Mutations Affecting the Development of the Peripheral Nervous System in Drosophila: A Molecular Screen for Novel Proteins

Corresponding author: Sergei N. Prokopenko, The Salk Institute for Biological Studies, MNL-T, P.O. Box 85800, San Diego, CA 92186-5800., prokopenko{at}salk.edu (E-mail)

Communicating editor: T. F. C. MACKAY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

In our quest for novel genes required for the development of the embryonic peripheral nervous system (PNS), we have performed three genetic screens using MAb 22C10 as a marker of terminally differentiated neurons. A total of 66 essential genes required for normal PNS development were identified, including 49 novel genes. To obtain information about the molecular nature of these genes, we decided to complement our genetic screens with a molecular screen. From transposon-tagged mutations identified on the basis of their phenotype in the PNS we selected 31 P-element strains representing 26 complementation groups on the second and third chromosomes to clone and sequence the corresponding genes. We used plasmid rescue to isolate and sequence 51 genomic fragments flanking the sites of these P-element insertions. Database searches using sequences derived from the ends of plasmid rescues allowed us to assign genes to one of four classes: (1) previously characterized genes (11), (2) first mutations in cloned genes (1), (3) P-element insertions in genes that were identified, but not characterized molecularly (1), and (4) novel genes (13). Here, we report the cloning, sequence, Northern analysis, and the embryonic expression pattern of candidate cDNAs for 10 genes: astray, chrowded, dalmatian, gluon, hoi-polloi, melted, pebble, skittles, sticky ch1, and vegetable. This study allows us to draw conclusions about the identity of proteins required for the development of the nervous system in Drosophila and provides an example of a molecular approach to characterize en masse transposon-tagged mutations identified in genetic screens.

THE peripheral nervous system (PNS) of Drosophila has been long used as an experimental paradigm to identify new genes and to further our understanding of the molecular mechanisms of neurogenesis (MODOLELL 1997 ; DAMBLY-CHAUDIERE and VERVOORT 1998 ; JAN and JAN 1998 ). Most known genes that affect PNS development were isolated serendipitously, since they affect easily identifiable morphological markers, namely bristle number in adults (JAN and JAN 1993 ). These genes are often nonessential or correspond to partial loss- or gain-of-function mutations in essential genes. Other players that are required for PNS development are essential genes that were isolated because loss of one gene copy causes a visible, but often unrelated haploinsufficient phenotype (e.g., Notch, Delta, and Enhancer of split; LINDSLEY and ZIMM 1992 ). Later, these genes were shown to affect embryonic neurogenesis when homozygous, and the functional analysis that followed their initial characterization gradually integrated them into developmental pathways of neurogenesis (JAN and JAN 1993 ).

A subset of PNS genes that remained largely unidentified until the late 1980s corresponds to those essential genes that do not cause a haploinsufficient phenotype when mutated. These genes were identified in genetic screens designed to isolate mutations that cause aberrant development of the embryonic PNS (SALZBERG et al. 1994 ; KOLODZIEJ et al. 1995 ; GAO et al. 1999 ). Effects of mutations in these genes are typically pleiotropic and do not affect PNS development only. Two classical examples include the daughterless (CAUDY et al. 1988 ) and numb (UEMURA et al. 1989 ) genes. However, because screening by immunohistochemical staining of fixed whole-mount embryos with monoclonal antibodies is quite tedious (JAN and JAN 1993 ), no such systematic screens were performed prior to 1992.

We set out to screen for genes that are essential and affect PNS development in embryos using chemical agents (SALZBERG et al. 1994 ) and P elements (KANIA et al. 1995 ; SALZBERG et al. 1997 ) as mutagens. Out of a total of 66 genes that affect PNS development, many genes were mapped and some were shown to be allelic to previously characterized genes on the basis of mapping information, similarity of phenotype, and complementation tests. However, numerous genes identified in P-element screens did not seem to correspond to known genes. The most direct approach to determine if these mutations correspond to novel genes and to establish the nature of mutations is to clone the genes adjacent to the P-element insertions. We selected 26 genes for cloning on the basis of several criteria. Combination of plasmid rescue, sequencing, and cDNA cloning gave us molecular information about the mapping position of P elements, their physical locations in the genome, the nature of mutations, the possible identity of encoded proteins, etc. Here, we report on the nature of these mutations and their adjacent genes. We demonstrate that 11 mutations correspond to known genes and report the cloning, sequence, and analysis of expression in the embryo of 10 novel genes.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

Stocks:
All stocks were maintained on a standard corn meal/agar medium (ASHBURNER 1989 ) at room temperature. P{lacZ,w⁺} P-element insertion lines used in this study derive from Istvan Kiss' collection of P elements on the second chromosome (TOROK et al. 1993 ) and from Peter Deák's collection on the third chromosome (SALZBERG et al. 1997 ), and were shown to be associated with phenotypes in the embryonic PNS (KANIA et al. 1995 ; SALZBERG et al. 1997 ). P-element insertion lines used in the screen are listed in Table 2.

Deficiencies and mutations used for complementation tests are listed in Table 1. Deficiencies and mutations were obtained from the Bloomington Stock Center, the Berkeley Drosophila Genome Project, and individual laboratories. Genetic nomenclature, gene names, and cytology are according to LINDSLEY and ZIMM 1992 and FlyBase (flybase.bio.indiana.edu; FLYBASE CONSORTIUM 1999 ).

In situ hybridization to polytene chromosomes:
Digoxigenin-labeled DNA probes were prepared using the DIG DNA labeling kit (Roche Molecular Biochemicals). Pretreatment and hybridization to polytene chromosomes were essentially as described (LANGER-SAFER et al. 1982 ). Following hybridization, probes were detected using an anti-digoxigenin antibody conjugated to alkaline phosphatase (Fab fragments, 1:200; Roche Molecular Biochemicals) and 4-nitroblue tetrazolium chloride with 5-bromo-4-chloro-3-indolyl-phosphate (Roche Molecular Biochemicals). The chromosomes were counterstained with Giemsa (Sigma, St. Louis) and mounted in Permount mounting medium (Fisher Scientific, Pittsburgh, PA).

