INTRODUCTION

One of the major challenges in developmental biology is tounderstand how a complex morphological pattern is generated.Pattern formation relies on cell lineage dependent mechanismsas well as signaling between cells. The ommatidial pattern ofthe Drosophilacompound eye is one of the most thoroughlystudied systems for elucidating the mechanism of patternformation. Many aspects of compound eye development inDrosophilaare dependent on cell-cell interactions, and mosaicanalysis has shown that photoreceptor cell recruitment isindependent of lineage (Ready et al., 1976; Lawrence andGreen, 1979). One of the best characterized events inommatidial assembly is R7 photoreceptor cell recruitment, inwhich a signal from the R8 photoreceptor induces a bipotentialprecursor to become an R7 cell rather than a non-neuronal conecell. Here we report our studies of a novel inductive signal fromthe R7 photoreceptor cell that is responsible for a binary switchbetween the expression of two different opsins in the R8 cell.

The compound eye of Drosophila contains approx. 750ommatidia arranged in a repetitive hexagonal array. Eachommatidium is an assembly of 12 accessory cells and eightphotoreceptor cells (R cells) (Hardie, 1986; Wolff and Ready,1993). The photoreceptor cells form rhabdomeres whichproject into the intraommatidial cavity and are arranged in a

trapezoidal pattern (Fig. 1A,B). Six genes encoding the opsinsexpressed in the adult fly have been isolated and characterized.The ninaEgene (Rhodopsin 1, Rh1) is expressed in the R1-R6cells which form the peripheral rhabdomeres that surround thecentral rhabdomeres of the R7 and R8 photoreceptor cells(Scavarda et al., 1983; O’Tousa et al., 1985; Zuker et al., 1985;Feiler et al., 1988). The Rh2 opsin gene is expressed in theocelli, simple eyes located on the vertex of the head, and mayalso be expressed in the male testes (Cowman et al., 1986;Feiler et al., 1988; Pollock and Benzer, 1988; Alvarez et al.,1996). The Rh3 and Rh4 opsin genes are expressed in non-overlapping sets of R7 cells (Fryxell and Meyerowitz, 1987;Montell et al., 1987; Zuker et al., 1987; Feiler et al., 1992).Rh3 is also expressed in a specialized class of R8 cells alongthe dorsal margin of the eye (Fortini and Rubin, 1990; Feileret al., 1992).

We and others have identified Rh5 and Rh6, two novelDrosophila opsin genes that are expressed in subsets of R8cells (Chou et al., 1996; Huber et al., 1997; Papatsenko et al.,1997). The expression of Rh5 in the R8 cell of an individualommatidium is strictly coordinated with the expression of Rh3in the R7 cell of the same ommatidium, at both the protein andtranscript level (Chou et al., 1996; Papatsenko et al., 1997).Likewise it has been proposed that the expression of Rh6in R8cells may be paired with the expression of Rh4in the R7 cells

607Development 126, 607-616 (1999)Printed in Great Britain © The Company of Biologists Limited 1999DEV8576

Opsin gene expression in the R7 and R8 photoreceptor cellsof the Drosophilacompound eye is highly coordinated. Wehave found that the R8 cell specific Rh5 and Rh6 opsins areexpressed in non-overlapping sets of R8 cells, in a precisepairwise fashion with Rh3 and Rh4 in the R7 cells ofindividual ommatidia. Removal of the R7 cells in sevenless,boss or sina mutants, disrupts Rh5 expression anddramatically increases the number of Rh6-expressing R8cells. This suggests that the expression of Rh5 may beinduced by an Rh3-expressing R7 cell, whereas Rh6expression is most likely a default state of the R8 cell. Wefound that the paired expression of opsin genes in the R7

and R8 cells occurs in a sevenlessand boss independentmanner. Furthermore, we found that the generation of bothRh3- and Rh4-expressing R7 cells can occur in the absenceof an R8 cell. These results suggest that the specification ofopsin expression in the R7 cells may occur autonomously,whereas the R7 photoreceptor cell may be responsible forregulating a binary developmental switch between inducedand default cell-fates in the R8 cell.

Key words: Retina, Drosophila melanogaster, Photoreceptor, Cell-fate determination, Rhodopsin, Opsin, Cell patterning

SUMMARY

Patterning of the R7 and R8 photoreceptor cells of Drosophila : evidence for

induced and default cell-fate specification

Wen-Hai Chou 1, Armin Huber 2, Joachim Bentrop 2, Simone Schulz 2, Karin Schwab 2, Linda V. Chadwell 1,Reinhard Paulsen 2 and Steven G. Britt 1,*1Institute of Biotechnology and Department of Molecular Medicine, University of Texas Health Science Center at San Antonio,,15355 Lambda Drive, San Antonio, Texas 78245-3207, USA2University of Karlsruhe, Institute of Zoology, Department of Cell and Neurobiology, Kornblumenstrasse 13, D-76128 Karlsruhe,Germany*Author for correspondence (e-mail: britt@uthscsa.edu)

Accepted 1 December 1998; published on WWW 20 January 1999

608

of the same ommatidia (Huber et al., 1997). The pairedexpression of the opsin genes in the R7 and R8 photoreceptorcells of individual ommatidia is likely to result from a specificdevelopmental signal. Because the expression of the Rh5transcript and protein are disrupted in sevenless(sev) mutantflies, we propose that the signal responsible for the patterningof R7 and R8 cells may arise in the R7 photoreceptor cell.

The Rh5 opsin has been a useful cell-type-specific-markerfor the examination of R7 and R8 photoreceptor cell patterning,and the cloning of the gene encoding Rh6 provides anopportunity to examine other aspects of this system. In thisstudy, we found that the expression of Rh6 in the R8 cell of anindividual ommatidium is faithfully coordinated with theexpression of Rh4 in the R7 cell. We also found evidence thatthe regulation of Rh5 versus Rh6 expression in R8 cells isdependent upon the presence of the R7 cell, and that thecoordinated expression of visual pigment genes within anindividual ommatidium is not dependent upon sevor boss. Wepropose that the expression of Rh5 is an induced state, and thatin the absence of an Rh3-expressing R7 cell, the R8 cellassumes a default state and expresses Rh6. Our results providestrong evidence for the presence of a novel developmentalsignal from the R7 photoreceptor cell that is responsible forregulating opsin gene expression in the R8 cell.

