rugose (rg), a Drosophila A kinase Anchor Protein, Is Required for Retinal Pattern Formation and Interacts Genetically With Multiple Signaling Pathways

Corresponding author: Tadmiri R. Venkatesh, City College of New York, 138th St. and Convent Ave., New York, NY 10031., venky{at}sci.ccny.cuny.edu (E-mail)

Communicating editor: T. C. KAUFMAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In the developing Drosophila eye, cell fate determination and pattern formation are directed by cell-cell interactions mediated by signal transduction cascades. Mutations at the rugose locus (rg) result in a rough eye phenotype due to a disorganized retina and aberrant cone cell differentiation, which leads to reduction or complete loss of cone cells. The cone cell phenotype is sensitive to the level of rugose gene function. Molecular analyses show that rugose encodes a Drosophila A kinase anchor protein (DAKAP 550). Genetic interaction studies show that rugose interacts with the components of the EGFR- and Notch-mediated signaling pathways. Our results suggest that rg is required for correct retinal pattern formation and may function in cell fate determination through its interactions with the EGFR and Notch signaling pathways.

IN multicellular organisms, cell fate specification and pattern formation are regulated by cell-cell interactions and mediated by the activation of specific signal transduction pathways. Genetic analysis in Drosophila has been instrumental in uncovering the various components of these signal cascades as well as the specific interactions between the components. The development of the retinal pattern of the Drosophila compound eye is particularly amenable to genetic dissection of pattern formation. The ommatidium, or "unit eye," is reiterated nearly 800 times in the compound eye. Each ommatidium contains eight specialized sensory neurons, the retinular photoreceptor cells (R1–R8), four cone cells, six pigment cells, and cells of the mechano-sensory bristle complex. The adult eye develops from a monolayer of epithelial cells, the eye imaginal disc. The anlagen for the eye imaginal disc arise as a group of about 20 cells in the embryo and proliferate through the first and second instar larval stages. Cellular pattern formation in the eye imaginal disc is marked by the anterior progression of a groove in the disc epithelium, known as the morphogenetic furrow (MF). As the MF progresses, cells within the furrow are induced to differentiate into photoreceptor cell clusters (READY et al. 1976 ; WOLFF and READY 1991 ). Pattern formation and cell fate determination in the developing eye are precisely coordinated with a stereotypical pattern of two mitotic cell cycles. The first mitotic domain is seen anterior to the MF where cells divide asynchronously and generate precursors for the photoreceptor cells R2, R3, R4, R5, and R8. The Drosophila TGF-ß homolog, DPP, is expressed in and promotes the G₁ arrest of cells in the MF (HORSFIELD et al. 1998 ). Immediately posterior to the MF, the second mitotic wave is seen as a compact band of S-phase cells and gives rise to the precursors for the cells R1, R6, R7, cone cells, pigment cells, and the cells of the bristle complex (WOLFF and READY 1991 ).

Early experiments using genetic mosaics led to the idea that cell fate specification in the developing eye is directed by cell-cell interactions rather than by strict lineage mechanisms (READY et al. 1976 ; LAWRENCE and GREEN 1979 ). Molecular genetic studies led to the characterization of several important signal cascades that regulate cell fate determination and pattern formation in the developing eye. Initially, the Sevenless (Sev) receptor tyrosine kinase (RTK)-activated signaling pathway was characterized as a pathway specific for the determination of the R7 cells and gave rise to the idea that unique signaling pathways may be involved in directing specific cell fates. Subsequently, it was shown that the Sev signaling pathway can also induce non-R7 cell fates when expressed ectopically in the appropriate precursor cells (ZIPURSKY and RUBIN 1994 ). Furthermore, another signal cascade mediated by EGFR (epidermal growth factor receptor) can direct the specification of an R7 cell fate in the absence of the Sevenless function. In the developing eye, activation of EGFR can direct the determination of all cell types (FREEMAN 1996 ; TIO and MOSES 1997 ) and the EGFR signal cascade is involved in a number of cellular processes including MF initiation, cell proliferation, cell spacing, and cell survival (FREEMAN 1997 ; DOMINGUEZ et al. 1998 ; LESOKHIN et al. 1999 ; FREEMAN and BIENZ 2001 ). EGFR-activated signaling pathway is involved in myriad developmental processes, including wing vein formation, midline-glial differentiation and establishment of axial polarity in the follicle cells (PRICE et al. 1989 ; KLAMBT et al. 1991 ; CLIFFORD and SCHUPBACH 1992 ; RAZ and SHILO 1992 , RAZ and SHILO 1993 ; NEUMAN-SILBERBERG and SCHUPBACH 1993 ; STURTEVANT et al. 1993 ; XU and RUBIN 1993 ; CLIFFORD and SCHUPBACH 1994 ; FREEMAN 1994B ; SCHUPBACH and ROTH 1994 ; ROTH et al. 1995 ). Both Sevenless and EGFR transduce signals via a common evolutionarily conserved downstream signal cascade, the RAS1-mitogen-activated protein kinase (MAPK) cascade, which includes RAS1, RAF, and MAPK kinase (MEK) and MAPK. Signals transduced through this protein phosphorylation cascade lead to the activation of transcription factors, which in turn regulate specific downstream targets in various cell types (DICKSON et al. 1992 ; BRUNNER et al. 1994 ; FLORES et al. 1998 , FLORES et al. 2000 ).

The signaling pathway activated by Notch (N), also plays many important roles during eye development. In the developing eye, Notch is involved in cell proliferation, cell fate determination, establishment of dorsoventral axis formation, and establishment of ommatidial polarity (CAGAN and READY 1989B ; FORTINI et al. 1993 ; FORTINI and ARTAVANIS-TSAKONAS 1994 ; PAPAYANNOPOULOS et al. 1998 ; COOPER and BRAY 1999 ). N is activated by the transmembrane ligand, Delta (Dl). Activated N, in combination with the transcriptional regulator, Suppressor of Hairless [S(H)], activates downstream targets. In addition to its role in eye development, N signaling is critical for a large array of cell-cell interactions and cell fate decisions during Drosphila development (reviewed by ARTAVANIS-TSAKONAS et al. 1999 ).

Genetic studies in Drosophila have shown that cAMP and PKA (A kinase or protein kinase A)-mediated signaling is required in a variety of processes, including oogenesis, establishment of tissue polarity in the embryo, imaginal disc morphogenesis, and synaptic function (LANE and KALDERON 1993 , LANE and KALDERON 1995 ; DAVIS et al. 1996 , DAVIS et al. 1998 ). In the developing eye, PKA plays an important role in the initiation of pattern formation and morphogenesis through its interactions with Hedgehog (Hh), DPP, and Wingless (Wg; HEBERLEIN and MOSES 1995 ; HEBERLEIN et al. 1995 ; PAN and RUBIN 1995 ; BURKE and BASLER 1996 ; DOMINGUEZ and HAFEN 1997 ; PIGNONI and ZIPURSKY 1997 ; HUANG and KUNES 1998 ). Signaling by Sev, EGFR, N, and the cAMP-PKA pathways leads to the activation of a variety of downstream transcription factors. Because these signaling pathways regulate a multitude of cellular functions in different tissues and due to the number of transcription factors involved, it became apparent that a single signaling pathway alone could not specify particular cell fates. Recent studies have demonstrated that the determination of specific cell fates involves the integration of inputs from multiple pathways and a combinatorial code of activated transcription factors generates specific cell types (FLORES et al. 2000 ; HALFON et al. 2000 ; SIMON 2000 ; XU et al. 2000 ; TOMLINSON and STRUHL 2001 ).

