Abstract
Photoreceptors form during Drosophila pupal development and acquire elaborate membrane structures, including the rhabdomeres and stalk membranes. Here, we show that the development of these cellular structures involves two distinct processes: the establishment of apical–basal polarity that requires Bazooka (Baz), and the regionalization of apical membrane into stalk membranes and rhabdomeres that requires Stardust (Sdt). In the absence of Baz, the apical–basal polarity is compromised in early pupal photoreceptors, and no identifiable apical membrane domain is formed. Sdt, in contrast, plays a more limited role in apical–basal polarity but is essential for the proper localization of transmembrane protein Crumbs (Crb), known to be required in the biogenesis of stalk membrane. Loss of Sdt causes strong defects in stalk membrane and rhabdomere resembling crb mutant phenotype. Thus, proteins required for establishing the early embryonic epithelial polarity are used later for the morphogenesis of photoreceptors, with Baz and Sdt functioning in different aspects of the formation of the apical–basal cellular architecture.
Different cell types exhibit different degrees of complexity in their morphogenesis. Epithelial cells are polarized to have distinct apical versus basolateral domains. Cascades of protein complexes have been found to specify this polarity (1). Many cells that are derived from epithelial cells have more complex morphology to carry out important functions, such as photoreceptors for vision. How these cells elaborate their different membrane domains is not well understood. In Drosophila, photoreceptors have clearly recognizable apical–basal polarity demarcated by zonula adherens (ZA) (2). Is this polarity established by the same protein complexes that specify epithelial polarity? How might this apical–basal polarity specification influence other structural specializations, e.g., the elaboration of rhabdomere perpendicular to the apical–basal axis? And, what are the mechanisms for regionalizing specific membrane domains in conjunction with the polarity?
One approach to address these questions in Drosophila is to test whether similar organizations of genetic pathways in simple and relatively well studied embryonic epithelial polarity also underlie the development of polarity in more specialized cell types like photoreceptors. Drosophila embryonic epithelium represents one of the simplest polarized cell types, and its apical–basal polarity is controlled largely in concert by three protein complexes (1). Baz(DmPar-3)/DmPar-6/DaPKC (atypical protein kinase C) plays the earliest and essential role in initiating the apical–basal polarity, whereas Sdt/Crb/Dlt (Discs lost) maintains the ZA formation and polarity by counteracting the activity of Dlg (Discs large)/Lgl (Lethal giant larvae)/Scrib (Scribble) (3, 4). Except for DaPKC, Crumbs (Crb), and Lgl, all of these proteins contain one or more PDZ domains important for mediating protein–protein interactions (1). The first insight that at least some of these proteins are also involved in the development of more specialized cell types is provided by the recent discovery that Crb plays an essential role in pupal photoreceptor morphogenesis (5, 6). During pupal development (pd), apical membranes in photoreceptors differentiate into two regions: the rhabdomere at the apical apex and stalk membranes connecting the rhabdomere and ZA (2) (also see Fig. 1L). In contrast to embryonic epithelia, Crb does not localize immediately adjacent to ZA in the photoreceptors as in epithelia; rather, it is exclusively on the stalk membrane. Loss of Crb does not cause severe apical–basal polarity defects; instead, Crb assumes a novel role of regulating the length of stalk membrane and the proper morphogenesis of rhabdomeres (5, 6).
Fig. 1.
Developmental expression pattern of Baz, DaPKC, and Sdt in pupal photoreceptors. (A) Baz colocalizes with Arm at 38% pd retina. (B) A single ommatidium at 43% pd showing Baz staining is basal to the phalloidin and Dlt. Phalloidin and Dlt stain the center of ommatidium, where the apical membranes of R1–R7 (outlined) converge. (C) A single ommatidium at ≈55% pd triple-stained with Baz, Dlt, and phalloidin. R1–R7 are outlined and labeled. Baz stains ZA, whereas Dlt and phalloidin stain the developing stalk membranes and rhabdomere, respectively. Staining of Baz at R7 ZA is consistently weaker than at R1–R6. Double staining with Baz and Arm at the same stage confirms that Baz is localized to ZA (data not shown). Only seven stalk membranes are visible by Dlt immunostaining. Based on the EM data (7), both stalk membranes of R1, R3, and R6 and one of R5 stalk membranes are too short to be visible under confocal microscope. (D) A single ommatidium at 60% pd double-stained with Baz and Dlt, showing that Baz staining is greatly elevated in cytoplasm and reduced at ZA (arrowheads). (E) A single ommatidium at 72% pd double-stained with Baz and Dlt. Baz apparently localizes into the rhabdomere region. (F) DaPKC colocalizes with Dlt in 36% pd photoreceptors. (G) DaPKC and Crb colocalize at stalk membranes at ≈55% pd. (H) At 71% pd DaPKC concentrates into rhabdomeres, with residual staining seen at stalk membrane (labeled with Dlt). (I) Sdt localizes to the center of ommatidia (four shown here) surrounded by ZA (labeled by Arm) at 30% pd. (J) Ommatidia at 72% pd double-stained with Sdt and Dlt to show that Sdt and Dlt precisely colocalize at stalk membrane. The same result is observed with Sdt and Crb (data not shown). (K and L) Summary of the developmental expression patterns of Baz and Sdt at early and late pupal stages. A single ommatidium at ≈37% pd is drawn in K, showing its composition of seven photoreceptors. Sdt, Crb, and Dlt localize at the apical region/membrane, which is demarcated by surrounding ZA where Arm and Baz localize. (L) A single photoreceptor at 70% pd is illustrated to show three functionally and structurally distinct domains of the developing apical membrane: ZA, stalk membrane, and rhabdomere; each is labeled by distinct protein markers.
