SEARCH

SEARCH BY CITATION

Keywords:

  • Drosophila retina;
  • cone cells;
  • cell shape;
  • cell contacts;
  • pattern formation;
  • hibris;
  • Nephrin

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Cell shapes and contacts are dynamically regulated during organogenesis to enable contacts with relevant neighboring cells at appropriate times. During Drosophila larval eye development, an apical contact is established between one pair of non-neuronal cones cells, precluding contact between the opposing pair. Concurrent with changes in cell shape, these contacts reverse in early pupal life. The reversal in cone cell contacts occurs in a posterior to anterior gradient across the eye, following the developmental gradient established in the larval eye imaginal disc. Hibris (Hbs), an Immunoglobulin cell adhesion molecule homologous to vertebrate Nephrin, is required for cone cell morphogenesis. In hbs null mutants, a majority of cone cells fail to both establish wild-type contacts and achieve mature cone cell shapes. hbs acts cell autonomously in the cone cells to drive these changes. The work presented here indicates hbs contributes to the remodeling of cell contacts and cell shapes throughout development. Developmental Dynamics 238:2223–2334, 2009. © 2009 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Spatial and temporal regulation of cell adhesion is essential for myriad processes and events throughout the development of multicellular organisms. When adhesion-mediated cell contacts are compromised, epithelial integrity deteriorates, cell–cell communication is disrupted, cell movements stall, and cell shapes become distorted. Throughout development, adhesion between cells must be malleable to enable organisms and epithelia to morph and grow. At a cellular level, this dynamic behavior permits cells to migrate and change their shapes and contacts, factors that facilitate communication by altering spatial configurations. These changes in cell shape enable cells to communicate with evolving subsets of cells as development proceeds. Contacts, and therefore adhesion, with neighboring cells are dynamic in the Drosophila eye, in which the growing subunits of the compound eye are constantly adding new cells to nascent ommatidia throughout larval and the first half of pupal development. Here, we describe the changes in cell shapes and contacts between the four non-neuronal cone cells of the eye, and discuss the role of one cell adhesion molecule, Hibris (Hbs), in regulating this event.

The Drosophila compound eye is a polarized, monolayer epithelium composed of approximately 800 unit eyes, or ommatidia (reviewed in Wolff and Ready,1993). Ommatidia are composed of 20 cells, including eight photoreceptor neurons (designated R1 through R8), four lens-secreting, non-neural cone cells, three classes of pigment cells, and the four-cell bristle units. The eye grows by accretion, whereby distinct classes of cells are recruited sequentially into the growing ommatidia. Maturation of ommatidia follows the morphogenetic furrow, a moving front of differentiation that initiates at the posterior margin of the eye and moves anteriorly over a 2-day period during larval development (Ready et al.,1976).

As the ommatidial preclusters mature, they progress through a series of stereotypic contacts with neighboring cells. These contacts are often transient, modified in response to signals from neighboring cells as the retina adopts its final, precise network of cells. The progression of cell contacts in the fly eye is particularly evident in the two pairs of cone cells, the anterior/posterior pair (A/P) and the polar/equatorial pair (Po/Eq). The A/P pair is recruited first, followed by the Po/Eq pair. The anterior and posterior cone cells establish a contact on the apical surface of the eye in late larval development, precluding contact between the polar and equatorial cone cells. This contact is maintained through early pupal life, at which point these apical contacts reverse, such that the polar and equatorial cone cells establish a contact and the anterior and posterior cone cells lose their contact (Cagan and Ready,1989). The mechanisms and molecules that regulate this transition and the manner in which cone cells are organized is a topic of growing interest (Bao and Cagan,2005; Hayashi and Carthew,2004; Bertet et al.,2004). Furthermore, these molecules and mechanisms are likely to be required ubiquitously throughout development for analogous events in both vertebrates and invertebrates.

Cell adhesion molecules (CAMs) perform a variety of roles, among them the establishment and maintenance of cell junctions that serve to preserve epithelial integrity and regulate cell shape. Members of the family of classic cadherins are integral components of the adherens junctions that connect epithelial cells and are recognized for their roles in establishing cell contacts and maintaining epithelial structure. Members of the cadherin superfamily mediate adhesion via interactions between the extracellular domains of cadherins expressed on adjacent cells (reviewed in Gumbiner,2005). Two members of this family, N-cadherin and E-cadherin, play prominent roles in cone cell junction formation and regulation of cone cell shape and contacts (Hayashi and Carthew,2004). Both the physical adhesive properties as well as differential localization of the classical cadherins are important in regulating the mutually dependent properties of cone cell shapes and contacts.

Members of the Immunoglobulin (Ig) superfamily of CAMs share a transmembrane domain and extracellular Ig repeats. The Neph/Nephrin subfamily of proteins, members of the Ig CAMs, mediate cell adhesion through their Ig repeats in homo- and heterotypic interactions. The Ig CAMs play a key role in modulating cell contacts and shapes, morphogenesis, pattern formation, cell growth, migration, and signaling in multiple systems (reviewed in Walsh and Doherty1993). Nephrin/Hbs plays critical roles in modulating cell adhesion in multiple systems. In the mammalian kidney, the second, more stringent level of filtration is accomplished by a meshwork of extracellular proteins that includes nephrin (Wartiovaara et al.,2004). Mutations in the nephrin gene, NPHS1A, are associated with congenital kidney disease. Furthermore, genetic removal of nephrin in a mouse model results in phenotypes similar to the defects observed in human patients (reviewed in Oh et al.,2004). The aPKC/Par3/Par6 cell polarity protein complex binds Nephrin, the vertebrate homolog of Hbs, suggesting a mechanism by which cell adhesion can be modulated in response to cell polarity or movement (Hartleben et al.,2008).

