SEARCH

SEARCH BY CITATION

Keywords:

  • Drosophila melanogaster;
  • CtBP;
  • Eyeless;
  • Distal antenna;
  • Distal antenna related;
  • Dachshund;
  • eye;
  • antenna

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Specification factors regulate cell fate in part by interacting with transcriptional co-regulators like CtBP to regulate gene expression. Here, we demonstrate that CtBP forms a complex or complexes with the Drosophila melanogaster Pax6 homolog Eyeless (Ey), and with Distal antenna (Dan), Distal antenna related (Danr), and Dachshund to promote eye and antennal specification. Phenotypic analysis together with molecular data indicate that CtBP interacts with Ey to prevent overproliferation of eye precursors. In contrast, CtBP,dan,danr triple mutant adult eyes have significantly fewer ommatidia than CtBP single or dan,danr double mutants, suggesting that the CtBP/Dan/Danr complex functions to recruit ommatidia from the eye precursor pool. Furthermore, CtBP single and to a greater extent CtBP,dan,danr triple mutants affect the establishment and maintenance of the R8 precursor, which is the founding ommatidial cell. Thus, CtBP interacts with different eye specification factors to regulate gene expression appropriate for proliferative vs. differentiative stages of eye development. Developmental Dynamics 239:2367–2385, 2010. © 2010 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Selector genes are required for the specification of cells, tissues, and organs. More strikingly, misexpression of selector genes can convert one cell type into another. Because all selector genes encode transcription factors, they are presumed to lie at (or near) the top of a hierarchy that regulates transcription of tissue-specific genes (Mann,2000).

One classic example of a selector gene is the eyeless (ey) gene of Drosophila melanogaster. In the absence of ey function, the eye fails to develop (Quiring et al.,1994). Conversely, misexpression of ey in cells fated to form antenna, wing or leg can convert them to an eye fate. Remarkably, the ectopic eye that forms as a result contains all of the cell types of an endogenous eye, and these cells develop specialized organelles and assort in the same patterns as in the endogenous eye (Halder et al.,1995). The ey gene encodes a Drosophila Pax6 (Paired Homeodomain-6) homolog, and is thus part of class of genes thought to have roles in eye specification throughout Metazoa (Gehring and Ikeo,1999; Gehring,2005).

Other than Ey, a small network of transcription factors that includes Twin of Eyeless (Toy), Eyes absent (Eya), Sine oculis (So), and Dachshund (Dac) cooperate to specify eye fate. All of these genes are required for eye development, and all can trigger ectopic eye formation, albeit to greater or lesser degrees. In terms of expression initiation, the factors in the eye regulatory network form a (mostly) linear hierarchy. For example, Ey is required for Eya expression, but Eya is not required for Ey expression (Halder et al.,1998). That they form an interconnected network once expressed is indicated by the fact that they can regulate each other's expression through feedback loops (e.g., Eya can activate Ey expression ectopically), and that they are capable of direct physical interaction (reviewed in Silver and Rebay,2005; Kumar,2009).

The probable transcription factors Distal antenna (Dan) and Distal antenna related (Danr) have recently been shown to participate in the eye regulatory network (Curtiss et al.,2007). As with other network members, Dan and Danr are required for normal eye development and are capable of converting antennal precursors to an eye fate. With respect to initiation of expression, Dan and Danr lie at the bottom of the eye specification factor hierarchy, and thus appear to act at or near the transition from specification to differentiation. However, Dan and Danr participate in expression feedback loops characteristic of the eye regulatory network. For instance, misexpression of either results in ectopic Ey expression in the antenna. Furthermore, both are capable of direct physical interactions with Ey and Dac.

Despite decades of work, very little is known about the mechanisms by which the eye specification transcription factor network regulates gene expression during eye development, or of the exact functions of the individual eye specification factors during the different stages of Drosophila eye development. During early larval stages, the cells of the eye-antennal disc proliferate asynchronously. Approximately halfway through the larval period, a structure called the morphogenetic furrow forms, marking the beginning of eye differentiation. The furrow propagates across the eye disc from posterior to anterior, leaving differentiating ommatidia in its wake. Cells ahead of the furrow continue to proliferate asynchronously, and arrest in G1 slightly anterior to the furrow (reviewed in Wolff and Ready,1993).

Ey is known to be expressed at early larval stages and later ahead of the furrow in proliferating eye precursors. Recent work suggests that Ey promotes proliferation of eye precursors (Bessa et al.,2002; Peng et al.,2009; Lopes and Casares,2010). In addition, Ey is directly involved in activating expression of so and the proneural gene atonal (ato) in the preproneural zone immediately ahead of the furrow (Niimi et al.,1999; Zhang et al.,2006). However, Ey is subsequently down-regulated at the furrow, and it is Eya, So, Dac, Dan, and Danr that participate in the transition of cells from proliferation to differentiation (Pignoni et al.,1997; Curtiss et al.,2007). The genes that are directly regulated by the eye specification factors and that participate directly in these developmental events are largely unknown, as are the mechanisms by which their transcription is regulated.

Besides the eye, Dan and Danr also participate in specification of another organ, the distal antenna. In loss-of-function dan or danr mutants the distal part of the antenna is transformed into a leg. Conversely, expression of Dan or Danr in distal leg precursors converts them to distal antennal structures. These phenotypes are similar to the phenotypes of mutants for the antennal selector genes, Distal-less and Homothorax (Casares and Mann,1998; Dong et al.,2000), as well as genes functioning downstream of these two factors (Duncan et al.,1998; Johnston et al.,1998; Dong et al.,2002). Evidence suggests that Dan and Danr lie at the bottom of the antennal specification hierarchy (Emerald et al.,2003; Suzanne et al.,2003), as they do in eye specification, again suggesting that they act at or near the transition from specification and differentiation.

The Dan and Danr protein sequences contain motifs that suggest a link to epigenetic mechanisms of regulating gene expression. Dan and Danr are closely related proteins that contain only two recognizable motifs, a Pipsqueak domain and a PxDLS sequence. Pipsqueak domains have a helix-turn-helix structure that is similar to a homeodomain, and have been shown to bind DNA in a sequence-specific manner when present in other proteins (e.g., Iwahara et al.,1998; Wang et al.,1999; Huang et al.,2002; Lours et al.,2003). Pipsqueak domains are often found in proteins that have been linked to chromatin modification (Siegmund and Lehmann,2002). The PEDLS sequence present in Dan and Danr is related to motifs known to bind to the transcriptional co-regulator C-terminal Binding Protein (CtBP; reviewed in Chinnadurai,2007). Accordingly, Danr has been shown to bind Drosophila CtBP (dCtBP) in a genome-wide two-hybrid screen (Giot et al.,2003).

CtBP is a highly conserved transcriptional co-regulator first identified for its ability to bind to the C-terminus of the adenovirus E1A oncoprotein (Turner and Crossley,2001; Chinnadurai,2007). Both vertebrate and Drosophila CtBPs have been characterized as co-repressors for both short and long-range repressors, including Hairy, Snail, Krüppel, and mammalian Basic Krüppel-like Factor/Krüppel-like Factor 3 (Nibu et al.,1998a,b; Poortinga et al.,1998; Turner and Crossley,1998). Whereas vertebrates have two CtBP genes, invertebrates including Drosophila have a single CtBP gene (dCtBP) that encodes two major protein isoforms, one long (∼50 kDa), and one short (∼42 kDa; Nibu et al.,1998a,b; Poortinga et al.,1998; Sutrias-Grau and Arnosti,2004; Mani-Telang and Arnosti,2007). Based on current knowledge, the major function of vertebrate CtBP homologs is to link DNA-binding transcription factors to the chromatin modification machinery (reviewed in Turner and Crossley,2001; Chinnadurai,2007).

Here, we use both molecular and genetic evidence to show that Dan and Danr bind directly to Drosophila CtBP, and that they form a complex or complexes with CtBP in eye-antennal discs that play functional roles in both eye and antennal development. Additional molecular and genetic experiments show that CtBP also complexes with Ey and Dac during eye development. CtBP loss-of-function mutant eyes have more ommatidia than wild-type eyes, and CtBP mutant clones overproliferate anterior to the morphogenetic furrow. Based on this result and the fact that ey has previously been shown to promote proliferation anterior to the furrow (Bessa et al.,2002; Peng et al.,2009; Lopes and Casares,2010), we speculate that the presence of CtBP in a complex containing Ey ahead of the furrow modulates the effects of Ey on proliferation, limiting the pool of ommatidial precursors. In contrast, CtBP,dan,danr triple mutant adult eyes have significantly fewer ommatidia than either CtBP single or dan, danr double mutant eyes, suggesting that CtBP/Dan/Danr complex(es) promote differentiation of eye precursors into ommatidia. Recruitment and maintenance of the founding R8 precursor is affected in CtBP single mutants and to a much greater extent in CtBP,dan,danr triple mutants, suggesting that a function of the CtBP/Dan/Danr complex during differentiation is to firmly establish R8 cell fate. Taken together, our results suggest that CtBP is part of multiple complexes with different eye specification factors to participate in the transition of eye precursors from proliferation to differentiation.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Dan and Danr Interact With CtBP In Vitro and In Vivo

Dan and Danr have been shown to physically interact with Ey, Dac, and with each other in GST-pulldown and in yeast two-hybrid assays (Curtiss et al.,2007). A genome-wide yeast two-hybrid screen (Giot et al.,2003) also predicted a direct physical interaction between Danr and CtBP. CtBP has been shown to interact with many transcription factors and co-factors by means of motifs related to the PxDLS CTBP-binding motif originally identified in the adenovirus E1A protein (Schaeper et al.,1995; reviewed in Chinnadurai,2007).