In this article, we present the isolation, phenotypic analysis, and genetic and molecular characterization of the rugose (rg) locus. Our results show that rugose gene function is required for correct retinal pattern formation and mutations at the rugose locus affect ommatidial organization, cone cell differentiation, and pigment cell number. Molecular analysis suggests that rg encodes a Drosphila A kinase anchor protein (DAKAP 550). A kinase anchor proteins (AKAPs) are important components of the cAMP-PKA-mediated signal transduction pathway. AKAPs modulate the specificity of PKA function by targeting and compartmentalization of PKA to specific subcellular structures (PAWSON and SCOTT 1997 ; SCOTT and PAWSON 2000 ). Studies on mammalian systems have led to the identification and molecular characterization of a large number of AKAPs, which mediate targeting of the PKA to various cellular organelles (GRAY et al. 1998 ; EIDE et al. 1998 ; VO et al. 1998 ). Tissue and cell type-specific isoforms of AKAPs have recently been characterized (DONG et al. 1998 ). In Drosophila, biochemical characterization and molecular cloning led to the identification of two AKAPs, DAKAP 550 and DAKAP 200; however, no physiological roles have been reported for these proteins (HAN et al. 1997 ; LI et al. 1999 ). Our data from the genetic and phenotypic analyses of rugose mutants implicate DAKAP 550 in retinal pattern formation. We also present data on the genetic interactions of rugose, which suggest that rugose interacts with components of the signaling pathway mediated by the EGFR and N. While this article was in preparation, SCHREIBER et al. 2002 identified rugose in a genetic screen for modifiers of Hairless (H), a Notch pathway antagonist and report that rugose interacts with EGFR and N signaling pathways.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Flies were cultured on standard cornmeal agar medium and maintained at 24°.

Drosophila stocks used in this study:
argos, argos^sty1/TM3, argos²⁵⁷, argos^sty2/TM6B, hs-argos#4; Argos modifiers, soba^sal/TM3, bul^6d7/TM6, 1040^{bul d}/TM6; ras1 (loss-of-function), sev^d2;E(e.1.B)th st ca/T21A; Ras^N17(dominant negative), isogenic, sev^d2;E(e.1.B)/T21A, w¹¹⁸;P[w+,sev hs-Ras^N17], A/Cyo, w¹¹⁸;P[w+,sev hs-Ras^N17], C/CyO, W¹¹⁸;P[w+,sev, hs-Ras^N17], E/CyO, ras^v12, CRI, Cyo,dp,pr,cn, P[ry, sev-ras1^v12]/Sco, CR2, CyO, P[sev-ras^v12]/, Sco, TM3, P[ry,sev-ras^v12]/, e,ftz, ry, Star, S²/CyO, S⁶⁴/CyO, S^11N23,cn, bw, sp/CyO, w₁S^hs8/CyO, cn bw/SM6a 1KE 4.28.1, cn bw/SM6a 1V KE 2.4.2, cn bw/SM6a VIII KE 4.1.1, VII 30 E1 cn bw/SM6a, E3X12X cn bw/SM6a, cn bw/SM6a VII KE3.22, cn bw/SM6a II KE 1.0.4, cn bw/SM6a VIII KE, 3.6. C, II.8.E1 cn bw/SM6a IX, KE 4.0.1, E3R12X cn bw/SM5, E3R34Xcn bw/SM5, E3RIIX cn bw/SM5, rolled, lt rl[1], rl[1], rl^ems64/SM1, rl^su23/CyO, deadhead, w,dhd^p8, DER, pn cn top¹ bw/CyO, co stw pwn/ Cyo, 2C82ICyO, IP02/Cyo, b pr tor^RL3cn/CyO, Y9, cn bw/SM1, Elp/Cyo; RII (PKA regulatory subunit), Df(2R)X¹/CyO, sparkling, w; spa^pol/spa^pol, Delta, Dl^RFTM6c, Dl(3R)^M2/TM6c, Dl^9B39es ca/TM6a, ssDl^6B37e/TM6c, Su(H)(12)/In(2LR)Cy, Cy[1]pr[1]; Hairless, H^E31/TM6B, H^c23/TM6b; rg duplication, ywf,yw,snf210; RDUP(w+)/+, w/w;2-3SbTM3/Drop. Deficiency stocks were from Bloomington Stock Center; rugose, rg¹/rg¹ was from Bowling Green Stock Center.

Mutagenesis and isolation of new rg alleles:
New alleles of rg were isolated by their failure to complement the rough eye phenotype of the hypomorphic rg¹ allele (LINDSLEY and ZIMM 1992 ). Three- to four-day-old Canton-S (CS) males were either irradiated with 1000 rad of -rays or treated with EMS (ASHBURNER 1989 ) and mated en masse to virgin Df (1) JC70/FM7 females. Non-FM7, F₁ females were screened for rough eyes under a dissecting microscope. F₁ females with rough eyes were individually mated to FM7/Y males and maintained for several generations as a stock. The rough-eyed males from this stock were tested for complementation with rg¹.

Isolation of P-element-induced alleles of rugose:
P-element-induced alleles of rugose were isolated by remobilizing a P element from the adjacent deadhead (dhd) locus. The P element in the dhd locus was remobilized employing the 2-3 transposase (ROBERTSON et al. 1988 ) in a hybrid dysgenic cross as described by SALZ et al. 1994 . dhd ^p2 males were mated to virgin C(1)Dx y f; Sb2-3/+ females. The F₁ progeny were mated inter se and F₂ males were screened for rough eyes. Rough-eyed males were tested for failure of complementation with the rg⁶ allele. P-element-excision-induced revertants of rugose were isolated following a second hybrid dysgenic cross as per ROBERTSON et al. 1988 .

Deficiency mapping:
The rg alleles were mapped to the 4E-F region of the X chromosome using the following deficiency stocks: Df (1) A113, Df (1) RC40, Df (1) JC70, Df (1) ovoG6, Df (1) svbEh, Df (1) DEB4D, and Df (1) GA56. Three-day-old rg males were crossed to virgin females carrying the specific deficiency chromosome over the FM7 balancer. The eye phenotype of the F₁ females with rg chromosome over the deficiency chromosome was scored.

Electron microscopy:
Tissues were prepared for transmission electron microscopic studies as described by LONGLEY and READY 1995 . Scanning electron microscopy was done as described by KARPILOW et al. 1989 .