There are four members of the Neph/Nephrin subfamily of Ig proteins in Drosophila: two homologs of the Neph ligand, irreC-roughest (rst) and kin of irreC-roughest (kirre, also known as dumbfounded); and two homologs of the Nephrin receptor, hibris (hbs) and sticks-and-stones (sns).

Expression studies in the Drosophila embryo demonstrate that hbs is expressed in migrating prohemocytes and is found in putative stellate cell precursors just prior to their intercalation into Malpighian tubules (Artero et al.,2006). Overexpression and loss-of-function hbs phenotypes in flies include rough eyes and a partial block of myoblast fusion in embryonic muscle (Artero et al.,2001; Dworak et al.,2001) The eye roughness in hbs-RNAi flies is a consequence of an underlying failure to properly sort interommatidial lattice cells (Bao and Cagan,2005).

Here, we identify a new role for hbs in Drosophila eye development: hbs contributes to the establishment of the mature configuration of cone cell contacts and cone cell shapes. Hbs participates in driving the exchange of contacts from the A/P pair of cone cells to the Po/Eq pair. In the absence of a functional hbs gene, the cone cells in many hbs mutant ommatidia establish novel configurations of cone cells. In addition, many hbs mutant cone cells retain a more immature, rounded morphology, failing to adopt the pointed morphology of mature cone cells. Mosaic analysis indicates that hbs is ommatidium autonomous and is required in at least one of the four cone cells for wild-type morphology. Finally, hbs modulates cone cells contacts without altering the asymmetrical distribution of N-cadherin. These data suggest that hbs acts in a novel fashion to modulate dynamic cell contacts.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Cone Cell Shape Is Correlated With Ommatidial Maturity

All cells that comprise the Drosophila retina undergo significant morphological changes as they adopt their fates and the characteristic features unique to their specified cell types. The shapes of the cone cells change dramatically during early pupal life throughout the course of a prominent morphological transformation that reverses the apical contacts between the two pairs of cone cells. The initial apical contact is established between the anterior (A) and posterior (P) cone cells after they migrate towards one another over the tips of the photoreceptor cells in the third instar eye disc (Cagan and Ready,1989). The “relaxed,” rounded configuration of the four cone cells in the third larval instar persists through early pupal development (Fig. 1A, A′, A″). The A and P cone cells maintain their rounded shapes longer than do the Po and Eq cone cells, and by 48 hr apf, the shapes of the Po and Eq cone cells are variable, with some having points at their poles (Fig. 1B″, lower cell of Po/Eq pair) and others remaining round (Fig. 1B″, upper cell of Po/Eq pair). At about 48 hr after puparium formation (apf; 18°), cone cells in ommatidia in the posterior third of the eye begin to transition from an A/P contact to a Po/Eq contact. As they begin to reverse contacts with the Po/Eq pair, the anterior and posterior cone cells adopt a more ellipsoid shape, becoming elongated along their D/V axes (compare Fig. 1A″ to Fig. 1A‴). During the transitional phase, the two members of the A/P pair seldom change their shapes synchronously. Rather, one cell of the pair lags behind the other in that the lagging cell retains its more juvenile, rounded state while its partner stretches more quickly along its D/V axis to adopt the mature, ellipsoid morphology (Fig. 1B‴). Unexpectedly, in the majority of these cases in a 50-hr eye, it is the posterior cone cell that is delayed; since the posterior of the eye is more mature than the anterior, the prediction would have been that the anterior cell is delayed. By 60 hr, this bias dissolves, such that a similar number of delays are seen in anterior and posterior cone cells. A similar asymmetric phenomenon has been observed in the pairs of primary pigment cells (Cagan and Ready,1989), which ensheathe the cone cells asymmetrically, although the relative rate of the anterior versus posterior primary pigment cell morphogenesis within a single ommatidium has not been documented.

thumbnail image

Figure 1. Changes in cone cell shape accompany reversal of their contacts during pupal morphogenesis. Confocal micrographs of cell junctions in fields of ommatidia, visualized with Arm-GFP, and accompanying tracings of 48 (A, A′), 60 (B, B′), 72 (C, C′), and 92 hour (D, D′) pupal eyes raised at 18°. Confocal micrographs of single ommatidia with distinctive configurations of cone cells (A″, A‴, B″, B‴, C″ and D″). (E) Color-coded key for identification of each cell type in ommatidium. Anterior is to the right for all panels. A, B: The anterior and posterior cone cells establish a contact late in larval life and maintain this contact up to approximately 60 hours of pupal development. In the initial phase, the apical profiles of all four cone cells are relatively round (A′ and B′, asterisk), however, the polar and equatorial cone cells may be asymmetric. For example, in A″, the upper cell is more rounded at its apex and the lower cell more pointed. C: As the four cone cells initiate their transition in contacts, the quartet forms an “X,” with the four cells meeting centrally, resembling wedges of a pie (C′, asterisk and A‴). Some transitional ommatidia have asymmetrical A/P cone cell pairs; for example, this posterior cone cell appears round and immature whereas the anterior cone cell is elongated along the Po/Eq axis (B‴). D: Newly established contacts are minimal (C′, arrowheads), but expand over the next 20 hr (D′, arrowhead). “Straggler” ommatidia, in which immature contacts exist between all four cone cells, yet the cone cell shapes are mature (elongated and pointed at their tips) are evident (C″). Once the polar and equatorial cone cells establish contacts, the anterior and posterior cone cells return to a more rounded, now oblong, shape. As the cone cells expand over the next 20 hours, they maintain rounded edges at their contacts with the primary pigment cells (D″).

