|Home | About | Journals | Submit | Contact Us | Français|
Correct patterning of cells within an epithelium is key to establishing their normal function. However, the precise mechanisms by which individual cells arrive at their final developmental niche remains poorly understood. We developed an optimized system for imaging the developing Drosophila retina, an ideal tissue for the study of cell positioning. Using this technique, we characterized the cellular dynamics of developing wild-type pupal retinas. We also analyzed two mutants affecting eye patterning and demonstrate that cells mutant for Notch or Roughest signaling were aberrantly dynamic in their cell movements. Finally, we establish a role for the adherens junction regulator P120-Catenin in retinal patterning through its regulation of normal adherens junction integrity. Our results indicate a requirement for P120-Catenin in the developing retina, the first reported developmental function of this protein in the epithelia of lower metazoa. Based upon our live visualization of the P120-Catenin mutant as well as genetic data, we conclude that P120-Catenin is acting to stabilize E-cadherin and adherens junction integrity during eye development.
The precise spatial arrangement of the cells within tissues is essential for their function. In some tissues, spatial restriction of cell fate is sufficient to generate the final pattern. In other tissues, such as the developing mammalian brain, the vertebrate retina, and the intestinal epithelium, cells migrate from their original position to their final niche (Batlle et al., 2002; Joyner and Zervas, 2006; Reese and Galli-Resta, 2002). These migrations can occur across significant distances but in most examples likely reflect more subtle cellular movements within the epithelium. Although some of the molecules that mediate these migrations are known, most tissues do not provide the needed accessibility to dissect individual cell movements in detail.
The Drosophila pupal retina is an ideal system in which to study cell positioning during development. The fully patterned retinal epithelium consists of a regular array of identical unit eyes. These ‘ommatidia’ are initially crudely arrayed within the larval eye field and are separated by a loose collection of ‘interommatidial precursor cells (‘IPCs’; (Cagan and Ready, 1989a; Wolff and Ready, 1993)). In the pupa, a precisely regulated combination of cell movements, death, and differentiation (Cagan and Ready, 1989a) corrals these IPCs into their final positions, yielding a honeycomb pattern that reorganizes the ommatidia into a hexagonal array.
These patterning steps are dependent on both signaling and adhesion. Cell–cell signaling regulates the number of cells within developing ommatidia as well as the specification of each cell type. For example, Notch pathway activity is required for differentiation of each of the 20 cell fates within the eye field (Cagan and Ready, 1989b; Nagaraj and Banerjee, 2007; Parody and Muskavitch, 1993). In addition, Notch activity is required within the IPCs for proper cell number as well as cell sorting (Cagan and Ready, 1989b; Miller and Cagan, 1998; Shellenbarger and Mohler, 1975). However, while the role of Notch in directing cell fate is well-established, less is understood of whether Notch also regulates cell morphogenesis.
The adhesion molecules Roughest and Hibris also play an essential role in patterning the retina as these two molecules are required to refine the IPC lattice to a hexagon (Bao and Cagan, 2005; Dworak et al., 2001; Reiter et al., 1996; Wolff and Ready, 1991). Single cell expression experiments with Roughest and Hibris indicate that both mediate the final positioning of cells within the hexagon through direct heterophilic adhesion and that both control adherens junction formation between IPCs (Bao and Cagan, 2005). For example, as IPCs re-arrange into their final pattern they briefly reduce their adherens junctions; these junctions are then re-assembled as patterning is completed (Bao and Cagan, 2005; Grzeschik and Knust, 2005). Ectopic expression of Hibris in the developing hexagonal lattice resulted in the premature re-appearance of these junctions as well as mis-patterning (Bao and Cagan, 2005). However, the mechanisms by which the adherens junctions are normally dynamically regulated are not known.
In this study, we present a method for visualizing development of the living pupal eye in situ. We utilize this method to extend previous observations on the cellular movements of the developing retina in wild-type and in two classical eye mutants that alter cellular positioning, one through cell signaling and a second through cell adhesion. Previous work has suggested that regulation of adherens junctions is important for patterning (Bao and Cagan, 2005; Cordero et al., 2007; Grzeschik and Knust, 2005; Tepass and Harris, 2007). To begin to address this issue, we utilize live visualization to demonstrate a role for P120-Catenin as a regulator of E-Cadherin as IPCs undergo the precise movements required to generate a hexagonal pattern within the eye field.
