How do initially equivalent cells in an epithelial sheet assume unique identities and give rise to different structures? What are the roles of cell migration, cell adhesion, and cell signaling in this process? How are these processes regulated at the molecular level? The somatic follicle cells in the Drosophila ovary are a well-characterized polarized epithelium in which these questions can be successfully addressed (Montell, 2003; Berg, 2005; Horne-Badovinac and Bilder, 2005). The viability, morphogenesis, and polarity of the egg chamber rely on the integrity of the polarized epithelium and extensive signaling among follicle cells and between follicle and germ line cells.
Oogenesis has been divided into 14 morphological stages (King, 1970), and the process has been comprehensively reviewed (Spradling, 1993). The follicle cells that encapsulate each germ line 16-cell cyst are derived from two somatic stem cells located near the apical tip of each ovariole (Margolis and Spradling, 1995). These cells proliferate until stage 6, when approximately 650 cells make up the follicle cell epithelium (Margolis and Spradling, 1995). Endoreplication occurs during stages 7–9, making the cells polyploid (Spradling, 1993). During stage 9, two important cell migrations begin. All but approximately 50 of the main body follicle cells begin to migrate posteriorly, becoming a cuboidal epithelium that covers the oocyte. The remaining 50 cells assume a squamous morphology and spread to cover the nurse cells. At the same time, the anterior polar cells signal to recruit adjacent terminal cells to form the border cell cluster that begins to migrate through the nurse cells (Montell, 2001, 2003). During stage 10B, the centripetally migrating cells, a population of the cuboidal epithelium at the anterior end of the egg, migrate between the nurse cell cluster and the oocyte to cover the anterior side of egg. These cells will secrete the collar, operculum, and micropyle. The border cells reach the oocyte during stage 10, migrate dorsally to a position near the oocyte nucleus, and participate in eggshell synthesis by forming the pore in the micropyle through which the sperm will enter the egg (Montell et al., 1992). Cells on the dorsal anterior side of the egg undergo dramatic morphological movements resulting in the secretion of a pair of dorsal respiratory appendages (Berg, 2005). During stage 14, the follicle cells undergo apoptosis and are lost from the egg as it moves down the oviduct (Nezis et al., 2002).
The elaborate signaling mechanisms that govern follicle cell migration, particularly of the border cells and those secreting dorsal appendages, have been studied in some detail (Dorman et al., 2004; Starz-Gaiano and Montell, 2004; Berg, 2005; Ward and Berg, 2005). Mutations causing defects in the integrity or epithelial polarity of the follicle cells frequently lead to over-proliferation, as well as disorganized, multilayered growth and subsequent degeneration of the egg chamber during mid-oogenesis (reviewed by Müller, 2000; Tepass et al., 2001; Bilder, 2004; Horne-Badovinac and Bilder, 2005).
Here we demonstrate that the Toll-like receptor, 18-wheeler, is expressed in the ovarian follicle cell epithelium in cells undergoing posterior and centripetal migration. Toll-like receptors are best known for their roles in innate immunity in Drosophila and many other organisms including mammals (Anderson, 2000). The Drosophila genome encodes nine family members, most of which are dynamically expressed during embryonic development and in the fat body (Tauszig et al., 2000; Kambris et al., 2002). Toll is the best characterized member of the family, both for its role in dorsal ventral patterning and in immunity (Belvin and Anderson, 1996; Brennan and Anderson, 2004). Functions for the remaining family members are less clear. The extracellular domains of Toll-like receptors contain multiple leucine-rich repeats, which are protein interaction motifs found in a wide variety of proteins (Kobe and Kajava, 2001). A role for Toll and Toll-like receptors in adhesion has been proposed, although evidence for such a role in vivo has been scant (Gerttula et al., 1988; Keith and Gay, 1990; Eldon et al., 1994; Wang et al., 2005). Here we show that mutations in 18w lead to altered ovarian follicle cell migration and to changes in egg morphology that can affect viability.
