PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Cell. Author manuscript; available in PMC Jun 10, 2012.
Published in final edited form as:
PMCID: PMC3117217
NIHMSID: NIHMS297789
Interlocked feedforward loops control cell type-specific Rhodopsin expression in the Drosophila eye
Robert J. Johnston, Jr,&# Yoshiaki Otake,&* Pranidhi Sood,# Nina Vogt,# Rudy Behnia,# Daniel Vasiliauskas,# Elizabeth McDonald,% Baotong Xie,% Sebastian Koenig,^ Reinhard Wolf,^ Tiffany Cook,% Brian Gebelein,% Edo Kussell,# Hideki Nagakoshi,+* and Claude Desplan#+
# Department of Biology, New York University, 100 Washington Square East, New York NY 10003, USA
* Graduate School of Natural Science and Technology, Okayama University 3-1-1, Tsushima-naka, Okayama, Okayama 700-8530, Japan
% Division of Developmental Biology, Cincinnati Children’s Hospital, 3333 Burnet Ave, MLC 7007, Cincinnati, OH 45229
^ Rudolf-Virchow-Zentrum, DFG Research Center for Experimental Biomedicine, University of Wuerzburg, Josef-Schneider-Str. 2, 97080 Wuerzburg, Germany
+ Corresponding authors: goshi/at/cc.okayama-u.ac.jp, cd38/at/nyu.edu
&Equal contributors
How complex networks of activators and repressors lead to exquisitely specific cell type determination during development is poorly understood. In the Drosophila eye, expression patterns of Rhodopsins define at least eight functionally distinct though related subtypes of photoreceptors. Here, we describe a role for the transcription factor gene defective proventriculus (dve) as a critical node in the network regulating Rhodopsin expression. dve is a shared component of two opposing, interlocked feedforward loops (FFLs). Orthodenticle and Dve interact in an incoherent FFL to repress Rhodopsin expression throughout the eye. In the R7 and R8 photoreceptors, a coherent FFL relieves repression by Dve while activating Rhodopsin expression. Therefore, this network uses repression to restrict, and combinatorial activation to induce cell type-specific expression. Further, Dve levels are finely tuned to yield cell type- and region-specific repression or activation outcomes. This interlocked FFL motif may be a general mechanism to control terminal cell fate specification.
Development of an organism requires regulatory networks that induce gene activation or repression with highly reproducible cellular specificity. In sensory systems, genetic programs produce distinct expression patterns of sensory receptors that dictate an organism’s capacity to interpret environmental cues.
The Drosophila eye is composed of ~800 ommatidia, or unit eyes (Figure 1A–B). Each ommatidium contains two main classes of photoreceptors (PRs), outer PRs (R1–6) and inner PRs (R7 and R8)(Figure 1A–G)(Wolff and Ready, 1993). The six outer PRs express the Rhodopsin1 (Rh1) protein and are used for motion detection (Hardie, 1985). The inner PRs are specialized for color vision (Gao et al., 2008; Yamaguchi et al., 2010). The Spalt transcription factors (Salm and Salr; referred to collectively as “Sal”) are critical for the induction of inner PR fate (Mollereau et al., 2001). The inner PR class is further differentiated into the R7 fate by Prospero (Pros) and Nf-yc, or the R8 fate by Senseless (Sens) (Cook et al., 2003; Morey et al., 2008; Xie et al., 2007).
Figure 1
Figure 1
The Dve K50-homeodomain transcription factor is expressed in yR7s and outer PRs
In contrast to the outer PRs that all express Rh1, the inner PRs display differential expression patterns of several Rhodopsins (Rhs). These Rh expression patterns determine two main ommatidial subtypes, designated ‘pale’ (p) and ‘yellow’ (y). In the p subtype, R7 expresses the UV-sensitive Rhodopsin3 (Rh3) and R8 expresses the blue-sensitive Rhodopsin5 (Rh5) (Figure 1H, 1J–K). In the y subtype, R7 expresses another UV-sensitive Rhodopsin4 (Rh4) and R8 expresses the green-sensitive Rhodopsin6 (Rh6) (Figure 1I–K)(For review, (Johnston Jr and Desplan, 2010)). These ommatidial subtypes are spatially randomized and occur at a p:y ratio of 35:65 (Franceschini et al., 1981). Subtype specification occurs in a three step process: 1. Stochastic expression of the PAS-bHLH transcription factor Spineless (Ss) in a subset of R7s determines yR7 fate (Wernet et al., 2006); 2. In the absence of Ss, R7s are specified as pR7s and signal p fate to R8, inducing Rh5; yR8 fate (i.e. Rh6 expression) is specified by default and is associated with yR7 (Chou et al., 1999); 3. R8 subtype fates are stabilized by a bistable feedback loop of the tumor suppressor Warts (Wts) and the growth regulator Melted (Melt)(Figure 1H–I)(Mikeladze-Dvali et al., 2005).
There is a breakdown of the rules for exclusive Rh expression in the dorsal third of the retina (hereafter, “dorsal third”). In this area, the transcription factors encoded by the iroquois-complex (iro-C) induce Rh3 expression in all R7s, leading to co-expression of Rh3 with Rh4 in yR7s (Figure 6A–B, 6D). yR7 subtype identity is, however, maintained as indicated by the expression of Rh4 as well as coordinate expression of Rh6 in R8 (Mazzoni et al., 2008). This co-expression of Rh3 and Rh4 broadens the sensitivity to UV wavelengths and may enable the fly to discriminate between the “solar” and “antisolar” halves of the sky, which is required for navigation in other insects (Rossel and Wehner, 1982).
Figure 6
Figure 6
Thresholded Dve levels control IroC-mediated activation of Rh3 expression
In addition to these specific inductive mechanisms, the K50 homeoprotein Orthodenticle (Otd) is expressed in all PRs and regulates Rh expression (Vandendries et al., 1996). In otd mutants, Rh3 and Rh5 expression is lost whereas Rh6 is de-repressed in outer PRs. Otd protein can bind canonical K50 binding sites found in the rh3, rh5, and rh6 promoters (Tahayato et al., 2003; Treisman et al., 1989). However, since Otd is expressed in all PRs, it is unclear how it regulates these genes in specific and differential ways.
Here, we identify a role for the divergent K50 homeobox gene, defective proventriculus (dve), in Rh regulation. In dve mutants, rh genes are de-repressed with cell type-specific differences in expressivity, i.e. the variation of mutant phenotypes between individuals. dve mutants display de-repression of Rh3 in all yR7s (complete expressivity) and de-repression of Rh3, Rh5, and Rh6 in random subsets of outer PRs (incomplete expressivity). We exploit these differences in expressivity to delineate network motifs that control Rh expression.
Our analysis reveals that dve is a shared component of two interlocked feedforward loops (FFLs). In an incoherent FFL, Otd interacts with Dve to yield repression of Rh3, Rh5, and Rh6 in outer PRs. In an opposing coherent FFL, Sal both represses Dve and activates Rh3 expression in R7s. Rh3, Rh5, and Rh6 are further restricted to specific PR subtypes by Sens, Pros, Ss (via Dve), and the Wts/Melt loop. We also show that Dve is expressed in yR7s at levels that are sufficient to repress Rh3 throughout most of the retina but low enough to allow for IroC-mediated activation of Rh3 in the dorsal third. The network elucidated here represents a paradigm for the complex, multilayered combinatorial control required for cell type-specific gene expression.
dve mutants display defects in low intensity light discrimination
The dve gene codes for a K50 homeoprotein involved in midgut development, wing generation, and joint formation. Mutations in the dve locus also exhibit defects in object fixation behavior, suggesting a role in visual function (Nakagoshi et al., 1998; Nakagoshi et al., 2002; Shirai et al., 2007).
