A Green Fluorescent Protein Reporter Genetic Screen That Identifies Modifiers of Hox Gene Function in the Drosophila Embryo

Corresponding author: Yacine Graba, CNRS, Université de la Méditerranée, Parc Scientifique de Luminy, Case 907, 13288 Marseille Cedex 09, France.

Communicating editor: T. C. KAUFMAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Hox genes encode evolutionarily conserved transcription factors that play fundamental roles in the organization of the animal body plan. Molecular studies emphasize that unidentified genes contribute to the control of Hox activity. In this study, we describe a genetic screen designed to identify functions required for the control of the wingless (wg) and empty spiracles (ems) target genes by the Hox Abdominal-A and Abdominal-B proteins. A collection of chromosomal deficiencies were screened for their ability to modify GFP fluorescence patterns driven by Hox response elements (HREs) from wg and ems. We found 15 deficiencies that modify the activity of the ems HRE and 18 that modify the activity of the wg HRE. Many deficiencies cause ectopic activity of the HREs, suggesting that spatial restriction of transcriptional activity is an important level in the control of Hox gene function. Further analysis identified eight loci involved in the homeotic regulation of wg or ems. A majority of these modifier genes correspond to previously characterized genes, although not for their roles in the regulation of Hox targets. Five of them encode products acting in or in connection with signal transduction pathways, which suggests an extensive use of signaling in the control of Hox gene function.

MULTICELLULAR organisms acquire specialized structures during development. These phenotypic traits have provided landmarks for the identification of genes critical for the generation of pattern diversity during development and evolution. The discovery of homeotic mutations that transform one part of the body into the likeness of another one (BATESON 1894 ) led to the genetic and molecular identification of Hox genes in a wide range of organisms covering the metazoan phyla, including hydra, nematodes, arthropods, and humans. Hox genes are clustered within complexes, are differentially expressed along the antero/posterior (A/P) axis, and specify the diversified morphogenesis of animal body parts (MCGINNIS and KRUMLAUF 1992 ; BOTAS 1993 ; GELLON and MCGINNIS 1998 ). They encode homeodomain (HD)-containing transcription factors thought to control different sets of subordinate targets required for morphogenetic processes (GARCIA-BELLIDO 1975 ; GRABA et al. 1997 ).

The HD, a 60-amino-acid DNA binding domain, also found in other classes of transcription factors (DUBOULE 1994 ), presents a stereotyped helical structure and its mode of interaction with DNA is largely invariant (GEHRING et al. 1994 ). The amino acid at position 50 in the third helix, also named the recognition helix, plays a fundamental role in DNA binding specificity (TREISMAN et al. 1989 ; SCHIER and GEHRING 1992 ). This amino acid is always a glutamine in Hox proteins, which likely accounts for the very similar or identical DNA binding properties of Hox proteins in vitro. This therefore raises the question of how Hox proteins, in spite of equivalent biochemical properties, select specific sets of transcriptional targets and reach distinct regulatory effects in vivo (HAYASHI and SCOTT 1990 ).

Studies of the Drosophila Extradenticle (Exd) HD-containing protein and of the vertebrate related Pbx proteins have provided substantial support for the idea that protein cofactors contribute in multiple ways to specify the function of Hox proteins (MANN and CHAN 1996 ). First, Exd can improve the affinity of a Hox protein to DNA. This is best illustrated in the case of Labial (Lab), a Hox protein that binds DNA very poorly on its own. When engaged in a complex with Exd, Lab adopts a conformation that overcomes an inhibitory activity of the hexapeptide motif (a short evolutionarily conserved motif lying upstream of the HD) on DNA binding and results in an increased affinity to target DNA (CHAN et al. 1996 ). Second, association to Exd can modify DNA binding specificity. In vitro, while Hox proteins all bind to very similar, if not identical, short TAAT core sequence, Hox-Exd complexes recognize a larger motif, TGATNNATNN, where the identity of central NN nucleotides depends of the Hox protein engaged in the complex (CHAN and MANN 1996 ). Such motifs exist in regulatory sequences of Hox downstream target genes [also termed Hox response elements (HREs)] and were found to be functional in vivo, since modifying the identity of the NN central nucleotides switches the responsiveness of the HRE from one Hox protein to another (CHAN et al. 1997 ). Another HD-containing protein, Homothorax (Hth) in Drosophila or Meis proteins in vertebrates, can associate to Exd/Pbx-Hox complexes to further improve DNA binding specificity (MANN and AFFOLTER 1998 ). Third, Exd has recently been proposed to modify transregulatory properties of Hox proteins. The proposal relies on the in vivo analysis of synthetic enhancers containing binding sites of identical affinities for either the Deformed (Dfd) or the Dfd-Exd complex (LI and MCGINNIS 1999 ; LI et al. 1999A ). The study of the Exd/Pbx family of proteins using bona fide or closely derived synthetic HREs has shown that diverse regulatory mechanisms are deployed to modulate Hox protein function in vivo.

Characterized HREs have been used in other approaches. For instance, a functional dissection of an HRE from the Dfd autoregulatory region has provided evidence that non-HD-containing proteins act as Hox cofactors (LI et al. 1999B ). HREs therefore appear as essential molecular tools to investigate the mechanisms conferring functional specificity to Hox proteins. In this article, we describe a genetic screen designed to identify functions that interfere with two Drosophila Hox proteins in controlling the transcription of target genes. Green fluorescent protein (GFP) reporter constructs were used in living embryos to follow the activity of HREs from Abdominal-A (AbdA) and Abdominal-B (AbdB) target genes, wingless (wg) and empty spiracles (ems). As the activity of an HRE reports the transcriptional activity of the Hox regulator, the rationale was to consider that mutations in genes coding for interfering functions should result in GFP pattern changes. Using a collection of deficiencies that uncover 60% of the Drosophila genome, we identified 33 genomic regions that affect the GFP fluorescence patterns. Most of them were analyzed using smaller deficiencies, P-element- or EMS-induced mutations, to identify genetic loci responsible for the observed phenotypes. As a result, we found eight candidate modifier genes not previously known to be involved in the homeotic regulation of wg and ems.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly stocks:
The 24B-Gal4 and arm-Gal4 transgenic lines were used as mesodermal and ubiquitous drivers. UAS-ttk p69 and UAS-ttk p88 transgenic lines were kindly provided by C. Klämbt, UAS-brk by M. Affolter, Hs-abdA and Hs-AbdBm by S. Sakonju, and the UAS-GFP (EGFP variant) transgenic lines by C. Desplan. Deficiency stocks and mutations were obtained from the Bloomington Stock Center, except P5975, which was from the Scott laboratory, and tsl⁴ and dally^E385, which were obtained from N. Dostatni and S. Kerridge. Cytology was taken from FlyBase and LINDSLEY and ZIMM 1992 . Deficiencies and mutations that alter wg HRE or ems HRE activity are listed in Table 1 and Table 2. The list of deficiencies and mutations that do not affect wg HRE or ems HRE activity are available upon request.