So far, Crb and Dlt are the only two embryonic epithelial polarity proteins with their expression patterns characterized in pupal photoreceptors. In this article, we continued exploring the Drosophila pupal photoreceptor as a model cell type by focusing on the expression pattern and potential roles of Bazooka (Baz) and Stardust (Sdt). We show that the overall morphogenesis of apical–basal cellular architecture in photoreceptor involves first the establishment of apical–basal polarity that requires Baz, and then the regionalization of apical membrane that requires Sdt for regulating the subcellular localization of Crb and Dlt. Whereas the role of Baz is conserved from epithelia to photoreceptors, its embryonic partner DaPKC does not seem to be part of a Baz complex in photoreceptors. In contrast, Sdt, Crb, and Dlt seem to function together as in embryos but assume a novel function in membrane specialization. Thus, these proteins are reused after embryogenesis to specify more specialized membrane structures in photoreceptor, but their functions show flexibility and variation unexpected from the embryonic epithelial study.
Materials and Methods
Fly Stocks and Genetics. Whole eye clones of sdt and crb were generated by the following crosses (15): y w sdtXP96 FRT-19A/FM6, y w B crossed with y w GMR-hid FRT-19A; ey-Gal4, UAS-FLP; and y w; FRT-82B crb11A22/TM3, Sb (gifts from S.-C. Lam and K.-W. Choi, Baylor College of Medicine, Waco, TX) crossed with y w; FRT-82B GMR-hid l(3)/TM6, Ubx P[y+]. baz eye clones were generated by crossing y w bazXi FRT-18D/FM6, y w B; Ubx-FLP with y w ubi-GFP FRT-18D; ey-FLP.
Pupal Retina Dissection and Immunostaining. White pupae <2or3h into the pupation were collected and kept at 18°C until desired age. For early pupal retina dissection, pupae were first removed from pupal cases and decapitated directly in the fixative (4% formaldehyde in 1× PBS). Pupal cuticle and brain tissues were then removed in the fixative. After 30–40 min of fixation, pupal heads were washed three times for 5 min with PBST (1× PBS/0.3% Triton X-100). For late pupal retina dissection, pupae were decapitated in Ringer's buffer. Heads were bisected and corneas were removed with tungsten needles (D. F. Ready, personal communication). Retinas were fixed with PLP fixative with 0.1% Saponin (Sigma) for 20–30 min, washed twice with 50 mM NH4Cl in 1× PBS, and washed twice with PBST. After a 30-min preincubation with 5% normal goat serum in PBST, staining with primary antibodies was carried out overnight at 4°C. After washes with PBST (three times for 20 min), incubations with secondary antibodies were carried out at room temperature for 2–4 h followed by washes with PBST (three times for 20 min). Mounting was done in antifading medium (Biomedia), and for late pupal retina, vacuum grease was used to prevent the coverslip from crushing the samples. The following primary antibodies were used: rabbit anti-Sdt (affinity purified, 1:100–200), mouse anti-Crb (Cq4) (1:10–100), rat anti-Crb (1:500, gift from U. Tepass, Univ. of Toronto, Toronto), rabbit anti-Dlt (1:500, gift from M. Bhat), mouse anti-Dlt (1:250–500, gift from K.-W. Choi), mouse anti-Arm (1:50–100), rabbit anti-Baz (1:500–1,000, gift from A. Wodarz, Universitate Dusseldorf, Dusseldorf, Germany), and rabbit anti-EGFP (1:1,000–2,000). Alexa-488-, Cy2-, Cy3-, RRX-, and Cy5-conjugated secondary antibodies were from The Jackson Laboratory and Molecular Probes. Alexa-488-, 633-, or TRITC-labeled phalloidins are from Molecular Probes and Sigma and are used at 1 unit/μl concentration together with primary antibodies.
Confocal and Electron Microscopy (EM). Confocal imaging was carried out on a Lecia TCS SP2 confocal microscope with a ×100 oil objective. EM was carried out according to standard protocols. EM negatives were digitized by scanning, and stalk membranes were hand-traced by using imagej software (http://rsb.info.nih/ij) under 200–300% magnification to reduce tracing errors. Only stalk membranes with clearly identifiable adherens junction (AJ) were measured. All images were processed with photoshop (Adobe Systems, Mountain View, CA).