To characterize the role of dve in the eye, we tested dve mutants for defects in their response to light. We examined two dve alleles using the ey-flp/FRT GMRhid technique to generate retinas composed exclusively of mutant tissue (Stowers and Schwarz, 1999). The dve186 molecular null allele is a deficiency that removes dve and 14 other genes (Terriente et al., 2008) whereas the dve1 hypomorphic allele is an intronic P-element insertion that causes a strong loss-of-function (Nakagoshi et al., 1998).
We conducted electroretinograms on wild type and dve mutant flies. Both dve alleles displayed response defects to light (Figure S1C–D). Next, we assessed the capacity of dve mutant flies to detect motion (Yamaguchi et al., 2008). In this behavioral assay, moving lines of different color and intensity are displayed to the fly. At the point of equiluminance, the fly ceases walking since it does not perceive motion resulting in a V-shaped response curve. dve mutants displayed a shallower, U-shaped response curve indicating defects in the ability to perceive more subtle differences in light contrast (Figure S1E–G).
Dve is expressed in outer PRs and yR7s
We analyzed Dve protein expression in the eye. Although Dve was first observed in PRs starting after the morphogenetic furrow, non-specific staining prevented clear identification of cells expressing Dve until ~50% pupation (data not shown). At that point and all later time points, strong Dve expression was observed in all outer PRs (Figure 1L–M, 5A–B) but was absent in R8 cells. However, weaker Dve expression was observed in a subset of R7s. In the adult eye, Dve was co-expressed in yR7s with Rh4 and Ss but not in pR7s with Rh3 (Figure 1L–M, 5A–B, S5B). Dve expression was also observed in the outer PRs of the mosquitoes Anopheles gambiae and Aedes aegypti, suggesting that it plays a conserved role in Rh regulation (Figure S1A–B).
Figure 5
Figure 5
Ss activates Dve to repress Rh3 in yR7s
Rh3 is de-repressed in all yR7s in dve mutants
We examined dve mutants for abnormal expression of Rhodopsins. Consistent with the expression of Dve in yR7, both dve186 and dve1 mutants displayed de-repression of Rh3 in all yR7s (defined by expression of Rh4) (Figure 2B, 2D, S5A, Table S1, a summary of all genotypes and expression patterns in this paper). dve mutant retinas are thus composed of pR7s expressing Rh3 and yR7s co-expressing Rh3 and Rh4. The ratio of pR7s to yR7s in dve mutants was the same as in wild type, suggesting that dve does not control subtype fate in general but rather specifically represses Rh3 in yR7s (Figure 2D, S5A). dve mutant eye tissue showed no overt morphological phenotypes and Rh5 or Rh6 were still exclusively expressed in subsets of R8 cells (data not shown).
Figure 2
Figure 2
Dve represses Rh3 expression in yR7s and Rh3, Rh5, and Rh6 in outer PRs
To test the cellular autonomy of dve function, we generated eyes containing clones of homozygous dve mutant tissue in a dve/+ background. In the main part of the eye, wild type yR7s only expressed Rh4 and never Rh3, whereas dve mutant yR7s always co-expressed Rh3 with Rh4 (Figure S2A). This suggests that dve is required autonomously in yR7s to repress Rh3.
Considering the role of Dve in repressing Rh3 in yR7s, we next tested if it was sufficient to repress Rh3 in pR7s. When Dve was expressed in all R7s (panR7>Dve), Rh3 was lost (Figure 2C, S5A).
The dve mutant phenotype in yR7s occurs with complete expressivity (i.e. Rh3 is expressed in all yR7s). This phenotype is in stark contrast with the incompletely expressive phenotype observed in outer PRs described below.
Rh3, Rh5, and Rh6 are de-repressed in random subsets of outer PRs in dve mutants
Expression of Rh3, Rh4, Rh5, and Rh6 is never observed in outer PRs in wild type animals (number of retinas>15, number of outer PRs>10,000; Figures 1J–K, 4C, 4G). Since Dve is strongly expressed in outer PRs, we examined Rh expression in dve mutant outer PRs and observed de-repression of Rh3, Rh5, and Rh6. In contrast to the de-repression of Rh3 in all yR7s, de-repression of Rh3, Rh5, and Rh6 occurred in random subsets of outer PRs with variable cellular specificity, timing, and levels (Figure 2E–M, S2E–F). Expression of Rh4 in yR7s and Rh1 in outer PRs was not affected (Figure 3B, Figure S2K).
Figure 4
Figure 4
Sal and Dve interact in a coherent feedforward loop to control Rh3 expression
Figure 3
Figure 3
Otd and Dve interact in an incoherent feedforward loop to control Rh expression
We developed custom software to identify individual cells within each ommatidium based on Phalloidin staining of the actin rich rhabdomere structures (Figure 2J). We assessed Rh expression by analyzing the contour plot of the intensity levels for a given image (Figure 2K, L). Based on these contours, we ascribed expression as being “on” or “off” based on brightness above an ommatidium-specific threshold (P.S, R.J. and E.K, in prep.).
The random expression of Rhs in dve mutant outer PRs occurred with particular cell type, regional, and temporal biases (Figure 2I, 2M, Figure S2E–F). Rh3 expression occurred more frequently in the R1 and R6 cells (Figure S2E–F) and decreased over time (Figure 2M). Further, Rh3 was expressed more often in the dorsal half compared to the ventral half of the retina (Figure 4D, Figure S2F). We discuss below the mechanism controlling this dorsal-ventral difference in frequency. In contrast, Rh5 exhibited almost no de-repression shortly after eclosion but showed a dramatic increase at 2 weeks (Figure 2M). Rh5 de-repression showed no consistent cell type or regional biases in frequency (Figure S2E–F). Rh6 displayed little de-repression at eclosion, but increased at 2 and 4 weeks (Figure 2M). Rh6 was more frequently observed in the R3 cell, but no consistent regional bias was seen (Figure S2E–F). Together, these observations suggest that distinct cell-specific, regional, and temporal mechanisms control random Rh3, Rh5, and Rh6 expression in dve mutant outer PRs.
We next tested the cell-autonomy of dve function. De-repression of Rh3, Rh5, and Rh6 was observed in dve mutant clones but not in wild type tissue (Figure S2B–D), suggesting that Dve is required cell autonomously in outer PRs to repress Rh3, Rh5, and Rh6.
We then asked whether Dve plays a permissive or instructive role in Rh repression. We have shown above that Dve is sufficient to repress Rh3 in R7s. We tested whether Dve was sufficient to repress Rh5 and Rh6 in R8 cells. We utilized an inducible “flip out” system (GMR>/FRT/white+ transcription stop/FRT/Gal4) to drive expression of Dve in adult flies to circumvent the disruption of PRs that occurs when Dve is ectopically expressed early (data not shown). We induced hs>flp in 1 day old adult flies with a 30 minute heat shock. Six weeks after induction of ectopic Dve expression, we observed an almost complete loss of Rh5 and Rh6 expression in R8s expressing Dve (>99%; Figure 2N). These results show that Dve is sufficient to repress Rh5 and Rh6 in R8 cells.
