white⁺ Transgene Insertions Presenting a Dorsal/Ventral Pattern Define a Single Cluster of Homeobox Genes That Is Silenced by the Polycomb-group Proteins in Drosophila melanogaster

Corresponding author: Dario Coen, Embryologie Moléculaire et Expérimentale, Université Paris Sud, Bâtiment 445, 91405 Orsay Cedex, France, dario.coen{at}emex.u-psud.fr (E-mail).

Communicating editor: V. G. FINNERTY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We used the white gene as an enhancer trap and reporter of chromatin structure. We collected white⁺ transgene insertions presenting a peculiar pigmentation pattern in the eye: white expression is restricted to the dorsal half of the eye, with a clear-cut dorsal/ventral (D/V) border. This D/V pattern is stable and heritable, indicating that phenotypic expression of the white reporter reflects positional information in the developing eye. Localization of these transgenes led us to identify a unique genomic region encompassing 140 kb in 69D1–3 subject to this D/V effect. This region contains at least three closely related homeobox-containing genes that are constituents of the iroquois complex (IRO-C). IRO-C genes are coordinately regulated and implicated in similar developmental processes. Expression of these genes in the eye is regulated by the products of the Polycomb-group (Pc-G) and trithorax-group (trx-G) genes but is not modified by classical modifiers of position-effect variegation. Our results, together with the report of a Pc-G binding site in 69D, suggest that we have identified a novel cluster of target genes for the Pc-G and trx-G products. We thus propose that ventral silencing of the whole IRO-C in the eye occurs at the level of chromatin structure in a manner similar to that of the homeotic gene complexes, perhaps by local compaction of the region into a heterochromatin-like structure involving the Pc-G products.

THE product of the white (w) gene is necessary for the deposition of pigments in the compound eye of Drosophila melanogaster. The expression of white is extremely sensitive to position effects, which can be observed when the gene is relocalized by germ line transformation as a P[w⁺] (HAZELRIGG et al. 1984 ; LEVIS et al. 1985 ) or a P[mini-white⁺] transgene (PIRROTTA 1988 ).

Three types of effects are observed with white⁺ derivative transgenes. Most frequent is the homogeneous reduction of pigmentation throughout the entire eye (LEVIS et al. 1985 ; PIRROTTA 1988 ). Less frequently, pigmentation is randomly reduced or absent in certain ommatidia (variegated or unstable position effect; HENIKOFF 1994 ). In rare cases, the reduced or absent pigmentation in certain ommatidia is distributed according to a reproducible pattern.

The variegated pigmentation patterns are usually observed when P[w⁺] transgenes are inserted in the proximity of heterochromatic loci (HAZELRIGG et al. 1984 ; LEVIS et al. 1985 ; WALLRATH and ELGIN 1995 ), resulting in position-effect variegation (PEV; reviewed in HENIKOFF 1990 ; REUTER and SPIERER 1992 ; SINGH 1994 ). These patterns are not heritable, as neither siblings derived from the same parents nor the two compound eyes of the same individual display identical mosaic patterns. P[w⁺] transgenes harboring cis-regulatory sequences of Polycomb-group (Pc-G) target genes, including Polycomb responsive elements (PREs; SIMON et al. 1993 ), have been shown to frequently exhibit a similar variegation, although inserted at euchromatic sites (FAUVARQUE and DURA 1993 ; CHAN et al. 1994 ; GINDHART and KAUFMAN 1995 ; ZINK and PARO 1995 ). This variegated phenotype is also distinct from classical PEV, as it is modified neither by most genetic modifiers of PEV nor by the removal of the Y chromosome. However, this new kind of euchromatic variegation was shown to be sensitive to the dosage of Pc-G and trithorax-group (trx-G) gene products (FAUVARQUE and DURA 1993 ; CHAN et al. 1994 ; GINDHART and KAUFMAN 1995 ; ZINK and PARO 1995 ) and was therefore called developmental-regulator effect variegation (DREV; FAUVARQUE and DURA 1993 ).

Pc-G genes have been genetically isolated as a class of negative trans-regulators responsible for the maintenance —but not the initiation—of homeotic gene repression (for reviews see PARO 1993 ; KENNISON 1995 ). The PC-G proteins act synergistically on homeotic gene regulation (JURGENS 1985 ). Five molecularly characterized members of this group (reviewed in SIMON 1995 ) have been shown to be co-localized at a number of discrete sites on polytene chromosomes (ZINK and PARO 1989 ; DECAMILLIS et al. 1992 ; MARTIN and ADLER 1993 ; RASTELLI et al. 1993 ; LONIE et al. 1994 ). The PC protein shares a domain homologous with the nonhistone heterochromatin-associated protein HP1 (PARO and HOGNESS 1991 ), which is necessary for PC attachment to the chromatin (MESSMER et al. 1992 ). By analogy with the model proposed for heterochromatin formation in the case of PEV (LOCKE et al. 1988 ), it was thus proposed that PC-G proteins act negatively on their target genes by inducing locally a clonally inherited heterochromatin-like structure, thus ensuring the clonal maintenance of the transcriptional repression state of these targets (PARO 1990 ). Derived models are described in PIRROTTA and RASTELLI 1994 ; ORLANDO and PARO 1995 ; and BIENZ and MULLER 1995 . It is likely that DREV depends on the local formation of a heterochromatin-like structure induced by multimeric complexes containing some or all Pc-G proteins.

Conversely, the trx-G gene proteins are required for continued transcriptional activation of homeotic genes (MAZO et al. 1990 ; BREEN and HARTE 1993 ). The molecularly characterized members of the TRX-G consist of a diverse set of proteins. While some proteins seem to be specific for regulating developmental genes (CHINWALLA et al. 1995 ), others appear to have a more general role in gene activation (TAMKUN et al. 1992 ; FARKAS et al. 1994 ; TSUKIYAMA et al. 1994 ; DINGWALL et al. 1995 ). It has been proposed that the TRX-G proteins maintain homeotic gene activity by keeping the DNA in an "open" chromatin configuration.

In contrast to PEV, DREV patterns can be stable and heritable. This reproducibility indicates that the phenotypic expression of the white⁺ transgene is reflecting positional information at work in the developing eye. In most cases, the patterned expression displays an anterior-posterior gradient (BHOJWANI et al. 1995 ; SUN et al. 1995 ). In rare cases, halolike patterns (COEN 1990 ) or dorsal-ventral (D/V) patterns can be observed (LEVIS et al. 1985 ; HAZELRIGG and PETERSON 1992 ; IRVINE and WIESCHAUS 1994 and personal communication; SUN et al. 1995 ; BRODSKY and STELLER 1996 ).