18-wheeler Is Expressed in the Ovarian Follicular Epithelium
The Toll-like receptor, 18w, is expressed in ovarian follicle cells and is detectable in egg chambers from stage 8 until the follicle cells are lost at the end of oogenesis (Fig. 1). In early oogenesis, three types of follicle cell types are readily distinguished both by morphology and by the expression of numerous enhancer detectors. These are the stalk cells that separate egg chambers, the terminal cells at the anterior and posterior of each egg chamber, and the main body follicle cells (see, for example, Manseau et al., 1997; Hrdlicka et al., 2002). By mid-oogenesis, five discrete cell types can be identified within the follicle cell epithelium: the anterior border cells, the squamous epithelial stretch cells, the centripetally migrating cells, the columnar epithelial cells, and the posterior terminal cells. These are specified by signals from the germ line and by signals from within the plane of the epithelium (Ray and Schupbach, 1996; Schweitzer and Shilo, 1997; Anderson, 1998; Van Buskirk and Schupbach, 1999; van Eeden and St Johnston, 1999; Gupta and Schüpbach, 2001; Montell, 2001; Roth, 2003).
During stage 8 18w transcripts accumulate in the main body follicle cells, but not in the terminal cells (Fig. 1A), a pattern that persists through stage 10A (Fig. 1B,C). Expression is also detected in the stalk cells (Fig. 2A,B). Later, as the main body follicle cells migrate posteriorly over the oocyte, they continue to express 18w, but expression is not detected in the squamous follicle cells left covering the nurse cells (Fig. 1B,C). By stage 10B, expression is restricted to a large patch of cells on the dorsal anterior side of the oocyte (Fig. 1D) and to the centripetally migrating cells (Fig. 1D,E). During stage 11, the group of cells expressing 18w becomes further refined, remaining highly expressed in the centripetal cells and in anterior cells along the dorsal midline (Fig. 1F–H). Expression remains high in cells secreting the collar (Fig. 1I) and in some of the cells secreting the dorsal appendages (Fig. 1J). Given the expression pattern at stage 11, we suspect that 18w is expressed by floor-forming cells of the dorsal appendages (Dorman et al., 2004). Thus, 18w is initially expressed in a broad domain of cells that is gradually refined to those cells responsible for secreting anterior structures of the eggshell. These domains also migrate as epithelial sheets.
The 18-wheeler Enhancer Detector, l(2)00053 Reports Ovarian Expression
The enhancer detector strain l(2)00053 faithfully reports 18w expression in both embryos and larval structures (Eldon et al., 1994; Williams et al., 1997). Ovaries from this strain also express β-galactosidase in the 18w pattern (Fig. 2). The gentler conditions of antibody staining allow ovarioles to remain intact, revealing that 18w is expressed in stalk cells (Fig. 2A). The broad distribution of 18w expression seen by in situ hybridization in early egg chambers is not detected with antibodies against β-galactosidase. By stage 9, β-galactosidase can be detected in a large domain of posteriorly migrating main body follicle cells (Fig. 2C,D). The one-sided expression seen in Figure 2C may represent a different perspective (lateral) than that seen in Figure 1C (dorsal). By stage 10, expression is restricted to anterior follicle cells, especially those on the dorsal side of the egg (Fig. 2E–J). Both in situ hybridization and β-galactosidase detection demonstrate that 18w is expressed in the anterior dorsal midline follicle cells (Fig. 2G, compare Fig. 1H). Anterior follicle cells secreting the collar and micropyle express β-galactosidase (Fig. 2H,I). Finally, some of the cells secreting the dorsal appendages show β-galactosidase expression (Fig. 2J, compare 1J).