Light microscopy:
The cobalt sulfide staining procedure as described by WOLFF and READY 1991 was used to visualize the apical surface of 50-hr pupal eyes. Eyes were dissected in 2% glutaraldehyde in PBS buffer for 10 min and then washed briefly in distilled water and transferred into 5% cobalt nitrate for 15 min. The tissues were rinsed in water for 5 sec and then transferred to a drop of 2% ammonium sulfide for 5 sec. The tissues were washed in distilled water for 5 min and mounted in glycerol. Slides were viewed using a Zeiss Axiophot microscope.

General procedures:
Standard molecular cloning techniques were performed according to SAMBROOK et al. 1989 with modifications as described in TENG et al. 1991 . Drosophila genomic DNA isolation was done according to the method described by TENG et al. 1991 .

RNA isolation and RT-PCR analysis:
RNA was isolated from whole flies using the Quantum Preps Master blaster RNA extraction reagent from Bio-Rad (Richmond, CA; catalog no. 732-6390). RT-PCR was done using the Retroscript kit (catalog no. 1710) from Ambion (Austin, TX). For RT-PCR reactions 2 µg of total RNA was used per reaction. The oligonucleotide primers used were: (1) RgS1: 5'-CTTGTCCTCACCTCAATCAA-3' and (2) RgAs4: 5'-AACTGCTGTTGCTGACCG-3'.

Actin expression was monitored as a positive control in RT-PCR experiments using the primers: (1) Actin S1: 5'-TGTTCGCAGCCACTAACC-3' and (2) Actin AS1: 5'-CCCATCTCTGTCCCAATCC-3'

Immunohistochemistry:
Late third larval instar eye discs and 60-hr pupal eyes were stained with anti-Cut antibody (BLOCHLINGER et al. 1993 ). Eye discs were dissected in PBS and fixed in 4% paraformaldehyde and then incubated in anti-Cut antibody for 3 hr. The tissues were washed in PBS and incubated in FITC-conjugated goat anti-mouse IgG for 2 hr and then washed in PBS and mounted in 90% glycerol containing 0.01% phenylenediamine. Slides were viewed using a Molecular Dynamics (Sunnyvale, CA) confocal microscope.

Molecular cloning of the rugose gene:
We used the following strategy for the molecular cloning of DNA from the rugose locus: We first mapped rugose to the chromosomal location 4E-4F by recombination and deficiency mapping. We next identified several P1 clones that mapped to the 4E-F region from the Drosophila genome project database. These clones contain a large insert (80 kb) of the Drosophila genomic sequence in the P1 phage vector (SMOLLER et al. 1991 ; PIERCE and STERNBERG 1992 ). The DNA from the entire P1 clone was digested with EcoRI restriction enzyme and used as a probe on Southern blots to look for restriction enzyme fragment length polymorphisms (RFLPs). The blots contained restriction digests of the total genomic DNA from the different mutant alleles of rugose isolated by us and DNA from the parental strains. A single P1 clone (FlyBase no. 82-63) detected RFLPs in the DNA from the mutant alleles. We identified a 3.5-kb EcoRI DNA fragment from the digests of P1 clone no. 82-63. This genomic DNA fragment is designated RGD1 and detects RFLPs on Southern blots when hybridized to genomic DNA from several rg mutants including EMS, -ray, and P-element-induced rugose alleles.

Isolation of rg cDNAs:
The 3.5-kb genomic fragment RGD1 was used as a probe to isolate cDNAs from the rugose locus. Using the RGD1 as a probe, we screened 10⁶ recombinant phage from an eye disc cDNA library in the phage gt10 (gift of Dr. G. Rubin, University of California, Berkeley) and isolated seven cDNA clones derived from the rg locus.

DNA sequence analysis of the rugose cDNA:
We determined the complete nucleotide sequence of the cDNA RgcD10 on both strands. The cDNA insert was isolated from the phage vector and subcloned into the Bluescript KS+ vector (Stratagene, Burlingame, CA) and sequenced by automated DNA sequencer at the Yale-Boyer DNA sequencing facility.

Expression of mRNA from the rugose locus:
To study the expression patterns of the rugose mRNA, we synthesized digoxygenin-UTP-labeled single-stranded probes. Sense and antisense probes were synthesized from the cDNA RgcD10 cloned into the Bluescript vector (Stratagene, KS+). Hybridization was done according to the method of TAUTZ and PFEIFLE 1989 and probes were hybridized in situ to whole mounts of embryos and developing eye discs. Sense strand probes showed very low levels of hybridization, which represented background levels.

Genetic interactions of rugose:
To identify the genes that interact with rg, we initially used stocks carrying autosomal deficiencies (Indiana University, Bloomington Stock Center) to uncover dominant interactions in an F₁ screen. rg/rg females were mated to males carrying the autosomal deficiency over a balancer. Nonbalancer F₁ males were screened for either enhancement or suppression of the rough eye phenotype of rg. Once a chromosomal region was identified, we searched the fly database to locate any known genes within the regions uncovered by the deficiency. We tested multiple alleles of the identified autosomal gene for its interactions with the rg alleles. For each interaction, multiple alleles of rg and the interacting mutations were tested (see Table 4).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation and characterization of rg mutants:
Although several alleles of the rg mutation are described in LINDSLEY and ZIMM 1992 , all but rg¹ are extinct. rg¹ is a hypomorphic allele and the rough eye phenotype is temperature sensitive. At 17° the eyes are almost normal with a very slight rough appearance, whereas at 25° the eyes are moderately rough. When flies are grown at 29° the eyes become severely rough. We have isolated a number of rugose mutant alleles using EMS, -rays and P elements as mutagenic agents. New rg alleles were isolated on the basis of their failure to complement the rg¹ allele. Table 1 summarizes the phenotypes of the various rg mutant alleles. The severity of the eye phenotype varies in the different alleles. On the basis of the degree of roughness of the eye we have classified the mutant eyes into mild, moderate, and severe classes. From our mutagenic screens, we have isolated several alleles that have severe eye phenotypes and behave as genetic nulls. Hypomorphic alleles or partial loss-of-function mutations show mild or moderate roughness of the eye. In the -ray and EMS mutagenesis, we screened 60,000 and 35,000 F₁ progeny, respectively. We also isolated six P-element-induced alleles of rg from screening 30,000 F₂ progeny.

rg^p2 and rg^p5 are insertion mutations:
P-element- induced mutations can be due to either insertions or deletions caused by imprecise excision of an element. Remobilization of the P element, resulting in the reversion of the mutant phenotype, can identify insertions. We carried out reversion analysis of the two P-induced alleles, rg^P2 and rg^P5. The data from the reversion analysis are presented in Table 2. The revertants show restoration of the cone cell number as well as the smooth eye characteristic of the wild-type eye (data not shown). The remobilization of the P element also yielded several partial revertants consistent with imprecise excision of the inserted P element. These results indicate that rg^P2 and rg^P5 are insertion mutants.