In situ hybridization to whole-mount embryos:
In situ hybridization to whole-mount Canton-S embryos was carried out as described (TAUTZ and PFEIFLE 1989 ) using digoxigenin-labeled antisense riboprobes (DIG RNA labeling kit; Roche Molecular Biochemicals). To generate riboprobes by run-off transcription, the following combinations of restriction enzymes (to linearize the template plasmid DNA) and RNA polymerases were used: aay antisense probe (5B cDNA, XhoI, T3 polymerase), aay sense (5B cDNA, BamHI, T7 polymerase), dmt antisense (16A cDNA, XhoI, T3 polymerase), dmt sense (16A cDNA, XbaI, T7 polymerase), glu antisense (glu11 cDNA, NotI, T3 polymerase), glu sense (glu11 cDNA, HindIII, T7 polymerase), melt antisense (8G cDNA, XbaI, T7 polymerase or HL03627 cDNA, NotI, T7 polymerase), melt sense (8G cDNA, EcoRV, T3 polymerase or HL03627 cDNA, XhoI, T3 polymerase), stich1 antisense (GM05287 cDNA, XbaI, T7 polymerase), and stich1 sense probe (GM05287 cDNA, XhoI, T3 polymerase).

Molecular biology:
Genomic DNA isolation from Canton-S flies, poly(A)⁺ RNA isolation from 0- to 20-hr-old Canton-S embryos, Southern and Northern analyses, and screening of cDNA libraries were performed according to standard protocols (SAMBROOK et al. 1989 ).

Plasmid rescue:
Genomic sequences flanking the sites of P{lacZ,w⁺} P-element insertions were isolated by plasmid rescue (PIRROTTA 1986 ) using BamHI, XbaI, and PstI (for 5' sequences) and EcoRI and SacII (for 3' sequences) restriction enzymes and Epicurian coli XL1-Blue supercompetent cells (Stratagene, La Jolla, CA). The typical number of transformant colonies with 3 µg of starting genomic DNA and one-third of a ligation reaction used for transformation ranged from 1 to 50.

Several tests were performed on each plasmid rescued genomic fragment to determine if they correspond to novel genes and if they can be used as probes to screen cDNA libraries to clone the corresponding genes. They were (1) checked molecularly by restriction analysis (Table 3), (2) checked cytologically by in situ hybridization to polytene chromosomes (data not shown), (3) analyzed by sequencing (Table 3), and (4) checked on a Southern of Canton-S genomic DNA for the absence of repetitive DNA (data not shown).

For each rescue, at least three colonies were checked by DNA miniprep and restriction analyses. In rare cases, when all three colonies exhibited different digestion patterns, three more colonies were analyzed. The lengths of isolated genomic fragments ranged from 150 bp to 15 kb (see Table 3). In some cases we found upon double digestion (using as a second enzyme XbaI for EcoRI and SacII rescues and HindIII for BamHI, PstI, and XbaI rescues) that a plasmid did not carry a fragment corresponding to a P-element backbone (2 kb for EcoRI and SacII rescues and 10 kb for all other rescues). The presence of a new band that had a size larger or smaller than expected suggested that there were rearrangements of genomic DNA associated with the P-element insertion.

Cytological location of each fragment was verified by in situ hybridization to polytene chromosomes. If a mapping position of a fragment did not correspond to the mapping position of a P-element line used for plasmid rescue, it was discarded.

Based on digestion pattern, a representative plasmid was chosen for sequencing. A single sequencing run was performed (see below). The sequences were used to perform BLAST searches against nucleotide and protein sequence databases. The sequence information from plasmid rescues also provided an independent verification of mapping positions of genomic fragments. If a mapping position of a genomic clone (cosmid, bacterial artificial chromosome, or P1) hit by plasmid rescue-derived sequence was different from a P-element mapping position, the plasmid rescue was excluded from further analysis. Genomic fragments listed in Table 3 have mapping positions identical to P-element lines from which they were derived. BLAST searches also allowed us to determine the origin of genomic fragments for multiple insertion lines (e.g., l(2)k00424). The results of BLAST searches are presented in Table 3.

cDNA cloning:
Plasmid rescue-derived genomic fragments (both 5' and 3', if available) were used to screen cDNA libraries. We used an adult head EXLX M(-) cDNA library (BRUCE A. HAMILTON, personal communication) to isolate 31HC and 31HE clones and an embryonic (9–12 hr) gt11 cDNA library (ZINN et al. 1988 ) to isolate all other cDNA clones (Table 4). For each set of probes, at least 400,000 plaques were screened.

cDNA clones derived from a gt11 library were subcloned into pBluescript II KS(+) (clone glu11) or pBluescript II SK(+) (all other clones) (Stratagene). cDNA clones derived from a EXLX library were converted into pEXLX plasmid clones by Cre-loxP automatic subcloning in vivo (PALAZZOLO et al. 1990 ) using BM25.8 Cre-expressing strain (Novagen).

For some genes, putative cDNA clones were identified through database searches among Drosophila expressed sequence tags (ESTs; RUBIN et al. 2000A ). These are cloned in pBluescript SK(+/-) (clones GM05287, HL03627, and LD13852; Table 4) or in pOT2a (clones GH03082, GH23250, GM14315, and LD47384).

Sequencing:
To determine the terminal sequences of plasmid-rescued genomic fragments the following primers were used: P-ele-R (5'-CGACGGGACCACCTTATGTTATTTC-3') for proximal ends of all rescues; 703 (5'-CGAAAAGTGCCACCTGACGTC-3') for distal ends of EcoRI and SacII rescues; and 1706 (5'-GCCAGCAACGCAAGCTTCTAG-3') for distal ends of BamHI, XbaI, and PstI rescues. To determine full-length sequence of cDNA clones, we used nested deletions generated with an ExoIII/mung bean nuclease deletion kit (Stratagene) in combination with primer walking. Dye primer and dye terminator sequencing (BigDye cycle sequencing ready reaction kits; PE Applied Biosystems, Foster City, CA) was carried out on an ABI Prism 377 DNA sequencer (PE Applied Biosystems). Nucleotide sequences were assembled using an Auto-Assembler (PE Applied Biosystems). All sequences were annotated and deposited in GenBank prior to the end of 1999 (see Table 3 and Table 4 for accession numbers).

