A Screen for Genes That Interact With the ... - Semantic Scholar

Copyright  2004 by the Genetics Society of America DOI: 10.1534/genetics.104.027250

A Screen for Genes That Interact With the Drosophila Pair-Rule Segmentation Gene fushi tarazu Mark W. Kankel,1 Dianne M. Duncan and Ian Duncan2 Department of Biology, Washington University, St. Louis, Missouri 63130 Manuscript received February 4, 2004 Accepted for publication March 23, 2004 ABSTRACT The pair-rule gene fushi tarazu (ftz) of Drosophila is expressed at the blastoderm stage in seven stripes that serve to define the even-numbered parasegments. ftz encodes a DNA-binding homeodomain protein and is known to regulate genes of the segment polarity, homeotic, and pair-rule classes. Despite intensive analysis in a number of laboratories, how ftz is regulated and how it controls its targets are still poorly understood. To help understand these processes, we conducted a screen to identify dominant mutations that enhance the lethality of a ftz temperature-sensitive mutant. Twenty-six enhancers were isolated, which define 21 genes. All but one of the mutations recovered show a maternal effect in their interaction with ftz. Three of the enhancers proved to be alleles of the known ftz protein cofactor gene ftz-f1, demonstrating the efficacy of the screen. Four enhancers are alleles of Atrophin (Atro), the Drosophila homolog of the human gene responsible for the neurodegenerative disease dentatorubral-pallidoluysian atrophy. Embryos from Atro mutant germ-line mothers lack the even-numbered (ftz-dependent) engrailed stripes and show strong ftz-like segmentation defects. These defects likely result from a reduction in Even-skipped (Eve) repression ability, as Atro has been shown to function as a corepressor for Eve. In this study, we present evidence that Atro is also a member of the trithorax group (trxG) of Hox gene regulators. Atro appears to be particularly closely related in function to the trxG gene osa, which encodes a component of the brahma chromatin remodeling complex. One additional gene was identified that causes pair-rule segmentation defects in embryos from homozygous mutant germ-line mothers. The single allele of this gene, called bek, also causes nuclear abnormalities similar to those caused by alleles of the Trithorax-like gene, which encodes the GAGA factor.

T

HE fushi tarazu (ftz) gene is a pair-rule segmentation gene that functions to define even-numbered parasegments in the early embryo. ftz null mutants die as late embryos that entirely lack these parasegments (Wakimoto et al. 1984). ftz is located within the Antennapedia complex (ANT-C) of Hox genes and, like the Hox genes, encodes a Q50 class homeodomain protein (Laughon and Scott 1984). ftz is initially expressed throughout the embryo at syncytial blastoderm. It is then repressed at the poles of the embryo by the maternal terminal genes, restricting expression to the region from 15 to 65% egg length. A pattern of seven “zebra” stripes then emerges by a process of interstripe repression (Carroll and Scott 1985; Weir and Kornberg 1985; Edgar et al. 1986; Yu and Pick 1995). After a distinct stripe sharpening process, the anterior edges of ftz stripes come to coincide precisely with the anterior edges of the even-numbered parasegments (Lawrence et al. 1987). ftz stripes fade during gastrulation and early

1 Present address: Massachusetts General Hospital Cancer Center, Charlestown, MA 02129. 2 Corresponding author: Department of Biology, Campus Box 1229, Washington University, 1 Brookings Dr., St. Louis, MO 63130. E-mail: [email protected]

Genetics 168: 161–180 (September 2004)

germ-band extension and are followed by a complex pattern of ftz expression in each segment of the developing nervous system (Carroll and Scott 1985; Hiromi et al. 1985). This expression serves to define the fates of specific neurons (Doe et al. 1987). Following expression in the nervous system, ftz is expressed in a ring within the developing hindgut (Krause et al. 1988). Its function here has not been determined. After expression fades in the hindgut, ftz is inactive for the rest of development. Characterization of the ftz 5⬘ region has revealed three regulatory elements. Furthest 5⬘ is a complex “upstream” element, through which ftz promotes its own transcription within the zebra stripes (Hiromi et al. 1985; Hiromi and Gehring 1987; Pick et al. 1990; Schier and Gehring 1992, 1993). More proximally a distinct “neurogenic” element is responsible for driving expression in the central nervous system (CNS; Hiromi et al. 1985). Finally, a short zebra element located close to the promoter drives a seven-stripe pattern at blastoderm (Hiromi et al. 1985). The zebra element has been studied extensively, and a number of proteins that bind it have been inferred from DNAse footprinting using nuclear extracts. Several of these proteins have been identified, including Caudal (Dearolf et al. 1989),

162

M. W. Kankel, D. M. Duncan and I. Duncan

GAGA factor (Topol et al. 1991; Bhat et al. 1996; Okada and Hirose 1998), Tramtrack (Harrison and Travers 1990; Brown et al. 1991), and Ftz-F1 (Ueda et al. 1990; Lavorgna et al. 1991). Despite this work, a good understanding of how ftz striping is achieved has not emerged. The striping patterns of the “primary” pair-rule genes eve, h, and run have all been shown to be controlled by the gap genes acting through stripe-specific regulatory elements. The general view has been that ftz striping is then controlled secondarily by the primary pair-rule genes. However, this view has been challenged by the finding that ftz stripes are established almost normally in null mutants for each of the primary pair-rule genes (Yu and Pick 1995). Several downstream targets of ftz have been identified, including Hox genes, segment polarity genes, and other pair-rule genes. However, the best-characterized target is ftz itself. Multiple binding sites for ftz are present within the upstream autoregulatory element (Pick et al. 1990). In a very elegant experiment, Schier and Gehring (1992) showed that ftz autoregulation results from a direct interaction between ftz protein and these sites. On the basis of the kinetics of response of other target genes to pulses of ectopic ftz expression, Nasiadka and Krause (1999) have argued that several other targets are directly controlled by ftz; these include targets that are both positively (en, prd, odd) and negatively (wg, slp) controlled. Surprisingly, the Ftz homeodomain (the only DNA-binding domain of Ftz) is not required for ftz regulation of most or all of these targets (Fitzpatrick et al. 1992; Copeland et al. 1996). This implies that Ftz can execute most of its segmentation functions solely by protein-protein interaction. Consistent with this, Ftz has been shown to require as cofactor the nuclear receptor Ftz-F1 (Han et al. 1993; Florence et al. 1997; Guichet et al. 1997; Yu et al. 1997), which is present uniformly in the early embryo. The Ftz and FtzF1 proteins interact directly with one another via specific domains (Schwartz et al. 2001; Suzuki et al. 2001; Yussa et al. 2001). Although normally both Ftz and FtzF1 bind directly to ftz targets (Florence et al. 1997), in the absence of the Ftz homeodomain interactions between Ftz and Ftz-F1 appear to be sufficient for proper target regulation. Ftz protein also interacts with the protein encoded by paired (prd), another of the pairrule genes (Ananthan et al. 1993; Copeland et al. 1996). This interaction appears to facilitate the recruitment of homeodomain-deleted Ftz to at least some normal ftz targets. Whether other Ftz cofactors are present is not known. Like other homeodomain proteins, Ftz is phosphorylated on multiple Ser and Thr residues (Krause et al. 1988). Phosphorylation of one of these residues, a Thr (T263) present in the N-terminal arm of the Ftz homeodomain, appears to be important for Ftz function (Dong et al. 1998). Substitutions of this residue do not appear to affect DNA binding (Dong et al. 1998), suggesting that T263 phosphorylation may

facilitate important interactions of Ftz with other proteins. To help clarify how ftz is regulated and how it controls its targets, we conducted a screen to identify genes required for normal ftz function. The screening strategy was to select dominant mutations that enhance the lethality of a ftz temperature-sensitive (ts) mutant. Enhancer screens of this type have a long and very successful history in Drosophila genetics. Among successful applications, such screens have been used to identify genes involved in position-effect variegation (Reuter and Wolff 1981), genes of the Polycomb (Duncan 1982) and trithorax groups (Kennison and Tamkun 1988), genes encoding components of receptor tyrosine kinase (Simon et al. 1991; Karim et al. 1996) and dpp signaling (Raftery et al. 1995), genes involved in cell death (Tanenbaum et al. 2000), and genes that interact with the homeotic genes Deformed (Harding et al. 1995; Gellon et al. 1997; Florence and McGinnis 1998) and spineless (Emmons et al. 1999). Here we report the recovery and characterization of 26 ftz enhancer mutations isolated from ⵑ5500 mutagen-treated third chromosomes. Three alleles of the ftz cofactor ftz-f1 were recovered, demonstrating the efficacy of the screen. Most of the mutations isolated cause enhancement of pair-rule segmentation defects in the ftzts background used for the screen and show a maternal effect in their interaction with ftz. Four of the enhancers recovered are alleles of Drosophila Atrophin (Atro), also known as Grunge (Erkner et al. 2002; Zhang et al. 2002). Embryos from Atro mutant germ-line mothers show strong ftz-like segmentation defects (Erkner et al. 2002; Zhang et al. 2002; this report) and an almost complete failure to activate engrailed within the ftz-dependent parasegments. These phenotypes likely result from impairment of the repression ability of Even-skipped (Eve), as Atro has been shown to serve as a corepressor for Eve (Zhang et al. 2002). Surprisingly, we find that Atro also plays a positive role. In addition to segmentation defects, Atro mutations enhance the loss-of-function phenotypes of Antennapedia complex and bithorax complex genes, indicating that Atro is a novel member of the trithorax group (trxG) of Hox gene positive regulators. Atro alleles cause additional phenotypes, including bristle and wing vein defects and leg outgrowths associated with ectopic dpp expression (Erkner et al. 2002; Zhang et al. 2002; this report), which are very similar to the phenotypes caused by mutations in trxG genes encoding components of the brahma (brm) chromatin remodeling complex (Tamkun et al. 1992; Papoulas et al. 1998). Atro appears to be particularly closely related to the Brm complex gene osa, as both genes play an important role in the regulation of wingless target genes (Treisman et al. 1997; Collins et al. 1999; Collins and Treisman 2000; Erkner et al. 2002). One additional gene, which we named bek, was identified that causes pair-rule segmentation defects in em-

ftz-Interacting Genes in Drosophila

bryos from homozygous germ-line mothers. At blastoderm, bek mutant embryos show variable defects in the width and resolution of ftz stripes; at later stages engrailed stripes frequently show gaps, most of which occur within the ftz-dependent parasegments. At syncytial stages, bek mutant embryos show variable defects in mitosis and nuclear arrangement that are almost identical to those described for mutants in Trithorax-like, which encodes the GAGA factor (Bhat et al. 1996).

MATERIALS AND METHODS Stocks: Flies were reared on standard yeast-agar-cornmeal medium (Lewis 1960). Routine crosses were performed at 25⬚. EMS mutagenesis was as described by Lewis and Bacher (1968). Unless indicated otherwise, mutations used are described in Lindsley and Zimm (1992). A TM3 balancer chromosome marked by Ubx was used extensively; this Ubx allele (UbxD105.17) was induced on TM3 by ENU mutagenesis. Preparation of larval cuticles: Twenty-four-hour egglays were allowed to age an additional 24 hr. Unhatched embryos were then washed in PBS-Tween and placed on the sticky side of Time tape. A needle was used to roll the embryo out of the chorion. The vitelline membrane was then punctured with the needle, and the embryo removed and placed in a drop of PBS-Tween. Embryos were then transferred to a drop of mounting medium consisting of 5 ml Shandon (Pittsburgh, PA) Immu-Mount, 7.5 ml saturated chloral hydrate in water and 1.25 ml lactic acid (10%), arranged, coverslipped, and cleared for 1 or 2 days at 42⬚–45⬚ on a slide warmer. Germ-line clones: Maternal germ-line clones were generated using the dominant female-sterile technique of Chou and Perrimon (1996). ftz enhancer mutations were recombined with P[mini w⫹; FRT]3L-2A (FRT 79D–F) or P[ry⫹⫽ neoFRT]82B (FRT 82B) and placed into stock with a y w hsFLP122 X chromosome. Germ-line clones were then produced by crossing these lines to stocks of P[ovoD1] P[FRT 79D–F] or P[FRT 82B] P[ovoD1] (both from the Bloomington Stock Center). Approximately 40 y w hsFLP122; enhancer mutation FRT/Balancer females were crossed to 20 ovoD1-FRT/ Balancer males and allowed to lay eggs for 3 days in glass vials. The vials were then heat-shocked at 37⬚ for 30 min to 1 hr to induce germ-line clones. Somatic clones: Clones of Atro tissue were induced by mitotic recombination using the FLP-FRT technique (Golic 1991; Xu and Rubin 1993). For cuticle preparations, Dp(1;3)sc J4, y⫹ was used as a clone marker. Legs were boiled in 10% KOH for several minutes, washed, dehydrated in n-propanol, and mounted in Euparal. Abdomens were prepared as described by Duncan (1982). Mutant Atro clones in imaginal discs were identified by the loss of green fluorescent protein (GFP) driven by the ubiquitin promoter (Ubi-GFP). lacZ reporters were used to monitor expression of dpp (Twombly et al. 1996), wg (Spradling et al. 1999), and hh (Tabata et al. 1992). The crosses were as follows: for dpp expression, y w hsFLP122; dpplacZ cn/SM6 ; Atro1757 FRT 79D–F/TM6B females were crossed to w ; dpp-lacZ cn/SM6 ; Ubi-GFP FRT 79D–F males; for wg expression, y w hsFLP122; wg-lacZ cn/SM6 ; Atro1757 FRT 79D–F/ TM6B females were crossed to w ; wg-lacZ cn/SM6 ; Ubi-GFP FRT 79D–F males; and for hh expression, y w hsFLP122; Atro1757 FRT 79D–F/TM6B females were crossed to y w hsFLP122; UbiGFP FRT 79D–F hh-lacZ males. Standard protocols for immunofluorescence and immunohistochemistry were used. lacZ expression was detected by using a monoclonal antibody against ␤-galactosidase (Promega, Madison, WI).