Phenotype of rugose mutants:
The wild-type Drosophila eye is a hexagonal array of about 800 ommatidia. This arrangement of the ommatidia gives a smooth appearance to the external surface of the eye (Fig 1A). Any perturbation of the reiterated pattern of the ommatidia results in a rough surface of the eye. In the wild-type eye each ommatidium contains eight photoreceptor cells (R cells), which occupy stereotypic positions (Fig 1C). The plasma membrane of the R cells is multiply folded to form a tightly compacted array of microvilli known as the rhabdomere. The rhabdomeres of the cells R1–R6 are arranged in an asymmetric trapezoidal manner and extend the length of the retina. The rhabdomeres of the two central cells, R7 and R8, are smaller in profile and shorter in length than the R1–R6 rhabdomeres. They are positioned in the same axial plane as the distal R7 rhabdomere, on top of the more basal R8 rhabdomere. The lens-secreting cone cells are found on top of the apical surface of the R cells. The cone cells extend into thin bag-like structures between the R cells and terminate at the floor of the retina where their feet contribute to the formation of the fenestrated basement membrane. The pigment cells ensheathe the R cells and the cone cells. The anterior and posterior primary pigment cells form a cup-like structure around the cone cells. The secondary and tertiary pigment cells form a hexagonal lattice enclosing the ommatidia within. Each ommatidium shares six secondary pigment cells and three tertiary cells with its neighbors. The feet of the secondary and tertiary pigment cells flatten into the fenestrated membrane and form the floor of the retina (TOMLINSON 1985 ; CAGAN and READY 1989A ). A scanning electron micrograph of the compound eye of a rg null allele (rg⁶) is presented in Fig 1B. The eyes of rg mutants are characterized by a rough external surface. The ommatidial facets of the rg mutants are irregular and not hexagonal as in the wild-type eyes. In addition, the mechano-sensory bristles are irregularly positioned and often multiple bristles are seen in place of a single bristle in each corner of the hexagonal facet seen in the wild type. Frequently in rg mutant eyes, the ommatidia exhibit abnormal numbers of R cells. The precise positions of the R cells are altered and the pigment cell lattice loses its hexagonal arrangement leading to the disorganization of the retina (Fig 1D). The missing retinal photoreceptors can be seen in the region below the fenestrated basement membrane as seen in transverse sections of the compound eyes (Fig 1F and Fig H). This mispositioning of the R cells may be due to the defects in the fenestrated basement membrane caused by the abnormal number and organization of the cone cells and the disorganized pigment cell lattice.

View larger version (71K): In this window In a new window Download PPT slide   Figure 1. Mutations in rugose result in a rough eye phenotype. (A) Scanning electron micrograph showing a smooth phenotype of the wild-type [Canton-S (CS)] compound eye. (B) Compound eye from rugose mutant (rg6), showing a rough eye phenotype. Anterior is to the right. Bar, 25 µm. rugose mutants have disorganized retinas. (C) Tangential section through a wild-type eye shows a highly ordered and repetitive ommatidial pattern. In D, a tangential section through rg6 mutant eye shows disorganized ommatidial pattern with aberrant R cell number and position. The pigment cell lattice is abnormal. Bar, 30 µm. (E–H) Transverse sections of wild-type (E and G) and rg mutant (F and H) compound eyes. Note the disrupted fenestrated membrane in the rg6 mutant retina (arrows in G and H). G and H show higher magnification views of the basement membrane regions revealing collapsed photoreceptor cells in the lamina region of the optic ganglion. R, retina; L, lamina; M, medulla; fm, fenestrated basement membrane. Bars, 40 µm (A and B) and 25 µm (G and H).

rugose mutants have abnormal numbers of cone cells:
To understand the cellular basis of the rough eye phenotype of the rg mutants we examined the number and organization of the cone cells in wild type and rg mutants. The number of cone cells can be determined by staining pupal stage eyes with cobalt sulfide (WOLFF and READY 1991 ). We have determined the cone cell number and organization in the various rugose mutant alleles. In the wild type, four cone cells are apical to the R cells and the individual cone cells can be identified by their specific position. On the basis of their positions, the four cone cells can be identified as an anterior cone cell, a posterior cone cell, an equatorial cone cell, and a polar cone cell (Fig 2A). The eyes of rg mutants show abnormal numbers of cone cells when compared to the wild type. The number of cone cells varies in the different mutant alleles and is dependent on the level of the residual rg gene function. Partial loss-of-function alleles such as rg¹⁰ show two or three cone cells in the majority of the ommatidia (Fig 2C), while strong alleles that behave as genetic nulls have either one or no cone cells in the majority of the ommatidia (Fig 2B). Some of the rg mutant alleles also exhibit aberrant numbers of secondary/tertiary pigment cells (Fig 2C).

View larger version (109K): In this window In a new window Download PPT slide   Figure 2. Cobalt sulfide staining and Cut expression reveals abnormal number of cone cells in rugose mutants. (A) Wild-type (CS) 60-hr pupal eye showing a complement of four cone cells (c) in each ommatidium. (B) In the null allele rg3 a majority of the ommatidia have no cone cells and a small number of ommatidia show a single presumptive cone cell surrounded by two primary pigment cells (arrow). (C) The rgp5 allele shows mostly one or two cone cells and an abnormal number of secondary pigment cells. Bar, 50 µm. (D–F) Cut expression in early pupal stage eyes reveals aberrant cone cell determination in rugose mutants. In the wild-type pupal eyes (D), four Cut positive cells are seen. (E) In the rugose hypomorphic allele rgg10, developing ommatidia show one or two Cut positive cells. (F) In the rg3 mutant, a strong allele, a majority of the cells fail to express Cut antigen. The weakly staining cells in C are likely to represent residual cone cells in the mutant allele. Bar, 30 µm.

Cut expression in rugose mutants:
Next, we used Cut protein expression to identify the differentiating cone cells in the developing eye. Cut protein is a transcription factor and its expression in the developing eye is specific to cells that assume a cone cell fate (BLOCHLINGER et al. 1993 ). To localize cone cells in the developing eye, 60-hr pupal eyes were stained with the anti-Cut antibody. In the wild-type pupal eye, four cone cells are labeled with the anti-Cut antibody in each ommatidium (Fig 2D) whereas the hypomorphic rugose allele rg¹⁰ shows either one or two Cut positive cells (Fig 2E). In the null allele, rg⁶, no Cut positive cells are seen showing the lack of differentiating cone cells (Fig 2F).