Establishment of ems-Gal4 and wg-Gal4 transgenic lines:
The identification of the wg enhancer that reproduces expression in parasegment (PS) 8 of the visceral mesoderm (VM) under the control of AbdA will be described elsewhere (A. GRIENENBERGER, S. MERABET, J. MANAK, I. ILTIS, H. BERENGER, M. P. SCOTT, J. PRADEL and Y. GRABA, unpublished observations). The 540-bp wg HRE was cloned from a Bluescript KS+ (Stratagene, La Jolla, CA) vector into a PCasper2Gal4 (a gift from L. Fasano) vector. This was achieved by directional cloning of a PvuII/SacII restriction fragment containing the wg HRE into a BglII Klenow-filled SacII linearized PCasper2Gal4 vector. The 1.2-kb ems HRE, provided by W. McGinnis in the Bluescript vector, was transferred following NotI/EcoRV restriction into a StuI/EcoRI linearized PCasper2Gal4 vector. Plasmid DNA for each construct was prepared and used for P-element-mediated germ-line transformation, as described by RUBIN and SPRADLING 1982 . Several transgenic lines with insertions on chromosomes X, II, and III were established. Homozygous viable insertions were selected, when possible, to generate recombinant chromosomes II and III (RII and RIII) that simultaneously carry the UAS-GFP, wg-Gal4, and ems-Gal4 transgenes.

The GFP reporter screen:
A homozygous viable RII chromosome was used for screening deficiencies of the third chromosome. All third chromosome deficiency stocks were initially crossed to introduce a Cy balancer. Fly lines homozygous for RII and heterozygous for each deficiency were established after two individual crosses (see Fig 3). The progeny of these stocks carry two copies of RII and a quarter of them are homozygous for a third chromosome deficiency. No homozygous viable reporter RIII chromosome was recovered. The screen of deficiencies of the second chromosome was done according to the same scheme as above. However, as a consequence of the lack of a homozygous viable RIII, lines carrying deficiencies of the second chromosome were heterozygous for RIII. Consequently a quarter of the progeny did not carry RIII. This situation does not add any complication to scoring of ectopic fluorescence. However, a deficiency that leads to the loss of expression should display half of the embryos with no fluorescence, instead of a quarter. This quantitative difference might appear difficult to evaluate in a genome-wide screen. This is an apparent difficulty as the reporter chromosome simultaneously drives wg- and ems-dependent fluorescence: scoring loss of one signal in the presence of the other allowed identification of deficiencies responsible for loss of one HRE response. To screen for X chromosome deficiencies, males homozygous for RII and heterozygous for RIII were crossed with females heterozygous for each deficiency. In the progeny, all embryos carried at least one copy of the reporter chromosome, and half of them were hemizygous for the deficiency. Overnight egg collections were done at 29° from agar plates and dechorionated in 50% bleach. The embryos were directly analyzed under a Leica MZ FLIII binocular at a x16 magnification. For the pilot experiments, overnight collection of Hs-abdA and Hs-AbdB eggs were placed in a 37° water bath for 30 min. A second identical heat pulse was given after 1 hr. Embryos were examined for GFP fluorescence 3 hr later.

View larger version (113K): In this window In a new window Download PPT slide   Figure 1. The GFP reporter constructs accurately reproduce wg and ems Hox-dependent expression. In all figures, anterior is to the left. (A and B) Wild-type embryonic expression of ems (A) and wg (B) revealed by in situ hybridization. Arrows indicate the cells in which Wg and ems are controlled by AbdA (PS8) and AbdB (A8). (C and D) lacZ expression (in situ detection of lacZ mRNA) driven by the HREs from ems (C) and wg (D). The patterns accurately and exclusively reproduce the aspects in the expression patterns that are controlled by Hox proteins. (E and F) The GFP signal observed in ems-Gal4/UAS-GFP or wg-Gal4/UAS-GFP embryos are located respectively in the spiracular chamber (E) and in PS8 of the midgut VM (F).

View larger version (57K): In this window In a new window Download PPT slide   Figure 2. Loss and gain of function of abdA and AbdB modify the activity of the GFP reporter constructs. (A) The GFP signal in a line homozygous for the RII chromosome. The signal corresponds to the sum of ems HRE- and wg HRE-driven fluorescence (compare to Fig 1E and Fig F). (B) GFP expression in the VM is absent in abdAJX2 mutant embryos. Expression in the spiracular chamber is not affected. (C) In AbdBM1 homozygous mutant embryos, A8 expression is lost, while expression in the midgut is not affected. (D) Ubiquitous expression of abdA induces ectopic GFP expression in the midgut (arrows). (E) Ubiquitous expression of AbdB induces ectopic expression in all trunk segments (arrows). Germ-band retraction in E is blocked as a result of heat-shock-induced AbdB expression.

View larger version (17K): In this window In a new window Download PPT slide   Figure 3. Schematic outline of the genetic screen. Crosses were performed to screen deficiencies of the third chromosome (A) and the X chromosome (B). R represents reporter chromosomes carrying the UAS-GFP, ems-Gal4, and wg-Gal4 transgenes. B3 represents a balancer of the third chromosome.

Whole-mount in situ hybridization and plasmid rescue:
Except for UAS/GFP and UAS/Gal4 genotypes for which collections were performed at 29°, embryos were collected at 22°. Embryos were prepared according to TAUTZ and PFEIFLE 1989 and in situ hybridization to whole embryos using digoxigenin DNA- or RNA-labeled probes was performed according to VINCENT et al. 1994 . After alkaline phosphatase detection, embryos were mounted in 80% glycerol, 100 mM Tris, pH 7.5 and observed under an Axiophot Zeiss microscope using Nomarski optics. Digoxigenin RNA-labeled probes were generated from cDNA clones according to the manufacturer's protocol (Boehringer Mannheim, Mannheim, Germany). Genomic DNA from the P5975 transgenic line was digested with EcoRI or BamHI, ligated, and transformed according to standard procedures. The genomic fragments of 3 and 5 kb obtained from the EcoRI and BamHI plasmid rescue experiments were used to screen a Drosophila embryonic cDNA library (a gift from N. Brown). The sequence of the isolated cDNAs matches the 5' coding sequence of the dally gene.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Generation of GFP reporter lines monitoring AbdA and AbdB transcriptional activities:
During embryogenesis, ems is expressed in a complex pattern, in the head, abdominal trachea, and posterior spiracles (Fig 1A; WALLDORF and GEHRING 1992 ). Transcription in the posterior spiracle is restricted to cells that will contribute to the filzkörper, the posterior end of the larval respiratory system. This aspect of ems expression results from AbdB activation and depends on a 1.2-kb HRE (Fig 1C; JONES and MCGINNIS 1993 ). wg is also transcribed in a complex pattern, including striped expression in the ectoderm and localized expression in the foregut, hindgut, and midgut (Fig 1B; BAKER 1987 ). Expression in the midgut is restricted to VM cells of PS8 and is controlled by AbdA (IMMERGLUCK et al. 1990 ). The HRE responsible for this control, to be described elsewhere (A. GRIENENBERGER, S. MERABET, J. MANAK, I. ILTIS, H. BERENGER, M. P. SCOTT, J. PRADEL and Y. GRABA, unpublished observations), is a 540-bp element that accurately reproduces wg expression and regulation in PS8 of the developing midgut (Fig 1D).

