![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Vol. 16, Issue 5, 2301-2312, May 2005
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Center for Basic Neuroscience and Department of Cell Biology, University of Texas Southwestern Medical Center, Dallas, TX 75390-9111
Submitted November 18, 2004;
Revised February 7, 2005;
Accepted February 12, 2005
Monitoring Editor: Suzanne Pfeffer
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
; Odorizzi et al., 1998; Piper and Luzio, 2001), glycoproteins destined for melanosomes (Berson et al., 2001), and mannose-6-phosphate receptors during their round-trip between the Golgi complex and late endosomes (Hirst et al., 1998). The mechanisms that regulate MVB biogenesis and their biological role in different cell types are still poorly understood.
Some molecular components that participate in MVB biogenesis have recently been identified (Gruenberg and Stenmark, 2004). For example, lysobisphosphatidic acid is a lipid enriched in internal vesicles of MVBs and may play a role in their formation (Matsuo et al., 2004). Furthermore, several genes necessary for the biogenesis of MVBs or the sequestration of cargo into their internal vesicles have been identified in yeast (Katzmann et al., 2001) as a subset of genes necessary for vacuolar protein sorting (vps genes). This subset, many members of which belong to the E class of vps genes, functions in the prevacuolar endosomal compartment (Rieder et al., 1996). Proteins encoded by these genes associate into three distinct endosomal sorting complexes required for transport of cargo to the vacuole, called ESCRT-I, -II, and -III (Katzmann et al., 2001; Babst et al., 2002a,b; Bishop et al., 2002; Katzmann et al., 2003; Bowers et al., 2004). Three proteins, Vps23p/Tsg101, Vps28p, and Vps37p, comprise the ESCRT-I complex in yeast and mammalian cells (Babst et al., 2000; Bishop and Woodman, 2001; Bache et al., 2004; Stuchell et al., 2004).
In mammalian cells, the ESCRT-I complex functions in the down-regulation of cell surface receptors. Partial depletion of Tsg101 or interference by anti-hVPS28 antibody injections causes an inhibition of epidermal growth factor (EGF) degradation, enhanced recycling of EGF receptors, and prolonged activation of the downstream kinases extracellular signal-regulated kinase (ERK)1 and ERK2 (Babst et al., 2000; Bishop et al., 2002). Tsg101 is recruited to endosomal membranes via its interaction with pVps27/Hrs (hepatocyte growth factor-regulated tyrosine kinase substrate). Hrs in turn binds to mono-ubiquitinated receptors and participates in their sorting into MVBs (Bache et al., 2003; Katzmann et al., 2003; Lu et al., 2003).
The mechanism that drives invagination of internal vesicles of MVBs has gained additional interest because its elements are hijacked by enveloped viruses such as HIV-1 (Marsh and Thali, 2003). HIV-Gag mimics the function of Hrs in MVB formation by binding to Tsg101 and thereby recruiting the ESCRT complexes to the site of virus budding; this recruitment is required for HIV budding (Garrus et al., 2001; Goila-Gaur et al., 2003; Martin-Serrano et al., 2003; Pornillos et al., 2003; von Schwedler et al., 2003; Stuchell et al., 2004).
Surprisingly, the most drastic phenotype exhibited by Tsg101 mutant cells from knockout mice is an early cell cycle arrest (Ruland et al., 2001; Wagner et al., 2003). This early phenotype has not yet been reconciled with ESCRT-I function in MVB formation or virus budding, suggesting that this complex may mediate additional functions that remain unexplored.
As in mammalian cells, MVBs have been implicated in the regulation of several signaling pathways in Drosophila (Krämer, 2002; Seto et al., 2002). For example, loss-of-function mutations in Drosophila hrs result in reduced EGF receptor degradation and increased mitogen-activated protein (MAP) kinase signaling (Lloyd et al., 2002). To begin to explore the function of ESCRT proteins during development, we describe here a mutation in the Drosophila homolog of vps28. Our data indicate a requirement for dVps28 during spermatogenesis, the development of early embryos, and the compound eye. These defects were observed despite the continued presence of MVBs in mutant cells and revealed a surprising connection between the functions of Vps28 and the actin cytoskeleton.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
). For expression in S2 cells, PCR-generated cDNAs encoding full-length dVps28 or dTsg101 were tagged with an N-terminal Myc or hemagglutinin (HA) epitope and expressed under control of the metallothionein promoter (Bunch et al., 1988). For the generation of transgenic flies, cDNA GH04443 was cloned into the pUASt transformation vector (Brand and Perrimon, 1993).
Fly Genetics
The hook11 null allele has been described previously (Krämer and Phistry, 1999). Other fly stocks were obtained from the Bloomington Stock Center (Bloomington, IN). The line with the P-element insertion l(2)k16503 carried an additional unidentified lethal mutation that was removed by recombination. The mutation l(2)k16503 was reverted to viability by introducing P-element transposase (Robertson et al., 1988). Small clones homozygous for dVps28l(2)k16503 were generated from heterozygous flies by using the Flp recognition target (FRT)/FLP technique (Xu and Rubin, 1993). To asses the effect of dVps28l(2)k16503 on eye color, clones were generated in a white+ background in FRT42D GMR-Hid l(2)CL-R1/FRT42D dVps28l(2)k16503 flies, thus eliminating cells in the eye that were heterozygous or homozygous wild type with respect to dVps28 (Stowers and Schwarz, 1999). To induce germline clones, FRT42A dVps28l(2)k16503/CyO females were crossed to Hs-FLP; FRT42A ovoD males (Chou and Perrimon, 1996), and the progeny were exposed to a 1-h 37°C heat shock in the larval stages.