The mini-white⁺ gene, with its constitutive modest expression level (PIRROTTA 1988 ), can be used as a reporter for the detection of both enhancer and silencer effects of cis-regulating sequences flanking its insertion point. These flanking sequences are supposed to direct the expression of the neighboring genes (BELLEN et al. 1989 ; BIER et al. 1989 ; WILSON et al. 1989 ). Taking advantage of this property of mini-white⁺ reporter genes, several groups have successfully performed screens for Drosophila genes that could be implicated in pattern formation (BHOJWANI et al. 1995 ; SUN et al. 1995 ; BRODSKY and STELLER 1996 ). It is noteworthy that, as noticed by BHOJWANI et al. 1995 , over half of the chromosomal loci where various P[w⁺] insertions display patterned white expression are binding sites for Pc-G gene products (HAZELRIGG et al. 1984 ; LEVIS et al. 1985 ; KASSIS et al. 1991 ; FAUVARQUE and DURA 1993 ; SUN et al. 1995 ). This suggests that the screen based on patterned expression of (mini-)white⁺ reporters appears prone to detect genes that are targets for Pc-G-mediated developmental regulation.

We have undertaken the collection of transgenic lines displaying a dorsally restricted expression pattern in the adult eye. Our working hypothesis was that the study of different transgenic lines, all displaying the same stable eye pigmentation pattern, would allow us to identify genes whose expression is subject to common developmental regulators. Moreover, we thought that the study of lines showing a differential expression of white⁺ reporter along the D/V axis of the eye would allow us to target our screen to genes that are involved in the establishment of the dorsal (vs. ventral) identity of ommatidial clusters in the adult eye. Localization of these transgenes, genetic analysis of the phenotypes induced by their insertion, and analysis of developmental expression patterns of the reporter genes allowed us to identify a single genomic region, 69D, showing the D/V effect. This region includes several homeobox-containing genes, coordinately regulated and implicated in similar developmental processes. In the developing eye, ventral repression of these genes is regulated by the Pc-G gene products, suggesting that regulation of the region we have identified may be exerted at the level of chromatin structure.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly strains and culture:
All strains were maintained on standard culture medium at 18°, 20°, or 25°. All variants used are described in LINDSLEY and ZIMM 1992 , except when stated in the article. All D/V transgenic lines described in this section were backcrossed with the y w^67c23 stock for ten generations.

T3 and T81 (renamed iro^T3 and iro^T81): These two strains contain an insertion of the P[w^d1] transgene. P[w^d1] carries the whole white⁺ gene with a direct tandem duplication comprising the 5' regulatory sequences and the first exon (COEN 1990 ). These strains were independently obtained after P-induced mobilization of an insertion of P[w^d1] (COEN 1990 ).

cre¹ (renamed mirr^cre1): We isolated this strain after mobilization of P[lacW] (BIER et al. 1989 ) with the stable source of transposase P[ry⁺ 2-3](99B) (ROBERTSON et al. 1988 ), designated hereafter as 2-3.

l(3)A5-3-42[1] (renamed mirr^cre2), 35^d (renamed mirr^cre3), Sc5 (renamed mirr^cre4), 59-12 (renamed mirr^cre5), R's (renamed mirr^cre6), Sc2 (renamed iro^Sc2), B6.8 (renamed iro^B6.8), T's, and Sc4: These strains, containing a P[lacW] insertion (BIER et al. 1989 ), were kindly provided by, respectively, the Bloomington Stock Center (Bloomington, IN); S. KERRIDGE, Laboratoire de Génétique—Centre Universitaire de Marseille, Marseille, France; M. BOUBE and D. CRIBBS (who also provided Sc2 and Sc4), Centre de Biologie du Développement Centre National de la Recherche Scientifique, Toulouse, France; P. MAROY, Department of Genetics, Atlila Jozsef University, Szeged, Hungary; R. COSSARD and R. TERRACOL, Laboratoire de Génétique du Développement et Evolution, Institute Jacques Monod, Paris, France; D. DORER, Division of Biomedical Sciences, Meharry College, Nashville, TN. The l(3)A5-3-42[1] line is described in HARTENSTEIN and JAN 1992 .

K's: This strain harbors a P[UAS-UbxW] insertion (K. D. IRVINE, personal communication). This transgene contains a pUAST-[Ubx IVa cDNA] fusion.

L's: This strain contains a P[Mtn W] insertion (L. THÉODORE, personal communication). This transgene harbors a functional Metallothionein (Mtn) transcriptional unit (MARONI et al. 1995 ).

J26.b16: This hobo enhancer trap line harbors the H[pHLw2] transgene (SMITH et al. 1993 ). Enhancer trap lines l(3)jD3, l(3)s2783, P[w⁺]33, l(3)j2E11, and l(3)j6C3, harboring a P[w⁺] insertion in 68F2–3, 69F5–6, 70C, 70C5–6, and 70D4–6, respectively, were obtained from the Bloomington Stock Center.

Enhancer trap lines 11F3 and 59A, harboring a P[w⁺] insertion in, respectively, the dptp69D and the Klc gene (DESAI et al. 1996 ), both located in 69D1–6, were obtained from C. DESAI. We also obtained deficiencies affecting the corresponding genes (Df(3L)8ex34, Df(3L)8ex25, dptp69D¹, and A6B (DESAI et al. 1996 ; C. DESAI, personal communication) from the same source.

Cloning of transposon-flanking regions:
DNA extraction from adult flies was performed as described by JUNAKOVIC et al. 1984 .

Inverse PCR (I-PCR) procedure: Genomic DNA of iro^T3 and iro^T81 was digested, ligated, and treated as described in DELATTRE et al. 1995 . Amplification reactions were performed on a Trio-Thermoblock Biometra Inc. (Tampa, FL) as follows: 35 cycles of 45-sec denaturation at 94°, 45-sec annealing at 45°, and 4-min extension at 72°, followed by 10 min at 72°.

The P-element-specific primers used were as follows (coordinates as in the P-element sequence; O'HARE and RUBIN 1983 ): primer 1, P108-P89, 5' CGTCCGCACACAACCTTTCC 3'; primer 2, P414-P433, 5' GGCTATACCAGTGGGAGTAC 3'; and primer 3, P31-7, 5' CGACGGGACCACCTTATGTTATTTC 3'.

A white-specific primer, localized at position 5009–5028 in white coordinates (O'HARE et al. 1984 ; GenBank accession number X02974), was also used: primer 4, 5' CGAATGCTCTCTCCATGCTC 3'.