Although functional PxDLS motifs can exhibit considerable sequence variability—Giot et al. (2003) identified 13 functional sequence variants—the PEDLS motifs in both Dan and Danr are perfectly conserved among Drosophila species (Fig. 1A). For D. melanogaster Dan, the PEDLS sequence lies 191 amino acids from the end of the protein; D. melanogaster Danr's PEDLS motif is 17 amino acids from the C-terminus. Dan and Danr are likely to have arisen from a gene duplication event, because they are only ∼45 kb apart in the genome and their sequences are more closely related to each other than to other known related proteins (Emerald et al.,2003). The duplication event appears to have occurred after the divergence between Nematocera and Brachycera suborders in Diptera, because the mosquito Aedes aegypti has only a single Dan/Danr ortholog. The predicted mosquito protein (included in the Dan alignment) also has a PEDLS sequence located 175 amino acids from the end of the protein. The presence of the PEDLS sequence among these species suggests that it has an important and conserved function.

thumbnail image

Figure 1. Dan and Danr interact with CtBP in vitro and in vivo. A: Alignments of partial predicted protein sequences for Danr (top) and Dan (bottom) orthologs in Drosophila species for which genome sequences are available. The Drosophila melanogaster sequence is boxed horizontally in gray. The highly conserved CtBP-binding PEDLS motif is boxed vertically in black. The single Aedes aegypti (mosquito) ortholog (bottom-most sequence in the Dan alignment) also contains a PEDLS sequence (boxed). B: GST-pulldowns for in vitro translated Dan (top panel) or Danr (bottom panel). GST-CtBP, GST-Danr, and GST-Dan interact with both Dan and Danr, but GST and GST-Six HD (a GST construct containing the Six domain and the homeodomain of Sine oculis) do not. C: Anti-Dan, but not anti-Tubulin, is capable of immunoprecipitating CtBP and Danr from eye-antennal disc extracts. D: Anti-Dan detects an ∼90 kDa band in extracts from wild-type eye-antennal discs. In dan,danrex56 homozygous eye-antennal discs this band is absent. The nonspecific band at ∼175 kDa serves as a loading control. E: Anti-CtBP, but not anti-Tubulin, is capable of immunoprecipitating Dan from eye-antennal disc extracts.

Download figure to PowerPoint

The data above prompted us to test whether Dan or Danr binds to CtBP in vitro, using GST pull-down assays. We incubated labeled in vitro translated Dan or Danr with GST, GST-Six HD (a truncated So protein containing only the Six domain and the Homeodomain), GST-CtBP, GST-Danr, and GST-Dan (Fig. 1B), the latter two serving as positive controls (Curtiss et al.,2007). Labeled in vitro translated Dan or Danr were both pulled down by GST-CtBP, GST-Dan, and GST-Danr, but not by GST-Six HD or by GST alone. Thus, CtBP binds directly with both Dan and Danr in vitro.

Next, we investigated whether CtBP, Dan and/or Danr can complex with each other in vivo. We performed co-immunoprecipitation (co-IP) experiments on lysates derived from eye-antennal disc/brain complexes, using either anti-Dan or anti-Tubulin antibody (Fig. 1C). On Western blots, a ∼50 kDa band was detected with anti-CtBP in immunoprecipitates obtained with anti-Dan but not anti-Tubulin. This corresponds to the size of the long CtBP isoform (Mani-Telang and Arnosti,2007). In addition, a ∼45 kDa band was detected with anti-Danr, which is the predicted size for Danr.

We also carried out the converse experiment, in which we performed co-IPs with anti-CtBP and probed Western blots with anti-Dan. The anti-Dan antibody recognizes a ∼90 kDa band on Western blots of wild-type eye-antennal disc lysates, and this band is missing from eye-antennal disc lysates from dan,danrex56 homozygotes (Fig. 1D; we have previously shown that Dan protein is undetectable by immunofluorescence in dan,danrex56 homozygous clones; Curtiss et al.,2007). A ∼90 kDa protein was detected by anti-Dan only in antibody only in immunoprecipitates obtained with anti-CtBP, but not in immunoprecipitates obtained with anti-Tubulin (Fig. 1E). These results suggest that Dan forms a complex with both Danr and CtBP. Due to lack of sufficient anti-Danr anti-sera, we were unable to perform the same experiment with Danr.

CtBP Interacts Genetically With dan and danr During Eye Development

Overexpression of genes involved in eye specification interferes with eye development, resulting in smaller, rougher eyes than wild-type (Curtiss and Mlodzik,2000; Curtiss et al.,2007). To test whether overexpression of CtBP interferes with eye development, we used the Gal4/UAS system (Brand and Perrimon,1993). P[GSV]A396 and P[GSV]A132 are bidirectional EP elements located in the first intron of CtBP that have previously been used to study CtBP function (Toba et al.,1999; Fang et al.,2006).

Similar to the effects of overexpressing ey, dac, dan, or danr, overexpression of CtBP in eye precursors under ey-Gal4 control disrupts eye development. The two CtBP EP lines gave similar effects with ey-Gal4, although P[GSV]A396 gave the strongest effects and was used in the experiments described below. One hundred percent of adult eyes from ey-Gal4/+; P[GSV]A396/+ (hereafter ey>CtBP) individuals raised at 30°C were smaller and rougher than wild-type eyes (Fig. 2A,B; 4 separate experiments, 169 flies total). Eyes ranged from 40–90% of wild-type size, with most (68%) greater than 50% wild-type size. In addition, the eyes were moderately rough and showed duplications of interommatidial bristles (Fig. 2B inset, arrows). Stern et al. (2007) also observed duplicated interommatidial bristles when they expressed CtBP under GMR-Gal4 control, corroborating our findings. These results provide evidence that CtBP functions in eye development.

thumbnail image

Figure 2. Overexpression of CtBP under ey-Gal4 control causes defects in eye and head capsule. A,B: Scanning electron micrographs of eyes from ey-Gal4/+; P[GSV]A396/+ flies. Inset in B is an enlargement of the boxed region in the larger picture. The ey>CtBP eyes are smaller than wild-type, contain extra interommatidial bristles (arrows in inset in B) and are occasionally split by cuticular structures (arrow in B). C,D: Scanning electron micrographs of dorsal views of heads from wild-type (A) and ey-Gal4/+; P[GSV]A396/+ (B) flies. The heads of ey>CtBP flies are occasionally split by a cleft (black arrow in D) and contain abnormalities in pattern such as the extra orbital bristles denoted by the white arrow in D.

Download figure to PowerPoint

Besides the eye defects, ey>CtBP flies had patterning defects in various regions of the head. These defects included: (1) loss of vibrissae and replacement by bristles resembling orbital or vertical bristles (not shown); (2) occasional splitting of the eye field, sometimes accompanied by formation of antennal-like structures in the cuticular field between eye domains (Fig. 2B, black arrow); (3) loss of distal antennal structures (not shown, ∼1% of flies); (4) apparent clefts in the vertex region of the head with loss of ocelli and mispatterning of bristles (31.7% of flies, N = 101; Fig. 2D, cleft denoted by black arrow; extra orbital bristles surrounded by a patch of orbital-like cuticle denoted by white arrow).