163

Antibody staining: Embryos were fixed in 4% formaldehyde in PMG buffer (100 mm PIPES, 2 mm EGTA, 1 mm MgSO4) as an emulsion with heptane on a shaking platform and stored in 100% methanol. Embryos were stained at room temperature with the following primary antibodies in PBS-Triton-X (PBSTx): mouse monoclonal anti-Ftz (Kellerman et al. 1990), mouse monoclonal anti-Engrailed (a gift of N. H. Patel), and rabbit polyclonal anti-Even-skipped (a gift of M. Frasch). The primaries were visualized with Cy3 anti-mouse ( Jackson Immunoresearch, West Grove, PA) and FITC anti-rabbit (Cappel). Imaginal discs were dissected in PBSTx, fixed in 4% formaldehyde in PBSTx, and stored in 100% methanol. Discs were stained at room temperature in PBSTx with the following primary antibodies: mouse monoclonal anti-␤-galactosidase (Promega), mouse monoclonal anti-Distal-less (Duncan et al. 1998), and rabbit anti-Homothorax (gift of A. Salzberg); and they were visualized with Cy3 anti-mouse and Cy5 anti-rabbit ( Jackson Immunoresearch). Propidium iodide staining of fixed embryos was for 20 min in 50 ␮g/ml propidium iodide in PBSTx after a 40-min pretreatment with RNAse A (10 ␮g/ ml PBSTx). All samples were photographed on a Leica SP2 confocal microscope.

RESULTS

The enhancer screen: As a first step in developing a screen for ftz-interacting genes, we wished to identify a genotype in which ftz activity is limiting for survival. Among several genotypes tested, we settled on the heteroallelic combination of ftz9H34, a null allele (Ju¨rgens et al. 1984), and ftz f47ts, a weak temperature-sensitive allele (Wakimoto et al. 1984) caused by an alanine-to-valine substitution in the hydrophobic region of the homeodomain (Fitzpatrick and Ingles 1989). As expected, the viability of ftz9H34/ftz f47ts heterozygotes is temperature dependent. For screening, we chose to use 26.5⬚, a temperature at which ftz9H34/ftz f47ts heterozygotes generated from our stocks have a viability to adulthood of ⵑ64% relative to the balancer classes. At this temperature, ftzf47ts/ftz9H34 heterozygotes exhibit variable segmentation phenotypes, ranging from normal up to having defects in three ftz-dependent larval denticle belts (Figure 1B; Table 2). The cross we used for screening is shown in Figure 2. Briefly, ftz9H34 ry506 e11/TM3, Sb Ubx males were mutagenized with EMS and mated en masse to sssta/TM3, Sb Ubx virgin females. F1 * ftz9H34 ry506 e11/TM3, Sb Ubx virgin females (* denotes a mutagenized chromosome) were collected and individually mated to ftz f47ts p p Ubx e4/TM1 males in vials at 26.5⬚. As indicated in Figure 2, only the * ftz9H34 ry506 e11/ftz f47ts p p Ubx e4 and * ftz9H34 ry506 e11/TM1 classes are potentially viable in this cross; both TM3, Sb Ubx/TM1 and TM3, Sb Ubx/ftz f47ts p p Ubx e4 heterozygotes are lethal due to markers carried by these chromosomes. Putative ftz enhancer mutations were identified on the basis that they showed a strong reduction in the number of * ftz9H34 ry506 e11/ftz f47ts p p Ubx e4 progeny, which are ebony in phenotype, relative to * ftz9H34 ry506 e11/TM1 progeny, which are normal in body color. Because ftz is expressed very early in development, it was anticipated

164


Figure 1.—Ventral views of the cuticle of mature embryos. (A) Wild type. Characteristic denticle belts are present on the three thoracic (T1– T3) and eight abdominal segments (A1–A8). (B) A ftzf47ts/ftz9H34 embryo raised at 26.5⬚. Note defects in the A5 and A7 denticle belts. (C) An embryo homozygous for the ftz null allele ftzR14. Note the complete absence of denticle belts from the oddnumbered abdominal segments and T2.

that many interacting gene products would be maternally provided. Accordingly, individual females carrying mutagenized chromosomes were tested in the screen. Cultures showing a strong reduction in ebony relative to non-ebony progeny were immediately retested by crossing individual * ftz9H34 ry506 e11/TM1 female progeny to ftz f47ts p p Ubx e4/TM1 males in vials at 26.5⬚. Several females were retested for each potential enhancer. If an enhancer mutation were on the third chromosome, then all retested females would be expected to display the mutant phenotype. However, if the mutation were on another chromosome, only about half the retested

Figure 2.—Screen for enhancers of ftz. ftz9H34 ry506 e11/TM3, Sb Ubx males were fed ethylmethane sulfonate (EMS) and crossed to females carrying the TM3, Sb Ubx chromosome. F1 * ftz9H34 ry506 e11/TM3, Sb Ubx females were then individually mated to three ftz f47ts p p Ubx e4/TM1 males at 26.5⬚. As indicated, only two types of progeny are potentially viable in this cross. Cultures having a reduced number of *ftz9H34 ry506 e11/ftz f47ts p p Ubx e4 heterozygotes (which are ebony) relative to *ftz9H34 ry506 e11/TM1 heterozygotes (which are not ebony) were selected and then retested.

females would be expected to exhibit the mutant phenotype. Only cases in which all or almost all the F1 daughters tested positive were kept for further study, so that only third-chromosome mutations were isolated in the screen. Out of ⵑ5500 mutagenized * ftz9H34 ry506 e11 third chromosomes tested, 27 were identified that showed reduced viability when heterozygous with ftz f47ts p p Ubx e4. We anticipated that some of these chromosomes might not carry enhancers of ftz, but rather might carry alleles that fail to complement background lethal mutations present on the ftz f47ts p p Ubx e4 chromosome. To test for this, a homozygous viable ftz f47ts chromosome (ftz f47ts p p e4 ca) was derived. When tested against this chromosome, 26 of the mutant chromosomes still showed low viability (not shown), suggesting that they were isolated due to an interaction with ftz. The basic properties of the 26 mutations are shown in Table 1. For each mutation, a maternal index (MI) value was calculated as the ratio of * ftz9H34 ry506 e11/ftz f47 p p Ubx e4 progeny to * ftz9H34 ry506 e11/TM1 siblings in crosses of * ftz9H34 ry506 e11/TM3, Sb Ubx females to ftz f47 p p Ubx e4/TM1 males. The MI values of the mutants isolated ranged from 0 to 38.5%, whereas the control MI value of the unmutagenized ftz 9H34 ry506 e11 chromosome was 63.8%. To determine whether reduced viability of the * ftz 9H34 ry506 e11/ftz f47 p p Ubx e4 class depends upon a maternal effect, the reciprocal cross was also carried out and a paternal index (PI) value, calculated as the same ratio of progeny types, was determined for each mutation. Most (21/26) of the mutations show PI values that are not significantly less than those seen for the control cross. For these mutations, a maternal effect is clearly required for their interaction with ftz. However, for 5 of the mutations, PI values are reduced relative to the control cross, indicating that these mutations can interact with ftz through their effects in the zygote. For one of these (how2455), the PI and MI values are essen-


165

TABLE 1 Maternal and zygotic effects of ftz interactors

Mutation

Location

Fails to complement

MI (%)

PI (%)

Zygotic effects

Controls ftz-f12478 ftz-f13443 ftz-f15308 Atro 519

75D4;75D5 75D4;75D5 75D4;75D5 66D

Df(3L)Cat Df(3L)Cat Df(3L)Cat Atro03928

63.8(394/618) 1.4(2/138) 1.2(2/163) 6.1(14/229) 2.6(1/39)

64.3(379/589) 72.5(111/153) 55.0(44/80) 54.0(81/150) 44.8(13/29)

No No No Weak

Atro 1757

66D

Atro03928

38.5(35/91)

107(119/111)

No

Atro 2295 Atro 3333

66D 66D

Atro03928 Atro03928

10.6(5/47) 0.0(0/145)

69.6(78/112) 61.9(39/63)

No No

bek1

99A5-6

Df(3R)3450 Df(3R)323a Df(3R)1075a Df(3L)29A6 Df(3R)eN19 hth05745

16.3(33/203)

80.7(109/135)

No

6.0(10/168) 3.4(4/117) 31.4(44/140)

33.7(56/166) 2.0(3/147) 56.8(46/81)

Yes Yes No

74.7(112/150) 29.3(36/123) 57.4(27/47) 70.7(94/133) 95.7(67/70) 71.4(60/84)

No Yes No No No No

eIF-4E 2188 67B1-2 94A1 how 2455 86C1-2 hth2232 20 723 1027 1427 1815 1819 2142 2828 2955 3150 3201 4038 4236 4758 5303

67A2;67D11-13 Between st and cu Between st and cu 68C8-11;69B4-5 82D5;82F3-6 95F7;96A2

Df(3L)AC1

9.5(18/189) 19.5(31/159) 25.5(35/137) Df(3L)vin7 0.5(1/197) Df(3R)6-7 9.3(13/140) Df(3R)crb87-4 6.4(8/125) Df(3R)crb87-5 on 3L 4.8(4/83) 75C1-2;75F1 Df(3L)Cat 25.6(42/164) 66D10-11;66E1-2 Df(3L)h-i22 2.5(2/81) 88A Df(3R)1-25-ca 0.0(0/204) On 3R 7.4(17/229) 93B2-13;94A3-8 Df(3R)eN19 5.0(8/161) Between cu and sr 7.1(9/127) On 3L 11.5(15/130) Unassigned 32.7(53/162)

51.0(49/96) Weak/no? 92.8(128/138) No 83.3(100/120) No 82.6(114/138) No 52.5(52/99) No 71.7(99/138) No 18.7(17/91) Yes 26.3(35/133) Yes 113(78/69) No

MI/PI

Maternal effects

Germ-line clones

0.02 Strong 0.02 Strong 0.11 Strong 0.06 Strong

Normal Dorsal fusions Normal ftz-like segment defects 0.36 Weak ftz-like segment defects 0.15 Moderate No eggs 0.0 Strong ftz-like segment defects 0.20 Moderate Weak ftz-like segment defects 0.18 Moderate Normal 1.7 No Normal 0.55 Weak/no? Weak posttransformation 0.13 Moderate Normal 0.67 Weak/no? Not tested 0.44 Weak/no? Not tested 0.01 Strong No eggs 0.10 Strong Not tested 0.09 Strong No eggs 0.09 0.28 0.03 0.0 0.14 0.07 0.38 0.44 0.29

Strong Moderate Strong Strong Moderate Strong Weak Weak/no? Moderate

Normal Not tested No eggs Normal Not tested Normal Normal Normal Not tested

Information on the chromosomal location and the maternal and zygotic effects of each mutation is summarized. The MI is calculated as the number of * ftz 9H34 ry 506 e 11/ftz f47 p p Ubx e 4 individuals relative to * ftz 9H34 ry 506 e 11/TM1 siblings in crosses of * ftz 9H34 ry 506 e 11/TM3, Sb Ubx females to ftz f47 p p Ubx e 4 /TM1 males. The PI is calculated as the same ratio, but from the reciprocal cross. The zygotic effects of each mutation are inferred from the PI: Values ⱖ54% are taken to indicate no zygotic effects, values of 45 and 51% likely indicate weak zygotic effects, and values ⬍34% are taken as indicating significant zygotic interactions. The maternal effect of each mutation is estimated from the MI/PI ratio: Ratios ⱕ0.11 are interpreted as strong maternal effects, ratios between 0.11 and 0.30 are interpreted as moderate maternal effects, and ratios ⬎0.30 are interpreted as indicating weak or no maternal effect. a From our own work. Df(3R)323 is deficient for bands 99A5–6;B7-11, Df(3R)1075 is deficient for bands 98F4–5;99A5,6, and Df(3R)1-25-c (now lost) is deficient for 88A.

tially the same (2.0 and 3.4%, respectively), indicating that this gene interacts purely zygotically. The remaining four mutations that show zygotic effects have a much lower MI than PI, indicating that both maternal and zygotic effects are important in their interaction with ftz. In summary, 25/26 of the ftz interactors identified in our screen appear to show a maternal effect. Mapping: The ftz enhancer mutations were first mapped by recombination using the multiply marked chromosome ru h th st cu sr e s ca. Reciprocal crossovers were recovered in each marker interval and placed into

stock. Recombinants that retained ftz9H34 were then tested for the presence of the ftz interactor by crossing heterozygous females to ftz f47ts p p Ubx e4/TM1 males at 26.5⬚. For 22 of the mutations, this procedure allowed for mapping between adjacent markers. Three mutations were localized only to chromosome arm (3L or 3R) and one mutation has not been mapped. Once an approximate map position was determined by recombination, mutations were then tested for lethality when heterozygous with deficiencies in the appropriate region. In this way, 18 of the ftz interactors were mapped