Biomolecular search and analysis tools:
Sequence similarity searches were performed using National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/BLAST/; WHEELER et al. 2000 ) and Berkeley Drosophila Genome Project (BDGP) (http://www.fruitfly.org/blast/) BLAST. Open reading frames (ORFs) were identified with NCBI ORF Finder (http://www.ncbi.nlm.nih.gov/gorf/) and codon preference was determined using SeqWeb (Wisconsin Package, Genetics Computer Group). To identify protein domains in predicted amino acid sequences, we used Motif (GenomeNet, Institute for Chemical Research, Kyoto University, Japan; http://www.motif.genome.ad.jp), SMART (EMBL, Heidelberg, Germany; http://smart.embl-heidelberg.de; SCHULTZ et al. 1998 , SCHULTZ et al. 2000 ), and ProfileScan (Swiss Institute for Experimental Cancer Research, Lausanne, Switzerland; http://www.isrec.isb-sib.ch/software/PFSCAN_form.html). Other biomolecular tools used were COILS (European Molecular Biology network—Swiss node, http://www.ch.embnet.org/software/COILS_form.html; LUPAS et al. 1991 ) to predict coiled-coil domains, SignalP (Center for Biological Sequence Analysis, The Technical University of Denmark, Lyngby, Denmark; http://www.cbs.dtu.dk/services/SignalP/; NIELSEN et al. 1997 ) to predict the presence and location of signal peptide cleavage sites, TMHMM (Center for Biological Sequence Analysis, The Technical University of Denmark, Lyngby, Denmark; http://www.cbs.dtu.dk/services/TMHMM-1.0/; SONNHAMMER et al. 1998 ) to predict transmembrane helices in proteins, PSORT (University of Tokyo, Tokyo, Japan, http://psort.nibb.ac.jp; NAKAI and HORTON 1999 ) to predict subcellular localization sites of proteins, and PESTfind (Pasteur Institute, Paris, France; http://bioweb.pasteur.fr/seqanal/interfaces/pestfind.html; RECHSTEINER and ROGERS 1996 ) to identify PEST regions.

	RESULTS AND DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

Rationale of the molecular screen:
To identify novel proteins required for the development of the peripheral nervous system, we decided to clone the affected genes identified in our forward genetic screens (KANIA et al. 1995 ; SALZBERG et al. 1997 ). The basis of the molecular screen (Fig 1) is the observation that P elements often insert in the 5' regions of genes (reviewed by BELLEN 1999 ). Therefore, genomic sequences flanking the sites of P-element insertions often provide information about the identity of genes affected by P elements. Typically, flanking genomic DNA is isolated by plasmid rescue, the 5'- and 3'-ends of rescued fragments are sequenced, and the sequences are used for BLAST searches. Three possible outcomes should be considered. First, P elements may affect known, previously identified genes. Indeed, P-element insertions are often partial loss-of-function mutations that cause mild phenotypes that are quite different from the phenotypes associated with severe loss-of-function alleles (e.g., emc^S009426 and other emc alleles, pbl^S054203; SALZBERG et al. 1997 ). Hence, relying on a similarity of phenotype with genes mapped to the region where the P element maps may not permit making an educated guess as to the identity of a gene. Thus, the molecular information obtained through plasmid rescue should greatly assist in the identification of affected genes. Second, BLAST searches may identify P-element insertions in genes that were cloned, but for which no mutations are available. Finally, P elements for which no significant matches were found in BLAST searches are good candidates for mutations affecting novel genes. Each presumably novel gene/mutation is then characterized using a combination of genetic (complementation tests), cytological (comparison of mapping positions of P elements, plasmid rescued genomic fragments, and candidate allelic genes), and molecular (positioning P-element insertions on the genomic sequence relative to neighboring genes, both known and predicted) approaches.

View larger version (22K): In this window In a new window Download PPT slide   Figure 1. Experimental design of the molecular screen. A general strategy for plasmid rescue using the P{lacZ,w+} P element inserted in the 5'-end of a gene transcription unit is shown. The genomic DNA and rescued genomic fragments are shown in red. For details, see MATERIALS AND METHODS and RESULTS. B, BamHI; E, EcoRI; G, BglII; P, PstI; S, SacII; X, XbaI.

Selection of P-element lines for the screen:
The screen is based on the assumption that the lethality caused by the insertion of a P element maps to the same molecular and cytological region as the P element itself. In other words, the P-element insertion itself, but not some other mutation on the chromosome, causes the observed phenotype. However, previous experience with P-element screens demonstrates that this is not always the case. P-element-encoded transposase is a mutagen (BEALL and RIO 1997 ) that may cause chromosomal breaks resulting in inversions, deletions, and other rearrangements elsewhere on a chromosome. In addition, 2-3 transposase activity may result in the introduction of multiple P elements on a chromosome. Indeed, the screen efficiency defined as the percentage of raw lines that contain a single insertion causing its associated phenotype can be as low as 31% (P-element lines from Kiss' collection; SPRADLING et al. 1999 ). Such considerations precluded us from selecting many lines for the molecular screen (see below).

We, therefore, established the following criteria to select P-element lines for the screen. First, P-element lines have to be single insertion lines and revertible. Second, if the P elements are not revertible, they should fail to complement deficiencies that on the basis of their breakpoints should lack the affected gene. Alternatively, they should fail to complement other independently isolated alleles from the same complementation group that map to the same cytological position as the P element. Third, occasionally a line with multiple insertions can be used to clone the gene. However, this was done only if we were able to demonstrate that only one P-element insertion is responsible for the lethality and phenotype.

A total of 72 novel P-element mutations representing 44 complementation groups were identified in our genetic screens (KANIA et al. 1995 ; SALZBERG et al. 1997 ). All these genes are essential, since P-element insertions cause embryonic, larval, or pupal lethality. Some of the novel genes have been already characterized by others and, hence, are not discussed here—barren (barr, BHAT et al. 1996 ), benchwarmer (bnch; A. KANIA and H. J. BELLEN, unpublished results), bunched (bun; short sighted, shs; TREISMAN et al. 1995 ; DOBENS et al. 1997 , DOBENS et al. 2000 ), gutfeeling (guf, SALZBERG et al. 1996 ; IVANOV et al. 1998 ), homothorax (hth; dorsotonals, dtl; RIECKHOF et al. 1997 ; PAI et al. 1998 ), pavarotti (pav, ADAMS et al. 1998 ), RpL30 (KANIA et al. 1995 ), sanpodo (spdo; DYE et al. 1998 ; SKEATH and DOE 1998 ), schnurri (shn; quo vadis, quo; NUSSLEIN-VOLHARD et al. 1984 ; ARORA et al. 1995 ; GRIEDER et al. 1995 ; STAEHLING-HAMPTON et al. 1995 ), and senseless (sens, NOLO et al. 2000 ). Among the remaining mutations we found 31 P-element strains representing 26 complementation groups (Table 2) that satisfy the required criteria. This number includes four lines selected during an initial stage of the project that were subsequently shown to be allelic to known genes—Star (S, KOLODKIN et al. 1994 ), escargot (esg, WHITELEY et al. 1992 ), extra macrochaetae (emc, ELLIS et al. 1990 ; GARRELL and MODOLELL 1990 ), and string (stg, EDGAR and O'FARRELL 1989 ; JIMENEZ et al. 1990 ) (see Table 2 and SALZBERG et al. 1997 ). We also selected three lines with multiple insertions—l(2)k00424, l(2)k05002, and l(2)k08104. Each of these lines carries two P elements; however, one of them is not responsible for the phenotype, since it either maps to another chromosome or complements deficiencies (Table 2).