Previous work by others and us (Monserrate and Brachmann, 2007; Vidal et al., 2006) has demonstrated the utility of live visualization. Here, we present an optimized approach to observing cells within the Drosophila retinal field. To fluorescently outline cells, we utilized transgenic animals containing a GFP-tagged α-Catenin (Oda and Tsukita, 1999) driven by the eye specific driver GMR-GAL4 (GMR > α-Catenin-GFP; (Freeman, 1996)). Expression of GMR > α-Catenin-GFP did not affect patterning of the eye compared to GMR-Gal4 alone (see Table S1 in supplementary materials). We did observe a mild suppression (Nfa-g) or enhancement (rstCT, p120ctn308) of specific mutant phenotypes, although the presence of GMR > α-Catenin-GFP did not introduce novel phenotypes. Pupae were mounted in a custom-designed chambered slide as a ‘hanging drop’ preparation to permit sufficient oxygen and moisture (Fig. 1A and B) and were then imaged over extended periods of time up to 10 h in length (see Methods). A subset of imaged pupae was allowed to develop inside of the slide and all survived to adulthood, indicating that both the slide and the imaging process permit development. The patterning of imaged retinas tended to eventually lag behind the patterning of non-imaged retinas dissected at the same developmental time; to minimize these effects we limited most of our imaging to 4 h.
The mature retinal array can be seen by 42 h after pupal formation (a.p.f.). At the apical surface, each ommatidium is composed of a floret of four central cone cells surrounded by two enwrapping primary pigment cells (1°s). The photoreceptor neurons lie below this ommatidial core and are not visible at the apical surface. These ommatidial cores are organized into a hexagonal array by an interweaving lattice of secondary pigment cells (2°s); at the vertices of this lattice alternate tertiary pigment cells (3°s) and sensory bristles (Fig. 1C). Patterning of the 2°/3° hexagonal lattice leads to precise organization of the ommatidial array that, in turn, permits accurate vision.
We focused on the apical surface of the retinal epithelium: evolving cell–cell contacts are initiated at the surface (Cagan and Ready, 1989a) and nearly all relevant surface signaling molecules are found at or above the adherens junctions. In particular, we imaged fifteen separate GMR > α-catenin-GFP retinas at stages between 18 and 31 h a.p.f., the developmental interval during which the interommatidial lattice is patterned. By 10 h a.p.f., the cone cells and underlying photoreceptor neurons have emerged to define approximately 750 ommatidia; the overall ommatidial array, however, is poorly patterned across the retina at this stage. These ommatidia are surrounded by a field of interommatidial precursor cells (IPCs) and bristles. The 1°s, 2°s, and 3°s are derived from this IPC pool; bristle development is independent of the ommatidia, though final placement of the bristles requires proper ommatidial patterning.
Two 1°s are recruited from the IPC pool to each ommatidial core through signals derived from the cone cells (Miller and Cagan, 1998; Nagaraj and Banerjee, 2007). In the three retina imaged during this process, we observed that as the apical profiles of nascent 1°s enlarged and enwrapped the cones, other cells adjacent to the cones were displaced. The enwrapment of the cone cells by the nascent primaries was continuous and steady throughout their formation until each primary contacted the other, completing the process (Fig. 1D, see Movie S1 in supplementary material).
Prior to 20 h a.p.f., IPCs are randomly arrayed in multiple layers between the ommatidia. In the seven retinas imaged between 20 and 23 h a.p.f. – the stage during which paired 1°s enwrap the cone cells – the remaining IPCs were observed to intercalate between each other to lie single file between adjacent ommatidia (Fig. 1E, see Movie S2 in supplementary material). This intercalation occurred asynchronously across the retina, as some portions of the field took longer than others to resolve.