166

M. W. Kankel, D. M. Duncan and I. Duncan TABLE 2 Distribution of segmentation defects in ftz f47ts/ftz 9H34 heterozygotes at 27.5ⴗ

Mutation

n

Wild type

Control ftz-f12478 ftz-f13443 ftz-f15308 Atro 519 Atro 1757 Atro 2295 Atro 3333 bek1 eIF-4E 2188 how 2455 hth2232 20 723 1027 1427 1815 1819 2142 2828 2955 3150 3201 4038 4236 4758 5303

93 123 131 173 92 78 135 150 134 142 65 92 152 160 121 175 122 134 144 102 130 141 142 123 82 102 117

43 5 2 15 0 8 41 2 10 59 48 24 42 13 58 1 34 3 10 39 28 7 16 21 50 42 44

One

Two

Three

Four

37 15 19 29 3 8 56 1 23 40 14 29 53 32 48 2 28 17 13 40 37 5 32 25 26 37 32

9 48 64 59 6 19 25 0 18 28 2 24 40 63 13 9 33 72 102 15 44 17 72 48 5 14 21

4 48 32 47 18 25 11 7 62 13 0 11 11 35 1 30 23 30 14 4 15 33 18 19 0 8 17

0 7 14 22 54 15 2 138 21 2 1 4 5 13 1 113 2 12 5 4 6 68 4 10 1 1 3

ftz⫺

% segment defects

% ⱖ3 segments affected

0 0 0 1 11 3 0 2 0 0 0 0 1 4 0 20 2 0 0 0 0 11 0 0 0 0 0

53.8 95.9 98.5 91.3 100 89.7 69.6 98.7 92.5 58.5 26.2 73.9 72.4 91.9 52.1 99.4 72.1 97.8 93.1 61.8 78.5 95.0 88.7 82.9 39.0 58.8 62.4

4.3 44.7 35.1 40.5 90.2 55.1 9.6 98.0 61.9 10.6 1.5 16.3 11.2 32.5 1.7 93.1 22.1 31.3 13.2 7.8 16.2 79.4 15.5 23.6 1.2 8.8 17.1

Distribution of mutant phenotypes in * ftz 9H34 ry 506 e 11/ftz f47ts p p Ubx e4 heterozygotes raised at 27.5⬚. Unhatched ∨ embryos from the following cross were examined: * ftz 9H34 ry 506 e 11/TM3, Sb Ubx 乆’s ⫻ ftz f47ts p p Ubx e 4/TM1 么’s at 27.5⬚. n, the number of embryos examined per genotype. “Wild type,” “One,” “Two,” “Three,” and “Four” are the number of embryos that have normal segmentation or that have defects in one, two, three, or four ftz-dependent segments, respectively. “ftz⫺” indicates the number of embryos with defects in all ftz-dependent segments.

to a cytological interval (Table 1). There is a potential problem with the mapping method used: For genes identified by a single hit there is a possibility that the lethal mapped by complementation may not in fact be the enhancer; rather, it may be a coincidental lethal linked to an enhancer. Assignment to lethal complementation groups: Once the ftz enhancers were mapped, each was tested for complementation with other enhancer mutations located in the same interval. Each enhancer was also tested for complementation with all lethal P-element insertions in its interval that were available from the Bloomington Stock Center collection. Two complementation groups that have multiple alleles were identified: Three of the mutations are allelic to ftz-f1, which encodes a known ftz cofactor, and four of the mutations are alleles of Atrophin (Atro), also known as Grunge. The remaining 19 mutations appear to be single-hit alleles (Table 1). Of these, three (eIF-4E2188, how2455, and hth2232)

are alleles of known genes, and the remainder appear to be novel. Although several of the ftz interactors map within cytological intervals that are known to contain trithorax group (trxG) genes, all the ftz enhancers tested complemented the corresponding trxG mutations (alleles tested were sls1, urd2, trx3, trxE2, kto3, mor2, brm5, ash6, ash2, vtd7, skd2, and dev2; all kindly provided by Dr. J. Kennison). Enhancement of embryonic cuticular phenotypes: All of the mutations recovered were tested to determine whether they enhance ftz segmentation defects in ftz9H34/ ftz f47ts larvae. Unhatched embryos from the following ∨ cross were examined: * ftz9H34 ry506 e11/TM3, Sb Ubx 乆 ⫻ f47ts p 4 ftz p e ca 么 at 27.5⬚. In control crosses at this temperature, ⵑ35% of the ftz9H34/ftz f47ts heterozygotes survive to adulthood, and almost all of the unhatched embryos are ftz9H34 ry506 e11/ftz f47ts p p e4 ca in genotype (data not shown). Forty-six percent of the unhatched control embryos are wild type for segmentation, 40% display single


ftz-dependent segment or hemisegment defects, and 14% display multiple ftz-dependent segmental disruptions (Table 2). The most severely affected embryos in this cross have defects in only three segments. The percentage of wild-type embryos is likely to be an underestimate, as only embryos that did not hatch were examined. Relative to controls, 19/26 of the ftz interactors display an increase in the frequency and severity of ftzlike segmentation defects at 27.5⬚. Of the seven mutations that fail to enhance segmentation defects, five behave similarly to the control. These may act by enhancing ftz loss-of-function in the CNS or hindgut. Surprisingly, two mutations, how2455 and 4236, appear to suppress the segmentation defects in ftz9H34/ftz f47ts heterozygotes. Embryonic cuticular phenotypes of maternal germline clones: As described above, 25/26 of the ftz enhancers display a maternal effect when heterozygous in their interaction with ftz (Table 1). To determine the effects of these mutations when homozygous in the female germ line, we used the dominant female-sterile technique of Chou and Perrimon (1996). We were able to place 20 of the mutations in cis to an appropriate FRT element and generate germ-line clones. Six mutations were not tested; these were difficult to recombine with appropriate FRTs because they are located very near centromeres or are poorly mapped. All mothers carrying homozygous germ-line clones were crossed to heterozygous mutant males; in selected cases, other types of males were used as well. As summarized in Table 1, four of the mutations cause ftz-like segmentation defects in embryos derived from homozygous germ-line clones. Three of these are alleles of Atro (Atro519, Atro3333, and Atro1757), whereas the fourth is the single allele recovered of a novel gene we call bek. Four of the mutations (1427, Atro2295, 1819, and 2955) cause the production of no or very defective eggs, one (ftz-f13443) causes variable dorsal-anterior defects, and 10 produce normal embryos. Homozygotes for the hth2232 mutation show weak posteriorly directed homeotic transformations of the larval thoracic and A1 denticle belts; this phenotype is the same for zygotes from heterozygous mothers or from homozygous mutant germ-line mothers. The embryonic defects we find in our analysis of germ-line clones are described by gene below. ftz-f1: Our three ftz-f1 alleles all fail to complement the existing ftz-f1 P-element insertion, ftz-f1l(3)03649 (Spradling et al. 1999). However, none of our alleles cause pair-rule segmentation defects in embryos from germ-line clones. One allele (ftz-f13443) causes apparent fusions of the dorsal head and thorax, giving the embryo a “comma” appearance. bek: With rare exception, embryos from bek1 homozygous mutant germ-line mothers die prior to hatching regardless of whether the fathers are bek1/⫹ or ⫹/⫹. Most of these embryos die early in development. Of those that survive to cuticle deposition, about half show

167

segmentation defects. The most common defect is deletion of the T3 denticle belt (which is not ftz dependent), but many embryos also show ftz-like pair-rule defects of varying severity (see Figure 3, E and F). The frequency of defects is sensitive to ftz⫹ dosage: In a cross of homozygous bek1 germ-line mothers to bek1/⫹ fathers, ⵑ45% of the embryos produced showed segmentation defects (number scored ⫽ 87), whereas in a cross to ftz9H34 bek1/⫹ fathers, the frequency of defects was 63% (number scored ⫽ 139). In the latter cross, defects were concentrated in the ftz-dependent segments. Staining with propidium iodide revealed numerous nuclear abnormalities early in development: These include mitotic asynchrony, chromatin bridges, and irregularities in nuclear shape and size (see Figure 4). These abnormalities are closely similar to those described for embryos from mothers mutant for the Trithorax-like gene, which encodes the GAGA factor (Bhat et al. 1996). At cellular blastoderm, nuclei are irregularly arranged at the cortex, with some nuclei lying more internally than others (Figure 4C). Ftz and Eve expression is variably affected, with stripes often showing abnormal spacing or width. Among embryos that survive to gastrulation, many show gaps in the En stripes. These gaps occur predominantly in the ftz-dependent parasegments. Atrophin (Atro): Three of the four Atro alleles recovered cause ftz-like segmentation defects in embryos from homozygous germ-line mothers. The phenotypes are similar to those described previously for other Atro alleles (Erkner et al. 2002; Zhang et al. 2002). Embryos from homozygous Atro519 or Atro3333 females display ftzlike embryonic phenotypes for all paternal chromosomes tested, including wild type (Figure 3, B–D). Some embryos from homozygous Atro3333 females (Figure 3E) show strong ftz-like segmentation defects (Figure 3C). However, most show five, rather than four, abdominal denticle belts (Figure 3B). Embryos from Atro519 female germ-line mosaics show a failure of the ftz-dependent denticle belts to develop, but also lack denticles along the ventral midline within the even-numbered segments (Figure 3D). Most eggs laid by mothers whose germ lines are homozygous for the Atro1757 allele appear not to be fertilized. The small proportion that do initiate development die prior to cuticle formation. Antibody stainings of early stages reveal strong ftz-like pair-rule defects in these embryos. Females whose germ lines are homozygous for the Atro2295 allele fail to lay eggs, likely because of the presence of secondary mutations on the Atro2295 chromosome. The ftz striping pattern is strongly affected in embryos from Atro mutant germ-line mothers. The three alleles that allow egg production have similar effects. By late blastoderm, ftz stripes 1, 2, and 4 are very strong and well defined. However, stripe 3 is very weak, and stripes 5, 6, and 7 are weak and fail to resolve properly (see Figure 5). Stripes 4–7 are narrower and better resolved dorsally than ventrally. The stripe patterns seen in Atro

168


Figure 3.—Cuticular phenotypes of ftz enhancer mutant embryos. Ventral views are shown for all embryos. (A) A ftz null mutant (ftzR14/ftzR14) embryo. (B) An Atro3333/⫹ embryo from a cross of homozygous mutant germ-line mothers to wildtype males. This embryo is typical of those produced in this cross. Five abdominal denticle belts are present. The most posterior belt is adjacent to the anal pads, and is likely, therefore, to be from A8. The identities of the more anterior belts are not entirely clear. However, it seems likely they are A2, A3, A4 and A6. This is suggested by the en activation pattern in mutant embryos (see Figure 5), in which a weak en stripe is often activated in A3 (parasegment 8), but not in other ftzdependent abdominal segments. (C) An embryo from the same cross as in B, but showing strong ftz-like segmentation defects. A fusion of the presumptive A4 and A6 denticle belts is present; similar fusions are often seen in ftz null mutant embryos. (D) An Atro519/⫹ embryo from a cross of homozygous mutant germ-line mothers to wildtype males. Embryos from this cross typically show strong ftz-like segmentation defects combined with the loss of denticles midventrally. (E) A bek1/⫹ embryo from a cross of homozygous mutant germline mothers to wild-type males. Note partial fusions of A4 and A5 and of A6 and A7. (F) An embryo from the same cross as in E, but showing strong ftz-like segmentation defects.

mutants are very different from those present at early stages in wild type, indicating that they do not arise simply because of a delay in stripe refinement. In many embryos from mothers having Atro-mutant germ lines there is a complete failure to activate en within the ftzdependent parasegments (see Figure 5, G–I). Presumably these embryos correspond to the class that shows the complete absence of ftz-dependent parasegments at the cuticle stage (see Figure 3). However, frequently there is weak activation of en in parasegments 4 and 8 (Figure 5, J–L). These parasegments are defined by ftz stripes 2 and 4, which are the two strongest ftz stripes in Atro-mutant embryos. Presumably this class of embryo corresponds to the predominant cuticular phenotype seen, in which five abdominal denticle belts are present (see Figure 3). Atro is a trithorax group (trxG) gene: In working with the Atro alleles, we noted that all four enhance the weak haltere-to-wing transformation caused by heterozygosity for Ubx alleles. This effect is illustrated in Figure 6 for the allele Ubx1 and for Df(3R)P9, a deficiency for the entire bithorax complex (BX-C). To determine whether

Atro alleles enhance other loss-of-function phenotypes of BX-C genes, we tested for interactions with other haplo-abnormal phenotypes of the BX-C. In wild-type males the A5 and A6 abdominal tergites are darkly pigmented and the A6 sternite is broad and has no bristles. When heterozygous with wild type, Df(3R)P9 causes the A5 tergite to be transformed toward A4 identity, which results in the incomplete pigmentation of A5 (Figure 7B). In addition, A6 is weakly transformed to A5, which causes a few bristles to develop in the A6 sternite, and A7 is weakly transformed to A6, resulting in the development of a small tergite in A7. These phenotypes result from lowered levels of Abdominal-B (Abd-B) expression (Duncan 1987; Celniker et al. 1990). The transformations of A5, A6, and A7 in Df(3R)P9/⫹ males are all enhanced by our Atro alleles (Figure 7C). We also tested the P-element insertion allele Atro03928 (Spradling et al. 1999). This allele behaves like our alleles in enhancing A5, A6, and A7 transformations in Df(3R)P9/⫹ males. Atro03928 also enhances the haltere-to-wing transformation in Df(3R)P9 heterozygotes, but the effect is weaker than that for our alleles.


