Drosophila stocks:

Unless otherwise noted, all crosses were carried out at 25° on standard cornmeal–molasses–agar medium. BL number refers to Bloomington Stock Center stock number. Stocks used: ato¹⁰⁹⁰ (also called ato^ts, described here), Df(3R)p13 [BL 1943 ato¹ and ato³ (Jarman et al. 1994, 1995)], atonal-lacZ enhancer trap lines (both 5′F:9.3 and 3′F:5.8) are described in Sun et al. (1998), hh^AC (Lee et al. 1992), Egfr^E1 (Baker and Rubin 1989, 1992), lilli^XS407 (gift from Amy Tang) (Tang et al. 2001), lilli^XS575 (gift from Amy Tang) (Tang et al. 2001), lilli^A17-2 (gift from Arno Muller) (Muller et al. 2005), lilli^S35 (Neufeld et al. 1998a), kis¹ (Kennison and Tamkun 1988), kis^k13416 (Roch et al. 1998; Srinivasan et al. 2005), roe¹ (Ma et al. 1996), roe³ (St Pierre et al. 2002), rn¹⁹ (St Pierre et al. 2002), rn¹⁶ (St Pierre et al. 2002), rn⁵ (Agnel et al. 1989), smo³ P{neo FRT} 40A (Vrailas and Moses 2006), tkv⁸ P{neo FRT} 40A (Vrailas and Moses 2006), smo³, tkv⁸ P{neo FRT} 40A (Vrailas and Moses 2006), da^UX P{neo FRT} 40A (synonym da¹⁰, gift from Claire Cronmiller) (Brown et al. 1996), da¹ (Bell 1954), eya² (Bonini et al. 1993), Star¹ (Lewis 1945; Higson et al. 1993), Dl^RevF10 Ser^RX82 P{neoFRT}82B (Zeng et al. 1998), osa³⁰⁸ P{neoFRT}82B (Treisman et al. 1997; Janody et al. 2004), trx^E2 P{neoFRT}82B (Janody et al. 2004), brm^T485 P{neoFRT}80B (Janody et al. 2004), and snr1^R3 P{neoFRT}82B (Zraly et al. 2003).

Mosaic clones:

Mosaic clones were generated using ey:FLP (Newsome et al. 2000), as described (Xu and Rubin 1993). Flip stocks (1) y⁻, w⁻, ey:FLP; Ubi-GFP, P{neoFRT}80B; (2) y⁻, w⁻, ey:FLP; Ubi-GFP, P{neoFRT}82B; and (3) y⁻, w⁻, ey:FLP; Ubi-GFP, P{neoFRT}40A were crossed to the following stocks as appropriate to generate mutant clones:

1. w⁻; kis^LM27, P{neoFRT}40A/Cyo

2. w⁻; kis^LM27, P{neoFRT}40A, 5′ ato-lacZ/Cyo

3. w⁻; kis^LM27, P{neoFRT}40A, 3′ ato-lacZ/Cyo

4. w⁻; kis^EC1, P{neoFRT}40A/Cyo

5. w⁻; smo³ P{neo FRT} 40A/Cyo

6. w⁻; tkv⁸ P{neo FRT} 40A/Cyo

7. w⁻; smo³, tkv⁸, P{neo FRT} 40A/Cyo

8. Dl^RevF10 Ser^RX82 P{neoFRT}82B/TM6B

9. w⁻; da^UX P{neo FRT} 40A/Cyo

10. w⁻; da^UX P{neo FRT} 40A, 5′ ato-lacZ/Cyo

11. w⁻; da^UX P{neo FRT} 40A, 3′ ato-lacZ/Cyo

12. w⁻; osa³⁰⁸ P{neoFRT}82B/TM6B

13. w⁻; brm^T485 P{neoFRT}80B/TM6B

14. w⁻; trx^E2 P{neoFRT}82B/TM6B

15. w⁻; snr1^R3 P{neoFRT}82B/TM6B

Mutagenesis screen:

A schematic of the genetic screen used is provided in supplemental Figure 2. For chemical mutagenesis, isogenic males of the genotype w⁻; p[w⁺]/p[w⁺]; Df(3R)p13/TM3 [where the Df(3R)p13 chromosome removes the atonal locus] were treated with 25 mm EMS (Sigma) as described (Ashburner 1989). These males were then crossed to w⁻; ato¹⁰⁹⁰/ato¹⁰⁹⁰ virgin females at 25°, and F₁ males or virgin females were scored under Leica dissecting microscopes for enhancement of the base Df(3R)p13/ato¹⁰⁹⁰ eye phenotype at 25° (see Figure 2E). Df(3R)p13 corresponds to Bloomington Stock no. 1943. p[w⁺] refers to the transposable element insertion P{EP}EP2178 from the Szeged Drosophila Stock Centre, which itself does not modify the base Df(3R)p13/ato¹⁰⁹⁰ eye phenotype at 25°. Mutations that affected this P-insertion (resulting in white-eyed flies) were used as a measure of the efficiency of mutagenesis. F₁ males or virgin females of the genotype w⁻; p[w⁺], */+; Df(3R)p13, e, */ato¹⁰⁹⁰, where * indicates the presence of the modifying mutation, were then crossed to w⁻; ato¹⁰⁹⁰/TM6B, and those mutants that retained the observed enhancement in these F₁ progeny were used to create independent stocks. Those mutants that failed to reproduce the observed enhancement were discarded. A full description of the genetic mapping is described in the supplemental Results.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.—