MATERIALS AND METHODS

Drosophila culture, stocks and geneticsAll fly strains were maintained in humidified incubators on 12 hourlight/dark cycles, on standard cornmeal-molasses-agar medium.Genetic nomenclature used in the text is that of Lindsley and Zimm(1992) and Flybase Consortium (1998). Two different white eyedstrains were used in this study as control flies (cn bw or w1118).Removal of the red pigments in the eye reduces tissue backgroundfluorescence that would otherwise interfere with the morphologicalanalysis. Specific genotypes used in this study were constructed fromavailable stocks by standard techniques using appropriate balancersand visible markers.

Monoclonal and polyclonal antibody productionMouse monoclonal antibodies against Rh4, Rh5 and Rh6 weregenerated against Keyhole Limpet Hemocyanin (Pierce, Rockford,IL) coupled synthetic peptides (Kohler and Milstein, 1975). Thepeptides used as haptens were based on the deduced amino acidsequence of the relevant visual pigment: (N-to-C terminus)LGVNEKSGEISSAQS corresponding to amino acids 352-366 ofRh4, REKHATSGTSGGQES corresponding to amino acids 353-367of Rh5 and GKDDLTSDSRTQATA corresponding to amino acids348-362 of Rh6. Hybridoma cell-lines were single-cell-cloned, andovergrown culture supernatants were used for immunohistochemistryas described below. Rabbit polyclonal antisera was generated againsta dihydrofolate-reductase fusion protein containing the C-terminalregion of Rh6 opsin (amino acids 331-369). The fusion protein wasexpressed, purified, and the antibodies were generated and purified aspreviously described (Huber et al., 1996).

MicroscopyImmunolabeling of ultrathin sections for electron microscopy wasperformed as described by Bentrop et al. (1997). Workingconcentrations of the primary antibodies were: anti-Rh1, 1:20; anti-Rh5, undiluted hybridoma supernatant; anti-Rh6, 1:50. Secondaryantibodies (Nanoprobes) were used at 1:100-1:150 (goat anti-rabbit;10 nm gold) or 1:150 dilution (goat anti-mouse; 1 nm gold). The

signal was amplified by silver enhancement as described by Danscher(1981). For the analysis of ommatidial cross sections by lightmicroscopy, the tissue was fixed as described above, embedded inEpon, and 1 µm sections were stained with toluidine blue.

The immunohistochemistry and confocal imaging were performedas previously described (Chou et al., 1996). Working concentrations ofthe primary antibodies were as follows: anti-Rh1 monoclonal Ab 4C5(1:50 dilution; de Couet and Sigmund, 1987), rabbit anti-Rh3 (1:10dilution; Feiler et al., 1992), rabbit anti-Rh4 (1:20 or 1:10 dilution;Feiler et al., 1992), monoclonal anti-Rh4 (clone 11E6; 1:10 dilution;IgG1 subclass), monoclonal anti-Rh5 (clone 7F1; 1:20 dilution; IgG1subclass), rabbit anti-Rh6 antibody (1:50 dilution) and monoclonalanti-Rh6 (clone 14C5; 1:20 dilution; IgM subclass). Secondaryantibodies and other immunological reagents were obtained fromJackson ImmunoResearch Laboratories, Inc. (West Grove, PA) andSouthern Biotechnology Associates, Inc. (Birmingham, AL). Thespecimens were incubated with the primary antibodies at 4°Covernight (or at room temperature for 1.5 hours), and the incubationswith the secondary antibodies were performed at room temperature for1 hour. The confocal images were collected using a Zeiss LSM-310(Thornwood, NY), or a Bio-Rad MRC-1024 (Hercules, CA) attachedto a Nikon Labphot-2 microscope (Melville, NY).

RESULTS

Rh6 is expressed in a subset of R8 photoreceptorcellsTo determine the expression pattern of the Rh6 opsin, wegenerated a polyclonal antiserum against the C-terminus of theRh6 protein. Immunofluorescence analysis showed that Rh6 islocalized to a subset of rhabdomeres that extend from themiddle of the retina to the level of the lamina (Fig. 1C). Doublelabeling with antibodies against both Rh6 and Rh1 also showedthat the Rh6 protein is restricted to the proximal retina, andthat Rh6 is only expressed in a subset of ommatidia (Fig. 1D).This result was confirmed by preparing dissociated ommatidiafrom adult fly heads and labeling them with antibodies againstRh6 and Rh1. As Fig. 1E shows, the R8 rhabdomere thatcontains Rh6 is located basally within the ommatidium and issurrounded by the rhabdomeres of the Rh1-expressing R1-6cells. We found no evidence of Rh6 expression in other regionsof the head, retina, or in the ocelli (data not shown). Thesefindings indicate that Rh6 is expressed specifically in a subsetof R8 photoreceptor cells.

Rh6 expression in R8 cells is coordinated with theexpression of Rh4 in adjacent R7 cells of individualommatidiaPrevious studies have shown that the Rh3 and Rh4 visualpigments of the R7 photoreceptor cells are expressed in a non-overlapping pattern (Montell et al., 1987; Feiler et al., 1992).To determine whether Rh5 and Rh6 are expressed in a similarmanner, we performed transmission immunoelectronmicroscopy on cross sections taken from the basal retina. Wefound that Rh5 and Rh6 are localized to the R8 rhabdomeresof different ommatidia (Fig. 2). Similarly, in double labelingexperiments performed on frozen sections (Fig. 3A) and onindividual dissociated ommatidia (Fig. 3D), the Rh5 and Rh6opsins are expressed in non-overlapping subsets of R8photoreceptor cells. 29% of ommatidia expressed Rh5 and71% expressed Rh6 (see Table 1).

To characterize the expression pattern of Rh6 in relation to

W. H. Chou and others

609Photoreceptor cell patterning in Drosophila

other known opsins, we examined frozen sections from theretinas of white-eyed flies using immunohistochemistry withantibodies directed at Rh3, Rh4 and Rh6. As shown in Fig. 3B,double labeling with antibodies against Rh3 and Rh6 illustratesthat the R7 and R8 cells labeled by these reagents are notpaired in most cases, suggesting that Rh3 and Rh6 areexpressed in different ommatidia. Conversely, when labelingwith antibodies against Rh4 and Rh6 there is a precise pairingof Rh4 and Rh6 within the R7 and R8 cells of an ommatidium(Fig. 3C). This result suggests that Rh4 and Rh6 are expressedin a paired manner similar to Rh3 and Rh5.