169

Figure 4.—Defects in embryos from mothers carrying germ lines homozygous for the bek1 mutation. Anterior is to the left in all panels. (A–C) Embryos stained with propidium iodide. (A) A wild-type embryo during midcellularization. Nuclei are of uniform size and distribution. (B) A syncytial-stage embryo from a bek1-mutant mother. Nuclei are of variable size and distribution, and chromatin bridges frequently join adjacent nuclei. (C) A gastrulating embryo from a bek1 mother showing a large patch that appears to have undergone an extra division relative to surrounding areas. Such asynchrony is seen frequently, but is variable in position. (D–F) Embryos stained for Eve (red) and Ftz (green). (D) A wild-type embryo at the beginning of gastrulation. Eve and Ftz stripes are regular in width and spacing. (E) An embryo at early gastrulation from a bek1-mutant mother. Note abnormal stripe widths and spacing, and the presence of gaps or holes in the stripes. As shown in F, these gaps result from irregular arrangement of nuclei at the cortex. (F) Part of an Eve stripe from E (see boxed area) viewed at higher magnification at two different focal planes. Note that what appears to be a gap in the left-hand image in F (arrow) contains a stained nucleus at a different focal plane. (G–I) A bek-mutant embryo at early germ-band extension stained for Eve (red) and En (green). Several of the En stripes are incomplete (I), with most defects occurring in the ftz-dependent parasegments (compare I with G). These gaps in the En stripes do not result from nuclei lying at different focal planes.

To examine further the role of Atro during abdominal development, somatic clones of Atro1757 were induced in the abdomens of adult males using the FLP-FRT technique. Atro1757 clones were similar in size to control clones and were generated at a frequency nearly identical to that of Atro⫹ controls (data not shown). Mutant clones in the abdomen often cause a transformation of A5 and A6 to the anterior (Figure 7E). This result is consistent with the enhancement of the haplo-abnormal phenotypes of Df(3R)P9 by Atro alleles and suggests that Atro⫹ is required for full activation of Abd-B in A5 and A6. As described previously (Erkner et al. 2002; Zhang et al. 2002), Atro mutant clones also cause extensive ectopic wing vein formation (Figure 8H), rough eyes, ectopic or missing bristles, disruptions of the dorsal midline in the thorax, and limb outgrowths (see below). We next tested for interactions between Atro alleles and mutations within the ANT-C. AntpNs is a gain-of-function allele that causes ectopic Antp expression in the eyeantennal imaginal disc (Jorgenson and Garber 1987), which results in a variable transformation of antenna to leg. In control AntpNs heterozygotes, only 15% of flies

show any aristal development (Table 3A). The frequency of flies having untransformed antennae increased nearly twofold in heterozygotes for the Atro519, Atro1757, and Atro3333 alleles (Table 3A). Heterozygosity for Atro2295 has no effect. Antp expression in AntpNs is driven by the endogenous Antp promoter (Jorgenson and Garber 1987). In contrast, ectopic expression in the Antp73b gain-of-function allele is driven by the promoter of stranded at second (sas), a nonhomeotic gene (Frischer et al. 1986; Schneuwly et al. 1987). This difference has been exploited as an assay of whether modifier genes act upstream or downstream of the Antp promoter (Tamkun et al. 1992). We find that all four Atro alleles fail to suppress the antenna-to-leg transformation in Antp73b heterozygotes (Table 3B), suggesting that Atro⫹ functions at the normal Antp promoter to enhance its expression, but not at the sas promoter. We also tested for interactions with the Sex combs reduced (Scr) gene of the ANT-C. Loss of Scr function in the T1 leg disc transforms the first leg toward second leg identity. This causes a reduction in number of sex

170


Figure 5.—Eve, Ftz, and En expression in embryos from mothers carrying germ-line clones homozygous for Atro3333. (A–C) Embryos at syncytial blastoderm stained for Eve (green) and Ftz (red). (A) A wild-type embryo midway into cellularization. Nuclei have not yet elongated and membrane ingrowth is about half completed (see inset). (B–C) An embryo from an Atro3333 germ-line mother. This embryo is slightly older than the embryo in A, showing some nuclear elongation and more complete membrane ingrowth. In B and C, ftz stripes 2 and 4 are very strong, whereas stripe 3 is very weak. Stripes 5–7 are weak and poorly resolved. Stripes 4–7 are fused to varying degrees ventrally. This pattern is seen in essentially all embryos from Atro-mutant germ-line mothers. (D–F) A wild-type embryo at early germ-band extension stained for Eve (green) and En (red). The Eve-dependent En stripes coincide with the sharpened Eve stripes at this stage. The intervening En stripes are Ftz dependent. (G–I) An embryo from an Atro3333 germ-line mother stained for Eve (green) and En (red). Note the complete absence of the Ftz-dependent En stripes. Very likely embryos of this type produce ftz-null cuticular phenotypes such as that shown in Figure 3C. ( J–L) Another embryo from an Atro3333 germ-line mother stained for Eve (green) and En (red). Weak En stripes are present in parasegments 4 and 8 (arrowheads), which correspond to the prominent Ftz stripes 2 and 4 in C. Presumably, embryos of this type produce weaker cuticular phenotypes like that shown in Figure 3B.

comb bristles on the T1 leg of males (Lewis et al. 1980). Scr is dose sensitive: Wild-type males show 11–12 teeth per T1 leg, whereas Scr⫺/⫹ males usually have 5–6 teeth. This reduction in sex comb teeth is significantly enhanced by heterozygosity for Atro alleles. Males heterozygous for the Scr deficiency, Df(3R)Scr, and any of the Atro alleles show a reduced number of sex comb bristles per leg relative to Df(3R)Scr/⫹ control males (Table 4). We also find that Atro alleles suppress the gain-offunction Scr allele ScrMsc. ScrMsc causes ectopic expression of Scr in the second and third leg discs, resulting in transformation of the second and third leg to first leg (Pattatucci and Kaufman 1991). All four Atro alleles cause significant suppression of this transformation in the second leg and all but one cause significant suppression in the third leg (data not shown). Taken together, these observations suggest that Atro⫹ is a positive regulator of Scr expression. Interactions of Atro with trithorax group gene members brahma and trithorax: The enhancement of BX-C

and ANT-C loss-of-function phenotypes by Atro alleles suggests that Atro is a member of the trxG of genes. This idea is supported by interactions between Atro and the trxG genes brahma (brm) and trithorax (trx). When heterozygous with wild type, brm mutations cause variable transformations of haltere to wing and A5 to A4 (Tamkun et al. 1992). Relative to control crosses, Atro alleles enhance the haltere-to-wing transformation phenotype in Df(3L)brm11 heterozygotes (Table 5A), but fail to enhance A5 to A4 transformations (data not shown). However, Atro alleles do strongly enhance A5 to A4 transformations when heterozygous with trx3 (Table 5B). Dominant interactions of Atro with the Polycomb group gene Polycomb-like: The trxG genes act antagonistically to the genes of the Polycomb group (PcG), which serve as maintenance repressors of the ANT-C and BX-C. Mutations in these genes cause several homeotic transformations that result from ectopic homeotic gene expression, the best known being transformations of second and third leg to first leg (Lewis 1978; Duncan and


171

Figure 6.—Enhancement of the haltere-to-wing transformation in Ubx heterozygotes by Atro1757. All halteres shown are from female flies. The halteres of Atro1757/⫹ heterozygotes (B) appear to be slightly enlarged relative to those of wild type (A). Ubx1/⫹ (C) and Df(3R) P9/⫹ (E) heterozygotes show very weak transformations of haltere to wing, as indicated by an increase in haltere size and the development of a few wing-type marginal bristles. For both of these genotypes, haltere size and the number of marginal bristles are strongly enhanced by heterozygosity for Atro1757 (D and F).

Lewis 1982). Since the trxG genes act antagonistically to the PcG genes, trxG mutations typically suppress the effects of mutations in PcG genes. To test whether Atro alleles act in this way, we determined the effects of the Atro alleles in the Polycomblike (Pcl) mutant heterozygote Pcl1/Pcl 4. In an otherwise wild-type background, Pcl1/ Pcl 4 heterozygotes show very strong transformations of the second and third legs to first legs (Duncan 1982). These transformations are suppressed significantly by heterozygosity for the Atro alleles Atro519, Atro1757, and

Atro3333. The Atro2295 allele has only a minor effect (Table 6). These interactions support the idea that Atro is a member of the trxG. The tests described above indicate that the four Atro alleles recovered do not form a consistent hypomorphic series. For example, the Atro1757 allele shows the weakest enhancement of ftz lethality, but causes the most severe phenotype in embryos from homozygous germ-line mothers and is the strongest suppressor of leg transformations in Pcl1/Pcl 4 heterozygotes. The Atro519 and

Figure 7.—Effects of Atro in the adult abdomen. (A) Abdominal cuticle from a wild-type male. The abdomen has been split middorsally and opened out so that hemitergites are arrayed around the outside and the sternites are located along the midline. The tergites of A5 and A6 are darkly pigmented. (B) Abdominal cuticle from a Df(3R)P9/⫹ male. A5 shows reduced pigmentation, indicating a weak transformation to A4, and A7 shows a tiny tergite (arrowhead), indicating a weak transformation to A6. (C) Abdominal cuticle from a Df(3R)P9/Atro1757 male. A5 is almost completely transformed to A4, and a tergite of moderate size is present in A7 (arrowhead). (D and E) Mitotic recombination clones in the A6 tergite marked by yellow (y ) owing to loss of Dp(1;3)sc J4, y⫹. (D) control Atro⫹/Atro⫹ clone showing the yellow phenotype of the clone. (E) Atro1757/Atro1757 clone showing a much greater loss of pigmentation, indicating a transformation to A4 or a more anterior segment.

172

M. W. Kankel, D. M. Duncan and I. Duncan TABLE 3 Interaction of Atro with gain-of-function alleles of Antennapedia Genotype

% normal arista A. Antp Ns

FRT79D-F/Antp Ns Atro 519/Antp Ns Atro 1757/Antp Ns Atro 2295/Antp Ns Atro 3333/Antp Ns

15 36 26 15 28

(31/211) (50/141) (32/123) (19/124) (33/117)

55 52 55 49 59

(118/221) (120/229) (130/238) (114/232) (132/223)

B. Antp 73b FRT79D-F/Antp 73b Atro 519/Antp 73b Atro 1757/Antp 73b Atro 2295/Antp 73b Atro 3333/Antp 73b

Interaction of Atro mutations with dominant alleles of the homeotic gene Antennapedia. The dominant effects of Atro mutations and an Atro ⫹ control (the FRT79D–F chromosome) in Antp Ns (A) and Antp 73b (B) heterozygotes are shown. “% normal arista,” the percentage of flies that have at least one wild-type arista. The number of flies scored is in parentheses.

Figure 8.—Effects of Atro in the legs and wing. (A) A T1 leg from a wild-type male. (B) A T1 leg containing an outgrowth from the ventral surface of the trochanter. The distalmost part of the outgrowth contains an Atro1757/Atro1757 clone. This is shown in a higher magnification of the outgrowth (C). The clone is marked by yellow (y ) (arrowhead) owing to loss of Dp(1;3)sc J4, y⫹. More proximal portions of the outgrowth are y⫹, indicating they are Atro1757/⫹. (D) A leg showing ventral fusion of the femur and tibia. This phenotype is caused frequently by Atro-mutant clones. (E) An outgrowth from the ventral region of the distal leg. As shown in the higher magnification in F, this outgrowth contains an Atro1757/Atro1757 clone distally (arrowhead indicates region containing y⫺ bristles). (G) A wild-type wing showing the normal venation pattern. (H) A wing containing many Atro3333/Atro3333 clones showing ectopic and missing wing vein elements.