Results
A Brief Overview of Apical–Basal Polarity Development of Pupal Photoreceptor. Each of the ≈800 ommatidia in Drosophila compound eye contains eight photoreceptors (R1–R8). At early pd, their apical surfaces converge and attach to each other at the center of each ommatidium (2), with R8 under R7 and normally not seen in distal sections. Around 37% pd, the ZA marker Armadillo (Arm) shows a characteristic staining pattern of seven “dots” in tangential sections (Fig. 1 A and K) surrounding the center of ommatidia where Crb and Dlt accumulates (5, 6) (Fig. 1 B and K). Later the photoreceptor apical surfaces detach from each other and differentiate into stalk membranes and rhabdomeres (7). At 55% pd Crb and Dlt begin to concentrate on the stalk membrane flanking the rhabdomere marked with phalloidin staining for F-actin (Fig. 1C), with ZA staining at the basal end of stalk membrane (5, 6) (Fig. 1L).
Baz but Not DaPKC Localizes to Adherens Junctions During Early Pupal Photoreceptor Development. In embryonic epithelial cells, Baz complex (Baz/DmPar-6/DaPKC) colocalizes with AJ complex and directs the apical movement of spot AJ complexes which initiates the apical–basal polarity during early embryogenesis (8, 9). The spot AJ complexes later coalesce around the apical cell boundary into a circumferential belt-like ZA, a process that requires Sdt/Crb/Dlt complex (9–12). Whereas at the level of resolution offered by confocal microscopy, Baz complex, Sdt/Crb/Dlt complex, and ZA markers like Arm and E-Cadherin all appear at the apical cell boundary (9), EM study shows that Crb localizes immediately apical to AJ (10) in embryonic epithelial cells.
Despite being absolutely required in establishing the apical–basal polarity of embryonic epithelia, proteins of Baz complex have not been examined in pupal photoreceptors. Moreover, in pupal photoreceptors Crb and Dlt are absent from ZA but localize to the stalk membrane (5, 6) (Fig. 1K). It is thus an open question as to whether these proteins are expressed in photoreceptors, where they localize subcellularly, and what their functions might be. To address these questions we first stained pupal retinas with anti-Baz and anti-DaPKC antibodies and found these two proteins are expressed throughout the pupal retina development. In photoreceptors of 38–43% pd, we found that Baz remains colocalized with Arm at ZA (Fig. 1 A) but does not overlap with the apical localization of Dlt (Fig. 1 B and C) and Crb (data not shown). Surprisingly, the localization of Baz at ZA is only maintained until 60% pupal stage, when Baz staining becomes progressively weaker at ZA with a concurrent accumulation in the cytosol (Fig. 1 D and L). At 70% pd, Baz staining disappears from ZA and becomes concentrated in the region of developing rhabdomere (Fig. 1E). Unexpectedly, we also found that subcellular localization of DaPKC, another component of Baz complex in embryonic epithelia (13, 14), does not overlap with Baz in early pupal stages. Instead, DaPKC localizes to the apical region and stalk membrane until 55% pd (Fig. 1 F and G). Only at later stages, DaPKC also begins to concentrate into the rhabdomere similarly to Baz (Fig. 1H). Thus, protein complex of Baz and DaPKC does not form a complex during early pupal photoreceptor development. For studying the apical-polarity specification, we focused on Baz.
Sdt Remains Colocalized with Dlt and Crb at Stalk Membrane. Sdt is a single PDZ domain membrane-associated guanylate kinase protein that binds to the short intracellular tail of transmembrane protein Crb and is required for the subcellular localization of Crb and Dlt (9, 12). We found Sdt is expressed throughout pupal stage and remains colocalized with Crb and Dlt in pupal photoreceptor (Fig. 1 I and J). At early pupal stage, Sdt is seen at the apical region of photoreceptors surrounded by Arm staining (Fig. 1I), and it overlaps with Dlt and Crb (data not shown). In late pupal photoreceptors, Sdt localizes to the stalk membrane, again in a pattern identical to that of Dlt (Fig. 1J) and Crb (data not shown). The developmental expression patterns of Sdt, Crb, and Dlt suggest that, unlike components of the Baz complex, these three proteins still form a complex in developing photoreceptors as in embryonic epithelia.
Fig. 1 K and L summarizes the developmental expression patterns of Baz, Sdt/Crb/Dlt complex, and ZA in pupal photoreceptors. Whereas in embryonic epithelia the subcellular localization of Baz and Sdt/Crb/Dlt could not be resolved by confocal microscopy, in pupal photoreceptors they clearly occupy different membrane domains.
Establishing the Apical–Basal Polarity in Pupal Photoreceptors Requires Baz. In embryonic epithelia, loss of Baz causes complete failure of ZA formation and loss of apical–basal polarity (11). Interestingly, in third instar larval eye discs, although Baz itself is expressed around the apical cell boundary just like Dlt (Fig. 2A), Dlt staining is not affected in baz clones located either in front of or behind morphogenetic furrow (Fig. 2B). Apparently, Baz is not required for the apical–basal polarity in larval photoreceptors. However, from late larval to early pupal stage the apical surfaces of R1–R8 turn 90° toward the center of ommatidium, and the apparent apical–basal axis of photoreceptor becomes perpendicular to the distal-proximal axis (2). It is possible that such morphological changes involve respecification of apical–basal polarity in pupal photoreceptors that may require Baz.
