|Home | About | Journals | Submit | Contact Us | Français|
Bicaudal C and trailer hitch are both required for dorsoventral patterning of the Drosophila oocyte. Each mutant produces ventralized eggs, a phenotype typically associated with failure of the oocyte to provide a dorsalization signal – the Gurken protein – to the follicle cells. Bicaudal C and trailer hitch are both implicated in post-transcriptional gene regulation. Bicaudal C acts in recruiting a deadenylase to specific mRNAs, leading to translational repression. The role of trailer hitch is less well defined, but mutants have defects in protein secretion, and show aberrant distribution of an endoplasmic reticulum exit site marker whose mRNA is associated with Trailer hitch protein. We show that Bicaudal C and trailer hitch interact genetically. Mutants of these two genes have shared defects in localization of gurken and other anteriorly-localized mRNAs, as well as altered microtubule organization which may underlie the mRNA localization defects. Bicaudal C and trailer hitch mutants also share a syndrome of actin-related abnormalities, including the formation of ectopic actin cages near the anterior of the oocyte. The cages sequester Gurken protein, blocking its secretion and thus interfering with signaling of the follicle cells to specify dorsal fate.
The process of pattern formation is central to developmental biology. Axial pattern is established in the egg or embryo, and more refined patterns are set up repeatedly in later development. One mechanism often used to elaborate axial pattern involves localized cytoplasmic determinants. These determinants are molecules that can direct a path of development, and are positioned at discrete sites in the oocyte. The genetic identification in Drosophila of axial patterning determinants allowed questions to be posed on two fronts. How do the determinants act? And how are the determinants deployed at specific sites? The primary structures of some of the determinants suggested likely roles. For example, the anterior patterning morphogen Bicoid (Bcd) contains a homeodomain and has features of a transcription factor (Berleth et al., 1988). Experimental evidence has proven this role (Driever and Nüsslein-Volhard, 1989; Driever et al., 1989; Struhl et al., 1989). The Gurken (Grk) protein, which specifies dorsal fate in the egg chamberNeuman-Silberberg and Schupbach, 1993; Ray and Schupbach, 1996), has features of a secreted signaling factor, and this role has been confirmed experimentally (Neuman-Silberberg and Schupbach, 1993; Ghiglione et al., 2002).
As for the question of how the determinants are selectively deployed, initial studies immediately pointed to mRNA localization as a key mechanism. The bcd mRNA is tightly restricted to the anterior of the oocyte (Berleth et al., 1988), while the oskar (osk) mRNA, which encodes a posterior patterning determinant (Lehmann and Nüsslein-Volhard, 1986), becomes positioned at the posterior pole of the oocyte (Ephrussi et al., 1991; Kim-Ha et al., 1991). The grk mRNA adopts a position at the junction between anterior and lateral cortex of the oocyte, adjacent to the nucleus and at the site destined to become the dorsal surface of the embryo (Neuman-Silberberg and Schupbach, 1993). Subsequent analyses have focused on how individual mRNAs are targeted for localization, with identification of localization signals within the mRNAs and proteins that bind the signals (Bashirullah et al., 1998; Johnstone and Lasko, 2001; Jambhekar and Derisi, 2007). Movement of the mRNAs relies on the cytoskeleton, and both microtubules and microtubule-based motors are required (Pokrywka and Stephenson, 1991; Clark et al., 1994; Pokrywka and Stephenson, 1995; Brendza et al., 2000; Duncan and Warrior, 2002; Januschke et al., 2002; Lopez de Heredia and Jansen, 2004). In at least one case, that of the grk mRNA, directed movement along the microtubules has been shown (Delanoue et al., 2007). The dependence on post-transcriptional regulation in deployment of the determinants is not limited to mRNA localization. Translational regulation is also essential, serving to ensure that the determinants are only synthesized once the mRNAs are properly localized (Lipshitz and Smibert, 2000).
Insights about these mechanisms have come in part from analysis of mutants displaying defects in one or more aspects of axial patterning. Disruption of any step in the synthesis and processing of the determinants has the potential to alter patterning, and genetic studies have been especially effective in revealing the full range of mechanisms used (St Johnston, 1995; Bashirullah et al., 1998). The mechanisms are not limited to events that affect the distribution or activity of the determinant mRNAs. In particular, because Grk is a secreted signaling molecule, the steps in its trafficking may be revealed by mutants with defects in dorsoventral patterning (Roth et al., 1995; Bokel et al., 2006).
The trailer hitch (tral) gene was identified in a screen for mutants affecting axial patterning (Wilhelm et al., 2005). Grk protein is mislocalized in oocytes of tral mutant females, appearing in cytoplasmic puncta. Based on the subcellular distribution of Tral – substantially colocalized with the ER marker GFP::KDEL – and the association of Tral with protein factors involved in mRNA regulation and with specific mRNAs encoding components of the secretory apparatus, Wilhelm et al. proposed that tral regulates expression of ER exit site component mRNAs on the surface of the ER (Wilhelm et al., 2005). Misregulation of one or more of these mRNAs would disrupt protein trafficking, causing the mislocalization of Grk protein.