Follicle Cell Migration Is Aberrant in 18-wheeler Mutant Clones
We wished to determine whether loss-of-function alleles of 18w produced defects in the follicle cell epithelium. Because 18w alleles show late embryonic/early larval lethality (Eldon et al., 1994), we used the FLP/FRT system initially developed by Golic and Lindquist (1989) to generate clones. Follicle cell–specific expression of FLP (Duffy et al., 1998) together with expression of nuclear GFP allowed tissue-specific clones to be produced and detected. Ovaries dissected from wild type and clone-bearing females were treated with antibodies to α-spectrin, an important cytoskeletal protein that localizes to the plasma membrane of follicle cells, both apically as α-βH heterodimers and laterally as α–β heterodimers (Lee et al., 1997). The importance of α-spectrin in both the germ line and follicle cells in ovarian development is well characterized (de Cuevas et al., 1996; Lee et al., 1997). α-spectrin reveals the shapes of cells in the follicle cell epithelium and allows us to assess their migration. Figure 3 demonstrates two striking delays in follicle cell migration. Posterior migration of main body follicle cells is normally complete by early stage 10A (Fig. 3A). Prior to completion of border cell migration (Fig. 3A, arrow), main body follicle cells are in position to begin migrating (Fig. 3A, arrowheads). However, when follicle cells are mutant for 18w, posterior migration of main body follicle cells is delayed (Fig. 3C). The arrow indicates that border cell migration is complete, but the follicle cells are still migrating posteriorly, a process normally completed by the end of stage 9. During stage 10B, border cells migrate to the position of the oocyte nucleus (Fig. 3B, arrow) and centripetal migration is in progress (Fig. 3B, arrowheads). However, even a small clone in the dorsal anterior follicle cells disrupts and delays normal migration (Fig. 3D). In this late stage-10B egg chamber, the normal ventral follicle cells are migrating, while the mutant dorsal cells have neither undergone the normal flattening nor begun to migrate centripetally (Fig. 3D, compare dorsal and ventral arrowheads). Border cell migration appears normal, even if they are part of a mutant clone, consistent with the observation that they do not express 18w.
Eggs Laid by Females Carrying 18-wheeler Mutant Clones Have Aberrant Morphology
To determine what effect delayed follicle cell migration has on eggs laid by females carrying follicle cell clones, we collected, photographed, and measured eggs from females expressing wild type, loss-of-function and putative gain-of-function forms of 18w (Fig. 4 and Table 1). Among these three genotypes, four aspects of egg morphology were altered: length, width, relative position of the dorsal appendages, and dorsal appendage morphology. In contrast to eggs laid by wild type females (Fig. 4, top row), eggs laid by females expressing a gain-of-function allele were shorter (Fig. 4, middle row), while eggs laid by females that fail to express functional 18w were both shorter and rounder (Fig. 4, bottom row). To quantify the differences among genotypes, egg length, egg width, and operculum length were measured. Egg volume and the position of the dorsal appendages relative to egg length were then calculated.
Table 1. Quantification of Differences in Egg Morphology Based on Tukey Post-Hoc Comparisons
DA % EL, position of dorsal appendages as a percent of egg length (0% = posterior, 100% = anterior).
Oregon R (n = 57)
Gain of function (n = 55)
Loss of function (n = 150)
OR vs. gof
P < 0.001
P = 0.997
P < 0.001
P < 0.001
P < 0.001
OR vs. lof
P < 0.001
P < 0.001
P = 0.480
P = 0.999
P = 0.001
gof vs. lof
P = 0.012
P < 0.001
P < 0.001
P < 0.001
P < 0.001
Significant differences in egg length exist among all three genotypes, with the shortest eggs laid by the females carrying loss-of-function clones and the longest by wild type females (Table 1). Egg width was significantly different between eggs laid by wild type and females carrying loss-of-function clones, but not between wild type females and those expressing a gain-of-function form of 18w (Table 1). The volumes were not significantly different between eggs laid by wild type and females carrying loss-of-function clones, suggesting that the shorter eggs laid by females carrying loss-of-function clones is not due to failure of nurse cell dumping. However, the eggs laid by females expressing the gain-of-function form of 18w had significantly smaller volumes. There was no increase in width to compensate for the decrease in length in these eggs as was observed in eggs laid by females carrying loss-of-function clones.