The remaining lines described in KANIA et al. 1995 and SALZBERG et al. 1997 were not included in the screen, since they did not satisfy the established selection criteria. Some of them correspond to genes identified by a single nonrevertible P-element insertion, which complements deficiencies uncovering the region of the insertion [e.g., l(2)k05422, l(2)k06712, l(2)k08807, l(3)S136802, misn^S081603]. In other cases, P elements map to cytological intervals with few or no deficiencies [e.g., l(3)S052908]. In addition, absence of independent alleles generated in other P-element or chemical mutagenesis screens did not allow us to genetically map the genes and therefore we were unable to conclude if the P-element insertion is responsible for the phenotype.

Characterization of P-element insertions using flanking genomic DNA sequences:
We used plasmid rescue to recover genomic DNA flanking the insertion sites of 30 P-element lines selected for the screen. For many lines (19 out of 30) we were able to recover DNA flanking both 5'- and 3'-ends of P elements (Table 3). Analysis of genomic sequences flanking P-element insertions provided several types of information. First, a significant sequence match found between a plasmid rescue and cDNA sequence of a known gene demonstrated that this gene is likely to be affected by the P element. Second, availability of genomic sequence information allowed us to physically associate P elements and their flanking genomic fragments with specific sites in the genome. Furthermore, we were able to precisely position the P elements relative to neighboring genes. This information allowed us to make predictions about the identity of genes affected by P elements, about allelic relationships between previously characterized mutations and those identified in our screens, and about other genes adjacent to P elements and linked molecularly to the gene of interest. Results of plasmid rescue experiments including molecular (analysis by gel electrophoresis), sequence (GenBank accession numbers for sequences of ends of plasmid rescued fragments), database (BLASTN and BLASTX search results), and genomic (prediction of P-element locations relative to neighboring genes) analyses are presented in Table 3.

Four classes of genes identified in the screen:
Sequence information derived from plasmid rescues allowed us to assign all genes to one of four classes: (1) previously characterized genes (11 genes), (2) first mutations in cloned genes (1 gene), (3) P-element insertions in genes that were phenotypically characterized, but not identified (1 gene), and (4) novel genes (13 genes).

Previously characterized genes:
Our initial analysis of mutations relied solely on the molecular information derived from genomic DNA flanking the sites of P-element insertions. Using this approach we identified 11 previously characterized genes. We provide both molecular and genetic evidence establishing new allelic relationships between 10 existing mutations (Table 3 and Table 5). We found that Cyclin E (CycE, RICHARDSON et al. 1993 ; KNOBLICH et al. 1994 ) is allelic to fondue (fond, KANIA et al. 1995 ), mindmelt (mm, KANIA et al. 1995 ) is allelic to muscleblind (mbl, BEGEMANN et al. 1997 ; ARTERO et al. 1998 ), patched (ptc, STURTEVANT 1948 ; NUSSLEIN-VOLHARD et al. 1984 ; HOOPER and SCOTT 1989 ; NAKANO et al. 1989 ) is allelic to rubberneck (rubr, KANIA et al. 1995 ), puckered (puc, RING and MARTINEZ ARIAS 1993 ; MARTIN-BLANCO et al. 1998 ) is allelic to hearty (hrt, SALZBERG et al. 1994 , SALZBERG et al. 1997 ), and three rows (thr, NUSSLEIN-VOLHARD et al. 1984 ; D'ANDREA et al. 1993 ; PHILP et al. 1993 ) is allelic to anarchist (anch, KANIA et al. 1995 ). CycE, mbl, and thr are directly affected by P-element insertions fond^k02514, mm^k07103, and anch^k07805b, respectively. In contrast, the rubr^k02507 P element is inserted in the first intron of ptc and the hrt^S023803 P element is inserted in the intron between exons 3 and 4 of puc. Since these introns do not contain any known or predicted genes (ADAMS et al. 2000 ), we considered the possibility that ptc is allelic to rubr and puc is allelic to hrt. In all five cases we were able to demonstrate by complementation tests that the mutations are indeed allelic (Table 2). For example, fond^k05002 insertion complements CycE mutation, but fails to complement another P-element allele, fond^k02514. fond^k02514/CycE⁰⁵²⁰⁶ flies are viable, but adult escapers have rough eyes and wing venation defects indicating that the two mutations are allelic. Similarly, mm^k07103 complements two hypomorphic alleles of mbl (Table 2), but fails to complement mbl^E27, a putative null allele affecting the coding sequence of the gene. There are few mm^k07103/mbl^E27 adult escapers that have wing blisters, wing venation defects, or unexpanded wings. Similarly, hrt^S023803 insertion complements puc¹, but fails to complement puc^A251.1F3 (Table 2), a P-element insertion that affects the same intron as the hrt^S023803 P element. In conclusion, results of complementation tests combined with the sequence data provide strong evidence that the mutations are correctly assigned as allelic to previously identified genes.

Previously, we and others have shown that the l(2)k06921, l(2)k08104, l(3)S024532, and l(3)S058701 mutations are allelic to known genes on the basis of complementation data (BERKELEY DROSOPHILA GENOME PROJECT, unpublished results; SALZBERG et al. 1997 ). Our plasmid rescue and sequence data confirm this allelism for the l(2)k08104 and l(3)S058701 P elements, which are allelic to escargot (esg, WHITELEY et al. 1992 ) and extra macrochaetae (emc, ELLIS et al. 1990 ; GARRELL and MODOLELL 1990 ), respectively, and are inserted in the 5'-untranslated regions (5'-UTRs) of the genes (Table 3). The l(3)S024532 P element is allelic to string (stg, EDGAR and O'FARRELL 1989 ; JIMENEZ et al. 1990 ) and is inserted in the stg gene (Table 2 and Table 3). The l(2)k06921 P-element insertion was shown (BERKELEY DROSOPHILA GENOME PROJECT, unpublished results) to be allelic to Star (S; floater, fltr; KOLODKIN et al. 1994 ) and is inserted 1060 nucleotides (nt) upstream of the S AUG on the (-) strand and 275 nt upstream of the asteroid (ast, HIGSON et al. 1993 ; KOTARSKI et al. 1998 ) AUG on the (+) strand (Table 3) and, therefore, may affect both genes.

cyrano: We and others (BYARS et al. 1999 ) found that cyrano (cyr) is allelic to raw (raw, NUSSLEIN-VOLHARD et al. 1984 ). raw is a dorsal-open group gene required for the regulation of Jun N-terminal kinase (JNK) signaling during dorsal closure that encodes a novel protein of 989 amino acids (aa). We used genomic sequences derived from cyr^k01021 to independently identify through database searches the GH23250 cDNA (RUBIN et al. 2000A ) as a candidate clone for raw/cyr (Table 4). It encodes a new smaller isoform (805 aa) of the RAW protein (Fig 2) generated by alternative splicing and the use of an upstream initiation methionine (BYARS et al. 1999 ). Interestingly, we were unable to align with each other any of the six plasmid rescue sequences derived from cyr^k01021 and cyr^k08801 P elements, suggesting that the two P elements may be far apart. Database analysis revealed that the distance between the two insertions is at least 22 kb. The cyr^k08801 P element is inserted in an intron of the raw gene, and cyr^k01021 is inserted 5 kb upstream of the raw AUG (GenBank accession no. AF186024, BYARS et al. 1999 ), but within the coding sequence of the alternatively spliced GH23250 cDNA.