Transgenic flies producing the Gal4 transcriptional activator under the control of ems or wg HREs were crossed with UAS-GFP flies and the resulting embryos were analyzed. For the ems-Gal4/UAS-GFP combination, the GFP signal first appears during germ-band retraction, 2–3 hr after the expression of ems is detected by in situ hybridization. One copy of the transgene is sufficient to produce a clear signal (Fig 1E), restricted to cells that will form the filzkörper. In wg-Gal4/UAS-GFP embryos, GFP is first detected in germ-band-retracted embryos, again 2–3 hr after wg transcripts are detected in the VM by in situ hybridization. The embryonic midgut displays high autofluorescence, which compromises the detection of GFP fluorescence when only one copy of the wg-Gal4 transgene is present (not shown). With two copies of the transgene, GFP fluorescence, accurately driven in PS8 of the VM, surpasses the midgut autofluorescence (Fig 1F). In summary, transgenic embryos for wg-Gal4/UAS-GFP or ems-Gal4/UAS-GFP accurately reproduce the expression of wg or ems that depends on AbdA or AbdB.

For the purpose of the screen, we generated second (RII) and third (RIII) reporter chromosomes that simultaneously carry ems-Gal4, wg-Gal4, and UAS-GFP. As expected, embryos containing the three transgenes display GFP profiles that correspond to the addition of signals driven by ems HRE and by wg HRE (Fig 2A). This allows us to follow simultaneously the activities of AbdB and AbdA on the expression of ems and wg HREs. To validate the experimental design, we tested the effect of loss and gain of AbdA and AbdB functions on the reporter RII and RIII lines. In embryos deficient for abdA or AbdB, wg and ems HREs are no longer active (Fig 2B and Fig C), which is consistent with the loss of wg and ems gene transcription in these mutants (IMMERGLUCK et al. 1990 ; JONES and MCGINNIS 1993 ). Conversely, providing AbdA or AbdB ubiquitously induces ectopic activation of wg or ems HREs (Fig 2D and Fig E) in patterns that resemble ectopic transcription of the endogenous genes. These reporter chromosomes are thus suited to detect changes in AbdA and AbdB activities and therefore can be used to screen for modifier genes.

Screen for genomic regions required for accurate activity of wg HRE or ems HRE:
To identify recessive mutations that modify ems or wg HRE activity, we first used the deficiency kit available from the Bloomington Stock Center. Crosses needed to perform the screen are given in Fig 3 and described in MATERIALS AND METHODS. Although the experimental design has the potential to search for loss and gain of wg HRE activity, the weak level of GFP fluorescence driven by the wg HRE relative to midgut autofluorescence led us to consider only deficiencies causing ectopic wg HRE activity. In contrast, ems HRE fluorescence is strong enough to screen deficiencies able to induce or to repress ems HRE.

Deficiencies leading to ectopic activation of the wg HRE:
Eighteen deficiencies leading to ectopic GFP signals were selected (Table 1). As wg HRE activity is first observed at the end of embryogenesis in a differentiated organ, we paid attention to whether the ectopic GFP signal occurred in a midgut of normal or affected morphology. Eight deficiencies [Df(1)DCB1-35b, Df(2R)PC4, Df(3L)GN24, Df(3L)vin4, Df(3R)by10, Df(3R)e-N19, Df(3R)crb87-4, and Df(3R)Dr-rv¹] induce variable defects in midgut morphogenesis. In embryos homozygous for Df(1)DCB1-35b and Df(3L)GN24, VM cells do not migrate dorsoventrally and the dorsal midgut closure fails, resulting in apparently larger wg-expressing cell clusters. The altered fluorescence pattern seen in these embryos (shown for Df(3L)GN24 in Fig 4A) is not likely to reflect a change in the activity of wg HRE, but rather results from altered midgut morphogenesis. For the other 6 deficiencies, the origin of midgut defects is not as well identified, and it could not be concluded whether altered fluorescence patterns result from impaired morphogenesis or defects in wg HRE activity per se. For 10 deficiencies, midgut morphogenesis was not altered, at least with regard to wg HRE pattern (Table 1). Among these deficiencies, three classes of phenotypes can be distinguished. The first, associated with Df(3L)Ten-m-AL29, results in a strong and enlarged GFP signal (Fig 4B). Embryos homozygous for two other deficiencies, Df(3R)B81 and Df(3R)96B, also display an extended GFP signal (Fig 4C and Fig D). The second class provokes ectopic fluorescence in cells that are not adjacent to the normal site of wg expression. Five deficiencies [Df(1)A113, Df(1)N73, Df(1)ct-J4, Df(1)B, and Df(3R)6-7] produce slightly different versions of this phenotype [shown for Df(1)ct-J4 and Df(1)B in Fig 4E and Fig F]. Finally, the third class displays ectopic activation in tissues other than the VM. Two deficiencies, Df(3L)iro-2 and Df(1)v-N48, lead to ectopic expression in the somatic musculature and in the midgut endoderm, respectively (Fig 4G and Fig H).

View larger version (132K): In this window In a new window Download PPT slide   Figure 4. Deficiencies leading to ectopic activation of the wg HRE. (A) In Df(3L)GN24, the GFP signal in the intestine is expanded (arrows) compared to wild-type embryos. Dorsal midgut closure is defective in embryos homozygous for this deficiency. (B) In Df(3L)Ten-m-AL29, comparison of the intensities of the wg HRE- and ems HRE-driven GFP signals shows that wg HRE activity (arrows) is enhanced and occurs in an area broader than that in wild-type embryos. (C–F) Df(3R)B81, Df(3R)96B, Df(3R)ct-J4, and Df(1)B, respectively. The wg HRE is activated in an enlarged domain (C and D) or in noncontiguous domains (E and F) of the VM. (G) In Df(3L)iro-2, ectopic GFP signal is observed outside the VM (arrow) in a tissue that most likely corresponds to somatic mesoderm. (H) In Df(1)v-N48, ectopic GFP signal is observed within endodermal cells of the intestine (arrow).

To narrow down the genomic regions responsible for these phenotypes, smaller or partially overlapping deficiencies were analyzed for 11 regions. Three small deficiencies that reproduce phenotypes observed in the primary screen were thus selected: Df(2R)P34, Df(3L)Ten-m-AL1, and Df(3R)awd-KRB mimic the effects of Df(2R)PC4, Df(3L)Ten-m-AL29, and Df(3R)B81, respectively, which allows positive assignment of the phenotypes to genes located in genomic regions 55E2;55F, 79E1;79E8, and 100C;100D. In eight cases, we were unable to recover the original phenotypes using smaller deficiencies. We could, however, narrow down, by exclusion, the genomic regions that putatively contain functions critical for wg HRE control. The refined genomic regions are given in Table 1.