Since the inner-PR specific Rh3, Rh5, and Rh6 are de-repressed in outer PRs, we wondered whether dve mutants exhibited altered PR fates. However, we observed no change in the expression of the cell type markers Pros (R7), Ss (yR7), Sens (R8), Sal (inner PRs), or Otd (all PRs) in dve mutants (Figure S2G–J). Furthermore, we observed no gross changes in rhabdomeric or axonal morphology of outer or inner PRs (Figure S2L). These observations suggest that Dve does not control general features of PR fate but rather specifically regulates the expression of Rh3, Rh5, and Rh6.
In wild type animals, Rh3, Rh5, and Rh6 are tightly regulated in their respective PR subtypes and are never co-expressed in the same cell. In R7, Pros excludes Rh5 and Rh6 from R7 whereas Sens excludes Rh3 from R8 (Cook et al., 2003; Xie et al., 2007). The Wts/Melt bistable loop controls exclusion between Rh5 and Rh6 in R8 subtypes (Mikeladze-Dvali et al., 2005). In dve mutants, however, all possible combinations of Rh3, Rh5 or Rh6 expression were observed in individual outer PRs (Figure 2H). This suggests that the mechanisms controlling Rh3, Rh5, and Rh6 in dve mutant outer PRs are independent for each gene and disconnected from the R7/R8 determination machinery and the Wts/Melt bistable loop.
Otd and Dve act in an incoherent feedforward loop to regulate Rh expression
Since Rh3, Rh5, and Rh6 are expressed throughout the eye in dve mutants, they might be controlled by a general activator. We tested Otd, which is expressed in all PRs, for this role. In otd mutants, Rh3 was lost in R7s, while Rh5 was lost in R8s; in contrast, Rh6 was de-repressed in outer PRs (Figure 3C, 3G)(Tahayato et al., 2003). Therefore, Otd and Dve regulate Rh6 in a similar way, but regulate Rh3 as well as Rh5 in opposite ways.
Since Otd and Dve are both K50-type homeodomain transcription factors (Figure S3A), we asked whether they bind the same cis-regulatory elements on the rh promoters. Previously, we have shown that Otd binds the rh3, rh5, and rh6 promoters via canonical K50 binding sites (TAATCC) (Figure S3B)(Tahayato et al., 2003). In gel shift assays, both Dve and Otd bound the rh3, rh5, and rh6 promoters and this binding was dependent on the K50 binding sites (Figure 3I–L).
Otd is required for dve reporter gene expression in the context of head capsule development (Carr et al., 2005). Since Otd is expressed in all PRs, we wondered if Otd is required for Dve expression in the eye. Indeed, in otd2 null mutant clones (Figure 3M) and the eye-specific otduvi mutant allele (Figure S3C), we observed complete loss of Dve expression both in outer PRs and in yR7s.
From these observations, we can formulate specific regulatory motifs in which Otd and Dve act to control Rh expression. Otd is required for Dve expression and Dve is required for repression of Rh3 and Rh5. However, Otd is also required for the activation of Rh3 and Rh5. Thus, Otd and Dve interact in an incoherent feedforward loop (iFFL) to regulate Rh3 and Rh5 (Mangan and Alon, 2003). In this iFFL, an upstream activator (Otd) upregulates a repressor (Dve), that represses target genes (Rh3 and Rh5). In parallel, the upstream activator induces the target genes (Figure 3N, ,7B).7B). In the Otd/Dve/Rh iFFL, repression by Dve completely supersedes activation by Otd yielding an end repressive outcome.
Figure 7
Figure 7
Models
Consistent with the iFFL, we observed a complete loss of Rh3 and Rh5 expression in otd; dve double mutants (Figure 3D, 3H), supporting a role for Otd as a general activator of Rh3 and Rh5. The iFFL is also supported by the observation that mutation of K50 binding sites in rh3 and rh5 reporter transgenes causes complete loss of expression in the retina (Tahayato et al., 2003).
We next confirmed the role of the iFFL using cell culture assays. In S2 cells, Otd was sufficient to induce expression from a reporter driven by either the rh3 or rh5 promoters. Dve alone had no effect on these promoters. However, co-expression of Otd and Dve led to a significant decrease in rh reporter expression, showing that Dve is sufficient to repress Otd-dependent activation of these rh genes (Figure 3O–P).
To further test the iFFL, we examined the effect of Otd and Dve on an artificial promoter containing six K50 binding sites and the heterologous tk minimal promoter (“6xK50”) (Figure S3B). Similar to the effects on the rh promoters, Dve repressed Otd-mediated expression suggesting that the K50 binding sites are sufficient to mediate regulation by the iFFL (Figure 3R).
Dve could repress expression either by preventing binding of the Otd activator (i.e. repression by exclusion), or by functionally repressing transcription (with simultaneous binding of activators and repressors). In gel shift assays, Dve and Otd could compete for binding to the rh3 promoter (Figure 3J, S3H). Dve was also sufficient to repress basal expression driven by the tk promoter (6xK50 construct) in the absence of Otd-mediated activation (Figure 3R). Thus, Dve likely represses expression by excluding binding of the obligatory Otd activator and functionally repressing transcription.
In contrast to the iFFL controlling Rh3 and Rh5, a simple linear pathway of Otd and Dve controls Rh6 expression. Both Otd and Dve are required for Rh6 repression and Otd is required for Dve expression. Further, otd;dve double mutants displayed de-repression of Rh6 in outer PRs at a similar frequency as dve and otd single mutants (frequency of outer PRs expressing Rh6 for otd=0.86+/−0.04, dve=0.81+/−0.05, otd, dve=0.85+/−0.01)(Figure 3H, ,7B7B).
Interestingly, in S2 cells, Otd was able to activate rh6 expression, and this activation was susceptible to Dve-mediated repression (Figure 3Q). Thus, ex vivo, the iFFL motif controlling Rh3 and Rh5 also regulates Rh6, but, in vivo, another factor (likely acting redundantly with Otd) induces Rh6 (Figure 7A–B).
Pph13 is required for Rh6 expression and is expressed in all PRs (Mishra et al., 2010; Zelhof et al., 2003). Thus, Pph13 is likely the general activator for Rh6. In S2 cells, Pph13 was sufficient to induce expression from a reporter driven by the rh6 promoter. Co-expression of Pph13 and Dve led to a significant decrease in rh6 reporter expression, showing that Dve is sufficient to repress Pph13-mediated activation of rh6 (Figure 3S).
Pph13 activates Rh6 independent of K50 binding sites (Mishra et al., 2010). We therefore tested whether K50 sites were required for Dve-mediated repression. Indeed, Dve was no longer able to repress Pph13-mediated activation from a rh6 promoter in which the K50 sites were mutated (Figure 3T). This result shows that K50 sites are required for Dve-mediated repression and supports the notion that Dve can functionally repress expression (with simultaneous binding of activators and repressors).
The linear pathway motif controlling Rh6 and the requirement of K50 sites for repression are also supported by the observation that mutation of K50 sites in a rh6 reporter transgene causes de-repression in outer PRs (Tahayato et al., 2003).