Histology
Antibodies against dVps28 were generated in rabbits. The C-terminal 134 amino acids of dVps28 (amino acids 83–236) were fused to glutathione S-transferase (GST). This fusion protein was expressed and purified using standard procedures before injection into rabbits (Ausubel et al., 1994). Eye discs were stained as described previously (Krämer et al., 1991) by using antibodies against Boss (anti-Boss NN1 at 1:3000; Krämer et al., 1991), dVps28 (1:5000), Delta (mAb202 at 1:50; Parks et al., 1995), Sevenless (1:2000; Banerjee et al., 1987), EGF receptor (1:500; Lesokhin et al., 1999), dpERK (1:500; Sigma-Aldrich, St. Louis, MO), and MYC epitope (9E10; 1:500; Covance Research Products, Berkeley, CA), and Alexa-labeled secondary antibodies (Molecular Probes, Eugene, OR). Signals were detected using a Bio-Rad MRC1024 confocal microscope with a 63x, 1.3 numerical aperture (NA) lens.
Embryos or testes were stained with TO-PRO-3 iodide (0.3 µM; Molecular Probes), phalloidin-Alexa488 (5 U/ml; Molecular Probes), or antibodies against AX-D5 (1:50; Karr, 1991), nuclear fallout (Nuf) (1:500; Riggs et al., 2003), Rab11 (1:2000; Dollar et al., 2002), or Dah (1:25; Zhang et al., 2000) as described previously (Theurkauf, 1994; Sullivan et al., 2000). To visualize F-actin with phalloidin-Alexa488, formaldehyde-fixed embryos were hand devitellinized. Staining in embryos was detected with a Leica DRM microscope equipped with a Zeiss Axiocam charge-coupled device camera with a 20x 0.5 NA lens or a Leica TCS SP2 confocal microscope with a 20x 0.7 NA or a 40x 1.2 NA 1.2. All digital images were adjusted for gain and contrast in Photoshop. To test for sperm motility, testes were dissected into Schneider's medium and directly inspected by phase contrast microscopy (Fabrizio et al., 1998).
For electron microscopy, adult eyes were fixed, embedded in plastic, sectioned as described previously (Van Vactor et al., 1991), and imaged with a JOEL 1200EX electron microscope. MVBs were recognized by their characteristic morphology, and their diameters were measured directly on the electron microscope.
Biochemistry
S2 cells were grown and transfected as described previously (Krämer and Phistry, 1996). Extracts were prepared and used for coimmunoprecipitation assays as described previously (Sevrioukov et al., 1999). dVps28 and associated proteins were precipitated using anti-dVps28 antibodies at 1:200, separated by SDS-PAGE, and detected on Western blots by using anti-HA or anti-Myc antibodies (1:1000; Covance Research Products) or anti-dVps28 (1: 2000) and enhanced chemiluminescence (Pierce Chemical, Rockford, IL).
For detection of dVps28 proteins in mutant embryos and first instars, a green fluorescent protein (GFP)-marked balancer (Casso et al., 2000) was used to distinguish homozygous and heterozygous animals. Extracts of 100 embryos or 50 larvae were prepared in 100 µl of SDS loading buffer, separated by SDS-PAGE, and Western blots were probed for dVps28 (1:5000) or tubulin (anti-tubulin DM1A 1:2000; Sigma-Aldrich). Eye pigments were measured as described previously (Ooi et al., 1997).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
To test whether the homology of Vps28 proteins extends to their biochemical activity, we investigated its binding to Vps23p/Tsg101 (Babst et al., 2000; Bishop and Woodman, 2001). The Drosophila homolog dTsg101 is encoded by cDNA GH09529 (Stapleton et al., 2002). Because antibodies are not yet available against endogenous dTsg101 protein, we coexpressed epitope-tagged versions of dTsg101 and dVps28 to test their interaction in S2 cells. Immunoprecipitation of dVps28 from S2 cells resulted in the coprecipitation of expressed HA-dTsg101 (Figure 1C, lane 5), which was increased after coexpression of Myc-dVps28 (lane 4). The interaction was specific because it was not observed with preimmune serum (lane 3), and the unrelated Hook protein was not pulled down (lane 2). These results indicated that, like its yeast and mammalian orthologs, dVps28 bound specifically to dTsg101.
A Mutation in the dVps28 Gene
The dVps28 coding unit CG12770 is contained within a single exon in the fly genome (Adams et al., 2000). The P-element–induced allele l(2)k16503 carries a transposon in the 5' untranslated region (UTR) of dVps28 (Spradling et al., 1999) and is homozygous lethal. Because excision of the P-element reverts lethality, we suspected that the lethality was due to the disruption of dVps28 function. We confirmed this by expressing a dVps28 cDNA under control of an arm-Gal4 driver (Sanson et al., 1996), which rescued the lethality of the l(2)k16503 mutation, either homozygous or hemizygous over the deficiency Df(2R)CA53, which removes the dVps28 gene (Tearle and Nüsslein-Volhard, 1987). We will therefore refer to the l(2)k16503 mutation as the dVps28l(2)k16503 allele.