Deficiencies leading to loss or ectopic activation of the ems HRE:
As ems expression precedes filzkörper organogenesis, it was possible to follow ems HRE activity irrespective of any defect in filzkörper morphogenesis. Expression was analyzed as early as possible during germ-band retraction, before filzkörper cells become organized into a morphological recognizable unit, and later in embryogenesis, to confirm early misregulation of the ems HRE as well as to detect later defects. Four deficiencies were found to cause ectopic activation of the ems HRE (Table 2). Embryos homozygous for Df(3L)fz-GF3b, Df(2R)X1, and Df(1)C52 show altered fluorescence during early germ-band retraction, as soon as the GFP signal is detected. Df(3L)fz-GF3b and Df(1)C52 produce very similar phenotypes, with GFP expression initiated in an enlarged cell cluster [shown in Fig 5A for Df(1)C52]. Since the ems HRE is active in a broader region and in nonclustered cells, Df(2R)X1 embryos display a somewhat different phenotype, which is still observed at the end of embryogenesis (Fig 5B). The fourth deficiency, Df(1)BA1, leads only to late defects in ems HRE activity (Fig 5C): the pattern during early germ-band retraction is normal (not shown); GFP fluorescence becomes detected at ectopic positions at the end of germ-band retraction, in cells that surround the filzkörper and most probably correspond to cells contributing to the stigmatophore, the external part of the posterior spiracles. Five deficiencies [Df(1)JC19, Df(1)64c18, Df(1)RK2, Df(3L)HR119, and Df(2L)H20] induce ectopic fluorescence in the head. We did not consider that these deficiencies specifically alter the activity of the ems HRE, but rather that a head-to-tail respecification has occurred in the mutant embryos. Consistent with this, Df(3L)HR119 (Fig 5D) uncovers the BicaudalD gene, the mutation of which results in two-tailed embryos. We did not list these deficiencies in Table 2. While only 4 deficiencies were identified as specifically inducing ectopic activation of the ems HRE, 11 were found to severely reduce [Df(2L)al, Df(2L)dp-79b, Df(2L)TW84, Df(3L)lxd6, and Df(3R)B81] or abolish [Df(2L)H20, Df(2R)H3E1, Df(2R)Jp1, Df(2R)AA21, Df(3L)h-i22, and Df(3R)e-R1] its activity [ Table 2; shown in Fig 5E and Fig F, and Fig 4C for Df(2L)dp-79b, Df(3R)e-F1, and Df(3R)B81].

View larger version (113K): In this window In a new window Download PPT slide   Figure 5. Deficiencies modifying the activity of the ems HRE. (A) In Df(1)C52, the GFP signal is broader than that in wild-type embryos (arrows). (B) In Df(2R)X1, ectopic GFP signal appears in nonclustered cells in a large region of the posterior part of the embryo (arrows). (C) Df(1)BA1 affects ems HRE activity only by the end of embryogenesis. The GFP signal appears broader (arrows) and most likely includes cells of the stigmatophore. (D) In Df(3L)HR119, ectopic fluorescence is observed in the anterior region of the embryo (arrow). The morphology of the embryo indicates a general anterior-to-posterior transformation, leading to filzkörper development in the head. (E and F) Df(2L)dp-79b and Df(3R)e-R1, respectively. Arrows indicate a reduction (E) or a complete loss (F) of the ems HRE activity.

Smaller and partially overlapping deficiencies were analyzed for 12 genomic regions. In eight cases, deficiencies produce phenotypes similar to primary deficiencies, narrowing down the critical genomic regions. In four cases, the smaller deficiencies tested did not alter ems HRE activity, which allows restricting, by exclusion, the critical genomic interval. Refined genomic regions are given in Table 2.

Candidate modifier genes of ems HRE and wg HRE activities:
To identify genetic loci that control ems and wg HREs, we tested a set of mutations that map within 17 reasonably small genomic regions identified from the deficiency screen. For regions containing a large number of mutations, we focused mainly on P-element insertions for which a ß-galactosidase reporter expression in the midgut or posterior spiracles was already reported. Lines carrying reporter and mutant homozygous chromosomes were analyzed. For 8 of the 17 genomic regions, we identified loci whose mutations alter wg HRE or ems HRE activity (Table 1 and Table 2).

Identification of modifiers of the wg HRE:

Nine of the 18 genomic regions initially found to contain recessive loci controlling wg HRE were screened. In four cases, our nonexhaustive search did not identify the modifier genes (Table 1). For the five remaining regions (5C6; 5D1, 7A6;7B2, 93D2;94, 96A21;96C2, and 100C;100D), mutations in single loci causing ectopic wg HRE activity were recovered (Fig 6, A–D). To assess whether the wg HREs accurately reflect the regulation of the endogenous gene, we tested wg expression in mutant embryos. We found very similar ectopic expressions of wg and wg HRE in all cases but one [l(3)01207, see below]. Finally, to ascertain that the candidate modifier gene is responsible for the observed phenotype, the effect of other alleles was determined and gain-of-function experiments were performed when possible.

View larger version (114K): In this window In a new window Download PPT slide   Figure 6. Mutations modifying activities of wg HRE or ems HRE. (A) tsl1, an EMS-induced mutation of tsl, causes ectopic GFP signal in the VM. (B) EP1201, a P-element insertion in a gene predicted to encode a calcium-binding protein, causes ectopic signal in cells anterior and noncontiguous to PS8. (C) ttk1e11, a P-element-induced deletion affecting the Ttk p69 isoform, leads to wg HRE activation in an expanded domain (arrows) compared to wild type. (D) l(3)01207, a P-element insertion in a gene predicted to encode an MKP, promotes wg HRE activity (arrows) in multiple domains. (E and F) Partial and total loss of the ems-driven GFP signals (arrows) in embryos homozygous for insertion in ds (ds05142, E) and in dally (P5975, F).

fs(1)K254¹ is a thermosensitive mutation that leads to pupal lethality at elevated temperatures and female sterility at lower temperatures. This EMS-induced mutation is likely to represent a hypomorphic allele of a gene whose molecular nature is unknown. Four other mutations, l(3)01207, EP1201, tsl¹, and ttk^1e11, are in transcriptional units already known or predicted from genome sequence data.

l(3)01207 mutation causes an enlargement of the wg domain (Fig 7B), while it induces a much broader wg HRE derepression (Fig 6D). Thus, although ectopic expression occurs in both cases, changes in HRE activity do not mimic changes in the endogenous wg pattern. We also noted that the GFP pattern in l(3)01207 mutant embryos is different from that seen in the corresponding primary deficiency (Fig 4D). This may suggest that more than one gene under the deficiency could modify wg HRE activity. l(3)01207 is a P-element insertion that has been mapped within the coding sequence of a gene encoding a putative mitogen-activated protein (MAP) kinase phosphatase (MKP). The lack of other MKP alleles prevented us from confirming its involvement in the control of wg HRE.

View larger version (110K): In this window In a new window Download PPT slide   Figure 7. Genes altering wg expression in the VM. wg transcripts were revealed by in situ hybridization. The arrow points to wg transcripts in midgut VM. (A) Wild-type wg expression in PS8. (B) The domain of wg expression is extended in l(3)01207 mutant embryos. (C and D) CBP represses wg expression in the VM. In EP1201 mutant embryos, ectopic expression is noncontiguous and anterior to PS8 (C). Ubiquitous expression of CBP in 24B-Gal4/EP1201 embryos results in the loss of wg expression (D). (E) The domain of wg expression in the VM is enlarged in tsl4 mutant embryos. (F) Ubiquitous expression of a constitutively active form of Ras in arm-Gal4/UAS-rasV12 embryos results in the loss of wg expression in the VM. (G and H) Ttk p69 represses wg expression in the VM. The ttk1e11 mutation leads to an enlarged wg expression domain (G), while ubiquitous expression of ttk p69, in arm-Gal4/UAS-ttk p69 embryos, results in the loss of wg expression in the VM.