View larger version (16K): In this window In a new window Download PPT slide   Figure 2. Domain structure of novel proteins. Proteins are represented schematically with predicted functional domains and motifs shown in color. Abbreviated names of proteins are indicated on the left, and their lengths (in aa) are indicated on the right. Length of a protein followed by a plus sign indicates that the coding sequence of a cDNA clone is incomplete at the 3'-end. MELT protein does not have any known functional domains. Proteins and the respective domains are drawn to scale. For exact location of domains and motifs refer to respective GenBank accession numbers (Table 4).

unchained: The molecular and genetic analyses of unchained (unch, KANIA et al. 1995 ) suggest that it is allelic to Sin3A (NEUFELD et al. 1998 ; PENNETTA and PAULI 1998 ). Mouse mSin3 is a transcriptional corepressor that forms a ternary complex with the Mad and Max basic helix-loop-helix leucine zipper proteins, recruits other corepressor proteins, and downregulates Myc target genes (AYER et al. 1995 ; SCHREIBER-AGUS and DEPINHO 1998 ). Drosophila embryos homozygous for Sin3A null mutation fail to hatch, but have no obvious defects in muscle or nervous system development (PENNETTA and PAULI 1998 ). We found (Table 3 and data not shown) that the unch^k15501 P element is inserted 24 nt upstream of the the 5'-end of the longest Sin3A mRNA (GenBank accession no. AF024603, NEUFELD et al. 1998 ) or 131 nt upstream of the beginning of the alternatively spliced Sin3A mRNA (GenBank accession no. AJ007518, PENNETTA and PAULI 1998 ). Furthermore, database searches using the B25 genomic fragment flanking unch^k15501 identified the LD13852 clone (RUBIN et al. 2000A ) as a putative unch cDNA (Table 4). It maps to the unch locus (Table 4) and is likely to be affected by the unch^k15501 insertion, which is located 33 nt upstream of its 5'-end (data not shown). The unch^k15501 insertion, although not revertible, fails to complement deficiencies and is therefore likely to be responsible for the unch phenotype (KANIA et al. 1995 ). On the basis of these observations we concluded that unch may be allelic to Sin3A. However, complementation tests between unch^k15501 and several Sin3A alleles (including the Sin3A^ex4 null allele, PENNETTA and PAULI 1998 ) showed that they complement each other (Table 2). Note that the Sin3A⁰⁸²⁶⁹, Sin3A^k05415, and Sin3A^k11222 P elements are inserted in the first intron of unch (PENNETTA and PAULI 1998 ; data not shown), whereas the unch^k15501 P element is inserted 5 kb upstream, in the very beginning of the unch transcription unit (see above). Therefore, the peculiar complementation data may be the result of intragenic complementation, and unch may indeed be allelic to Sin3A. Further phenotypic and genetic analyses will be required to resolve this matter.

The above issues are further complicated as we cannot exclude that the unch^k15501 insertion may affect the amphiphysin gene (RAZZAQ et al. 2000 ), which is transcribed from a complementary strand in the orientation opposite to Sin3A. The unch^k15501 P element is inserted 0.8 kb upstream of the 5'-end of amphiphysin mRNA (Table 3) and may affect the amphiphysin promoter or enhancer elements.

First mutations in cloned genes:
We identified only one P element affecting a gene for which there were no mutations available. Calreticulin (Crc) is the gene mutated by the potp^S114307 P-element insertion. The potp^S114307 P element affects the Crc gene (Table 3), since it is inserted in the 5'-UTR of Crc mRNA, 28 nt upstream of initiator methionine.

Vertebrate calreticulins are major Ca²⁺-binding proteins of endoplasmic reticulum implicated in the regulation of intracellular Ca²⁺ signaling, glycoprotein folding, integrin-mediated cell adhesion, and steroid-dependent gene expression (COPPOLINO et al. 1997 ; KRAUSE and MICHALAK 1997 ). The Drosophila Crc gene (SMITH 1992 ) was previously shown to map to 86B-C (GAMO et al. 1998 ) or 85E1-5 (CHRISTODOULOU et al. 1997 ). The potp^S114307 P element was mapped to 85E (SALZBERG et al. 1997 ). To resolve the ambiguity of the mapping of Crc, we identified through database searches the 1.5-kb LD07621 cDNA clone (RUBIN et al. 2000A ) corresponding to Crc and mapped it to 85E1 (data not shown). These observations, together with the ability to revert the lethality by a precise excision of potp^S114307 (SALZBERG et al. 1997 ), indicate that it is indeed a mutation in the Crc gene.

A Crc mutation was reported to cause hypersensitivity of flies to diethylether anesthesia (Crc^eth-as311, GAMO et al. 1998 ). Our data show that Crc is an essential gene, since the potp^S114307 insertion is homozygous lethal and results in loss of neurons, decreased staining with the MAb 22C10 in the central nervous system (CNS), disorganization of the PNS, and pathfinding defects during embryonic development (SALZBERG et al. 1997 ).

P-element insertions in genes that were identified, but not characterized molecularly:
This class of genes includes those P-element insertions that may serve as cloning tools for previously identified mutations that are not transposon tagged. The pebble (pbl) gene was identified in a chemical mutagenesis screen for mutations affecting the pattern of the embryonic cuticle (JURGENS et al. 1984 ) and is required for cytokinesis during postblastoderm mitoses (HIME and SAINT 1992 ; LEHNER 1992 ). We identified two alleles of pbl in our P-element screen (SALZBERG et al. 1997 ), cloned the gene, and showed that it encodes a putative guanine nucleotide exchange factor for Rho1 G protein (RhoGEF, PROKOPENKO et al. 1999 ).