The ato¹⁰⁹⁰ allele (referred to as ato^ts) is a temperature-sensitive mutation that affects Atonal protein expression. (A–F) Stereomicroscope pictures of adult compound eyes, anterior right, dorsal up, same magnification. Genotypes are listed in lower right of each panel. Temperature cultured is listed in upper right of each section. (A) Wild-type adult eye from fly cultured at 29° shows normal appearance. (B) Adult eye from heterozygous ato^ts/+ mutant fly cultured at 29° shows wild-type appearance. (C) Adult eye from heterozygous Df(3R)p13/+ [referred to as Df(ato)] mutant fly cultured at 29° shows wild-type appearance. (D) Adult eye from transheterozygous ato^ts/Df(ato) mutant fly cultured at 29° shows virtually no eye. (E) Adult eye from transheterozygous ato^ts/Df(ato) mutant fly cultured at 25° shows a small, rough eye. (F) Adult eye from transheterozygous ato^ts/ato¹ mutant fly cultured at 25° shows a small, rough eye similar to that of ato^ts/Df(ato) flies cultured at 25°. (G–I) Third instar eye imaginal discs showing Atonal protein expression, anterior right. (G) Wild-type eye disc cultured at 29° shows normal Atonal expression pattern. Arrows indicate Atonal protein expression in single R8 nuclei. (H) Transheterozygous ato^ts/Df(ato) mutant eye disc from fly cultured at 25° shows decreased Atonal protein expression, particularly posterior to the furrow in R8 nuclei (arrowhead). (I) Transheterozygous ato^ts/Df(ato) mutant eye disc from fly cultured at 29° shows strongly decreased Atonal protein expression, both posterior to the furrow in R8 nuclei (arrowhead), and anterior to the furrow in the broad Atonal stripe (arrow). (J–L) Stereomicroscope pictures of adult compound eyes, anterior right, dorsal up, same magnification. All flies were cultured at 25°. (J) hh^AC/+; ato^ts/Df(ato) mutant eye. The eye appears smaller than ato^ts/Df(ato) eyes alone at this temperature, indicating dominant enhancement by hh^AC. (K) Egfr^E1/+ adult eyes show a slightly rough and small eye alone. (L) Egfr^E1/+; ato^ts/Df(ato) mutant eye. The eye appears smaller than ato^ts/Df(ato) eyes alone at this temperature, indicating dominant enhancement by Egfr^E1.

Atonal antibody preparation:

The ato reading frame fused to the His₆ tag and inserted into the pRSET vector (a gift from Andrew Jarman, see Jarman et al. 1995) was used to express and isolate Atonal protein. The protein was purified and used to immunize two guinea pigs (Covance) and two rabbits (Zymed).

For Roe antibody production, genomic antibodies were produced from Strategic Diagnostics (http://www.sdix.com/) according to the manufacturer's protocol. The following amino acid sequence was used from the Roe N-terminal specific sequence: VSSSSGSSSTAGPAPTGSTRGRKSRIYPNPNQHIIVTSNSVDNGGIRMQNATLNERNSQNSSAGAAGVGNVTTAAGSGNGISSSTSSPHHMTQLDVKDTK.

Immunohistochemistry and microscopy:

Eye disc preparations were as described (Tio and Moses 1997) mounted in Vectashield (Vector Laboratories, H-1000), and imaged by confocal microscopy using a Nikon Eclipse TE2000-U laser-scanning confocal microscope. Primary antibodies: guinea pig anti-Atonal (1:1000, described in this study), rabbit anti-Roughened Eye (1:50, described in this study), rat anti-ELAV (1:500, 7E8A10 from Developmental Studies Hybridoma Bank), rabbit anti-Kismet-L (1:100, gift of J. Tamkun; Srinivasan et al. 2005), mouse anti-Daughterless (1:50, gift of C. Cronmiller; Brown et al. 1996), and rabbit anti-β-galactosidase (1:1000, Cortex Biochem CA2190). Secondary antibodies were from Jackson ImmunoResearch: goat anti-mouse Cy5 (1:500, 115-175-003) and goat anti-rabbit TRITC (1:250, 111-025-003).

Adult eyes were immersed in 100% ethanol, and digital photographs were taken under a Leica mZ 12.5 stereomicroscope, using an attached Leica digital camera.

Characterization of a temperature-sensitive loss-of-function atonal genetic background:

The ato¹⁰⁹⁰ mutation (henceforth referred to as ato^ts, a generous gift from V. Rodrigues) was initially isolated in an EMS screen designed to identify second-site modifiers of a lozenge phenotype and was subsequently tested for complementation with the Df(3R)p13 deficiency [henceforth referred to as Df(ato)], which removes the atonal locus. Heterozygous ato^ts/+ flies show no dominant effect on adult eye morphology at either 18° or 29° (Figure 2B). Similarly, eyes from heterozygous flies containing Df(ato) also show no dominant effect on adult eye morphology at 29° (Figure 2C). However, when ato^ts is crossed in trans to Df(ato), adult eye morphology is severely disrupted at 29° (Figure 2D), showing loss of nearly all photoreceptors and a strongly reduced eye size. When these flies are raised at either 25° or 18°, eye size and morphology are strongly rescued (Figures 2E and data not shown), although these eyes still display a rough eye phenotype. Df(ato) has been previously used to examine atonal-specific loss-of-function effects in transheterozygous mutant combinations (Jarman et al. 1994, 1995; Baker et al. 1996). However, Df(ato) deletes genes other than atonal. To verify that the phenotypic effects observed from our ato^ts allele were due to a decrease specifically in atonal gene function, we analyzed transheterozygous combinations of ato^ts with two other atonal mutants (ato¹ and ato³). Both of these alleles are strong loss-of-function mutations in the atonal gene (Jarman et al. 1994, 1995). In each case, transheterozygous mutant combinations of ato^ts/ato¹ and ato^ts/ato³ produced small rough eyes at 25° very similar to the phenotype observed in the ato^ts/Df(ato) genotype at 25° (Figure 2F and data not shown). Thus, these data suggest that the small rough eye phenotype of the ato^ts/Df(ato) genotype is due to loss of atonal gene function within the Df(ato) stock and not due to effects from removal of other loci within this deficiency.

We next analyzed Atonal protein expression in the ato^ts/Df(ato) mutant background at two different temperatures (25° and 29°) and found that the precise expression and refinement of Atonal is disrupted. At 25° there is a defect in the refinement of Atonal protein into R8 cells (compare arrows in Figure 2G to arrowhead in Figure 2H). This is consistent with what has been previously reported for Atonal protein and atonal mRNA in the ato¹/Df(ato) loss-of-function atonal genetic background (Jarman et al. 1995). At 29° this effect is more dramatic, with a reduction in Atonal expression both ahead of the furrow (arrow in Figure 2I) and no expression within R8 cells posterior to the furrow (arrowhead in Figure 2I).