Cone cell development is abnormal in rugose mutants:
To ascertain if the aberrant number of cone cells seen in rg mutants was due to failure of cone cell differentiation, we examined developing pupal stage eyes using transmission electron microscopy. The results are presented in Fig 3. In the developing eye, the R7 cell is the last photoreceptor to differentiate, resulting in a mature eight-cell cluster. The cone cells differentiate during the late third larval instar stages of development. Once the R7 cell nucleus completes its apical migration, two cone cell nuclei move up from the basal region and flank the R cell cluster. The cone cells spread over the apical tips of the photoreceptor cells and become the anterior and posterior cone cells (Fig 3A). These cone cells contact each other centrally above the R cells. Subsequently, two more cells, the equatorial and polar cone cells, are added resulting in four cone cells. During the early pupal stages, in a wild-type eye a stereotypic arrangement of the R cells and the cone cells can be seen at the four-cone cell stage (Fig 3A). In rugose mutants, cone cell specification is abnormal and consequently some of the cone cells fail to differentiate. In Fig 3B and Fig C, transmission electron micrographs show sections through developing eyes of different rg mutant alleles at the four-cone cell stage. In the hypomorphic allele, rg¹⁰, a single cone cell is seen (Fig 3B), while in the null allele rg³ no cone cells are seen (Fig 3C). In all of the mutant alleles tested, an undifferentiated precursor cell occupies the position of the cone cell and fails to differentiate.

View larger version (130K): In this window In a new window Download PPT slide   Figure 3. Electron microscopic study of cone cell (A and B) and R cell development (D and E). Transmission electron micrographs show abnormal development of cone cells in rugose mutants. (A) Wild-type early pupal stage developing eye at the four-cone cell stage. In the wild type, the four cone cells occupy their characteristic positions in relation to the photoreceptor cells (R1–R8). Note at this stage all the R cells except R4 touch the central R8 cell. (B) In the rg10 hypomorphic allele, a single anterior cone cell is seen. A small portion of the edge of the neighboring ommatidium is seen at the upper left. (C) In the developing ommatidium of the rugose null allele rg3 at the same stage all of the four cone cells fail to develop. The section is slightly more apical than the sections in A and B. The R cells (R1–R8) are visibly unaltered and occupy their respective positions. acc, anterior cone cell; pcc, posterior cone cell; ecc, equatorial cone cell; plcc, polar cone cell. Stars in C indicate the positions of the presumptive cone cells that failed to differentiate. Bar, 0.5 µm. (D and E) Photoreceptor cell determination is unaffected in rugose mutants. Transmission electron micrographs of developing late third instar eye disc from wild type (D) and the rg3 mutant (E) show that the normal development of the R cells is unaffected in the rugose mutant. Bar, 0.5 µm.

rugose mutants show normal photoreceptor cell specification:
To determine if mutations at the rugose locus affect R cell differentiation, we followed the sequential specification of the R cells in the third instar eye discs. Transmission electron microscopic analysis showed normal differentiation of R cell clusters in both wild-type and rugose mutant eye discs (Fig 3D and Fig E). In addition, immunochemical experiments involving a variety of R cell-specific probes also showed normal specification of the R cells in the developing third instar eye imaginal discs (data not shown). This further supports our conclusion that rugose mutations do not affect R cell differentiation.

rugose mutants are pleiotropic:
In addition to phenotypes in the eye, rugose mutants exhibit wing vein defects and embryonic lethality with partial penetrance. The penetrance is variable depending on the particular rugose allele. Loss of rugose function leads to an incomplete wing vein L5 (not shown). rugose mutants also exhibit an embryonic semilethal phenotype. The extent of lethality varies with the strength of the rugose allele. In the null alleles rg⁶ and rg³, for example, 20–40% of the fertilized eggs fail to hatch, suggesting defects in the embryonic stages. The cellular and molecular basis for the lethality is not clear at this point. The pleiotropic nature of rugose mutations is consistent with the multiple roles of cAMP-PKA-mediated signaling in Drosophila.

Molecular cloning of DNA from the rugose locus:
To clone the DNA sequences from the rugose locus we identified a 3.5-kb EcoRI genomic DNA fragment (designated RGD1), from the P1 genomic clone (FlyBase ID no. 82-63, see MATERIALS AND METHODS). We used the 3.5-kb RGD1 fragment as a probe on Southern blots containing genomic DNA from wild-type parental strains and various rugose mutant alleles. The results are shown in Fig 4A. As expected, the genomic DNA from the wild type (CS) and the dhd^p8 showed a 3.5-kb band (Fig 4, lanes 10 and 11). The dhd gene resides in the 4E-F region of the X chromosome and is unrelated to rugose. The dhd^p8 was a parental strain for the rugose P alleles. The band is slightly larger in the dhd^p8. The genomic DNA from various rg mutant alleles showed restriction fragment length polymorphisms with bands ranging from 5 to 11 kb (Fig 4, lanes 1–9), suggesting that the RGD1 was derived from the rugose locus.

View larger version (92K): In this window In a new window Download PPT slide   Figure 4. (A) Localization of rugose locus by Southern analysis. Wild-type and rg mutant genomic DNA were digested with EcoRI and the resulting fragments were separated by electrophoresis, transferred to nytran, and subsequently probed with RGD1, the 3.5-kb EcoRI fragment from the P1 clone. The genomic DNA in the lanes shown is from flies of the following strains: lane 1, rg11; 2, rg10; 3, rg9; 4, rg8; 5, rg7; 6, rg1; 7, rg1; 8, rgp9; 9, rgp6; 10, dhdp8; 11, Canton-S (CS) males. A 3.5-kb band is detected in the genomic DNA from wild-type (CS) parental strain and in the dhdp8 strain. The dhd strain was the parental strain for the P alleles of rg. The dhd mutation maps close to rg in the 4E-F region of the X chromosome but is unrelated to rugose. The genomic DNA from the rg mutant strains shows polymorphisms. (B) RT-PCR analyses of transcripts from wild type and rugose mutants. Poly(A)+ RNA was isolated from whole flies, and PCR primers that yield an expected 1.2-kb product were used in the RT-PCR reactions. In the null allele, rg6, no rg PCR product is seen (arrowhead, lane 6), but actin expression is normal (lane 14). Other mutant alleles show altered patterns. Lane 1, kb ladder; 2, liver RNA control (430 bp); 3, CS; 4, rg1; 5, rg5; 6, rg6; 7, rg; 8, rg9; 9, kb ladder; 10, liver control; 11, CS; 12, rg1; 13, rg5; 14, rg6. The arrowhead under markers in lane 8 shows the strong band of the actin product (630 bp) that was used as an internal control.

Isolation of cDNAs from the rg locus:
We used the 3.5-kb RGD1 genomic fragment as a probe to isolate rugose cDNA clones. We isolated seven cDNA clones from screening an eye disc-specific cDNA library and the longest among them is cDNA clone RgcD4, which contains an insert that is 3.0-kb in length. The seven cDNA clones fall into two classes based on Southern analysis data. The first group of cDNA inserts from the cDNA clones, RgcD2, RgcD4, RgcD9, and RgcD10, hybridize to each other on Southern blots and detect RFLPs on Southern blots of the genomic DNA from several rg mutant alleles. These results suggested that the cDNAs were derived from the rg coding regions. The cDNAs, RgcD5 and RgcD8, compose the second group and do not cross-hybridize with the cDNAs from the first group.

rugose encodes a Drosophila A kinase anchor protein:
We determined the DNA sequence of the 1.9-kb cDNA clone RgcD10 and performed a BLAST analysis to look for homologies to other known sequences in the DNA sequence database. Our data analyses showed that the RgcD10 cDNA sequence is identical (99–100%) to the Drosophila A kinase anchor protein, DAKAP 550 (data not shown). HAN et al. 1997 first reported the cloning and sequence of a 7.2-kb partial cDNA for DAKAP 550. The complete sequence of the DAKAP 550 has been reported in the annotation of the Drosophila genome by the Drosophila genome project (http://www.fruitfly.org; gene ID no. CG6775). The DAKAP 550 gene is composed of 26 exons in a 29-kb transcription unit and the complete cDNA is 11.2 kb in size. Our DNA sequence data analysis shows that the 1.9-kb cDNA RgcD10 is derived from exons 4–10 of DAKAP 550.