The centripetally migrating follicle cells express 18w and are responsible for secreting the operculum. Therefore, operculum defects might be expected when 18w expression is altered. While opercula appeared larger in eggs laid by females with loss-of-function mutant clones, careful measurement revealed that they were the same length as those seen in eggs laid by wild type females (Table 1). However, the consequence of normal size opercula on shorter eggs is the significant posterior shift of the position of the dorsal appendages as a percent of egg length (Table 1). One caveat to these results is that boundaries of loss-of-function clones are variable (Fig. 3), so not all eggs have clones in all centripetally migrating cells. In contrast, eggs laid by females expressing a gain-of-function form of 18w had significantly shorter opercula (Table 1). When combined with the shorter length of eggs laid by these females, this produces a significant anterior shift in the position of the dorsal appendages as a percent of egg length (Table 1). It is not yet clear whether the smaller opercula are due to fewer or smaller cells migrating centripetally. Dorsal appendage morphogenesis appears disrupted as well. When compared to wild type eggs (Fig. 4, top row), eggs laid by females carrying loss-of-function clones frequently show reduced or absent paddles (Fig. 4, bottom row).
Not included in the data presented here are the eggs that could not be accurately measured because they collapsed. As many as a quarter of the eggs were deflated in some collections, and all collections of eggs from females carrying loss-of-function clones contained at least a few such eggs. Deflated eggs were not observed in wild type collections. The rate of successful hatching even among the normally hydrated eggs laid by females carrying loss-of-function clones was reduced by 20–40% relative to parallel wild type collections. We have not yet determined whether the lack of hatching success represents aborted development or failure of fertilization. Further analysis will be necessary to determine whether sperm entry occurs in these defective eggs. Hatching data were not obtained for gain-of-function carrying flies, since the high temperature employed for stable GAL4 expression poses risks for male fertility.
All the cells expressing 18w are post-mitotic, and thus 18w expression may reflect an early stage in fate determination. The intriguing pattern of 18w expression in posteriorly migrating cuboidal epithelial cells, centripetally migrating cells, and likely the floor cells secreting the dorsal appendages is consistent with a role in cells that migrate as sheets. We note that although we detect expression of the 18w enhancer detector in the stalk cells of young egg chambers, it does not appear to be expressed in the polar or terminal cells, nor in the border cell clusters that they become. It is not clear whether stalk cell expression is correlated with a role for 18w in cell migration. We have observed some aberrantly long stalks separating egg chambers in females bearing heat shock–induced loss-of-function clones. However, we need marked clones to carry out a statistical analysis of this effect. The T155 driver we used to induce marked mutant clones is not expressed in stalk cells.
Failure of expression of 18w in main body follicle cells in mid-oogenesis results in delayed cell migrations. These delays are apparently affecting the morphology of the eggs that are produced by females carrying loss-of-function clones. The follicle cells are responsible for the shape of the egg in addition to secreting the chorion (Spradling, 1993). Delayed cell migration may prevent the follicle layer from constraining egg diameter during nurse cell dumping and would produce shorter, rounder eggs. The shorter eggs laid by females expressing a gain-of-function form of 18w could be caused by a failure of follicle cells to allow sufficient egg expansion during nurse cell dumping. These eggs, however, do not have the open anterior “cup” phenotype associated with many “dumpless” phenotypes (Spradling, 1993).
A striking percentage of deflated eggs was observed in some collections from females carrying loss-of-function clones. They could not be measured accurately, and were not included in our statistical analysis of egg shape. These eggs frequently took up the purple color of the grape juice egg lay plates and often showed dramatically enlarged opercula. Some eggs in these collections were extremely fragile and prone to rupture during normal handling. All these characteristics are consistent with a failure of the follicle cell epithelium to secrete an intact chorion. The chorion is secreted by the follicle cells during the last day of oogenesis, stages 9–14 (Spradling, 1993), so delayed migration could disrupt this carefully choreographed process.
What does delayed cell migration reveal about the normal function of 18w in the ovary? Cell migration depends upon cells recognizing their position and polarity, and reconfiguring their cytoskeletal components to allow changes in shape and adhesive properties. Cell polarity could involve detecting gradients of signaling molecules and transducing that signal to the cytoskeleton. It is also possible that 18w acts as an adhesion molecule promoting migration. Multiple signaling pathways act in the developing egg chamber (reviewed by Dobens and Raftery, 2000). If the 18W receptor were acting as a competence factor, then its loss would make cells less responsive to signals, either slowing their migration or reducing their directionality. If 18W were acting as an adhesion molecule, then its loss might reduce cells' ability to remodel their cytoskeleton or to change their cell–cell contacts for normal migration.