Although they are reduced and morphologically abnormal, adult eyes still form in the ato^ts/Df(ato) mutant backgrounds, even though the only functional copy of the atonal gene comes from the ato^ts mutant. Taken together, these data suggested that (1) the ato^ts allele is a temperature-sensitive, loss-of-function allele that still produces functional Atonal protein at both 25° and 29°; (2) the ato^ts/Df(ato) atonal eye phenotype is due to decreased (although not absent) Atonal protein expression, where R8 refinement within the developing retina is strongly impaired; and (3) the rough eye phenotype of transheterozygous ato^ts/Df(ato) flies is not saturated at 25° and could potentially be enhanced at this temperature by second-site dominant modifier mutations (at least to the severity of the phenotype at 29°).

To examine whether the ato^ts/Df(ato) rough eye phenotype at 25° (hereafter referred to as the ato phenotype) could be dominantly modified as described above, we first examined whether loss-of-function mutations in genes known to be involved in ato gene regulation could dominantly enhance this phenotype. The Hedgehog (hh) signal transduction pathway is involved in numerous cellular processes (Ingham and McMahon 2001; Stark 2002; Lum and Beachy 2004), including resolution of Atonal protein expression in the developing fly eye (Borod and Heberlein 1998; Dominguez 1999; White and Jarman 2000). We analyzed the effect of loss of a single copy of the hh gene on the ato phenotype at 25°. We found that a null mutation in the hh gene (hh^AC, which alone has no heterozygous dominant adult eye phenotype) shows a dominant genetic enhancement of the ato phenotype (Figure 2J), indicating that this phenotype is indeed sensitive to loss-of-function mutations in cellular components that regulate Atonal protein expression.

We then tested whether the ato phenotype is sensitive to gain-of-function mutations that regulate Atonal protein expression as well. Increased nuclear translocation of MAP kinase, a downstream component of the Egfr pathway, has been shown to disrupt Atonal protein expression and spacing (Kumar et al. 2003; Vrailas et al. 2006), and increased Egfr expression directed within the morphogenetic furrow also shows decreased Atonal expression (Chen and Chien 1999). Further, a gain-of-function mutation in Egfr (Egfr ^E1) results in decreased Atonal expression within the developing retina (Jarman et al. 1995). Egfr ^E1, which alone has slight rough eye phenotype (Figure 2K), also shows a dominant genetic enhancement of the ato phenotype (Figure 2L). These results suggest that the ato phenotype is also sensitive to dominant enhancement by gain-of-function mutants as well.

A genetic screen for enhancers of the ato phenotype:

Using the ato phenotype at 25°, we performed a chemical (EMS) mutagenesis screen, examining a total of 22,151 flies for dominant enhancement of the phenotype. We recovered a total of 48 dominant enhancers, 15 of which separated into five total complementation groups (Table 1). Further, of the 48 enhancers we identified, only 1 of these enhancers was a mutation in the Star gene, which itself exhibits a dominant rough eye phenotype. This suggests that the multiallele complementation groups we identified in our screen most likely reflect genes directly involved in regulating atonal gene function and/or expression and are not additive or indirect effects from mutations in general modifiers of eye morphology.

The enhancers we identified in our genetic screen generally fell within one of three groups with regard to atonal expression: (1) those that differentially regulated atonal expression anterior and posterior to the morphogenetic furrow, (2) those that similarly regulated atonal expression anterior and posterior to the morphogenetic furrow, and (3) those that only regulated atonal expression posterior to the furrow.

Differential regulation of atonal expression—hedgehog (hh) and daughterless (da):

We mapped three of the third chromosome dominant enhancers of the ato phenotype to the hh locus (Table 1). hh is a segment polarity gene (Nüsslein-Volhard and Wieschaus 1980) that encodes a secreted ligand required for Hedgehog pathway signaling and is involved in multiple developmental processes (Ingham and McMahon 2001; Lum and Beachy 2004), including the progression of the morphogenetic furrow (Heberlein et al. 1993; Ma et al. 1993) and the regulation of atonal gene expression in the developing fly retina (Dominguez 1999). Previous studies have shown that loss of hh signaling has opposite effects on atonal expression in the developing retina (Dominguez 1999).

We mapped two of the second chromosome dominant enhancers of the ato phenotype to the da locus (Figure 3, J–L; Table 1). Both alleles of da moderately enhance the ato phenotype (Figure 3, J and K), and both alleles failed to complement two independent loss-of-function da mutations (da^UX, da¹, Table 1). We therefore tested for dominant enhancement of the ato phenotype with these independent loss-of-function da mutants and found that they also dominantly enhanced the ato phenotype (e.g., Figure 3L). We conclude that the two alleles we identified in our screen are also loss-of-function da alleles.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.—

Genetic enhancement of the atonal loss-of-function eye phenotype. (A–L) Stereomicroscope pictures of adult compound eyes, anterior right, dorsal up, same magnification. All panels show ato^ts/Df(ato) mutant eyes raised at 25°. Enhancer mutations listed in lower right-hand panel. (A) ato^ts/Df(ato) mutant eye shows a small, rough appearance. (B and C) Loss-of-function mutations in lilliputian (lilli). (B) lilli^GD17/+; ato^ts/Df(ato) mutant eye. (C) lilli^S35/+; ato^ts/Df(ato) mutant eye. Both lilli mutants dominantly enhance the small, rough eye phenotype. (D–F) Loss-of-function mutations in kismet (kis). (D) kis^LM27/+; ato^ts/Df(ato) mutant eye. (E) kis¹/+; ato^ts/Df(ato) mutant eye. (F) kis^k13416/+; ato^ts/Df(ato) mutant eye. All three kis mutations enhance the small, rough eye to varying degrees. (G–I) Loss-of-function mutations in roughened eye (roe). (G) ato^ts/Df(ato), roe^KM29 mutant eye. (H) ato^ts/Df(ato), roe^SM8 mutant eye. (I) ato^ts/Df(ato), roe^BM10 mutant eye. All three roe mutants strongly enhance the small, rough eye phenotype. (J–L) Loss-of-function mutations in daughterless (da). (J) da^IB21/+; ato^ts/Df(ato) mutant eye. (K) da^AB12/+; ato^ts/Df(ato) mutant eye. (L) da^UX/+; ato^ts/Df(ato) mutant eye. All three da mutations moderately enhance the small, rough eye phenotype.

daughterless (da) encodes the only type I basic helix-loop-helix (bHLH) transcription factor in flies (Smith and Cronmiller 2001). Type II bHLH proteins generally exhibit a more restricted expression pattern, while the type I bHLH family of proteins (such as Daughterless) are more widely expressed throughout development (Massari and Murre 2000). Type II bHLH proteins bind to type I bHLH transcription factors to form functional heterodimers to regulate target gene expression (Jarman et al. 1993; Kophengnavong et al. 2000; Massari and Murre 2000).