PCR, with primers 1 and 2, on iro^T3 DNA (digested with EcoRI and ligated) allowed the amplification of a 1.2-kb DNA fragment flanking the 5' insertion point.

Two successive PCRs—first with primers 1 and 2 on iro^T81 DNA (digested with Nde2) and second with primers 3 and 4 on the preceding amplification product—allowed the amplification of a 0.7-kb DNA fragment flanking the 5' insertion point.

Plasmid rescue procedure: Cloning by plasmid rescue was performed on iro^Sc2, mirr^cre2, and mirr^cre3 DNA digested with EcoRI; on mirr^cre1 and mirr^cre4 DNA digested with SacII; and on iro^B6.8 digested with BglII according to PIRROTTA 1986 . This allowed the cloning of genomic DNA fragments (flanking the insertion point of P[lacW]) of 1.6 kb, 10 kb, 4.5 kb, 4.7 kb, 5.7 kb, and 2.0 kb, respectively, for iro^Sc2, mirr^cre1, mirr^cre2, mirr^cre3, mirr^cre4, and iro^B6.8. The same procedure was applied on J26.b16 for the cloning of a 6.6-kb BamHI fragment flanking the H[pHLw2] insertion point (SMITH et al. 1993 ).

Sequencing of transposon-flanking regions:
Clones containing the genomic DNA flanking mirr^cre2, mirr^cre3, mirr^cre4, and DH1 insertions were sequenced in an ABI 373 automatic sequencer (ABI Adv. Biotechnologies, Inc., Columbia, MD) using a primer complementary to the P-element inverted repeat (IR = CGATCGGGACCACCTTATGTTATTTCATCAT).

A 0.5-kb ClaI-XhoI genomic DNA fragment obtained from mirr1 genomic clone and including the 5' end of mirr cDNA (MCNEILL et al. 1997 ; H. MCNEILL, personal communication), was inserted in pBluescript and sequenced using T3 oligonucleotide as primer.

Molecular mapping of D/V transgenes:
Clones containing the genomic DNA flanking iro^T3, iro^T81, iro^B6.8, iro^Sc2, and J26.b16 transgenes were used as probes to hybridize Southern blots containing EcoRI restriction fragments obtained by digestion of genomic DNA of the region (GOMEZ-SKARMETA et al. 1996 ; MCNEILL et al. 1997 ; R. DIEZ DEL CORRAL and J. MODOLELL, unpublished results). The restriction pattern of the transposon-flanking clones was then compared to that of genomic clones to determine the position of the insertions within the EcoRI restriction fragments. iro^T3 is located upstream from the 5' end of ara cDNA. iro^T81 and iro^B6.8 are inserted in ara second and third introns, respectively. iro^Sc2 is located upstream from the 5' end of caup cDNA.

The exact positions of mirr^cre2, mirr^cre3, mirr^cre4, and DH1 insertions, relative to mirr cDNA, were determined by sequencing cloned genomic DNA flanking the P[lacW] insertion site and a mirr1 genomic subclone including the 5' end of mirr cDNA (H. MCNEILL, personal communication). Sequence alignment revealed that mirr^cre3, mirr^cre4, and both mirr^cre2 and DH1 are located 343 bp, 282 bp, and 474 bp upstream from the 5' end of mirr cDNA, respectively.

Preparation of P1 DNA and Southern analysis:
Bacterial clones containing single P1 clones (clones covering the 69C2–70A5 region: DS02752, DS00099, DS00285, DS07487, DS08512, DS00044, DS08062, DS00298, DS00073, DS06094, DS07359, DS08991, DS04287, DS02826, DS00334, DS07050, DS04368, DS06456, DS00722, DS04746, and DS06041, Drosophila Genome Center, Berkeley, CA; Figure 3) were inoculated into 500 ml of Luria broth (LB) medium containing 25 µg/ml kanamycin and 1 mM IPTG (isopropyl 1-thio-ß-D-galactopyranoside) and grown for ~6 hr at 37° (until OD₅₅₀ = 1.3–1.5). Plasmid DNA was extracted according to the maxi-prep kit protocol (QIAGEN Inc., Chatsworth, CA). Southern blots of these P1 plasmids were hybridized with each flanking genomic DNA fragment cloned as probes.

View larger version (17K): In this window In a new window Download PPT slide   Figure 1. —Embryos immunostained for ß-galactosidase. Stained embryos that do not display the ftz pattern have the boxed genotype.

View larger version (98K): In this window In a new window Download PPT slide   Figure 2. —Eye pigmentation pattern of some D/V transgenic lines. (A) iroT81; (B) L's; (C) mirrcre1; (D) iroSc2; (E) iroB6.8; (F) T's. Orientation is anterior to the left and dorsal to the top. In A, a red sector appears ventrally on a yellow background. In B, mottling is visible in the ventral part of the eye. For all other D/V lines shown, all the ventral ommatidia are white.

View larger version (14K): In this window In a new window Download PPT slide   Figure 3. —Physical map of the D/V region. An EcoRI (R) restriction map of the D/V region is shown. The irorF209 element insertion point has been taken as the origin of coordinates (GOMEZ-SKARMETA et al. 1996 ). The triangles indicate the positions of the D/V insertions. Overlapping lines above the map indicate P1 clones used to localize the insertions. The dotted lines indicate that the positions of the ends of cloned DNA in the P1 bacteriophages have not been determined. The exact size of the DNA linking the ara/caup and mirr regions has been determined from the data of a genomic walk of the region (R. DIEZ DEL CORRAL and J. MODOLELL, unpublished results). Arrows under the DNA line show the structure of the ara, caup, and mirr transcripts (GOMEZ-SKARMETA et al. 1996 ; MCNEILL et al. 1997 ). The position of the insertions was determined by restriction and Southern analysis or by genomic sequencing (see MATERIALS AND METHODS). DH1 is a P[lacW] insertion presenting a D/V pattern in the eye and initially mapped to cytological position 69C8–11 (SUN et al. 1995 ).

In situ localization on polytene chromosomes:
Preparation of chromosome spreads and cytogenetic localization of transgene insertion sites of every transgenic line listed were performed as described in FAUVARQUE and DURA 1993 . A 1.5-kb DNA fragment containing the sixth exon of the white gene (fragment SalI +12725 to +14240 in white coordinates) labeled with biotin-dUTP was used as a probe.

Cloned DNA fragments flanking iro^T3 and mirr^cre2 insertions were also labeled with biotin-dUTP and hybridized to chromosomes of the w¹¹¹⁸ stock.

Methods of molecular biology:
All standard molecular techniques (such as restriction digestion, agarose gel electrophoresis, Southern blotting, and hybridization) were performed as described in MANIATIS et al. (1989).