EP1201 is a viable P-element insertion 30 bp upstream of the first exon of a transcriptional unit encoding a putative calcium-binding protein (CBP). This P-element insertion is homozygous viable, indicating that the deregulation of wg induced by EP1201 in the VM (Fig 7C) does not cause lethality. Deficiencies that uncover CBP do not alter midgut morphogenesis (BILDER and SCOTT 1995 ), suggesting that the mutation of CBP does not cause any visible midgut phenotype. The closest predicted transcription units are located 15 kb upstream and 10 kb downstream of the P-element insertion site, indicating that the EP1201 mutation most likely affects the activity of the CBP gene. This is supported by the strong reduction of CBP expression in EP1201 mutant embryos (data not shown). To further confirm a role for CBP in wg regulation, we took advantage of the EP line to overexpress CBP using the UAS/Gal4 system. EP1201/24B-Gal4 embryos, in which CBP is provided in the whole mesoderm, no longer express wg in the midgut (Fig 7D). Loss and gain of CBP activity therefore produce opposite effects on wg expression, strongly arguing that CBP is involved in the regulation of wg in the VM.

tsl¹ is a point mutation in torso like (tsl). Because tsl alleles have been described so far as displaying a strictly maternal effect, with no identified zygotic requirement, it has been somewhat unexpected to find this gene in our screen. Of note, however, tsl is expressed in the embryonic VM (MARTIN et al. 1994 ), and alteration of wg expression in the VM does not necessarily result in a visible midgut phenotype or in lethality, as shown by the CBP mutation. To further assay whether tsl can modify wg HRE activity, we used another EMS-induced allele of tsl, tsl⁴. The two alleles cause similar enlargement of wg expression (shown for tsl⁴ in Fig 7E). Tsl is a ligand of the Torso receptor tyrosine kinase (RTK) receptor, suggesting that Ras- and MAP kinase (MAPK)-mediated signaling is involved in wg regulation. We tested this hypothesis and found that expressing a constitutive active form of Ras, Ras^V12, abolishes wg transcription in the VM (Fig 7F). This observation, together with the GFP ectopic expression seen in tsl alleles, supports a function for Ras/MAPK signaling in counteracting AbdA-mediated activation of wg in the VM.

The repressive role of tramtrack (ttk), suggested by the extended wg expression in ttk^1e11 mutant (Fig 7G), was further tested by gain-of-function experiments. ttk encodes two isoforms of a well-characterized zinc finger containing DNA binding transcription factors, p69 and p88 (XIONG and MONTELL 1993 ). ttk^1e11 is a mutation that specifically affects the function of the p69 isoform. Ubiquitous expression of Ttk p69, in UAS-ttk p69/arm-Gal4 embryos, results in loss of wg expression in the VM, supporting repression by this isoform (Fig 7H). We next addressed whether the Ttk p88 could also act as a repressor by analyzing wg transcription in a mutant specific for p88, ttk¹, as well as after ubiquitous expression of p88. In both cases, the expression of wg was not affected (not shown), indicating that Ttk p69 and not Ttk p88 represses wg in PS8 VM. Finally, we assayed the effect of ubiquitously provided Ttk p69 on the expression of dpp, an Ultrabithorax target gene in the VM. In arm-Gal4/UAS-ttk p69 embryos, dpp is still expressed (not shown), indicating that Ttk p69 does not repress dpp but specifically represses wg in the VM.

Identification of modifiers of the ems HRE:

Nine of the 15 genomic regions modifying ems HRE activity were screened to identify candidate modifier genes (Table 2). For five of these regions, we did not recover loci whose mutations induce either loss or ectopic GFP signals. In the four remaining genomic regions (21B8-C1;21C8-D1, 38A1;38A3, 66E1;66E2, and 100C;100D), mutations that cause partial or complete loss of ems HRE activity were identified: ds⁵¹⁴² (Fig 6E), P5975 (Fig 6F), scw¹¹, and ttk^1e11 (not shown).

The scw¹¹, ds⁵¹⁴², ttk^1e11 mutations identify screw (scw), dachsous (ds), and ttk p69 as candidate modifiers. Scw is a ligand for transforming growth factor (TGF)-ß receptor signaling and Ds is a calcium-dependent cell adhesion molecule. P5975 is a P-element insertion that we cytogenetically mapped to 66E. Isolation of flanking genomic DNA and sequencing showed that P5975 is an insertion within the 5' end of the coding sequence of dally. To check whether the loss of ems HRE activity reflected loss of endogenous ems expression, ems transcription was analyzed by in situ hybridization in the four mutants. With the exception of scw¹¹, which globally altered morphogenesis and thus made it difficult to unambiguously recognize ems-expressing cells, we concluded that ems transcription (shown in Fig 8B and Fig C, for ds⁵¹⁴² and dally ^E385) and ems HRE respond similarly in mutants for the ds, scw, dally, and ttk candidate modifier genes.

View larger version (69K): In this window In a new window Download PPT slide   Figure 8. Genes altering ems expression in A8 ectoderm. ems transcripts were detected by in situ hybridization. The arrow points to the eighth tracheal pit and the arrowhead to the position of filzkörper prospective cells. (A) Wild-type ems expression. (B) In ds5142 mutant embryos, ems expression in prospective cells of the filzkörper is reduced. (C) Loss of ems expression is observed in dallyE385 mutant embryos. (D) Expression of Brk in prospective cells of the filzkörper, in ems-Gal4/UAS-brk embryos, results in strong reduction of ems expression.

Our observation that scw and ttk are required for ems transcriptional control is consistent with already reported roles for these two genes in filzkörper morphogenesis (XIONG and MONTELL 1993 ; ARORA et al. 1994 ). We therefore did not analyze this question further. As the scw¹¹ mutation suggests an involvement of the TGF-ß transduction pathway in the control of ems, we analyzed the expression of ems after localized expression of Brinker (Brk), a potent repressor of Dpp/TGF-ß target genes. In ems-Gal4/UAS-brk embryos, expression is lost from the prospective filzkörper cells (Fig 8D), supporting a Dpp/TGF-ß signaling requirement for the activation of ems in the spiracular chamber.

ds and dally were not previously linked to filzkörper morphogenesis. To further establish the involvement of dally, another EMS-induced loss-of-function allele, dally^E385 (S. KERRIDGE, personal communication), was analyzed and found to induce loss of ems expression similar to that of P5975. ds⁵¹⁴² is a P-element insertion in the first intron of the ds transcriptional unit. We did not test other ds mutations for an effect on ems expression, because the strongest available alleles all derive from the ds⁵¹⁴² chromosome. Viable excisions of the P element were frequently obtained (J. P. COUSO, personal communication), which indicates that the ds⁵¹⁴² chromosome does not contain mutation in another essential gene, arguing thus that ds is most likely the gene responsible for the loss of ems expression in ds⁵¹⁴² homozygous embryos.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A novel genetic screen to identify modifiers of embryonic Hox gene function:
Our knowledge of how Hox proteins gain specificity in vivo has significantly increased these past few years. Molecular approaches have focused mostly on the contribution of two cofactors, Exd/Pbx and Hth/Meis (MANN and AFFOLTER 1998 ). It seems probable, however, that other factors contribute to modify the DNA binding properties of Hox proteins in vivo and that other levels of regulation are most certainly crucial for Hox gene function. Although genetics has the potential to identify genes that functionally interfere with Hox genes, irrespective of the molecular mechanisms involved, only a few genetic screens have been performed. Screens reported so far were based on the modification of adult homeotic phenotypes and were designed to identify dominant dose-sensitive modifiers of Hox gene from the Antp complex, Antp, Dfd, and Proboscipedia (KENNISON and RUSSELL 1987 ; KENNISON and TAMKUN 1988 ; HARDING et al. 1995 ; BOUBE et al. 1997 ; GELLON et al. 1997 ; FLORENCE and MCGINNIS 1998 ).