To explore the functional relationships among dan, danr, and CtBP, we tested the effects of overexpressing CtBP in individuals heterozygous for danems3, danrex35, and dan danrex56 mutations (Fig. 3; Table 1). Importantly, danems3, danrex35, and dan danrex56 heterozygotes all have normal-sized eyes with no roughness. We crossed ey>CtBP flies to either control w1118 flies or to flies containing dan/danr mutations. This experiment was performed several times, with similar results. In one particular experiment carried out at 25°C, 31% (N = 35) of ey>CtBP individuals had eyes that were smaller and rougher than wild-type eyes (Fig. 3A,B). In contrast, in the same experiment only 16% (N = 19) of ey>CtBP, danems3/+ (Fig. 3C), and only 13% (N = 30) of ey>CtBP/+; danrex35/+ (Fig. 3D) individuals had smaller than wild-type eyes, and the eye roughness and extra interommatidial bristle phenotypes were also considerably suppressed. These data suggest that CtBP interacts in the same developmental process as Dan and Danr, and that dan and danr are required for CtBP function.

thumbnail image

Figure 3. Dan and Danr are required for CtBP function, and vice versa. A–H: Scanning electron micrographs of eyes from the following genotypes: (A,B) ey-Gal4/+; P[GSV]A396/+; (C) ey-Gal4/+; P[GSV]A396/danems3; (D) ey-Gal4/+; P[GSV]A396/danrex35; (E) ey-Gal4, UAS-dan/+; (F) ey-Gal4, UAS-dan/CtBP87De-10; (G) ey-Gal4, UAS-danr/+; (H) ey-Gal4, UAS-danr/CtBP87De-10. Inset in A is an enlargement of the boxed region in the larger picture, showing the extra interommatidial bristles found when CtBP is overexpressed (arrows). A single copy of either the danems3 (C) or the danrex35 (D) mutations suppresses the small eye phenotype resulting from CtBP overexpression (A,B). Conversely, a single copy of the CtBP87De-10 mutation suppresses the small eye phenotype resulting from overexpression of either dan (compare E,F) or danr (compare G,H)

Download figure to PowerPoint

Table 1. Effects of Overexpressing CtBP
ey>CtBP/+25°Cw1118danems3/+danrex35/+  
% smaller than WT31%16%13%  
 N = 35N = 19N = 30  
ey>dan/+30°Cw1118CtBP87De-10/+CtBPrev19/+  
% smaller than 50% WT size36%9%17%  
N = 14N = 11N = 12  
ey>danr/+30°Cw1118CtBP87De-10/+   
% smaller than WT90%55%   
 N = 58N = 22   
ey>CtBP/+25°Cw1118ey2/+eyR/+dacP/+dac4/+
% smaller than WT size46%35%53%39%46%
N = 24N = 20N = 17N = 56N = 13
Size range60-80%80-100%80-95%60-80%60-80%
WT sizeWT sizeWT sizeWT sizeWT size
ey>ey/+30°Cw1118CtBP87De-10/+   
% smaller than WT size100%54%   
N = 63N = 26   

We also performed converse experiments in which we compared ey-Gal4, UAS-dan/+ (ey>dan) and ey-Gal4, UAS-danr/+ (ey>danr) to ey>dan, CtBP87De-10/+, and ey>danr, CtBP87De-10/+ (CtBP87De-10 heterozygotes have normal-sized eyes with no roughness). As described previously (Curtiss et al.,2007), misexpression of either dan or danr in eye precursors results in small, rough eyes in a significant percentage of the population. Removing a copy of CtBP in this background suppresses these defects in terms of both eye size and roughness (Table 1; compare Fig. 3E,F). Similarly, removing one copy of CtBP suppresses the small rough eye phenotype of ey>danr (Table 1; compare Fig. 3G,H). These data suggest that CtBP is required for the functions of dan and danr.

CtBP Functions in Complexes With Eyeless and Dachshund During Eye Development

It has previously been shown that Dan physically interacts with Ey and Dac in vitro and that dan mutations interact genetically with ey and dac mutations (Curtiss et al.,2007). However, it is not known whether Dan can complex with the Ey and Dac proteins in vivo. To answer this question, we performed co-IP experiments with anti-Dan as above, but probed the Western blots with antibodies specific for Ey and Dac (Fig. 4A). An ∼115 kDa protein and a ∼130 kDa protein were detected by anti-Ey and anti-Dac antibodies, respectively, only in immunoprecipitates obtained with anti-Dan, but not in immunoprecipitates obtained with anti-Tubulin. These apparent molecular weights have previously been reported for Ey (Callaerts et al.,2001; Punzo et al.,2001) and for Dac (Chen et al.,1997).

thumbnail image

Figure 4. CtBP interacts in eye-antennal discs with Dan and Ey, and interacts genetically with ey and dac, but not with eya or so. A: Anti-Dan, but not anti-Tubulin, is capable of immunoprecipitating Ey and Dac from eye-antennal disc extracts. B: Sepharose beads conjugated to anti-FLAG are capable of pulling down Dan and CtBP from extracts generated from eye-antennal discs expressing FLAG-tagged Ey; Sepharose beads alone do not pull down CtBP from the same extracts. C: Anti-CtBP, but not anti-Tubulin, is capable of immunoprecipitating Ey from eye-antennal disc extracts. D–L: Scanning electron micrographs of eyes from the following genotypes: (D,G) ey-Gal4/+; P[GSV]A396/+; (E) ey-Gal4/+; P[GSV]A396/+; ey2/+; (F) ey-Gal4/dac4; P[GSV]A396/+; ey-Gal4/+; P[GSV]A396/+; (H) ey-Gal4/eya2015; P[GSV]A396/+; (I) ey-Gal4/so3; P[GSV]A396/+; (J,K) ey-Gal4, UAS-ey/+; (L) ey-Gal4, UAS-ey/CtBP87De-10. A single copy of the ey2 (E) or dac4 (F) mutations suppress the small eye phenotype resulting from CtBP overexpression (D,G). Conversely, a single copy of eya2015 (H) or of so3 (I) enhances or has no effect on the CtBP overexpression phenotype, respectively. A single copy of the CtBP87De-10 mutation represses the effects of eyeless overexpression (compare L with J,K).

Download figure to PowerPoint

Conversely, we were able to pull down Dan from eye-antennal discs expressing a FLAG-tagged version of Ey (kind gift of Rui Chen), using Sepharose beads conjugated to anti-FLAG; Sepharose beads alone did not pull down Dan (Fig. 4B). Thus, results from GST-pulldowns (Curtiss et al.,2007) and co-IP assays indicate that Dan can directly bind Ey and Dac, and suggest that these factors form a complex or complexes in vivo.

Because Dan associates with both CtBP and Ey in vivo (Figs. 1C,D, 4A), we wondered whether CtBP and Ey might also associate in vivo. We, therefore, used anti-CtBP to probe Western blots containing co-IPs from eye-antennal discs expressing FLAG-tagged Ey. CtBP was detected in co-IPs obtained with anti-FLAG conjugated Sepharose (Fig. 4B), suggesting that CtBP and Ey associate in eye-antennal disc tissue during larval stages. The top-most band runs at an apparent molecular weight of ∼50 kDa, which corresponds to the long CtBP isoform. It is unclear whether the lower bands are degradation products, or if Ey is also capable of associating with the short ∼42 kDa CtBP isoform (Mani-Telang and Arnosti,2007). Finally, we were able to pull down Ey from eye-antennal discs using anti-CtBP (Fig. 4C). We have not reliably detected Dac in co-IPs obtained with anti-CtBP. These results suggest that CtBP interacts with both Dan and Ey during eye development.

We next tested whether CtBP also interacts functionally with ey or dac during eye development. Removing a copy of ey or dac in an ey>CtBP background using the ey2, eyR, dac4, dac3, or dacP alleles reproducibly suppresses the ey>CtBP phenotype (Table 1; Fig. 4D–F; not shown). Although the percentages of ey>CtBP,ey2/+; ey>CtBP,eyR/+; ey>CtBP,dacP/+, and ey>CtBP,dac4/+ with smaller than wild-type eyes is not suppressed compared with ey>CtBP alone, the eyes in these genotypes tend to be larger and much less rough. Eyes of ey2, eyR, dacP, and dac4 heterozygotes are normal in size and show no roughness. These data suggest that, like Dan and Danr, Ey and Dac are required for CtBP function.

Not all eye specification factors have the same relationship to CtBP as Dan, Danr, Ey, and Dac. For instance, ey>CtBP, eya2015/+ flies have slightly smaller and rougher eyes than ey>CtBP flies (compare Fig. 4G,H), suggesting that CtBP and Eya have antagonistic functions. Likewise, removing a copy of so (so3) in an ey>CtBP background has no discernible effect on eye size or roughness (compare Fig. 4G,I).

Finally, we showed that reducing the function of CtBP suppresses the effects of expressing ey under ey-Gal4 control. We have previously demonstrated that ey>ey flies have smaller, rougher eyes than wild-type (Curtiss and Mlodzik,2000; Curtiss et al.,2007). In one particular experiment, 100% (N = 63) of ey>ey flies had smaller eyes than wild-type. In contrast, only 54% (N = 26) of ey>ey, CtBP87De-10/+ flies had smaller eyes than wild-type (Table 1; compare Fig. 4J–L), suggesting that CtBP is required for Ey function. Taken together, these data suggest that Ey and Dac function in a complex or complexes with CtBP.

CtBP Is Required for Development of Eye-Antennal Disc-derived Tissues

Ey, Dac, Dan, and Danr have known roles in eye and antennal specification. CtBP is a well-characterized co-regulator that participates in many developmental contexts including oncogenesis and apoptosis (Chinnadurai, 2007), but no published studies have looked for a role in either eye or antennal specification in Drosophila. Because CtBP interacts with Ey, Dac, Dan, and Danr in vitro and in eye-antennal discs, we asked whether CtBP is necessary for eye or antennal development.