Fig. 2.
Baz is required for establishing the apical–basal polarity in pupal photoreceptors. (A) Baz antibody (red) stains the apical cell boundary in third instar retina. (B)A baz clone (GFP negative) before the morphogenetic furrow in third instar eye disk double-stained with GFP (green) and Dlt (red) showing that Dlt staining is not disrupted in baz clones. Similar results are seen in baz clones after the morphogenetic furrow at same stage (data not shown). (C) Dotted are two ommatidia at 37% pd: a wild-type ommatidium at right with normal pattern of Arm and Baz staining (arrowheads) and a mosaic ommatidium at left containing only one spot of wild-type Arm and Baz staining (arrowhead). Note the severely disrupted Arm and phalloidin staining (highlighted by arrows) in mosaic ommatidium compared with the wild type at right. (D) Two mosaic (dotted) and one wild-type (bottom left) ommatidia at 47% pd triple-stained with GFP (to label the wild-type photoreceptors), Arm, and phalloidin. Arrowheads highlight the wild-type Arm staining. Arrows point to the abnormal staining patterns of Arm and phalloidin in mosaic ommatidia. Also note the abnormal cell shape and Arm/phalloidin staining patterns of wild-type photoreceptors (GFP-positive) in the mosaic ommatidia. (E) Three of seven confocal Z sections (total of 5.5 μm) of a mosaic ommatidium (upper right) at 44% pd double-stained with GFP (to label wild-type photoreceptors) and Dlt. Arrows point to the Dlt staining between wild-type photoreceptors, and arrowhead highlights the mispositioned Dlt staining in a mutant photoreceptor (GFP negative). (F) A 40% pd retina contains a relatively large patch of baz clone (GFP negative) double stained with GFP (green) and Elav (red) antibodies. Four ommatidia (dotted) in the center of clone contain only baz photoreceptors, surrounded by mosaic ommatidia. (G) A light microscopy section of adult eye containing baz clones. Wild-type ommatidia are blue tinted, whereas malformed ommatidia with missing photoreceptors and deformed rhabdomeres are left black and white.
Indeed, we found Baz is essential in establishing the apical–basal polarity in pupal photoreceptors. As shown in Fig. 2 C and D, at either 37% or 47% pd Arm is severely disrupted in baz mutant photoreceptors within mosaic ommatidia, indicating a complete loss of ZA formation, which is the hallmark of apical–basal polarity in photoreceptors. Phalloidin, which stains F-actin in developing rhabdomeres, normally marks a solid patch at the center of wild-type ommatidia at these stages (Fig. 2 C and D). But inside mosaic ommatidia phalloidin staining becomes fragmented, randomly positioned, or lost in both wild-type and baz photoreceptors, with reduced Arm staining sometimes misplaced over the edge of fragmented phalloidin staining (Fig. 2 C and D). Because Sdt/Crb/Dlt are established as apical markers in photoreceptors by this study and previous work (5, 6), we also looked at the staining of Dlt in baz mutant photoreceptors. Shown in Fig. 2E are three representative confocal Z sections (total depth of 5.5 μm) of a mosaic ommatidium. Although the apical localization of Dlt remained in wild-type photoreceptors, it is either lost or becomes randomized in baz photoreceptors (arrowhead in Fig. 2E).
To exclude the possibility that the severe polarity phenotype in baz mutant photoreceptors may be due to a deficit in adopting neuronal cell fate, we stained both third instar larval eye discs and 40% pd pupal retinas containing baz clones with anti-Elav antibody, a pan-neuronal marker that stains all of the neurons. At both stages, Elav staining demonstrates a wild-type pattern and, with few exceptions, correct cell numbers in either mosaic ommatidia or ommatidia composed by all baz cells (Fig. 2F). Such data argue that baz photoreceptors are of neuronal fate, and the polarity defects are not due to cell fate change.
In addition, we found the size of baz clones in eye discs is much smaller than their wild-type twin spots (data not shown), suggesting baz cells may suffer from certain proliferation disadvantage in eye discs. Consistently, although baz clones can be readily identified in early pupal retina, such clones were never found in late pupal retinas. We suspect that impaired apical–basal polarity in baz photoreceptors causes either strong cell growth defects or even cell death at later pupal stages. Adult eyes containing baz clones were found having many malformed ommatidia containing fewer than seven rhabdomeres (Fig. 2G), probably because of the eventual loss of baz photoreceptors in mosaic ommatidia after pd.