This study reports a screen for modifiers of two genes from the Bithorax complex, abdA and AbdB, which presents several differences from screens for Hox modifiers conducted so far. First, taking advantage of GFP to monitor the activity of bona fide HREs in living embryos, our screen searched for modifiers that affect Hox gene function during embryogenesis. This was not attempted in the past, most probably because of difficulty in scoring embryonic homeotic phenotypes in the context of a genome-wide screen. Second, while previous screens were designed to identify dominant modifiers of Hox gene function, our screen identifies recessive modifier genes. Our screening conditions are therefore genetically distinct, allowing recovery of genes that would not be identified in a dominant dose-sensitive screen. Third, previous screens were based on the modification of a homeotic phenotype, which can result from misregulation of a large number of Hox downstream target genes. Our screen is based on changes in the regulation of individual target genes. In principle it could allow the identification of genes that would be specifically required for only a small subset of the morphogenetic program controlled by a Hox protein. This type of "reporter genetic screen" appears suited to identify modifiers of any transcription factor, providing that a bona fide in vivo response element has been identified. In particular, it would be of special interest when the inactivation of the transcription factor does not produce any visible phenotype, which is best illustrated by the CBP mutation that alters wg expression but does not apparently cause a midgut defect.

In this study we surveyed 60% of the genome and found 11 genomic regions acting as recessive activators of ems HRE, 4 acting as recessive repressors of ems HRE, and 18 acting as recessive repressors of wg HRE. A comprehensive summary of the wg and ems screens is given in Fig 9. So far, the only known gene in addition to AbdB required for ems activation is lines (CASTELLI-GAIR 1998 ). Df(2R)H3E1 that uncovers lines has been recovered from the screen for AbdB modifiers. A search for discrete mutations that reproduce the deficiency phenotypes allowed identification of four ems HRE modifier genes: dally, ds, scw, and ttk. Although ttk and scw have already been linked to filzkörper development, none of the four genes had previously been involved in the control of ems expression in posterior spiracles. The screen for AbdA modifiers was restricted to genomic regions leading to ectopic activation of the wg HRE, therefore containing functions that repress the enhancer. Accordingly, genomic regions or genes already known to play a role in wg activation, such as abdA, exd, hth, or genes coding for components of the Dpp signaling pathway, were not recovered. Five mutations at specific loci reproduce the phenotypes caused by original deficiencies. Four of these mutations identify tsl, ttk, and genes encoding a putative MPK and a putative CBP as candidate modifiers of wg HRE. None of these genes has so far been involved in the regulation of wg in the VM.

View larger version (41K): In this window In a new window Download PPT slide   Figure 9. Genomic regions and genes modifying activities of the wg HRE and ems HRE. The five major chromosome arms are schematized. Numbers represent the cytological divisions on each chromosome. Open boxes represent deficiencies classified as having no effect on wg HRE or ems HRE activity. Deficiencies that result in early embryonic lethality or in head-to-tail respecification have arbitrarily been classified in the "no effect" category. Green dotted boxes and green boxes, respectively, identify deficiencies that lead to ectopic activation of the wg HRE, with or without significant alteration of midgut morphogenesis. Red and red dashed boxes represent deficiencies that lead to ectopic or loss of ems HRE activity, respectively. Blue boxes represent deficiencies that simultaneously alter wg HRE and ems HRE activity. Stars indicate location of mutations that modify wg HRE or ems HRE.

The wg enhancer screen deserves an additional comment. Df(1)N73 has been reported to promote the formation of an additional central midgut constriction anterior to the endogenous one (BILDER and SCOTT 1995 ). As the central constriction forms at the boundary of wg and dpp expression domains, the prediction is that the additional constriction originates from a new dpp/wg boundary created by an ectopic expression of wg. Consistent with this prediction, we found that Df(1)N73 and the EMS mutation fs(1)K254¹ that maps within this genomic region led to ectopic wg HRE activity, suggesting that ectopic wg expression might be responsible for the phenotype described by BILDER and SCOTT 1995 . However, we also identified deficiencies that led to ectopic activity of the wg HRE, but that were not recorded in the screen of BILDER and SCOTT 1995 to produce an extra constriction. If in these deficiencies ectopic wg expression follows ectopic wg HRE activity, it would suggest that ectopic expression of wg does not necessarily result in an extra constriction, indicating that precisely defined site, level, and timing of wg expression might be required for constriction formation.

In summary, our genetic screen identified eight candidate modifier genes controlling the activity of the ems and wg HREs and the expression of the endogenous wg and ems genes as well. It appears therefore that many functions are required for accurate control of target genes by Hox proteins, which stands in contrast with the few Hox modifier genes identified so far.

Local specification of Hox gene function:
Hox proteins have distinct regulatory activities in different tissues and in different cell types within a tissue (CHAUVET et al. 2000 ). At the molecular level, this phenomenon is best illustrated by the regulation of wg by AbdA. First, while AbdA is expressed in the VM, somatic musculature, and ectoderm (KARCH et al. 1990 ), it activates wg in the VM exclusively (REUTER et al. 1990 ). Second, AbdA is expressed from PS8 to PS12 of the embryonic VM (BIENZ and TREMML 1988 ), but activates wg transcription in cells from PS8 only. The expression pattern of ems also illustrates that the transcriptional activity of a Hox protein is specified locally. AbdB is expressed in A7 and A8 abdominal segments in both the ectoderm and the mesoderm (TREMML and BIENZ 1989 ; DELORENZI and BIENZ 1990 ) and activates ems in only a few ectodermal cells of A8 (HU and CASTELLI-GAIR 1999 ). Thus, Hox proteins control their target genes in both a tissue-specific and a position (A/P coordinate)-specific manner. Little is known, however, about factors/mechanisms required for the regionalization of Hox protein transcriptional activity.

Like already reported screens for Hox modifiers, our screen has the potential to identify mutations in genes required for Hox gene expression and for Hox protein function. Four deficiencies [Df(3L)iro-2, Df(1)v-N48, Df(1)ct-J4, and Df(1)N73] led to an activation of wg HRE outside of the wild-type domain of AbdA expression, which might simply result from an ectopic expression of AbdA. The same can occur for the 11 deficiencies that cause loss of ems HRE activity. For the other selected deficiencies, HREs are derepressed within the normal domains of AbdA or AbdB expression, which might result rather from changes in AbdA or AbdB regulatory activities.