We used two independent CtBP mutant alleles (CtBP87De-10 and CtBP03463 – see Experimental Procedures for details) for our analysis. Although Poortinga et al. (1998) describe CtBP03463 as homozygous lethal at the pharate adult stage, in our hands both CtBP87De-10 and CtBP03463 are lethal before late larval stages. We generated flies trans-heterozygous for CtBP87De-10 and CtBP03463, and flies containing these mutations in trans with a deletion encompassing the CtBP locus [Df(3R)ry615]. In addition, we generated CtBP87De-10 homozygous loss-of-function clones using the FLP/FRT system (Xu et al.,1993) together with an ey-FLP transgene (Newsome et al.,2000), which induces FRT-mediated recombination in both eye and antennal portions of the eye-antennal discs. We also combined this with the Minute technique (Newsome et al.,2000) to generate individuals in which wild-type or CtBP87De-10 clones comprised most of the adult head (+/M, CtBP87De-10/M).

Homozygous mutant or wild-type tissue was marked in the larval eye-antennal disc by the absence of GFP, and examination of +/M or CtBP87De-10/M discs using fluorescence microscopy confirmed that the majority (>95%) of the eye-antennal disc cells were mutant (not shown). Clones were marked in the adult eye by the absence of pigment, but were not marked in other regions of the adult head. Wild-type clones produced by means of ey-FLP/FRT with or without the Minute technique resulted in no discernible defects (Fig. 5A and not shown). In all cases, a substantial number of CtBP trans-heterozygotes survived to the pharate adult stage, but none was ever observed to emerge from its pupal case; pharate adults were thus dissected from pupal cases for analysis. Many CtBP87De-10/M, and almost all CtBP87De-10 mosaic individuals survived to adulthood. Mutant phenotypes in the thorax of all of the trans-heterozygous combinations resembled those seen in other studies (Poortinga et al.,1998; Barolo et al.,2002; Biryukova and Heitzler,2008; Stern et al.,2009). CtBP87De-10/M individuals and CtBP trans-heterozygotes all had similar phenotypes in the eye, the bristles of the head, and the antennae (described in more detail below).

thumbnail image

Figure 5. CtBP mutants affect bristle development on the head capsule and antennae, and also affects eye size and development of interommatidial bristles. A–C: Scanning electron micrographs of dorsal views of heads from (A) eyFLP; FRT82 M/FRT82, (B) eyFLP; FRT82 M/FRT82 CtBP87De-10, and (C) CtBP87De-10/CtBP03463 flies. Panel (A) shows the normal pattern of bristles on the dorsal side of the head (fro, frontal; f-or, fronto-orbital; ioc, interocellar; oc, ocellar; orb, orbital; pv, post-vertical; ver, vertical). In this example of a stronger CtBP mutant phenotype (B), several of the vertical bristles are duplicated (arrows), most of the frontal bristles are missing, and all of the ocellar, interocellar and post-vertical bristles are missing. In this example of a weaker CtBP phenotype (C), vertical bristles and one of the orbital bristles are duplicated (arrows), and some of the fronto-orbital and interocellar bristles are missing. D,E: Light micrographs of wild-type (D) and CtBP87De-10/Df(3R)ry615 (E) antennae. In the CtBP- mutant, a2 and a3 appear bigger than wild-type, fewer bristles are found on the 2nd antennal segment, and the arista is shorter and misshapen compared with wild-type. F,G: Scanning electron micrographs of wild-type (F) and CtBP87De-10/CtBP03463 (G) eyes. Insets are enlargements of the boxed region shown in the larger picture. The CtBP- mutant eye is larger and is missing interommatidial bristles (inset). H: Bar graph showing the mean number of ommatidia for each of the indicated genotypes. Error bars show standard error of the mean. Genotypes were assigned to statistical groups using the Tukey-Kramer post hoc test. Groups sharing at least one letter are not significantly different; groups not sharing any letter are significantly different (P < 0.05).

Download figure to PowerPoint

CtBP is required for development of head bristles.

Head bristle phenotypes of the CtBP mutant combinations tested here were similar to those described for the thorax (Barolo et al.,2002; Stern et al.,2009), with some bristles duplicated, others missing, and some normal. Duplicated bristles were of two varieties: extra bristles (additional bristle and socket) or double bristles (two bristles in the same socket). In stronger mutant combinations (e.g., CtBP87De-10/M, Fig. 5B) extra or double bristles were always observed for one or more vertical (ver) bristles (arrows); most post-vertical (pv) bristles were missing; orbital (orb) bristles were mostly present, but occasionally missing or duplicated (not shown); all interocellar bristles, and most ocellar, frontal, and postorbital bristles were missing. Weaker combinations (e.g., CtBP87De-10/CtBP03463; Fig. 5C) had similar but milder head bristle phenotypes, with occasional duplicated vertical and orbital bristles (arrow), occasional double and/or misoriented orbital (arrow) and postvertical bristles (not shown), and variable loss of interocellar and frontal bristles. Barolo et al. (2002) previously mentioned similar head bristle phenotypes at a much-reduced frequency in CtBP87De-10/+ heterozygotes, and at a higher frequency in CtBP87De-10/CtBP03463 pharate adults.

In addition to the bristle phenotypes described above, the ocelli and surrounding ocellar cuticle appeared somewhat larger in CtBP- mutants compared with wild-type (compare Fig. 5A with Fig. 5B,C), although the distance across the dorsal head between eyes appeared to be approximately the same as in wild-type heads. The larger size of the ocellar region does not appear to be due to transformation of other regions of the head: although bristles are often missing or altered, distinctive regions of cuticle (e.g., the ridged frons cuticle, the orbital cuticle, etc) all appeared to be present. Rather, the relative sizes of the various regions may be changed in CtBP- mutants. These results suggest a minor role for CtBP in development of the morphology and bristles of the head.

CtBP is required for antennal bristle development and for development of the aristae.

We also noted defects in antennae in the CtBP loss-of-function mutants. Wild-type antennae comprise five segments plus the arista (Fig. 5D; a4 is not readily visible in this preparation). Second and third antennal segments (a1, a2, and a3) appeared somewhat larger in CtBP loss-of-function mutant combinations (compare Fig. 5D,E). In addition, the aristae appeared to be slightly shorter, with the main shaft tapering more prominently along the proximal–distal axis (arrow in Fig. 5E). Branches from the main shaft were curved. Unlike dan and danr single or double mutants, we saw no bracted bristles or other obvious evidence of antenna-to-leg transformation. Finally, the number of bristles on a2 was reduced; the loss of bristles appeared more severe in CtBP87de10/Df(3R)ry615 individuals than in other mutant combinations. Johnston's organs in a2 appeared normal (not shown). These data suggest a minor role for CtBP in development of the size and morphology as well as the bristles of the antennae.

CtBP is required to limit eye size, and for interommatidial bristle formation.

CtBP mutant combinations have larger than wild-type eyes (compare Fig. 5F,G) with an increase in the number of ommatidia. In our hands, wild-type (Oregon R) eyes (N = 5) have a mean of 740 ommatidia, similar to what has been found by others (∼750; Wolff and Ready,1993). In contrast, CtBP03463/Df(3R)ry615 (N = 2), CtBP87De-10/Df(3R)ry615 (N = 3) and CtBP87De-10/CtBP03463 (N = 3) have means of 938, 818, and 906 ommatidia, respectively (Fig. 5H). A significant overall difference among the mean number of ommatidia of the 4 genotypes was detected (one-way analysis of variance [ANOVA], P < 0.0001). A Tukey-Kramer post hoc test revealed multiple significant differences among individual pairs of genotypes (Fig. 5H). In particular, the test demonstrates that CtBP03463/Df(3R)ry615, CtBP87De-10/CtBP03463, and CtBP87De-10/Df(3R)ry615 eyes have statistically significantly more ommatidia than wild-type eyes. These results suggest that CtBP functions to prevent eyes from growing too large.

Besides the effects on eye size, the CtBP mutant eyes were mildly rough and showed a loss of interommatidial bristles (Fig. 5F,G insets). The latter effect is opposite to the extra interommatidial bristles that result from overexpression of CtBP (Fig. 2B; Stern et al.,2007) and suggests that CtBP is required for interommatidial bristle formation.

CtBP regulates proliferation ahead of the morphogenetic furrow during eye development.

Loss of CtBP could result in an increased number of ommatidia for several reasons. It is possible that the entire head is larger in CtBP- mutants. However, based on our observations (e.g., Fig. 5), the CtBP- eyes appear to be proportionally larger than other structures that make up the head capsule. Although at present we have no definitive evidence, as noted above CtBP may also affect growth of other discrete regions of the eye-antennal disc, including the ocelli (compare Fig. 5A with Fig. 5B,C) and parts of the antennae (compare Fig. 5D with Fig. 5E). In addition, we have not noticed major patterning defects in CtBP mutant heads. For instance, as described above the extra ommatidia in CtBP- eyes do not appear to replace particular regions of head cuticle. Nor have we observed obvious ectopic dorsal/ventral margins, which can serve as organizers of ectopic growth in the eye-antennal disc (c.f., Cavodeassi et al.,1999; Gutierrez-Avino et al.,2009), as assessed by staining CtBP trans-heterozygous eye-antennal discs for markers of the D/V boundary.