Novel genes:
We identified 13 novel genes and cloned or identified candidate cDNAs for 10 genes. The identity of the respective proteins, their domain structure, and RNA expression are described in the following sections. The information on cDNA clones including their names, lengths, mapping positions, GenBank accession numbers, and results of Northern analysis and sequence analysis is presented in Table 4 and summarized in Table 5. Cloning and functional characterization of bonus (bon) will be published elsewhere (R. B. BECKSTEAD, S. N. PROKOPENKO and H. J. BELLEN, unpublished results). The identity of three remaining genes remains unknown.

l(2)k00424: The l(2)k00424 strain carries two P-element insertions that were mapped at 30D1-2 and 44F1-2. However, only one insertion (at 44F1-2) is responsible for the lethality and possibly the organizational defects observed in the dorsal cluster of neurons in the PNS (KANIA et al. 1995 ) given the complementation data with deficiencies (Table 2). The genomic sequences flanking this insertion did not show any homologies in database searches. Hence, the l(2)k00424 gene remains unidentified. The second genomic fragment isolated from this strain maps at 30D1-2. This P element is inserted in the 5'-UTR, 21 nt upstream of the initiator methionine of the FKBP59-bp1 gene, which encodes the FK506-binding protein FKBP59, a member of the immunophilin family of proteins (ZAFFRAN 2000 ).

bumper-to-bumper: The bumper-to-bumper (btb) gene was identified by a single revertible P-element insertion that leads to pathfinding and connectivity defects and affects dorsoventral migration of lateral chordotonal neurons (KANIA et al. 1995 ). The P-element insertion in btb^k09901 may affect a predicted gene, CG5380, which encodes a DNA-directed RNA polymerase III. Genomic sequences flanking the 5'-end of the P element show homology to RNA polymerase III subunits from several species (Table 3). This P-element insertion is revertible (KANIA et al. 1995 ) and is located 418 nt upstream of CG5380 AUG. Both P element and a plasmid rescued genomic fragment were mapped to 47A-47B14.

on-the-rack: on-the-rack (rack) was identified by a single revertible P-element insertion that causes loss of LCh5 neurons and affects morphology of neurons in the lateral cluster (KANIA et al. 1995 ). Genomic sequence flanking rack^k15001 P-element insertion did not give any significant matches in database searches (Table 3). Our attempts to clone rack were unsuccessful, because the rack^k15001 P-element insertion is associated with DNA rearrangements affecting the P element (Table 3).

astray: astray (aay) was identified by a single revertible P-element insertion (aay^S042314) that causes severe defects in the axonal trajectories in the embryonic PNS (SALZBERG et al. 1997 ). We cloned the full-length aay cDNA, which encodes a 3-phosphoserine phosphatase (Table 4). L-3-phosphoserine phosphatase (PSPase) catalyzes the last rate-limiting step in the biosynthesis of serine—Mg²⁺-dependent hydrolysis of L-phosphoserine to serine as well as an exchange reaction between L-serine and L-phosphoserine. The AAY protein is homologous to PSPases from mammals, plants, yeast, and bacteria (Table 4 and data not shown) and is most similar to human phosphoserine phosphatase (COLLET et al. 1997 ). AAY and human PSPase belong to a new class of phosphotransferases (COLLET et al. 1998 ) characterized by a conserved N-terminal DXDX(T/V) motif (aa 67–71 in AAY, Fig 2). The first aspartate in this motif is absolutely conserved among all proteins and has been implicated in covalent binding of phosphate and formation of a phosphoenzyme catalytic intermediate (COLLET et al. 1999 ). The two proteins share two other highly conserved motifs (in AAY, aa 155–159 and 203–207, 225–232, Fig 2). Conserved residues within these motifs play an important role in catalysis, as demonstrated by site-directed mutagenesis of other phosphotransferases (P-type ATPases and human PSPase, LINGREL and KUNTZWEILER 1994 ; COLLET et al. 1999 ), and are likely to form a catalytic pocket, as shown by 3-D structure analysis of Pseudomonas haloacid dehalogenase (LI et al. 1998 ).

During embryonic development, ASTRAY is expressed in a complex pattern (Fig 3, A–E). During stage 5 (Fig 3A), ASTRAY is expressed in a highly specific pattern consisting of 7 stripes capped on the dorsal side by a longitudinal stripe. It is also expressed abundantly in the area surrounding the pole cells and the invagination in which the pole cells migrate (Fig 3A and Fig B, arrows). The expression during germ band extension is characterized first by 7 broad stripes (Fig 3B) and later by 10 stripes that eventually fuse to form a peculiar pattern (data not shown). This expression then fades and gives rise to a pattern of segmentally repeated small clusters of cells (Fig 3C, arrowheads), a ring of large cells around the anterior gut (Fig 3, C–E, arrows), and low levels of expression in most of the gut in more mature embryos (Fig 3E).

View larger version (75K): In this window In a new window Download PPT slide   Figure 3. Expression of novel genes during embryonic development. Wild-type embryos were hybridized with aay (A–E), dmt (F–I), glu (J–N), melt (O–S), and stich1 (T–W) antisense RNA probes. Sense probes used in parallel in situ hybridization experiments gave either no staining or a low level of background staining (data not shown). Stages of embryonic development are indicated in each lower right corner. For details on generating RNA probes see MATERIALS AND METHODS.

How does a mutation in PSPase (aay^S042314 P element is inserted in the 5'-UTR of aay) lead to the axon guidance phenotype observed in the PNS? Serine is used not only as a building block for protein synthesis but also as a precursor of phospholipids (phosphatidylserine and sphingomyelin) and glycolipids. Loss of AAY may cause abnormalities in membrane biogenesis in specific cells that would affect transmembrane signaling in neuronal growth cones. Clearly, other alternative hypotheses are possible. It is interesting to note that L-serine does not cross the blood-brain barrier well (SMITH et al. 1987 ) and that defects in the serine biosynthesis pathway leading to low serine levels in cerebrospinal fluid have been described in patients with Williams syndrome (3-phosphoserine phosphatase deficiency, JAEKEN et al. 1997 ) and congenital encephalopathy (3-phosphoglycerate dehydrogenase deficiency, JAEKEN et al. 1996 ).

chrowded: chrowded (chrw) was identified by a single revertible P-element insertion (chrw^k06908) that causes organizational and morphological defects in the PNS (KANIA et al. 1995 ). We identified the LD47384 clone (RUBIN et al. 2000A ) as a candidate chrw cDNA. It encodes a Rab-related small G protein (Fig 2) that is identical to the ORF of the predicted gene CG3870 (RUBIN et al. 2000B ). The chrw^k06908 P element does not directly affect the LD47384 cDNA, but is inserted in close proximity to its 5'-end (Table 3).