Mutations in da were previously shown to negatively regulate Atonal protein expression in the developing Drosophila retina (Brown et al. 1996), but were also suggested to abolish R8 expression as well (Brown et al. 1996). To determine the effects of da mutation on ato transcriptional regulation from the two different ato enhancer elements, we created loss-of-function homozygous somatic da mutant clones (Xu and Rubin 1993) and analyzed ato-lacZ reporter expression within these clones. We used a da null allele, da^UX, and the eyeless:Flip specific recombinase to generate clones in the developing retina. In these clones, we observed expanded Atonal protein expression posterior to the morphogenetic furrow (arrows in Figure 4, A and B). We next analyzed atonal transcriptional regulation from the 3′ genomic enhancer element within da^UX mutant clones and found that the expression of this reporter is strongly increased and expanded in these clones posterior to the morphogenetic furrow (Figure 4, C and D). Analysis of atonal transcriptional regulation from the 5′ genomic enhancer element showed the opposite effect, with nearly a complete loss of expression of atonal transcription within the intermediate groups and single R8 nuclei (Figure 4, E and F). Taken together, these results suggest that da gene function is differentially required for atonal transcriptional regulation, negatively regulating atonal transcription at the 3′ atonal genomic enhancer, while simultaneously positively regulating atonal transcription at the 5′ atonal genomic enhancer.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.—

daughterless functions differentially at both atonal enhancer elements. (A–F) Third instar larval retinas, anterior right. All panels show da^UX homozygous mutant clones marked by the absence of GFP (green). (A) Atonal protein expression (red) in da mutant clone. Arrow denotes the expansion of Atonal protein toward posterior within clones. (B) Atonal protein expression (white) from A. (C) β-galactosidase protein expression (red) as driven by the 3′ atonal genomic enhancer element. Note the expansion of β-galactosidase expression in the posterior. (D) β-galactosidase protein (white) from C. (E) β-galactosidase protein expression (red) as driven by the 5′ atonal genomic enhancer element. Note the loss of β-galactosidase expression. (F) β-galactosidase protein (white) from E.

Similar regulation of atonal expression—lilliputian (lilli):

We mapped two of the second chromosome dominant enhancers of the ato phenotype to the lilli locus (Figure 3, B and C; Table 1). Both alleles show moderate-to-strong enhancement of the ato phenotype (e.g., Figure 3B), and both alleles failed to complement Bloomington Stock Collection Deficiency Df(2L)JS17, which deletes the lilliputian gene. Both alleles also failed to complement four independent loss-of-function lilli mutations (lilli^XS407, lilli^A17-2, lilli^XS575, and lilli^S35) (Neufeld et al. 1998b; Tang et al. 2001). We therefore tested for dominant enhancement of the ato phenotype with these independent loss-of-function lilli mutants and found that all four alleles also dominantly enhanced the ato phenotype, although not as severely as the two we recovered in our screen (e.g., Figure 3C). We therefore concluded that these two mutants were alleles of lilliputian (lilli^GD17 and lilli^AG5), and that both behave genetically as loss-of-function mutants. Further, we conclude that it is loss-of-function at the lilli locus, and not at a potential second site, that is associated with the enhancement of the ato phenotype in these experiments.

We analyzed atonal transcriptional regulation from both the 3′ and 5′ atonal genomic enhancer elements and found that transcription from both of these elements is reduced in homozygous lilli mutant clones (data not shown), suggesting that lilli gene function is positively required for ato transcriptional regulation within and posterior to the morphogenetic furrow at both the 3′ and 5′ atonal genomic enhancers. Characterization of the significance of this interaction will be described elsewhere (G. DiStefano and D. Marenda, unpublished results).

kismet (kis):

We mapped four of the second chromosome dominant enhancers of the ato phenotype to the kis locus (Figure 3, D–F; Table 1). All four alleles show a mild enhancement of the ato phenotype (e.g., Figure 3D), and all four alleles failed to complement Exelixis Deficiency Df(2L)ED19, which deletes the kismet gene. All four alleles also failed to complement two independent loss-of-function kis mutations (kis¹ and kis^k13416) (Kennison and Tamkun 1988; Roch et al. 1998). We tested for dominant enhancement of the ato phenotype with these independent loss-of-function kis mutants, and found that these two alleles also dominantly enhanced the ato phenotype (Figure 3, E and F). We therefore concluded that these four mutants were alleles of kismet (kis^LM27, kis^EC1, kis^AS21, and kis^AO9), and all behave genetically as loss-of-function mutants. Further, we conclude that it is loss-of-function at the kis locus, and not at a potential second site, that is associated with the enhancement of the ato phenotype in these experiments.

kismet was initially identified in a screen for suppressors of a dominant homeotic phenotype associated with Polycomb (Kennison and Tamkun 1988) and has critical functions in embryonic development and segmentation (Daubresse et al. 1999; Srinivasan et al. 2005). kis encodes multiple nuclear proteins that are related to chromatin remodeling factors and are largely associated with transcriptionally active chromatin (Srinivasan et al. 2005). In the Drosophila eye, mutations in kismet were recovered as enhancers of a gain-of-function kinase suppressor of Ras (ksr) rough eye phenotype (Therrien et al. 2000).