To investigate whether or not the paired expression of Rh4and Rh6 in the R7 and R8 cells of individual ommatidia is auniform occurrence throughout the retina and not an artifact of

sectioning, we performed double labeling experiments ofdissociated ommatidia using antibodies against Rh3 and Rh6.As Fig. 3E shows, Rh3 and Rh6 are expressed in the R7 andR8 cells of different ommatidia in most cases (upper twoquadrants of Fig. 3E, see also Table 1). Two exceptional classesof ommatidia were also noted. The lower left quadrant of Fig.3E shows an ommatidium which expresses Rh3 in both the R7and R8 rhabdomeres. This is likely to be an ommatidium fromthe dorsal margin of the eye (Fortini and Rubin, 1990; Feileret al., 1992). The lower right quadrant of Fig. 3E shows a singleommatidium in which Rh3 and Rh6 are expressed in adjacentR7 and R8 photoreceptor cells. This type of ommatidium wasfairly rare, and we found only a small number of ommatidiawith this staining pattern (Table 1).

Fig. 1. Expression pattern of Rh6 in theDrosophila retina.(A) A differentialinterference contrast (DIC) image of alongitudinal section through the retina showingthe levels of the lamina (l), retina (r) and cornea(c). (B) (left side) A DIC image of a singledissociated ommatidium alongside a diagramindicating the position of the rhabdomereswithin the ommatidium and (right) in crosssections of the basal and apical region. (C) Thesuperimposed DIC and confocalimmunofluorescence image of a retina labeledwith antibodies against Rh6, showing that Rh6is localized to short rhabdomeres at the base ofthe retina. Likewise, double labeling oflongitudinal sections (D) and a dissociatedommatidium (E), with antibodies against Rh1and Rh6 confirms that Rh6 is present inrhabdomeres at the base of the retina that aresurrounded by the rhabdomeres of the Rh1-expressing R1-6 cells. Scale bar in A, 50 µm.

Table 1. Antibody labeling of dissociated ommatidiaStrain Percentage of ommatidia labeled with antibodies against specific opsins Total counted

cn1 bw1 Rh5 Rh629% 71% 214*

cn1 bw1 Rh3 alone Rh6 alone Rh3‡,§ Rh3 and Rh631% 53% 10% 6% 240

cn1 bw1 Rh4 and Rh6 Rh6 alone Rh4 alone Rh4‡96% 1.4% 0.3% 0.3% 274¶

w sev14 Rh5 Rh6 unlabeled12%** 86% 2% 597

w; boss1cu Rh5 Rh6 unlabeled10%** 81% 9% 268

w; boss1cu Rh3 Rh6 Rh3 and Rh6 unlabeled4%§ 90% 0.4% 6% 242

*Includes only ommatidia labeled with either Rh5 or Rh6 antibody.‡Labeling in both R7 and R8.§Rh3-labeled R8 cells are probably from the dorsal margin.¶Includes only ommatidia labeled with Rh4 and/or Rh6 antibody.**2/3 of the ommatidia labeled with Rh5 antibody were aberantly positioned.

610 W. H. Chou and others

Fig. 2. Localization of Rh5 and Rh6 tothe rhabdomeres of the R8photoreceptor cells. In tissue sectionscut from the basal region of the retina,the rhabdomeres of the R8 cells can beseen projecting between those of R1and R2. Immunoelectron microscopy ofretinas from white eyed flies (w) wasperformed to determine the subcellulardistribution of Rh5 (A, and at higherpower in B) and Rh6 (C, and at higherpower in D). In both cases there isspecific labeling of the rhabdomere ofthe R8 photoreceptor cell. Anti-Rh5(mouse monoclonal) labeling wasdetected with a 1 nm gold-coupledsecondary antibody, and anti-Rh6(rabbit polyclonal) was detected with10 nm gold-coupled secondaryantibody. A control retina incubatedwithout primary antibodies but withboth secondary antibodies fails to labelthe retina (E), whereas anti-Rh1 (rabbit-polyclonal detected by 10 nm gold-coupled secondary) specifically labelsthe R1-6 rhabdomeres which expressthe Rh1 visual pigment (F). Scale barsin B, D and F correspond to 0.5, 0.5 and1.0 µm respectively.

Fig. 3. Paired expression of Rh4 and Rh6 inadjacent R7 and R8 cells within the sameommatidia. In longitudinal sections of cn bwretinas, Rh5 and Rh6 are expressed in non-overlapping sets of R8 cells (A). Rh3 and Rh6are expressed in the R7 and R8 cells ofdifferent ommatidia (B), whereas Rh4 andRh6 appear to be expressed in adjacent R7and R8 cells in the same ommatidia (C). Thisobservation was confirmed in dissociatedommatidia where we found that Rh5 and Rh6are expressed in the R8 cells of differentommatidia (D). Likewise, Rh3 and Rh6 areexpressed in different ommatidia in mostcases (E, two upper quadrants). In the lowerleft quadrant of E, an ommatidium withlabeling against Rh3 in both the R7 and R8cell is shown. The lower right quadrant of Eshows a single ommatidium in which Rh3 andRh6 are expressed in the R7 and R8 cells,respectively, of the same ommatidium. Fshows numerous examples of the normalpaired expression of Rh4 and Rh6 in the R7and R8 cells of individual ommatidia. Scalebar in F, 25 µm.

611Photoreceptor cell patterning in Drosophila

Dissociated ommatidia double labeled with antibodiesagainst Rh4 and Rh6 are shown in Fig. 3F. In virtually everyommatidium which expresses Rh4 in the R7 cell, Rh6 isexpressed in the R8 cell. With the few exceptions noted aboveand in Table 1, Rh6 is expressed in a non-random manner, ina subset of R8 cells in ommatidia that contain Rh4-expressingR7 cells. This indicates that over most of the compound eyethe R7 and R8 cells of individual ommatidia exist as ‘even’(Rh4/Rh6 expressing) or ‘odd’ (Rh3/Rh5 expressing) pairs.These appear to correspond to the R7 yellow (R7y) and R7 pale(R7p) types of ommatidia, respectively, that have beendescribed in Calliphora, Muscaand Drosophila(Kirschfeld etal., 1978; Franceschini et al., 1981; and reviewed by Chou etal., 1996).

In the absence of R7, the R8 cells assume a defaultstate and express Rh6If an inductive event between the R7 and R8 cells is responsiblefor the coordinated expression of the opsin genes in these twocells, then we would predict that removal of the R7 cell might

have an effect on the expression of opsin genes in the R8 cell.To test this hypothesis, we previously examined the expressionof Rh5 in sevmutant flies, that lack R7 photoreceptor cells, andfound that Rh5 expression is disrupted (Chou et al., 1996). Thisresult suggested that the expression of Rh5 in a subset of R8cells is likely to be an induced event that is dependent upon asignal from an Rh3-expressing R7 cell.