Atro3333 alleles behave similarly in strongly enhancing ftz segmentation defects and causing ftz-like segmentation defects in embryos from homozygous germ-line mothers. However, Atro3333 is the stronger in its enhancement of segmentation defects in ftz9H34/ftz f47ts heterozygotes, while Atro519 is the stronger in its interactions with AntpNs, Scr, Pcl, brm, and trx. The Atro2295 allele is the weakest by most tests, but the strongest in its interaction with brm. Atro mitotic clones in the leg: Mitotic recombination clones homozygous for Atro alleles frequently show de-

fects in the ventral regions of the legs (Erkner et al. 2002; this report). Often the ventral surfaces of the femur and tibia are fused together, so that the leg cannot be properly extended (Figure 8D). This latter phenotype is also caused by homothorax (hth) loss-of-function clones when they are induced midway in larval development (I. Duncan, unpublished results), suggesting that Atro⫹ may be required for normal expression or function of hth. Mitotic clones also frequently cause outgrowths from the ventral surface of the legs. Although these outgrowths can occur anywhere from the coxa to the tibia along the proximo-distal axis (Figure 8), they are by far the most frequent in the coxa-trochanter region (Figure 8, B and C), where they are associated with loss of expression of the proximal determinant teashirt and ectopic expression of Distal-less (Erkner et al. 2002). The distal-most portion of each leg outgrowth is always mutant for Atro but more proximal regions are usually a mix of Atro and Atro⫹ tissue or are entirely Atro⫹ (see Figure 8, C and F). This nonautonomy suggested that Atro clones might activate expression of wingless (wg) or decapentaplegic (dpp), since the overlapping distribution of these morphogens defines the proximodistal limb axis (Lecuit and Cohen 1997). To test these possibilities, wg and dpp expression were examined in discs containing Atro mutant clones. wg is normally expressed in the antero-ventral region of the leg imaginal disc and is required for ventral cell fates, whereas dpp is expressed in the antero-dorsal region of the leg imaginal disc and is required for dorsal cell fates (Jiang and Struhl 1996 and references therein). Atro1757 clones were monitored for wg-lacZ and


173

TABLE 4 Interaction of Atro with the homeotic gene Scr Genotype FRT 79D-F/Df(3R)Scr Atro 519/Df(3R)Scr Atro 1757/Df(3R)Scr Atro 2295/Df(3R)Scr Atro 3333/Df(3R)Scr

n⫽3

n⫽4

n⫽5

n⫽6

n⫽7

n⫽8

No. of legs

0 1 2 0 1

3 39 28 32 9

38 53 56 62 66

40 7 14 6 23

18 0 0 0 1

1 0 0 0 0

100 100 100 100 100

sc/leg ⫾SD 5.76 4.66 4.82 4.74 5.14

⫾ ⫾ ⫾ ⫾ ⫾

0.82 0.62 0.69 0.56 0.62

Shown are the dominant effects of the Atro alleles on the number of the male sex comb bristles on the first thoracic legs of males heterozygous for Df(3R)Scr. The distribution of the number of sex comb teeth per leg (n ⫽ 3 for three sex-comb bristles; n ⫽ 4 for four sex-comb bristles, etc.) is presented for each Atro allele and for an Atro ⫹ control (the FRT79D–F chromosome).

dpp-lacZ expression in all regions of the leg imaginal disc. The large majority of clones have no effect on the expression of either reporter and appear to integrate well into the disc, as they have wiggly borders. However, a few ventral clones in the proximal portions of the leg discs have smooth borders and alter the nearby folding pattern of the disc. These clones show ectopic expression of dpp-lacZ. These results are consistent with those of Erkner et al. (2002) and suggest that Atro⫹ is required to repress dpp expression in the proximal and ventral regions of the leg discs. Both dpp and wg expression in anterior compartment cells is induced by the diffusible morphogen hedgehog (hh), which is expressed in the posterior compartment of the leg disc (Basler and Struhl 1994; Tabata and Kornberg 1994). To determine whether ectopic dpp expression results from ectopic hh expression in the anterior/ventral region, we examined hh expression in Atro1757 clones in the leg imaginal disc. Clones of Atro1757 cells fail to ectopically activate hh-lacZ expression in anterior regions of the leg disc (data not shown). In addition, hh expression is unaffected in clones in posterior regions of the leg disc (data not shown). Thus, Atro⫹ is not required for hh repression in the anterior compartment of leg discs.

teins known to function in close association with the target protein are not identified (see, for example, the very large-scale screen for ras-interacting genes conducted by Karim et al. 1996). These may be proteins that are produced in excess or that are functionally redundant. Often dominant-negative alleles are recovered, which can circumvent these problems. Typically, the mutations identified in enhancer or suppressor screens comprise multiple alleles of a few genes and single alleles of many others. Genes defined by multiple alleles usually turn out to encode proteins that are closely related in function to the target gene. Although TABLE 5 Interaction of Atro with the brm and trx genes

Genotype

A. Interaction of Atro with Df(3L)brm11 FRT79D-F/Df(3L)brm11 20 (48/244) Atro 519/Df(3L)brm11 56 (55/99) Atro 1757/Df(3L)brm11 41 (46/111) Atro 2295/Df(3L)brm11 66 (78/118) Atro 3333/Df(3L)brm11 44 (53/121)

DISCUSSION

The strategy of identifying interacting genes by screening for dominant enhancers has a long and successful track record in Drosophila genetics. The general strategy in these screens is to select dominant enhancers and suppressors in a genetic background already compromised in the process of interest. The rationale is that in such “sensitized” backgrounds, reductions in the level of expression of interacting genes by 50% will in many cases have an effect on the phenotype. The key advantage of enhancer and suppressor screens is that they allow the recovery in heterozygotes of mutations that would normally be recognized only by their recessive effects. Although of tremendous utility, enhancer and suppressor screens have their limitations: Often pro-

% showing transformation of haltere to wing

Genotype

% showing transformation of A4 to A5 B. Interaction of Atro with trx

FRT79D-F/trx 3 Atro 519/trx 3 Atro 1757/trx 3 Atro 2295/trx 3 Atro 3333/trx 3

9.3 (12/129) 32 (9/28) 19 (5/27) 26 (7/27) 23 (9/39)

(A) Interaction with Df(3L)brm11. The percentage of flies showing patches of wing tissue in the haltere is shown for the indicated Atro/Df(3L)brm11 heterozygotes as well as for an Atro ⫹/Df(3L)brm11 control (FRT79D–F/Df(3L)brm11). (B) Interaction with trx 3. The percentage of flies showing patches of fourth abdominal segment (A4) tissue in A5 is shown for the indicated Atro/trx3 heterozygotes as well as for an Atro ⫹/trx 3 control (FRT79D–F/trx 3).

174

M. W. Kankel, D. M. Duncan and I. Duncan TABLE 6 Interaction of Atro alleles with the Polycomb group gene Pcl

Genotype n ⫽ 0 n ⫽ 1 n ⫽ 2 n ⫽ 3 n ⫽ 4 n ⫽ 5 n ⫽ 6 n ⫽ 7 n ⫽ 8 n ⫽ 9 n ⫽ 10 n ⫽ 11 No. of legs Pcl 1/Pcl 4 Atro 519 Atro 1757 Atro 2295 Atro 3333

7 6 23 5 16

1 2 7 0 1

0 6 16 5 5

1 10 9 7 9

10 27 13 9 8

5 24 10 15 19

12 10 6 12 17

18 9 3 12 13

21 6 6 15 8

26 0 4 12 3

2 0 3 6 0

1 0 0 2 1

104 100 100 100 100

sc/leg 6.73 4.43 3.39 6.08 4.60

⫾ ⫾ ⫾ ⫾ ⫾

2.55 1.93 2.90 2.66 2.69

The distribution of the number of sex comb teeth (sc) per second thoracic leg is shown for the indicated Atro/⫹ heterozygotes in a Pcl 1/Pcl 4 background.

the “single-hit” alleles are usually not well studied, in some cases they have turned out to identify genes involved in more general aspects of gene expression. The results of our ftz enhancer screen are consistent with this pattern. Of ⵑ5500 third chromosomes screened, only two genes were identified by multiple alleles: These were ftz-f1 (three alleles) and Atro (four alleles). The identification of ftz-f1 is important, as it demonstrates the efficacy of the screen. ftz-f1 encodes an orphan nuclear receptor protein that is an obligate dimerization partner for the segmentation functions of ftz. Embryos that lack maternally supplied Ftz-F1 display segmentation defects that are indistinguishable from ftz⫺ embryos (Guichet et al. 1997; Yu et al. 1997). The Ftz and Ftz-F1 proteins have been shown to interact cooperatively to activate expression of the ftz downstream target engrailed (Florence et al.1997; Guichet et al. 1997; Yu et al. 1997). Surprisingly, germ-line clones homozygous for any of the three ftz-f1 alleles isolated in our screen produce embryonic cuticles that are wild type for segmentation. Very likely, all three are weak alleles that enhance segmentation defects in ftz9H34/ftz f47ts heterozygotes, but have little effect on segmentation in ftz⫹ embryos. Possible roles for Atro are considered at the end of the discussion. In addition to alleles of ftz-f1 and Atro, 19 single-hit mutations were recovered. One of these is of particular interest, as it causes ftz-like segmentation defects. Mothers carrying germ-line clones homozygous for the bek1 mutation produce embryos that show variable segmentation defects. When such mothers are crossed to ⫹/⫹ or bek1/⫹ males, about half the embryos produced show some segmentation defect, most frequently deletion of the T3 denticle belt. However, when these mothers are crossed to ftz9H34 bek1/⫹ fathers, the frequency of defects increases to 63%, and the defects produced show a strong ftz-like pair-rule modulation. Stripes of en expression often show gaps and deletions in embryos from mothers with bek1 homozygous germ lines. These defects occur primarily, but not exclusively, in the ftz-dependent parasegments. Taken together, these observations suggest that Bek might function as a coactivator for Ftz in the regulation of en. However, bek must have other functions as well. Most embryos from bek1 mutant germ-

line mothers die early in development. Numerous nuclear abnormalities are present in these embryos, including irregularities in nuclear shape and size, chromatin bridges, mitotic asynchrony, and irregularities in nuclear localization. These abnormalities are closely similar to those described for embryos from mothers mutant for the Trithorax-like gene, which encodes the GAGA factor (Bhat et al. 1996). Ftz and Eve expression are also often abnormal, with stripes showing abnormal spacing or width. Additional alleles will need to be recovered for a better assessment of the role of the bek gene. Another single-hit enhancer is an allele of homothorax (hth), which encodes a TALE class homeodomain protein that functions along with Extradenticle (Exd) as a Hox protein cofactor. Hth associates with Exd and causes it to be transported to the nucleus, where Hth/ Exd heterodimers bind cooperatively with Hox proteins to tripartite-binding sites within target genes ( Jacobs et al. 1999; Ryoo et al. 1999; Ferretti et al. 2000; Gebelein et al. 2002). We were surprised to isolate a hth allele in our screen because Ftz lacks a motif (the hexapeptide or YPWM motif) present in Hox proteins that mediates interactions with Exd. If Hth does associate with Ftz, the association must not be obligatory for ftz function since hth⫺ embryos from hth⫺ germ-line clone mothers do not show regular pair-rule segmentation defects. However, such embryos do show a preferential loss of the ftz-dependent engrailed (en) stripes during midembryogenesis, which leads to variable segmental fusions (Rieckhof et al. 1997). Whether this loss of en stripes occurs because of a direct impairment of ftz function or because of some secondary effect is not clear. “GST pulldown” experiments are in progress to test for associations of Exd, Hth, and Ftz. Surprisingly, our hth2232 chromosome appears to have a weak maternal effect in its interaction with ftz (see Table 1). In previous studies (Rieckhof et al. 1997; Kurant et al. 1998), it has been shown that hth⫹ is not expressed maternally. Our results may be due to the presence of a linked mutation that has not been separated from hth2232. Two other single-hit alleles identify genes involved in translational regulation. These are held-out wings (how) and eukaryotic initiation factor-4E (eIF-4E). Surprisingly, our alleles of these genes do not enhance segmentation


defects, although they do cause strong enhancement of the lethality of ftz9H34/ftz f47ts heterozygotes. Indeed, our allele of how appears to suppress segmentation defects. how encodes a KH (hnRNP K homology) domain RNAbinding protein that is expressed at high levels in mesodermal derivatives (Lo and Frasch 1997; Baehrecke 1997; Zaffran et al. 1997). The activity of How has been best characterized in its interaction with the mRNA of stripe (sr) in tendon cells. how encodes two alternative isoforms that have opposing effects on sr RNA levels: A long form destabilizes sr mRNA, causing it to accumulate at low levels, whereas a shorter form stabilizes sr RNA and causes elevated accumulation (Nabel-Rosen et al. 1999, 2002). Both isoforms interact with the 3⬘-UTR of sr. Although the interacting RNA sequence has not been further localized, Nabel-Rosen et al. (2002) note that the sr 3⬘-UTR contains a sequence related to an element [the tra-2 and GLI element (TGE)] in the Caenorhabditis elegans tra-2 mRNA that is regulated by Gld-1, a homolog of how. The similarity consists of an 11-base match between the sr 3⬘-UTR and the 27-base TGE repeat. Interestingly, the ftz 3⬘-UTR contains a similar sequence, showing a match of 11 bases out of 15 in the same part of the TGE. This sequence overlaps the ftz polyadenylation sequence, but is distinct from the instability sequence defined by Riedl and JacobsLorena (1996). Although on its own this match would not seem significant, its coincidence with the same region of similarity between sr and the tra-2 TGE suggests that How might directly regulate ftz translation. Why does our how allele not affect segmentation? Transcription of how at blastoderm occurs only ventrally, in the region destined to become mesoderm. Therefore, the enhancing effects of how might be restricted to the mesoderm, which was not examined in our study. Alternatively, How might influence the function of ftz in the CNS or hindgut. eIF-4E is a subunit of the translation initiation complex, eIF-4F, which binds the 5⬘ ends of mRNAs and facilitates binding of the small ribosomal subunit. eIF4E binds the 7-methyl-guanosine cap at the 5⬘ end of the RNA. In mammalian cells, eIF-4E is a key point of translational regulation and is rate limiting for translation (see Gingras et al. 1999 for review). Therefore, we may have recovered an eIF-4E allele in our screen because it causes reduced levels of ftz translation. Why would this not enhance segmentation defects? One possibility is suggested by the finding that normal segmentation depends on a balance between Ftz and Eve levels (Kellerman et al. 1990; Hughes and Krause 2001). Therefore, if Eve and Ftz levels were reduced similarly by eIF-4E 2188, there might be no enhancement of segmentation defects. The enhanced lethality seen could then be due to impairment of other functions of Ftz, perhaps in the CNS or hindgut, that do not involve a balance with Eve. Fourteen additional single-hit enhancers were recov-