To better understand how kismet function is required for atonal expression in the developing retina, we first analyzed Kismet protein expression in developing third larval instar retinas. Using an antibody that is specific to the long form of the Kismet protein, Kismet-L, we detected ubiquitous Kismet expression in developing third instar retinas (Figure 5A). Kismet is expressed anterior to the morphogenetic furrow (Figure 5A), within the furrow (arrowhead in Figure 5B) and within developing photoreceptor cells posterior to the furrow (arrow in Figure 5B). Kismet is also expressed in later developing structures, including cone cells (arrow in Figure 5C). Thus, Kismet expression matches that of a protein that may be generally required for multiple events in Drosophila eye development and not specific to atonal expression within or posterior to the morphogenetic furrow.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.—

Kismet expression and function in the developing retina. (A–C) Wild-type retinas. (D–O) kis^LM27 homozygous mutant clones marked by the absence of GFP (green). All panels show third instar larval eye discs, anterior right. (A–C) Kismet protein expression in the developing eye. (A) Kismet protein appears nuclear and is expressed throughout the retina. Arrowhead marks the morphogenetic furrow. (B) Kismet protein expression in nuclei ahead of the furrow, within the morphogenetic furrow (arrowhead), and within nuclei posterior to the furrow (arrow). (C) Kismet expression within cone cells (arrow). (D–F) Kismet protein expression within loss-of-function kis^LM27 mutant clones. (D) Kismet protein expression is in red. Loss of GFP expression (green) marks the cells mutant for kis. Note loss of Kismet protein within kis^LM27 mutant clones. (E) GFP expression (white) from D. (F) Kismet protein expression (white) from D. (G–I) Atonal protein expression within loss-of-function kis^LM27 mutant clones. (G) Atonal protein (red) is lost in kis mutant clones (arrow in clone) predominantly in the broad stripe of expression anterior to the furrow (compare arrow in clone to arrow above in normal tissue). (H) GFP expression (white) from G. (I) Atonal expression (white) from G. Note the presence of R8 nuclei within kis mutant clones (arrow). (J–L) β-galactosidase protein expression as driven by the 3′ atonal genomic enhancer element within loss-of-function kis^LM27 mutant clones. (J) β-galactosidase protein expression (red) is reduced but not absent in kis mutant clones. (K) GFP expression (white) from J. (L) β-galactosidase protein expression (white) from J. (M–O) β-galactosidase protein expression as driven by the 5′ atonal genomic enhancer element within loss-of-function kis^LM27 mutant clones. (M) β-galactosidase protein expression (red) is reduced but not absent in kis mutant clones. (N) GFP expression (white) from M. (O) β-galactosidase protein expression (white) from M.

To further characterize which phase in eye development kis gene function is required, we created loss-of-function homozygous somatic kis mutant clones. We first analyzed Kismet protein expression within these clones. In kis^LM27 homozygous mutant clones, Kismet protein is absent from the mutant tissue (Figure 5, D–F), indicating that this particular allele is a protein null. This was also true for kis^EC1 homozygous mutant clones (data not shown). We analyzed Atonal protein expression within kis^LM27 homozygous mutant clones and found that Atonal expression is markedly reduced (Figure 5, G–I). Atonal expression was most severely reduced in the broad stripe of Atonal expression anterior to the morphogenetic furrow as well as within the intermediate groups, (arrows in Figure 5G). Some Ato expression within single R8s still remained (arrow in Figure 5I). Similar results were obtained with kis^EC1 homozygous mutant clones (data not shown). These data suggest that kis gene function is required for normal Atonal protein expression.

The Kismet protein is a transcription factor, thought to play a role in altering gene expression through changes in chromatin structure (Srinivasan et al. 2005). Therefore, we next looked at atonal transcriptional regulation at both the 3′ and 5′ regulatory elements within kis mutant tissue. In kis^LM27 homozygous mutant clones, we found that the amount of β-galactosidase protein expressed from the 3′ ato-lacZ element was strongly reduced (Figure 5, J–L). Similarly, we found that the amount of β-galactosidase protein expressed from the 5′ ato-lacZ element was also strongly reduced (Figure 5, M–O). Taken together, these data strongly suggest that kis gene function is normally required for atonal transcriptional activation and/or maintenance at both the 3′ and 5′ atonal gene enhancer elements.

On the basis of Kismet protein expression within the developing eye, we sought to further analyze the role of kis gene function in other phases of eye development, both anterior and posterior to the morphogenetic furrow. Unlike homozygous somatic mutant clones in genes required for cell division and cell growth, homozygous mutant kis clones are not small and scarce within the developing retinas (Figure 5D), suggesting that kis gene function is not required anterior to the morphogenetic furrow for cell division and/or growth functions. To further analyze gene expression anterior to the furrow, we examined the expression of the Hairy protein. Anterior to the furrow, the expression of hairy (h) marks the “preproneural domain” (Greenwood and Struhl 1999), which is expressed just prior to atonal expression in the developing retina. Within homozygous mutant kis clones, h expression was normal (Figure 6, A–C), suggesting that kis function is also not required for patterning events immediately ahead of the morphogenetic furrow. kis loss-of-function clones were previously reported to have no effect on photoreceptor differentiation posterior to the morphogenetic furrow (Janody et al. 2004). Consistent with these data, we also observe no significant effect on ELAV expression within kis mutant clones located posterior to the morphogenetic furrow (data not shown), suggesting that kis gene function is also not required for photoreceptor differentiation posterior to the morphogenetic furrow.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.—

Kismet function in the morphogenetic furrow. (A–F) and (H and I) show kis^LM27 homozygous mutant clones marked by the absence of GFP (green). (G) Wild-type disc. All panels show third instar larval eye discs, anterior right. (A–C) Hairy protein expression within loss-of-function kis^LM27 mutant clones. (A) Hairy expression (red) is not altered in kis mutant clones. (B) GFP expression (white) from A. (C) Hairy expression (white) from A. (D–F) Daughterless protein expression within loss-of-function kis^LM27 mutant clones. (D) Daughterless expression (red) is reduced but not absent in kis mutant clones. (E) GFP expression (white) from D. (F) Daughterless expression (white) from D. (G) Wild-type Scabrous expression (white) in normal disc. (H) Scabrous expression (Red) in kis mutant clone. Scabrous expression is decreased but not absent. (I) Scabrous expression (white) from H.

To further determine if kis function is restricted to events occurring within and near the morphogenetic furrow, we analyzed the expression of other proteins expressed within and near the furrow, including the expression of Daughterless and Scabrous. Daughterless protein was previously shown to be expressed throughout the eye disc, with stronger expression within the morphogenetic furrow (Brown et al. 1996). In homozygous mutant kis clones, we observed that Daughterless expression was reduced, but not absent (Figures 6, D–F). Scabrous expression is a target of Atonal and is also expressed within and near the intermediate groups in the furrow (Baker et al. 1990; Mlodzik et al. 1990; Frankfort and Mardon 2002). In homozygous mutant kis clones, we observed that like Daughterless expression, Scabrous expression was also reduced, but not absent (Figure 6, G–I). Taken together, these data suggest that kismet gene function is restricted to regulating the expression of factors within and near the morphogenetic furrow of the developing larval retina.