To determine whether the expression of Rh6 within the R8cell is also an induced event, or whether there is a defaultpathway of opsin specification in the R8 cell, we examined theexpression of Rh6 insev14 mutant flies. Interestingly, while theR8 cells of normal flies express either Rh5 or Rh6 (Fig. 4A),almost all R8 cells express Rh6 insev14 flies (Fig. 4B, Table 1).Our finding that the proportion of R8 cells expressing Rh6increases dramatically in sev14 flies suggests that in the absenceof an R7 cell, or an Rh3-expressing R7 cell, the R8 cell assumesa ‘default’ fate to become an Rh6-expressing R8 cell. Curiously,we found that Rh5 expression was not completely eliminated insev14 flies. We found rare examples of Rh5-containing cells insome sections (indicated in Fig. 4B), and many of theserhabdomeres were either longer than normal or aberantlylocalized to the apical region of the retina. Although there areno Rh3-expressing R7 cells in the ommatidia in which theserare Rh5-expressing cells are found, there must be somemechanism responsible for their induction.

To determine whether the changes in the expression of theRh5 and Rh6 proteins in sev14 mutant flies is reflected in thesteady-state transcript levels of the genes, we compared thelevel of mRNAs in wild-type and sev14 mutant flies. As shownin the left three lanes of Fig. 5, the transcripts for both Rh5(upper panel) and Rh6 (middle panel) were detectable inmRNAs prepared from the heads of wild-type (cn bw) flies. No

Fig. 4. Loss of R7 photoreceptor cells disrupts Rh5 and Rh6expression. Compared to the normal pattern of Rh5 and Rh6expression (A), mutants which lack R7 cells show a dramaticreduction in the number of cells expressing Rh5, whereas Rh6 isexpressed in virtually every R8 cell. This occurs in sev14 (B), boss1

(C) and sina2/sina3 mutants (D). In each of these mutantbackgrounds, the expression of Rh5 is not entirely eliminated. Thereare rare examples of Rh5 expression in cells that have abnormallylong rhabdomeres (B, arrow), as well as those that have rhabdomeresthat are displaced into an apical position (B, arrowhead). Themorphology of the Rh6-expressing cells is similarly affected (readilyapparent in B, C and D).

Fig. 5.Regulation of Rh5and Rh6mRNA expression.Northern analysis revealedthat the Rh5(upper panel)and Rh6(middle panel)genes are transcribed as 1.4and 1.35 kb mRNAs,respectively. The lowerpanel shows the 0.6 kbtranscript of RP 49, whichwas used as a loadingcontrol (O’Connell andRosbash, 1984). Both theRh5and Rh6transcripts arepresent in the heads (H) butnot the bodies of white eyedcn bwflies. There is notranscript for either genedetectable in the heads ofeyes absent (eya) flies,which lack retinal structures(Sved, 1986), consistentwith both Rh5and Rh6being expressed specificallyin the retina in adult flies.The blots also show that theRh5transcript isdramatically reduced in w; sev14 mutant flies that lack R7 cells,whereas the Rh6gene is still expressed in this mutant.

612

transcripts were observed in mRNAs prepared from the bodiesof these animals, or from the heads of eyes absent(eya) mutantflies, which lack retinal tissues (Sved, 1986). This retinalspecific expression is consistent with both Rh5 and Rh6encoding visual pigments of the compound eye. In mRNAsprepared from the heads of sev14 mutant flies, the level of Rh5transcript, but not the Rh6transcript, was dramatically reduced.These results suggest that the alterations in the expression ofthe Rh5 opsin in sev14 mutants are likely to be regulated at thelevel of transcription.

The normal expression pattern of Rh5 and Rh6 isdependent on the presence of the R7 cell, not thesev or boss proteinsTo determine whether the change in R8 cell fate is due to thelack of the R7 cell or to the lack of the sev protein itself wefirst examined frozen sections taken from bossand seven inabsentia (sina) mutant flies (Reinke and Zipursky, 1988;Carthew and Rubin, 1990). We found that in both boss1 (Fig.4C) and sina2/sina3 (Fig. 4D) mutants Rh5 expression isdisrupted, whereas the proportion of R8 cells expressing Rh6is substantially increased in a manner identical to that seen insev14 mutants. This result was confirmed by examiningdissociated ommatidia of boss1 mutant flies (Table 1). Theseresults are consistent with our observations in sevmutant flies,and provide additional support for the presence of a defaultstate. In addition, these experiments indicate that in the absenceof a fully differentiated R7 cell, the presence of sev (in bossmutants), boss (in sevmutants) or both sev and boss (in sinamutants) is insufficient to rescue normal Rh5 expression.

To determine whether sevis required for the expression ofRh5, we examined mutants in which the R7 cells were rescuedin a sev-independent manner. Ras1 is an essential componentof the sev signaling cascade, and it has been shown that theexpression of an activated form of this molecule (Ras1Val12) in

R7 cell precursors is sufficient to rescue R7 cell formation insevmutant flies (Fortini et al., 1992). As Fig. 6A shows, wefound that when Ras1Val12 expression is driven by the sevenhancer in a sev14 mutant background, Rh5 expression wasrestored. Rh5 expression in this mutant background occurredpredominantly in a normal paired manner beneath therhabdomeres of Rh3-expressing R7 cells. Similar experimentsperformed in a boss1 mutant background yielded identicalresults (data not shown). These findings indicate that there isno specific requirement for the sev or boss proteins in inducingthe expression of the Rh5 opsin in R8 cells. To determinewhether the pairing of opsin expression in the R7 and R8 cellsis precisely maintained in sev14 mutants when the R7 cells arerescued by the expression of Ras1Val12, we examined opsinexpression at the level of dissociated ommatidia. We found thatin most cases, Rh3-expressing R7 cells pair with Rh5-expressing R8 cells (Fig. 6B), and likewise for cells expressingRh4 and Rh6 (data not shown). However, there are someexamples where Rh3 and Rh6 are found in the sameommatidium (Fig. 6C), and rare instances where Rh4 and Rh5are also expressed in the same ommatidium (Fig. 6D). In thelater case, although Rh4 and Rh5 are expressed in the sameommatidium, the rhabdomeres containing Rh4 and Rh5 are notarranged in tandem as we have observed in the normal flies.