We conclude that loss of Baz in early pupal photoreceptors causes a failure in establishing the apical–basal polarity, and no apical membrane domains are properly specified in baz photoreceptors. Because ZA formation involves cell–cell adhesions, it is expected that baz mosaic ommatidia should have certain cell nonautonomous effects. For instance, in Fig. 2C, there is only a single dot of Baz/Arm staining in mosaic ommatidium. Apparently, a wild-type photoreceptor is unable to form continuous ZA with one of its neighboring baz photoreceptors; otherwise, two dots of Baz/Arm staining should be seen on each side. The shape and polarity of wild-type photoreceptors are frequently affected in mosaic ommatidia (Fig. 2 C–E), although their polarity defects are generally much less severe. Such phenotypes likely represent secondary defects in multicellular assembly of baz mosaic ommatidia, resulting from the primary polarity defects in baz photoreceptors.
Sdt Plays a Limited Role in Developing the Apical–Basal Polarity of Photoreceptors. To investigate whether Sdt is required for apical–basal polarity in pupal retina, we generated sdt whole eye clones (15) and looked for potential ZA defects in photoreceptors. In early sdt photoreceptors at 38% pd, ZA formation is only mildly disrupted. Arm staining pattern seems less organized and often extends basal-laterally (Fig. 3A) as compared with the wild type at the same stage (Fig. 1 A). Overall, the phenotype appears to be similar to that of crb (5, 6). At late pupal stages the ZA defects in sdt photoreceptors are no longer detectable. Arm staining is formed between cell–cell junctions (Fig. 3 B and C) and extends fully along the distal-proximal axis (Fig. 3E) amid strong mislocalization of Dlt (Fig. 3 C and E; see also Fig. 4D). EM analysis in adult sdt eye confirms normal AJ structure in most, if not all, photoreceptors (Fig. 3D). Such recovery seems to be more complete than in crb photoreceptors, as in crb photoreceptors ZA often fails to fully extend proximally (5, 6). Perhaps, in sdt photoreceptors the deficit of ZA integrity is alleviated by residual membrane distribution of Crb (Fig. 4D). Compared with the phenotype seen in baz photoreceptors, these data indicate that the role of Sdt/Crb/Dlt in establishing apical–basal polarity in photoreceptors is much more limited.
Fig. 3.
Polarity defects in sdt mutant pupal photoreceptors. All panels are of sdt genotype, except for B (wild type). (A) Mislocalization of Arm and Baz in sdt photoreceptors at 38% pd. Five ommatidia are shown. Arrows, laterally mislocalized staining of Baz and Arm. (B and C) A close-up view of a wild-type (B) and an sdt (C) ommatidium at 73% pd double-stained by Arm and Dlt. Although Dlt staining is disrupted in sdt photoreceptor (C), the Arm staining is nearly indistinguishable from the wild type (B). (D) EM section showing normal ZA structure (arrows) in adult sdt photoreceptors. (Scale bar, 1 μm.) (E) A lateral view of two sdt ommatidia (asterisks) at 73%, double-stained with Arm and Dlt. Arrowheads highlight a ZA extending fully along the distal-proximal axis. This image is a projection of a Z-section series of 7.3 μm.
Fig. 4.
Mislocalization of Baz, Crb, and Dlt in sdt pupal photoreceptors. All panels here are of sdt genotype, except for E (crb). (A) Premature accumulation of Baz at 43% pd in sdt photoreceptors. Arrowheads point to Baz staining at ZA locations, including one that has diffused laterally (arrowhead). Note the diffuse appearance of Baz, Dlt, and phalloidin stainings at the center of ommatidium. (B) Localization of Baz in rhabdomere is not affected in sdt photoreceptors at 74% pd. (C)A sdt mutant ommatidium at 71% pd showing the mislocalization of Crb into the developing rhabdomere (arrowheads). For reasons unknown, Crb staining is particularly strong in R7 sdt photoreceptor (arrow). (D) Mislocalization of Dlt and Crb in 74% pd sdt photoreceptors. Note that mislocalized Crb and Dlt do not fully overlap with each other. (E) Mislocalization of Sdt and Dlt in 72% pd crb photoreceptors. Note the high cytoplasmic stainings of Sdt and Dlt.
Sdt Is Required for Preventing Premature Baz Localization in Rhabdomeres. Loss of Sdt in embryos disrupts both Arm and Baz into random aggregates, but these two proteins remain apical and associated with each other (9), marking a remnant of apical–basal polarity in the absence of Sdt/Crb/Dlt complex. Similarly, in 38% pd sdt photoreceptors, Baz localization is also disrupted, but it remains colocalized with Arm (Fig. 3A). However, at a slightly later stage of 43% pd, a dramatic accumulation of Baz is seen in the apical region of photoreceptors that costains with Dlt and phalloidin (Fig. 4A, compare with Fig. 1B for wild type). And although Baz is still seen at ZA, the level of staining is reduced (arrowheads in Fig. 4A). Such premature accumulation of Baz at apical region is never seen in control wild-type retinas of the same age that were processed in parallel (Fig. 1B). Because in wild-type photoreceptors only after 55% pd Baz starts relocating from ZA to the rhabdomere (Fig. 1C), this phenotype shows that at 43% pd Baz has already become capable of localizing into the apical region of photoreceptors where rhabdomeres develop, but is normally limited by Sdt/Crb/Dlt complex acting as a potential membrane barrier. At late stages such as 74% pd, the rhabdomere localization of Baz is normal in sdt photoreceptors (Fig. 4B).