In the four retina imaged after 23 h a.p.f., cell movement resulting in the exchange of cell neighbors was limited to cells competing for the future 3° niche. Previous work has speculated that 3° cells arise when a single cell in the presumptive 3° vertex reaches across its neighbors to contact three primaries (Bao and Cagan, 2005). Between 24:15 h a.p.f. and 27:30 h a.p.f., we observed that 37.7% of the 3° cells were formed in this manner (N = 61; see Supplementary Table S2). In addition, 29.51% of vertexes exchanged contacts such that an apparently established cell in the 3° niche was often subsequently displaced by one of its neighbors (N = 61; Fig. 1F, see Movie S3 in supplementary material). Eventually, a single cell was established stably in the 3° niche at all vertices. Overall, stable 3°s were established at any stage between 24 and 30 h a.p.f. when all 3°s were fully resolved.
Cell death was observed throughout the 3° formation stage and has been well-described elsewhere (Monserrate and Brachmann, 2007). Although the formation and placement of the 3° is largely independent of cell death (Rusconi et al., 2004), we observed several examples where the death of an IPC appeared to eliminate competition for a 3° position: death of a potential 2° pulled its neighbor fully into a 2° niche and away from a position that would have competed for a 3° niche (data not shown).
Loss of Notch pathway activity disrupts IPC patterning, leading to loss of 1°s and unpatterned 2°/3°s (Shellenbarger and Mohler, 1975; Cagan and Ready, 1989b). The Notch allele Notchfacet-glossy (Nfa-g) specifically reduces Notch activity in the pupal eye (Shellenbarger and Mohler, 1975; Cagan and Ready, 1989b), permitting us to explore these defects in more detail. Adult Nfa-g eyes have a glossy appearance, resulting from an excess of interommatidial cells that is apparent in the pupal eye (Fig. 2D; (Cagan and Ready, 1989b)). The role of Notch activity in directing this patterning process is not understood, nor do we understand the underlying cellular defects that lead to the scattered IPC phenotype. We therefore utilized our live imaging approach to examine the morphogenesis of genotypically Nfa-g IPCs.
Previous studies of Nfa-g utilizing electron microscopy reported that 1° differentiation initiated normally in Nfa-g retinas but that over time the nascent 1°s separated to allow other cells direct apical access to the cone cells (Cagan and Ready, 1989b). Separation of nascent Nfa-g 1°s could occur either by an eventual relaxation in the shapes or contact by the initial precursor 1°s or by active invasion by adjacent IPCs to displace the original 1°-like cells from their position. To distinguish between these two models and to better understand the role of Notch in cell morphogenesis, we imaged six retinas of Nfa-g males between 18and30 h a.p.f.. We observed that, similar to our controls, IPCs adjacent to the cone cells increased in apical area and began to enwrap the cones as 1° precursors. However, unlike our controls this behavior was not limited to the original two 1°-like cells but instead was consistently observed in multiple cells (Fig. 2, see Movie S4 in supplementary material). Even when two 1°-like cells were able to fully enwrap the cone cell cluster and contact each other, adjacent IPCs typically intercalated between them to directly contact the inner cone cell core. In some examples, an invading cell was able to completely displace a previously established 1°-like cell from the cone cells cluster it was contacting.
The movements of Nfa-g cells surrounding the cones were remarkably and abnormally dynamic. For example, we observed examples in which a single cell (i) initially contacted the cone cells of one ommatidium, (ii) was subsequently displaced by adjacent cells exhibiting 1°-like behavior, and (iii) then invaded a second ommatidial niche to contact its constituent cone cells (Fig. 2, Movie S4). Also, 1° formation was completed by 22 h a.p.f in control retinas; in Nfa-g retinas, dynamic behavior surrounding the cone cell clusters occurred as late as 30 h a.p.f. These images indicate that genotypically Notch ommatidia fail to limit the number of 1° precursors and Notch 1° precursors fail to hold their positions once established.
In addition to requiring signal transduction pathways such as Notch, IPC patterning also requires selective adhesion to direct IPCs into their proper niches. The IgG domain transmembrane adhesion molecule Roughest (Ramos et al., 1993) is expressed in 2°/3° precursors; its heterophilic binding to Hibris in 1°s is required for correct assembly of the 2°/3° lattice (Reiter et al., 1996; Bao and Cagan, 2005). Loss of roughest activity through the truncation mutant roughestCT (rstCT) approximately phenocopied Nfa-g mutants: rstCT retinas exhibited ectopic IPCs that were poorly patterned (Reiter et al., 1996; Wolff and Ready, 1991). Unlike Nfa-g, however, rstCT retinas contain properly differentiated 1°s (Fig. 3C).