Generation of mirr^cre1 derivatives:
w¹¹¹⁸/Y; 2-3, Sb/TM6, Ubx males were crossed with w¹¹¹⁸; mirr^cre1 females. w¹¹¹⁸/Y; Sb, 2-3/mirr^cre1 F₁ males were then individually mated to w¹¹¹⁸; TM3, Sb/T(2;3)apterous^Xa females. Eye color was examined in the F₂ progeny: exceptional [white] F₂ males were individually mated to w¹¹¹⁸; TM3, Sb/T(2;3)apterous^Xa females. Sibling [white; Sb] F₃ male and female progeny were mated and, in their progeny, [white; Sb⁺] individuals were scored for viability or phenotypical defects.

Histochemical staining:
ß-Galactosidase activity was detected in adult ovaries by 5-bromo-4-chloro-3-indolyl-ß-D-pyranoside (X-gal) staining according to LEMAITRE et al. 1993 .

Imaginal discs and brains were dissected from late third-instar larvae and stained with X-gal as described by LEMAITRE and COEN 1991 .

Immunostaining of embryos:
Embryos were stained with an antibody directed against ß-galactosidase as described by INGHAM and MARTINEZ-ARIAS 1986 . In control experiments on embryos of a w¹¹¹⁸ stock, no staining was detected with this antibody.

In situ hybridization in whole embryos:
The in situ hybridization of (mini^-)white⁺ or lacZ transcripts in whole embryos was performed as described by BONNETON and WEGNEZ 1995 . The white antisense probe used was a 0.86-kb fragment (fragment SalI +11866–+12725) including the fourth and fifth exon of the white gene (DELATTRE et al. 1995 ) and labeled with digoxigenin (Dig-dUTP). In control experiments on embryos of a w¹¹¹⁸ stock, no staining was detected with this probe.

Testing for the effect of PEV modifiers:
Females from the tested line were mated to males mutant for a modifier of PEV. The effect of these modifiers on eye pigmentation pattern was observed in males issued from this cross. Suppressors tested were Su(var)2-101, Su(var)2-501, Su(var)205-5, Su(var)2-b4801, Su(var)2-b204, Su(var)2-b701, Su(var)2-b801, Su(var)2-201, Su(var)2-1001, Su(var)3-316, Su(var)3-103, Su(var)3-c1101, Su(var)3-902, Su(var)3-401, and Su(var)3-111. Enhancers were E(var)8, E(var)102-1, E(var)129-1, E(var)166-7, E(var)56-9, E(var)70-2, E(var)90, E(var)3-101, and a duplication of Su(var)3 in 21A. These modifiers of PEV were kindly provided by G. REUTER (Institute of Genetics, Martin Luther University, Halle, Germany) and J. GAUSZ (Institute of Genetics, Biological Research Center, Szeged, Hungary). Their effect on PEV was confirmed on w^vco.

To generate progeny with an extra Y chromosome (XXY females) or with no Y chromosome (XO males), females from the tested lines were mated to males with attached XY, w¹¹¹⁸ (R. LEVIS; described in DORER and HENIKOFF 1994 ).

Testing for the effect of trx-G or Pc-G gene mutations:
To test the effect of a dosage reduction of Pc-G or trx-G gene products on eye pigmentation, females from the tested line were mated to males heterozygous mutant for the Pc-G or trx-G gene tested (mutant/Balancer). In the progeny, eye pigmentation patterns of sibling males that did or did not display the balancer chromosome marker were compared.

To test the effect of the ph⁴¹⁰ mutation on lacZ expression patterns in larvae, ph⁴¹⁰ w females were mated to w/Y; P[lacW]/TM3 males. Mutant ph⁴¹⁰ w/Y; P[lacW]/+ male and heterozygous ph⁴¹⁰ w/+; P[lacW]/+ female larvae derived from this cross were stained with X-gal and compared for lacZ expression patterns.

Crosses allowing analysis of the effect of the ph⁶⁰⁰ mutation on mirr^cre3 expression pattern in embryos are described in Figure 1.

A fushi tarazu-lacZ (ftz-lacZ) fusion was used to trace the X chromosomes that do not bear the ph mutation. In the progeny of the first cross, males presenting the D/V pigmentation pattern with a Bar phenotype were crossed to ph⁶⁰⁰ w/FM7c ftz-lacZ females. F₂ embryos, immunostained for ß-galactosidase, that do not display the ftz pattern are mutant for ph and bear one copy of the mirr^cre3 transgene (ph⁶⁰⁰ w; mirr^cre3/+ males).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

white⁺ transgenes showing a D/V restriction of white expression pattern co-localize in a single chromosomal region, 69D:
In the course of P[w⁺] transgene-mediated mutagenesis, we have recovered three independent transgenic lines (iro^T3, iro^T81, and mirr^cre1; see Table 1) displaying a peculiar pattern of white expression in the adult eye: the pigmentation is normal in the dorsal half of the eye, but white expression is strongly or completely repressed in the ventral half, with a clear D/V boundary (Figure 2A and Figure C). The cytological localization of these insertions revealed that all three were clustered in the same chromosomal site, 69D1–3 (Table 1).

We systematically collected, from various sources, flies harboring a white⁺-derived transgene with a similar D/V pattern (D/V transgenes). We thus obtained 11 other independent transgenic lines presenting this pattern in the adult eye (Table 1 and Figure 2). Differences in the extent of the pigmented area and in the pigmentation level can be seen among these transgenic lines. In most cases (iro^T3, iro^T81, mirr^cre1, mirr^cre2, mirr^cre3, mirr^cre4, mirr^cre5, mirr^cre6, iro^Sc2, L's, K's, and J26.b16), the pigmented area corresponds to the dorsal half of the eye, with a pigmentation level ranging from orange to red (Figure 2, A–D). However, two of the lines, iro^B6.8 and T's, display a more limited pigmentation, gradually fading from the dorsal to the equatorial part of the eye (Figure 2E and Figure F). In most lines, white gene expression is completely abolished in the ventral half of the eye (ommatidia are white, Figure 2, C–F). In iro^T3 and iro^T81 lines (bearing a transgene containing the complete white gene sequence), the ventral half of the eye is yellow, with a ventral-posterior red sector and some additional mottling (Figure 2A). The L's line also displays some mottling in the ventral half of the eye (Figure 2B). For every P[lacW] insertion that is homozygous viable, the pigmentation level is higher when the transgene is homozygous than when it is heterozygous.