This abnormal patterning may be explained by the formationof multiple R7 cells in individual ommatidia, which has beenobserved in sev mutants expressing the Ras1Val12 transgene(Fortini et al., 1992). Flies ectopically expressing seven-up(svp) or bosshave also been shown to generate multiple R7cells within individual ommatidia. These ommatidia appear tocontain both Rh3 and Rh4-expressing R7 cells as indicated byRh4-lacZtransgene expression (Van Vactor et al., 1991; Hiromiet al., 1993). To determine whether the ‘unpaired’ expressionof Rh3 and Rh6 or Rh4 and Rh5 in some ommatidia was dueto the presence of multiple R7 cells expressing Rh3 and Rh4

W. H. Chou and others

Fig. 6. R7 and R8 photoreceptor cell patterningoccurs independently of sevenless. In sev14

mutant flies, in which the R7 cells have beenrescued by expression of activated Ras1Val12

driven by the sevenhancer, the pairing of Rh3and Rh5 is restored (A). Dissociated ommatidiafrom this strain show examples of normallycoordinated Rh3/Rh5 pairing in individualommatidia (B), as well as instances ofabnormal Rh3/Rh6 and Rh4/Rh5 expression insome ommatidia (C and D respectively). Thismost likely reflects the generation of multipleR7 cells within an ommatidium, which E showsmay consist of both Rh3 and Rh4-expressingmembers. In triple labeling experiments, weconfirmed that the abnormal occurrence ofRh3/Rh6 as well as Rh4/Rh5 combinations maybe explained by the presence of individualommatidia that express Rh3/Rh4/Rh6 (F) orRh3/Rh4/Rh5 (G). Scale bar in A, 50 µm.

613Photoreceptor cell patterning in Drosophila

in different cells within the same ommatidium, we performedadditional double and triple labeling experiments ondissociated ommatidia from this genetic background. We foundthat most ommatidia contain single R7 cells that produce eitherRh3 or Rh4 (Fig. 6E left of panel). In ommatidia containingmultiple R7 cells, we found examples where the R7 cells eitherhave the same or different opsins (Fig. 6E, middle and right ofpanel respectively). In triple labeling experiments, we foundommatidia expressing Rh4 in one or both R7 cells that arepaired with Rh6 expression in the R8 cell (Fig. 6F, twoommatidia on left). In ommatidia containing both Rh3- andRh4-expressing R7 cells, the R8 cells express either Rh6 (Fig.6F, three right ommatidia) or Rh5 (Fig. 6G). These resultssuggest that the geometry of the R7 and R8 cells with respectto each other may play a role in determining whether the R8cell will express Rh5 or Rh6. This result is completelyconsistent with the idea that an intercellular signal from the R7cell is involved in regulating opsin gene expression in the R8cell.

The expression of Rh3 and Rh4 in the R7photoreceptor cells is not dependent on thepresence of the R8 photoreceptor cellsFrom the evidence presented above we propose that a novel,and as yet unknown, developmental signal from the R7 cellregulates a binary switch between the expression of Rh5 andRh6 in the R8 cells, and that in the absence of such a signalthe R8 cell adopts a default fate and expresses Rh6. Animmediate question raised by this model concerns themechanism by which opsin expression in the R7 cells isspecified. The finding that both Rh3 and Rh4-expressing R7cells can be generated within a single ommatidium in sev-

Ras1Val12 flies suggests that the specification of opsinexpression in the R7 photoreceptor cell may occurautonomously (Fig. 6E-G). If the R7 cell’s opsin expression isindeed autonomous and independent of the R8 cell, we wouldpredict that removal of the R8 cell would not affect the abilityof R7 to assume either an Rh3- or Rh4-expressing state.However, because the R8 photoreceptor cell is essentially thefounder cell of the developing compound eye, and is the sourceof an inductive signal (boss) that is required for R7photoreceptor cell recruitment, this approach is problematic.Nonetheless, previous studies have shown that ectopicexpression of svp in the R8 cells, at a time following theinduction of the R1-6 cells but prior to R7 cell recruitment,leads to the generation of ommatidia that lack both centralphotoreceptor cells (R7 and R8) (Hiromi et al., 1993; Krameret al., 1995). These studies have suggested an alternative meansto generate flies having ommatidia that lack only the R8 cells,by introducing thesev-Ras1Val12 transgene to rescue the R7cells in the background of flies expressing svp1 under thecontrol of scabrous (sca).

The R7 and R8 cells are distinguishable within theommatidium based upon their position and the visual pigmentsthey express. In the apical region of the retina in wild-type flies,the rhabdomeres of the R7 cells project from the photoreceptorcell body between the rhabdomeres of the R1 and R6 cells(Figs 7A, 1B). In the basal region of the retina, therhabdomeres of the R8 cells project between the rhabdomeresof the R1 and R2 photoreceptor cells (Figs 7B, 1B). Flieslacking both central photoreceptor cells were generated usingthe GAL4/UAS system, by crossing flies carrying the sca-GAL4 transgene to those carrying the UASG-svp1fusion gene(Brand et al., 1994; Kramer et al., 1995). As shown in both

Fig. 7.Populations of Rh3- and Rh4-expressingR7 cells can be generated in the absence of R8cells. In the apical region of the retina of wildtype (Canton S) flies, the rhabdomere of the R7cell is projected between those of R1 and R6(A), whereas in the basal region of the retina theR8 cell projects its rhabdomere between R1 andR2 (B). Flies that lack both R7 and R8 weregenerated by expressing svp1under the controlof the scapromoter (sca-GAL4/UAS-svp1).C and D show that ommatidia in both the apicaland basal retina lack central rhabdomeres.Consistent with the absence of the R7 and R8cells throughout the retina, longitudinal sectionsrevealed no labeling against either the Rh3 andRh4 (G) or the Rh5 and Rh6 opsins (H).Introduction of the sev-Ras1Val12 transgene intothis genetic background (RasVal12/sca-GAL4/UAS-svp1) results in the generation of single aswell as multiple R7 cells within individualommatidia, that are normally positioned withrespect to R1 and R6 in some cases (arrowheadsin E). In the absence of R8 cells, some of the R7cell rhabdomeres extend into the basal region ofthe retina (arrowheads in F). These R7 cells mayexpress either Rh3 or Rh4, and multipleexamples of both R7 cell types were noted (I).In the absence of the R8 cells, there is nolabeling against the Rh5 or Rh6 opsins (J).Scale bar in J, 50 µm.