Based on single cell expression experiments, heterophilic binding between Hibris and Roughest was proposed to selectively enhance adhesion between IPCs and 1°s and drive patterning of the retina through the maximization of IPC/1° contacts (Bao and Cagan, 2005). This role for Roughest could manifest itself in overall IPC movement within the epithelium or, alternatively, strictly by biasing IPC movement towards correct cell niches. To further explore the role of Roughest in IPC patterning, we imaged rstCT males between 18 and 28 h a.p.f. (5 movies total). Early development of rstCT retinas was indistinguishable from wild-type with the exception that occasional rstCT 1°s from separate ommatidia contacted each other, an event almost never observed in wild-type. Most of these 1°/1° contacts were resolved over time.
As anticipated, during the stage of intercalation in wild-type rstCT IPCs did not undergo the normal cell intercalation that would produce a single layer of cells between ommatidia. Surprisingly, mutant cells did exchange neighbors and altered their cell shapes at a rate similar to wild-type, demonstrating their ability and tendency toward movement. However, these movements were undirected and hence there was no resolution of the IPC lattice into a single layer of cells (Fig. 3, see Movies S5 and S6 in supplementary material). Later movies of rstCT indicated that these undirected cell movements continued through at least 24–28 h a.p.f. This data indicates that Roughest is necessary for either the directed movement of cells or stabilization of cells as they achieve a correct position, but that cell movement itself is independent of Roughest activity.
Recent work has led to the proposal that adherens junctions also play an important role in patterning the pupal eye (Bao and Cagan, 2005; Cordero et al., 2007; Grzeschik and Knust, 2005). Nfa-g and rstCT cells did not exhibit disruption of α-Catenin-GFP, suggesting that the adherens junctions were correctly regulated. Indeed, mutations specifically affecting the adherens junctions but not the integrity of the retinal epithelium have not been reported. P120-Catenin, encoded by p120ctn, is an armadillo repeat domain-containing protein that binds to the juxtamembrane domains of classical cadherins to regulate adherens junction stability and activity (reviewed in Anastasiadis, 2007; Xiao et al., 2007). In mammals, it is essential for viability and modulates the levels and adhesive properties of cadherins (reviewed in McCrea and Park, 2007). In Drosophila and C. elegans, deletion of the p120ctn locus enhanced mutations in cadherin but was non-essential for viability (Myster et al., 2003; Pacquelet et al., 2003; Pettitt et al., 2003). Recent work, however, has found that p120ctn is required to regulate neuron morphology (Li et al., 2005). In contrast, despite earlier reports to the contrary (Magie et al., 2002), recent studies ascribe no phenotype to p120ctn in Drosophila epithelia. This led to the suggestion that P120-Catenin solely plays a supporting role in cadherin-based adhesion (Fox et al., 2005; Myster et al., 2003; Pacquelet et al., 2003).
The surface phenotype of adult fly eyes homozygous for the null p120ctn allele p120ctn308 (Myster et al., 2003) was wild-type in appearance. However, examination of the pupal retina indicated that genotypically p120ctnmutant eyes have ectopic lattice cells and a partially penetrant mis-patterning of the 3° niche (Fig. 4B). Pupae bearing the p120ctn308 chromosome in trans to a deficiency covering the region (Df(2R)244) exhibited a phenotype similar to homozygous p120ctn308 retinas (data not shown), consistent with previous reports (Myster et al., 2003) that p120ctn308 represents a null allele. The p120ctn308 mutation was generated by imprecise excision of a P-element found in the parent line KG01086 (Myster et al., 2003). Retinas bearing a single copy of p120ctn308 in trans to KG01086 exhibited a wild-type phenotype indicating that the p120ctn308 phenotype was a direct result of the excision event (data not shown). Lastly, ubiquitous expression of a full-length p120ctn-GFP transgene in a p120ctn null background completely rescued the p120ctn null phenotype (Fig. 4). Taken as a whole this data indicates that the eye phenotype is a direct result of a loss of p120ctn and represents the first reported developmental requirement of p120ctn in an epithelium of lower metazoa.