To address the reason for overgrowth, we asked whether CtBP mutants affect the number of cells generated during larval eye-antennal disc development, increasing the pool of cells from which ommatidia form. Any additional cells that may be present in CtBP- mutants would likely be generated ahead of the furrow: because the number of ommatidia is determined in the furrow, increased proliferation behind the furrow would not be expected to lead to additional ommatidia, but instead to additional cells available for recruitment into existing ommatidia. In cases where extra cells are recruited into ommatidia or are otherwise allowed to survive (e.g., if apoptosis fails during pupal stages), the resulting adult eyes are very rough (c.f., Wolff and Ready,1991). CtBP- eyes exhibit little or no roughness.

Consistent with the idea of additional prefurrow cells in CtBP- mutants, we noticed that CtBP87De-10 prefurrow clones appeared larger than wild-type clones in third instar larval eye-antennal discs, often tapering off in size close to and behind the furrow (Fig. 6A,B). We also tested CtBP87De-10,dan,danrex56 triple clones (described in more detail below), and these also appeared larger compared with either wild-type clones or dan, danrex56 double mutant clones (Fig. 6C,D).

thumbnail image

Figure 6. CtBP87De-10 clones are larger than wild-type clones. A–D: Fluorescence micrographs of eye-antennal discs containing wild-type (A); CtBP87De-10 (B); CtBP87De-10,dan,danrex56 (C) and dan,danrex56 (D) clones. E: Bar graph showing the mean ratio of prefurrow clone area divided by prefurrow total area for each of the indicated genotypes. Error bars show standard error of the mean. Genotypes were assigned to statistical groups using the Tukey-Kramer post hoc test. Groups sharing at least one letter are not significantly different; groups not sharing any letter are significantly different (P < 0.05). F: Bar graph showing the mean ratio of number of anti-PH3-stained cells in prefurrow clones divided by total area of prefurrow clones, for each of the indicated genotypes. Error bars show standard error of the mean.

Download figure to PowerPoint

We measured the areas of wild-type, CtBP87De-10, CtBP87De-10,dan, danrex56, and dan,danrex56 prefurrow clones, and divided the total prefurrow clone area by the total prefurrow area to obtain a ratio. Mean ratios for CtBP87De-10 (0.45 ± 0.05), and CtBP87De-10,dan,danrex56 (0.48 ± 0.05) were larger than for wild-type (0.31 ± 0.04) or dan,danrex56 (0.32 ± 0.04; Fig. 6E), and we detected a significant overall difference among the means of the ratios (one-way ANOVA, P < 0.02). A Tukey-Kramer post hoc test found a significant difference between the wild-type and CtBP87De-10,dan, danrex56 ratios.

To determine whether the difference in clone size might be caused by an increase in proliferation in CtBP87De-10 mutant tissue, we stained eye-antennal discs containing wild-type or CtBP87De-10 clones with anti-Phospho-Histone H3 (anti-PH3) to mark cells in mitosis. We then counted the number of anti-PH3–positive cells in all clones ahead of the furrow, and divided this number by the total prefurrow clone area in each disc. We obtained a mean ratio of 0.0024 ± 0.0003 for homozygous CtBP87De-10 clones, vs. 0.0015 ± 0.0003 for wild-type clones, a difference that is statistically significant (t-test; P < 0.05; Fig. 6F). These results suggest that CtBP controls the number of cells anterior to the furrow by regulating their proliferation.

This result is particularly interesting in light of the fact that Ey has been implicated in promoting proliferation in eye precursors anterior to the furrow (Bessa et al.,2002; Peng et al.,2009; Lopes and Casares,2010), in the same region where we have shown that CtBP is limiting proliferation. In addition, we have demonstrated that CtBP and Ey can associate in eye-antennal disc extracts (Fig. 4). Based on this information, we propose that the purpose of the interaction between CtBP and Ey ahead of the furrow in the larval eye disc is to restrain the effects of Ey on eye precursors, allowing their proliferation to be tightly controlled.

To summarize our results, CtBP87De-10 and CtBP87De-10,dan, danrex56 triple mutant clones ahead of the furrow in the larval eye disc are larger than wild-type clones. Furthermore, CtBP87De-10 single mutant clones contain more mitotic figures anterior to the morphogenetic furrow, where cells are proliferating but have not yet begun to differentiate, suggesting that more cells are being produced in these clones. In the case of the CtBP single mutant clones, this is consistent with the fact that CtBP trans-heterozygous or CtBP87De-10/M mosaic mutant adult eyes contain increased numbers of ommatidia. Thus, in CtBP single mutants increased numbers of eye precursors are produced, and are able to differentiate into increased numbers of ommatidia.

The fact that CtBP87De-10,dan, danrex56 triple mutant clones are similar to CtBP87De-10/M single mutant clones in being larger than either wild-type or dan,danrex56 double mutant clones, is consistent with the fact that Dan and Danr are not expressed in anterior regions of the eye-antennal discs, and do not appear to have a role there (Emerald et al.,2003; Suzanne et al.,2003; Suzanne,2004; Curtiss et al.,2007). Thus, the large size of CtBP87De-10,dan,danrex56 clones is likely to be due solely to the effects of CtBP on proliferation.

CtBP Functionally Interacts With dan and danr During Eye and Antennal Development

It has previously been shown that dan and danr have roles in eye and antennal specification (Emerald et al.,2003; Suzanne et al.,2003; Curtiss et al.,2007). Here, we have demonstrated that CtBP can bind directly to Dan and Danr in vitro, that CtBP can interact in eye-antennal disc extracts with Dan (Fig. 1) and that CtBP has a role in development of eye-antennal disc-derived structures (Figs. 5, 6). To test whether CtBP, Dan, and Danr interact functionally during eye and antennal development, we constructed danems3,CtBP87De-10 and danrex35,CtBP87De-10 double mutants, and dan,danrex56,CtBP87De-10 triple mutants. Because these double and triple mutant combinations are homozygous lethal, we used the eyFLP/FRT system together with the Minute technique to generate mosaic individuals in which the homozygous double and triple mutant clonal tissue comprised most of the eye-antennal-derived structures in the adult head (+/M; CtBP87De-10/M; danems3/M; CtBP87De-10, danems3/M; danrex35/M; CtBP87De-10, danrex35/M; dan, danrex56/M; and CtBP87De-10, dan, danrex56/M).

The effects on head bristles in these double and triple mosaic mutants appeared identical to those observed for the CtBP87De-10/M mosaics (not shown), suggesting that CtBP is required for normal head bristle development but Dan and Danr are not. This is consistent with the fact that Dan and Danr are not expressed in the regions of the eye-antennal disc that give rise to these bristles, and that dan and danr mutants have no discernible effect on head bristle development (Emerald et al.,2003; Suzanne et al.,2003; Suzanne,2004; Curtiss et al.,2007). In contrast, as described below, the CtBP, dan, and danr double and triple mutants had much stronger effects on antennal and eye development than any of the single mutants or the dan, danrex56 double mutant.

CtBP, Dan, and Danr interact genetically to promote antennal specification.

Strikingly, whereas CtBP87De-10/M, danems3/M, danrex35/M, and dan danrex56/M mosaics have relatively minor defects in antennae, the CtBP87De-10,danems3/M and CtBP87De-10, danrex35 double and CtBP87De-10,dan, danrex56 triple mutant mosaics result in strong antenna-to-leg transformations. As previously described (Emerald et al.,2003), danems3 (Fig. 7A) and danrex35 (Fig. 7B) single mutants, and the dan danrex56 double mutant (Fig. 7C) exhibit defects that indicate mild antenna-to-leg transformation of the distal antenna. These defects are characterized by a reduction in the size of the a3 antennal segment, the presence of bristles on the a3 antennal segment and an enlarged basal cylinder of the arista with bracted bristles, which are diagnostic of leg cuticle. As described above, CtBP mutants have bristle loss on a2 and defects in aristal morphology, but no obvious changes in size of the basal cylinder of the arista, and no bracted bristles. In contrast, double and triple dan/danr/CtBP mutant clones show significant enlargement all along the length of the arista to form a structure that strongly resembles a leg; bracted bristles were observed on these structures, confirming leg identity (Fig. 7D–F). These data provide evidence that dan/danr/CtBP interact functionally during antennal specification.

thumbnail image

Figure 7. CtBP interacts genetically with dan and danr for antennal development. A–F: Light micrographs of the following genotypes: (A) eyFLP; FRT82 danems3/FRT82 M; (B) eyFLP; FRT82 danrex35/FRT82 M; (C) eyFLP; FRT82 dan danrex56/FRT82 M; (D) eyFLP; FRT82 CtBP87De-10, danems3/FRT82 M; (E) eyFLP; FRT82 CtBP87De-10, danrex35/FRT82 M; (F) eyFLP; FRT82 CtBP87De-10, dan danrex56/FRT82 M. A–C: Single dan (A) and danr (B) or double dan danr (C) mutants show mild transformation of distal antenna to leg, as demonstrated by the smaller relative size of a3 and the fact that a3 contains bristles (white arrow), by the enlarged a5 with bracted bristles (black arrow), and by the defects in aristae. D–F: Double CtBP, dan (D), and CtBP, danr (E), and triple CtBP, dan danr mutants show a dramatically stronger antennal-to-leg transformation, with a strong enhancement in size of a5 (black arrows) and corresponding shortening of the arista. For wild-type and CtBP single mutant, see Fig. 2D,E).