In dVps28l(2)k16503 mutants, expression of the dVps28 protein is severely reduced. In dVps28l(2)k16503 embryos 20 h after egg laying (AEL), the level of dVps28 protein is reduced compared with wild-type embryos (Figure 2B), and dVps28 protein becomes undetectable by 50 h AEL. The rate of disappearance of the protein is indistinguishable between dVps28l(2)k16503 homozygous and hemizygous larvae, indicating that dVps28l(2)k16503 is a strong mutation eliminating most or all of dVps28 function. Unless otherwise indicated, the dVps28l(2)k16503 allele was used in all subsequent experiments that refer to a dVps28 mutation.
|
dVps28 Is Required for Proper Eye Development
Because the cellular phenotypes of mutations affecting membrane trafficking can be readily characterized in the compound eye (Lloyd et al., 1998; Stewart, 2002), we used variations of the FLP/FRT technique that eliminate most heterozygote or wild-type cells (Stowers and Schwarz, 1999) to generate compound eyes almost entirely composed of dVps28 mutant cells (Figure 3). In such dVps28 mutant eyes, we noticed a disorganization of ommatidia that is externally visible as roughness of the compound eye (Figure 3, B and F) and a darker eye color (Figure 3B) compared with wild-type (Figure 3A). This subtle alteration in eye color is characteristic of rough eyes because roughness changes the light-guide effects of ommatidia (Franceschini, 1972; Pichaud and Desplan, 2001). Consistent with the idea that the observed eye color change is a consequence of the rough eye pheno-type, we found that levels of both types of pigments, ommochromes and drosopterins, were not significantly different in dVps28 and wild-type eyes (our unpublished data).
|
The orange gene (or) encodes one of the subunits of the adaptor protein (AP)-3 complex in Drosophila necessary for transport to pigment granules (Lloyd et al., 1998; Mullins et al., 2000). Because in yeast the AP-3 complex is specifically required for a pathway to the vacuole distinct from the vps28-dependent route (Cowles et al., 1997; Katzmann et al., 2001), we tested the genetic interactions between or49h2 and dVps28. The phenotypes of these mutants are distinct, because or49h2 eyes exhibit loss of pigmentation but no roughness (Figure 3C). In double mutants, these genes exhibited no interactions; the or49h2 eye color defect was not modified in the or49h2, dVps28 double-mutant eyes (Figure 3D). Similarly, roughness of dVps28 mutant eyes was not significantly enhanced by loss of AP-3 function. These data suggest that the AP-3– and Vps28-dependent pathways transport different sets of cargo in Drosophila.
The external roughness of the dVsp28 compound eye reflected a variety of cellular defects as revealed by thin sections of plastic-embedded mutant compound eyes. Some ommatidia were missing photoreceptor cells (Figure 3F, arrowheads) and others contained extra cells (an extra pigment cell is indicated by the asterisk in Figure 3F). Still others contained a full complement of cells, yet they were not oriented correctly (compare the ommatidia indicated by the double arrow in Figure 3F). All of these phenotypes are consistent with misregulation of EGF receptor signaling (Wolff, 2003). It is important to note, however, that the majority of ommatidia had no visible defects, indicating that loss of dVps28 function had only minor effects on cell-cell interactions mediated through EGF receptor or other cell surface molecules during cell specification in the developing eye disk.
Mammalian cells with reduced function of the hVps28p-binding protein Tsg101 exhibit a prolonged activation of the MAP kinase pathway after EGF-induced activation (Babst et al., 2000). To directly test such a role of dVps28, we generated patches of dVps28 mutant cells in eye imaginal discs and assayed the levels of phosphorylated ERK (Figure 4). In eye discs, phosphorylated ERK detected by anti-dpERK antibodies exhibits a dynamic pattern that serves as a sensitive readout of receptor tyrosine kinase signaling (Wasserman et al., 2000; Spencer and Cagan, 2003). The comparison between wild-type and dVps28 mutant cells revealed no significant differences in the level of dpERK staining (Figure 4D). Furthermore, no significant difference was observed in ERK phosphorylation after immunoprecipitation from wild-type or dVps28 mutant larvae (our unpublished data). These results demonstrated that any changes in MAP kinase signaling suggested by developmental defects in the fly eye were too subtle too be detected by these methods.
|
To directly assess trafficking from the cell surface to lysosomes, we analyzed the distribution of two ligands, Delta and Boss, that have been used previously to visualize changes in endocytic trafficking in the fly eye (Krämer, 2002). Much of the Delta ligand detected in eye discs is present in endosomes (Parks et al., 1995), and levels are sensitive to mutations that alter lysosomal delivery (Krämer and Phistry, 1999; Lai, 2002; Le Borgne and Schweisguth, 2003). We did not observe a difference between Delta staining in wild-type and dVps28 mutant cells (Figure 4E).