This also applies to wg HRE modifier genes. In embryos deficient for CBP, wg becomes activated anteriorly in PS7, outside the normal AbdA domain, which could simply result, as stated above, from an ectopic expression of AbdA in PS7. The situation seems more complicated, however, since dpp expression in the VM, which is repressed by AbdA, is not affected by the EP1201 mutation (data not shown). This suggests that the effect of this mutation is unlikely to result from ectopic AbdA expression. The three other candidate modifiers (MKP, Tsl, and Ttk), most likely affect AbdA regulatory activity, as their mutations induce ectopic expression in the normal domain of AbdA expression. Ttk p69 is a good candidate to provide such regionalizing function by directly controlling wg expression, as putative binding sites for this transcriptional repressor are present in the wg HRE sequence (not shown). The recent finding that Ttk p69 associates with the dMi2 subunit of the NuRD chromatin remodeling complex (MURAWSKY et al. 2001 ) suggests that the homeotic control of wg expression involves local modulation of chromatin structure.

In summary, the large number of genomic regions leading to ectopic activation of wg HRE and of ems HRE indicates that spatial limitation of Hox transcriptional activity is an important factor in the control of their function, although the mechanisms are not known. Our screen thus provides appropriate tools to investigate this question.

Signaling and Hox gene function:
Cell signaling and Hox transcriptional control are essential determinants for patterning and for coordinated regulation of many developmental processes. How these determinants functionally interact has been analyzed mostly in terms of reciprocal transregulation, control of Hox gene transcription by signaling pathways (REUTER et al. 1990 ; MALOOF et al. 1998 ; MILLER et al. 2001 ), and, conversely, control of genes encoding signaling molecules by Hox proteins (PANGANIBAN et al. 1990 ; MILLER et al. 2001 ). Functional interactions between Hox and signaling pathways, however, are not restricted to transcriptional cross-talks. For example, the pointed (pnt) and odd paired (opa) genes are activated in distinct domains of the VM, respectively in the third and fourth midgut chambers, through a concerted control by AbdA and signaling pathways: Wg allows AbdA to activate pnt, while Dpp prevents AbdA from promoting opa expression (BILDER et al. 1998 ). Several genes identified from our screen as candidate modifiers of ems or wg HRE activity are functionally connected to signal transduction pathways.

Three of the four candidate genes identified from the ems screen encode molecules acting in or acting in connection with signal transduction pathways. The Scw protein is a secreted factor of the TGF-ß family (ARORA et al. 1994 ). The loss of ems expression induced by Brk, a potent repressor of the Dpp/TGF-ß target gene, strongly supports this hypothesis. The involvement of additional signaling pathways in the regulation of ems is more indirectly suggested by the identification of ds and dally that act in connection with several signaling pathways. ds codes for a calcium-dependent cell adhesion molecule of the cadherin superfamily and genetically interacts with shotgun and rhomboid (STURTEVANT and BIER 1995 ; GREAVES et al. 1999 ), two genes involved in epidermal growth factor (EGF) signaling, as well as with armadillo (arm), which produces a nuclear effector of the Wg transduction pathway (GREAVES et al. 1999 ). dally encodes a heparin sulfate proteoglycan involved in the reception of Wg (LIN and PERRIMON 1999 ; TSUDA et al. 1999 ) and TGF-ß signals (JACKSON et al. 1997 ). Although additional experiments are required to firmly establish the involvement of the Wg and EGF pathways, the integration of multiple signals seems to be required for accurate ems regulation by AbdB.

Two modifier genes obtained from the wg screen are presumably involved in the signal transduction cascade. The first, tsl, encodes a ligand for the RTK Torso receptor (SAVANT-BHONSALE and MONTELL 1993 ) and the second encodes a putative MKP. Signaling by Ras/MAPK could thus be part of the genetic network that controls wg expression in the midgut, which we confirmed by showing that wg transcription is impaired by a constitutive active form of Ras. Interestingly, the Ras/MAPK pathway has been implicated in regulation of the Ubx and lab enhancer in the central midgut (SZUTS et al. 1998 ), and the ETS-domain-containing transcription factor Pointed, which acts as a nuclear effector of the Ras/MAPK pathway, is expressed in the third midgut chamber (BILDER et al. 1998 ).

Several modifiers of wg and ems HRE activities identified in this study encode molecules acting in signal transduction cascades. This indicates that signaling processes play important roles in the control of Hox gene function and extends previous observations from a screen for modifiers of a dominant Pb phenotype (BOUBE et al. 1997 ). Understanding how cell signaling and transcriptional control by Hox protein are mechanistically integrated requires further study. A recent report showed that nuclear effectors of signaling pathways directly coregulate Hox downstream targets (MARTY et al. 2001 ), indicating that, at least in some instances, they could act as Hox cofactors.

	ACKNOWLEDGMENTS

We thank S. Kerridge and J. Castelli-Gair for critical reading of the manuscript and K. Matthews and the Bloomington Drosophila Stock Center for providing the deficiency kit and for numerous additional fly stocks. We are also grateful to L. Fasano, C. Klambt, C. Desplan, N. Brown, W. McGinnis, S. Sakonju, and N. Dostatni for fly stocks and reagents. We also thank Marie Leborgne for screening some of the P-element insertions. This work was supported by the Centre National de la Recherche Scientifique, by grants from the Association pour la Recherche contre le Cancer (ARC) and the Ligue Nationale Contre Le Cancer, and by a fellowship from the Ministère de la Recherche et de la Technologie and ARC to S.M.

Manuscript received November 2, 2001; Accepted for publication June 11, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ARORA, K., M. S. LEVINE, and M. B. O'CONNOR, 1994  The screw gene encodes a ubiquitously expressed member of the TGF-beta family required for specification of dorsal cell fates in the Drosophila embryo. Genes Dev. 8:2588-2601.[Abstract/Free Full Text]

BAKER, N. E., 1987  Molecular cloning of sequences from wingless, a segment polarity gene in Drosophila: the spatial distribution of a transcript in embryos. EMBO J. 6:1765-1773.[Medline]

BATESON, W., 1894 Materials for the Study of Variation. MacMillan & Co., London.