Download figure to PowerPoint

CtBP, Dan, and Danr interact genetically during eye development.

As with the antenna, whereas CtBP87De-10/M, danems3/M, danrex35/M, and dan danrex56/M mosaics have relatively minor defects in the eye, the CtBP87De-10,danems3/M and CtBP87De-10,danrex35 double and CtBP87De-10,dan,danrex56 triple mutant mosaics result in strong eye phenotypes (Fig. 8). We counted ommatidia in the eyes of adult flies containing wild-type or CtBP, dan, or danr homozygous single, double, and triple mutant clones generated by means of eyFLP and the Minute technique (Fig. 8A), and detected a significant overall difference among the mean number of ommatidia of the eight genotypes (one-way ANOVA, P < 0.0001). A Tukey-Kramer post hoc test revealed multiple significant differences among individual pairs of genotypes.

thumbnail image

Figure 8. CtBP interacts genetically with dan and danr for eye development. A: Bar graph showing the mean number of ommatidia for each of the indicated genotypes. Error bars show standard error of the mean. Genotypes were assigned to statistical groups using the Tukey-Kramer post hoc test. Groups sharing at least one letter are not significantly different; groups not sharing any letter are significantly different (P < 0.05). B–I: Scanning electron micrographs of eyes of the following genotypes: (B) eyFLP; FRT82/FRT82 M; (C) eyFLP; FRT82 CtBP87De-10/FRT82 M; (D) eyFLP; FRT82 danems3/FRT82 M; (E) eyFLP; FRT82 CtBP87De-10, danems3/FRT82 M; (F) eyFLP; FRT82 danrex35/FRT82 M; (G) eyFLP; FRT82 CtBP87De-10, danrex35/FRT82 M; (H) eyFLP; FRT82 dan danrex56/FRT82 M; (I) eyFLP; FRT82 CtBP87De-10, dan danrex56/FRT82 M. Eyes mosaic for the double CtBP, dan mutant are significantly smaller than those mosaic for single CtBP or dan mutants; eyes mosaic for the double CtBP, danr mutant are significantly smaller than those mosaic for single CtBP or danr mutants; and eyes mosaic for the triple CtBP, dan danr mutant are significantly smaller than those mosaic for single CtBP or the double dan danr mutants.

Download figure to PowerPoint

CtBP87De-10/M mosaic eyes have a mean of 722 ommatidia (N = 3), which is larger than control +/M mosaic eyes (mean = 692 ommatidia; N = 4; compare Fig. 8B,C), although not statistically significantly so. This result is consistent with the significantly larger eyes observed in CtBP trans-heterozygous combinations (Fig. 5F–H), and supports the idea that CtBP is required to prevent eyes from growing too large. The fact that the CtBP87De-10/M mosaic eyes are not statistically significantly larger than +/M mosaic eyes is likely to reflect the fact that these eyes are not entirely composed of homozygous CtBP mutant tissue.

Consistent with a previous description of dan/danr mutant phenotypes (Curtiss et al.,2007), the danems3/M (Fig. 8D) and dan,danrex56/M (Fig. 8H) mosaic eyes have comparable numbers of ommatidia compared with control +/M mosaic eyes, and danrex35/M mosaic eyes (Fig. 8F) have statistically significantly fewer ommatidia than the control. However, and strikingly, CtBP87De-10,danems3/M (Fig. 8E), CtBP87De-10,dan,danrex56/M (Fig. 8I), and CtBP87De-10,danrex35/M (Fig. 8G) mosaic eyes have statistically significantly fewer ommatidia and are much rougher compared with danems3/M, dan,danrex56/M, or danrex35/M mosaic eyes, respectively.

Thus, although CtBP87De-10,dan, danrex56 triple mutant clones ahead of the furrow in the larval eye disc are larger than wild-type clones (Fig. 6), CtBP87De-10,dan,danrex56/M mosaic adult eyes are significantly smaller than either CtBP87De-10/M single or dan,danrex56/M double mutant eyes. These results appear at first glance to be contradictory. However, as described above, it is most likely that the effect on triple mutant clone size far anterior to the furrow is due solely to loss of CtBP, because Dan and Danr are not expressed in this part of the disc (Curtiss et al.,2007). Therefore, although more cells are produced in the CtBP,dan,danr mutant clones, they either die instead of differentiating, or many fewer ommatidia are generated compared with CtBP single or dan,danr double mutant eyes. This genetic interaction suggests that, in addition to its early role in preventing overproliferation of eye precursors ahead of the furrow, CtBP also functions together with Dan and Danr at later stages to promote ommatidial cell survival and/or recruitment.

CtBP, Dan, and Danr interact to promote formation of the founding R8 photoreceptor.

To understand the reasons for the reduction in ommatidia in the CtBP/dan/danr double and triple mutants, we sectioned CtBP87De-10,dan,danrex56/M mosaic adult eyes to determine whether particular ommatidial cell types are affected in this genotype. In addition, we examined ommatidial development in CtBP87De-10, dan, danrex56/M mosaic eye-antennal imaginal discs using an antibody against the neuronal marker Elav, which stains photoreceptors. Apical sections of wild-type eyes reveal the stereotypic structure of a typical ommatidium (Fig. 9A, inset): 6 “outer” photoreceptors (R1–R6) with large rhabdomeres surrounding 1 “inner” photoreceptor (R7) with a small rhabdomere. In more basal sections the R8 photoreceptor, which lies below R7, is found in the center of the outer photoreceptors and has a small rhabdomere (Wolff and Ready,1993).

thumbnail image

Figure 9. CtBP, Dan, Danr function together during recruitment of the founding R8 photoreceptor. A: Apical section of an eyFLP; FRT82 CtBP87De-10,dan danrex56/FRT82 M adult eye at the R7 level. Inset shows a single wild-type ommatidium, enlarged, and numbers mark rhabdomeres of the individual R1-R7 photoreceptors. Black arrow: normal-appearing ommatidium, white arrow: gap in the ommatidial array, red arrow: ommatidium with too few photoreceptors, green arrow: ommatidium with too few outer photoreceptors and two inner photoreceptors, yellow arrow: ommatidium containing an inner photoreceptor with a large rhabdomere. B: Wild-type late third instar eye-antennal disc stained with anti-Elav (red). C–J: Eye-antennal discs containing wild-type (+/M; E,H), CtBP87De-10 (CtBP87De-10/M; C,F,I) or CtBP87De-10,dan danrex56 (CtBP87De-10,dan danrex56/M; G,J) clones of cells (β-galactoside negative) generated using the FLP/FRT system together with eyFLP and the Minute technique. Discs were stained with anti-β-gal (green), anti-Elav (red) and either anti-Ato (E–G) or anti-Sens (H–J). Gaps occur in the array of developing ommatidia in CtBP87De-10/M and CtBP87De-10,dan danrex56/M discs (arrows in C,D). In CtBP87De-10/M discs, initial Ato expression ahead of the furrow is wider in some places than in wild-type or +/M discs (brackets in E′,F′), and there are gaps in the array of R8 precursors expressing Ato (arrows in E′, F′). In CtBP87De-10,dan danrex56/M discs, initial Ato expression ahead of the furrow is consistently wider than in wild-type or +/M discs (bracket in G′), and there large gaps in the array of R8 precursors expressing Ato (arrows in G′). Whereas Sens expression in R8 equivalence groups (asterisks in H′–J′) appears normal, gaps in single R8 Sens expression correspond with loss of ommatidial differentiation (arrows in H,H′,I,I′,J,J′). Furthermore, the number of single R8 cells with strong Sens expression is reduced in CtBP87De-10/M and CtBP87De-10,dan danrex56/M discs (compare width of brackets in H′–J′).

Download figure to PowerPoint

Previous studies have demonstrated that dan,danrex56 double mutant ommatidia infrequently have the following defects: loss of one or more photoreceptors, two “inner” photoreceptors with small rhabdomeres rather than one, “inner” photoreceptors with a large rhabdomere (suggesting conversion of an “inner” photoreceptor to an “outer” photoreceptor fate, and gaps in the arrangement of ommatidia (Curtiss et al.,2007). Although CtBP87De-10,dan,danrex56/M eyes have significantly fewer ommatidia than dan,danrex56 eyes (Fig. 8A, H,I), sections reveal that CtBP87De-10, dan,danrex56/M eyes have similar defects in ommatidial structure compared with dan,danrex56 double mutants (Fig. 9A), albeit at a much higher frequency. Whereas many ommatidia have a normal complement of photoreceptors (black arrow), others show loss of photoreceptors (red and green arrows), gain of “inner” photoreceptors with small rhabdomeres (green arrow), apparent conversion of “inner” to “outer” photoreceptor fate (yellow arrow), and gaps in the ommatidial array (white arrow). Consistent with the smaller size of CtBP87De-10,dan,danrex56/M eyes, CtBP87De-10,dan,danrex56/M triple mutant eyes have significantly more gaps in the ommatidial array than dan,danrex56/M double mutant eyes (not shown).