CHRW is distantly related to a number of Rab proteins from Drosophila, mammals, plants, and yeast (Table 4 and data not shown), but has a longer ORF (261 aa compared to 200–220 aa in most Rab proteins). Rab proteins compose a separate family within the Ras superfamily that consists of >30 members (in mammals, Rab1–25 and others) implicated in different aspects of intracellular vesicular trafficking (reviewed by ZERIAL and HUBER 1995 ). Rab proteins that share at least 80% identity are included in the same subfamily (e.g., Rab1, Rab2, etc.). In contrast, CHRW shares not more than 40–42% identity (51–54% similarity) with known Rab proteins, including several Rab proteins cloned in Drosophila (SASAMURA et al. 1997 ; SATOH et al. 1997 ). Therefore, we propose that CHRW is a Rab-related protein. Alternatively, it may represent the first member of an unidentified family of Rab proteins. Like other small G proteins, CHRW contains four conserved regions that form the GTP-binding and GTPase catalytic site and the C-terminal CCX cysteine motif as well as the Rab-specific effector region. It is possible that CHRW does not function as a Rab G protein, since it lacks two conserved amino acids present in Rab proteins—G41 in the effector region (A in CHRW) and A151 in region IV (T in CHRW). Further genetic and molecular analyses are necessary to demonstrate that CHRW is indeed a Rab-related protein and that the chrw^k06908 P element affects the Rab gene identified with the LD47384 cDNA.

dalmatian: We identified dalmatian (dmt) in our chemical mutagenesis screen as a mutation that leads to a loss of neurons, disorganization of the PNS, and formation in the ectoderm of small round cells that stain darkly with MAb 22C10 (SALZBERG et al. 1994 ). We used a revertible P-element allele of dmt, dmt^S048103 (SALZBERG et al. 1997 ), to clone the full-length cDNA (Table 4). This cDNA is likely to correspond to the dmt gene, since the dmt^S048103 P element is inserted 11 nt upstream of its 5'-end. dmt encodes a novel protein that is not related to any known or predicted proteins from other species. The only conspicuous feature of DMT is the presence of four nuclear localization sequences (Fig 2). In addition, the PSORT algorithm (NAKAI and HORTON 1999 ) predicts a 78% probability of DMT being a nuclear protein. Analysis of expression pattern by in situ hybridization revealed that at cellular blastoderm, DALMATIAN is expressed at low levels (data not shown). Expression levels increase during germ band extension and the gene is widely expressed at stage 11 (Fig 3F). From stage 12 (Fig 3G), expression becomes more restricted to the PNS and the CNS—ventral nerve cord (Fig 3G and Fig I, arrowheads) and brain (Fig 3, G–I, asterisks). In addition, DALMATIAN is expressed in the pole cells (data not shown), the gut (Fig 3H, arrowheads), and the posterior spiracles (Fig 3H, arrow). In summary, dmt is expressed in numerous tissues including the CNS throughout most of embryonic development and encodes a novel presumably nuclear protein of an unknown function.

gluon: gluon (glu) was identified by a single nonrevertible P-element insertion (glu^k08819) that fails to complement a deficiency uncovering the region where the transposon is inserted (KANIA et al. 1995 ). The embryonic phenotype of glu is characterized by subtle organizational defects in ventral and lateral PNS resulting in irregularly shaped clusters. Sequencing of the genomic fragments flanking glu^k08819 revealed that the P element affects a gene related to SMC (structural maintenance of chromosomes) proteins (Table 3), which are required for chromosome condensation during mitosis (reviewed by STRUNNIKOV 1998 ; HIRANO 1999 ). The nearly full-length glu cDNA encodes a predicted ORF of 1409 amino acids related to members of a highly conserved family of SMC proteins (Table 4)—Xenopus XCAP-C (HIRANO and MITCHISON 1994 ), human CAP-C (NISHIWAKI et al. 1999 ), Saccharomyces cerevisiae Smc4p (KOSHLAND and STRUNNIKOV 1996 ), and Schizosaccharomyces pombe cut3p (SAKA et al. 1994 ). Therefore, GLU is a Drosophila XCAP-C/Smc4-like protein. Similar to other Smc4p-like proteins GLU contains an N-terminal nucleotide-binding motif, two central coiled-coil domains, and a C-terminal DA-box that has DNA-binding capability (Fig 2). The presence of a nuclear localization sequence and PSORT algorithm prediction (91% probability) suggest that GLU is a nuclear protein as demonstrated for XCAP-C (HIRANO and MITCHISON 1994 ) and cut3p (SAKA et al. 1994 ; SUTANI et al. 1999 ).

GLU is likely to be a component of the 13S condensin complex described in Xenopus (HIRANO et al. 1997 ) and S. pombe (SUTANI et al. 1999 ). This complex consists of two SMC subunits (in Xenopus, XCAP-E and XCAP-C) and three non-SMC subunits (in Xenopus, XCAP-D2, XCAP-G, and XCAP-H). 13S condensin as well as its individual components is absolutely required for mitotic chromosome condensation in vitro and in vivo (reviewed by STRUNNIKOV 1998 ; HIRANO 1999 ). Two other components of Drosophila condensin have been identified—Barren (XCAP-H homolog), which is required for sister chromatid segregation during mitosis (BHAT et al. 1996 ), and dSMC2 (XCAP-E homolog, HIRANO 1999 ).

glu has a typical "mitotic" expression pattern similar to other genes implicated in cell cycle regulation or mitosis (e.g., stg, CycA, and barr; EDGAR and O'FARRELL 1989 ; LEHNER and O'FARRELL 1989 , LEHNER and O'FARRELL 1990 ; BHAT et al. 1996 ). Prior to cellularization (Fig 3J), there is maternally provided GLUON RNA that is enriched at the posterior end of the embryo (Fig 3J, arrow), where the pole cells form during telophase of mitotic cycle 10 (stage 4). During germ band extension, there are high levels of GLUON in dividing neuroblasts in the PNS and the CNS (Fig 3K). GLUON expression in the brain (Fig 3, K–N, asterisks) and the ventral nerve cord (Fig 3L and Fig M, arrowheads) persists throughout embryogenesis. By the end of stage 16 (Fig 3N), when most of the embryonic cells have stopped dividing, GLUON RNA expression in most tissues is much lower than at earlier embryonic stages. However, GLUON continues to be expressed at elevated levels in tissues that will resume proliferation during larval development—in neuroblasts in the brain (Fig 3N, asterisk) and in gonads (Fig 3N, arrow). Hence, the expression pattern of GLUON is very similar to BARREN, which encodes another condensin subunit (BHAT et al. 1996 ), but its in vivo role remains to be determined.