The Kismet protein belongs to the trithorax group of transcriptional regulators (Daubresse et al. 1999), which include trithorax, kohtalo, skuld, and members of the Brahma complex (brm, osa, snr1, and moira) (Tamkun et al. 1992; Dingwall et al. 1995; Elfring et al. 1998; Collins et al. 1999). Mutations in brm, snr1, and osa all affect eye development (Treisman et al. 1997; Brumby et al. 2002; Marenda et al. 2003), and mutations in the trithorax genes trithorax, brahma, kohtalo, and skuld were recently identified in a mosaic genetic screen used to identify genes required for retinal patterning (Janody et al. 2004). Our identification of kis in our genetic screen led us to hypothesize that other trithorax group genes might also be involved in regulating Atonal expression. To test this possibility, we created clones in the developing retina for loss-of-function mutations in brahma (brm^T485), osa (osa³⁰⁸), snr1 (snr1^R3), and trithorax (trx^E2).

brm mutant clones were very small, consistent with what has been previously reported (Janody et al. 2004). Analysis of multiple small brm mutant clones showed a complex regulation on Atonal protein expression (Figure 7, A–F). Within brm mutant clones in and near the morphogenetic furrow, Atonal protein expression was unaffected in both the broad stripe of expression and in single R8 nuclei as well (arrows in Figure 7, A–C). However, within ∼10% of brm mutant clones posterior to the morphogenetic furrow, Atonal protein expression in single R8 nuclei was inappropriately expressed (arrows in Figure 7, D–F). For both osa and snr1 mutant clones, Atonal protein expression was unaffected (arrowheads in Figure 7, G–L), suggesting that although these genes are required for retinal patterning, they are not significantly affecting Atonal expression within the morphogenetic furrow. We did observe loss of Atonal protein expression in clones of cells mutant for trx (Figure 7, M–O). Within trx mutant clones, Atonal expression is absent at both the broad stripe as well as within single R8 nuclei. Taken together, our results suggest that of the trithorax group proteins we tested, only Kismet and Trithorax are required for Atonal expression in the morphogenetic furrow.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.—

Atonal expression in trithorax group loss-of-function mutant clones. (A–O) Third instar eye discs showing Atonal protein expression in different loss-of-function backgrounds. Mutant clones are marked by loss of GFP expression (green) in all panels. (A–F) Atonal protein expression (red) in brm^T485 mutant clones. (A) Atonal expression is still present in brm mutant clones within the broad stripe of Atonal expression. Arrows denote Atonal expression in clone. (B) Atonal expression is still present in single R8 nuclei in brm mutant clones. Arrows denote Atonal expression in brm clones. (C) Atonal expression (white) from B. (D) Atonal expression is not reduced in R8 nuclei (arrow) posterior to the furrow. (E) GFP (white) expression from D. (F) Atonal expression (white) from D. Arrows denote two ectopic Atonal positive nuclei posterior to the morphogenetic furrow. (G–I) Atonal protein expression (red) in osa³⁰⁸ mutant clones. Atonal expression is still present within the clones both ahead of the furrow (arrow) and within R8 nuclei (arrowheads). (H) GFP (white) expression from G. (I) Atonal expression (white) from G. (J–L) Atonal protein expression (red) in snr1^R3 mutant clones. Atonal expression is still present within the clones both ahead of the furrow and within R8 nuclei (arrowheads). (K) GFP (white) expression from J. (L) Atonal expression (white) from J. (M–O) Atonal protein expression (red) in trx^E2 mutant clones. Atonal expression is lost within these clones (arrows in M and O). (N) GFP (white) expression from M. (O) Atonal expression (white) from M.

Regulation of Atonal expression at a single enhancer—roughened eye (roe):

We mapped four of the third chromosome dominant enhancers of the ato phenotype to the roe locus (Figure 3, G–I; Table 1). Mutations in roe are strong enhancers of the ato phenotype (Figure 3, G–I). Each of these alleles was on a chromosome that also contained the Df(ato) deletion, and these flies were crossed to a small deletion (rn¹⁶, St Pierre et al. 2002) that removes the C terminus of the rotund gene and the entire roughened eye gene. These adults produced a small eye phenotype that is stronger than the ato eye phenotype used in our screen (compare Figure 3A and G–I to Figure 8, C and D). Heteroallelic mutant combinations of the roughened eye (roe) gene produce viable adults with small rough eyes (Figure 8, A and B) (St Pierre et al. 2002). We therefore concluded that the four alleles that made up this complementation group were alleles of the roughened eye gene (roe^SM8, roe^BM10, roe^KM29, roe^KK16; Table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.—

roughened eye function in regulating Atonal expression in the eye. (A–D) Stereomicroscope pictures of adult eyes, anterior right, dorsal up. (A) roe¹/roe³ adult eyes show a mild small and rough eye phenotype. (B) roe³/rn¹⁶ adult eyes show a moderate small and rough eye phenotype. (C) roe^SM8, Df(ato)/rn¹⁶ and (D) roe^BM10, Df(ato)/rn¹⁶ both show an enhanced small and rough eye compared to roe³/rn¹⁶ adult eyes. (E–L) Third instar larval eye discs from wild-type (E–G) and roe³/rn¹⁶ (H–L) genotypes. (E and F) Colocalization of wild-type Atonal (green) and Roe (red) protein expression. (F) Arrowhead denotes Atonal expression anterior to Roe expression. Black arrows denote first column of R8 nuclei where Atonal and Roe protein are colocalized. Blue arrow denotes second column of R8 nuclei that are surrounded by Roe protein expression. Red arrow denotes third column of R8 nuclei. (G) Wild-type Atonal expression (white). (H) Atonal expression (white) in roe³/rn¹⁶ background. Note loss of Atonal expression in R8 nuclei. (I and J) Atonal expression (green) and β-galactosidase protein expression (red) as driven by the 3′ atonal genomic enhancer element in roe³/rn¹⁶ mutant retinas. (J) Magnified view of I. (K and L) Atonal expression (green) and β-galactosidase protein expression (red) as driven by the 5′ atonal genomic enhancer element in roe³/rn¹⁶ mutant retinas. (L) Magnified view of K.