175

ered. All but 3 of these show convincing maternal effects, validating our strategy of screening for enhancement in the progeny of mothers carrying mutagentreated chromosomes. Some of these mutations, most notably 1427 and 3150, interact very strongly with ftz, enhancing both segmentation defects and lethality. Despite being one of the strongest enhancers isolated, mutant 3150 produces normal embryos from homozygous germ-line mothers. Mothers carrying homozygous germ-line clones for 1427 fail to lay eggs, rendering it difficult to assess the requirement of this gene for ftz function. Five of the 14 single-hit enhancers do not appear to enhance segmentation defects. However, two of these enhancers (4236 and 4758) do cause moderately strong enhancement of lethality, suggesting that they may, perhaps like eIF-4E, have some general function in gene expression. Duffy et al. (1996) tested deficiencies covering some 65% of the genome for dominant maternal interaction with the pair-rule gene runt (run). Surprisingly, there appears to be little overlap between the ftz-interacting genes identified in our screen and the run-interacting third chromosome deficiencies identified in their screen. Of the four deficiencies they found to interact strongly with run, two delete regions that contain ftz enhancers. However, it seems likely that the interacting genes are different in these cases, since for both deficiencies the run-interacting gene appears to be a Minute [M(3)67C and M(3)86D], whereas none of our ftz enhancers are Minute’s. Moreover, of the five deficiencies tested that remove strong ftz enhancers (ftz-f1, 1427, 2955, 3150, and 4038), none interacts strongly with run. Four of the ftz enhancers isolated in our screen proved to be alleles of Atrophin (Atro; Erkner et al. 2002; Zhang et al. 2002). Polyglutamine tract expansion within one of the human homologs of Atro (Atrophin-1) causes the neurodegenerative disease dentatorubral-pallidoluysian atrophy (Koide et al. 1994; Nagafuchi et al. 1994). Humans possess at least one additional Atrophin family member, Atrophin-2, which encodes a protein that can heterodimerize with Atr1 (Yanagisawa et al. 2000; Waerner et al. 2001). The functions of the mammalian Atrophin proteins are not well characterized. However, a role in gene repression seems likely, as Atr1 binds Eto1, a corepressor complex component, and overexpression of Atr1 can repress transcription of a reporter gene in tissue culture cells (Wood et al. 2000). In addition, Atr2 has been shown to interact with the histone deacetylase Hdac1 (Zoltewicz et al. 2004). Zhang et al. (2002) present compelling evidence that Atro also functions as a corepressor in Drosophila. They report that eve mutations show strong dominant lethality when crossed to mothers heterozygous for Atro alleles. In the eve/⫹; Atro/⫹ embryos produced in this cross, oddnumbered en stripes are expanded, suggesting a weakening in the ability of Eve to repress paired, runt, or sloppypaired, other pair-rule genes involved in specifying these

176


stripes. Zhang et al. (2002) also show that Atro binds to the minimal repression domain of Eve, and that artificial recruitment of Atro to a target gene can cause repression in vivo. A failure in the repressive activity of Eve may account for the absence of even-numbered en stripes we describe for embryos from Atro mutant germ-line mothers. In normal development, the even-numbered en stripes form as a result of differential repression of ftz and odd-skipped (odd) by Eve. Ftz is an activator of en, whereas Odd is a repressor. The even-numbered en stripes form where odd, but not ftz, has been repressed by Eve (Manoukian and Krause 1992; Fujioka et al. 1995). If there were a failure of Eve to repress odd, zones expressing ftz but not odd would not form, and the evennumbered en stripes would not be established. Exactly this mechanism appears to be responsible for a reduction in even-numbered en stripes in mutants for the Rpd3 histone deacetylase (Mannervik and Levine 1999). However, it is also possible that the even-numbered en stripes fail to appear in Atro⫺ embryos because of a defect in the ability of Ftz to activate en. It is important to note that the odd-numbered en stripes are established almost normally in Atro mutant embryos (although they are wider than normal; see Figure 5). These stripes are thought to be defined by differential repression of sloppy-paired, runt, and paired by Eve (Fujioka et al. 1995); the presence of these stripes in Atro⫺ embryos indicates that Atro is not required for all repressive activities of Eve. Although the atrophin proteins have largely been viewed as dedicated corepressors, our results indicate that Atro also functions in a positive fashion. We show that Atro is a member of the trxG of Hox gene positive regulators (Kennison and Tamkun 1988; Kennison 1995; Pirrotta 1998; Simon and Tamkun 2002). Mutations in trxG genes enhance the phenotypes of lossof-function alleles of the Hox genes and suppress the Hox gain-of-function phenotypes caused by mutations in Polycomb group genes. In otherwise wild-type backgrounds, trxG mutations also often cause weak transformations similar to those caused by Hox gene loss of function. Consistent with these effects, Atro mutations enhance haltere-to-wing transformations in Ubx heterozygotes, anteriorly directed transformations of the posterior abdominal segments in BX-C deficiency (Df(3R)P9) heterozygotes, and T1-to-T2 leg transformations in Df(3R)Scr heterozygotes. They also suppress the effects of the PcG gene Pcl and enhance the effects of the trxG mutations brm and trx. Finally, somatic clones homozygous for Atro alleles show partial transformations of the fifth and sixth abdominal segments to the anterior, transformations that are likely caused by loss of expression of the Abd-B gene of the BX-C. The genes of the trxG play diverse roles in promoting the transcription of the Hox genes and other loci. Several encode components of chromatin-remodeling complexes that function to render genes more accessible to

activators or to facilitate their transcription (for recent review see Simon and Tamkun 2002), whereas others assist enhancer-promoter interactions (e.g., Bickel and Pirrotta 1990; Mahmoudi et al. 2002). One of the best-characterized trxG genes is brahma (brm), which encodes a Drosophila homolog of the yeast SWI2/SNF2 protein (Tamkun et al. 1992). SWI2/SNF2 is the catalytic DNA-stimulated ATPase subunit of a large multiprotein complex that functions in chromatin remodeling. Brm is also part of a large protein complex, which consists of a core of 10 tightly associated proteins as well as several loosely associated factors (Papoulas et al. 1998; Kal et al. 2000). This complex can function in vitro to alter nucleosome spacing and to enhance transcription (Kal et al. 2000). Three of the Brm complex proteins are encoded by trxG genes (brm, osa, and mor) and four (Brm, Snr1, BAP155, and BAP60) are conserved in the yeast remodeling complexes SWI/SNF and RSC. The proteins of the Brm complex show a surprising degree of functional specialization: Mutations in some show strong trxG phenotypes, whereas others (e.g., snr1; Marenda et al. 2003) show no clear homeotic phenotypes. Although the Brm complex is primarily involved in gene activation, it also functions in repression, particularly for genes that are targets of wingless signaling (Treisman et al. 1997; Collins et al. 1999; Collins and Treisman 2000). At least two other trxG protein complexes are known, one containing Ash1 and the other Ash2 (Papoulas et al. 1998). Ash1 functions as a multifunctional histone methyl transferase whose activity may recruit Brm complexes to target genes (Beisel et al. 2002). The function of Ash2 is not yet known. Atro appears particularly closely related to the Brm complex in function. The parallels between Atro and the Brm complex component Osa are the most striking, as both proteins appear to be intimately involved in regulating wingless (wg) targets. osa encodes a subunit present in some, but not all, Brm complexes (Collins et al. 1999). Treisman and colleagues (Treisman et al. 1997; Collins et al. 1999; Collins and Treisman 2000) have shown that Osa-containing brahma chromatinremodeling complexes are required for the normal expression of several wg targets, including dpp, Dll, nubbin, en, and the UbxB midgut enhancer. In each case, loss of osa function causes ectopic expression of the target, indicating that osa is required for the normal repression of these targets. brahma (brm) and moira (mor), which encode other components of the Brm complex, are also required for this repression, at least for nubbin and the UbxB enhancer (Collins and Treisman 2000). Although Atro has not been as well studied, our results and those of Erkner et al. (2002) suggest that Atro is also involved in wg target regulation. In the leg, wg specifies ventral characteristics (Struhl and Basler 1993). Atro⫺ clones located ventrally in the leg show numerous pattern deletions, whereas clones located dorsally develop almost normally. In the proximo-ven-


tral portion of the leg, some Atro⫺ clones show ectopic expression of dpp or Dll (Erkner et al. 2002; this report), both targets of wg signaling (Brook and Cohen 1996; Jiang and Struhl 1996; Johnston and Schubiger 1996; Theisen et al. 1996; Lecuit and Cohen 1997) that also require osa for their normal repression. The EGL-27 protein, which shows sequence similarity with the N-terminal portion of Atro, may function similarly, as it is required for normal Wnt signaling in C. elegans (Herman et al. 1999). There are additional similarities between Atro alleles and Brm complex mutations. Atro alleles cause ectopic wing venation and bristle defects in homozygous clones; similar phenotypes are caused by loss-of-function of the Brm complex genes brm (Brizuela et al. 1994; Elfring et al. 1998), mor (Brizuela and Kennison 1997), osa (Treisman et al. 1997; Collins et al. 1999), and snr1 (Marenda et al. 2003; Zraly et al. 2003). Atro alleles suppress the antenna-to-leg transformation caused by AntpNs, in which expression is driven by the endogenous Antp P2 promoter, but not Antp73b, in which expression is driven by the promoter of a non-Hox gene. The same promoter specificity is shown by alleles of brm (Tamkun et al. 1992), mor (Brizuela and Kennison 1997), and osa (Va´zquez et al. 1999). Although no clear pair-rule modulation has been described for Brm complex members, severe segmentation defects are seen in embryos from osa homozygous germ-line mothers (Treisman et al. 1997; Va´zquez et al. 1999) and from mothers heterozygous for two partially complementing alleles of brm (Brizuela et al. 1994). For embryos from osa homozygous germ-line mothers, expression of gap gene proteins is normal, but the pair-rule stripes of eve are abnormal (Treisman et al. 1997). Gap gene expression also appears normal in embryos from Atro homozygous germ-line mothers (Zhang et al. 2002; but see Erkner et al. 2002), but expression of eve and other pair-rule genes is abnormal. The similar effects of Atro alleles and of mutations affecting Brm complex subunits suggest a close functional relationship between the two. Atro is not among the core Brm complex components identified by Papoulas et al. (1998). However, Atro could be one of the high-molecular-weight Brm complex components identified by Kal et al. (2000) or a core component of another Brm-like complex that has yet to be characterized. Consistent with the latter possibility, the N-terminal portion of Atro shares homology with the N-terminal portion of mammalian metastasis-associated protein 2 (Mta2), a component of the NURD chromatin-remodeling complex (Xue et al. 1998; Zhang et al. 1999). NURD has nucleosome remodeling and histone deacetylase activity. The N-terminal regions of Atr2 and EGL-27 also contain Mta2 homologous regions, and in Atr2 this region has been shown to interact with Hdac1, suggesting that Atr2 may be part of a novel histone deacetylase complex (Zoltewicz et al. 2004). These observations suggest

177

that Atro might also be part of a chromatin-modifying complex. Although Atro has been considered only as a transcriptional repressor, our finding that Atro is a member of the trithorax group indicates that it also plays a positive role. In previous reports (Erkner et al. 2002; Zhang et al. 2002), defects in the ftz striping pattern in Atro⫺ embryos have been interpreted as resulting from a failure in the repression ability of segmentation gene products. However, our results suggest these defects could result from a failure in stripe activation. The ftz striping pattern in Atro-mutant embryos bears a striking resemblance to the ftz pattern in embryos lacking the pairrule gene runt (run) or the maternal gene caudal (cad). In run⫺ embryos ftz stripes 2–6 look essentially the same as in Atro mutant embryos. However, stripe 1 is less well developed and stripe 7 is better developed in run⫺ embryos than in Atro-mutant embryos (Carroll and Scott 1986; Tsai and Gergen 1995; Yu and Pick 1995). In cad⫺ embryos, stripes 1–4 are very similar to these stripes in Atro-mutant embryos, but stripes 5–7 are less well developed (Macdonald and Struhl 1986). Both run⫹ and cad⫹ are known to function as activators of the ftz zebra element (Dearolf et al. 1989; Tsai and Gergen 1995). These observations suggest that Atro might serve as a coactivator for Run and/or Cad, perhaps by mediating recruitment of the Brm complex or other chromatin remodeling complexes to the ftz zebra element. We thank members of the Duncan lab for their support and advice over the course of this work. This article was improved by the comments of Jennifer Brisson, Gabriella Farkas, Boris Leibovitch, and the anonymous reviewers. We also thank Kathy Matthews and Kevin Cook of the Bloomington Stock Center for countless stock requests, Dr. Jim Kennison for sending trxG gene mutations, and Dr. Ross Cagan for providing a set of third-chromosome lethal P-element insertion lines. This work was supported by a National Institutes of Health grant GM32318 to I.D.