Download figure to PowerPoint

Finally, although the morphology of the cone cells is generally tightly correlated with the exchange of contacts (cells are generally more rounded prior to the reversal and generally more pointed following the reversal), this is not an absolute correlation, as the morphology of the cone cells in the last ommatidia to transition, an event that occurs between 72–76 hr (Fig. 2A; see below), more closely resembles the mature, pointed ommatidia than the immature, round ommatidia (Fig. 1C″, compare to immature, rounds cells in Fig. 1A″ and mature cells in Fig. 1D″).

thumbnail image

Figure 2. Cone cell contacts reflect the anterior-posterior gradient of eye development. A: Percent of anterior-posterior (yellow bar), transitional (orange bar), and polar-equatorial (red bar) cone cell contacts in the anterior (a), center (c), and posterior (p) thirds of pupal eyes at time points indicated. B: Table of standard deviations for data shown in A. See Experimental Procedures section for sample size.

Download figure to PowerPoint

In addition to the two morphologies described above (initial A/P contacts and mature Po/Eq contacts), a third configuration in which the four cone cells meet at a common point in the center, forming an “X” (Fig. 1C′, asterisk), is also evident. Fewer ommatidia of this configuration are seen, and a peak in their number corresponds to the time at which the cone cells transition from A/P to Po/Eq contacts. This state likely represents the transitional conformation since (1) at some point during the transition, all four cells should appear to contact one another, and (2) the greatest number of ommatidia in the “X” configuration is evident roughly at the peak phase of A/P to Po/Eq transitions (56–64 hr apf, Fig. 2A). During this transitional conformation, one of two configurations is seen: either all four cone cells are similar in shape and size and the two lines forming the “X” are straight, such that the quartet resembles four slices of a pie (Fig. 1A‴), or the anterior and posterior cone cells are disproportionate in size and/or shape (Fig. 1B‴) and the lines forming the “X” are somewhat curved (Fig. 1B″). Once a contact is established between the polar and equatorial cone cells (Fig. 1C and 1C′), an event that is achieved in all ommatidia by 76 hr apf (note that virtually all transitions are complete by 72 hr apf and the few remaining stragglers reverse over the next four hours), this contact expands along the AP axis of the ommatidium (Fig. 1D″).

Cone Cell Contact Reversal Follows the Posterior to Anterior Gradient of Eye Development

The reversal of cone cell contacts between the A/P and Po/Eq cone cell pairs is not uniform across the eye. Rather, the developmental gradient established by the posterior to anterior movement of the morphogenetic furrow in the third larval instar (reviewed in Wolff and Ready,1993) persists to this point of pupal development. The gradient is distinctly more subtle during this phase than it is in the eye imaginal disc, becoming evident only when the state of the cone cell contacts is surveyed over a 32-hr period and quantitated in posterior, central, and anterior thirds of the eye. Reversal of cone cell contacts at the level of adherens junctions occurs over a period of about 28 hr (18°, Fig. 2A, B), with the initial transitions occurring in the posterior third of the eye at 48 hr apf and the final ommatidia reversing their contacts in the anterior third of the eye at about 76 hr apf. By 56 hr apf, greater than 50% of ommatidia in the posterior two thirds of the eye have either completed or started to transition, whereas the 50% mark is not reached in the anterior third of the eye until 12 hr later, or 68 hr apf (Fig. 2A).

Throughout this transition phase, the maturational gradient is not uniform: between the posterior and central thirds of the eye, the gradient is relatively shallow, whereas it is significantly steeper between the posterior two thirds and anterior third of the eye (Fig. 2A). For example, at 56 hr apf, the percentages of ommatidia with A/P cone cell contacts are 46 and 34% in the center and posterior thirds of the eye, respectively, whereas 84% of ommatidia in the anterior third still have the immature A/P contact. Indeed, it is not until 12 hr later, or 68 hr apf, that the number of ommatidia with A/P cone cell contacts in the anterior third drop to these levels (36%; Fig. 2A). The anterior third of the eye “catches up” with the posterior two thirds of the eye at the very end of the reversal phase, with between 2–5% of ommatidia in the X configuration at 76 hr apf (4 hr earlier, at 72 hr, there is still a distinct difference between the posterior two thirds and anterior one third). The transition from A/P to Po/Eq contacts seems to reach completion more quickly in the posterior than in the anterior third of the eye due to both a 4-hr delay in initiation of the switch in the anterior coupled with an overall slower transition in the anterior third (approximately 20 hr) relative to the posterior third (16 hr).

Little is known about the molecules that regulate the changes in cone cell shapes and contacts described here. Given that N-cadherin and E-cadherin regulate both of these processes (Hayashi and Carthew 2005), it is likely that additional cell adhesion molecules also play a role in regulating these morphogenetic changes.

Identification and Characterization of New hbs Alleles

A role for hbs in cone cell development was recognized in a new allele of hbs, hbs66 (below), and confirmed and characterized in this and two additional, new alleles, hbs1 and hbsEP (see Experimental Procedures section for details). hbs66 and hbs1 are both X-ray-induced alleles, whereas hbsEP is a P-element-induced allele. Exon 1 of hbs contains both the transcription and translation start sites (Artero et al.,2001; Dworak et al.,2001). The hbs1 allele, a deletion, removes at least part of exon 1 (Fig. 3A), as demonstrated by PCR analysis. RT-PCR analysis indicates that this allele does not produce any transcript and is, therefore, designated a molecular null allele (data not shown). In hbs66, there is a single base pair deletion in exon two that removes T620 (Fig. 3A, B), causing a frame shift that is predicted to both change the downstream seven amino acids and to truncate the protein. Genetically, hbs66 behaves like a hypomorphic allele (discussed below and see Tables 1, 2). The P-element in the hbsEP allele is inserted just upstream of the transcription start site of exon 1 (Fig. 3A) and likely decreases functional transcript, as this allele behaves like a genetic null (see below and Tables 1, 2).

thumbnail image

Figure 3. Genetic and molecular analyses demonstrate hbs is distinct from nearby defined and predicted genes. A: Approximate locations of lesions in hbsEP, hbs66, and hbs1 alleles. Key identifies conserved domains, as described (Artero et al.,2001; Dworak et al.,2001). hbsEP contains a P-element inserted immediately upstream of exon 1 (arrowhead), hbs66 is a single base pair deletion in exon 2 (arrow), and hbs1 is a deletion that removes exon 1, indicated by the bar. B: Alignment of wild type and hbs66 nucleic acid and protein sequences. A single base pair deletion at T620 (arrowhead above wild type sequence) causes a frame shift that is predicted to prematurely truncate the Hbs protein. *, stop codon. Sequence affected by the hbs66 mutation is underlined. C: Genes are illustrated as boxes and color-coded as follows. Black: genes tested by genetic complementation; gray: genes sequenced in hbs mutant alleles with no mutations identified; white: hbs, coding mutations identified in hbs mutant alleles.