The localization of these transgenes by in situ hybridization on polytene chromosomes of salivary glands was carried out using a fragment of the white gene as a probe. The 12 D/V transgenes considered here were found located in the same cytological interval of bands on the third chromosome: 69D1–3 (Table 1). We have cloned the genomic DNA flanking the insertion point of nine of the D/V transgenes. In situ localization of the genomic DNA fragments flanking two insertions (iro^T3 and mirr^cre2) was performed on polytene chromosomes of the w¹¹¹⁸ stock, which is devoid of P-element insertions. In both cases, the localization was identical to that obtained with the white probe on the corresponding transgenic lines.

Meiotic recombination rates between different D/V insertions were estimated by the yield of [white] recombinants produced by females heterozygous for two different D/V transgenes. This analysis showed that there could be up to 1.5% recombination between the most distant transgenes (iro^T3 and mirr^cre2), whereas no recombinants were obtained between iro^T3, iro^T81, and iro^Sc2. Intermediate recombination frequencies were obtained between other combinations of D/V transgenes when tested by pair (data not shown). D/V insertions are thus genetically separable, suggesting that the D/V phenotype was not the result of several P[w⁺] insertions into a single site.

By Southern blotting of genomic DNA from P1 contiguous clones covering the 69C2–70A4 chromosomal region (Drosophila Genome Project; SMOLLER et al. 1991 ; HARTL et al. 1994 ), we have mapped some transgene insertion points using the flanking genomic DNA fragments as probes. The insertion site of iro^T3, iro^T81, iro^B6.8, and iro^Sc2 was thus restricted to a single P1 clone (Figure 3): DS08512 (hybridizing the 69D1 band). The genomic DNA flanking the mirr^cre3 insert hybridizes with three P1 clones (Figure 3): DS00044, DS08062 (hybridizing the 69D1–3 bands), and DS00298 (hybridizing the 69D2–3 bands). The localization of mirr^cre3 can thus be restricted to the 69D2–3 interval (as it is not included within the DS08512 P1 clone). Genomic DNA flanking J26.b16 hybridizes the P1 plasmids DS00044 and DS00298 and is thus localized in 69D2–3.

Three homeobox-containing genes were recently isolated in the 69D chromosomal region. Two transcription units, araucan (ara) and caupolican (caup) were detected within the iroquois (iro) locus (DAMBLY-CHAUDIERE and LEYNS 1992 ; GOMEZ-SKARMETA et al. 1996 ). They encode related proteins that contain a novel class of homeodomain. Both proteins contribute to iro function. Another related gene, mirror (mirr), has been isolated in the 69D region (MCNEILL et al. 1997 ). It encodes a protein with a homeodomain very similar to those of ARA and CAUP, although the iro proteins are related more closely to each other than to MIRR (GOMEZ-SKARMETA et al. 1996 ). The similarity of mirr to ara and caup and its coincident location in 69D led to the identification of mirr as another member of the iro complex (GOMEZ-SKARMETA et al. 1996 ; MCNEILL et al. 1997 ).

P[lacW] insertions in mirr (mirr^P1 and mirr^P2), are expressed in the dorsal half of the eye (MCNEILL et al. 1997 ; BRODSKY and STELLER 1996 ). A P[lacZ] insertion in ara (iro^rF209; GOMEZ-SKARMETA et al. 1996 ) expresses lacZ in the dorsal half of the eye imaginal disc (Figure 7D). The size of the genomic region separating those two loci (ara/caup and mirr) was estimated at less than 75 kb, since caup and mirr probes hybridized with the DS00044 P1 clone. This has been confirmed (R. DIEZ DEL CORRAL and J. MODOLELL, unpublished results) by a genomic walk between the ara/caup and mirr previously isolated genomic clones (GOMEZ-SKARMETA et al. 1996 ; MCNEILL et al. 1997 ) demonstrating that mirr is located 70 kb proximal to caup. The ara, caup, and mirr transcription units are thus included within a 140-kb genomic region (Figure 3). We mapped the D/V insertions for which genomic flanking DNA was cloned (except mirr^cre1). This showed that iro^T3, iro^T81, and iro^B68 are inserted into ara; iro^Sc2 into caup; and mirr^cre2, mirr^cre3, and mirr^cre4 into mirr. J26.b16 insertion point is located between caup and mirr and does not affect a transcription unit previously described in that region. The D/V insertions (except T's) that have not been localized on the molecular map have been linked to insertions in iro or mirr by genetic recombination or complementation studies: L's and K's, respectively, map close to iro^T3, and mirr^cre1, mirr^cre5, and mirr^cre6 are mirr mutants (genetic analysis of the DIV region and data not shown). In conclusion, we found that P[lacW] insertions showing a D/V pattern are all clustered in a single genomic region of at least 140 kb.

View larger version (79K): In this window In a new window Download PPT slide   Figure 4. —Phenotypical defects of mirr mutant adults. (A) Wing and bristle defects of one viable [white] derivative of mirrcre1: mirrcre111: left panel, dorsal view of a w1118 adult showing wild-type wings and thoracic bristles; right panel, dorsal view of a mirrcre111 adult displaying reduced size, outheld wings with missing alulae, and absence of some macrochaetes on the thorax. (B) High magnification of a wild-type wing hinge (Canton strain). The alula is indicated with an arrow. (C) mirrcre2/mirrcre4 adult wing-hinge defect. Severe distortion of the wing hinge and almost complete absence of alula are observed on both wings of mirrcre2/mirrcre4 escapers. Wings are oriented with proximal to the left and anterior to the top.

View larger version (86K): In this window In a new window Download PPT slide   Figure 5. —D/V transgene expression patterns during embryogenesis. In situ hybridization, with a white antisense probe, on whole mount embryos. All embryos are shown at retracted germ band stage and are oriented anterior to the left and dorsal to the top. (A) mirrcre3; (B) iroSc2; (C) iroB6.8; (D) J26.b16. In A, expression of mirrcre3 reporter gene is detected in the central nervous system (CNS), in the anterior part of each metameric unit of the ventral nerve cord, and in the brain. The reporter gene is also expressed in the proventriculus and in the dorsal epidermis of each segment. This expression pattern is identical to that revealed with a mirr cDNA probe (MCNEILL et al. 1997 ). In B, at the retracted germ band stage, iroSc2 reporter expression domains are the same as mirr—except in the CNS, where there is no detectable staining. In C, when the germ band is retracted, iroB6.8 reporter expression pattern is partly identical to that of iroSc2 (no expression in the CNS)—but in addition, groups of cells are stained in the dorsal-lateral epidermis. In D, J26.b16 transgene expression is very similar to that of iroSc2, but a weak staining is detectable in the ventral neural ectoderm.