To better understand how Roe protein function is required for atonal expression in the developing retina, we raised an antibody to the N-terminal portion of the Roe protein and analyzed Roe protein expression in developing retinas. roe mRNA was previously reported as being expressed within a band of 4–6 cells within the morphogenetic furrow (St Pierre et al. 2002). To more precisely define the area within the eye disc in which the Roe protein is expressed, we costained wild-type developing retinas for both Roe and Atonal protein expression (Figure 8, E and F). Consistent with previous reports, we also detected Roe expression within a band of 4–6 cells within the morphogenetic furrow (Figure 8E). Roe protein expression is predominantly nuclear. Interestingly, Roe expression first appears just behind (posterior to) the broad stripe of Atonal expression that is anterior to and within the morphogenetic furrow (arrowhead in Figure 8F). Roe protein first colocalizes both within the first column of nascent R8 nuclei immediately posterior to the broad stripe of Atonal expression (black arrow in Figure 8F) as well as within those nuclei surrounding this R8 cell. Roe protein then appears to be solely expressed within those nuclei immediately surrounding the second column of R8 cells (blue arrow in Figure 8F) and is no longer expressed near the third column of R8 cells (red arrow in Figure 8F).

On the basis of expression of the Roe protein, we predicted that Roe protein function with regard to atonal expression may be limited to the 5′ atonal regulatory element. As an initial test of this hypothesis, we analyzed Atonal protein expression in a strong roe loss-of-function background that had been previously used to analyze loss of roe function in the developing retina (roe³/rn¹⁶, Figure 8B) (St Pierre et al. 2002). This genetic background includes both a deletion of the rotund locus that removes function for both the rotund and roe genes (rn¹⁶) and also a point mutation that specifically removes roe gene function alone (roe³) (St Pierre et al. 2002). In this background, Atonal protein expression is reduced, but not absent. Atonal protein is still present in a broad stripe anterior to the furrow (Figure 8H), however, refinement of Atonal into single R8 nuclei is nearly absent (compare Figure 8, G and H), suggesting that mutation of the roe gene specifically affects atonal expression from the 5′ regulatory element. To verify this, we analyzed atonal transcriptional regulation at both the 3′ and 5′ regulatory elements within the roe loss-of-function background (Figure 8, I–L). ato transcription as assayed from the 3′ atonal regulatory element appears normal in roe loss-of-function eye discs (Figure 8, I and J). However, transcription from the 5′ atonal regulatory element is severely reduced (Figure 8, K and L), and ato transcription in single R8 cells is absent, consistent with what we observe with Atonal protein expression in this background. When taken together, these results suggest that the Roe protein is necessary for atonal transcriptional regulation specifically at the 5′ atonal regulatory element within the morphogenetic furrow, and that in roe mutants, single R8 cells are not specified.

Differential regulation of atonal transcription:

In our screen, we identified mutations in both the hedgehog and daughterless genes as enhancers of our loss-of-function atonal phenotype. Both of these mutants alter Atonal protein expression differentially anterior to the morphogenetic furrow as compared to posterior to the morphogenetic furrow. Regulation of atonal expression by the hedgehog pathway is well described (Dominguez 1999; Suzuki and Saigo 2000; White and Jarman 2000; Lim et al. 2008). Lim et al. (2008) describe recent evidence for Daughterless-mediated atonal expression in the developing retina that may further explain the results we have observed in our study.

We have shown that daughterless function is required to repress atonal transcription at the 3′ atonal enhancer. Importantly, this repression is only required for 3′ atonal transcription posterior to the morphogenetic furrow, as 3′ atonal enhancer expression is not upregulated anterior to the furrow in da mutant clones (Figure 4). At the 5′ atonal enhancer, daughterless function is required to promote atonal transcription. Atonal forms heterodimers with Daughterless (Jarman et al. 1993), and it is this heterodimerization that is required for bHLH protein function (Kophengnavong et al. 2000; Massari and Murre 2000). Since the 5′ atonal enhancer elements undergo autoregulation (Sun et al. 1998), this suggests that Daughterless protein function at the 5′ element may play an important role in this autoregulation. However, our data suggest that posterior to the morphogenetic furrow, while the Daughterless protein is required to form heterodimers with Atonal to positively regulate atonal transcription from the 5′ atonal enhancer, it is simultaneously required to transcriptionally repress atonal expression at the 3′ atonal enhancer element. Lim et al. (2008) have recently suggested that Daughterless forms homodimers and/or heterodimers with unknown bHLH proteins to repress atonal expression. Our data fit well with this model of atonal regulation and suggest that Daughterless homodimers may be required in cells surrounding developing R8 photoreceptors to repress atonal transcription from the 3′ atonal enhancer element.

Similar regulation of atonal transcription:

In our screen, we identified mutations in both the lilliputian and kismet genes as enhancers or our loss-of-function atonal phenotype. Mutations in both of these genes have a similar effect on atonal transcriptional regulation at the two atonal enhancer elements, as each is required to promote atonal transcription at both enhancers. Interestingly, mutations in both of the human homologs of these genes are involved in forms of mental retardation. Mutation in Fmr2/AF4 gene family members (the human homolog of lilliputian) are involved in both acute lymphoblastic leukemia and FRAXE nonsyndromic X-linked mental retardation syndrome (Gu et al. 1996; Gu and Nelson 2003). Mutations in Chd7 (the human homolog of kismet) are estimated to be causative for nearly two-thirds of all CHARGE syndrome diagnoses (Sanlaville and Verloes 2007), a rare, autosomal dominant disorder. Of further interest, patients with CHARGE syndrome also display eye and ear abnormalities, including coloboma, a symptom that is also attributed to mutations in the mammalian homolog of the hedgehog gene (Schimmenti et al. 2003; Fuerst et al. 2007). The human homolog of atonal (math5) is expressed and critically required for both proper retinal and auditory development (Brown et al. 1998, 2001; Wang et al. 2001; Yang et al. 2003; Le et al. 2006; Hufnagel et al. 2007; Saul et al. 2008) and thus may be a good candidate for a critical target gene that is disrupted in CHARGE syndrome.