Transcript analysis at the rg locus:
To determine whether the mutational lesions in rugose mutant alleles was reflected in transcriptional changes, we analyzed the RNA from wild type and rg mutants using RT-PCR. We used primers that were expected to generate a RT-PCR product of 1.2 kb and our results from these experiments are shown in Fig 4B. In wild type (lane 3), a PCR product of 1.2 kb is seen, whereas the RNA from the rg⁶ allele (lane 6) shows no 1.2-kb PCR product band. This supports the genetic data that show that rg⁶ is a null allele. In the other rg mutant alleles the pattern is polymorphic, suggesting that the mutational lesions result in altered transcript profiles. The same RNA preparation showed a strong band in all the rg mutant alleles and the wild type with the actin primers used as a control (lanes 11–14).

Expression of rugose mRNA in the embryo:
To follow the expression of rugose mRNA we used synthesized antisense single-strand probes from the RgcD10 cDNA. The expression of rugose mRNA was dynamic during development and was detected in all the embryonic stages. The early expression at stage 5 showed enrichment anteriorly and expression was also seen ventrally in the posterior regions. The late stage 5 showed strong expression in the region of the cephalic furrow formation. During stage 6, the anterior localization of the message is strong including the cephalic furrow region. At stage 7, rg expression becomes maximized in the region of the anterior midgut and persists through stage 8. During the later stages of development (stages 13–17), the expression is pronounced in the developing nervous system (Fig 5C). The RgcD10 expression was also seen in the antenna-maxillary complex. In the developing eye imaginal disc, RgcD10 was expressed throughout the disc and in the region of the morphogenetic furrow (Fig 5E and Fig F). The expression of the RgcD10 transcript is highly reduced or absent in the rg mutant eye discs (Fig 5G). The rg transcript expression patterns seen in our experiments are similar to the rg expression pattern reported by SCHREIBER et al. 2002 and consistent with the DAKAP expression patterns reported by HAN et al. 1997 .

View larger version (113K): In this window In a new window Download PPT slide   Figure 5. Expression of rugose mRNA in embryos and eye imaginal disc. Single-stranded digoxygenin-labeled probes were synthesized from the cDNA RgcD10 and were hybridized in situ to whole mounts of embryos and developing eye discs. The early expression at stage 5 is shown in A. In B, the rugose mutant rg6 of a comparable stage 5 embryo is shown and expression is greatly reduced compared to wild type (A). The residual expression may be due to maternal contribution. During the later stages of development (stages 13–17), the expression is pronounced in the developing nervous system (C). A comparable stage embryo from the rugose mutant allele rg6 shows highly reduced expression compared to background levels (D). Sense strand probes showed very low levels of hybridization and represented background levels (not shown). In the wild-type developing eye imaginal disc, the rugose mRNA is expressed throughout the disc (E and F). Stronger expression is seen in the region of the morphogenetic furrow (arrow in F) and the disc margins. Posterior to the morphogenetic furrow, a low level of expression is seen in the differentiating photoreceptor cells. In the eye disc of the rg mutant rg6, no expression is seen (G). Bars, 80 µm.

Genetic interactions of rugose:
Genetic interactions uncover interrelationships between components of a cellular pathway or interactions between different cellular pathways. Genetic modifier screens offer a simple and powerful method of identifying genetic interactions of the gene of interest (CHANUT et al. 2000 ; HUANG and RUBIN 2000 ; REBAY et al. 2000 ; THERRIEN et al. 2000 ). We have utilized the rough eye phenotype of the rg mutants to identify the genetic interactions of rugose. We first conducted a dominant modifier F₁ screen by using autosomal deficiencies from the Bloomington Stock Center deficiency kit. We placed the hemizygous rg males in double mutant combination with a single copy of the autosomal deficiency and scored for either suppression or enhancement of the rough eye phenotype. We screened a total of 118 autosomal deficiency stocks (51 on chromosome II and 68 on chromosome III) and covering 60% of the autosomes. The deficiency chromosomes that had a modifier effect on the rough eye phenotype are shown in Table 3. Genes that mapped within the deficiency breakpoints were identified (FlyBase and LINDSLEY and ZIMM 1992) and mutant alleles were obtained and tested for their interactions with various rugose mutant alleles. We looked for the effects of a 50% reduction in the dosage of the interacting locus. For some of the interacting genetic loci, we also tested the effect of one copy of the gain-of-function allele on rugose phenotype. To do this, we used transgenic lines carrying heat-shock promoter constructs ectopically expressing the wild-type gene product. Our results suggest that rugose interacts with the components of the Drosophila EGFR- and Notch-activated signal transduction cascades (Table 4).

rugose mutations interact with the components of the EGFR signal pathway:
Argos (aos) is a secreted protein having an EGF motif. Partial loss-of-function mutations in argos result in rough eyes with supernumerary cone and R cells and extra-wing-vein phenotypes. Complete loss-of-function mutants of argos are embryonic lethals. While loss-of-function argos mutants have increased numbers of cone cells, heat-shock promoter-driven overexpression of Argos leads to a reduction in the number of cone cells. Cell culture experiments have shown that Argos is a negative regulator of the DER (KRETZSCHMAR et al. 1992 ; OKANO et al. 1992 ; FREEMAN 1994A ; SAWAMOTO et al. 1994 ; SCHWEITZER et al. 1995 ; GOLEMBO et al. 1996 ). argos mutations act as strong suppressors of the rugose mutation. The rough eye phenotype of rg is completely suppressed by a single copy of the argos loss-of-function mutation. rg/Y; argos/+ double mutants have smooth eyes and normal complement of R cells as well as cone cells (Fig 6E). Consistent with this result is our finding that heat-shock promoter-driven overexpression of Argos acts as a dominant enhancer of the rg mutant phenotype (data not shown). In a genetic screen for second-site modifiers of the argos phenotype, WEMMER and KLAMBT 1995 identified two interacting genes and suggested that they may function in the Argos-mediated signaling pathway. Mutations in bulge and soba act as dominant suppressors of the rough eye phenotype of the argos amorphic allele argos⁷ as well as the rough eye phenotype caused by the heat-shock-induced overexpression of the Argos protein. A single copy of bulge and soba dominantly enhance the rough eye phenotype of the rg mutants (Fig 6F). These results are consistent with our finding that argos mutations act as strong suppressors of rugose.