The Boss ligand is expressed in the central R8 cells in developing ommatidia, binds to the Sevenless receptor, and is internalized into neighboring R7 cells (Krämer et al., 1991). Thus, the level of Boss detected in R7 cells is sensitive to mutations such as hook and dor that alter trafficking from the cell surface to lysosomes (Figure 4, F–H; Krämer and Phistry, 1996; Sevrioukov et al., 1999; Chang et al., 2002). When patches of dVps28 mutant cells were compared with dVps28+ cells, no difference could be detected in the level of Boss staining in R7 cells of otherwise wild-type (Figure 4I) or hk11 mutant eye discs (Figure 4J). Similarly, staining of eye discs with antibodies against the Sevenless receptor (Banerjee et al., 1987) or the EGF receptor (Lesokhin et al., 1999) did not reveal differences between wild-type and dVps28 mutant cells (our unpublished data). Together, these experiments demonstrated that despite the developmental defects observed in dVps28 mutant ommatidia, any effect on the lysosomal delivery of these ligands or receptors were too subtle to be detected by these methods.
MVBs Are Present in dVps28 Cells
Because Vps28 is a subunit of the ESCRT-I complex implicated in sorting of proteins into MVBs, we examined the ultrastructure of dVps28 mutant cells. We analyzed mutant cells in the retina by electron microscopy and compared them with cells in wild-type eyes. The most notable difference was the presence of large multivesicular structures in dVps28 cells (Figure 5). We measured the diameter of these MVBs along the longest axis in ultrathin sections from three wild-type and three dVps28 mutant eyes. The average diameter of wild-type MVBs was 395 ± 8 nm (mean ± SEM), whereas their size was significantly increased to 722 ± 36 nm in dVps28 cells (Figure 5C). This 1.8-fold increase in average diameter corresponds to a more than fivefold increase in the volume of MVBs. Vacuoles of increased size have been observed in hrs mutant cells presumably due to a decrease in the formation of internal vesicles (Lloyd et al., 2002). Importantly, in MVBs of dVps28 mutant cells, internal vesicles were abundant (Figure 5B). Some of the internal vesicles looked larger than those in wild-type MVBs, but this was not a statistically significant difference; in wild-type cells, the diameter of internal vesicles averaged 67 ± 13 nm (45 ± 23 vesicles per MVB, n = 7 MVBs) compared with 69 ± 14 nm (40 ± 14 vesicles per MVB, n = 7 MVBs) in dVps28 cells. The abnormal size of MVBs that we described is consistent with a function of dVps28 in MVBs, but our data indicate that, in Drosophila, dVps28 may not be strictly required for the biogenesis of MVBs.
|
dVps28 Is Required in the Female Germline
The lethal phase of dVps28l(2)k16503 homozygous or hemizygous dVps28l(2)k16503 mutants was during the first instar. Mutant embryos hatched, but most died as first instars (81%, n = 80). Some larvae had initiated (9%) or completed (9%) the formation of the second instar mouth hooks but died before molting. This developmental requirement is consistent with the increase in expression of dVps28 from first to second instar (Figure 1B).
We reasoned that a maternal contribution might be sufficient to mask an embryonic requirement for dVps28. Using the FLP/FRT ovoD system (Chou and Perrimon, 1996), we generated germline mosaics to eliminate the maternal dVps28 contribution. Embryos produced from homozygous dVps28 germ cells in which the maternal contribution was knocked out (mKO dVps28) exhibited major defects early in development (Figure 6). About one-half of these eggs (41–68% in different collections) were not fertilized as indicated by the absence of mitotic nuclear divisions 3 h AEL. We confirmed that arrested eggs were not fertilized by staining with the AX-D5 antibody (Karr, 1991) that recognizes a sperm-tail antigen (Figure 6D, inset). Note that the entire sperm tail enters the egg in Drosophila fertilization (Karr, 1991). The AX-D5 epitope was absent in all embryos that failed to undergo mitotic nuclear divisions.
|
Sperm entry and fertilization can be prevented by mutations that cause malformed micropyles, the openings through which Drosophila sperm enters an egg (Suzanne et al., 2001). In dVps28 eggs, we found that micropyles were not obviously malformed. In the majority of mKO dVps28 mutant eggs (96% mKO dVps28 mutant and wild-type embryos), the pore in the micropyle through which the sperm enters was visible by DIC imaging (Figure 6, A and B, insets), indicating that this was not the cause of the failure of these eggs to be fertilized.
The development of mKO dVps28 embryos that did undergo nuclear divisions was blocked at various stages after fertilization but before complete cellularization. A frequently observed phenotype was an irregular distribution of nuclei in the embryo (compare Figure 6A to C and D). The subset of mutant embryos that initiated cellularization exhibited two additional phenotypes. First, pole cells were frequently absent; <3% of mKOdVps28 embryos had the usual complement of cells in the characteristic position of pole cells (Figure 6, E and F). Similar phenotypes have previously been observed as a consequence of incomplete dumping of nurse cell content into the developing oocyte (e.g., in armadillo mutant germline; Cox et al., 1996). We therefore examined dVps28 germline clones but did not detect any dumping phenotypes (McKearin and Krämer, unpublished data). Nevertheless, due to the variability of mKO dVps28 embryonic phenotypes, it is possible that some reflect earlier defects that originated during oogenesis.
Cells that formed during cellularization of mKO dVps28 embryos displayed a droplet-like shape (Figure 6F) that differed from the cuboidal shape of wild-type cells (Figure 6E). This phenotype suggests a lack of membrane tension during the late phase of cellularization (Foe et al., 1993). It is interesting to note that three phenotypes observed in embryos lacking dVps28, disorganized nuclear migration, lack of pole cells, and misshapen cells correspond to cellular processes that require the function of the actin cytoskeleton (Hatanaka and Okada, 1991; Riggs et al., 2003). This points to a possible connection between dVps28 function and the regulation of actin dynamics.
The Actin Cytoskeleton in Early Embryos Requires dVps28
To directly address the possibility that dVps28 plays a role in actin dynamics, we visualized the F-actin distribution in mKOdVps28 embryos. Actin accumulates in characteristic patterns in precellular blastoderm embryos (Figure 7 A and B, inset; Foe et al., 1993). These patterns were aberrant in mKOdVps28 embryos that developed to this stage. In many of these embryos, large regions were devoid of F-actin (Figure 7, B–E, arrowheads). These correlated with the appearance of prominent, enlarged actin bundles (Figure 7E, arrows). Even in embryos in which such gross abnormalities were not apparent, closer examination revealed an uneven distribution of actin (Figure 7F, arrowheads). We noticed that defects in the actin cytoskeleton often correlated with enlarged nuclei (Figure 7, E and F, asterisks) suggestive of failures in nuclear divisions. Such nuclei seem to be displaced toward the interior of the embryos (Figure 7F and 6, C and D). This is similar to other mutants in which nuclear divisions fail with the resulting oversized nuclei either remaining at or retreating to the interior of the embryo (Sullivan et al., 1993). Importantly, such a failure of nuclear divisions is not a prerequisite for the actin gaps that also were found in the presence of normal-looking interphase (Figure 7C) or metaphase nuclei (Figure 7B). These data therefore suggest that loss of dVps28 function alters the actin cytoskeleton, and the defects in nuclear division, distribution, and axial migration are secondary defects.
|
Mutations in three genes have previously been demonstrated to have defects in the localization of actin to invaginating furrows, similar to our observations with mKO dVps28 embryos. Defective actin hexagon (Dah) is a membrane-associated protein that localizes to invaginating furrows in syncytial blastoderm embryos and during cellularization. In embryos lacking maternal Dah protein, actin fails to properly localize to the furrow structures (Zhang et al., 1996). Nuf and Rab11 are two additional proteins involved in this process (Rothwell et al., 1998; Dollar et al., 2002; Pelissier et al., 2003). They are necessary for the delivery of vesicles from recycling endosomes to the invaginating furrows. In their absence, Dah is not recruited to furrows and as a consequence similar actin defects were observed as those induced by the lack of Dah protein itself (Riggs et al., 2003).
Prompted by the similarity of these phenotypes to our observations, we investigated the distribution of these three proteins in mKO dVps28 embryos. We found that in mKO dVps28 embryos, Nuf and Rab11 accumulated around centrosomes apical to prophase nuclei during late cortical divisions, indistinguishable from wild-type embryos (Figure 8, A–D). By contrast, Dah distribution in mKO dVps28 embryos is altered. Rather than being recruited to the invaginating furrows, as it is in wild-type embryos (Figure 8, E and G, an example is indicated by the arrow in the cross-sectional view in G), Dah remains distributed in small puncta and is found in aberrant accumulations within the mKO dVps28 embryos (Figure 8H, arrows). These data indicate that dVps28 is required for the normal localization of Dah to furrows by a mechanism that is independent of Rab11 and Nuf's localization to recycling endosomes.
|
dVps28 Is Required in the Male Germline
A connection between dVps28 function and the actin cytoskeleton also was observed during spermatogenesis. Lethality of dVps28 was rescued by expression of dVPS28 under control of arm-gal4 driver (Sanson et al., 1996), which provides widespread expression. Transgenic rescue restored viability and female fertility. Rescued males, however, lacked dVps28 expression in the male germ line (Figure 9, A–D). In wild-type spermatocysts, dVps28 is strongly expressed and found throughout the cytoplasm (Figure 9A). This staining was absent in the testis of dVps28 males that were rescued by expression of dVps28 under arm-Gal4 control. When such males were crossed to wild-type virgin females, 100% of the eggs were unfertilized as judged by AX-D5 staining (Karr, 1991), indicating that these males did not produce functional sperm.
|
The earliest stage of spermatogenesis visibly affected by the dVps28 mutation was sperm individualization. Expression of dVps28 was relatively low in the early stages of spermatogenesis in wild-type cells. Although we cannot rule out subtle earlier defects, spermatogenesis proceeded through these stages in dVps28 mutant testis to yield a syncytium containing a bundle of spermatid nuclei indistinguishable from wild-type cells (Figure 9). Such syncytial nuclei are resolved into separate sperm during sperm individualization (Fabrizio et al., 1998). This process encompasses a dramatic reorganization of the plasma membrane driven by an actin-dependent synchronous movement of a structure called the cystic bulge (Rogat and Miller, 2002; Noguchi and Miller, 2003). Its progress along the spermatid tails can be visualized by fluorescently-labeled phalloidin. This staining highlights the characteristic actin cones that first assemble around the bundle of 64 spermatid nuclei before moving toward the distal end (Figure 9, E and F). In wild-type testis the moving actin cones were tightly synchronized (Figure 9G), whereas they seemed disorganized in dVps28 mutant testis (Figure 9H). As a consequence, individual motile sperm were absent in the majority of dissected dVps28 testes, and the remaining rare males had only a few motile sperm. Our observation that dVps28 males were completely sterile indicates a strict requirement of dVps28 in spermatogenesis.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
; Raiborg et al., 2003). Vps28's role in these processes has been addressed by several approaches. In yeast, loss of vps28 function causes malformed prevacuolar endosomal compartments (Rieder et al., 1996) and defects in protein sorting to the interior vesicles within MVBs (Katzmann et al., 2001). Consistent with these findings, injection of anti-hVps28 antibodies into mammalian cells resulted in reduced lysosomal delivery of EGF receptors and accumulation of ubiquitinated cargo (Bishop et al., 2002). Vps28p binds directly to other ESCRT proteins (Babst et al., 2000; Bishop and Woodman, 2001; Martin-Serrano et al., 2003; Bowers et al., 2004), and the Vps28 binding site in Tsg101 is necessary for ES-CRT-1 function (Martin-Serrano et al., 2003). Together, these experiments point to a critical requirement for Vps28p in ESCRT-1 function.
Here, we describe a mutation in dVps28, which encodes one of the ESCRT-I subunits in Drosophila. Consistent with the complex's conserved function from yeast to mammalian cells (Babst et al., 2000; Bishop and Woodman, 2001; Bishop et al., 2002; Bache et al., 2003; Katzmann et al., 2003), loss of dVps28 function caused morphological changes in MVBs and developmental defects in the compound eye that indicate a subtle misregulation of cell signaling molecules. However, for the ligands and receptors tested, we did not detect significant changes in their cell surface levels or their delivery to lysosomes in dVps28 cells, indicating that any changes were too subtle to be detected by the immunofluorescence methods used.
One possible explanation for this observation is that some Vps28 functions are partially fulfilled by another protein. It is not likely that this hypothetical protein is similar to Vps28p in sequence because we did not identify another Vps28-like molecule in the completed genome sequences of Drosophila melanogaster. Another possibility is that the dVps28l(2)k16503 allele does not completely inactivate dVps28 function. Although we cannot rule out this possibility, it is unlikely for two reasons. First, the dVps28l(2)k16503 allele is a strong mutation that causes lethality early in development at the transition from first to second instar, much earlier than a null mutation in the hrs gene, which regulates the sorting of receptors into MVBs (Lloyd et al., 2002). Furthermore, the lethal phase and the loss of dVps28 protein were indistinguishable between larvae homozygous for dVps28l(2)k16503 and those that were hemizygous over a deficiency of the region. This indicates that the dVps28l(2)k16503 allele removes most if not all of dVps28 function.
Another possibility for the relatively mild phenotypes is the perdurance of dVps28 protein. This is suggested by the dramatic phenotypes after removal of the maternal contribution in mKO dVps28 embryos: many remained unfertilized and the remaining embryos were arrested in their development before reaching the cellular blastoderm. It is interesting to compare this phenotype to that of embryos lacking any maternal Hrs contribution. Hrs binds to mono-ubiquitinated membrane proteins, whose sorting into MVBs is thought to be mediated by Hrs binding to ESCRT-1 (Bishop et al., 2002; Bache et al., 2003; Katzmann et al., 2003; Lu et al., 2003). Consistent with this notion, mKO hrs embryos exhibited a reduced down-regulation of cell surface receptors and a resulting enhanced activation of the MAP kinase pathway (Lloyd et al., 2002). Importantly, these defects were observed after cellularization has completed and during gastrulation. Our finding that mKO dVps28 embryos exhibited major defects before completion of cellularization indicates functions of dVps28 in addition to the down-regulation of membrane proteins mediated by the interaction of ESCRT-I with Hrs.
Surprisingly, the earliest defect we found in fertilized eggs was an uneven distribution of nuclei. In Drosophila, the first 13 nuclear divisions occur without accompanying cell divisions. The first five of these syncytial divisions occur close to the center of the embryo, with nuclei subsequently spreading out along the anterior-posterior axis. This process, referred to as axial expansion, depends on the function of the actin cytoskeleton (Hatanaka and Okada, 1991; von Dassow and Schubiger, 1994; Wheatley et al., 1995). In mKO dVps28 embryos, this process often was disorganized resulting in an uneven distribution of nuclei.
Other dVps28 phenotypes emerged during cellularization. After axial expansion, the nuclei move to the embryo's cortex where the last four of the syncytial nuclear divisions occur. The resulting nuclei are surrounded by invaginating membranes in a process that requires the actin–myosin network (Foe et al., 1993; Royou et al., 2004). At this stage, two defects were evident in embryos lacking dVps28. Many of the cells that formed were droplet shaped instead of the usual cuboidal shape. The actin cytoskeleton is required for normal cellularization and interfering with is function, by cytochalasin (Royou et al., 2004, and references therein), or altering the activity of the Rho or Cdc42 GTPases (Crawford et al., 1998), interferes with normal cellularization. Furthermore pole cells, usually the first cells to form, were absent in most embryos. The lack of pole cells has previously been observed in embryos in which axial expansion is perturbed upon interference with the actin cytoskeleton (Hatanaka and Okada, 1991).
All of these early phenotypes are consistent with a role of dVps28 in directly or indirectly organizing the actin cytoskeleton. A role for endosomes in actin remodeling during cellularization has previously been established in embryos mutant for nuf or rab11 (Riggs et al., 2003). The small GTPase Rab11 localizes to recycling endosomes (Sonnichsen et al., 2000; Dollar et al., 2002), and Nuf is a homolog of Arfo2 that directly binds Rab11 and acts in recycling endosomes (Hickson et al., 2003). Mutants eliminating either of these proteins cause gaps in which actin fails to be recruited to the furrows during cortical nuclear divisions, similar to our observations in mKO dVps28 embryos (Figure 7). In all three of these mutants, the actin defects may be a consequence of the failure to recruit Dah protein to invaginating furrows (Figure 8; Rothwell et al., 1999; Riggs et al., 2003). Dah has significant similarity to dystrobrevin and dystrophin (Zhang et al., 1996). Dystrophin plays a critical role in anchoring the actin cytoskeleton to membranes (Michalak and Opas, 1997), and consistent with a similar function for Dah, embryos lacking maternally contributed Dah fail to properly assemble the actin cytoskeleton at furrows (Zhang et al., 1996).
A subset of the defects in rab11 mutant embryos has been linked to a requirement for trafficking through the recycling endosome during cellularization (Pelissier et al., 2003). Consistent with this notion, defects in rab11 and nuf embryos only become apparent during cortical divisions (Rothwell et al., 1998; Riggs et al., 2003). In mKO dVps28 embryos, by contrast, defects were detected in the distribution of nuclei, long before cellularization initiates (Figure 6). This indicates a function of dVps28 in actin remodeling independent of recycling to the cell surface and independent of Rab11 and Nuf. This is consistent with our finding that in mKO dVps28 embryos, recycling endosomes are not affected, as judged by Rab11 and Nuf localization (Figure 8), suggesting that dVps28 is required for furrow localization of Dah down-stream or in parallel to Rab11 and Nuf's function in recycling endosomes. Such a model is difficult to reconcile with the canonical function of Vps28 as an ESCRT-1 subunit involved in targeting proteins into the interior vesicles of late endosomes (Katzmann et al., 2002; Raiborg et al., 2003).
It is unlikely that Dah mediates all effects of dVps28 on the actin cytoskeleton. No defects are observed in embryos lacking Dah before cycle 10 (Zhang et al., 1996), long after defects have become apparent in mKO dVps28 embryos (Figure 7). Furthermore, Dah is not required during spermatogenesis, another developmental context in which we observed effects of dVps28 on the organization of the actin cytoskeleton. Spermatogenesis in dVps28 mutant testis progresses until bundles of 64 syncytial spermatids are formed (Figure 8). These spermatids are separated by a process called individualization that requires complex membrane rearrangements (Fabrizio et al., 1998). At the site of these rearrangements, syncytial membrane, cytoplasm, and vesicles accumulate in the cystic bulge (Noguchi and Miller, 2003). In wild-type testis, an actin-dependent process drives the cystic bulge away from the 64 spermatid nuclei toward the distal end. The loss of synchrony in the movement of the cystic bulge is the first detectable defect during spermatogenesis in dVps28 mutant testis (Figure 8). Because dah mutants have no phenotype in males (Zhang et al., 1996), these results indicate an independent connection between dVps28 function and the actin cytoskeleton.
Actin acts at many stages in the endocytic pathway (Engqvist-Goldstein and Drubin, 2003). In yeast, the initial internalization step requires actin polymerization (Kubler and Riezman, 1993; Wendland et al., 1998; Kaksonen et al., 2003), and in mammalian cells, early endosomes move on the tip of actin tails (Merrifield et al., 1999). Additionally, late endocytic organelles require the actin cytoskeleton for fusion in yeast (Eitzen et al., 2002) and mammalian cells (Kjeken et al., 2004). Furthermore, a screen of the yeast genome for mutations interfering with protein sorting to the vacuole identified several regulators of the actin cytoskeleton (Bonangelino et al., 2002).
Importantly, several of these mutants identified on the basis of a vacuolar sorting phenotype also exhibited defects in the organization of the actin cytoskeleton. For example, aberrant actin patches and a reduction of actin cables were observed in yeast lacking Vps36p (Bonangelino et al., 2002), one of the subunits of the ESCRT-II complex (Babst et al., 2002b). It will be interesting to see whether a similar functional connection between the actin cytoskeleton and the ESCRT-I complex may underlie the enigmatic phenotypes of Tsg101 mutations in mice that cause cell cycle arrest and early embryonic lethality (Ruland et al., 2001; Krempler et al., 2002).
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
Footnotes |
|---|
* These authors contributed equally to this study. 
Present address: Developmental and Cell Biology, University of California, Irvine, 5205 McGaugh Hall, Irvine, CA 92697-2300.
Present address: Massachusetts General Hospital, Neuroscience Center, Molecular Neurogenetics Unit, Bldg. 149, 13th St., #6116, Charlestown, MA 02129.
Address correspondence to: Helmut Krämer (helmut.kramer{at}utsouthwestern.edu).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and Struhl, K. ((1994). ). Current Protocols in Molecular Biology, New York: John Wiley & Sons.
Babst, M., Katzmann, D. J., Estepa-Sabal, E. J., Meerloo, T., and Emr, S. D. ((2002a). ). ESCRT-III: an endosome-associated heterooligomeric protein complex required for MVB sorting. Dev. Cell 3, , 271–282.[CrossRef][Medline]
Babst, M., Katzmann, D. J., Snyder, W. B., Wendland, B., and Emr, S. D. ((2002b). ). Endosome-associated complex, ESCRT-II, recruits transport machinery for protein sorting at the multivesicular body. Dev. Cell 3, , 283–289.[CrossRef][Medline]
Babst, M., Odorizzi, G., Estepa, E. J., and Emr, S. D. ((2000). ). Mammalian tumor susceptibility gene 101 (TSG101) and the yeast homologue, Vps23p, both function in late endosomal trafficking. Traffic 1, , 248–258.[CrossRef][Medline]
Bache, K. G., Brech, A., Mehlum, A., and Stenmark, H. ((2003). ). Hrs regulates multivesicular body formation via ESCRT recruitment to endosomes. J. Cell Biol. 162, , 435–442.
Bache, K. G., Slagsvold, T., Cabezas, A., Rosendal, K. R., Raiborg, C., and Stenmark, H. ((2004). ). The growth-regulatory protein HCRP1/hVps37A is a subunit of mammalian ESCRT-I and mediates receptor down-regulation. Mol. Biol. Cell 15, , 4337–4346.
Banerjee, U., Renfranz, P. J., Hinton, D. R., Rabin, B. A., and Benzer, S. ((1987). ). The Sevenless+ protein is expressed apically in cell membranes of developing Drosophila retina; it is not restricted to cell R7. Cell 51, , 151–158.[CrossRef][Medline]
Berson, J. F., Harper, D. C., Tenza, D., Raposo, G., and Marks, M. S. ((2001). ). Pmel17 initiates premelanosome morphogenesis within multivesicular bodies. Mol. Biol. Cell 12, , 3451–3464.
Bishop, N., Horman, A., and Woodman, P. ((2002). ). Mammalian class E vps proteins recognize ubiquitin and act in the removal of endosomal proteinubiquitin conjugates. J. Cell Biol. 157, , 91–101.
Bishop, N., and Woodman, P. ((2001). ). TSG101/mammalian VPS23 and mammalian VPS28 interact directly and are recruited to VPS4-induced endosomes. J. Biol. Chem. 276, , 11735–11742.
Bonangelino, C. J., Chavez, E. M., and Bonifacino, J. S. ((2002). ). Genomic screen for vacuolar protein sorting genes in Saccharomyces cerevisiae. Mol. Biol. Cell 13, , 2486–2501.
Bowers, K., Lottridge, J., Helliwell, S. B., Goldthwaite, L. M., Luzio, J. P., and Stevens, T. H. ((2004). ). Protein-protein interactions of ESCRT complexes in the yeast Saccharomyces cerevisiae. Traffic 5, , 194–210.[CrossRef][Medline]
Brand, A. H., and Perrimon, N. ((1993). ). Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development 118, , 401–415.[Abstract]
Bunch, T. A., Grinblat, Y., and Goldstein, L. S. ((1988). ). Characterization and use of the Drosophila metallothionein promoter in cultured Drosophila melanogaster cells. Nucleic Acids Res. 16, , 1043–1061.
Casso, D., Ramirez-Weber, F., and Kornberg, T. B. ((2000). ). GFP-tagged balancer chromosomes for Drosophila melanogaster. Mech. Dev. 91, , 451–454.[CrossRef][Medline]
Chang, H. C., Newmyer, S. L., Hull, M. J., Ebersold, M., Schmid, S. L., and Mellman, I. ((2002). ). Hsc70 is required for endocytosis and clathrin function in Drosophila. J. Cell Biol. 159, , 477–487.
Chou, T. B., and Perrimon, N. ((1996). ). The autosomal FLP-DFS technique for generating germline mosaics in Drosophila melanogaster. Genetics 144, , 1673–1679.[Abstract]
Cowles, C. R., Odorizzi, G., Payne, G. S., and Emr, S. D. ((
35 kDa (asterisk) is not known. In a parallel blot, tubulin served as a loading control. (C) Coimmunoprecipitation demonstrated binding of dVps28 to dTsg101. S2 cells were transfected with Myc-dVps28, HA-dTsg101, HA-Hook, or combinations as indicated. Expressed proteins were detected in the input samples consisting of S2 cells lysates with immunoblots (IB) by using anti-dVps28 (top) or anti-HA (middle). Top, samples that were transfected with Myc-dVps28 (lanes 2, 3, 4, and 6) were diluted fivefold before loading to generate comparable signals. After immunoprecipitation (IP) with anti-dVps28 antibodies (bottom) only HA-dTsg101 (lanes 4 and 5) but not HA-Hook (lane 2) was detected by Western blotting by using anti-HA. Endogenous dVps28 was sufficient to pull-down a small amount of dTsg101 (lane 5), but no coimmunoprecipitation was detected when preimmune serum was used (lane 3).