View larger version (175K): In this window In a new window Download PPT slide   Figure 6. —iroT3 expression pattern during embryogenesis. In situ hybridization, with a white antisense probe, on whole mount embryos. Embryos are oriented anterior to the left, dorsal views. (A) Embryo of the Canton strain. At retracted germ band stage, endogenous white+ expression is strong in the Malpighian tubules. (B) iroT3 embryo. iroT3 expression pattern is mostly identical to that of the white gene, since additional staining is detected only in the head.

View larger version (348K): In this window In a new window Download PPT slide   Figure 7. —Expression patterns of D/V transgenes at the third instar larvae. lacZ expression detected by X-gal staining of larval tissues. (A–D) Eye imaginal discs, oriented anterior to the left and dorsal to the top. lacZ expression is restricted to the dorsal half of the eye disc, as shown here for mirrcre3 (A); iroSc2 (B); T's (C); and irorF209 (D). (E–H) Wing imaginal discs, oriented dorsal to the top and posterior to the right. (E) mirrcre3; (F) iroSc2; (G) iroB6.8; (H) J26.b16. lacZ expression is detected in the prospective notum, N; alula, AL; pleura, PL; and in domains that may correspond to the prospective L1, L3, and L5 longitudinal vein regions (E–G). (I and J) Legs imaginal discs of iroSc2 (I); and iroB6.8 (J). (K) iroSc2 proventriculus. (L) mirrcre3 brain.

To determine whether the "D/V effect" (ventral repression of white⁺ transgenes' expression) extends beyond 140 kb, we analyzed the eye pigmentation patterns of eight enhancer trap lines harboring a P[w⁺] insertion in the 68–70 interval (see MATERIALS AND METHODS). One line located in 69C (Sc4) displayed a pigmentation restricted to the anterior-equatorial part of the eye. The other lines, located outside of the 69C–69D1–3 interval, had uniformly pigmented eyes. This study allowed us to confine the D/V effect distally to 69C and proximally to 69D3–6.

We have mapped, by functional complementation, deficiencies covering dptp69D and Klc genes, both located in 69D1–6 (DESAI et al. 1996 ), with respect to deficiencies of the IRO-C (data not shown). This genetic analysis showed that the dptp69D and Klc genes are proximal to mirr. Moreover, a dptp69D genomic clone (DESAI et al. 1996 ) hybridized to the P1 clone DS08062, but neither to DS00298 nor to DS00044 (Figure 3), confirming that the dptp69D and Klc genes are proximal to mirr. As P[w⁺] insertions in dptp69D and Klc genes are not affected by the D/V effect, this effect is thus limited to the distal region of these genes.

Genetic analysis of the D/V region:
Seven out of 14 D/V insertions are homozygous viable and do not lead to any visible phenotype in adults.

With low penetrance, the iro^B6.8 strain displays a wing phenotype resembling that of iro¹ and iro^rF209 homozygotes (DAMBLY-CHAUDIERE and LEYNS 1992 ; GOMEZ-SKARMETA et al. 1996 ; LEYNS et al. 1996 ). More than 30% of iro^B6.8/iro² trans-heterozygous survive. They display outheld wings and either duplicated or missing thoracic bristles. iro^rF209/iro^B6.8 trans-heterozygous are viable, and some of them display outheld wings. This suggests that iro^B6.8 is a hypomorphic allele of iro, weaker than iro^rF209.

Six D/V insertions are homozygous lethal or semilethal and are allelic (Table 2). We have called the functional unit affected by these insertions crépuscule (cre; "twilight" in French). These insertions do not complement the lethality of the mirr^P2 insertion, an early larval homozygous lethal allele of mirr (BRODSKY and STELLER 1996 ; MCNEILL et al. 1997 ), which shows that they are alleles of the mirr gene. We have therefore renamed these alleles mirr^cre1–mirr^cre6.

The escapers homozygous for the mirr^cre1 insertion display thoracic macrochaetae duplications. We have generated transposase-induced [white] derivatives of mirr^cre1. Revertants for the mirr^cre1-associated defects were recovered this way, showing that the mutant phenotype is a consequence of the P-transgene insertion. Partly viable [white] derivatives of mirr^cre1 were also recovered. They have outheld wings with missing or reduced alulae and either loss or duplication of thoracic bristles (Figure 4A). These defects are similar to those displayed by escapers homozygous for mirr^cre5 or trans-heterozygous for some combinations of mirr lethal or sublethal alleles (Table 2 and Figure 4B and Figure C).

The wing and bristle phenotypes of mirr mutants are reminiscent of the dominant Dichaete (D) phenotype (BRIDGES and MORGAN 1923 ). Previously described D alleles (D¹, D³, and D⁴) are not separable from chromosomal rearrangements (Table 2). We have seen that D¹ and D³ do complement D⁴ lethality. Recently, the dominant wing phenotype of D alleles, associated with the breakpoints in 70–71, was shown to be due to a mutation located in 70D and encoding a Sox-domain protein (RUSSELL et al. 1996 ). The name Dichaete was retained for that gene. We have obtained from this laboratory (A. T. CARPENTER, Department of Genetics, University of Cambridge, England, personal communication) two allelic lethals, Sail1 (Sai¹) and Sail2 (Sai²), mapped to 69D (Table 2) and associated with a dominant outheld-wing phenotype. The mirr alleles fail to complement the lethality of D¹ and D³ or the lethality of Sai¹ and Sai², whereas they did complement the lethality associated with D⁴ or other D alleles affecting 70D (Table 2 and A. T. CARPENTER, personal communication). The 69D–70D interval comprises thus at least four genes (ara, caup, mirr, and D) susceptible to giving dominant or recessive D phenotypes when mutated independently. We propose to rename the alleles affecting the mirr gene mirr^D1, mirr^D3, mirr^Sai1, and mirr^Sai2. The mirr gene would be implicated, notably, in wing and peripheral nervous system development, according to the phenotype of its mutations.

This complementation and phenotypic analysis, together with previous study on IRO-C (GOMEZ-SKARMETA et al. 1996 ), shows that the D/V region in 69D contains at least three different functional units (ara, caup, and mirr), consistent with the molecular data.

Developmental expression patterns:
Expression patterns were analyzed for some representative D/V transgenes from oogenesis to larval stages.

Four lines containing the lacZ reporter gene were analyzed for ß-galactosidase activity in ovaries. mirr^cre2 and mirr^cre3 display an identical expression pattern: lacZ expression is detected from the beginning to the end of oogenesis. At stage 10, it is restricted to the follicle cells surrounding the anterior-dorsal part of the oocyte, the region where the nucleus is located (data not shown). No expression was detected in ovaries of iro^Sc2 and iro^B6.8 lines.