View larger version (186K): In this window In a new window Download PPT slide   Figure 6. rugose interacts with argos and Star. Argos mutation acts as a strong suppressor of the rugose eye phenotype. Scanning electron micrographs of compound eyes are shown. (A) Wild type shows a smooth eye phenotype. (B) rg1/Y eye shows a moderately rough eye phenotype. (D) argossty1/+ shows an eye surface similar to wild type. (C) bulD/+, a suppressor of the argos mutation, shows a mild rough eye phenotype. (E) rg10/Y ; argossty1/+, a single copy of the argos mutation, acts as a strong suppressor of the rough eye phenotype of rg. (F) rg1/Y; bulD/+, a single copy of the argos suppressor, acts as a dominant enhancer of the rg phenotype. Anterior is to the right. Bar, 25 µm. Star mutations act as strong enhancers of the rugose phenotype (G–I). (G) S54/+; (H) rg6/Y; S54/+ double mutant shows severe enhancement of the rough eye. (I) rg6/Y; Shs.30/+ eye showing the complete suppression of the rugose phenotype by the heat-shock promoter-induced expression of wild-type Star. Anterior is to the right. Bar, 25 µm.

Star (S):
Star, rhomboid, and spitz belong to the "spitz" group of genes and encode an essential function necessary for ventral midline development (ANDERSON and NUSSLEIN-VOLHARD 1984 ). In addition to the recessive lethal embryonic phenotype, S mutations are haplo- insufficient and show a dominant, rough eye phenotype (RENFRANZ and BENZER 1989 ). During development, S is required in a wide variety of tissues and S mutations show genetic interactions with genes from multiple signaling pathways. S encodes a putative membrane protein that, in combination with Rhomboid (rho), participates in the processing of the EGFR ligand, Spitz (HEBERLEIN et al. 1993 ; KOLODKIN et al. 1994 ; PICKUP and BANERJEE 1999 ; LEE et al. 2001 ; URBAN et al. 2001 ). In our modifier screen we identified an S deficiency as a strong enhancer of the rugose rough phenotype. We have tested a number of S alleles for their interactions with multiple alleles of rugose (Table 4). Mutations at the S locus act as strong enhancers of the rugose eye phenotype and conversely heat-shock promoter-driven overexpression of the wild-type Star protein acts as a dominant suppressor of the rugose mutant phenotype (Fig 6, G–I). Consistent with these results, in our experiments, rhomboid (rho) mutations act as dominant enhancers of the rough eye phenotype of rugose mutants. rho encodes a novel intramembrane serine protease and is involved in the proteolytic processing of the EGFR ligand Spitz. (BIER et al. 1990 ; STURTEVANT et al. 1993 , STURTEVANT et al. 1996 ; STURTEVANT and BIER 1995 ; LEE et al. 2001 ; URBAN et al. 2001 ). A single copy of the rho mutation acts as an enhancer of the rg eye phenotype and a single copy of the hs-rho acts as a weak suppressor (data not shown).

EGFR:
The Drosophila homolog of EGFR (DER) is an RTK that activates a highly conserved signal transduction cascade in a variety of tissue and cell types during Drosophila development. The activation of the DER is dependent on the tissue/cell type-specific ligands at the specific developmental stage. In the developing eye, DER function is required for the determination of all retinal cell types. A single copy of the mutations in DER acts as a mild enhancer of the rugose eye phenotype (Fig 7). In addition, Ellipse, a dominant mutation in DER (DER^E), acts as a suppressor suggesting that rugose interacts with the DER-mediated signal cascade.

View larger version (185K): In this window In a new window Download PPT slide   Figure 7. Rugose interacts with EGFR. (A) Scanning electron micrograph shows a wild-type eye. (B) rg6/Y shows a rough eye. (C) rg6/Y; top1/+ eye shows a slight enhancement of the phenotype. (D) top1/+ eye appears wild type. (E) Elp/+ eye shows a rough eye phenotype. (F) rg6/Y; Elp/+ eye shows that a single copy of Ellipse acts as a suppressor of the rough eye phenotype. Anterior is to the right. Bars, 25 µm.

Ras1:
ras1 is a Drosophila homolog of the human ras genes (H-ras, Ki-ras, and N-ras). Ras1 is a GTPase, which functions as the key transducer in several of the receptor tyrosine kinase-activated cellular signal transduction pathways. In the developing eye, ras1 is required for the specification of photoreceptors as well as cone cells. Reduction or loss of Ras1 activity results in the failure of photoreceptor cell determination. A constitutively active form of Ras1 (Ras^v12) results in the overrecruitment of retinal cells. We tested the effects of the ras1 mutations on the rg eye phenotype. Our results show that a 50% reduction in ras1 activity acts as a dominant enhancer of the rg rough eye phenotype (Fig 8E). In addition, a single copy of the dominant negative mutant form of Ras^N17 acts as a strong enhancer of the rg eye phenotype (Fig 8F). In our experiments, a single copy of the constitutively active Ras^v12 was a weak suppressor of the rough eye phenotype of rg. These data suggest that Ras1 and rugose interact in a dose-dependent manner and may function synergistically in retinal pattern formation.

View larger version (154K): In this window In a new window Download PPT slide   Figure 8. rugose interacts with ras1 and rolled mutations. Scanning electron micrographs show (A) a smooth wild-type eye; (B) compound eye from the ras1 allele sevd2; E (e.1b)/+, with an almost wild-type appearing surface; (C) compound eye from the dominant negative mutant allele of ras1, w118, [w+,sevhs-RasN17]/Cyo, showing a rough phenotype; and (D) rg6/Y mutant eye showing a rough eye phenotype. (E and F) Compound eyes from the double mutant flies show that ras mutations act as strong enhancers of the rg eye phenotype. (E) rg6/Y; ras sevd2; E (e.1B)/+. (F) rg6/Y; w118, [w+,sevhs-RasN17]/+. Anterior is to the right. Bar, 25 µm. rugose interacts with rolled (MAP kinase). (G) Scanning electron micrograph shows a wild-type (CS) compound eye. (H) Compound eye from rl/+ appears wild type. (I) The rolled gain-of-function allele (rlEMS24) has a mild rough eye phenotype. (J) The compound eye from rg3/Y shows a rough eye phenotype. (K) A single copy of the rolled gain-of-function mutation suppresses the rough eye phenotype of rugose. Compound eye from rg3/Y; rlEMS24/+ double mutant shows suppression of the rough eye phenotype. A single copy of the rl loss-of-function mutation acts as an enhancer of the rugose eye phenotype. (L) Compound eye from rg3/Y; rlsu23/+ shows an enhanced rough eye phenotype. Anterior is to the right. Bar, 25 µm.