In addition to understanding what 18w is doing in the ovary, we would also like to know how its expression is regulated. It is possible that early expression in the stalk cells is regulated differently than later expression in the main body cells. Such temporal differences in control of gene expression during oogenesis have been noted for Broad Complex transcripts (Deng and Bownes, 1997). The expression of 18w in the centripetally migrating cells suggests that 18w normally responds to the anterior TGF-β signal, Dpp, which is responsible for setting the operculum size (Twombly et al., 1996; Dobens and Raftery, 1998). The expression domain of 18w along the anterior dorsal midline is likely to be affected by EGF receptor signaling in addition to Dpp. 18w-expressing cells contribute to dorsal appendage synthesis and subtle appendage defects are observed when that contribution is missing. Ultimately, we wish to understand how 18w contributes to the extremely complex signaling that occurs in the dorsal anterior region of the mid- to late-stage egg chamber to execute the elaborate morphogenetic events of eggshell patterning (Dobens and Raftery, 2000; Berg, 2005).
We have presented evidence that a Toll-like receptor, 18-wheeler, is expressed in the follicle cell epithelium and plays a role in the timely migration of main body and centripetally migrating follicle cells. Eggs laid by females carrying 18w mutant follicle cell clones show defects in the eggshell, both structurally and morphologically. Further analysis will be required to examine how 18w interacts with other signaling and adhesion molecules to contribute to normal epithelial migration and to the production of viable eggs.
Germ line and follicle cell clones were generated using the FLP/FRT technique (Golic and Lindquist, 1989). P-element excision alleles of 18w were recombined onto the P[mini w+; FRT]2R-G13 chromosome (Chou and Perrimon, 1996; Syed, 2000) in trans to P[mini w+; FRT]2R-G13 P[mini w+; Ubiquitin-GFPnls]. Clones were either induced in the presence of P[ry+; hs FLP]12 by heat shock for 2 hr in a circulating 37°C water bath during early second and again during mid third larval instars (usually days 2 and 7 after a one day egg lay) or by growth from second instar at 30°C in the presence of T155:GAL4, UAS-FLP (Duffy et al., 1998). Loss-of-function mutants resulted from excision of the P-element l(2)00053 (Karpen and Spradling, 1992). Line Δ21 makes no transcript and carries a 1.5-kb deletion of upstream regulatory sequences. Line Δ72 retains part of the P-element and carries a large deletion downstream of the P-element that removes most, if not all, of the 18w open reading frame. Weak transcript accumulation is detected in homozygous mutant embryos using a probe hybridizing to the 3′ untranslated region, but not using a probe hybridizing to the open reading frame. Δ76 makes no detectable 18-wheeler transcript, and carries a 2.8-kb deletion that removes 2.3 kb of an upstream sequence and 450 bp of an open reading frame (Syed, 2000). All three lines gave similar results in the analyses presented here and are referred to throughout as 18w loss-of-function mutants. Clones of Df(2R)017 also produced similar phenotypes. The gain-of-function transgene, P[mini-w+; UAS-18-wheeler10b], was the generous gift of J.-M. Reichhart. It carries the C951Y change, analogous to the C781Y change in Toll that created the strong gain-of-function allele, Toll10b (Schneider et al., 1991). For the statistical analyses of egg morphology presented here, the T155:GAL4 driver was employed (Hrdlicka et al., 2002). GAL4 lines 198Y, C329b, C355, C204, C289b11, and 109-28, all obtained from the Bloomington Stock Center, were also tested. GAL4 expressing lines were maintained at 19–22°C, but vials were transferred to 28–30°C during third larval instar to insure efficient UAS-18w10b expression.