The expression pattern of D/V transgenes in embryo was visualized by in situ hybridization with an antisense RNA white probe (for mirr^cre1, mirr^cre3, mirr^cre6, iro^Sc2, iro^B6.8, iro^T3, iro^T81, L's, K's, and J26.b16 lines) and, in addition, by immunodetection of ß-galactosidase (for mirr^cre1, mirr^cre3, mirr^cre6, and iro^Sc2 lines that carry the lacZ reporter; data not shown). The slight differences in staining observed between the two methods of detection may be due to the perdurance of the ß-galactosidase protein rather than to differences in the expression domain of the reporter genes tested. In fact, in situ hybridization achieved either with a lacZ or a white antisense probe on the mirr^cre3 line gave exactly the same pattern, therefore showing that the two reporters of the P[lacW] construct give the same expression pattern in the embryo, as previously described for transgenes subject to position effect (KASSIS et al. 1991 ; FAUVARQUE et al. 1995 ).

For most lines, transgene expression is very dynamic and very specific. The expression patterns, although similar, differ from one line to the other, with common subpatterns shared by certain lines. In all cases, transgene expression is first detected very early in development and persists throughout embryogenesis. Expression patterns of P[lacW] insertions in mirr, ara, and caup reflect those revealed with the cDNA probes for the three corresponding genes (GOMEZ-SKARMETA et al. 1996 ; MCNEILL et al. 1997 ; J. L. GOMEZ-SKARMETA, personal communication). A representative for ara, caup, mirr, and also J26.b16 expression is shown in Figure 5. Expression patterns of K's and L's insertions are hardly detectable (data not shown). It should be noted that throughout embryogenesis, iro^T3 expression pattern was found to be mostly identical to that of the white gene (compare Figure 6A TO B). iro^T81 expression pattern is also identical to that of the white gene, with an additional strong staining in the primordia of the proventriculus. This shows that white⁺ transgene expression in iro^T3 and iro^T81 mostly reflects that of the endogenous white gene.

We have studied lacZ reporter gene expression patterns in third instar larvae. Tissues expressing the reporter gene in each line are listed in Table 3.

For all D/V lines, lacZ expression in the eye imaginal disc reflects (mini^-)white⁺ expression pattern in the adult eye: ß-galactosidase activity is detected only in the dorsal half of the disc (Figure 7, A–D). However, inside this domain, differences in the expanse of the ß-galactosidase activity can be observed among the D/V transgenic lines. In fact, the higher the lacZ expression level, the closer the ß-galactosidase staining is to the D/V border (for instance, compare Figure 7A TO C). A gradual pigmentation pattern (from dorsal to equatorial) was also observed in some cases (Figure 2E and Figure F) in the adult eye.

In the wing imaginal disc, lacZ reporter is expressed in domains that are precursors of the notum and part of the dorsal hinge, including the precursor of the alula (Figure 7, E–H and data not shown). For certain lines (mirr^cre3, mirr^cre4, iro^B6.8, and iro^Sc2), lacZ expression is also detected in restricted areas in the wing pouch. These domains might correspond to the prospective longitudinal veins (respectively, proximal L1 and distal L3 veins for mirr^cre3 and mirr^cre4; L3 veins and proximal L1 and L5 veins for iro^Sc2; and L3 and L5 veins for iro^B6.8; Figure 7, E–G).

Most D/V lines do not show lacZ expression in leg imaginal discs except the iro^Sc2, iro^B6.8, and (weakly) T's lines (Figure 7I and Figure J). For every line, lacZ is also expressed in other larval tissues in specific and similar patterns (Table 3 and Figure 7K and Figure L). The ß-galactosidase accumulation patterns of P[lacW] inserted into ara, caup, and mirr mostly reflect the expression pattern of these genes in third instar larvae (GOMEZ-SKARMETA et al. 1996 ; MCNEILL et al. 1997 and data not shown).

The similarity of the expression patterns at all developmental stages suggests that the genes included in the D/V region may be implicated in common developmental processes and coordinately regulated.

Interaction with modifiers of variegation, Pc- and trx-G genes:
The ventral silencing of (mini^-)white⁺ in D/V transgenic lines, which is sometimes associated with variegation and mottling (Figure 2A and Figure B), could be related either to PEV or DREV. Therefore, we tested the effect of mutations in genes involved in these phenomena on the expression pattern of the D/V transgenes.

The effect of 15 suppressors and 9 enhancers of PEV (listed in MATERIALS AND METHODS) was tested on the iro^T81 line pigmentation pattern. None of these mutations produced any effect on the D/V pattern. The effect of Su(var)205-5 was also assayed on mirr^cre3 and iro^Sc2 and did not affect pigmentation. The presence or absence of a Y chromosome, which is mainly heterochromatic, is also known to affect PEV (GOWAN and GAY 1933 ). We have shown that, unlike PEV, the D/V pattern of mirr^cre3 or iro^Sc2 lines was not influenced by variation in the number of Y chromosomes. Moreover, PEV decreases when breeding temperature of the flies increases and, reciprocally, increases when breeding temperature decreases. For iro^T3 and iro^T81 lines, we observed that, in contrast to PEV, the proportion of pigmented ommatidia increased when the flies were raised at 18° compared to 25°; the small ventral pigmented area is enlarged. This particular response to the elevation of temperature had previously been reported for zeste variegation (CHEN 1948 ) and for insertions of transgene in which mini-white⁺ expression was under the control of polyhomeotic or AbdB gene regulatory sequences (FAUVARQUE and DURA 1993 ; ZINK and PARO 1995 ). Thus, the ventral repression of white expression in D/V strains is mediated by a mechanism different from the centric heterochromatin inactivation in PEV.

69D is a binding site for the product of at least four members of the Pc-G: PC, PH, PSC, and PCL (ZINK and PARO 1989 ; DECAMILLIS et al. 1992 ; MARTIN and ADLER 1993 ; RASTELLI et al. 1993 ; LONIE et al. 1994 ). We therefore tested the effect of the mutation of several members of the Pc-G on the D/V pattern. Interactions with the alleles listed in Table 4 were studied with the iro^T81 line. The effect of a polyhomeotic mutation (ph⁴¹⁰) was assayed on all D/V lines. In all cases, we observed that a reduction of the dosage of Pc-G gene products leads to a diminution of the ventral repression (Figure 8, A–D), which is completely relieved in some cases (e.g., iro^T81 in combination with Sce¹ or ph⁴¹⁰; Figure 8A and Figure C). In addition, the Pc¹⁶ mutation, which has no effect on the mirr^cre1 eye pigmentation pattern by itself, strongly enhances the derepressive effect of the ph⁴¹⁰ mutation (data not shown). This suggests that Pc and ph products act synergistically for the ventral repression of mirr^cre1 expression in the eye, as they do for homeotic gene regulation (DURA et al. 1985 ). These results show that the ventral repression of D/V transgenes' expression in the 69D region is mediated by the Pc-G gene products and thus could be maintained by a mechanism analogous to DREV.