Download figure to PowerPoint

Table 1. Pupal Eye Defects in hbs Alleles
GenotypeAbnormal contactsWrong no. cone cellsFused ommatidiaTotal defectsNo. ommatidia scored (no. eyes scored)
hbs6635.01.52.538.91,433 (4)
hbs661238.33.40.742.3436 (6)
hbs143.41.42.247.1491 (6)
hbs11240.23.21.544.9535 (6)
hbsEP51.74.11.156.91,158 (8)
hbsEP1259.44.31.265.41,605 (8)
Table 2. Adult Eye Defects in hbs Alleles
GenotypeTissue polarity defectsMissing/extra R cellsFused ommatidiaTotal defectsNo. ommatidia scored (no. eyes scored)
hbs662.70.70.13.5921 (12)
hbs66123.01.804.81,578 (9)
hbs14.31.50.15.91,231 (9)
hbs1123.51.304.8721 (6)
hbsEP3.93.10.27.21,315 (9)
hbsEP121.62.4041,092 (6)

hbs1, hbs66, and hbsEP form a single complementation group, as they fail to complement each other for all epithelial phenotypes when placed in trans to one another. Flies homozygous for hbs1, hbs66, and hbsEP exhibit essentially the same degree of penetrance as does each allele in trans to Deletion 12. Using cone cell defects (contacts and number), photoreceptor number, ommatidial orientation, and fused ommatidia as a measure of penetrance, hbsEP is the strongest allele, followed by hbs1, and then by the weakest allele, hbs66 (Tables 1 and 2). Given that RT-PCR analysis indicates hbs1 is a molecular null, it is surprising that it is less penetrant than hbsEP, suggesting a reduction in Hbs is more deleterious than an absence of Hbs or, alternatively, that the hbsEP line has tightly linked modifiers that enhance the phenotype. hbs459 and hbs2593 are also members of this complementation group but background mutations in these two lines precluded analysis of the cone cell phenotype (Supp. Fig. 1, which is available online). An outcrossed allele derived from hbs459 does show the same cone cell contact defects described for hbs1, hbs66, and hbsEP (data not shown). Notably, the cone cell phenotype is evident only when hbs levels are reduced by more than 50%, as no cone cell phenotype is generated in flies carrying two copies of the hbs RNAi transgene driven by the strong eye-specific driver, GMR-Gal4 (data not shown), a genotype that is reported to decrease hbs transcript by 50% (Bao and Cagan,2005).

hbs Is Required for Multiple Developmental Events

hbs participates in numerous events throughout development. In addition to its previously described roles in muscle development (Artero et al.,2001; Dworak et al.,2001), Malpighian tubule development and physiology (Denholm et al.,2003), and cell sorting (Bao and Cagan2005), hbs also plays a role in epithelial patterning and orientation of specialized epithelial structures. hbs contributes to patterning the cellular lattice of the compound eye during early pupal life, giving rise to a rough eye in adults. Two populations of cells are affected in these mutants: the cone cells and the interommatidial cells (IOCs). Note that Bao and Cagan (2005) report abnormal configurations of cone cells. In approximately 60% of ommatidia from hbsEP eyes, the cone cells fail to undergo the transition in contacts between the A/P and Po/Eq pairs (Table 1, Fig. 4B). In one class of hbs mutant ommatidia, the anterior and posterior cone cells retain their immature contact and juvenile morphologies (Fig. 4C). Other mutant ommatidia resemble the morphology of wild-type cone cells that are transitioning from A/P to Po/Eq contacts and have one round and one elongated A/P cone cell; in addition, the cone cell–cone cell contacts remain rounded (Fig. 4D). A third class of hbs mutant ommatidia arrests at the “X” transitional state, in which the cone cell–cone cell contacts appear more linear (Fig. 4E). In the most common class of hbs mutant ommatidia, the polar and equatorial cone cells extend around the rounded cell of the A/P pair (likely preventing the rounded cell from stretching along its D/V axis) to establish a contact on the anterior or posterior side of the cone cell quartet, forming a bull's-eye configuration with one centrally located cell surrounded by three cells (Fig. 4F). Finally, occasional variants of the described mutant cone cell shapes and contacts arise; for example, the Po/Eq cone cells can completely enwrap an anterior or posterior cone cell, separating members of the A/P cone cell pair (Fig. 4G).