LITERATURE CITED Ananthan, J., R. Baler, D. Morrissey, J. Zuo, Y. Lan et al., 1993 Synergistic activation of transcription is mediated by the N-terminal domain of Drosophila fushi tarazu homeoprotein and can occur without DNA binding by the protein. Mol. Cell. Biol. 13: 1599–1609. Baehrecke, E., 1997 who encodes a KH RNA binding protein that functions in muscle development. Development 124: 1323–1332. Basler, K., and G. Struhl, 1994 Compartment boundaries and the control of Drosophila limb pattern by hedgehog protein. Nature 368: 208–214. Beisel, C., A. Imhof, J. Greene, E. Kremmer and F. Sauer, 2002 Histone methylation by the Drosophila epigenetic transcriptional regulator Ash1. Nature 419: 857–862. Bhat, K. M., G. Farkas, F. Karch, H. Gyurkovics, J. Gausz et al., 1996 The GAGA factor is required in the early Drosophila embryo not only for transcriptional regulation but also for nuclear division. Development 122: 1113–1124. Bickel, S., and V. Pirrotta, 1990 Self-association of the Drosophila zeste protein is responsible for transvection effects. EMBO J. 9: 2959–2967. Brizuela, B. J., and J. A. Kennison, 1997 The Drosophila homeotic

178


gene moira regulates expression of engrailed and HOM genes in imaginal tissues. Mech. Dev. 65: 209–220. Brizuela, B. J., L. Elfring, J. Ballard, J. Tamkun and J. A. Kennison, 1994 Genetic analysis of the brahma gene of Drosophila melanogaster and polytene chromosome subdivisions 72AB. Genetics 137: 803–813. Brook, W. J., and S. M. Cohen, 1996 Antagonistic interactions between Wingless and Decapentaplegic responsible for dorsal-ventral pattern in the Drosophila leg. Science 273: 1373–1377. Brown, J. L., S. Sonoda, H. Ueda, M. P. Scott and C. Wu, 1991 Repression of the Drosophila fushi tarazu (ftz) segmentation gene. EMBO J. 10: 665–674. Carroll, S. B., and M. P. Scott, 1985 Localization of the fushi tarazu protein during Drosophila embryogenesis. Cell 43: 47–57. Carroll, S. B., and M. P. Scott, 1986 Zygotically active genes that affect the spatial expression of the fushi tarazu segmentation gene during early Drosophila embryogenesis. Cell 45: 113–126. Celniker, S. E., S. Sharma, D. J. Keelan and E. B. Lewis, 1990 The molecular genetics of the Bithorax complex of Drosophila : cisregulation in the Abdominal-B domain. EMBO J. 9: 4277–4286. Chou, T., and N. Perrimon, 1996 The autosomal FLP-DFS technique for generating germline mosaics in Drosophila melanogaster. Genetics 144: 1673–1679. Collins, R. T., and J. E. Treisman, 2000 Osa-containing Brahma chromatin remodeling complexes are required for the repression of Wingless target genes. Genes Dev. 14: 3140–3152. Collins, R. T., T. Furukawa, N. Tanese and J. E. Treisman, 1999 Osa associates with the Brahma chromatin remodeling complex and promotes the activation of some target genes. EMBO J. 18: 7029–7040. Copeland, J. R., A. Nasiadka, B. H. Deitrich and H. M. Krause, 1996 Patterning of the Drosophila embryo by a homeodomaindeleted Ftz polypeptide. Nature 379: 162–165. Dearolf, C. R., J. Topol and C. S. Parker, 1989 The caudal gene product is a direct activator of fushi tarazu transcription during Drosophila embryogenesis. Nature 341: 340–343. Doe, C. Q., Y. Hiromi, W. J. Gehring and C. S. Goodman, 1987 Expression and function of the segmentation gene fushi tarazu during Drosophila neurogenesis. Science 239: 170–175. Dong, J., L.-H. Hung, R. Strome and H. M. Krause, 1998 A phosphorylation site in the Ftz homeodomain is required for activity. EMBO J. 17: 2308–2318. Duffy, J. B., J. Wells and J. P. Gergen, 1996 Dosage-sensitive maternal modifiers of the Drosophila segmentation gene runt. Genetics 142: 839–852. Duncan, D. M., E. A. Burgess and I. Duncan, 1998 Control of distal antennal identity and tarsal development in Drosophila by spineless-aristapedia, a homolog of the mammalian dioxin receptor. Genes Dev. 12: 1290–1303. Duncan, I., 1987 The bithorax complex. Annu. Rev. Genet. 21: 285–319. Duncan, I. M., 1982 Polycomblike : a gene that appears to be required for the normal expression of the Bithorax and Antennapedia gene complexes of Drosophila melanogaster. Genetics 102: 49–70. Duncan, I., and E. B. Lewis, 1982 Genetic control of body segment differentiation in Drosophila, pp 533–554 in Developmental Order: Its Origin and Regulation. The 40th Symposium of the Society for Developmental Biology, edited by S. Subtelny. Alan Liss, New York. Edgar, B. A., M. P. Weir, G. Schubiger and P. H. O’Farrell, 1986 Repression and turnover pattern fushi tarazu RNA in the early Drosophila embryo. Cell 47: 747–754. Elfring, L. K., C. Daniel, O. Papoulas, R. Deuring, M. Sarte et al., 1998 Genetic analysis of brahma: the Drosophila homolog of the yeast chromatin remodeling factor SWI2/SNF2. Genetics 148: 251–265. Emmons, R. B., D. Duncan, P. A. Estes, P. Kiefel, J. T. Mosher et al., 1999 The spineless-aristapedia and tango bHLH-PAS proteins interact to control antennal and tarsal development in Drosophila. Development 126: 3937–3945. Erkner, A., A. Roure, B. Charroux, M. Delaage, N. Holway et al., 2002 Grunge, related to human Atrophin-like proteins, has multiple functions in Drosophila development. Development 129: 1119–1129. Ferretti, E., H. Marshall, H. Popperl, M. Maconochie, R. Krumlauf et al., 2000 Segmental expression of Hoxb2 in r4 requires

two separate sites that integrate cooperative interactions between Prep1, Pbx, and Hox proteins. Development 127: 155–166. Fitzpatrick, V. D., and C. J. Ingles, 1989 The Drosophila fushi tarazu polypeptide is a DNA-binding transcriptional activator in yeast cells. Nature 337: 666–668. Fitzpatrick, V. D., A. Percival-Smith, C. J. Ingles and H. M. Krause, 1992 Homeodomain-independent activity of the fushi tarazu polypeptide in Drosophila embryos. Nature 356: 610–612. Florence, B., and W. McGinnis, 1998 A genetic screen of the Drosophila X chromosome for mutations that modify Deformed function. Genetics 150: 1497–1511. Florence, B., A. Guichet, A. Ephrussi and A. Laughon, 1997 FtzF1 is a cofactor in Ftz activation of the Drosophila engrailed gene. Development 124: 839–847. Frischer, L. E., F. S. Hagen and R. L. Garber, 1986 An inversion that disrupts the Antennapedia gene causes abnormal structure and localization of RNAs. Cell 47: 1017–1023. Fujioka, M., J. B. Jaynes and T. Goto, 1995 Early even-skipped stripes act as morphogenetic gradients at the single cell level to establish engrailed expression. Development 121: 4371–4382. Gebelein, B., J. Culi, H. D. Ryoo, W. Zhang and R. S. Mann, 2002 Specificity of Distalless repression and limb primordia development by abdominal Hox proteins. Dev. Cell 3: 487–498. Gellon, G., K. W. Harding, N. McGinnis, N. 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. Gingras, A.-C., B. Raught and N. Sonenberg, 1999 eIF4 initiation factors: effectors of mRNA recruitment to ribosomes and regulators of translation. Annu. Rev. Biochem. 68: 913–963. Golic, K. G., 1991 Site-specific recombination between homologous chromosomes in Drosophila. Science 252: 958–961. Guichet, A., J. W. R. Copeland, M. Erdelyl, D. Hlousek, P. Zavorszky et al., 1997 The nuclear receptor homologue Ftz-F1 and the homeodomain protein Ftz are mutually dependent cofactors. Nature 385: 548–552. Han, W., Y. Yu, N. Altan and L. Pick, 1993 Multiple proteins interact with the fushi tarazu proximal enhancer. Mol. Cell. Biol. 13: 5549–5559. Harding, K. W., G. Gellon, N. McGinnis and W. McGinnis, 1995 A screen for modifiers of Deformed function in Drosophila. Genetics 140: 1339–1352. Harrison, S. D., and A. A. Travers, 1990 The tramtrack gene encodes a Drosophila finger protein that interacts with the ftz transcriptional regulatory region and shows a novel embryonic expression pattern. EMBO J. 9: 207–216. Herman, M. A., Q. Ch’ng, S. M. Hettenbach, T. M. Ratliff, C. Kenyon et al., 1999 EGL-27 is similar to a metastasis-associated factor and controls cell polarity and cell migration in C. elegans. Development 126: 1055–1064. Hiromi, Y., and W. J. Gehring, 1987 Regulation and function of the Drosophila segmentation gene fushi tarazu. Cell 50: 963–974. Hiromi, Y., A. Kuroiwa and W. J. Gehring, 1985 Control elements of the Drosophila segmentation gene fushi tarazu. Cell 43: 603– 613. Hughes, S. C., and H. M. Krause, 2001 Establishment and maintenance of parasegmental compartments. Development 128: 1109– 1118. Jacobs, Y., C. A. Schnabel and M. L. Cleary, 1999 Trimeric association of Hox and TALE homeodomain proteins mediates Hoxb2 hindbrain enhancer activity. Mol. Cell. Biol. 19: 5134–5142. Jiang, J., and G. Struhl, 1996 Complementary and mutually exclusive activities of Decapentaplegic and Wingless organize axial pattern during Drosophila limb development. Cell 86: 401–409. Johnston, L. A., and G. Schubiger, 1996 Ectopic expression of wingless in imaginal discs interferes with decapentaplegic expression and alters cell determination. Development 122: 3519–3529. Jorgenson, E. M., and R. L. Garber, 1987 Function and misfunction of the two promoters of the Drosophila Antennapedia gene. Genes Dev. 1: 544–555. Ju¨rgens, G., E. Wieschaus, C. Nu¨sslein-Volhard and H. Kluding, 1984 Mutations affecting the pattern of the larval cuticle of Drosophila melanogaster. II. Zygotic loci on the third chromosome. Roux’s Arch. Dev. Biol. 193: 283–295. Kal, A. J., T. Mahmoudi, N. B. Zak and C. P. Verrijzer, 2000 The