These results show a critical role for the Otd/Dve/Rh iFFL and linear pathway motifs in regulating Rh expression. Otd activates Rh3 and Rh5 expression in all PRs (Figure 7A). To prevent aberrant expression, Otd also activates Dve which is sufficient to repress Rh3, Rh5, and Rh6. These motifs alone should yield repression of Rh3, Rh5, and Rh6 in all PRs (Figure 7B). Other mechanisms must therefore be utilized to relieve repression by Dve to allow for Rh expression in inner PRs. Further, Rh3, Rh5, and Rh6 are only expressed in random subsets of outer PRs in dve mutants. Thus, outer PRs lack a factor(s) that is present in inner PRs and is sufficient to induce expression in all cells. Next, we discuss the mechanisms utilized to relieve repression and induce Rh expression in inner PRs.
Sal acts in a coherent feedforward loop to repress Dve and activate Rh3 expression
Otd is expressed in all PRs and activates Dve. However, Dve is expressed in outer PRs and yR7s but not in the other inner PRs (i.e. pR7s and R8s). Sal, a key determinant of inner PR fate, is expressed in a nearly complementary expression pattern, i.e. Sal is expressed in inner PRs and excluded from outer PRs in adult retinas. In sal mutants, inner PRs lose Rh3, Rh4, Rh5, and Rh6 and gain the outer PR-specific Rh1 (Figure S4A–C)(Mollereau et al., 2001). Since Dve represses Rh expression, the loss of Rh3, Rh5, and Rh6 in sal mutants might be due to de-repression of Dve in inner PRs. We thus asked whether Sal represses Dve expression in inner PRs. In sal null mutants, Dve was strongly de-repressed in all R7s and R8s, resulting in Dve expression at equal levels in all PRs (Figure 4A). Further, ectopic expression of Sal in all PRs resulted in repression of Dve in outer PRs (Figure 4B). Thus, Sal represses Dve in inner PRs, allowing Rh3, Rh5 and Rh6 to be expressed in these cells.
In dve mutants, Rh3 is expressed in random subsets of outer PRs whereas it is expressed in all R7 PRs. Thus, R7s must contain an additional factor that induces Rh3 expression in all cells. Since Sal is expressed in inner PRs but not outer PRs, we tested whether it played this role by providing Sal to outer PRs in dve mutants. To obviate possible cell fate transformations upon early Sal expression, we utilized the panR7>Gal4 driver to induce late Sal expression in outer PRs at the time of Rh expression. This driver contains a hybrid rh3/rh4 promoter that drives expression in all R7s in wild type animals but induces expression in all outer PRs in dve mutants (data not shown). In dve mutant eyes expressing Sal in all outer PRs (dve; outers>sal), we observed expression of Rh3 in nearly all outer PRs (Figure 4C–G). Sal is thus sufficient to induce Rh3 expression in all outer PRs in the absence of the Dve repressor.
We next wondered whether Sal was required to activate Rh3 in R7s. In dve mutants, Rh3 was expressed in all R7s. In sal, dve double mutant clones, Rh3 was only expressed in a subset of R7s, demonstrating that Sal is required to induce Rh3 expression in all R7s in dve mutants (Figure 4H–K).
Expression of Rh3 in R7s therefore requires both activation by Sal (and Otd) and exclusion of the repressor Dve. Since Sal not only represses Dve but also activates Rh3, Sal and Dve interact in a coherent feedforward loop (cFFL) to control Rh3. In this cFFL, an upstream regulator (Sal) prevents expression of a repressor (Dve) that can repress a target gene (Rh3). In parallel, the regulator (Sal) induces expression of the target gene (Rh3) (Figure 4F) (Mangan and Alon, 2003).
This cFFL counters the repressive Otd/Dve/Rh3 iFFL to induce Rh3 expression in all inner PRs (Figure 4L–N). The Otd/Dve/Rh3 iFFL and the Sal/Dve/Rh3 cFFL are interlocked FFLs with the Dve repressor as a shared component. We have shown four states for this interlocked FFL motif in two different cellular contexts (i.e. R7s and outer PRs). (1) In the absence of Otd (in otd mutant R7s and otd; dve mutant outer PRs), Rh3 is lost showing that Otd is an obligate activator (Figure 3C–D). (2) In the absence of Sal (in sal mutant R7s and wild type outer PRs), Otd activates Dve which represses Rh3 (Figure 4C, 4H, 4L). (3) In the absence of Sal and Dve (in sal, dve mutant R7s and dve mutant outer PRs), Otd activates Rh3 in a random subset of cells (Figure 4D, 4I, 4M). (4) In the absence of Dve and presence of Sal and Otd (in dve mutant R7s and dve mutant with ectopic expression of Sal in outer PRs), Sal and Otd induce Rh3 expression in all cells (Figure 4E, 4J, 4N).
In wild type outer PRs, the interlocked FFL motif is in state 2, yielding repression of Rh3 (Figure 4L, ,7D).7D). In R7s, the interlocked FFL motif is in state 4, yielding activation of Rh3 (Figure 4N, ,7C).7C). This motif predicts Rh3 activation by Sal and Otd in all inner PRs. Rh3 expression is limited to pR7s by repression by Sens in R8s (Xie et al., 2007) and by Dve in yR7s (described below) (Figure 7E–H).
Rh5 is activated in inner PRs
We next asked whether Sal was similarly able to control Rh5 and Rh6 expression. In dve mutant eyes expressing sal in all PRs (dve; outers>sal), we observed a random pattern of Rh5 and Rh6 expression in outer PRs (data not shown). This observation suggests that, in contrast to Rh3, Sal is not sufficient to activate Rh5 and Rh6 expression in all cells.
We thus examined the regulatory logic controlling specific Rh5 expression in pR8s. The Pros transcription factor represses Rh5 expression in R7s (Figure S4D–H) (Cook et al., 2003). We re-examined Rh5 in pros mutants and observed de-repression specifically in pR7s (i.e. Rh5 was not expressed in Rh4-expressing yR7s)(Figure S4E). Since Dve is expressed in yR7s, we tested whether it acts there to repress Rh5 in pros mutants. In dve; pros double mutant retinas, Rh5 was de-repressed into all R7 cells, including Rh4-expressing yR7s (Figure S4G–H). These data suggest that Rh5 is activated in all inner PRs (Figure 7C).
Thus, the logic controlling Rh5 is similar to that controlling Rh3, i.e. broad activation coupled with repression in specific cells to yield precise expression. Though activated by an inner PR-specific factor and Otd, Rh5 expression is limited to pR8s by repression in R7s by Pros (and Dve) and in yR8s by the Wts pathway (Figure 7E–F, 7I–J) (Cook et al., 2003; Mikeladze-Dvali et al., 2005).
Rh6 is activated in R8s
We next assessed the logic controlling Rh6 expression in yR8s. In pros mutants, Rh6 was randomly expressed in p or yR7s (Figure S4D, S4F). This phenotype suggests that Rh6 is not susceptible to repression by the low levels of Dve in yR7. Further, the random expression of Rh6 in pros mutants suggests that Rh6 is not activated in all inner PRs like Rh3 and Rh5. We have shown previously that Sens is necessary and sufficient to induce consistent Rh6 expression in R8s (Xie et al., 2007). Taken together, we conclude that Rh6 expression requires activation in all R8s by Sens. Rh6 expression is excluded from pR8s by a repressor that is active in the absence of Wts pathway activity (D. Jukam and C.D., in prep.). Rh6 expression is excluded from R7s by Pros and from outer PRs by high levels of Dve. These mechanisms yield Rh6 expression in yR8s (Figure 7D–F, 7I–J).