thumbnail image

Figure 4. hbs is required for epithelial patterning. A–G: Apical cell junctions visualized by α-Arm. A: At the completion of pattern formation in wild-type eyes, the polar and equatorial cone cells contact one another (arrow). In addition, the polar and equatorial cone cells adopt identical morphologies, as do the anterior and posterior cone cells (anterior is to the right). B: In many mutant hbsEP ommatidia, the polar and equatorial cone cells fail to achieve a central contact between the anterior and posterior cone cells; instead, they often reach around the lateral circumference of either the anterior or posterior cone cell to contact one another. The cone cell around which they wrap retains a less mature, rounded appearance (asterisks; see also F). Occasional extra interommatidial cells are seen (arrows). C–G: Hbs mutant ommatidia display a range of defects in cone cell shapes and contacts. C: Immature, rounded A/P cone cells. D: Asymmetrical A/P cone cells, with rounded contacts. E: A/P cone cell contact, with linear contacts between cone cells. F: As described in B, the Po/Eq cone cells achieve a lateral, rather than central, contact. G: Po/Eq cone cells enwrap an anterior or posterior cone cell. H, I: Tangential sections through dorsal hemisphere of adult eyes and corresponding schematics. Blue trapezoids represent ommatidia with wild-type chirality. H: Seven of the eight photoreceptor rhabdomeres can be seen in this apical section of a wild-type eye. The outer six, which have large rhabdomeres, are arranged in a “trapezoid”; these trapezoids are precisely aligned with one another. I: hbsEP mutant ommatidia are often tilted somewhat off-axis. J: The macrochaete are uniformly spaced in wild-type thoraxes, but are mispatterned in hbs thoraxes (K), leading to bald patches (arrows). L: Wing hairs point distally in wild-type wings, but in hbs66 mutants (M), the hairs are not uniformly aligned along the proximal/distal axis. Occasional ectopic bristles are seen (arrow).

Download figure to PowerPoint

In addition to its requirement in cone cell development, hbs also plays a role in patterning the interommatidial lattice. Bao and Cagan (2005) describe an IOC phenotype in hbs RNAi lines in which the interommatidial precursor cells have a “relaxed” morphology and, therefore, limited contacts with their neighbors, missing and misplaced bristle groups, extra cells at vertices, small apical profiles of secondary pigment cells, and extra IOCs that are stacked lengthwise between neighboring ommatidia. At the end of pattern formation, there is a slight surplus of IOCs in hbs66 mutant eyes. These surplus cells are either aligned lengthwise, side-by-side, as shown for the RNAi phenotype (Bao and Cagan,2005), or they are aligned end-to-end. Ommatidia in hbsEP adult eyes exhibit rare defects in photoreceptor recruitment and ommatidial chirality. In addition, ommatidia are aligned almost precisely perpendicular to the dorso-ventral midline of the eye in wild-type eyes (Fig. 4H; ±4°, Fiehler and Wolff,2007), whereas in hbs mutant eyes, a variable number of ommatidia are slightly misaligned relative to the D/V axis (Fig. 4I). The origin of this phenotype has not been explored, but the two most likely possibilities are defects in the rotational movements that orient the ommatidia and/or aberrant “packing” of the interommatidial cells, which has been shown to give rise to orientation defects (Fiehler and Wolff,2007). Chirality defects are rare, with less than 5% of ommatidia exhibiting either reversal on the A/P axis or symmetry within the R3/R4 pair (i.e., two R3 and no R4 cells or two R4 and no R3 cells; Supp. Fig. 1E). The occasional ommatidium is inverted on its D/V axis. Finally, fewer than 3% of ommatidia fail to recruit a full complement of photoreceptors (Table 2).

Thoracic microchaetae are mis-patterned in hbs mutants in that there are bald patches towards the anterior, along the midline, and at the posterior lateral margins (Fig. 4K, compare to wild type in Fig. 4J). The hairs in hbs mutant wings are properly distributed but are slightly off-axis, a phenotype reminiscent of the off-axis ommatidial phenotype described above. (Fig. 4M, compare to wild type in Fig. 4L). Occasional bristles on the thorax and hairs on the abdomen also appear to be misaligned, although it is unclear if these defects are within the natural range observed in wild type. Aspects of cell adhesion that are mediated by Hbs impact multiple developmental events in Drosophila.

Mosaic Analysis Reveals That hbs Is Cell-Autonomously Required in the Cone Cells for the Correct Configuration of Contacts

hbs is expressed in three cell types in the Drosophila eye: the photoreceptors, cone cells, and primary pigment cells (Dworak et al.,2001; Bao and Cagan,2005). A role for any of these three cell types in patterning the cone cells is reasonable, given the dynamics of the eye's development and architecture, as follows. Following recruitment of the cone cells by the photoreceptors, these two sets of cells maintain contact, so the photoreceptors could contribute non-autonomously to cone cell patterning. Alternatively, the cone cells could signal to one another to direct movements within the quartet. Similar to the photoreceptors, the primary pigment cells also establish and maintain direct contact with the cone cells, and could, therefore, provide non-autonomous cues to pattern the cone cells.

To distinguish between these possibilities and to identify the cell type(s) that require hbs for normal patterning, we conducted a mosaic analysis. The genotypes of each cell were scored in genetically mosaic ommatidia that were either phenotypically mutant or phenotypically wild-type. Analysis of 35 phenotypically wild-type (Fig. 5D) and 34 phenotypically mutant ommatidia (Fig. 5E) revealed that hbs acts autonomously within ommatidia to promote wild-type contacts between cone cells: all phenotypically mutant ommatidia contained at least one cone cell that was mutant for hbs (Fig. 5E), suggesting that hbs is required in the cone cells for proper switching of contacts. Although phenotypically wild-type ommatidia lacking hbs in all four cone cells were seen, this was not unexpected given that the cone cell defects are not fully penetrant in hbs null mutant eyes. Finally, no link was found between the genotypes of photoreceptor neurons or primary pigment cells and the phenotype of the associated cone cells.

thumbnail image

Figure 5. Hbs is required in the cone cells to achieve normal contacts. A: hbs mutant clone viewed in an apical plane; unlabeled cells are genetically mutant for hbs whereas pseudocolored (green) cells are genetically wild-type. Genotype of wild-type cells was determined based on underlying, labeled nuclei. Cell junctions are visualized with α-Arm, genetically wild-type cells are marked with nuclear GFP. Cell types are indicated in C. Red arrowhead: doubled interommatidial cells. B: More basal focal plane of clone shown in A; wild-type cone cell nuclei (e.g., quartet of green nuclei at top, asterisk) are evident in this focal plane. C: Schematic illustrating morphology and position of nuclei and photoreceptor rhabdomeres used in scoresheet (D, E) for mosaic analysis. 1–8, photoreceptors; ant. post., anterior and posterior primary pigment cells; Po, Eq, A, P, polar, equatorial, anterior, and posterior cone cells. D, E: Scoresheets indicating genotypes of each cell type for ommatidia scored in mosaic analysis. Filled (black) cells are genetically wild-type for hbs whereas unfilled (white) cells are genetically mutant for hbs. Ommatidia shown in D are phenotypically wild-type for their cone cell contacts whereas those in E exhibit mutant cone cell contacts.