614

apical and basal sections (Fig. 7C,D respectively) all of theommatidia in sca-GAL4/UASG-svp1flies lack both central R7and R8 photoreceptor cells. Some ommatidia also lack one ortwo outer photoreceptor cells, and some have an extra outerphotoreceptor cell. As would be expected in animals that lackall of the central R7 and R8 photoreceptors, Rh3 and Rh4expression are absent (Fig. 7G), and Rh5 and Rh6 expressionare absent as well (Fig. 7H).

To test whether the Rh3- and Rh4-expressing states of theR7 cell can be acquired in the absence of the R8 cell, weintroduced the sev-Ras1Val12 transgene into the sca-GAL4/UASG-svp1background. As expected, in Ras1Val12/sca-GAL4/UAS-svp1flies the formation of the R7 cells wasrescued (Fig. 7E). In cross sections through the apical retina,we observed numerous examples of central photoreceptorcells having small rhabdomeres, several examples ofommatidia with multiple central photoreceptor cells, andsome ommatidia in which the R7 cell had not been rescued.Although the ommatidial morphology is somewhat irregularin this genotype due to the presence of extra R7 cells, as wellas the gain or loss of some outer photoreceptor cells, wefound numerous examples of ommatidia having normalmorphology in which the R7 rhabdomeres projected betweenthose of R1 and R6. Interestingly, in the basal region of theretina we also found numerous examples of centralphotoreceptor cells (Fig. 7F). In morphologically normalommatidia, the rhabdomeres of the central photoreceptorcells were also projected between those of R1 and R6 in manycases, suggesting that the rhabdomeres of the R7 cell mayextend into the basal region of the retina in the absence ofR8, in a manner similar to that observed above (Fig. 6D,G).Consistent with this assumption, we found that sections of theretinas of these animals failed to label with antibodies againstRh5 and Rh6 (Fig. 7J). Furthermore, we found that both Rh3-and Rh4-expressing R7 cells are present in these animals, andthat many of their rhabdomeres do indeed extend into thebasal region of the retina (Fig. 7I). These results provide verystrong evidence that at the time of R7 cell induction, thepresence of the R8 cell is not required for the generation ofthe two types of R7 cells. These findings are consistent withthe idea that while opsin specification in the R8 cell isdependent on the presence of the R7 cells, the generation ofthe two classes of R7 cells may occur autonomously.

DISCUSSION

In this paper, we have shown that Rh6 is expressed in a subsetof R8 photoreceptor cells, and that its expression is coordinatedwith the expression of Rh4 in the R7 cell of the sameommatidium. This patterning of opsin expression in the R7 andR8 cells is similar to what we and others have observed for theexpression of Rh3 and Rh5 (Chou et al., 1996; Papatsenko etal., 1997) As we have demonstrated, this unique patterning isdramatically altered in mutants that lack R7 cells, resulting inthe disruption of Rh5 expression and the expression of Rh6 ina dramatically increased proportion of R8 cells. These resultsare consistent with a model in which a developmental signalfrom the Rh3-expressing R7 cell induces the expression of Rh5in the R8 cell, and represses the expression of Rh6. In theabsence of this signal the R8 cell assumes a default state and

expresses Rh6. An immediate question raised by these resultsis the nature of this novel signaling pathway and the identityof the molecules that mediate it. As we have described in thispaper, it is quite clear that neither sev nor boss are required forthe establishment of the paired expression of visual pigmentsin the R7 and R8 cells, and that the presence of either boss orsev or both is insufficient to restore Rh5 expression in theabsence of an R7 cell. These results indicate that there must beother novel signaling molecules that are responsible forpatterning of the R7 and R8 cells.

In the course of these experiments, we have observed anumber of apparent exceptions to the predominant even orodd pairing of opsins expressed in the R7 and R8photoreceptor cells. In particular, we found instances ofexpression of Rh3 and Rh6 in the R7 and R8 cells ofindividual ommatidia (Fig. 3E and Table 1), occurring in 6%of cases. Because Rh5 is not expressed in these R8 cells, andbecause Rh5 expression appears to be an induced state, it maybe that the geometry of the R7 and R8 cells in theseexceptional ommatidia was somehow abnormal at the timewhen this signal is required. Thus, these R8 cells may havefailed to receive the inductive signal, and then assumed theRh6-expressing default state. Consistent with this, studies ofommatidia having multiple R7 cells (Figs 6D-G) seem tosupport the idea that the signal to induce Rh5 expression mustbe highly spatially restricted. Additionally, we have neverobserved ommatidia from wild type strains that showedpaired expression of Rh4 and Rh5. This suggests that whilethe inductive signal may fail in some rare cases, Rh5 cannever be induced in a normal eye by an Rh4-expressing R7cell because it does not transmit the inductive signal.Curiously, in mutant strains that lack R7 cells, Rh5expression is not completely abolished (Fig. 4B). Many ofthese Rh5-expressing cells are morphologically abnormal, inthat they have long rhabdomeres or rhabdomeres in an apicalposition. Such photoreceptor cells have been observedpreviously in sevmutants and shown to be R8 cells, basedupon their axonal projections and the timing of their terminalmitoses (Campos-Ortega et al., 1979). Perhaps these cells arereceiving or improperly processing an inductive signal, basedon their abnormal position within the eye. This result doesshow however, that Rh5 expression is not absolutelydependent upon the presence of the R7 cell. The signal usedfor the induction of Rh5 could be used elsewhere within theeye for other purposes, or the signal from the R7 cell may notbe direct. Identification of the genes required to establish R7and R8 cell patterning should allow us to resolve some ofthese questions.

If the R7 cell is the primary cell to acquire a specific cellfate and to communicate that to the secondary R8 cell, then isthis primary cell fate decision a stochastic event, occurring ata frequency of 30%/70% over the majority of the compoundeye, or, alternatively, is it an induced event? Since there is noevidence for any strict lineage relationship betweenphotoreceptor cells, and the apparent random distribution ofRh3 and Rh4 argues against a strong inductive mechanism, thisprocess seems to be regulated by a stochastic mechanism(Fortini and Rubin, 1990). Our finding that both Rh3- and Rh4-expressing R7 cells are generated in flies that lack R8 cells, isconsistent with the idea that the Rh3 versus Rh4 cell-fatedecision occurs autonomously. We propose that once the R7

W. H. Chou and others

615Photoreceptor cell patterning in Drosophila

photoreceptor cell-fate has been specified, this even or oddidentity is then communicated to the R8 cell. This putativesignal would regulate a binary cell-fate decision in R8 cells,between the proposed induced (Rh5 expressing) and default(Rh6 expressing) states.