hoi-polloi: hoi-polloi (hoip) was identified by a single nonrevertible P-element insertion (hoip^k07104) that fails to complement a deficiency uncovering the region where the transposon is inserted (KANIA et al. 1995 ). hoip^k07104 causes subtle fasciculation and organization defects characterized by misplaced cells within few neuronal clusters. Database searches using genomic sequences flanking the site of hoip^k07104 insertion identified HOIP as a member of a conserved family of YEL026W-like proteins (Table 3). We identified the GH03082 clone (RUBIN et al. 2000A ) as a candidate near full-length hoip cDNA (Table 4). It encodes a 127-aa protein related to human non-histone chromosome protein 2-like 1 (NHP2L1) protein (SAITO et al. 1996 ), Caenorhabditis elegans YEL026W (GenBank accession no. Q21568), and S. cerevisiae Snu13p (GOTTSCHALK et al. 1999 ). These proteins compose an evolutionary conserved family of YEL026W-like proteins found in phyla from plants to humans and distantly related to families of NHP2-like proteins and ribosomal L7Ae proteins (see Table 4 and MAIORANO et al. 1999 ). We propose that HOIP is likely to be a functional homolog of these proteins, since it is very similar to human and C. elegans proteins (79 and 74% identity, respectively; Table 4). HOIP contains a central region corresponding to the ribosomal L7Ae signature (aa 72–89, Motif prediction, see Fig 2) that is highly conserved (83–100% identical) among family members. Human NHP2L1 was shown to bind directly to the 5' stem-loop of U4 small nuclear RNA and is an essential component of a spliceosome (GOTTSCHALK et al. 1999 ). On the basis of these observations we hypothesize that HOIP is an RNA-binding protein, component of a spliceosome, and is required for pre-RNA splicing.

melted: The melted (melt) gene was identified by a single revertible P-element insertion (melt^S144114) that results in abnormal morphology and mild loss of peripheral neurons (SALZBERG et al. 1997 ). We used plasmid-rescued fragments to clone two partially overlapping melt cDNAs (8C and 8G, see Table 4). In addition, through database searches we identified the HL03627 clone (RUBIN et al. 2000A ) as a candidate melt cDNA. We used the full-length sequence of the three clones to assemble a 2.903-kb melt cDNA that is incomplete at the 3'-end. BLAST searches with the sequence of MELT gave no significant results, except a limited homology to a predicted protein from C. elegans (Table 4). In addition, MELT does not contain any functional domains or motifs. Early in embryogenesis (stage 5, Fig 3O), MELTED RNA is expressed in 8 or 9 stripes and in the invaginating ventral furrow (Fig 3O, arrow). During germ band extension, MELTED is expressed in discrete domains in each segment of the embryo (Fig 3P, arrowheads). Later, this pattern is refined to several rows of ectodermal cells in the anterior of each segment (Fig 3, Q–S, arrowheads). There are also low levels of expression in the brain and the gut (data not shown). In conclusion, MELT is a novel protein of unknown function, which, on the basis of its expression pattern, may be required for ectodermal patterning.

skittles: The skittles (sktl) cDNA was isolated in an attempt to clone the fata morgana (fam) gene. fam was identified by several P-element alleles that result in morphological defects of lateral and v' chordotonal neurons in the PNS (KANIA et al. 1995 ). We and others have shown (KNIRR et al. 1997A ; HASSAN et al. 1998 ) that sktl is nested in the first intron of inscuteable (insc; not enough muscles, nem; BURCHARD et al. 1995 ; KRAUT and CAMPOS-ORTEGA 1996 ). fam P-element insertions (e.g., fam^k07505) fail to complement independently generated alleles of insc (Table 2) and affect two genes (KNIRR et al. 1997A ; HASSAN et al. 1998 ). Therefore, the name fata morgana does not refer to either of the two genes, but rather describes the composite phenotype caused by the loss of both. The isolated sktl cDNA (Table 4) encodes a putative phosphatidylinositol-4-phosphate 5-kinase that is longer than the published sequence of SKTL protein (KNIRR et al. 1997B ). SKTL contains a nuclear localization sequence (aa 577–593) and a PEST region (aa 49–78) and has a 78% probability to localize to the nucleus (PSORT prediction). Expression pattern and genetic and functional analyses of sktl have been published elsewhere (KNIRR et al. 1997A ; HASSAN et al. 1998 ).

sticky ch1: The sticky ch1 (stich1) gene was originally identified by a single EMS-induced allele (stich1^D233) as a mutation that affects morphology of neurons (SALZBERG et al. 1994 ). Later, we identified an independently isolated P-element allele of stich1, stich1^S143702 (SALZBERG et al. 1997 ), and showed that the stich1^S143702 insertion is revertible (this study, Table 2). We used one of the genomic sequences flanking the site of the stich1^S143702 insertion to identify through database searches the GM05287 clone (RUBIN et al. 2000A ) as a putative stich1 cDNA. It encodes a predicted basic helix-loop-helix (bHLH) protein similar to Hairy and Enhancer of split-related transcriptional repressors (Table 4, reviewed in FISHER and CAUDY 1998 ). Common functional domains shared among these proteins are the bHLH domain and the adjacent orange domain (Fig 2), which is thought to confer functional specificity among Hairy-related proteins (DAWSON et al. 1995 ). The bHLH domain is most closely related to that of Drosophila HEY, mouse Hey1 (LEIMEISTER et al. 1999 ), and mouse HRT2 (NAKAGAWA et al. 1999 , see Table 4). The basic region is most closely related to that of rat SHARP-1 (ROSSNER et al. 1997 ), human DEC1 (SHEN et al. 1997 ), and mouse Stra13 (BOUDJELAL et al. 1997 ), but the position of a conserved proline residue is shifted in STICH1 2 amino acids toward the N terminus, suggesting that it may have a different DNA-binding specificity. Since the GM05287 cDNA is incomplete at the 3'-end, we do not know if it contains a C-terminal WRPW motif found in all Hairy family proteins and required for interaction with Drosophila non-HLH corepressor protein Groucho (reviewed in FISHER and CAUDY 1998 ). Finally, the predicted protein is much longer (at least 610 aa) than other Hairy and Enhancer of split-related proteins (250–400 aa) and, therefore, it may not be their functional homolog.

The gene has a complex expression pattern during embryonic development (Fig 3, T–W). During stage 8, the RNA is expressed in the anterior and posterior midgut primordia (Fig 3T, asterisks). Expression in the gut continues throughout embryonic development (Fig 3U; hindgut in Fig 3V and Fig W, arrows). During germ band retraction, expression is initiated in many tissues in a prominent segmentally repeated pattern (Fig 3U, arrowheads). L