Download figure to PowerPoint

Over-expression of hbs in single cone cells using the Flp-on technique (Ito et al.,1997) demonstrated that hbs over-expression in single cells does not have any effect on cone cell contacts, although the shapes of the hbs over-expressing cone cells are occasionally slightly abnormal (data not shown).

hbs Does Not Affect the Formation or Maintenance of Adherens Junctions

The adherens junction proteins, N-cadherin and E-cadherin, are required to achieve mature cone cell contacts (Hayashi and Carthew, 2005). N-cadherin is specifically localized to the contacts between the cone cells whereas E-cadherin is localized to all cell–cell contacts on the apical surface of the pupal eye. To explore a potential role for hbs in adherens junction formation or maintenance, the localization of N-cadherin and E-cadherin was evaluated in hbs mutant clones. Both proteins are properly localized, even in ommatidia with abnormal contacts, indicating that adherens junctions form normally and are properly maintained between hbs mutant cone cells.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

The switching of contacts between the A/P and Po/Eq cone cells and the accompanying shape changes are reiterated throughout development in diverse systems. For example, strong parallels are seen between the cone cell quartet and quartets of cells undergoing germ band elongation in the Drosophila embryo. As the germ band elongates, two cells initially contact one another along the A/P axis (Bertet et al.,2004), similar to the A/P cone cell contact. As junctions are remodeled, the four cells of each quartet contact each other in a fashion similar to the X configuration of cone cells described here. Ultimately, a new contact is established between the dorsal and ventral cells; this contact subsequently elongates similar to the Po/Eq cone cell contacts in the pupal eye. The similarities in cell shapes and contacts between cone cells and cells undergoing germ band elongation likely reflect the stability of the final quartet configuration.

The notion that certain configurations of cells are more stable than others was described almost a century ago by D'Arcy Thompson in On Growth and Form (1917); similar ideas were expressed by Plateau in 1873 using soap bubbles as a system to explore stability (Plateau,1873). In his classic treatise on the mathematics of biological form, Thompson illustrates the “X” configuration of cells we describe here and comments on the instability of such an arrangement of cells. In contrast, the mature configuration of cone cells recapitulates the “tetracitula” (Rusconi, ca. 1817, as cited in Thompson,1917), noted by Thompson to represent a stable configuration. More recently, Hayashi and Carthew (2004) explored this notion that cell shapes are largely dictated by cell surface mechanics, and that the cell configurations and shapes most commonly seen in nature are a consequence of cells attempting to minimize their surface area. In their report, Hayashi and Carthew (2004) demonstrate that differential localization of cadherins in the cone cells promotes shapes and configurations that also minimize surface area, and that these phenotypes are identical to those achieved by a purely physical system, soap bubbles. While proteins are required to orchestrate the orderly transition of contacts between living cells, it seems that the basic principle of physics that seeks to minimize surface area between cells also predicts the shapes adopted by cells. Only when this system is altered, such as in hbs mutants, can we see aberrations in this perfect, stable symmetry.

Hbs acts as a cell adhesion molecule during Drosophila development and, therefore, likely acts in a similar capacity to modulate cell–cell adhesion between cone cells. The mechanism by which Hbs regulates the transition of cone cell contacts, whether it disrupts the A/P contact or promotes the Po/Eq contact or both, and what additional molecules are involved, are not known. Several models are suggested by studies on Hbs and related molecules in Drosophila embryonic development. For example, embryonic muscle development relies on selective cell adhesion, mediated by Nephrin-like molecules, in two subsets of cells that fuse with one another, the muscle founder cells and the fusion-competent myoblasts. The myoblasts express sns and hbs, while the founder cells express kirre and rst (reviewed in Chen and Olson,2005). These proteins bind in a heterotypic manner to regulate adhesion. Selective cell adhesion could also dictate the reversal of contacts between the cone cells. If this is the case, the process would have to be mediated by Hbs and Kirre, as neither Sns nor Rst is detectable on the apical cone cell membranes (B.K. Grillo-Hill, unpublished observations). Furthermore, mutations in rst do not affect cone cell contacts (Wolff and Ready,1991, R. Cagan, personal communication). This scenario, in which expression of a ligand on just one cell of each pair interacts with the receptor in its partner cell, seems less likely given that hbs is expressed in all four cone cells (Bao and Cagan,2005). While the expression and mutant phenotype of kirre in the eye has not been explored, Hbs and Kirre can bind one another in S2 cells (Dworak et al.,2001). Artero and colleagues (Artero et al.,2001) suggest that Hbs could disrupt binding of Sns to Rst and/or Kirre, but again, this model is unlikely given the apparent absence of Sns and Rst in cone cells. Alternately, it is possible that hbs itself does not directly mediate cell adhesion, but transduces signals to downstream partners through its intracellular tail to effect proper cone cell contact switching.