Recent analysis of Kismet protein on salivary gland chromosomes suggests that Kis plays a global role in transcription by Pol II (Srinivasan et al. 2005). Consistent with the idea of global transcription, we observe Kis staining in all cells of the developing Drosophila retina, suggesting that Kis would also play a role in global transcription in the eye. However, our analysis of kismet mutant clones suggests that kis gene function is limited to events occurring within and near the morphogenetic furrow. We observe effects on Atonal, Daughterless, and Scabrous protein expression within the furrow, but see no effects on marker expression within kis mutant clones either posterior or anterior to the furrow. We have analyzed Kis protein expression within these clones and found that it is absent, suggesting that these mutants are effectively protein nulls within this tissue, and thus the normal expression of these markers is not likely due to residual kismet gene expression. kis mutants also show a limited number of phenotypes within Drosophila embryos (Daubresse et al. 1999). Taken together, these data suggest that (1) there may be redundant factors within the developing retina that can compensate for kis gene function within cells posterior and anterior to the furrow or (2) that Kismet protein function is only required in cells within and near the furrow, and does not have a function in areas outside these cells.

Members of the trithorax group of proteins appear to have a variety of functions within the developing Drosophila retina (Janody et al. 2004). Further, many of the trithorax genes required for specific cellular processes in the eye have a ubiquitous expression pattern in the eye (Kuzin et al. 1994; Treisman et al. 1997; Janody et al. 2003; Zraly et al. 2003), much like the Kismet protein. This expression data might suggest a dynamic functional and developmental regulation of these factors. In support of this, we have analyzed Kismet protein expression within loss-of-function genetic backgrounds of developmental signals known to regulate eye development and Atonal expression, including the Notch and Hedgehog pathways, as well as in daughterless and eyes absent mutants and find no difference in Kismet expression in any of these mutants (supplemental Figure 1). Thus, the rapid transitions in gene expression between different phases of retinal development (anterior and posterior to the furrow) must involve functional regulation of these ubiquitously expressed transcription factors and not the regulation of their gene expression within different phases.

While mutations in kis show no affect on photoreceptor differentiation posterior to the furrow, or on patterning events anterior to the furrow, mutations in other ubiquitously expressed trithorax group genes show differential effects in these areas. Mutations in trithorax group genes brahma, moira, skd, and kto all show loss of photoreceptor differentiation posterior to the furrow, while mutations in the trithorax group gene osa show less severe effects on photoreceptor differentiation in this area of the retina (Janody et al. 2004). Ahead of the furrow, brm and osa mutants have no effect on the expression of the anterior markers hth, ey, and tsh, while mutations in trx, skd, and kto all significantly affect these markers (Janody et al. 2004). Further, while we do not observe any effect on Atonal expression in the furrow within brm, osa, or snr1 mutant clones, we do see decreased Atonal expression in trx mutant clones. Immediately ahead of the furrow, within the preproneural domain, only trx, kto, and skd show effects on the markers hairy and eya, and only trx mutants showed an effect on dac expression (Janody et al. 2004). Similarly, recent studies have shown atonal expression is affected in clones mutant for skd and kto (Lim et al. 2007). Within skd and kto clones, single R8 neurons are selected even without fully resolved intermediate groups (Lim et al. 2007), an occurrence that also happens in kis mutant clones (Figure 5).

Taken together with our results, these data now begin to separate specific functions within the trithorax group of proteins into a developmental requirement for specific events during retinal development. Thus, ahead of the furrow, only trx, skd, and kto appear to be required for specific transcriptional regulation of the targets tested so far. Within the furrow, trx, skd, kto, and kis all show an effect on atonal expression, while brm, osa, and snr1 do not. Finally, posterior to the furrow, trithorax group members brm, moira, osa, skd, kto, and trx are all required for photoreceptor differentiation.

Transcriptional regulation in the developing retina requires rapid changes in gene expression as the morphogenetic furrow moves across the eye disc (Moses 1991, 2003; Kumar and Moses 1997). The movement of the morphogenetic furrow allows for a new column of photoreceptor cells to be specified roughly every 2 hr for uniform movement of the morphogenetic furrow (Campos-Ortega and Hofbauer 1977), and every 70 min for nonuniform movement (Basler and Hafen 1989). Thus, the developmental regulation of different target-specific transcription factors must be tightly controlled, as cell type specific transcription will rapidly change as the morphogenetic furrow moves across the retina. Regulated changes in chromatin structure must therefore also be tightly regulated, as the morphogenetic furrow moves across the retina. The different requirement for trithorax group proteins ahead of, within, and posterior to the morphogenetic furrow may reflect differential changes in chromatin structure necessary to accommodate rapid changes in transcription during retinal development. Thus, these different transcriptional events may be anticipated by cells in regions of the developing retina before the morphogenetic furrow has reached these cells.

Expression of atonal at only one enhancer element:

In our screen, we identified mutations in the roughed eye gene as an enhancer or our loss-of-function atonal phenotype. We show that Roe protein expression is restricted to the cells expressing the 5′ atonal regulatory element marker within morphogenetic furrow, specifically colocalized within and surrounding the first column of nascent R8 cells. These data suggest that roe gene function is required for the proper formation of R8 nuclei and is consistent with Roe protein expression colocalizing with Atonal protein expression in the R8 cells. However, Roe protein is also present in nuclei surrounding the nascent R8 cells. It has been previously shown that Delta protein expression is disrupted in roe mutant retinas (St Pierre et al. 2002), moving from a punctate expression pattern to a more diffuse pattern. Posterior to the furrow, Delta expression becomes restricted to the R8 cell, and this activates the Notch pathway in surrounding cells to repress atonal transcription (Lee et al. 1996; Li and Baker 2001; Doroquez and Rebay 2006). Thus, if the Roe protein surrounding the R8 cells functions to repress Delta expression within these cells, loss of Roe may lead to increased Delta expression within these cells and thus cause the Notch pathway to become activated in the R8 cell itself, repressing atonal transcription.

In the developing retina, the restricted expression of cell-specific factors (such as roughened eye) is vital to the proper development of a specific subset of retinal precursor cells. However, the broader transcriptional regulation of these cell-specific transcription factors by ubiquitous transcription factors (such as kismet) remains less clear. Our results presented here suggest that although these factors are expressed ubiquitously, many of them have very specific functions in distinct phases of retinal development. These results provide additional insights into the complex regulation of atonal expression in the developing retina.