To better understand the role of p120ctn in patterning the fly eye, we imaged p120ctn null mutant development from 24 to 29 h a.p.f. (four retinas total). While the final p120ctn308 phenotype was fairly subtle, our live imaging revealed surprisingly dramatic differences with wild-type development. In particular, live imaging of p120ctn308 pupal eyes showed consistent, transient separation of IPCs accompanied by a loss of α-Catenin-GFP fluorescence from the membranes at their contact face (Fig. 5, Movies S7 and S8 in Supplementary Material). This apparent breakdown of coherent junctions occurred almost exclusively between IPC:IPC and IPC:1° junctions and presumably accounted for their ability to achieve or maintain stable positions. To quantitate this difference, we followed the dynamics of 3° emergence throughout the stage of IPC patterning (see Table S2 in supplementary materials). We observed a clear difference in the ability of local IPCs to achieve and – in particular – to retain a position in the 3° niche. This instability and ectopic movement presumably accounts for the errors in 3° patterning observed in the mid-pupa. Control movies showed no comparable separation or loss of fluorescence. Other parameters such as cell movements were on the whole indistinguishable from wild-type. This result indicates that p120ctn308 IPCs are capable of forming adherens junctions but are unable to maintain them during dynamic cell movements.
We did not observe clear genetic interactions – either as trans heterozygotes or as dominant modifier activity – between p120ctn and Egfr, wingless, roughest, Notch, shotgun, α-Catenin, or the small GTPases (reducing Rho1 or Cdc42) (data not shown). However, a closer genetic analysis of the relationship between p120ctn, shotgun, and Rho1 yielded surprising results. Based on both cell culture and in vivo data, mammalian P120-Catenin has been proposed to regulate both RhoA and E-cadherin (Anastasiadis et al., 2000; Davis and Reynolds, 2006; Grosheva et al., 2001; Noren et al., 2000; Perez-Moreno et al., 2006).
Interestingly, the phenotype observed with complete loss of p120ctn activity – using the null deletion allele p120ctn308 – was further enhanced by removing a functional genomic copy of shotgun (shgR69) or Rho1 (Rho172O) (Fig. 6A–C). Removal of Rho1 resulted in additional ectopic cells and an increase in the frequency of patterning errors (Fig. 6E). In the case of the shg interaction, the hexagonal IPC pattern was disrupted with extra cells present in double layers around bristle cells (Fig. 6A). The severity of this interaction prevented its quantification. Neither shgR69 nor Rho172O, both null alleles, gave a dominant phenotype on their own (data not shown; Fig. 6D). The ability of mutations in shotgun or Rho1 to further enhance a null mutation in p120ctn indicates that both DE-Cadherin and Rho1 act, at least in part, through a pathway that is independent of P120-Catenin. We did, however, observe a consistent difference in E-cadherin localization. While full loss of p120ctn led to at most a slight decrease in Armadillo (see Fig. 7B) and E-cadherin (not shown), we noted that E-cadherin protein was discontinuous at the membranes of p120ctn308 cells (see Fig. 7). This was best observed when comparing loss of P120-Catenin next to a rescue construct of P120-Catenin in neighboring clonal patches (see Fig. 7C).
In this study, we further characterize the cell movements required to pattern the developing pupal retina. While many of these movements have been inferred from dissected tissue (Bao and Cagan, 2005; Cagan and Ready, 1989a; Cagan and Ready, 1989b; Wolff and Ready, 1993), we observed that the developing retina was more dynamic than expected in both wild-type and mutant flies. For example, others and we had speculated (Bao and Cagan, 2005; Reiter et al., 1996) that the roughest phenotype was due to a loss of cell movement. However, rstCT IPCs were observed to actively exchange contacts and neighbors despite the fact that this exchange did not productively pattern the retina. In an earlier study, scanning electron micrographs showed that rstCT IPCs extend filopodia from their apical surface in a manner identical to wild-type (Frohlich, 2001). Combined with our live visualization studies, this data indicates that rstCT cells have an active cytoskeleton and can participate in cell rearrangement but cannot functionally recognize 1°s. Alternatively, rstCT cells may fail to establish junctions that stabilize a final position; consistent with this latter possibility, we previously reported the ability of ectopic Hibris to direct precocious adherens junctions (Bao and Cagan, 2005). While SEM studies of Nfa-g remain to be conducted, the similarity of the movements of Nfa-g IPCs to rstCT IPCs is striking. In fact, Notch is required for localization of Roughest protein (Gorski et al., 2000; Reiter et al., 1996) perhaps accounting for their phenotypic similarity.