ftz-Interacting Genes in Drosophila Drosophila Brahma complex is an essential coactivator for the trithorax group protein Zeste. Genes Dev. 14: 1058–1071. Karim, F. D., H. C. Chang, M. Therrien, D. A. Wassarman, T. Laverty et al., 1996 A screen for genes that function downstream of Ras1 during Drosophila eye development. Genetics 143: 315–329. Kellerman, K. A., D. M. Mattson and I. Duncan, 1990 Mutations affecting the stability of the fushi tarazu protein of Drosophila. Genes Dev. 4: 1936–1950. Kennison, J. A., 1995 The Polycomb and trithorax group proteins of Drosophila: trans-regulators of homeotic gene function. Annu. Rev. Genet. 27: 289–303. Kennison, J. A., and J. W. Tamkun, 1988 Dosage-dependent modifiers of homeotic mutations in Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 85: 8136–8140. Koide, R., T. Ikeuchi, O. Onodera, H. Tanaka, S. Igarashi et al., 1994 Unstable expansion of CAG repeat in hereditary dentatorubral-pallidoluysian atrophy (DRPLA). Nat. Genet. 6: 9–13. Krause, H. M., R. Klemenz and W. J. Gehring, 1988 Expression, modification, and localization of the fushi tarazu protein in Drosophila embryos. Genes Dev. 2: 1021–1036. Kurant, E., C. Y. Pai, R. Sharf, N. Halachmi, Y. Sun et al., 1998 dorsotonals/homothorax, the Drosophila homologue of meis1, interacts with extradenticle in patterning of the embryonic PNS. Development 125: 1037–1048. Laughon, A., and M. P. Scott, 1984 Sequence of a Drosophila segmentation gene: protein structure homology with DNA-binding proteins. Nature 310: 25–31. Lavorgna, G., H. Ueda, J. Clos and C. Wu, 1991 FTZ-F1, a steroid hormone receptor-like protein implicated in the activation of fushi tarazu. Science 252: 848–851. Lawrence, P. A., P. Johnston, P. Macdonald and G. Struhl, 1987 Borders of parasegments in Drosophila embryos are delimited by the fushi tarazu and even-skipped genes. Nature 328: 440–442. Lecuit, T., and S. M. Cohen, 1997 Proximal-distal axis formation in the Drosophila leg. Nature 388: 139–145. Lewis, E. B., 1960 A new standard food medium. Dros. Inf. Serv. 34: 117–118. Lewis, E. B., 1978 A gene complex controlling segmentation in Drosophila. Nature 276: 565–570. Lewis, E. B., and F. Bacher, 1968 Method of feeding ethyl methane sulfonate (EMS) to Drosophila males. Dros. Inf. Serv. 43: 193. Lewis, R. A., B. T. Wakimoto, R. E. Dennel and T. C. Kaufman, 1980 Genetic analysis of the Antennapedia gene complex (ANT-C) and adjacent chromosomal segments of Drosophila melanogaster. II. Polytene chromosome segments 84A1–B1,2. Genetics 95: 383–397. Lindsley, D. L., and G. G. Zimm, 1992 The Genome of Drosophila melanogaster. Academic Press, San Diego. Lo, P. C. H., and M. Frasch, 1997 A novel KH-domain protein mediates cell adhesion processes in Drosophila. Dev. Biol. 190: 241–256. MacDonald, P. M., and G. Struhl, 1986 A molecular gradient in early Drosophila embryos and its role in specifying the body pattern. Nature 324: 537–545. Mahmoudi, T., K. R. Katsani and C. P. Verrijzer, 2002 GAGA can mediate enhancer function in trans by linking two separate DNA molecules. EMBO J. 21: 1775–1781. Mannervik, M., and M. Levine, 1999 The Rpd3 histone deacetylase is required for segmentation of the Drosophila embryo. Proc. Natl. Acad. Sci. USA 96: 6797–6801. Manoukian, A. S., and H. Krause, 1992 Concentration dependent activities of the even-skipped protein in Drosophila embryos. Genes Dev. 6: 1740–1751. Marenda, D. R., C. B. Zraly, Y. Feng, S. Egan and A. K. Dingwall, 2003 The Drosophila SNR1 (SNF5/INI1) subunit directs essential developmental functions of the Brahma chromatin remodeling complex. Mol. Cell. Biol. 23: 289–305. Nabel-Rosen, H., N. Dorevithc, A. Reuveny and T. Volk, 1999 The balance between two isoforms of the Drosophila RNA-binding protein how controls tendon cell differentiation. Mol. Cell 4: 573–584. Nabel-Rosen, H., G. Volohonsky, A. Reuveny, R. Zaidel-Bar and T. Volk, 2002 Two isoforms of the Drosophila RNA binding protein, How, act in opposing directions to regulate tendon cell differentiation. Dev. Cell 2: 183–193. Nagafuchi, S., H. Yanagisawa, E. Ohsaki, T. Shirayama, K. Tado-

179

koro et al., 1994 Structure and expression of the gene responsible for the triplet repeat disorder, dentatorubral and pallidoluysian atrophy (DRPLA). Nat. Genet. 8: 177–182. Nasiadka, A., and H. M. Krause, 1999 Kinetic analysis of segmentation gene interactions in Drosophila embryos. Development 126: 1515–1526. Okada, M., and S. Hirose, 1998 Chromatin remodeling mediated by Drosophila GAGA factor and ISWI activates fushi tarazu gene transcription in vitro. Mol. Cell. Biol. 18: 2455–2461. Papoulas, O., S. J. Beek, S. L. Moseley, C. M. McCallum, M. Sarte et al., 1998 The Drosophila trithorax group proteins BRM, ASH1 and ASH2 are subunits of distinct protein complexes. Development 125: 3955–3966. Pattatucci, A. M., and T. C. Kaufman, 1991 The homeotic gene Sex combs reduced of Drosophila melanogaster is differentially regulated in the embryonic and imaginal stages of development. Genetics 129: 443–461. Pick, L., A. Schier, M. Affolter, T. Schmidt-Glenewinkel and W. J. Gehring, 1990 Analysis of the ftz upstream element: germ layer-specific enhancers are independently autoregulated. Genes Dev. 4: 1224–1239. Pirrotta, V., 1998 Polycombing the Genome: PcG, trxG, and chromatin silencing. Cell 93: 333–336. Raftery, L. A., V. Twombly, K. Warton and W. M. Gelbart, 1995 Genetic screens to identify elements of the decapentaplegic signaling pathway in Drosophila. Genetics 139: 241–254. Reuter, G., and I. Wolff, 1981 Isolation of dominant suppressor mutations for position-effect variegation in Drosophila melanogaster. Mol. Gen. Genet. 182: 516–519. Rieckhof, G. E., F. Casares, H. D. Ryoo, M. Abu-Shaar and R. S. Mann, 1997 Nuclear translocation of Extradenticle requires homothorax, which encodes an Extradenticle-related homeodomain protein. Cell 91: 171–183. Riedl, A., and M. Jacobs-Lorena, 1996 Determinants of Drosophila fushi tarazu mRNA instability. Mol. Cell. Biol. 16: 3047–3053. Ryoo, H. D., T. Marty, F. Casares, M. Affolter and R. S. Mann, 1999 Regulation of Hox target genes by a DNA bound Homothorax/Hox/Extradenticle complex. Development 126: 5137– 5148. Schier, A. F., and W. J. Gehring, 1992 Direct homeodomain-DNA interaction in the autoregulation of the fushi tarazu gene. Nature 356: 804–807. Schier, A. F., and W. J. Gehring, 1993 Analysis of a fushi tarazu autoregulatory element: multiple sequence elements contribute to enhancer activity. EMBO J. 12: 1111–1119. Schneuwly, S., A. Kuroiwa and W. J. Gehring, 1987 Molecular analysis of the dominant homeotic Antennapedia phenotype. EMBO J. 6: 201–206. Schwartz, C. J. E., H. M. Sampson, D. Hlousek, A. Percival-Smith, J. W. R. Copeland et al., 2001 FTZ-Factor1 and Fushi tarazu interact via conserved nuclear receptor and coactivator motifs. EMBO J. 20: 510–519. Simon, J. A., and J. W. Tamkun, 2002 Programming off and on states in chromatin: mechanisms of Polycomb and trithorax group complexes. Curr. Opin. Genet. Dev. 12: 210–218. Simon, M. A., D. D. L. Bowtell, G. S. Dodson, T. R. Laverty and G. M. Rubin, 1991 Ras1 and a putative guanine nucleotide exchange factor perform crucial steps in signaling by the sevenless protein tyrosine kinase. Cell 67: 701–716. Spradling, A. C., D. Stern, A. Beaton, E. J. Rhem, T. Laverty et al., 1999 The Berkeley Drosophila genome project gene disruption project. Single P-element insertions mutating 25% of vital Drosophila genes. Genetics 153: 135–177. Struhl, G., and K. Basler, 1993 Organizing activity of wingless protein in Drosophila. Cell 72: 527–540. Suzuki, T., H. Kawasaki, R. T. Yu, H. Ueda and K. Umesono, 2001 Segmentation gene product Fushi tarazu is an LXXLL motifdependent coactivator for orphan receptor FTZ-F1. Proc. Natl. Acad. Sci. USA 98: 12403–12408. Tabata, T., and T. B. Kornberg, 1994 Hedgehog is a signaling protein with a key role in patterning Drosophila imaginal discs. Cell 76: 89–102. Tabata, T., S. Eaton and T. B. Kornberg, 1992 The Drosophila hedgehog gene is expressed specifically in posterior compartment cells and is a target of engrailed regulation. Genes Dev. 6: 2635– 2645.

180


Tamkun, J. W., R. Deuring, M. P. Scott, M. Kissinger, A. Pattatucci et al., 1992 brahma : a regulator of Drosophila homeotic genes structurally related to the yeast transcriptional activator SNF2/SWI2. Cell 68: 561–572. Tanenbaum, S. B., S. M. Gorski, J. C. Rusconi and R. L. Cagan, 2000 A screen for dominant modifiers of the irreC-rst cell death phenotype in the developing Drosophila retina. Genetics 156: 205–217. Theisen, H., T. E. Haerry, M. B. O’Connor and J. L. Marsh, 1996 Developmental territories created by mutual antagonism between Wingless and Decapentaplegic. Development 122: 3939–3948. Topol, J., C. R. Dearolf, K. Prakash and C. S. Parker, 1991 Synthetic oligonucleotides recreate Drosophila fushi tarazu zebra-stripe expression. Genes Dev. 5: 855–867. Treisman, J. E., A. Luk, G. M. Rubin and U. Heberlein, 1997 eyelid antagonizes wingless signaling during Drosophila development and has homology to the Bright family of DNA-binding proteins. Genes Dev. 11: 1949–1952. Tsai, C., and P. Gergen, 1995 Pair-rule expression of the Drosophila fushi tarazu gene: a nuclear receptor response element mediates the opposing regulatory effects of runt and hairy. Development 121: 435–462. Twombly, V., R. K. Blackman, H. Jin, J. M. Graff, R. W. Padgett et al., 1996 The TGF-␤ signaling pathway is essential for Drosophila oogenesis. Development 122: 1555–1565. Ueda, H., S. Sonoda, J. L. Brown, M. P. Scott and C. Wu, 1990 A sequence-specific DNA-binding protein that activates fushi tarazu segmentation gene expression. Genes Dev. 4: 624–635. Va´zquez, M., L. Moore and J. A. Kennison, 1999 The trithorax group gene osa encodes an ARID-domain protein that genetically interacts with the Brahma chromatin-remodeling factor to regulate transcription. Development 126: 733–742. Waerner, T., P. Gardellin, K. Pfizenmaier, A. Weith and N. Kraut, 2001 Human RERE is localized to nuclear promyelocytic leukemia oncogenic domains and enhances apoptosis. Cell Growth Differ. 12: 201–210. Wakimoto, B. T., F. R. Turner and T. C. Kaufman, 1984 Defects in embryogenesis in mutants associated with the Antennapedia gene complex of Drosophila melanogaster. Dev. Biol. 102: 147–172. Weir, M. P., and T. Kornberg, 1985 Patterns of engrailed and fushi tarazu transcripts reveal novel intermediate stages of Drosophila segmentation. Nature 318: 433–439.

Wood, J. D., F. C. Nucifora, Jr., K. Duan, C. Zhang, J. Wang et al., 2000 Atrophin-1, the dentato-rubral and pallido-luysian atrophy gene product, interacts with ETO/MTG8 in the nuclear matrix and represses transcription. J. Cell Biol. 150: 939–948. Xu, T., and G. M. Rubin, 1993 Analysis of genetic mosaics in developing and adult Drosophila tissues. Development 117: 1223–1237. ˆ te´ et al., 1998 Xue, Y., J. Wong, G. T. Moreno, M. K. Young, J. Co NURD, a novel complex with both ATP-dependent chromatinremodeling and histone deacetylase activities. Mol. Cell 2: 851– 861. Yanagisawa, H., M. Bundo, T. Miyashita, Y. Okamura-Oho, K. Tadokoro et al., 2000 Protein binding of a DRPLA family through arginine-glutamic acid dipeptide repeats is enhanced by extended polyglutamine. Hum. Mol. Genet. 9: 1433–1442. Yu, Y., and L. Pick, 1995 Non-periodic cues generate seven ftz stripes in the Drosophila embryo. Mech. Dev. 50: 163–175. Yu, Y., W. Li, K. Su, M. Yussa, W. Han et al., 1997 The nuclear hormone receptor Ftz-F1 is a cofactor for the Drosophila homeodomain protein Ftz. Nature 385: 552–556. Yussa, M., U. Lo¨hre, K. Su and L. Pick, 2001 The nuclear receptor Ftz-F1 and homeodomain protein Ftz interact through evolutionarily conserved protein domains. Mech. Dev. 107: 39–53. Zaffran, S., M. Astier, D. Gratecos and M. Semeriva, 1997 The held out wings (how) Drosophila gene encodes a putative RNAbinding protein involved in the control of muscular and cardiac activity. Development 124: 2087–2098. Zhang, S., L. Xu, J. Lee and T. Xu, 2002 Drosophila Atrophin homolog functions as a transcriptional corepressor in multiple developmental processes. Cell 108: 45–56. Zhang, Y., H. Ng, H. Erdjument-Bromage, P. Tempst, A. Bird et al., 1999 Analysis of the NuRD subunits reveals a histone deacetylase core complex and a connection with DNA methylation. Genes Dev. 13: 1924–1935. Zoltewicz, J. S., N. J. Stewart, R. Leung and A. S. Peterson, 2004 Atrophin 2 recruits histone deacetylase and is required for the function of multiple signaling centers during mouse embryogenesis. Development 131: 3–14. Zraly, C. B., D. R. Marenda, R. Nanchal, G. Cavalli, C. Muchardt et al., 2003 SNR1 is an essential subunit in a subset of Drosophila Brm complexes, targeting specific functions during development. Dev. Biol. 253: 291–308. Communicating editor: T. Kaufman