Rolled (rl):
Rolled is a mitogen-activated protein kinase that functions at the last step in the Ras-MAPK phosphorylation cascade. Activated Rolled activates downstream transcription factors and thus plays a key role in the EGFR-mediated signaling required for cell determination and pattern formation (BRUNNER et al. 1994 ). We constructed rll/+; rg/Y double mutants to test for genetic interactions with rg, and a single copy of the rolled loss-of-function mutation enhances the rough eye phenotype of rg (Fig 8L). A single copy of the dominant gain-of-function rolled mutation acts a suppressor of the rough eye phenotype (Fig 8K). Taken together these results suggest that rugose interacts with the components of the signal cascade activated by EGFR.

sparkling (spa):
sparkling is a Drosophila homolog of the vertebrate pax-2 gene and is involved in cone cell specification. The runt family transcription factor, Lozenge, has been shown to directly regulate spa and Lozenge is a key downstream mediator of the Notch and EGFR pathways (FLORES et al. 1998 , FLORES et al. 2000 ). A single copy of the spa mutation acts as dominant enhancer of the rg eye phenotype (see Table 4; data not shown). These results are consistent with our data showing rg interacts with the EGFR and Notch signaling pathways.

rugose interacts with Delta-Notch pathway:
The Delta-Notch pathway is involved in a variety of cell fate decisions during development. A single copy of the Delta mutation acts as strong dominant enhancer of the rough eye phenotype of rugose (Fig 9E). Similarly, a single copy of the Suppressor of Hairless [Su(H)] mutation also acts as an enhancer of the rugose eye phenotype (data not shown). Conversely, a single copy of the Notch pathway antagonist, Hairless (H), acts as dominant suppressor of the rg phenotype (Fig 9F). The results suggest that rugose interacts with the components of the Notch signaling pathway. Similar results have been recently reported by SCHREIBER et al. 2002 .

View larger version (191K): In this window In a new window Download PPT slide   Figure 9. rugose interacts with Notch pathway components. Scanning electron micrographs of compound eyes are shown. (A) Wild type shows a smooth surface. (B) rg6/Y compound eye shows a rough eye phenotype. (C) The Hairless mutant compound eye HE31/+ is similar to wild type. (D) Dl6B37/TM6c shows a mild rough eye. In E a compound eye from an rg6/Y; Dl6B37/+ double mutant shows strong enhancement of the rough eye phenotype. In F, a compound eye from rg6/Y; HE31/+ shows suppression of the rugose rough eye phenotype. Anterior is to the right. Bar, 25 µm.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In this article, we have characterized the rugose locus and its genetic interactions. We show that rg is required for correct retinal pattern formation and those mutations at the rugose locus result in rough eyes due to an aberrant ommatidial organization. This phenotype may be the consequence of a malformed retinal basement membrane and a defective pigment cell lattice. Analyses of the cellular basis of this phenotype revealed that, in rugose mutants, the cone cell differentiation is abnormal, leading either to the lack of or to the reduced numbers of cone cells. The cone cell number is sensitive to the rugose gene dosage, suggesting that rugose function is required for normal differentiation of the cone cells. Developmental studies employing both light and transmission electron microscopy revealed that, in rugose mutants, cone cell development is defective even at the earliest stages and suggest that rugose is a positive effector of the cone cell determination. Cone cells belong to the R7 equivalence group and begin to differentiate immediately following R7 cell specification. Our studies also showed that, in rugose mutants, the number of secondary/tertiary pigment cells is increased. This may be due to the alternative recruitment of the cone cell precursors to the pigment cell fate. Although the ommatidia are disorganized in rugose mutants, photoreceptor cell specification and differentiation are unaffected. Our molecular analyses data reveal that rugose encodes a Drosophila A kinase anchor protein (DAKAP 550), a component of the cAMP-PKA-mediated signaling pathway. Protein kinase A (A kinase or PKA) is a key transducer of the cAMP signaling cascade and it phosphorylates multiple downstream targets.

Direct evidence for the role of AKAPs in the regulation of signaling comes from studies in mammalian systems and has led to the characterization of a large family of AKAPs. Altered localization of PKA by ectopic expression in tissue culture cells results in the altered pattern of phosphorylation of PKA substrates such as ion channels and CREB (FELICIELLO et al. 1997 ; GAO et al. 1997 ; GRAY et al. 1998 ). Our data presented here suggest that rugose, a Drosophila AKAP, is important for correct retinal pattern formation in the developing eye, a function likely to be mediated through its effects on the subcellular localization of PKA. If true, mutations in rugose would be expected to result in the mislocalization of PKA, leading to altered signaling. Determining the cellular and subcellular distribution of PKA in rugose mutants will test this. At this point the precise role of the cAMP-PKA signaling pathway in cell fate specification in the Drosophila eye is unclear. The cAMP-PKA signal cascade may play a role in cell fate determination through its interactions with the Notch and EGFR pathways. Recent studies have shown that cell fate determination in developing eye requires the integration of inputs from multiple signaling pathways and a combinatorial code involving the activation of specific transcription factors (FLORES et al. 2000 ; HALFON et al. 2000 ; XU et al. 2000 ; TOMLINSON and STRUHL 2001 ). Our data from the genetic modifier screen show that rugose interacts with components of the EGFR and Notch signaling pathways in a dosage-dependent manner. Mutations that reduce either EGFR or N signaling behave as dominant enhancers of the rough eye phenotype. Conversely, gain-of-function mutations that increase EGFR or N signaling act as dominant suppressors of the rugose phenotype. The molecular basis of these observed interactions is not clear at the present time. We propose that these interactions between PKA signaling and the EGFR and Notch pathways are likely to be indirect.

Similar to our results, SCHREIBER et al. 2002 identified rugose as a modifier of the rough eye phenotype induced by Hairless overexpression and have shown that rugose interacts with the Notch and EGFR pathway components, confirming our results on the genetic interactions. In a separate study, HUANG and RUBIN 2000 identified AKAP 200, the second Drosophila AKAP (LI et al. 1999 ), in a misexpression screen for genes that modulate RAS1 signaling suggesting interactions between the RAS1 signaling and the cAMP-PKA pathways. Others have also reported interactions between the cAMP-PKA, RAS-MAPK, and the Notch pathways in a variety of systems (FISHMAN et al. 1997 ; JACKSON and BLOCHLINGER 1997 ; FITZGERALD et al. 2000 ; LIEBMANN 2001 ; VOGT WEISENHORN et al. 2001 ). However, molecular details of interactions between the various pathways are not yet clear. It is possible that rugose interacts with the EGFR and N pathway components at the protein level. In this model, Rugose, as an AKAP, may be involved in anchoring components of other signaling pathways. Although initially it was thought that AKAPs were specific for PKA, it has been shown subsequently that AKAPs are multivalent and can anchor a wide variety of signaling components (KLAUCK et al. 1996 ; PAWSON and SCOTT 1997 ). Taken together our results suggest that rugose could exert its effects on cone cell specification and retinal pattern formation through its interactions with the EGFR and Notch pathways by facilitating the cross-talk between the different signaling pathways.