Sdt Is Required for the Proper Localization of Crb and Dlt at Stalk Membrane. The subcellular localization patterns of Sdt, Crb, and Dlt suggest that these three proteins may form a complex in photoreceptors as they do in embryonic epithelia. In embryonic epithelia, Sdt, Crb, and Dlt mutually depend for their subcellular localization (9), as both Crb and Dlt stainings are reduced to sparse spots at the cell surface in sdt embryos, and similar disruption of Sdt is also observed in crb embryos (9). Similarly, sdt mutant photoreceptors show strong mislocalization of Dlt and Crb (Figs. 3C and 4 C and D); at ≈73% pd, Dlt staining is diffuse around the apical region (Figs. 3C and 4D) and Crb is mislocalized to the developing rhabdomere with only faint apical staining (Fig. 4 C and D). However, the protein levels of Crb and Dlt are not significantly reduced in sdt photoreceptors, unlike embryonic epithelia. And although loss of Sdt still leaves Dlt and Crb preferentially in the apical side of photoreceptors, Dlt and Crb do not appear to overlap completely with each other (Fig. 4D). We also looked at the staining patterns of Sdt and Dlt in crb mutant photoreceptors and found at 72% pd both Sdt and Dlt stainings are largely cytoplasmic, with some puncta seen around the apical side (Fig. 4E).
Sdt-Regulated Localization of Crb Is Essential for the Morphogenesis of Stalk Membrane and Rhabdomere. In crb photoreceptors, loss of Crb causes both shortening of stalk membrane and defects in rhabdomere shape (5, 6). Consistent with the notion that Sdt-regulated localization of Crb on stalk membrane should be critical for the biogenesis of stalk membrane, by EM analysis we found stalk membranes in adult sdt photoreceptors are reduced by 40% compared with the wild type (Fig. 5 A, B, and D–F), and such reduction is comparable to that of crb photoreceptors (44%, Fig. 5 C and F; 50% in ref. 6). Similar to the loss of Crb phenotype, stalk membranes in sdt photoreceptor are often shorter and smoother with no deep folds compared with wild-type stalk membranes (Fig. 5 D and E). These observations support the notion that Sdt, Crb, and Dlt act together, although it remains possible that Sdt may also act through proteins other than Crb and Dlt to regulate the stalk membrane biogenesis.
Fig. 5.
Morphogenesis defects of stalk membrane and rhabdomere in adult sdt photoreceptors. (A–C) EM section of wild-type (A), sdt (B), and crb (C) adult ommatidia. y w sn FRT-19A whole eye clones serve as wild-type control. (Scale bars, 2.5 μm.) (D and E) Examples of tracing the contour of stalk membrane in sdt (D) and wild-type (E) photoreceptors. Arrowhead, AJ. (F) Length of stalk membrane in wild-type (n = 64), sdt (n = 157), and crb (n = 22) photoreceptors. (G) EM section of a single rhabdomere of adult sdt photoreceptor. Asterisks highlight the vesicle like structures at the bottom of rhabdomere. (Scale bar, 1 μm.) (H–J) Close-up view at the base of rhabdomere in sdt (H), wild-type (I), and crb (J) adult photoreceptors. Arrows in H point to the overextensions of microvilli at the base of rhabdomere. Asterisk in H indicates a potential extension of microvillus base whose connection to the base membrane is apparently off the section plane. Arrowheads in I and J indicate the neck and the base membrane of microvilli.
The rhabdomere is a specialized membrane structure packed with ≈60,000 microvilli (7). The shape of rhabdomeres in sdt photoreceptors is also disrupted as in crb mutant. sdt rhabdomeres become irregularly shaped, frequently expanded, split, or in contact with neighboring rhabdomeres (Fig. 5B). Unlike crb mutants, however, loss of Sdt also disrupts the base membrane of microvilli in rhabdomere (Fig. 5 G and H). In sdt photoreceptors, the microvilli base membrane seems to be severely disrupted by vesicle-like membrane structures concentrating around the base of the rhabdomere (Fig. 5G). At high magnification, some of these vesicles are clearly the overextension of the individual microvilli base membrane (Fig. 5H), whereas some isolated vesicles likely represent the off-the-section plane base membrane overextensions. We speculate that such phenotypes may be caused by the mislocalization of Crb in the rhabdomeres because overexpression of Crb expands the apical membrane in epithelia as well as stalk membrane in photoreceptors (6, 16).
Discussion
Recent work (5, 6) demonstrates a function of Crb in photoreceptor morphogenesis. Here, we further examined the roles of Baz and Sdt, two proteins also of essential epithelial polarity functions, in pupal retina development. Our data suggest that polarity and cellular architecture in pupal photoreceptors develop in two distinct processes. First, the specification of the apical–basal polarity requires Baz. Later, the development of the elaborate apical membrane structure requires Sdt/Crb/Dlt. Compared with the simple embryonic epithelial cells, photoreceptors depend on a well conserved function of Baz in establishing apical–basal polarity. However, other proteins, such as DaPKC and Sdt/Crb/Dlt, show novel subcellular localization and functions that are not readily predicted from embryonic studies.
