:1966-1977[Abstract/Free Full Text].

ELLIS, H. M., D. R. SPANN, and J. W. POSAKONY, 1990  extramacrochaetae, a negative regulator of sensory organ development in Drosophila, defines a new class of helix-loop-helix proteins. Cell 61:27-38[Medline].

FISCHER-VIZE, J. A., G. M. RUBIN, and R. LEHMANN, 1992  The fat facets gene is required for Drosophila eye and embryo development. Development 116:985-1000[Abstract].

FOE, V. E., 1989  Mitotic domains reveal early commitment of cells in Drosophila embryos. Development 107:1-22[Abstract].

FOGARTY, P., R. F. KALPIN, and W. SULLIVAN, 1994  The Drosophila maternal-effect mutation grapes causes a metaphase arrest at nuclear cycle 13. Development 120:2131-2142[Abstract].

FOGARTY, P., S. D. CAMPBELL, R. ABU-SHUMAYS, B. DE SAINT PHALLE, and K. R. YU et al., 1997  The Drosophila grapes gene is related to checkpoint gene chk-1 rad27 and is required for late syncytial division fidelity. Curr. Biol. 7:418-426[Medline].

FORTINI, M. E. and S. ARTAVANIS-TSAKONAS, 1994  The suppressor of Hairless protein participates in Notch receptor signaling. Cell 79:273-282[Medline].

FRISE, E., J. A. KNOBLICH, S. YOUNGER-SHEPHERD, L. Y. JAN, and Y. N. JAN, 1996  The Drosophila Numb protein inhibits signaling of the Notch receptor during cell-cell interaction in sensory organ lineage. Proc. Natl. Acad. Sci. USA 93:11925-11932[Abstract/Free Full Text].

FUSE, N., S. HIROSE, and S. HAYASHI, 1994  Diploidy of Drosophila imaginal cells is maintained by a transcriptional repressor encoded by escargot.. Genes Dev. 8:2270-2281[Abstract/Free Full Text].

GARRELL, J. and J. MODOLELL, 1990  The Drosophila extramacrochaetae locus, an antagonist of proneural genes that, like these genes, encodes a helix-loop-helix protein. Cell 61:39-48[Medline].

GELLON, G., K. W. HARDING, N. MCGINNIS, M. M. MARTIN, and W. MCGINNIS, 1997  A genetic screen for modifiers of Deformed homeotic function identifies novel genes required for head development. Development 124:3321-3331[Abstract].

GHO, M., Y. BELLAICHE, and F. SCHWEISGUTH, 1999  Revisiting the Drosophila microchaete lineage: a novel intrinsically asymmetric cell division generates a glia cell. Development 126:3573-3584[Abstract].

GHYSEN, A. and C. DAMBLY-CHAUDIERE, 1989  Genesis of the Drosophila peripheral nervous system. Trends Genet. 5:251-255[Medline].

GINIGER, E., K. TIETJE, L. Y. JAN, and Y. N. JAN, 1994  lola encodes a putative transcription factor required for axon growth and guidance in Drosophila.. Development 120:1385-1398[Abstract].

GO, M. J. and S. ARTAVANIS-TSAKONAS, 1998  A genetic screen for novel components of the notch signaling pathway during Drosophila bristle development. Genetics 150:211-220[Ab