Binary cell-fate decisions have been characterized in anumber of different systems. In Drosophila, the glial cellsmissinggene regulates the binary decision between neuronal andglial cell fates (Hosoya et al., 1995; Jones et al., 1995). Similarly,lineage based cell-fate determination may also occur in a binarymanner (Jan and Jan, 1998). Likewise, R7 photoreceptor cellrecruitment via the boss/sev pathway also serves as a potentialexample of a binary cell-fate decision (Banerjee and Zipursky,1990; Zipursky and Rubin, 1994). The main difference betweenthese examples and R8 photoreceptor cell opsin specification isthat, in the later case, the two alternative cell-fates are only subtlydifferent from one another. Nonetheless, the regulatedexpression of different visual pigments in different types ofphotoreceptor cells has obvious biological importance becausethis is the basis for color vision in many organisms. As we havediscussed previously, physiological studies of Musca andCalliphora suggest that the pairing of different types of R7 andR8 cells may play an important role in tuning photoreceptorspectral sensitivity, by coordinating the expression of a screeningpigment in Rh4-expressing R7 cells that serves as an optical filterfor Rh6-expressing R8 cells (Kirschfeld et al., 1978; Hardie,1979; Chou et al., 1996).

The current work has special relevance within the field ofphotoreceptor cell differentiation, in which there is growingevidence from other experimental systems that supports theidea that local interactions are involved in regulating theexpression of visual pigments in specific cells. In studies ofboth the chicken and goldfish retinas, it has been shown thatthe expression of a specific opsin in a cell is more closelyrelated to the cell’s position than to the cell’s birthdate (Bruhnand Cepko, 1996; Stenkamp et al., 1997). These results suggestthat extrinsic signals are intimately involved in thedetermination of photoreceptor cell fate and the specificationof photoreceptor cell opsin phenotype. Thus, our finding of alikely inductive signaling mechanism between the DrosophilaR7 and R8 photoreceptors cells, that is involved in thespecification of the opsin phenotype, may ultimately provide ameans to identify signaling molecules mediating similarsignals in other organisms and organ systems.

This work was supported by a grant (R01-EY10759) from theNational Eye Institute to S. G. B., and a grant from the EuropeanUnion (BMH4-CT97-2341) to R. P. The authors thank Anna Lazzellof the UTHSCSA Institutional Hybridoma Facility in the Departmentof Microbiology for generating the monoclonals against Rh4, Rh5 andRh6 used in this study; Peggy Miller in the Department of PathologyElectron Microscope facility for the Epon embedding and sectioningof ommatidial cross sections for light microscopy; and MarthaLundell at the University of Texas San Antonio for providing us withaccess to and assistance with their Bio-Rad confocal microscope. Wethank Rich Carthew, Ernst Hafen, Yash Hiromi, Todd Laverty, GerryRubin, Larry Zipursky and the Bloomington stock center for flystocks; and Charles Zuker and Ann Leslie for antibodies. The 4C5monoclonal antibody developed by de Couet and Sigmund wasobtained from the Developmental Studies Hybridoma Bankmaintained by the University of Iowa, Department of BiologicalSciences, Iowa City, IA 52242.

REFERENCES

Alvarez, C. E., Robison, K., and Gilbert, W.(1996). Novel Gq alpha isoformis a candidate transducer of rhodopsin signaling in a Drosophila testes-autonomous pacemaker. Proc. Natl. Acad. Sci USA93, 12278-12282.

Banerjee, U., and Zipursky, S. L.(1990). The role of cell-cell interaction inthe development of the Drosophila visual system. Neuron4, 177-187.

Bentrop, J., Schwab, K., Pak, W. L., and Paulsen, R.(1997). Site-directedmutagenesis of highly conserved amino acids in the first cytoplasmic loopof Drosophila Rh1 opsin blocks rhodopsin synthesis in the nascent state.EMBO J. 16, 1600-1609.

Brand, A. H., Manoukian, A. S., and Perrimon, N. (1994). Ectopicexpression in Drosophila. In Drosophila melanogaster: Practical Uses inCell and Molecular Biology(ed. L. S. B. Goldstein and E. A. Fyrberg), pp.635-654. Academic Press, San Diego, CA.

Bruhn, S. L., and Cepko, C. L. (1996). Development of the pattern ofphotoreceptors in the chick retina. J. Neurosci. 16, 1430-1439.

Campos-Ortega, J. A., Jurgens, G., and Hofbauer, A.(1979). Cell clonesand pattern formation: Studies on sevenless, a mutant of Drosophilamelanogaster. Wilhelm Roux’s Archives186, 27-50.

Carthew, R. W., and Rubin, G. M. (1990). seven in absentia, a generequired for specification of R7 cell fate in the Drosophila eye. Cell 63,561-77.

Chou, W. H., Hall, K. J., Wilson, D. B., Wideman, C. L., Townson, S. M.,Chadwell, L. V., and Britt, S. G. (1996). Identification of a novelDrosophila opsin reveals specific patterning of the R7 and R8 photoreceptorcells. Neuron17, 1101-1115.

Cowman, A., Zuker, C., and Rubin, G.(1986). An opsin gene expressedin only one photoreceptor cell type of the Drosophila eye. Cell 44, 705-710.

Danscher, G. (1981). Localization of gold in biological tissue. Aphotochemical method for light and electronmicroscopy. Histochemistry71,81-88.

de Couet, H. G., and Sigmund, C.(1987). Monoclonal antibodies provideevidence that rhodopsin in the outer rhabdomeres of Drosophilamelanogaster is not glycosylated. Eur. J. Cell. Biol.44, 50-56.

Feiler, R., Bjornson, R., Kirschfeld, K., Mismer, D., Rubin, G. M., Smith,D. P., Socolich, M., and Zuker, C. S.(1992). Ectopic expression ofultraviolet-rhodopsins in the blue photoreceptor cells of Drosophila: visualphysiology and photochemistry of transgenic animals. J. Neurosci.12, 3862-3868.

Feiler, R., Harris, W. A., Kirschfeld, K., Wehrhahn, C., and Zuker, C. S.(1988). Targeted misexpression of a Drosophila opsin gene leads to alteredvisual function. Nature333, 737-741.