Because of the limitation that baz photoreceptors do not survive in late pupal stages, the functional implications of the dynamic relocalization of Baz from ZA to rhabdomere after 55% pd remain unexplored. Apparently, ZA integrity in late pupal photoreceptors is independent of Baz; Baz may play different roles in early and late pupal photoreceptor development. Nonetheless, the role of Baz in cell polarization seems to be well conserved across different cell types and different organisms. Recent study on cultured hippocampal neurons has shown that spatially localized mammalian homologues of Baz complex (mPar-3/mPar-6/aPKC) are required for neuronal polarization and axonogenesis (26). Drosophila photoreceptor is also polarized along the distal-proximal axis with its axon projecting into the optical lobe. Whether Baz is also involved in this distal-proximal polarization of photoreceptor remains to be tested. Because in early pupal photoreceptors DaPKC colocalizes with Sdt/Crb/Dlt complex instead of Baz, it will be of interest to identify the protein components working with Baz to regulate the apical–basal polarity of photoreceptors. A recent report has shown that a Drosophila PAK (p21-activated kinase) protein Mbt is localized to the AJ and regulates the photoreceptor morphogenesis through the Cdc42 pathway (27). Loss of function phenotype of Mbt in photoreceptors seems similar to baz mutant, although somewhat weaker, raising the possibility that part of the Baz function in specifying apical–basal polarity is to regulate the actin cytoskeleton through Rho GTPase pathways.
Unlike the Baz complex, the Sdt/Crb/Dlt complex in embryonic epithelia also seems to act as a complex in pupal photoreceptors. Although the subcellular localization of Sdt, Crb, and Dlt differs from simple epithelia, these proteins remain mutually dependent for their proper localization on the stalk membranes of photoreceptors. The function of Sdt/Crb/Dlt in pupal photoreceptor is not at the level of maintaining the apical–basal polarity as in embryonic epithelia, but for specifying and controlling the growth of special membrane domains such as stalk membrane. Consistent with this hypothesis, Crb is also required in the growth of apical membrane of salivary tubes (17). Overall, while protein–protein interactions in Sdt/Crb/Dlt complex are preserved in pupal photoreceptors, the function of the complex apparently changes and adapts to the process of developing specific cellular architecture. To further understand this process, it will be of interest to characterize the mechanisms that are responsible for targeting Sdt/Crb/Dlt complex specifically to the stalk membrane in pupal photoreceptors.
The exact mechanisms of rhabdomere defects in crb and sdt photoreceptors are still unclear (5, 6), as biogenesis of rhabdomere and stalk membrane is a complex process yet to be fully characterized (18, 19). Probably regionalizing the apical membrane into stalk membrane and rhabdomere requires a balance between these two subdivided membrane domains, and defects in stalk membrane biogenesis also impair the rhabdomere development. In addition, premature localization of Baz into the rhabdomeres caused by loss of Sdt may also contribute to the defects of rhabdomere genesis. How Sdt functions in preventing Baz from accumulating prematurely in rhabdomere remains to be characterized.
It is noteworthy that in crb or sdt mutant photoreceptors the stalk membrane is only shortened, and despite structural defects the rhabdomere still forms. Proteins other than the Sdt/Crb/Dlt complex may be involved in the membrane specialization of the apical surface in photoreceptors. One candidate could be DaPKC based on its transient subcellular localization on stalk membranes. Interactions between mammalian homologues of DmPar-6/DaPKC and Sdt/Crb/Dlt complexes have just been reported (20), consistent with the colocalization of DaPKC with Sdt/Crb/Dlt on stalk membrane. In addition, human homologues of Crb are implicated in retinitis pigmentosa and Leber congenital amaurosis, two heritable forms of human disease of photoreceptor degeneration (21–24). In Drosophila, crb photoreceptors also degenerate under intense lighting conditions (25), resembling the phenotype in human diseases. It would be of interest to find out whether mutations in the human homologues of Sdt, Dlt, and possibly aPKC lead to photoreceptor degeneration.
Acknowledgments
We thank U. Tepass, S.-C. Lam, K.-W. Choi, A. Wodarz, E. Knust, S. Syfried, the Developmental Studies Hybridoma Bank, and the Bloomington Stock Center for antibodies and fly stocks. We also thank D. Ready and C. H. Yang for advice on pupal retina and imaginal disk dissection techniques, laboratory members for their comments and help, and the Ralston laboratory at the University of California, San Francisco, for the use of its electron microscope. Y.H. is a research associate and L.Y.J. and Y.-N.J. are Investigators of the Howard Hughes Medical Institute.
Abbreviations: ZA, zonula adherens; pd, pupal development; Baz, Bazooka; Sdt, Stardust; Crb, Crumbs; EM, electron microscopy.