Spineless induces Dve to repress Rh3 in yR7s
Sal acts in inner PRs to repress Dve and activate Rh3. Rh3 is limited to pR7s via repression by Sens in R8s (Xie et al., 2007) and by low levels of Dve in yR7. We next addressed how Dve expression is specifically induced in yR7s. We have shown previously that specification of yR7 fate is dependent on the activity of the ss gene. ss is expressed at mid-pupation in ~65% of R7s, correlating with the frequency of yR7 fate in adults (Wernet et al., 2006) (Figure 5A, S5B).
Dve is expressed in yR7s at low levels and repressed in all other inner PRs due to Sal activity. We posited that Dve might be re-activated by Ss in yR7s and act as a downstream effector to repress Rh3. Indeed, Dve and Ss were always found in the same subset of R7s (Figure 5A, S5B). In ss null mutant retinas, Dve expression was lost in all R7s but maintained in outer PRs (Figure 5B–C, S5B). Ss is required autonomously in R7 cells, since dve expression was lost in ss mutant clones but not in neighboring wild type tissue (Figure 5E). Further, ectopic expression of Ss was sufficient to induce Dve expression in R7s (Figure 5D, S5B). Together, these results show that Ss activates expression of Dve in yR7s.
To address whether Dve requires Ss function to repress Rh3 expression, we expressed Dve in all R7s in ss mutants (panR7>dve; ss). The repressive function of Dve does not require Ss function since expression of Dve was sufficient to repress Rh3 in ss null mutants leading to R7s without Rh3 or Rh4 (Figure 5F–G, S5A). Together, these data show that the subtype determinant Ss activates Dve to repress Rh3 in yR7s (Figure 7G–H).
Dve expression levels are finely tuned to allow for Rh3 expression in yR7s in the dorsal third
In the dorsal third of the retina, IroC induces Rh3 in Rh4-expressing yR7s (Figure 6A–B, 6D)(Mazzoni et al., 2008). This situation resembles the dve loss-of-function phenotype in R7s (Figure 2B, ,6D),6D), suggesting that IroC genes repress Dve expression in yR7 in the dorsal third of the eye. However, IroC>lacZ labeled yR7 cells in the dorsal third still displayed Dve expression in a subset of R7s (Figure 6C). Therefore, rather than regulating Dve expression, IroC must act downstream of, or in parallel to Dve to either activate Rh3 expression or to repress Dve function (Figure 7K).
Dve is expressed in yR7 at levels that are significantly lower than in outer PRs (Figure 1L–M, 5A–B). Although these levels are sufficient to repress Rh3 in yR7s in the main part of the retina, they might allow Rh3 to be susceptible to upregulation by IroC in the dorsal third. If this were the case, increasing expression levels of Dve in R7s should prevent Rh3 co-expression with Rh4 in dorsal third yR7s. Indeed, ectopic expression of Dve at high levels (panR7>Dve) caused repression of Rh3 in yR7s in the dorsal third of the retina (Figure 6D–E, S5C). This suggests that higher levels of Dve are able to overcome the function of IroC. Maintaining low level expression of Dve in yR7s is therefore critical to allow IroC-mediated activation of Rh3 in dorsal third yR7s.
Since Dve at high levels is able to repress IroC-mediated activation of Rh3 in yR7 cells, it could play a similar role in outer PRs. IroC is expressed throughout the entire dorsal eye field before PR specification. Later, during pupal stages, IroC expression becomes more restricted to the dorsal third in all PRs. At eclosion, this expression ceases in all PRs except in dorsal third R7s. In dve mutants, random expression of Rh3 in outer PRs occurs predominantly in the dorsal part of the retina (Figure S2E–F) and we hypothesized that this is due to the activity of IroC. To test this, we generated IroC null mutant clones in dve whole mutant eyes. We observed a strong decrease in the frequency of dorsal Rh3 expression in IroC null mutant tissue such that both ventral and dorsal regions of the eye now expressed Rh3 at low frequency (Figure 6F). This effect was specific to Rh3 since Rh5 was unaffected under the same conditions (data not shown).
Since IroC is expressed in all outer PRs early in development but ceases by the time of Rh expression, IroC may not activate Rh3 directly. IroC may induce chromatin modifications at the Rh3 locus or the expression of another transcription factor that persists to activate Rh3 expression later.
At least two mechanisms contribute to the random activation of Rh3 in outer PRs in dve mutants: 1. Otd activates expression at low frequency throughout the retina and 2. IroC increases the frequency of expression in the dorsal part. For both cases, Dve at high levels ensures that Rh3 is not expressed in outer PRs in wild type animals.
Our studies have revealed that the Dve transcription factor acts as a broad Rh regulator repressing Rh3, Rh5, and Rh6. In dve mutants, Rh3 is de-repressed in all R7s whereas Rh3, Rh5, and Rh6 are de-repressed in random subsets of outer PRs. These differences in phenotypic expressivity result from the distinct combinations of regulators that occur in specific cellular contexts. In outer PRs, the general activator Otd induces expression of Rh3 and Rh5 (and possibly Rh6) in random subsets of cells in dve mutants. In R7s, Sal acts with Otd to induce expression of Rh3 in all cells in dve mutants. Interestingly, Otd and Sal not only regulate Rhs but also Dve. Together, these genes interact in interlocked feedforward loops that are required for wild type gene expression. Next, we discuss the regulatory motifs controlling Rh expression, particularly Rh3.
Interlocked feedforward loops control Rh expression
Though Otd activates the expression of Rh3, it also induces expression of a negative regulator Dve, forming an iFFL motif (Figure 7B). iFFL motifs are observed in many biological contexts and have been implicated in temporal pulse responses, response acceleration, dose-dependent biphasic responses, and fold-change detection (Alon, 2007; Goentoro et al., 2009; Kaplan et al., 2008; Kim et al., 2008; Mangan and Alon, 2003). Since iFFLs respond to fold-change differences in inputs, the repressor (Dve) may also act to normalize noise in levels of the input (Otd) (Goentoro et al., 2009). In other words, variation in Otd input levels would induce comparable changes in Dve repressor levels, preventing aberrant activation of Rh3. Thus, in dve mutants, variation in Otd input levels would go unchecked, yielding random expression of Rh3 output.
When the Otd/Dve/Rh3 iFFL is integrated with the Sal/Dve/Rh3 cFFL, a combinatorial binary switch for gene regulation is established. The Otd/Dve/Rh3 iFFL provides repression throughout the eye due to the activity of Dve. Specific expression of Sal in inner PRs triggers the Sal/Dve/Rh3 cFFL leading to repression of Dve and activation of Rh3 (i.e. with Sal acting together with Otd). Thus, Dve serves as a common node that integrates the two loops, forming a motif of interlocked feedforward loops that ensures Rh3 expression in the presence of Otd and Sal and in the absence of Dve (Figure 4N, ,7C).7C). In the absence of Sal as in outer PRs, this motif induces high Dve levels to repress aberrant Rh expression, even in the presence of activators such as Otd and IroC (Figure 4L, ,7D7D).
Though the network logic is slightly different, cell type-specific expression of Rh5 and Rh6 requires both inhibition of Dve-mediated repression and combinatorial activation provided by broad and specific activators similar to Rh3 regulation (Figure 7).