Flybase Consortium.(1998). FlyBase: a Drosophila database. Available fromhttp://flybase.bio.indiana.edu/. Nucleic Acids Res.26, 85-88.

Fortini, M. E., and Rubin, G. M. (1990). Analysis of cis-acting requirementsof the Rh3 and Rh4 genes reveals a bipartite organization to rhodopsinpromoters in Drosophila melanogaster. Genes Dev.4, 444-463.

Fortini, M. E., Simon, M. A., and Rubin, G. M. (1992). Signalling by thesevenless protein tyrosine kinase is mimicked by Ras1 activation. Nature355, 559-561.

Franceschini, N., Kirschfeld, K., and Minke, B. (1981). Fluorescence ofphotoreceptor cells observed in vivo. Science213, 1264-1267.

Fryxell, K. J., and Meyerowitz, E. M. (1987). An opsin gene that isexpressed only in the R7 photoreceptor cell of Drosophila. EMBO J. 6, 443-451.

Hardie, R. C. (1979). Electrophysiological analysis of fly retina. I:Comparative properties of R 1-6 and R 7 and 8. J. Comp. Physiol.129, 19-33.

Hardie, R. C. (1986). The photoreceptor array of the dipteran retina. TrendsNeurosci.9, 419-423.

Hiromi, Y., Mlodzik, M., West, S. R., Rubin, G. M., and Goodman, C. S.(1993). Ectopic expression of seven-up causes cell fate changes duringommatidial assembly. Development118, 1123-1135.

Hosoya, T., Takizawa, K., Nitta, K., and Hotta, Y.(1995). glial cells missing:a binary switch between neuronal and glial determination in Drosophila.Cell 82, 1025-1036.

Huber, A., Sander, P., and Paulsen, R.(1996). Phosphorylation of the InaDgene product, a photoreceptor membrane protein required for recovery ofvisual excitation. J. Biol. Chem.271, 11710-11717.

Huber, A., Schulz, S., Bentrop, J., Groell, C., Wolfrum, U., and Paulsen,R. (1997). Molecular cloning of Drosophila Rh6 rhodopsin: the visualpigment of a subset of R8 photoreceptor cells. FEBS Lett.406, 6-10.

616

Jan, Y. N., and Jan, L. Y.(1998). Asymmetric cell division. Nature392, 775-778.

Jones, B. W., Fetter, R. D., Tear, G., and Goodman, C. S.(1995). glial cellsmissing: a genetic switch that controls glial versus neuronal fate. Cell 82,1013-23.

Kirschfeld, K., Feiler, R., and Franceschini, N. (1978). A photostablepigment within the rhabdomeres of fly photoreceptors no. 7. J. Comp.Physiol.125, 275-284.

Kohler, G., and Milstein, C. (1975). Continuous cultures of fused cellssecreting antibody of predefined specificity. Nature256, 495-497.

Kramer, S., West, S. R., and Hiromi, Y.(1995). Cell fate control in theDrosophila retina by the orphan receptor seven- up: its role in thedecisions mediated by the ras signaling pathway. Development121, 1361-1372.

Lawrence, P. A., and Green, S. M.(1979). Cell lineage in the developingretina of Drosophila. Dev. Biol.71, 142-52.

Lindsley, D. L., and Zimm, G. G. (1992). Genome of Drosophilamelanogaster.Academic Press, San Diego.

Montell, C., Jones, K., Zuker, C., and Rubin, G.(1987). A second opsingene expressed in the ultraviolet sensitive R7 photoreceptor cells ofDrosophila melanogaster. J. Neurosci.7, 1558-1566.

O’Connell, P. O., and Rosbash, M.(1984). Sequence, structure, and codonpreference of the Drosophila ribosomal protein 49 gene. Nucl. Acids Res.12, 5495-5513.

O’Tousa, J. E., Baehr, W., Martin, R. L., Hirsh, J., Pak, W. L., andApplebury, M. L. (1985). The Drosophila ninaEgene encodes an opsin.Cell 40, 839-850.

Papatsenko, D., Sheng, G., and Desplan, C.(1997). A new rhodopsin in R8photoreceptors of Drosophila:evidence for coordinate expression with Rh3in R7 cells. Development124, 1665-73.

Pollock, J. A., and Benzer, S.(1988). Transcript localization of four opsin

genes in the three visual organs of Drosophila; RH2 is ocellus specific.Nature333, 779-782.

Ready, D. F., Hanson, T. E., and Benzer, S.(1976). Development of theDrosophila retina, a neurocrystalline lattice. Dev. Biol.53, 217-240.

Reinke, R., and Zipursky, S. L.(1988). Cell-cell interaction in the Drosophilaretina: the bride of sevenless gene is required in photoreceptor cell R8 forR7 cell development. Cell 55, 321-330.

Scavarda, N. J., O’Tousa, J., and Pak, W. L.(1983). Drosophila locus withgene-dosage effects on rhodopsin. Proc. Natl. Acad. Sci. USA80, 4441-4445.

Stenkamp, D. L., Barthel, L. K., and Raymond, P. A. (1997).Spatiotemporal coordination of rod and cone photoreceptor differentiationin goldfish retina. J. Comp. Neurol.382, 272-284.

Sved, J.(1986). eyes absent. Dros. Inf. Serv.63, 169.Van Vactor, D. L., Jr., Cagan, R. L., Kramer, H., and Zipursky, S. L.

(1991). Induction in the developing compound eye of Drosophila: multiplemechanisms restrict R7 induction to a single retinal precursor cell. Cell 67,1145-1155.

Wolff, T., and Ready, D. F.(1993). Pattern formation in the Drosophila retina.In The Development of Drosophila melanogaster, (ed. M. Bate and A. M.Arias), pp. 1277-1325. Cold Spring Harbor Laboratory Press, Plainview,New York.

Zipursky, S. L., and Rubin, G. M. (1994). Determination of neuronal cellfate: lessons from the R7 neuron of Drosophila. Annu. Rev. Neurosci.17,373-397.

Zuker, C., Montell, C., Jones, K., Laverty, T., and Rubin, G.(1987). Arhodopsin gene expressed in photoreceptor cell R7 of the Drosophila eye:homologies with other signal-transducing molecules. J. Neurosci.7, 1550-1557.

Zuker, C. S., Cowman, A. F., and Rubin, G. M.(1985). Isolation andstructure of a rhodopsin gene from D. melanogaster. Cell 40, 851-858.

W. H. Chou and others