View larger version (108K): In this window In a new window Download PPT slide   Figure 8. —Effect of mutant background for Pc-G and trx-G genes on the D/V eye pigmentation pattern. (A) w/Y; iroT81/Sce1. (B) w/Y; iroT81/Pc16. (C) ph410 w/Y; iroT81/+. (D) ph410 w/Y; mirrcre3/+. (E) w/Y; iroT81/mor1. (F) w/Y; iroT81/Df(3R)red-31.

69D is a binding site for the trx gene products (CHINWALLA et al. 1995 ). Mutations of trx-G gene members were analyzed for their influence on the D/V pigmentation pattern. trx-G mutations tested with iro^T81 were Df(2L)net- PMF (including kis), kis², mor¹, trx^E2, Df(3R)red-31 (including trx and urdur), Df(2R)Ba-MP, Su(Pc)37D, Df(3L)kto², Dll³, brm², Df(3L)fz-CAL5, and Df(3L)fz-D21 (both deficiencies, including dev). Two of these mutations produced a modification of the iro^T81 pigmentation pattern: mor¹ and Def(3R)red-31 reduce, respectively, weakly and strongly, the number of pigmented ommatidia in the ventral part of the eye (Figure 8E and Figure F). The eye pigmentation of mirr^cre1 was not significantly altered by the trx-G mutations tested (trx^E2, brm², mor¹, and Df(3R)red-31).

In another example of DREV, it has been shown that the number of pigmented ommatidia is dependent on the relative "balance" of Pc-G and trx-G gene products (GINDHART and KAUFMAN 1995 ). We have thus further investigated the influence of trx-G mutations by testing whether they can suppress the effect of Pc-G gene mutation on the D/V pattern. This was achieved by comparing the eye pigmentation pattern of ph⁴¹⁰ w/w; iro^T81/+ and ph⁴¹⁰ w/w; iro^T81/trx-G^- females (mutations tested were mor¹, kis², trx^E2, and Df(3R)red-31). A partial suppression of the ventral derepression due to ph⁴¹⁰ was observed in combination with mor¹ and Df(3R)red-31. Thus, relieving the ventral repression does not permit revealing of an effect of additional trx-G mutations (kis² and trx^E2) that could have been undetected in a ph⁺ background because of a too strong repressive effect of PC-G proteins.

We have tested the effect of the ph⁴¹⁰ mutation (which leads to a strong ventral derepression of mini-white⁺ in the eye of D/V adults) on lacZ expression in mirr^cre3 third instar larvae. We compared the lacZ expression pattern of ph⁴¹⁰ w/Y; mirr^cre3/+ males to that of sibling ph⁴¹⁰ w/+; mirr^cre3/+ females (whose lacZ expression is identical to that of w/w; mirr^cre3/+ females). In these males, a clear ectopic staining was observed in the ventralmost part of the eye imaginal disc (Figure 9A and Figure B), but no effect of ph⁴¹⁰ was detected in other larval tissues. This tissue-specific derepression of the lacZ reporter perfectly reflects mini-white⁺ derepression in the eye of adults of the same genotype (see Figure 8D), showing that ph is required for the ventral repression of reporter genes' expression from the third larval stage onwards.

View larger version (95K): In this window In a new window Download PPT slide   Figure 9. —ph function is necessary for maintenance of mirr expression pattern in embryos and larvae. (A and B) lacZ expression in third instar larvae eye imaginal disc of w/Y; mirrcre3/+ male (A) and ph410 w/Y; mirrcre3/+ male (B). In addition to the staining in the dorsal half in a wild-type background (A), an ectopic staining is detected in the ventral-most part of the disc in a ph410 background (B). Discs are oriented anterior to the left and dorsal to the top. (C and D) Immunostaining of ß-galactosidase in whole mount embryos. All embryos are oriented with anterior to the left and dorsal to the top. (C) mirrcre3 expression pattern in a ph+ background. (D) mirrcre3 expression pattern in a ph null background. A strong decrease of staining is seen in the ventral nerve cord, when compared to C, at the retracted germ band stage.

In embryos mutant for a member of the Pc-G genes, ectopic expression of homeotic genes first appears during germ band elongation (STRUHL and AKAM 1985 ; WEDEEN et al. 1986 ; DURA and INGHAM 1988 ; MCKEON and BROCK 1991 ; SIMON et al. 1992 ). If Pc-G genes act on the D/V pattern in the same way that they act on the regulation of homeotic genes, they should be responsible for the maintenance, but not for the initiation, of this pattern. We thus tested the effect of a null mutation in ph (ph⁶⁰⁰) on lacZ expression pattern in mirr^cre3 embryos. At the head involution stage (stage 15), the staining in the ventral nerve cord of these embryos is strongly decreased compared to ph⁺ embryos (Figure 9C and Figure D). No other effect of ph⁶⁰⁰ was detected, either before or after this stage or in other tissues. The same positive regulator effect of ph function on genes expressed in the central nervous system was previously described (DURA and INGHAM 1988 ). Our result suggests that ph function is required, very likely indirectly, for the positive regulation of mirr^cre3 expression specifically in the neural cells of retracted germ band embryos. In the case of homeotic gene regulation, the absence of ph product leads both to an ectopic expression of homeotic genes in the epidermis and to a diminution of expression in the ventral nerve cord from the extended germ band stage (DURA and INGHAM 1988 ).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

All white⁺ transgenes showing a dorsal restricted expression of white localize in a single chromosomal region, the D/V region:
Our results show that all the (mini^-)white⁺ transgenes displaying a D/V expression pattern, where the white gene is expressed in the dorsal half of the eye and repressed in the ventral part, are confined to a single genomic region, 69D1–3. This region seems to be unique, since no transgene showing the same pattern of white expression has, to our knowledge, ever been found elsewhere in the genome of Drosophila. In fact, recently, P[lacW] transgenic lines that express (mini^-)white⁺ strictly in the dorsal half of the eye have been isolated by others (SUN et al. 1995 ; BRODSKY and STELLER 1996 ; CHOI et al. 1996 ; MCNEILL<