Accordingly, CtBP87De-10,dan, danrex56/M larval eye-antennal discs stained for the neuronal marker Elav consistently show 5–6 gaps in the developing ommatidial array (Fig. 9D). Surprisingly, given that CtBP87De-10/M adult eyes are larger than +/M adult eyes, CtBP87De-10/M eye-antennal discs consistently have 1–2 gaps in the ommatidial array (Fig. 9C, arrow). These results confirm that CtBP, Dan, and Danr function together to promote ommatidial recruitment, and also suggest that loss of CtBP by itself affects ommatidial recruitment to a minor degree. The fact that CtBP mutant adult eyes have extra ommatidia despite a failure to differentiate a few ommatidia suggests that the effect of CtBP mutants on proliferation is even greater than is evident from the adult eye phenotype.

Ommatidial development depends on specification of the founding R8 photoreceptor, which subsequently recruits other photoreceptors and ommatidial accessory cells. R8 specification in turn depends on expression of the pro-neural gene atonal (ato), which encodes a bHLH transcription factor. Expression of the Ato protein initiates in all cells in a narrow band anterior to the morphogenetic furrow (e.g., Fig. 9E′, bracket). In wild-type eye-antennal discs, Ato is sequentially restricted to evenly spaced clusters of 8–10 cells referred to as intermediate groups, followed by groups of 2–3 cells that form an R8 equivalence group, and finally a single cell, the R8 precursor. Ato is expressed in R8 precursors for 2–3 rows, and subsequently fades (Jarman et al.,1994,1995; Baker et al.,1996; Dokucu et al.,1996; Frankfort and Mardon,2002). The senseless (sens) gene is an Ato target whose expression is initiated in the R8 equivalence group, becomes restricted to the R8 and is also required for R8 development (Frankfort et al.,2001; Pepple et al.,2008).

The ommatidial phenotypes described for dan,danrex56 mutants have previously been ascribed to a reduction in Ato expression anterior to the furrow, leading to inefficient and incomplete R8 specification (Curtiss et al.,2007). Because it seemed plausible that the stronger phenotype in CtBP87De-10,dan,danrex56/M might result from an even stronger effect on Ato expression resulting from synergism among CtBP, Dan, and Danr, we stained CtBP87De-10/M and CtBP87De-10,dan,danrex56/M eye-antennal discs with anti-Ato.

Ato expression in +/M discs (Fig. 9E) appeared identical to wild-type discs (not shown). Defects in CtBP87De-10/M were variable and minor: in some cases the initial swath of cells expressing Ato ahead of the furrow appeared wider than in wild-type discs (compare the width of brackets in Fig. 9E,F), and there were occasional gaps in the array of single R8 precursors (arrows in Fig. 9F), consistent with the gaps in the array of developing ommatidia described above. In the CtBP87De-10,dan, danrex56/M mutant discs the initial swath of Ato-expressing cells ahead of the furrow is consistently wider (bracket in Fig. 9G), and there were many gaps in the single R8 precursor array (arrows in Fig. 9G). Expression of Ato in intermediate groups and R8 equivalence groups appeared normal in both CtBP87De-10/M and CtBP87De-10,dan,danrex56/M mutant discs. We do not currently know what effect the increase in the number of Ato-expressing cells ahead of the furrow may have on R8 development; however, our data clearly shows that R8 establishment sometimes fails in CtBP87De-10/M and even more strikingly in CtBP87De-10,dan,danrex56/M mutant discs.

To determine whether the gaps observed in the ommatidial array in CtBP87De-10,dan,danrex56/M and to a lesser extent in CtBP87De-10/M eye-antennal discs in fact result from gaps in the formation of R8 precursors, we stained the discs with antibodies against the R8 marker Sens. In both CtBP87De-10/M and CtBP87De-10,dan, danrex56/M discs the gaps in the ommatidial array revealed by anti-Elav expression corresponded with gaps in R8 precursor formation (Fig. 9I,J, arrows). This indicates that the gaps in ommatidia observed in these genotypes results from failure to establish R8 precursors. As with Ato, expression of Sens in R8 equivalence groups appeared normal in CtBP87De-10/M and CtBP87De-10,dan, danrex56/M mutant discs.

In addition to the gaps, we also noticed that, whereas in wild-type (not shown) or +/M (Fig. 9H, bracket) discs Sens expression in single R8 precursors remains strong for ∼4 rows before fading slightly (Fig. 9H′, bracket), Sens expression in CtBP87De-10/M and particularly in CtBP87De-10,dan,danrex56/M discs is strong for only ∼2–3 rows before it fades (dramatically in CtBP87De-10, dan,danrex56/M discs; Fig. 9I′ and 9J′, brackets) and finally rebounds somewhat by approximately the 8th row. This result suggests that CtBP, Dan, Danr are required to maintain high levels of Sens expression in R8 precursors, notably during the period when each R8 precursor is recruiting additional photoreceptors into the nascent ommatidial clusters. This may reflect a role for these factors in “locking” in the correct fate (perhaps by means of chromatin modification) following R8 specification, and likely explains the loss of photoreceptors and other defects observed in CtBP87De-10,dan,danrex56/M mutant eye sections (Fig. 9A).

Importantly, the fact that early expression of Ato in intermediate groups and R8 equivalence groups, and of Sens in R8 equivalence groups, appears normal in both CtBP87De-10/M and CtBP87De-10,dan,danrex56/M discs (Fig. 9F–J), suggests that the roles of CtBP, Dan, Danr lie in the transition from the R8 equivalence group to a single R8 precursor. Taken together, our results strongly indicate that a complex containing CtBP, Dan and possibly Danr functions during ommatidial recruitment, promoting the establishment of R8 precursors from the R8 equivalence group, as well as the maintenance of the R8 fate that is required for proper recruitment of other cell types into ommatidia.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

CtBP is a transcriptional co-regulator known to play many roles during Drosophila development. Here, we demonstrate that CtBP can bind the probable transcription factors Dan and Danr in vitro, and that CtBP is part of a protein complex (or complexes) in the eye-antennal disc that includes Dan as well as the retinal determination factors Ey and Dac. As has been previously shown for the thorax, CtBP has a role in bristle development in the head capsule. In addition, CtBP regulates eye size, at least in part by limiting proliferation in eye precursors ahead of the morphogenetic furrow. We speculate that CtBP may interact with Ey to regulate proliferation of eye precursors at these early stages. Finally, CtBP synergizes with Dan and Danr for antennal specification and for maintenance/recruitment of eye precursors into ommatidia. Our results suggest that the CtBP co-regulator complexes with eye and antennal specification factors at multiple stages to regulate cell fate specification.

A Role for CtBP in Regulating Proliferation in Eye Precursors

The eyes of several CtBP loss-of-function mutant combinations have statistically significantly more ommatidia than wild-type eyes, and CtBP87De-10 clones ahead of the furrow show a statistically significant increase in the number of cells undergoing mitosis per unit area. These results suggest a role for CtBP in downregulating proliferation of eye precursors, although we cannot currently exclude additional processes such as apoptosis that might contribute to the increase in eye size in CtBP- mutants.

The evidence presented here suggests that CtBP is required to down-regulate proliferation of eye precursors ahead of the morphogenetic furrow: CtBP- clones are larger and show more mitotic figures than wild-type clones in the most anterior regions of the eye field (Fig. 6). As described below, several factors are known to promote proliferation ahead of the furrow, and have been connected to CtBP in some way. Many of these factors can cause massive overgrowth when overexpressed. Thus, the role of CtBP may be to counteract the activity of one or more of these factors, to ensure that cells do not proliferate out of control.

Factors that have previously been linked to CtBP and are known to regulate proliferation of eye precursors ahead of the furrow include the Wingless (Treisman and Rubin,1995; Lee and Treisman,2001; Baonza and Freeman,2002), Notch (Dominguez and de Celis,1998; Papayannopoulos et al.,1998; Morel et al.,2001; Barolo et al.,2002; Nagel et al.,2005,2007; Bray,2006; Kovall,2008; Protzer et al.,2008; Borggrefe and Oswald,2009; Fortini,2009), and JAK/STAT (Chao et al.,2004; Dominguez and Casares,2005; Reynolds-Kenneally and Mlodzik,2005; Gutierrez-Avino et al.,2009) signaling pathways. We have not ruled out a possible interaction between CtBP and these signaling pathways in the context of eye precursor proliferation.

However, the combination of Hth, Tsh, and Ey also regulates proliferation in ahead of the morphogenetic furrow (Bessa et al.,2002; Peng et al.,2009; Lopes and Casares,2010), and we have demonstrated here that Ey and CtBP are part of a complex in eye-antennal disc cells, and that they interact genetically during eye development. Circumstantial evidence also suggests links between CtBP and Hth and Tsh. For instance, Drosophila and mouse Tsh homologs both contain a PxDLS motif, and have been shown to interact in vitro with Drosophila and mouse CtBP, respectively (Saller et al.,2002; Manfroid et al.,2004). In addition, the Drosophila Cdc25 homolog encoded by string, which triggers mitosis, appears to be a target of Hth, although it is not clear if it is a direct target (Lopes and Casares,2010). DamID experiments with CtBP in Kc cells have also identified string as a potential direct CtBP target (Bianchi-Frias et al.,2004). Based on current data, we, therefore, propose that CtBP interacts with the Hth/Tsh/Ey complex in eye precursors ahead of the furrow.