References
- 1.Johnson, K. & Wodarz, A. (2003) Nat. Cell Biol. 5 12–14. [DOI] [PubMed] [Google Scholar]
- 2.Longley, J. R. & Ready, D. F. (1995) Dev. Biol. 171 415–433. [DOI] [PubMed] [Google Scholar]
- 3.Bilder, D., Schober, M. & Perrimon, N. (2003) Nat. Cell Biol. 5 53–58. [DOI] [PubMed] [Google Scholar]
- 4.Tanentzapf, G. & Tepass, U. (2003) Nat. Cell Biol. 5 46–52. [DOI] [PubMed] [Google Scholar]
- 5.Izaddoost, S., Nam, S.-C., Bhat, M. A., Bellen, H. J. & Choi, K.-W. (2002) Nature 416 178–182. [DOI] [PubMed] [Google Scholar]
- 6.Pellikka, M., Tanentzapf, G., Pinto, M., Smith, C., McGlade, C. J., Ready, D. F. & Tepass, U. (2002) Nature 416 143–149. [DOI] [PubMed] [Google Scholar]
- 7.Kumar, J. P. & Ready, D. F. (1995) Development (Cambridge, U.K.) 121 4359–4370. [DOI] [PubMed] [Google Scholar]
- 8.Müller, H.-A. J. & Wieschaus, E. (1996) J. Cell Biol. 134 149–163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hong, Y., Stronach, B., Perrimon, N., Jan, L. Y. & Jan, Y. N. (2001) Nature 414 634–638. [DOI] [PubMed] [Google Scholar]
- 10.Tepass, U. (1996) Dev. Biol. 177 217–225. [DOI] [PubMed] [Google Scholar]
- 11.Grawe, F., Wodarz, A., Lee, B., Knust, E. & Skaer, H. (1996) Development (Cambridge, U.K.) 122 951–959. [DOI] [PubMed] [Google Scholar]
- 12.Bachmann, A., Schneider, M., Theilenberg, E., Grawe, F. & Knust, E. (2001) Nature 414 638–643. [DOI] [PubMed] [Google Scholar]
- 13.Wodarz, A., Ramrath, A., Grimm, A. & Knust, E. (2000) J. Cell Biol. 150 1361–1374. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Petronczki, M. & Knoblich, J. A. (2001) Nat. Cell Biol. 3 43–49. [DOI] [PubMed] [Google Scholar]
- 15.Stowers, R. S. & Schwarz, T. L. (1999) Genetics 152 1631–1639. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Wodarz, A., Hinz, U., Engelbert, M. & Knust, E. (1995) Cell 82 67–76. [DOI] [PubMed] [Google Scholar]
- 17.Myat, M. M. & Andrew, D. J. (2002) Cell 111 879–891. [DOI] [PubMed] [Google Scholar]
- 18.Chang, H.-Y. & Ready, D. F. (2000) Science 290 1978–1980. [DOI] [PubMed] [Google Scholar]
- 19.Sang, T.-K. & Ready, D. F. (2002) Development (Cambridge, U.K.) 129 143–154. [DOI] [PubMed] [Google Scholar]
- 20.Hurd, T. W., Gao, L., Roh, M. H., Macara, I. G. & Margolis, B. (2003) Nat. Cell Biol. 5 137–142. [DOI] [PubMed] [Google Scholar]
- 21.den Hollander, A. I., ten Brink, J. B., de Kok, Y. J. M., van Soest, S., Ingeborgh van den Born, L., van Driel, M. A., van de Pol, D. J. R., Payne, A. M., Bhattacharya, S. S., Kellner, U., et al. (1999) Nat. Genet. 23 217–221. [DOI] [PubMed] [Google Scholar]
- 22.den Hollander, A. I., Heckenlively, J. R., van den Born, L. I., de Kok, Y. J. M., van der Velde-Visser, S. D., Kellner, U., Jurklies, B., van Schooneveld, M. J., Blankenagel, A., Rohrschneider, K., et al. (2001) Am. J. Hum. Genet. 69 198–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.den Hollander, A. I., Johnson, K., de Kok, Y. J. M., Klebes, A., Brunner, H. G., Knust, E. & Cremers, F. P. M. (2001) Hum. Mol. Genet. 10 2767–2773. [DOI] [PubMed] [Google Scholar]
- 24.Lotery, A. J., Jacobson, S. G., Fishman, G. A., Weleber, R. G., Fulton, B., Namperumalsamy, P., Héon, E., Levin, A. V., Grover, S., Rosenow, J., et al. (2001) Arch. Ophthalmol. 119 415–420. [DOI] [PubMed] [Google Scholar]
- 25.Johnson, K., Grawe, F. Grzeschik, N. & Knust, E. (2002) Curr. Biol. 12 1675–1680. [DOI] [PubMed] [Google Scholar]
- 26.Shi, S.-H., Jan, L. Y. & Jan, Y. N. (2003) Cell 112 63–75. [DOI] [PubMed] [Google Scholar]
- 27.Schneeberger, D. & Raabe, T. (2003) Development (Cambridge, U.K.) 130 427–437. [DOI] [PubMed] [Google Scholar]