Cell adhesion plays a key role in patterning the developing pupal retina (Bao and Cagan, 2005). In normal patterning, DE-cadherin staining between IPCs decreased during later stages of IPC re-arrangements, only to increase a few hours later as patterning was completed and cell contacts were finalized (Bao and Cagan, 2005; Grzeschik and Knust, 2005). The loss of roughest resulted in uniform DE-cadherin staining during this time, suggesting that one method by which Roughest may affect retinal patterning is through modulation of E-cadherin levels (Bao and Cagan, 2005; Cordero et al., 2007). More recently, BMP family signaling was found to regulate retinal patterning, in part by positively regulating E-cadherin (Cordero et al., 2007). Using live visualization, we found that P120-Catenin positively regulated DE-cadherin-based junctions, further demonstrating a role for DE-cadherin regulation in fine cellular patterning within the eye.
The mechanism by which P120-Catenin regulates cadherin based junctions in Drosophila remains unclear. In mammals, P120-Catenin has been shown to regulate cadherin by modulating its endocytosis (Hoshino et al., 2005; Miyashita and Ozawa, 2007; Xiao et al., 2005) and degradation (Davis et al., 2003; Ireton et al., 2002; Xiao et al., 2003).We note, however, that Drosophila cadherins lack the di-leucine motif that P120-Catenin masks to prevent endocytosis in mammals. Mammalian P120-Catenin also acts as an inhibitor of the small GTPase Rho by regulating RhoGAPs such as p190RhoGap (Wildenberg et al., 2006) or Rho itself ((Anastasiadis et al., 2000); reviewed in (Anastasiadis, 2007). During Drosophila embryogenesis, however, p120ctn failed to show functional interactions with mutations in Rho1 (Fox et al., 2005). In contrast, we detected an interaction between p120ctn and Rho1 during eye development, but this interaction was also inconsistent with the mammalian data. If the p120ctn null phenotype was the result of a loss of Rho inhibition, then we would have expected that removal of a functional genomic copy of Rho would suppress the p120ctn phenotype. Instead our results are consistent with a model in which P120-Catenin and Rho1 act in parallel pathways to regulate eye development.
Live visualization of development allows for high-resolution examination of cell movement during morphogenesis. As labeling techniques continue to improve, the use of in situ visualization will provide an increasingly sophisticated understanding of the physical properties of cells during their morphogenesis.
All crosses and staging were conducted at 25 °C unless otherwise noted. Stocks used were Canton-Special (wild-type), and shgR69 (Godt and Tepass, 1998). UAS-α-Catenin-GFP #8 (Oda and Tsukita, 1999) was provided by the Kyoto Stock Center. GMR-Gal4 (Freeman, 1996), Nfa-g (Welshons, 1965), p120ctn308 (Myster et al., 2003), Ubi-p120ctn-GFP (Myster et al., 2003), rstCT (Wolff and Ready, 1991), and Rho172O (Strutt et al., 1997) were obtained through the Bloomington Drosophila Stock Center. For imaging and genetic interactions, the following recombinant second chromosomes were constructed: UAS-α-Catenin-GFP, GMR-Gal4 1104; p120ctn308, UAS-α-Catenin-GFP; p120ctn308, shgR69; and p120ctn308, Rho172O. Clones were made in a p120ctn308/p120ctn308; Ubi-p120ctn-GFP/+ background.
Animals were raised at 25 °C and the pupal case removed around the eyes. The pupa was then inserted at a 45° angle into a slit in a 2% agarose pad contained within a ~1.5 mm thick Sylgard® (Dow Corning Corporation, Midland, MI, USA) washer on a 20mm × 60mm No. 1 coverslip, with the eye down and pressed tightly against the coverslip (Fig. 1A and B). This coverslip was then placed over a 0.8 cm deep × 2.1 cm wide × 2.1 cm long well routed in a 0.9 cm deep × 2.7 cm wide × 7.8 cm long block of acrylic containing moistened tissue cut to fit to maintain pupal humidity. This chambered slide was used on a Zeiss Axioplan2 microscope using a 63× objective, a FITC filter and a 75 Watt Xenon bulb with the intensity reduced using a 1.3% ultraviolet neutral density filter. The room containing the microscope was maintained at 25 °C. A Z-series of images was acquired by hand every 15 min using a CCD camera (Quantix Photometrics, Tucson, AZ, USA) and ImagePro Plus 5.1 (MediaCybernetics, Bethesda, MD, USA). When assembling the movie, in focus regions of each time point were isolated and spliced together into a single in focus image using Adobe Photoshop CS (Adobe Systems, San Jose, CA, USA). For each composite image the levels were adjusted to give the clearest image possible. Images were assembled into a Quick-time movie using Adobe ImageReady CS (Adobe Systems, San Jose, CA, USA) with a delay between frames of 0.8 s.
Pupae were aged at 25 °C. Retinas were dissected into PBS, fixed in 4% paraformaldehyde or 4% formaldehyde in PBS and washed in PBS-T (PBS/0.2% Triton X-100). Retinas were then incubated overnight at 4 °C with primary antibodies diluted in PAXDG (PBS containing 1% BSA, 0.3% Triton X-100, 0.3% deoxycholate, and 5% goat serum). Afterwards, retinas were washed in PBS-T and incubated overnight at 4 °C with secondary antibodies diluted in PAXDG, followed by washing in PBS-T. Retinas were mounted in Vectashield mounting media (Vector Labs, Burlingame, CA, USA). Antibodies used were: rat anti-DE-cadherin (1:10), mouse anti-discs large (1:500), and mouse anti-Armadillo (1:3 or 1:5) from the Developmental Studies Hybridoma Bank at the University of Iowa. anti-Armadillo was tagged with Alexa Fluor 488 or Alexa Fluor 568 using the appropriate Protein Labeling Kit (Invitrogen, Carlsbad, CA). Secondary Alexa Fluor 488 and Alexa Fluor 568 conjugated anti-mouse, and anti-rat were used (1:1000; Invitrogen, Carlsbad, CA).
Images of retinas, dissected at or after 40 h a.p.f., were scored by counting the number of cells in a hexagon formed by connecting the centers of the 6 neighboring ommatidia surrounding a single ommatidium. Errors in patterning were defined as multiple cells occupying a single 3° niche or double-layers of two cells in the secondary niche. Missing 2°s and bristles were not scored as errors as these would be detected in a simple count of cell number per hexagonal region. Likewise, misplaced and extra bristles were not scored as errors since they merely replaced a 3° and did not alter the hexagonal pattern of the ommatidium.
Retinas imaged between 24:15h a.p.f. and 27:30 h a.p.f. were scored by examining the configuration of the cells competing for the tertiary position. Analysis was limited to vertexes that remained in focus and in the field of view throughout the period scored. For each timepoint, each vertex was evaluated as either a tertiary vertex or an unresolved vertex. A tertiary vertex was defined as a single cell in the vertex position contacting three primaries. If a vertex was not a tertiary vertex then it was unresolved.
Vertexes were scored by examining the pattern of tertiary versus unresolved configurations over time. Vertexes where the configuration remained the same throughout the period were scored as stable. Vertexes that began unresolved, and changed to a tertiary configuration for the rest of the period were scored as being resolved to single tertiary. The reverse situation was scored as stable unresolved tertiaries. Vertexes that switched between unresolved and tertiary configurations multiple times were scored as dynamic, unresolved tertiaries. The results for the p120ctn and wild-type movies are tabulated in Supplemental Table S2.
We are grateful to Richard Carthew for providing flies. We would like to thank Craig Micchelli, Sarah Larson, Midori Seppa, Sujin Bao, and members of our lab for helpful comments and discussion. This work was supported by the National Institutes of Health Grant NIH R01 EY1149; D. Larson was supported by National Institutes of Health, Institutional National Research Service Award 5-T32-EY13360-06, from the National Eye Institute.
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.mod.2007.11.007.