CtBP and the Transition From Proliferation to Differentiation in the Drosophila Eye

CtBP87De-10 single mutant clones ahead of the furrow in the larval eye disc are larger than wild-type clones and show evidence of eye precursor overproliferation. Accordingly, adult eyes of CtBP trans-heterozygous combinations or of CtBP87De-10/M mosaic individuals have more ommatidia than wild-type eyes. In contrast, whereas CtBP87De-10,dan,danrex56 triple mutant clones are similar to CtBP87De-10 single mutant clones in being larger than wild-type clones ahead of the furrow, CtBP87De-10,dan, danrex56 adults have small rough eyes. This suggests either that the CtBP87De-10,dan,danrex56 triple mutant cells fail to be efficiently recruited into ommatidia and/or eventually undergo apoptosis. In support of the former hypothesis, the phenotypic analysis shown in Figure 9 demonstrates that recruitment of the R8 photoreceptor, which is required to recruit all other ommatidial cells, is affected by loss of CtBP, and is more strongly affected in the CtBP,dan, danr triple mutant.

Given the dynamic and overlapping expression patterns of the retinal determination factors, one intriguing hypothesis that fits our data is that a complex containing CtBP may have different members at different stages of eye development. For instance, a complex containing CtBP, Ey as well as possibly Tsh and Hth anterior to the furrow, might participate in maintaining a “poised” chromatin structure with respect to eye specific genes, in which genes involved in eye differentiation are not yet expressed and the cells are kept in a proliferative state. It has been suggested that vertebrate Pax6 is a “pioneering” factor for the lens lineage (Yang et al.,2006), and other “pioneering” factors have been shown to promote a “bivalent” state in which developmental genes are silenced, but “poised” for activation (e.g., Bernstein et al.,2006). The down-regulation of Ey close to the furrow, and the initiation of Dan, Danr, and Dac expression in the same region would be expected to change the composition of the complex containing CtBP and lead to changes in transcription that reflect the transition from proliferation to differentiation.

At present, we do not know what genes might be direct targets of the complexes containing CtBP and the eye specification factors. Some possibilities include the cell cycle regulator string (mentioned above), and the pre-proneural gene atonal, which is known to be regulated by Ey, So, Dan, and Danr (Ey and So are direct regulators; Zhang et al.,2006; Curtiss et al.,2007), and which plays a critical role in the transition from proliferation to differentiation of eye precursors. Thus, future work on Drosophila CtBP will shed light on the functions of this important transcriptional regulator, as well as on important transitions during development.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Alleles

The danems3, danrex35, dan danrex56 mutations have been described previously (Emerald et al.,2003; Curtiss et al.,2007). CtBP87De-10 and CtBP03463 were both obtained from the Bloomington Drosophila stock center. CtBP87De-10 (Hilliker et al.,1980) is reported to be a nonsense mutation that truncates the protein to 228 amino acids (Flybase). Using an antibody generated against amino acids 8-383 (Stern et al.,2009; kind gift of Y. Nibu), we do not detect CtBP protein in CtBP87De-10 homozygous clones, suggesting that the truncated protein is not stable or that the precise epitope of this antibody lies between amino acids 228-383. This, together with the fact that the eyFLP; CtBP87De-10/M phenotype resembles the CtBP87De-10/Df(3L)ry615 phenotype, suggests that CtBP87De-10 is a strong loss-of-function allele. CtBP03463 has a P-element insertion located 5′ to the CtBP coding region (Poortinga et al.,1998).

P[GSV]A396 and P[GSV]A132 are bidirectional EP elements located in the first intron of CtBP (Toba et al.,1999; Fang et al.,2006) and were kind gifts from K. Cadigan.

Flies expressing FLAG-tagged Eyeless were a kind gift from R. Chen. The FLAG tag is a C-terminal fusion incorporated into a genomic construct containing the entire eyeless gene. The construct is inserted at chromosome region 68A, and rescues an eyeless mutant (R. Chen, personal communication).

GST Pulldowns

GST fusion proteins for Dan and Danr have been previously described (Curtiss and Mlodzik,2000); GST-CtBP (Nibu et al.,1998b) and GST-SixHD (Kenyon et al.,2005) were kind gifts from J. Posakony and F. Pignoni, respectively. Biotinylated in vitro translated proteins were generated using the TnT Quick-Coupled in vitro Transcription/Translation kit and Transcend Biotinylated tRNA (Promega). GST Pulldowns were performed using the MagnetGST Pull-Down System (Promega) according to the manufacturers protocol. Samples were analyzed by Western blotting, probing with HRP-conjugated streptavidin (1:10,000, Jackson Immunoresearch). Chemiluminescent detection was performed and analyzed using a ChemiDoc XRX imaging system and Quantity One software (Bio-Rad).

Western Blots and Immunoprecipitation

Western blots were performed as previously described (O'Keefe et al.,2009). Membranes were probed with rat anti-Dan (1:300), rat anti-Danr (1:500), mouse anti-Tubulin (1:500, Developmental Studies Hybridoma Bank [DSHB]), rabbit anti-Ey (1:100, gift of U. Walldorf), rabbit anti-CtBP (1:1,500, gift of D. Arnosti), or mouse anti-Dac (1:200, DSHB), washed and probed with horseradish peroxidase (HRP)-conjugated secondary antibodies. Chemiluminescent detection and analysis was performed as described above.

Immunoprecipitations were performed by dissecting eye-antennal disk/brain complexes from 200 wild-type third instar larvae. Eye-antennal disc/brain complexes were lysed in 500 μl RIPA buffer containing protease inhibitors for 10 min with vigorous shaking at room temperature. Lysates were cleared by centrifugation at 10,000 rpm for 5 min at 4°C, and the supernatant was equally divided into two tubes. Anti-Dan, anti-CtBP (Stern et al.,2009; kind gift of Y. Nibu), or anti-Tubulin (DSHB; 2 μg for 300 μL of lysate) was added and incubated for 2 hr at 4°C. 40 μL of Protein G agarose beads (Invitrogen) was added and incubated for an additional 2 hr at 4°C. The beads were washed three times, shaking vigorously at room temperature in wash buffer (4.2 mM Na2HPO4, 2 mM KH2PO4, 140 mM NaCl, 10 mM KCl). Beads were then resuspended in 40 μl 1× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer (Bio-Rad) and boiled 5 min, and 20 μl of each sample was analyzed by Western blotting as described above.

For the anti-FLAG immunoprecipitations, 200 eye-antenna disc/brain complexes were lysed as above. The lysate volume was increased to 1 ml by the addition of 500 μl lysis buffer (50 mM Tris pH7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100). The lysate was then divided equally. Forty microliter anti-FLAG M2 beads (Sigma) were added to one aliquot, while 40 μl protein G beads were added to the other. Both tubes were incubated overnight at 4°C, and then processed as described above.

Immunohistochemistry

Rabbit anti-Phospho-Histone H3 (1:400, Cell Signaling Technologies) stainings were performed according to the manufacturers protocol. Other primary antibodies used were guinea pig anti-Atonal (1:2,000, gift from D. Marenda), guinea pig anti-Senseless (1:1,000, gift from G. Mardon), and rat anti-Elav (1:100, Developmental Studies Hybridoma Bank). Fluorescent secondary antibodies (Jackson Immunoresearch) were diluted 1:200. Images were obtained with a Zeiss Axioimager Z3 fluorescent microscope with Apotome and analyzed using Zeiss Axiovision, ImageJ and Adobe Photoshop software. Clonal and nonclonal areas were measured using Axiovision software.

Analysis of the Adult Phenotypes

For the analysis of the adult eye and vertex phenotypes, adult flies were dissected out of the pupal case (where necessary), frozen 20 min at −80°C and then imaged with a Hitachi TM-1000 Tabletop Scanning Electron Microscope. For the analysis of the adult antenna phenotypes, antennae were dissected in PBS, transferred to methanol for 20 min, then mounted in 80% glycerol. Images were obtained and analyzed as described above.

Semi-thin sections of adult eyes were generated as previously described (O'Keefe et al.,2009).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Ken Cadigan, Graeme Mardon, Daniel Marenda, Yutaka Nibu, Francesca Pignoni, James Posakony, Uwe Walldorf, the Bloomington Drosophila Stock Center, and the Developmental Studies Hybridoma Bank for reagents. We especially thank Rui Chen for sharing reagents before publication. Thanks also to Kathryn Hanley for continuing patient help with statistics, and to Peter Cooke in the NMSU Electron Microscopy Laboratory for help with SEM. Thanks to Andreas Jenny and to Richard Cripps and to anonymous reviewers for insightful comments on the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES