INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The magnitude of yolk uptake into the oocyte during vitellogenesis suggests a heavy involvement of the endocytic machinery; indeed, the clathrin-coated vesicle was originally described in the vitellogenic mosquito oocyte (Roth and Porter, 1964). Because the morphological features are so striking, descriptions of vitellogenesis have been made in a broad range of oviparous species, including birds (Perry and Gilbert, 1979; Perry et al., 1984), frogs (Opresko and Wiley, 1987; Wall and Patel, 1987), fish (reviewed by Wallace and Selman, 1990), and insects (Cummings and King, 1970; Mahowald, 1972; Giorgi and Jacob, 1977a,b; Raikhel, 1984; van Antwerpen et al., 1993) (reviewed by Raikhel and Dhadialla, 1992). In several cases, immunocytochemical and ultrastructural studies using labeled yolk protein precursors or fluid phase markers have followed the fate of the proteins as they are sorted to the yolk platelets (Giorgi and Jacob, 1977a; Raikhel, 1984; Busson et al., 1989). The initial steps in the yolk uptake pathway are similar to those described for general receptor-mediated endocytosis (Goldstein et al., 1985; Mukherjee et al., 1997). Vitellogenins (Vgs) are taken up through clathrin-coated pits, and they accumulate initially in vesiculotubular structures (early endosomal structures), which coalesce into primary yolk bodies (analogous to late endosomal structures). In contrast to general endocytic pathways where the internalized ligands are degraded in lysosomes, yolk proteins are stored as yolk granules for later use during embryogenesis. The yolk granules appear to be modified lysosomes with relatively high pH; during embryogenesis, the pH of the yolk granule drops to levels more typical of lysosomes (Fagotto, 1995).

Until recently, the location of the vitellogenin receptor (VgR) throughout this process had not been examined directly. Generally, the fate of the receptor had been inferred by following fluid phase markers and labeled vitellogenins. Tubules labeled with the fluid phase marker but not the yolk proteins were suggested to be receptor recycling tubules by analogy to the morphologically similar structures identified as recycling tubules in other endocytic systems (Geuze et al., 1983). Recent purification of VgRs or the genes encoding VgRs now permits such an analysis. Shen et al. (1993) found the chicken VgR in endocytic structures (coated pits, vesicles, and tubules); however, recycling compartments were not described. Snigirevskaya et al. (1997) also described endocytic compartments and putative recycling compartments that contained the mosquito VgR. In this study, we undertook an analysis of the Drosophila yolk protein receptor distribution during oogenesis. We address the relationship of the expression and intracellular distribution of the receptor to the development of the oocyte and to the onset of vitellogenesis. Moreover, the availability of yl mutants in Drosophila has enabled us to examine the distribution of receptors defective in yolk uptake.

VgRs from birds, insects, fish, and frogs (reviewed by Schneider, 1996; Sappington and Raikhel, 1998), all belong to the low-density lipoprotein receptor superfamily. The conservation of vitellogenin receptors across such diverse phyla suggests a conserved mechanism not only in yolk uptake but also potentially in the regulation of vitellogenin receptors. For example, in many insects, oocyte development arrests just before Vg uptake. Hormonal and environmental cues induce the oocyte to resume development and to begin vitellogenesis. Juvenile hormone (JH), in particular, seems to play a key role in various insects in releasing the oocyte from this block (reviewed by Raikhel and Dhadialla, 1992). Recent work also suggests that ecdysteroids may be involved in the inception of vitellogenesis (Richard et al., 1998); however, the cellular mechanisms underlying the initiation of yolk uptake are unresolved. Thus, in addition to the endocytic profile, we were interested in the regulation of yl during oogenesis. In this paper, we describe the expression patterns of yl RNA and protein in normal flies as well as in mutants defective in oogenesis. Based on this analysis, we address the relationship of the expression and intracellular distribution of the yl RNA and protein to the development of the oocyte and the onset of vitellogenesis. We also identify gene regulatory regions sufficient for germ line expression of yl.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Fly Culture

Drosophila cultures were maintained at 24°C on standard cornmeal molasses agar unless otherwise specified. All yl alleles have been described previously (DiMario et al., 1987). We thank Rod Nagoshi (University of Iowa) for sending the female sterile yl alleles generated by J. Mohler (1977), Beat Suter (McGill University) for the BicD flies, and Tom Wilson (Colorado State University) for the ap⁴ flies. We also thank Julie Feder for assistance in use of the confocal microscope.

RNA In Situ Hybridization

Whole-mount RNA in situ hybridization was carried out as described by Tautz and Pfeifle (1989) with modifications of the methods of Ephrussi et al. (1991). Random-primed digoxygenin-labeled DNA probes were prepared as described by the manufacturer (Boehringer Mannheim, Indianapolis, IN). The yl cDNAs used for the in situ hybridizations have been described previously (Schonbaum et al., 1995).

yl-lacZ Reporters

To construct the CHZY191 (20/400) line, a 375-bp region, from 20 to 395 bp upstream of the strong yl transcription start site (our unpublished results), was generated by PCR amplification and cloned into the pCasper-hs43-lacZ Vector (Thummel and Pirrotta, 1991). The sequence of the PCR-amplified region was confirmed. The CHZY195 (20/1700) line was made by adding a 1.3-kb BamHI fragment (395/1685) to the CHZY191 construct and isolating the clone with the fragment inserted in the correct orientation. The CHZY196 line was prepared by cloning a SphI-BamHI (215/1685) fragment into the casper-hs43-lacZ vector. DNAs were purified (Midi-Preps; Qiagen, Hilden, Germany) and coprecipitated with a helper P transposase plasmid (phs). DNAs were injected into y w¹¹¹⁸ hosts and selected for white⁺ phenotype as described previously (Schonbaum et al., 1995). Transgenic lines were tested for -galactosidase activity with 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (Margolis and Spradling, 1995).

Immunostaining

Whole-mount protein immunostaining was modified from a procedure provided by H. Ruohola-Baker (Ruohola et al., 1991). Ovaries were dissected in PBS and fixed for 15 min to 1 h with 4% paraformaldehyde in PBS. The ovaries were washed four times with PBT (PBS plus 0.1% Tween 20) and once with PBTxT (PBT plus 0.3% Triton X-100, 0.1% Tween 20) and blocked for 1-2 h at room temperature with PBTxT plus 2% BSA (Sigma, St. Louis, MO). The ovaries were incubated with anti-Yl whole serum (1:100-1:300, diluted in PBTxT + 2%BSA) or affinity-purified anti-Yl antibodies (1:100) overnight at 4°C. The rat anti-Yl antibodies have been described previously (Schonbaum et al., 1995). The ovaries were washed extensively with four 1.5-h washes with PBTxT and then incubated overnight at 4°C with preabsorbed secondary antibody (fluorescein-conjugated goat anti-rat immunoglobulin, diluted 1:400-1:500 in PBTxT. The secondary antibodies had been preabsorbed for >2 h against fixed embryos. Finally, the ovaries were washed again four times with PBTxT, rinsed with PBT, and mounted in Aquamount (Polysciences, Warrington, PA) or in 70% glycerol. Samples were viewed on a Zeiss (Thornwood, NY) Laser Scan confocal microscope.

Immunogold Labeling

Ovaries were dissected and fixed in 0.1 M NaPO₄, pH 7.4, containing 4% formaldehyde (electron microscopy [EM[ grade; Electron Microscopy Sciences, Fort Washington, PA) for 10 min at room temperature and then 1 h at 4°C. For thin sections two techniques were used. For cryothin sections, fixed samples were washed in 0.2 M sucrose in 0.1 M NaPO₄ buffer, pH 7.4, and then allowed to equilibrate in 20% polyvinylpyrrolidone in 1.84 M sucrose. Thin sections were collected on tungsten loops with 2.3 M sucrose and mounted on Formvar- and carbon-coated nickel grids. For plastic sections, fixed samples were washed three times for 5 min with 0.1 M NaPO₄, dehydrated through a graded ethanol series at decreasing temperatures, infiltrated with graded alcohol and Lowicryl K4M resin according to the manufacturer's instructions (Electron Microscopy Sciences) at 25°C. Samples were polymerized using UV light for 3-5 d at 25°C. Ultrathin sections were mounted onto Formvar- and carbon-coated nickel grids. For immunostaining all solutions were centrifuged briefly (5000 rpm for 5 min) or were clarified with 0.2-µm filters. Sections were washed in PBS for 15 min and blocked for 3 h at room temperature with 2% BSA in PBS. They were then incubated in a humidified chamber at room temperature for 2 h with affinity-purified rat anti-Yl antibody diluted 1:150 in blocking solution. Sections were washed three times for 10 min each in PBS drops and then incubated for 2 h with 15-nm gold-conjugated goat anti-rat immunoglobulin G (Amersham, Arlington Heights, IL) diluted 1:5 in blocking solution. After washing in PBS as before, the sections were postfixed briefly with 1% glutaraldehyde in PBS and then washed briefly with water. Finally, the plastic sections were stained with uranyl acetate and lead citrate, and the cryothin sections were stained and dried in 2% polyvinylalcohol with 0.03% uranyl acetate, and they were then examined with a JEOL (Peabody, MA) 100CX II transmission electron microscope.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Yolkless RNA Expression

The adult development of the oocyte begins with an asymmetric germ line stem cell division that gives rise to a cystoblast (Figure 1). The cystoblast undergoes four incomplete divisions in region 1 of the germarium to yield a complex of 16 cells interconnected by cytoplasmic bridges (ring canals). Only 1 of these 16 cells differentiates as an oocyte, whereas the remainder develop as nurse cells. As the germ line-derived cystocytes develop within the germarium, the 16-cell cluster is ensheathed by a layer of somatically derived follicle cells. The follicle cells and oocyte signal to each other, influencing the development of the chamber and its coordinate axes (reviewed by Morgan and Mahowald, 1996; van Eiden and Johnston, 1999). The vitellogenic stages (Figure 1B), by definition, initiate when yolk begins to accumulate in the oocyte (Cummings and King, 1969; King, 1970). Finally, toward the end of ovary development, the follicle cells secrete the proteinaceous vitelline membrane and chorion that cover the egg.

View larger version (61K): [in this window] [in a new window]   Figure 1.   Oogenesis in Drosophila melanogaster. (A) Drawing of a germarium (modified from Morgan and Mahowald, 1996). (B) Stages of vitellogenesis from inception (stage 8) to completion (stage 10) (modified from Cummings and King, 1969).

Pole cell transplantations and germ line clonal analysis indicated a germ line-dependent function for the yolkless gene (Waring et al., 1983; Perrimon et al., 1986). Consistent with these studies, in situ hybridization to yl RNA in ovaries demonstrated that yl accumulates only in the germ line-derived nurse cells and oocyte (Figure 2A). In addition, the RNA in situ data showed that the gene is expressed very early during the differentiation of the oocyte. yl RNA is seen as early as region 2A in the germarium, soon after the germ line cells stop dividing and before the nurse cell-oocyte complex has been enveloped by follicle cells (Figure 2A, inset). The RNA is concentrated in a single cell within the 16-cell cyst. This cell is clearly the oocyte in region 3 (stage 1) ovarian cysts, and we assume that the single cell accumulating yl RNA in regions 2A and 2B is the presumptive oocyte.

View larger version (125K): [in this window] [in a new window]   Figure 2.   yl RNA expression in the germ line cells. (A) Whole-mount RNA in situ hybridization shows yl RNA is expressed in wild-type germ line cells and accumulates in the oocyte as early as region 2 in the germarium (g). Inset, Higher magnification of a germarium with yl RNA localized to the presumptive oocyte of region 2 and region 3 cystocytes. (B) yl RNA levels in wild-type ovaries increase in the nurse cells during stages 9 and 10. (C) In egl1/egl1mutant chambers, which have 16 nurse cells but no oocyte, yl RNA is still expressed but it is not localized to any of the cells. (D) Similar results are seen with a hypomorphic BicD allele [BicDR29/Df(2)TW119], although some RNA accumulation in the most posterior cell is seen in very early stage chambers. (E) Staining is not seen in the yl null [Df(1)gl/Df(1)KA9] control. (F) -Galactosidase is detected in the germ line cells of a yl-lacZ transgenic ovary. The yl-lacZ reporter construct contains the first 400 bp upstream of the yl transcription start site.

During later stages of wild-type oocyte development, yl RNA continues to be found in the oocyte (Figure 2B). However, in contrast with the earlier stages of development, yl RNA levels become more pronounced in the nurse cells of stage 9 and 10 chambers. There does not appear to be any specific localization of the RNA within the oocyte at any time.

Transport of nurse cell-derived RNAs into the oocyte has been noted for several genes important in the development of the oocyte (e.g., BicD [Suter et al., 1989], osk [Ephrussi et al., 1991; Kim-Ha et al., 1991]; fs(1)K10 [Cheung et al., 1992], CycB [Dalby and Glover, 1992], and orb [Lantz et al., 1992]). Transport of yl RNA into the oocyte is also suggested by examination of yl RNA distribution in egl and BicD mutants. In both of these mutants, the oocyte does not differentiate normally, and instead of 15 nurse cells and an oocyte, a cluster of 16 nurse cells is formed. yl RNA is detected in the germ line of early stage egl and BicD chambers (Figure 2, C and D); however, the RNA is now distributed almost equally among the 16 cells. Some yl RNA localization is still evident in the hypomorphic BicD^PA66 (Figure 2D) and the BicD^R26 alleles (our unpublished results), even though the posteriorly positioned cell does not fully develop into an oocyte. Similar effects on osk and orb RNA localization have been observed in BicD and egl mutants (Ephrussi et al., 1991; Ran et al., 1994).

yl-lacZ reporter gene constructs confirmed that yl RNA is transcribed in the nurse cells. Furthermore, they identified a minimal enhancer region sufficient for germ line-specific expression of yl. Previously, a genomic DNA fragment that contained the yl gene, including a region 1.7 kb upstream of the start site, was shown to be sufficient for rescue of the yl mutant phenotype (Schonbaum et al., 1995). We cloned portions of this 1.7-kb region into a reporter vector bearing an hsp70 basal promoter element linked to the lacZ gene (Thummel and Pirrotta, 1991). The -galactosidase expressed from these constructs accumulates in the nucleus because of the presence of a nuclear localization signal. Transgenic animals bearing either the 1.7-kb region (our unpublished results) or the first 400 bp upstream of the yl start site (Figure 2F) fused to the reporter showed -galactosidase activity in all nurse cell nuclei. Animals with a 200-bp region upstream of the transcription start site also exhibited germ line-specific expression of the reporter; however, the signal was weaker and somewhat variable (our unpublished results). Transgenic lines with 1.5 kb of upstream sequences but lacking the first 200 bp upstream of the transcription start site did not show any ovarian -galactosidase staining (our unpublished results).

Yolkless Protein Expression

The Yl protein distribution mirrors the RNA pattern in previtellogenic stages of oocyte development (Figure 3A). Yl can be detected by stage 1 (region 3 in the germarium; cf. Figure 1). Like the RNA pattern, the protein is concentrated in the oocyte. Optical sectioning of wild-type chambers by confocal microscopy indicates that Yl protein is diffusely distributed throughout the oocyte up through stage 7 (Figure 3B). Although the majority of the Yl protein is in the oocyte, we can detect some Yl in nurse cells adjacent to the oocyte (Figure 3A, inset; our unpublished results).

View larger version (71K): [in this window] [in a new window]   Figure 3.   Whole-mount in situ immunolocalization of Yl in wild-type oocytes. (A) Whole-mount in situ immunolocalization shows that the Yl protein is expressed in the previtellogenic stages. Expression is uniform in the stage 1-5 oocytes. Inset, Protein levels in the stage 2 chamber are highest in the oocyte, but lower levels are apparent in the nurse cells just adjacent and connected to the oocyte via ring canals. (B) In a stage 7 chamber, before vitellogenesis has begun, the receptor is distributed throughout the oocyte. (C) A stage 9 chamber, where vitellogenin is being accumulated, shows Yl enrichment at the cortex of the oocyte. (D) By the end of stage 10, the protein appears to be mostly cortically localized. (E) Df(1)g/Df(1)KA9 yl null controls. Bar, 10 µm.

Before the vitellogenic stages, endocytic structures, such as coated vesicles, are not found along the oocyte-follicle cell border (Figure 4; Mahowald, 1972). However, soon after the transition to the vitellogenic stages, defined as stage 8 (Cummings and King, 1969), endocytic structures become prominent (Mahowald, 1972). Yl can be seen accumulating at the surface (cortex) of the stage 8-9 oocyte (Figure 3C) with moderate Yl staining in the internal regions of the oocyte. As oogenesis proceeds, the cortical staining intensifies, and optical sectioning by confocal microscopy shows little Yl present within the center of the stage 10 oocytes. By the end of stage 10, the receptor is almost exclusively cortical (Figure 3D). In addition, at stage 10, there appears to be little Yl protein in the nurse cells, even though yl RNA levels in the nurse cells are elevated (Figure 2B).

View larger version (148K): [in this window] [in a new window]   Figure 4.   Wild-type previtellogenic oocyte. (A) Electron micrograph of a stage 6 oocyte, showing the characteristic spherical oocyte nucleus (N), the absence of coated vesicles along the follicle cell (F)-oocyte (O) border (shown at higher magnification in B), and the presence of a dispersed endoplasmic reticulum (arrows), which is absent from nurse cells (NC). Magnification: A, 8100×; B, 42,000×.

Cortical accumulation of the Yl protein can be disrupted by mutations in yl. Several alleles (yl¹⁵, yl¹⁸, yl²⁰, and yl²¹), have been identified in which full-length 210-kDa protein is synthesized (Figure 5), but the protein remains distributed throughout the oocyte (Figure 6, A-C). These mutants fail to accumulate appreciable levels of yolk proteins and they have dramatically reduced numbers of endocytic structures (DiMario et al., 1987). Females bearing weak yl alleles (yl⁹), which have reduced but still significant levels of yolk protein accumulation (DiMario et al., 1987), exhibit cortical localization of the receptor (our unpublished results).

View larger version (49K): [in this window] [in a new window]   Figure 5.   Western blot analysis of yl mutants. Western analysis of ovary proteins using anti-Yl antibodies shows that the 210-kDa Yl protein is absent in yl alleles with a strong phenotype (yl11,yl14, yl28), but it is present in other alleles with strong reductions in yolk uptake (yl15,yl20, yl21) as well as in alleles with weak phenotypes (yl18). Two alleles, yl16 and yl29, expressed truncated proteins of 130 kDa (*) and 175 kDa, respectively. The low-molecular-weight band is a cross-reacting protein. It is not detected using affinity-purified anti-Yl antibodies; however, it serves as a useful loading control.

View larger version (121K): [in this window] [in a new window]   Figure 6.   Yl distribution in yolk uptake mutants. In strong yl mutants, which express full-length Yl protein yet fail to take up yolk, the Yl protein does not become cortically localized during the vitellogenic stages. (A) Early stage 9 yl21 oocyte; (B) stage 10 yl21 oocyte; (C) stage 10 yl15 oocyte. (D) Cortical Yl localization also fails to occur in stilWE42 mutant ovaries. Bar, 20 µm.

Mutations in other genes with effects on vitellogenesis have been described. We tested two of these for effects on Yl expression. Hypomorphic mutants of the stand still (stil) gene arrest oogenesis at a stage that resembles stages 9-10. stil oocytes are smaller than expected, and endocytosis appears to be impaired as monitored by trypan blue uptake (Gutzeit and Arendt, 1994). Immunostaining for Yl revealed that the receptor is present but not cortically localized in stil ovaries (Figure 6D). Vitellogenesis is also blocked in certain apterous (ap) mutants with depressed JH levels. apterous acts nonautonomously to affect ovary development. Oocytes in ap⁴ females arrest at stage 7, and yolk uptake is not seen (Postlethwait, and Weiser, 1973). Administration of juvenile hormone restores oocyte development and yolk uptake (Postlethwait, and Weiser, 1973; Gavin and Williamson, 1976). In arrested 1- to 2-d ap⁴ oocytes, Yl expression is uniformly distributed, as in wild-type previtellogenic oocytes (our unpublished results).

Subcellular Distribution of Yl during Vitellogenesis

We next studied the distribution of Yl at the ultrastructural level to clarify the relocalization of Yl during the transition from pre- to postvitellogenic stages of oogenesis. Using immunogold staining of sections of ovarian chambers embedded in Lowicryl or cryosections, we have been unable to detect the Yl receptor in egg chambers before the inception of yolk uptake. This is surprising, especially because the receptor was readily detected at the light microscopic level in these stages after similar paraformaldehyde fixation. In contrast, our EM immunogold staining methods were able to detect Yl in vitellogenic oocyte stages (Figures 7 and 8A). Colloidal gold particles were detected in both endocytic and tubular structures in the cortex of the oocyte. More internally, Yl is detected around the perimeter of immature yolk spheres and in tubular-like projections adjacent to the yolk granules. The receptor is also found in the flocculent associated body that lies to one side of the mature yolk sphere (Figure 8B).

View larger version (145K): [in this window] [in a new window]   Figure 7.   Ultrastructural localization of Yl at the oocyte cortex. Immunoelectron micrograph of cryothin sections of the cortex of a stage 10 oocyte, stained with 5-nm colloidal gold-labeled secondary antibodies to Yolkless antibody. Antigen is present in coated vesicles (v) and tubules (t), as well as at the surface of nascent yolk platelets (arrow). The vitelline membrane (VM) is at the top. Magnification, 84,000×.

View larger version (112K): [in this window] [in a new window]   Figure 8.   Ultrastructural localization of Yl. (A) Cryothin section showing the presence of Yolkless in tubules radiating from the surface (top) of a stage 10 oocyte and at the surface of nascent yolk spheres (arrow) (10-nm colloidal gold-labeled secondary antibody). (B) Yolkless antigen is found in the fluffy layer of large yolk granules in a stage 10 oocyte. Magnification: A, 42,000×; B, 42,000×.

At the transition to stage 8, when endocytosis of yolk is beginning, multivesicular bodies (MVBs) are heavily labeled for Yl (Figure 9). In some instances, the receptor is interspersed with an electron-dense material (Figure 9A); in other cases, the electron-dense mass has coalesced, and the receptor is segregated to the periphery of the MVB (Figure 9B). The dense mass found in these MVBs reacts positively for yolk proteins (our unpublished results). These multivesicular structures are abundant in early vitellogenic stage chambers but are found more rarely in stage 9-10 oocytes. They always show a strong staining with the Yl antibody. Yl staining of the oocyte cortex is found in both early and late vitellogenic stages.

View larger version (169K): [in this window] [in a new window]   Figure 9.   Immunogold labeling of Yl in multivesicular bodies. In stage 8 to early stage 9 chambers, multivesicular bodies with dense labeling of Yolkless were frequently observed. In some instances (A), the labeling is distributed throughout the MVB; in other MVBs (B), electron-dense yolk is segregated from the receptor-labeled regions. In B, the cortex of the oocyte is at the right, showing active endocytosis. Magnification: A, 53,000×; B, 53,000×.

Cortical staining is abolished in yl mutants that express relatively high levels of a full-length mutant Yl protein. Instead, the defective protein was uniformly distributed and appeared to reside predominantly in the endoplasmic reticulum (Figure 10). Similar results were seen for the yl²⁰ and yl¹⁵ mutants.

View larger version (217K): [in this window] [in a new window]   Figure 10.   Immunogold labeling of Yl in a yl mutant. In the yl21 mutant, Yl labeling is not localized to the cortical regions but instead is found predominantly in the lumen of the endoplasmic reticulum (lines). Magnification, 42,000×.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Yl Expression

Based on the yolkless mutant phenotype (DiMario et al., 1987) and on similarity of the sequence of the yolkless gene to the vertebrate vitellogenin receptor, we proposed previously that yolkless encoded a vitellogenin receptor in Drosophila melanogaster (Schonbaum et al., 1995). This was subsequently confirmed by the cloning of a very closely related vitellogenin receptor from another dipteran, the mosquito Aedes aegypti (Sappington et al., 1996). The similarity of vitellogenin receptors from insects to birds suggests conserved mechanisms for regulating yolk uptake into eggs. We have now analyzed the expression patterns of yl RNA and protein to identify potential regulatory steps during Drosophila vitellogenesis.

RNA in situ hybridization, Yl immunolocalization, and lacZ reporter studies showed that yl RNA and protein are expressed in germ line cells very early during oogenesis, long before the protein is required for vitellogenin uptake. Clearly, expression of the vitellogenin receptor is not the limiting component for yolk uptake. The lacZ reporter studies also identified sequences in the first 400 bases upstream of the transcription start site that are necessary and sufficient to direct expression in the germ line cells. We have not ruled out additional regulatory elements downstream of the transcription start site. It will be interesting to see whether vitellogenin receptor genes from other species are regulated by a conserved set of germ line transcription factors.

The expression studies also showed that yl RNA is transcribed in the nurse cells and then transported into the oocyte, as has been seen with a number of other Drosophila genes that have roles in oogenesis (reviewed by Lasko, 1999). Transport of yolkless RNA into the oocyte likely occurs via a microtubule-based transport system that has been implicated in the movement of other RNAs from the nurse cells into the oocyte (Cooley and Theurkauf, 1994). RNAs produced by the oocyte nucleus do not depend upon the microtubule transport system (Saunders and Cohen, 1999). Recent surveys of Drosophila genes indicate that up to 10% of germ line-expressed RNAs may be transported into the oocyte, but patterns of transport vary between genes (Dubowy and Macdonald, 1998). Some RNAs are transported efficiently and early during oocyte differentiation, whereas others are transported slowly or later and do not accumulate in the oocyte until late previtellogenic stages or postvitellogenic stages. yolkless falls into the class of genes whose RNA is transported efficiently and very early. We see no obvious localization of the yl RNA to a particular region of the oocyte; thus, the yl RNA should possess sequences solely involved in RNA transport. Distinct RNA transport and localization signals have been identified in the 3' untranslated RNA sequences of nanos and osk RNAs (Kim-Ha et al., 1993; Gavis et al., 1996). It will be interesting to compare the regions of yl RNA that are required for transport to those of other oocyte localized RNAs.

Yl Localization

Consistent with its proposed role as a vitellogenin receptor, Yl protein is present at the cortex of vitellogenic stage oocytes. Yl was seen in coated vesicular and tubular (early endosomal) structures. Yl was also seen associated with smaller yolk granules where the receptor was present in tubular projections. These projections likely represent a sorting and recycling tubule. In more mature yolk granules, Yl labeling at the perimeter of the granule may represent the fraction of the receptor that was not recycled. The mature yolk granule appears to be a modified lysosome with reduced hydrolytic activity (Fagotto, 1995). Thus, receptors that were not recycled would end up associated with the yolk granule.

In contrast to the vitellogenic stages, Yl is uniformly distributed through the oocyte during previtellogenic stages. During these stages, there is no evidence of endocytosis at the oocyte-follicle cell border, and oocytes do not internalize vitellogenins. The failure to take up yolk in previtellogenic stages does not result from the absence of the receptor protein. Is yolk uptake caused by relocalization of the receptor? Redistribution of vitellogenin receptors upon the onset of vitellogenin uptake is also seen in chickens. In small previtellogenic chicken oocytes, the VgR is initially detected in vesicular structures within the interior of the oocyte, with little receptor present at the cell surface (Shen et al., 1993). However, during the phase of rapid vitellogenin uptake, the chicken receptor relocalizes mainly to the cortex of the oocyte. When the subcellular distribution of Yolkless was examined to determine whether the receptor was present in a specific compartment during previtellogenic stages, we were unable to detect Yl using immunogold techniques, even though by confocal microscopy, we could see that Yl was distributed throughout the oocyte. Although there appears to be lower protein levels in previtellogenic stages, the whole-mount results still suggest that the protein was abundant enough to be observed by EM immunolabeling, especially by stage 7. In addition, immunogold labeling of mutant Yl receptor in internal regions of vitellogenic oocytes (Figure 10) suggests that the inability to detect Yl in previtellogenic stages reflects a difference between Yl synthesized during previtellogenic stages and that synthesized during vitellogenic stages. We propose that Yl protein is masked before the transition to vitellogenesis, possibly in the relatively abundant endoplasmic reticulum in the oocyte (Figure 4).

The mechanism underlying VgR relocalization in insects and birds is unknown. One example of regulated endocytosis is seen in the response of mammalian adipose and muscle cells to insulin. Insulin stimulates relocalization of the GLUT4 glucose transporter to the cell surface. Before reception of the signal, the GLUT4 receptor is enriched in intracellular vesicles (reviewed by Pessin et al., 1999). The redistribution of GLUT4 appears to be a case of regulated exocytosis, as occurs during synaptic vesicle fusion; it may also involve the selective retention of the transporter within a distinct intracellular compartment. Interestingly, the GLUT4 C-terminal domain appears to be masked in Lowicryl sections before the insulin-stimulated redistribution (Wang et al., 1996). Is a similar mechanism regulating vitellogenin uptake? Hormonal and/or environmental stimuli initiate yolk uptake in many insects (reviewed by Raikhel and Dhadialla, 1992). In particular, JH can stimulate vitellogenin uptake both in vivo and in vitro in many insects, although its mechanism of action in the ovary is unknown (Tedesco et al., 1981; Raikhel and Lea, 1985). Consistent with this role for JH, Yl is not cortically localized in JH deficient apterous⁴ oocytes. It is not known whether yolk uptake in vertebrate systems is hormonally mediated.

Unlike the GLUT4 system, the onset of vitellogenin uptake appears to involve regulation of general endocytosis, not just a specific receptor. Uptake of fluid phase markers such as ferritin, trypan blue, and horseradish peroxidase is not detected in previtellogenic ovaries of dipterans (Mahowald, 1972; Giorgi, 1979; Raikhel and Lea, 1985), suggesting a complete or significant reduction in endocytosis. Because the oocyte and the polar follicle cells signal via cell surface receptors during previtellogenic stages (Gonzalez-Reyes et al., 1997, Newmark et al., 1997; Gonzalez-Reyes and St. Johnston, 1998), there must be some membrane trafficking within in the oocyte. Thus, there may not be a complete block in endocytosis. It is not clear, however, whether these signaling events are dependent on endocytic activity.

Although we were unable to detect Yl by immunogold labeling of previtellogenic oocytes, we were able to observe the protein easily in early vitellogenic stage chambers. In particular, multivesicular bodies were prominently labeled during the early stages. The presence of vitellogenin receptors in MVBs has not been reported in other ultrastructural studies that have examined receptor location during vitellogenin uptake (Shen et al., 1993, Snigirevskaya, 1997). However, MVBs have been noted in ultrastructural studies of insect vitellogenesis (Mahowald, 1972; Giorgi and Jacob, 1977a; Raikhel and Lea, 1986), and during Xenopus vitellogenesis, gold-labeled vitellogenin is incorporated into multivesicular bodies (Wall and Patel, 1987). MVBs are a distinct feature of general endocytosis (Mukherjee et al., 1997), where they are considered a late endosomal compartment.

We do not yet know the series of events by which Yl becomes associated with the MVB. By analogy to mammalian endocytic systems, receptor (Yl) and ligand (yolk protein) would be internalized, forming a vesiculotubular early endosome. The receptor then would be recycled back to the plasma membrane, whereas the yolk would remain as the endosome matured. The MVB could represent a late endosomal compartment. We see the labeled MVBs frequently in early vitellogenic stages (stage 8 to early stage 9) but only rarely in later stages. The higher abundance of Yl-positive MVBs in early vitellogenic stages may reflect a difference in the rate of trafficking in early versus late vitellogenic stages. For example, as vitellogenesis is starting, recycling may not be efficient, leading to higher levels of the receptor in a later MVB compartment. Similarly, a reduction in the rate of maturation of late endosome to mature yolk granule could lead to enrichment of labeled MVBs in early stages. It is possible that the Yl in MVBs also represents a fate for "masked" receptors synthesized during previtellogenic stages. Instead of being sorted to the plasma membrane, Yl synthesized during previtellogenic stages would end up being sorted to late endosomal structures, such as MVBs. From there, the receptors could be recycled to the cell surface, or they could end up along the perimeter of the yolk granules, possibly in the fluffy layer at the surface of large yolk granules (Figure 8B).

To identify other gene products involved in the relocalization of Yl, we examined Yl distribution in mutants with disrupted vitellogenesis. Although Yl distribution was affected in certain stil mutants, this is likely an indirect effect, because stil is a chromatin-associated protein (Sahut-Barnola and Pauli, 1999). However, recent work in Drosophila has begun to identify various endocytic components used in development (e.g., adaptins [Dornan et al., 1997; Gonzalez-Gaitan and Jackle, 1997; Ooi et al., 1997], rabs [Sasamura et al., 1997; Satoh et al., 1997], vps41 homologue [Warner et al., 1998], and hook [Kramer and Phistry, 1999]). The increase in markers that can be used to definitively label endosomal compartments and the increase in the number of endocytic mutants should facilitate the analysis of yolk uptake in flies. This will also likely shed light on vitellogenic mechanisms in other egg-laying animals.