The loss-of-function hbs eye phenotype described here differs from two previously published loss-of-function reports. One study describes the hbs459 and hbs2593 alleles as producing rough eyes, although a detailed analysis of the underlying phenotypes was not conducted (Artero et al.,2001). Our analysis of these alleles reveals a defect in photoreceptor number, which can consequently lead to a cascade of patterning defects in cone cells and IOCs and, ultimately, a rough eye. Given that the penetrance of the photoreceptor differentiation defects is greatly reduced in outcrossed alleles (Supp. Fig. 1), second site mutations contribute significantly to the rough eye phenotype of hbs459 and hbs2593. Notably, the most penetrant phenotype of the outcrossed hbs459 stock resembles the hbs cone cell phenotype reported here. In the second report, Bao and Cagan (2005) used RNAi to generate a hbs mutant phenotype and found that an approximately 50% reduction of hbs results in supernumerary secondary pigment cells that are generally stacked lengthwise, side-by-side, similar to the distinctive phenotype that has been reported for mutations in rst (Wolff and Ready,1991). We observed IOCs that stack in a similar fashion along the borders of hbs null clones, raising the possibility that stacking may be a consequence of juxtaposing cells with different levels of Hbs side-by side (Fig. 5A, red arrowhead). Our analysis of flies carrying two copies of the hbs RNAi transgene driven by the eye-specific GMR driver did not reveal any cone cell defects, suggesting that Hbs levels likely must be reduced by more than 50% to affect cone cell contacts (data not shown). Dworak et al. (2001) note that overexpression of hbs also causes a rough eye. This phenotype is due, at least in part, to a missing photoreceptor phenotype (Dworak et al.,2001). Whether there is also an independent cone cell contact phenotype has not been investigated but it would not be unexpected since increased cell adhesion interferes with the ability of cells to interact with one another.

While hbs does not seem to be essential for establishing cell junctions, it clearly modulates cell adhesion in multiple contexts, from promoting adhesion between fusion competent myoblasts and muscle founder cells to remodeling contacts among the cone cell quartet. The ultimate goal of, or necessity for, cone cell contact switching is unclear. Lens material is produced in hbs mutant eyes, indicating mutant cone cells can still perform their function and, therefore, achieve their appropriate fates. Cone cells may switch contacts as a result of other morphogenetic events occurring during pupal eye development. For example, cone cell contacts are remodeled during roughly the same time frame that the primary pigment cells are enwrapping the cone cells at the apical surface; it is possible hbs promotes or maintains contacts between the anterior and posterior cone cells and their respective primary pigment cells. As this contact extends along the D/V axis, the area of contact between each cone cell and its respective primary pigment cell continues to increase, thereby diminishing in size and/or strength the central contact between the A/P cone cells. These changes favor the formation of a Po/Eq contact to maintain the stable tetracitula conformation. This is consistent with the mutant phenotype we observe in hbs, in which those mutant cone cells that remain round and often fail to elongate along the D/V axis also fail to contact a primary pigment cell. Switching of cone cell contacts may render the contacts more stable and resistant to physical stresses imposed on the eye during pupal development. Similarly, remodeling the contacts may contribute to the overall morphogenesis of the retina, much in the same way as the rearrangements between quartets of cells in the elongating germ band cause an increase in the length of that structure.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Genetics

Fly culture and crosses were carried out at 25°C according to standard procedures unless otherwise noted. Isogenized w1118 flies were used as wild type. All stocks noted with a number were obtained from the Bloomington Stock Center. Fly strains used: Df(2R)Jp1 (no. 3518), Df(2R) knSA4 (no. 6380), Df(2R)XTE-18 (no. 2468), b, pr, c, px, sp (no. 4195), P{w[+mW.hs]=arm-GFP.P}83 (no. 8555), EGUF 42D (no. 5251); hbs459, hbs2593, hbs361 (gift of M. Baylies); UAS-hbs, Deletion11 (Δ11) and Deletion12 (Δ12) (gift of H. Sink); P[act5Cy+>GAL4] P{UAS-GFP.S65T]/CyO, y+ (gift of R. Cagan); eyFlp; FRT42D, neoR, ubi-GFP (gift of H. Chang); hbs66, hbs1, hbsEP (described here). The Flp/FRT system was used to generate homozygous mutant clones of hbsEP (Golic and Lindquist,1989; Xu and Rubin,1993). hbs-RNAi was performed as described (Bao and Cagan,2005). hbs was overexpressed in single cell clones using the Flp-on technique (Ito et al.,1997).

Molecular and Genetic Analysis of hbs Alleles

inappropriate contacts66 (icon66), an X-ray allele, was isolated in a FLP/FRT screen designed to identify mutations that cause a rough eye (for screen details, see Wolff and Rubin,1998). icon1, also an X-ray-induced allele (4,000 rads), was generated in an isogenized w1118 background in a non-complementation screen for alleles that failed to complement the rough eye phenotype of icon66 (10,500 haploid genomes screened). A third allele, iconEP, was identified in the GMREP collection of P-element insertions (Wolff et al.,2007; collection was a gift of B. Hay). These three alleles are allelic to hibris, as described below, and are henceforth referred to as hbs66, hbs1, and hbsEP.

hbs is located at cytological position 51D (Flybase). The rough eye phenotypes of hbs66 and hbs1 were genetically mapped to between 2–83 (54C) and 2–75 (52B), respectively, using the b, pr, c, px, sp mapping line. hbs66 was cytologically mapped and found to contain an inversion with breakpoints at 50B and 52B (T. Laverty). Systematic mapping of hbs66 and hbs1 using deficiencies Df(2R)Jp1 (51C3-52F5-9), Df(2R)knSA4 (51C3-51D6), Df(2R)XTE-18 (51E1-52D1), and Deletions 11 and 12 define the interval for these three alleles as a 132-kb region (51D6–51E1), an interval that includes, from proximal to distal, atypical protein kinase C (aPKC), CG12424, parcas, CG7639, CG10265, and hbs. In complementation tests, hbs66 and hbs1 failed to complement the rough eye phenotype of hbs459 and hbs2593. hbs66 and hbs1 were wild type in trans to the following genes: CG12424 (H. Broihier, unpublished), aPKC and parcas (pcsgs) (Fig. 3A). All predicted exons in CG10265 and CG7639 were sequenced. The hbs coding region, 5′ and 3′ UTRs, and introns two through 11 were sequenced using standard PCR-based sequencing methods; genomic DNA for sequencing was obtained from homozygous mutant adult flies.