In Situ Hybridizaton
Ovaries from well-fed 2–7-day-old females were dissected and fixed in 4% formaldehyde following standard protocols (Tautz and Pfeifle, 1989; Wasserman and Freeman, 1998). A 1.1-kb Eco RV fragment of the 18w open reading frame was the template for single-stranded digoxygenin labeled RNA probes (Roche Applied Science, Indianapolis, IN). Hybridizations were carried out using adaptations of standard protocols (Wasserman and Freeman, 1998). Ovaries were mounted in 70% glycerol and photographed on a Nikon E-600 microscope with Digital Spot RT camera (Diagnostic Instruments, Sterling Heights, MI). Images were processed and figures were prepared in Photoshop (Adobe Systems, Inc., San Jose, CA).
Ovaries were fixed as above for in situ hybridization, and incubated with antibodies following standard protocols (Patel, 1994). Mouse monoclonal anti-β-galactosidase (Invitrogen, La Jolla, CA) was used at 1:2,000 dilution, with a Vectastain ABC kit (Vector Laboratories, Burlingame, CA) and diaminobenzadine as the substrate. Ovaries were viewed and images captured as above. Mouse monoclonal anti-α-spectrin (3A9, ascites fluid, generously provided by R. Dubreuil) (Dubreuil et al., 1987) was used at 1:2,500 dilution. Cell supernatant, obtained from the Developmental Studies Hybridoma Bank at the University of Iowa, was used at 1:10 dilution. Alexafluor 555 conjugated secondary antibody (Molecular Probes, Eugene, OR) was used at 1:7,500 dilution for confocal microscopy on ovaries mounted in Aqua Polymount (Polysciences, Warminster, PA). Images of 75 clones from three different mutant lines were captured on a Leica TCS SP2 confocal microscope (courtesy of the Department of Biological Sciences, California State University, Fullerton, CA) or an Olympus IX81 Fluoview 1000 confocal microscope (courtesy of Olympus America, Inc., San Diego, CA). Images were processed and figures were prepared in Photoshop (Adobe Systems, Inc., San Jose, CA).
Eggs from 4–10-day-old well-fed females were collected on grape juice agar plates containing a smear of yeast paste. Collections were made at room temperature (22–24°C) for females with loss-of-function clones, and at 28–30°C for GAL4-expressing strains. Age-matched Oregon R flies were used as controls. Eggs were photographed using an SMZ 1500 Nikon dissecting microscope with a Spot RT digital camera. Measurements were made using Spot Advance software calibrated with a stage micrometer. Volume was calculated using the equation for a prolate spheroid, length × width2 × π/6 (Pitnick et al., 2003). Images were processed and figures were prepared in Photoshop (Adobe Systems, Inc., San Jose, CA).
Descriptive statistics were determined using Excel (Microsoft Corp., Redmond, WA). Differences among genotypes were assessed based on general linear models and Tukey's post-hoc comparisons using Statistica (StatSoft, Tulsa, OK). To analyze length, volume, and dorsal appendage data, data were log transformed to generate normal distributions.
We thank Marie Alpuerto for help with ovary dissections, Stephen Karl for assistance with confocal microscopy (CSUF), Dr. Esteban Fernandez-Juricic for advice and assistance with statistical analyses, and Dr. Judy Brusslan and members of the lab for critical reading of the manuscript. We are grateful to Ron Dubreuil for the 3A9 α-spectrin monoclonal antibody. A subsequent aliquot was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. The Bloomington Stock Center supplied numerous fly lines provided by members of the Drosophila community. Olympus America generously provided confocal microscope time and expertise. We also thank the staff of the Faculty Technology Center at CSULB for assistance with Photoshop. Two anonymous reviewers are gratefully acknowledged for the improvements their critiques prompted. This work was supported by NSF/MCB 97-23899, a Scholarly and Creative Activities Award from California State University, Long Beach to E.D.E., and the generosity of Dr. Toni Stanton. C.D.K. was an undergraduate fellow of the Arnold and Mabel Beckman Foundation Beckman Scholars Program and of the Howard Hughes Medical Institute, 52002663.