BIENZ, M. and G. TREMML, 1988  Domain of Ultrabithorax expression in Drosophila visceral mesoderm from autoregulation and exclusion. Nature 333:576-578.[Medline]

BILDER, D. and M. P. SCOTT, 1995  Genomic regions required for morphogenesis of the Drosophila embryonic midgut. Genetics 141:1087-1100.[Abstract]

BILDER, D., Y. GRABA, and M. P. SCOTT, 1998  Wnt and TGFbeta signals subdivide the AbdA Hox domain during Drosophila mesoderm patterning. Development 125:1781-1790.[Abstract]

BOTAS, J., 1993  Control of morphogenesis and differentiation by HOM/Hox genes. Curr. Opin. Cell Biol. 5:1015-1022.[Medline]

BOUBE, M., C. BENASSAYAG, L. SEROUDE, and D. L. CRIBBS, 1997  Ras1-mediated modulation of Drosophila homeotic function in cell and segment identity. Genetics 146:619-628.[Abstract]

CASTELLI-GAIR, J., 1998  The lines gene of Drosophila is required for specific functions of the Abdominal-B HOX protein. Development 125:1269-1274.[Abstract]

CHAN, S.-K. and R. S. MANN, 1996  A structural model for a homeotic protein-extradenticle-DNA complex accounts for the choice of HOX protein in the heterodimer. Proc. Natl. Acad. Sci. USA 93:5223-5228.[Abstract/Free Full Text]

CHAN, S.-K., H. PÖPPERL, R. KRUMLAUF, and R. S. MANN, 1996  An extradenticle-induced conformational change in a HOX protein overcomes an inhibitory function of the conserved hexapeptide motif. EMBO J. 15:2476-2487.[Medline]

CHAN, S. K., H. D. RYOO, A. GOULD, R. KRUMLAUF, and R. S. MANN, 1997  Switching the in vivo specificity of a minimal Hox-responsive element. Development 124:2007-2014.[Abstract]

CHAUVET, S., S. MERABET, D. BILDER, M. P. SCOTT, and J. PRADEL et al., 2000  Distinct hox protein sequences determine specificity in different tissues. Proc. Natl. Acad. Sci. USA 97:4064-4069.[Abstract/Free Full Text]

DELORENZI, M. and M. BIENZ, 1990  Expression of Abdominal-B homeoproteins in Drosophila embryos. Development 108:323-329.[Abstract]

DUBOULE, D., 1994 Guidebook to the Homeobox Genes. Oxford University Press, London.

FLORENCE, B. and W. MCGINNIS, 1998  A genetic screen of the Drosophila X chromosome for mutations that modify deformed function. Genetics 150:1497-1511.[Abstract/Free Full Text]

GARCIA-BELLIDO, A., 1975  Genetic control of wing disc development in Drosophila. Ciba Found. Symp. 29:161-182.

GEHRING, W. J., Y. Q. QIAN, M. BILLETER, K. FUROKUBO-TOKUNAGA, and A. F. SCHIER et al., 1994  Homeodomain-DNA recognition. Cell 78:211-223.[Medline]

GELLON, G. and W. MCGINNIS, 1998  Shaping animal body plans in development and evolution by modulation of Hox expression patterns. Bioessays 20:116-125.[Medline]

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

GRABA, Y., D. ARAGNOL, and J. PRADEL, 1997  Drosophila Hox complex downstream targets and the function of homeotic genes. Bioessays 19:379-388.[Medline]

GREAVES, S., B. SANSON, P. WHITE, and J. P. VINCENT, 1999  A screen for identifying genes interacting with armadillo, the Drosophila homolog of beta-catenin. Genetics 153:1753-1766.[Abstract/Free Full Text]

HARDING, K. W., G. GELLON, N. MCGINNIS, and W. MCGINNIS, 1995  A screen for modifiers of Deformed function in Drosophila. Genetics 140:1339-1352.[Abstract]

HAYASHI, S. and M. P. SCOTT, 1990  What determines the specificity of action of Drosophila homeodomain proteins? Cell 63:883-894.[Medline]

HU, N. and J. CASTELLI-GAIR, 1999  Study of the posterior spiracles of Drosophila as a model to understand the genetic and cellular mechanisms controlling morphogenesis. Dev. Biol. 214:197-210.[Medline]

IMMERGLÜCK, K., P. A. LAWRENCE, and M. BIENZ, 1990  Induction across germ layers in Drosophila mediated by a genetic cascade. Cell 62:261-268.[Medline]

JACKSON, S. M., H. NAKATO, M. SUGIURA, A. JANNUZI, and R. OAKES et al., 1997  dally, a Drosophila glypican, controls cellular responses to the TGF-ß-related morphogen, Dpp. Development 124:4113-4120.[Abstract]

JONES, B. and W. MCGINNIS, 1993  The regulation of empty spiracles by Abdominal-B mediates an abdominal segment identity function. Genes Dev. 7:229-240.[Abstract/Free Full Text]

KARCH, F., W. BENDER, and B. WEIFFENBACH, 1990  abdA expression in Drosophila embryos. Genes Dev. 4:1573-1587.[Abstract/Free Full Text]

KENNISON, J. A. and M. A. RUSSELL, 1987  Dosage-dependent modifiers of homeotic mutations in Drosophila melanogaster.. Genetics 116:75-86.[Abstract/Free Full Text]

KENNISON, J. A. and J. W. TAMKUN, 1988  Dosage-dependent modifiers of Polycomb and Antennapedia mutations in Drosophila. Proc. Natl. Acad. Sci. USA 85:8136-8140.[Abstract/Free Full Text]

LI, X. and W. MCGINNIS, 1999  Activity regulation of Hox proteins, a mechanism for altering functional specificity in development and evolution. Proc. Natl. Acad. Sci. USA 96:6802-6807.[Abstract/Free Full Text]

LI, X., C. MURRE, and W. MCGINNIS, 1999a  Activity regulation of a Hox protein and a role for the homeodomain in inhibiting transcriptional activation. EMBO J. 18:198-211.[Medline]

LI, X., A. VERAKSA, and W. MCGINNIS, 1999b  A sequence motif distinct from Hox binding sites controls the specificity of a Hox response element. Development 126:5581-5589.[Abstract]

LIN, X. and N. PERRIMON, 1999  Dally cooperates with Drosophila Frizzled 2 to transduce Wingless signaling. Nature 400:281-284.[Medline]

LINDSLEY, D. L., and G. G. ZIMM, 1992 The Genome of Drosophila melanogaster. Academic Press, San Diego.

MALOOF, J. N., J. WHANGBO, J. M. HARRIS, G. D. JONGEWARD, and C. KENYON, 1998  A Wnt signaling pathway controls Hox gene expression and neuroblast migration in C. elegans. Development 126:37-49.[Abstract]

MANN, R. S. and M. AFFOLTER, 1998  Hox proteins meet more partners. Curr. Opin. Genet. Dev. 8:423-429.[Medline]

MANN, R. S. and S.-K. CHAN, 1996  Extra specificity from extradenticle: the partnership between HOX and PBX/EXD homeodomain proteins. Trends Genet. 12:258-262.[Medline]

MARTIN, J.-R., A. RAIBAUD, and R. OLLO, 1994  Terminal pattern elements in Drosophila embryo induced by the torso-like protein. Nature 367:741-745.[Medline]

MARTY, T., M. A. VIGANO, C. RIBEIRO, U. NUSSBAUMER, and N. C. GRIEDER et al., 2001  A HOX complex, a repressor element and a 50 bp sequence confer regional specificity to a DPP-responsive enhancer. Development 128:2833-2845.[Abstract/Free Full Text]

MCGINNIS, W. and R. KRUMLAUF, 1992  Homeobox genes and axial patterning. Cell 68:283-302.[Medline]

MILLER, D. F., S. L. HOLTZMAN, A. KALKBRENNER, and T. C. KAUFMAN, 2001  Homeotic Complex (Hox) gene regulation and homeosis in the mesoderm of the Drosophila melanogaster embryo: the roles of signal transduction and cell autonomous regulation. Mech. Dev. 102:17-32.[Medline]

MURAWSKY, C. M., A. BREHM, P. B