Coding mutations were identified in hbs1 (a deletion that removes at least part of exon 1; Fig. 3B) and hbs66 (a single base pair deletion in exon two that removes T620 and causes a frame shift). However, the caveat remains that mutations may also exist in the first intron of these alleles. RT-PCR (HSRT-20, Sigma) performed on total RNA isolated from 30 hbs1 larvae (Trizol, Invitrogen) demonstrated hbs1 does not produce any transcript (data not shown); this allele is, therefore, designated a molecular null allele. hbs66 does produce a transcript.

The insertion site of the hbsEPP-element was mapped using plasmid rescue (J. Fetting, unpublished data), as described (Hay et al.,1997). The P-element is inserted immediately upstream of the transcription start site of exon 1 (Fig. 3B), and likely decreases levels of functional transcript, as this allele behaves as a genetic null (Tables 1 and 2).

Histology and Immunohistochemistry

Pupal eyes were dissected and fixed according to standard protocols (Wolff,2000). Tissue was incubated in primary antibody overnight at 4°C at a concentration of 1:10 for mouse anti-Armadillo (Arm), rat anti-N-cadherin (N-cad), and mouse anti-E-cadherin (E-cad) antibodies (Developmental Studies Hybridoma Bank). Secondary antibodies conjugated to Alexafluor fluorescent dyes (Molecular Probes) were used at a concentration of 1:300. Arm-GFP pupal eyes were visualized by fixing tissue for 20 min in 4% paraformaldehyde followed by three 10-min rinses in PBS. Fluorescent images were collected using a Leica TCS SP2 confocal microscope. Adult eyes were prepared for sectioning using standard methods, as described (Wolff,2000).

Analysis of Cuticular Structures

Wing hair orientation was analyzed in adult wings, dissected and mounted in 1:1 lactic acid:70% ethanol. Ten wings were analyzed for each of the following hbs alleles: hbs1, hbs66, and hbsEP. Thoracic bristles were examined in anesthetized flies embedded in 2% agarose, cooled to 60°C, and covered with water.

Cone Cell Development and Timeline

White prepupae were collected and staged for noted lengths of time at 18°C. Contacts between cone cells were visualized at the level of the adherens junctions using transgenic flies expressing GFP-tagged Armadillo (Arm-GFP) under the control of the endogenous promoter (Bloomington stock no. 8555). These flies do not exhibit defects in pattern formation (B.K. Grillo-Hill, unpublished data). To score and quantitate the phenotypes, eyes were divided into thirds (anterior, central, and posterior) along the anterior-posterior axis, with approximately 10 columns in each third. Four randomly selected groups of seven ommatidia (a central ommatidium and its six nearest neighbors) were scored for each third. N = 7–9 eyes and n = 544–741 ommatidia (650 ommatidia, average) per time point.

Mosaic Analysis

The Flp/FRT system was used to generate hbs homozygous mutant clones (Golic and Lindquist,1989; Xu and Rubin,1993). Cone cell contacts were scored in mosaic ommatidia. GFP was used to mark genetically wild-type cells (eyFlp; FRT42D, neoR, ubi-GFP), and genotypes of photoreceptors, cone cells, and primary pigment cells determined. Cell types were identified by the unique nuclear shape and position each class of cells occupies within the ommatidial unit.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

We are grateful to S. Abmayr, M. Baylies, H. Broihier, R. Cagan, M. Erdelyi, K.F. Fischbach, B. Hay, H. Sink, and the Bloomington Stock Center for providing invaluable fly stocks, and K.F. Fischbach and S. Abmayr for antibodies. We are particularly grateful to T. Laverty for his cytological analysis of hbs66. We are indebted to A.S. Rawls and J.B. Guinto for performing the screen that identified the hbsEP line, and to J.L. Fetting for molecular mapping of this allele. We thank K.P. Johnson for assistance with dissections, H. Chang and members of the Wolff and Cagan labs for valuable discussions, and G. Miura and J. Rusconi for insightful comments on the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
DVDY_21981_sm_SupFigS1.tif1711KSupp. Fig. 1. Extant hbs alleles have second site modifiers. Tangential sections of adult eyes and corresponding schematics. A: Wild-type eye. Blue circles are the rhabdomeres, or photosensitive membranes of the photoreceptor cells; blue trapezoids in schematics connect the rhabdomeres to emphasize ommatidial orientation. Blue trapezoids identify ommatidia with wild-type chirality. B–D: Photoreceptor number defects (black asterisks) in multiple hbs alleles appear to be due to a background mutation. B: hbs459/hbs2593 ommatidia exhibit the expected orientation defects (0.8%) in addition to unexpected missing or extra photoreceptors (16.8%). Similar phenotypes are seen in (C) hbs459/Δ12 (2.0% tissue polarity defects and 16.6% missing/extra R cells) and (D) hbs2593/Δ12 (2.0% tissue polarity defects and 16.6% missing/extra R cells). A rare inversion on both anterior/posterior and dorsal/ventral axes is designated by a black trapezoid in C. E: Ommatidia in an outcrossed line derived from hbs459 (hbs459*) have a full complement of photoreceptors (3.3% missing/extra R cells). Ommatidia are frequently oriented somewhat off-axis, and rare tissue polarity-type defects are evident (0.8%; yellow forms represent R4/R4-type symmetrical ommatidia), phenotypes that are seen in the hbs alleles described here.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.