Dve levels are finely tuned to control Rh3 expression
In the outer PRs, Dve at high levels acts as a buffer against aberrant expression activated by Otd in the ventral part or Otd with IroC in the dorsal part. In contrast, Dve levels appear to be finely tuned in yR7s to allow for wild type Rh3 expression. In yR7s, Dve at low levels is sufficient to repress Rh3 despite the presence of the activators Sal and Otd. However, in the dorsal third of the eye, IroC together with Sal and Otd override Dve at low levels to induce Rh3 expression. Increasing Dve levels in R7s is sufficient to repress Rh3 in the dorsal third containing IroC, Sal, and Otd. Thus, Dve levels are finely tuned to repress Rh3 in yR7s throughout the majority of the retina, yet allow for expression in yR7s in the dorsal third region.
IroC acts to induce Rh3 expression in all yR7s of the dorsal third. However, in dve mutant outer PRs, IroC does not induce expression in all cells but rather only increases the frequency of random Rh3 expression. It appears that Sal in R7s is critical to control this difference in effect. IroC with Sal and Otd induce Rh3 expression despite low levels of Dve in all dorsal third yR7s. However, in dve mutant outer PRs, IroC and Otd without Sal yield random expression. We have thus elucidated several combinations of regulatory factors that yield Rh expression in all cells, random subsets, or no cells in particular cellular contexts (Table S1).
High levels of Dve are used in outer PRs to ensure repression of Rh expression. It appears that this role is likely conserved in other insect species, since Dve is expressed in outer PRs in two mosquito species tested. Interestingly, these mosquito species display regionalized expression of Rhs in inner PRs in contrast to the stochastic expression pattern observed throughout the fly retina (Hu et al., 2009). This suggests that mosquitoes may use a different mechanism to determine inner PR Rh expression compared to flies, but likely use the same mechanism (i.e. Dve) to prevent inner PR-specific Rh expression in outer PRs.
Global activation allows for evolution of novel expression patterns
The fly eye contains 13 known types of PRs based on Rh expression, morphology, and positioning including 6 types of outer PRs (R1–6), 4 types of R7s (pR7, yR7, dorsal yR7, DRA R7) and 3 types of R8s (pR8, yR8, DRA R8). Interestingly, these cell types are all competent to express Rh3, Rh5, and Rh6, suggesting the presence of an activator that is expressed in all PRs. This activator is Otd for Rh3 and Rh5.
Why would Rh expression be activated throughout the eye? One possibility is that global activation allows for rapid changes in Rh expression through evolution. Evidence for rapid changes in Rh expression patterns is observed in flies, mosquitoes, butterflies, and bees. Each genus or family has a characteristic number and arrangement of PRs. However, individual species have diversified their Rh expression patterns, likely adapting specific visual capacity to their particular environments (Briscoe, 2008; Hornstein et al., 2000; Hu et al., 2009; Wakakuwa et al., 2005).
A general repressor such as Dve may facilitate the generation of cryptic regulatory interactions. In flies, repression is utilized to exclude Rh expression from specific cell types (ex: Dve in outer PRs). Underlying this repression, new cryptic genetic interactions can arise that are only revealed when the repressor itself is absent. Thus, evolution could occur via a two step mechanism: (1) novel cryptic genetic interactions arise that are masked by repression, and (2) the repressor is removed in specific PRs, allowing for new Rh expression patterns. In this model, dve mutants reveal cryptic regulatory interactions that could yield new expression patterns if Dve-mediated repression was alleviated.
Alternatively, Dve-mediated repression of Rh3, Rh5, and Rh6 may have enabled the break down of regulatory interactions in outer PRs. Other insects use multiple combinations of spectrally restricted Rhs (orthologs of Rh3, Rh5 and Rh6) to achieve broad spectrum light detection in their equivalent of outer PRs (Briscoe, 2008). This broad expression supports the notion that Rh3, Rh5, and Rh6 are activated throughout the retina. Thus, the general repression of these Rhs in outer PRs by Dve in flies may be a recent change. Upon this change, specific mechanisms activating Rhodopsins in outer PRs would be rendered ineffective due to repression by Dve. We have shown here that Dve at high levels can repress Rh expression despite the presence of several activators (i.e. Otd, Sal, IroC). Thus, mechanisms that previously induced Rh expression could break down under the mask of Dve-mediated repression. In this model, random expression in dve mutants is the result of the remnants of previous regulatory interactions.
The increasing catalog of Rh expression patterns in insect species combined with a comprehensive study of the known regulatory factors should discern between these models.
Drosophila strains and genotypes
Flies were raised on standard corn meal-molasses-agar medium and grown at 25C. For simplicity, genotypes are simplified throughout the text and figures. See Table S2 for complete genotypes.
Antibody staining
Retinas were dissected, fixed, washed, incubated with primary and then secondary antibodies and mounted for visualization. For detailed protocol and complete list of antibodies used, see Supplemental Experimental Procedures.
Quantification of expression
Using cell specific markers, antibody staining frequency was assessed (n=number of retinas scored; N=total number of cells quantified). For dve mutant outer PRs, Rh expression was scored for 3 retinas with a minimum of 75 ommatidia (450 total outer PR cells) per retina. Scoring was conducted using custom software or manually. The scoring and statistics methods are described in the Supplemental Experimental Procedures.
Behavioral analysis of outer PR function
Individual flies walked on a styrofoam ball floating in an air stream. Rotations of the ball served as a measure of the optomotor response. The visual stimulus consisted of a moving grating of vertical blue and green stripes. The intensity of the blue stripes was kept constant, while the intensity of the green stripes varied. This setup allows detection of strong defects in motion detection or a shift in the point of isoluminance between two colors. For more details, see Supplemental Experimental Procedures.
Electroretinograms
Electroretinograms were conducted as in (Hotta and Benzer, 1969).
Luciferase Assays
The rh promoter-based luciferase reporter assays were described previously (Xie et al., 2007). Additional constructs and methodology are described in the Supplemental Experimental Procedures.
Electrophoretic Mobility Shift Assays
Binding assays were performed as previously described (Li-Kroeger et at Dev Cell 2008). Protocol, proteins, and probes are described in the Supplemental Experimental Procedures.
01
Figure S1. Dve is expressed in outer PRs in mosquitoes; dve mutants display defects in light response and the perception of subtle light contrast, related to Figure 1
For A–B, Yellow arrowheads indicate R7s. Yellow arrows indicate R8s. White arrowheads indicate outer PRs. * indicates non-specific, non-nuclear staining.
A. Dve is expressed in outer PRs but not inner R7 and R8 PRs in Aedes aegypti mosquitoes. R7s were identified based on the unique shape of their nuclei. R8s were identified based on relative position to R7s (i.e. with one outer PR nuclei between, as in Drosophila) and their slight displacement from the outer PR nuclei in the Z-axis (Hu et al., 2009).
B. Dve is expressed in outer PRs but not inner R7 and R8 PRs in Anopheles gambiae mosquitoes. We concluded that R8 does not express Dve since there is no discernible signal above background in the nucleus of this cell. R7s were identified based on the unique shape of their nuclei. R8s were identified based on their central position in the ommatidium (Hu et al., 2009).
C. dve1 mutants at 4 weeks (blue traces) display a decreased response to white light compared to wild type animals (black traces)
D. dve186 mutants at 4 weeks (red traces) display a decreased response to white light compared to wild type animals (black traces)
E. Wild type animals at 0 weeks display a sharp V-shaped optomotor response curve at the point of equiluminance. For datapoints 3–8, the values are significantly different from each other, consistent with the V-shape (ANOVA, P value = 0.0234).
F. dve1 mutants at 0 weeks display a shallow trough U-shaped optomotor response curve at the point of equiluminance. For datapoints 3–8, the values are not significantly different from each other, consistent with the U-shape (ANOVA, P value = 0.2836).
G. dve186 mutants at 0 weeks display a shallow trough U-shaped optomotor response curve at the point of equiluminance. For datapoints 3–8, the values are not significantly different from each other, consistent with the U-shape (ANOVA, P value = 0.4424).
Figure S2. The dve mutant phenotypes are cell-autonomous; Quantification of outer PRs phenotypes; General PR fate is not disrupted in dve mutants, related to Figure 2
For A–D, GFP indicates wild type tissue and non-GFP indicates dve mutant tissue.
A. De-repression of Rh3 in yR7s is only observed in dve mutant tissue. Panel 1, triple stain; Panel 2, Rh3 and Rh4; Panel 3, Rh3 alone; and Panel 4, Rh4 alone. The dotted line indicates the clone boundary.
B–D. Random de-repression of Rh3 (B.), Rh5 (C.), and Rh6 (D.) in outer PRs is only observed in dve mutant tissue. In C., the arrow indicates a single mutant cell clone that expresses Rh5.
E. Frequency of random Rh expression. For all ommatidia scored, cells exhibiting a statistically significant increase in the frequency of expression compared to all cells are indicated in green. Cells exhibiting a statistically significant decrease in the frequency of expression compared to all cells are indicated in red.
F. Frequency of random Rh expression in the dorsal half compared to the ventral half. Cells exhibiting a statistically significant increase in the frequency of expression in the dorsal half compared to ventral half are indicated in green. Cells exhibiting a statistically significant decrease in the frequency of expression in the dorsal half compared to the ventral half are indicated in red.
For G–L, left panel indicates wild type; right panel indicates dve186 mutant.
G. Ss is expressed in a subset of R7s in wild type and dve mutant adult retinas. Pros is expressed in all R7s in wild type and dve mutant adult retinas.
H. Sal is expressed in all R7s and R8s in wild type and dve mutant retinas at 50% pupation.
I. Sens is expressed in all R8s in wild type and dve mutant retinas at 50% pupation.
J. Otd is expressed in all PRs in wild type and dve mutant retinas at 50% pupation.
K. Rh1 is expressed in all outer PRs in wild type and dve mutant adult retinas.
L. The axons of outer PRs project to the lamina in wild type and dve mutant adult retinas (as marked by Rh1>lacZ). The axons of inner PRs project to the medulla in wild type and dve mutant retinas (as marked by 24B10 antibody staining).
Figure S3. Otd and Dve interact in an iFFL to control Rh expression, related to Figure 3
A. Schematics of the Dve and Otd proteins used in the gel shifts in Figure 3.
The yellow box labeled HD indicates the K50-homeodomain.
B. Schematics of rh and 6xK50 promoters used in the gel shifts and cell culture experiments in Figure 3 and S3.
C. Dve expression is lost in otduvi mutants at 50% pupation.
For D–F, binding of Dve-s is dependent on K50 binding sites. WT=wild type promoter; M=mutation of K50 site. The number of bands is reduced for binding of the single homeodomain Dve-s protein compared to the two homeodomain Dve protein (see Figure 3). These data suggest that the presence of two homeodomains in the Dve protein yields differently ordered DNA/protein interactions.
D. rh3 promoter
E. rh5 promoter
F. rh6 promoter
G. Binding of Otd-s to the rh3 promoter is dependent on K50 binding sites. Arrow indicates the band shifted upon Otd-s binding. WT=wild type promoter; M=mutation of K50 site.
H. Otd competes with Dve for binding to the rh3 promoter.
Additionally, deletion of K50 sites in the rh3 promoter ablates Otd- and Dve- mediated regulation of this promoter in S2 cells (data not shown).
Figure S4. Rh expression in sal and pros mutants, related to Figure 4
A. Rh1 is expressed in all PRs in sal mutants.
B. Rh5 and Rh6 expression is lost in sal mutants.
C. Rh3 and Rh4 expression is lost in sal mutants.
For D–H, images are from 4 week old adults.
D. In pros mutants, Rh5 and Rh6 are expressed in R7s.
E. In pros mutants, Rh5 is expressed in pR7s and not in yR7s (marked by Rh4).
F. In pros mutants, Rh6 is expressed in random pR7s and yR7s (marked by Rh4).
G. In pros; dve double mutants, Rh5 is expressed in all R7s whereas Rh6 is expressed in random pR7s and yR7s.
H. In pros; dve double mutants, Rh5 is expressed in all R7s including yR7s (marked by Rh4).
Figure S5. Quantification of R7 phenotypes, related to Figure 5
A. Frequency of Rh expression in R7s
B. Frequency of Dve (and Ss for right column) expression in R7s
C. Frequency of Rh expression in yR7s in the dorsal third
Table S1. Summary of the regulatory mechanisms controlling Rh expression, related to Figure 7
Acknowledgments
We are very grateful to S. Britt, Y.N. Jan, C. Zuker, and the Bloomington Stock Center for reagents. We thank U. Alon, R. Carthew, O. Hobert, P. O’Farrell, F. Payre, and E. Siggia for helpful comments on the manuscript. We thank all members of the Desplan laboratory for suggestions throughout the project. R.J. was supported by a Jane Coffin Childs Memorial Fund for Medical Research post-doctoral fellowship. N.V. was supported by Deutsche Forschungsgemeinschaft (DFG). R.B. was supported by the European Molecular Biology Organization (EMBO) and Human Frontier Science Program (HFSP). D.V. was supported by the NIH 5F32EY016309-03. B.X. was supported by the University of Cincinnati Research Council. T.C. was supported by NIH T32 HD046387 (ECM), Research to Prevent Blindness, and the Ziegler Foundation for the Blind. B.G. was supported by an NIH grant (GM079428). E.K. was supported by a Burroughs Wellcome Fund Career Award at the Scientific Interface. This work was funded by a grant from the National Eye Institute/NIH R01 13010 to C.D. and conducted in a facility constructed with the support of a Research Facilities Improvement Grant from the NCRR, NIH.
Footnotes
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
  • Alon U. Network motifs: theory and experimental approaches. Nat Rev Genet. 2007;8:450–461. [PubMed]
  • Briscoe AD. Reconstructing the ancestral butterfly eye: focus on the opsins. J Exp Biol. 2008;211:1805–1813. [PubMed]
  • Carr M, Hurley I, Fowler K, Pomiankowski A, Smith HK. Expression of defective proventriculus during head capsule development is conserved in Drosophila and stalk-eyed flies (Diopsidae) Dev Genes Evol. 2005;215:402–409. [PubMed]
  • Chou WH, Huber A, Bentrop J, Schulz S, Schwab K, Chadwell LV, Paulsen R, Britt SG. Patterning of the R7 and R8 photoreceptor cells of Drosophila: evidence for induced and default cell-fate specification. Development. 1999;126:607–616. [PubMed]
  • Cook T, Pichaud F, Sonneville R, Papatsenko D, Desplan C. Distinction between color photoreceptor cell fates is controlled by Prospero in Drosophila. Dev Cell. 2003;4:853–864. [PubMed]
  • Franceschini N, Kirschfeld K, Minke B. Fluorescence of photoreceptor cells observed in vivo. Science. 1981;213:1264–1267. [PubMed]
  • Gao S, Takemura SY, Ting CY, Huang S, Lu Z, Luan H, Rister J, Thum AS, Yang M, Hong ST, et al. The neural substrate of spectral preference in Drosophila. Neuron. 2008;60:328–342. [PMC free article] [PubMed]
  • Goentoro L, Shoval O, Kirschner MW, Alon U. The incoherent feedforward loop can provide fold-change detection in gene regulation. Mol Cell. 2009;36:894–899. [PMC free article] [PubMed]
  • Hardie RC. Functional organization of the fly retina. In: Autrum H, Ottoson D, Perl ER, Schmidt RF, Shimazu H, Willis WD, editors. Progress in Sensory Physiology. Springer; Berlin: 1985. pp. 1–79.
  • Hornstein EP, O’Carroll DC, Anderson JC, Laughlin SB. Sexual dimorphism matches photoreceptor performance to behavioural requirements. Proc Biol Sci. 2000;267:2111–2117. [PMC free article] [PubMed]
  • Hotta Y, Benzer S. Abnormal electroretinograms in visual mutants of Drosophila. Nature. 1969;222:354–356. [PubMed]
  • Hu X, England JH, Lani AC, Tung JJ, Ward NJ, Adams SM, Barber KA, Whaley MA, O’Tousa JE. Patterned rhodopsin expression in R7 photoreceptors of mosquito retina: Implications for species-specific behavior. J Comp Neurol. 2009;516:334–342. [PMC free article] [PubMed]
  • Johnston RJ, Jr, Desplan C. Stochastic Mechanisms of Cell Fate Specification that Yield Random or Robust Outcomes. Annu Rev Cell Dev Biol 2010 [PMC free article] [PubMed]
  • Kaplan S, Bren A, Dekel E, Alon U. The incoherent feed-forward loop can generate non-monotonic input functions for genes. Mol Syst Biol. 2008;4:203. [PMC free article] [PubMed]
  • Kim D, Kwon YK, Cho KH. The biphasic behavior of incoherent feed-forward loops in biomolecular regulatory networks. Bioessays. 2008;30:1204–1211. [PubMed]
  • Mangan S, Alon U. Structure and function of the feed-forward loop network motif. Proc Natl Acad Sci U S A. 2003;100:11980–11985. [PubMed]
  • Mazzoni EO, Celik A, Wernet MF, Vasiliauskas D, Johnston RJ, Cook TA, Pichaud F, Desplan C. Iroquois complex genes induce co-expression of rhodopsins in Drosophila. PLoS Biol. 2008;6:e97. [PMC free article] [PubMed]
  • Mikeladze-Dvali T, Wernet MF, Pistillo D, Mazzoni EO, Teleman AA, Chen YW, Cohen S, Desplan C. The growth regulators warts/lats and melted interact in a bistable loop to specify opposite fates in Drosophila R8 photoreceptors. Cell. 2005;122:775–787. [PubMed]
  • Mishra M, Oke A, Lebel C, McDonald EC, Plummer Z, Cook TA, Zelhof AC. Pph13 and orthodenticle define a dual regulatory pathway for photoreceptor cell morphogenesis and function. Development. 2010;137:2895–2904. [PubMed]
  • Mollereau B, Dominguez M, Webel R, Colley NJ, Keung B, de Celis JF, Desplan C. Two-step process for photoreceptor formation in Drosophila. Nature. 2001;412:911–913. [PubMed]
  • Morey M, Yee SK, Herman T, Nern A, Blanco E, Zipursky SL. Coordinate control of synaptic-layer specificity and rhodopsins in photoreceptor neurons. Nature. 2008;456:795–799. [PMC free article] [PubMed]
  • Nakagoshi H, Hoshi M, Nabeshima Y, Matsuzaki F. A novel homeobox gene mediates the Dpp signal to establish functional specificity within target cells. Genes Dev. 1998;12:2724–2734. [PubMed]
  • Nakagoshi H, Shirai T, Nabeshima Y, Matsuzaki F. Refinement of wingless expression by a wingless- and notch-responsive homeodomain protein, defective proventriculus. Dev Biol. 2002;249:44–56. [PubMed]
  • Rossel S, Wehner R. The bee’s map of the e-vector pattern in the sky. Proc Natl Acad Sci U S A. 1982;79:4451–4455. [PubMed]
  • Shirai T, Yorimitsu T, Kiritooshi N, Matsuzaki F, Nakagoshi H. Notch signaling relieves the joint-suppressive activity of Defective proventriculus in the Drosophila leg. Dev Biol. 2007;312:147–156. [PubMed]
  • Stowers RS, Schwarz TL. A genetic method for generating Drosophila eyes composed exclusively of mitotic clones of a single genotype. Genetics. 1999;152:1631–1639. [PubMed]
  • Tahayato A, Sonneville R, Pichaud F, Wernet MF, Papatsenko D, Beaufils P, Cook T, Desplan C. Otd/Crx, a dual regulator for the specification of ommatidia subtypes in the Drosophila retina. Dev Cell. 2003;5:391–402. [PubMed]
  • Terriente J, Perea D, Suzanne M, Diaz-Benjumea FJ. The Drosophila gene zfh2 is required to establish proximal-distal domains in the wing disc. Dev Biol. 2008;320:102–112. [PubMed]
  • Treisman J, Gonczy P, Vashishtha M, Harris E, Desplan C. A single amino acid can determine the DNA binding specificity of homeodomain proteins. Cell. 1989;59:553–562. [PubMed]
  • Vandendries ER, Johnson D, Reinke R. orthodenticle is required for photoreceptor cell development in the Drosophila eye. Dev Biol. 1996;173:243–255. [PubMed]
  • Wakakuwa M, Kurasawa M, Giurfa M, Arikawa K. Spectral heterogeneity of honeybee ommatidia. Naturwissenschaften. 2005;92:464–467. [PubMed]
  • Wernet MF, Mazzoni EO, Celik A, Duncan DM, Duncan I, Desplan C. Stochastic spineless expression creates the retinal mosaic for colour vision. Nature. 2006;440:174–180. [PubMed]
  • Wolff T, Ready DF. Pattern formation in the Drosophila retina. In: Bate M, Martinez-Arias A, editors. The Development of Drosophila melanogaster. Cold Spring Harbor Laboratory Press; Cold Spring Harbor, NY: 1993. pp. 1277–1325.
  • Xie B, Charlton-Perkins M, McDonald E, Gebelein B, Cook T. Senseless functions as a molecular switch for color photoreceptor differentiation in Drosophila. Development. 2007;134:4243–4253. [PubMed]
  • Yamaguchi S, Desplan C, Heisenberg M. Contribution of photoreceptor subtypes to spectral wavelength preference in Drosophila. Proc Natl Acad Sci U S A. 2010;107:5634–5639. [PubMed]
  • Yamaguchi S, Wolf R, Desplan C, Heisenberg M. Motion vision is independent of color in Drosophila. Proc Natl Acad Sci U S A. 2008;105:4910–4915. [PubMed]
  • Zelhof AC, Koundakjian E, Scully AL, Hardy RW, Pounds L. Mutation of the photoreceptor specific homeodomain gene Pph13 results in defects in phototransduction and rhabdomere morphogenesis. Development. 2003;130:4383–4392. [PubMed]