We have also characterized mutants with dorsoventral patterning defects, focusing on tral and Bicaudal C (BicC) which we show interact genetically. Our analysis reveals multiple defects shared by the mutants. These include aberrant localization of grk and other mRNAs, altered microtubule organization within the oocyte, and a syndrome of microfilament-related abnormalities. The most striking of the microfilament defects is the appearance of ectopic actin cages that sequester Grk. Despite the extensive similarities, the BicC mutants do not share the effect of tral mutants on ER exit sites. We suggest that formation of the actin cages causes Grk mislocalization and thus interferes with dorsoventral patterning, although our results do not rule out a contribution of ER exit site dysfunction to disruption of Grk activity.
w1118 was used as wild type. PCG10686KG08052 (tral1)(Bellen et al., 2004; Wilhelm et al., 2005), from the Bloomington Stock center, was extensively backcrossed to w1118 with no change in phenotype. In the course of making an imprecise excision mutant of tral1 we characterized a precise excision line, in which wild type viability and fertility was restored. GFP Protein Trap lines P(PTT-GA)CG10686G89 (tralG89) and P(PTT-GA)CG10686G140 (Morin et al., 2001) were from Lynn Cooley and Bill Chia, P[Sar1::GFP] flies were from Jim Wilhelm, P[protein disulfide isomerase (PDI)::GFP] was from John Sisson, the tral genomic rescue construct was from Mani Ramaswami, me31BΔ1 flies were from Akira Nakamura, and BicCRU35 and cniAA flies were from Trudi Schüpbach. All other stocks were from the Bloomington Stock Center. tralG89 and tral1 were recombined with P[FRT(whs)]2A, and germ line clones were analyzed using the dominant female sterile technique (Chou et al., 1993).
Ovaries were fixed for 5 min and stained as described (Findley et al., 2003; Snee and Macdonald, 2004). Egg chambers for microtubule staining were fixed with freshly made formaldehyde and were washed at least five times with PBST (PBS plus 0.1% Triton X-100), then with PBS containing 1% Triton X-100 and 3% normal goat serum for 1 h before incubating with the tubulin antibody overnight. Tubulin fluorescence in the oocyte was measured along lines, 15 pixels wide, stretching from the anterior to the posterior using the plot profile tool in ImageJ (National Institutes of Health). The data were transferred to Microsoft Excel for analysis.
Rabbit polyclonal Tral antibodies were raised against an N terminal region (residues 30–376) of Tral encoded by the BamHI to BamHI fragment (nucleotides 512–1548 of Genbank accession NM_140328) of the tral EST GH08269 (Drosophila Genomics Resource Center). Primary antibodies used were rabbit anti-Tral (1/500), rabbit anti-BicC (1/1000), mouse anti-Grk 1D12 (1/10), mouse anti-Kelch 1B (undiluted), mouse anti-HtsRC (1/10), mouse anti-Hts 1B1 (1/10), mouse anti-Syntaxin (1/10), mouse anti-BiP (1/10), mouse anti-KDEL (1/400, Stressgen, MI), rabbit anti-Lava Lamp (Lva) (1/5000), mouse anti-Rho1 (1/10), mouse anti-Spectrin (1/10), and mouse anti-β-tubulin E7 (1/100 of antibody concentrate). We also used AlexaFluor 488- or 568-conjugated phalloidin (Molecular Probes).
FM1-43 labeling of the plasma membrane was performed by dissecting ovaries into 5 μg/ml FM1-43 (Invitrogen) in Schneider’s insect medium (Invitrogen) then processing them for live imaging as described (Snee and Macdonald, 2004).
Ovaries were fixed in 4% formaldehyde in PBS for 20 min, washed in PBST, and processed for in situ hybridization as described (Tautz and Pfeifle, 1989), using antisense probes labeled with digoxigenin, and previously described modifications (Kim-Ha et al., 1991). After the post hybridization washes the samples were incubated with anti-digoxigenin-HRP (diluted 1/200; Roche) for 1–2 hours, were washed multiple times in PBST and incubated in Tyramide-Cy5 reagent (Perkin Elmer) diluted 1/50 in the manufacturers amplification diluent for 30 min, then were washed in PBST several times and mounted on slides.
For genetic analysis of tral we used two P element insertion mutants, tralG89 (a GFP Trap) (Morin et al., 2001) and tral1, (Wilhelm et al., 2005) as well tral1Δ1, a weaker allele obtained by incomplete excision of the tral1 P element (Materials and methods). Other alleles of tral have transposon insertions in the same genomic region (the first intron of tral) Wilhelm et al., 2005), or have a deletion that removes the 5′ part of the gene (Barbee et al., 2006); all fail to complement the mutants used here. Our results confirm and extend the genetic characterization of tral. (Wilhelm et al., 2005; Barbee et al., 2006) In addition to the detailed molecular analysis described below, we provide four additional insights into tral genetics. First, we find that tral mutants are substantially defective in egg laying. The rare hemizygous tral1/Df(3L)iro-2 females produce eggs at a rate substantially less than wild type. These females accumulate many late stage oocytes, but most are not laid. Second, the eggs retained by tral1/Df(3L)iro-2 and tral1Δ1/tral1Δ1 females have multiple morphological defects: many have fused dorsal appendages (Wilhelm et al., 2005); Sup. Fig. 1B, Sup. Table 1), many are smaller than wild type, and a small proportion have open anterior chorions (Sup. Fig. 1C). Third, clonal analysis with tral1 and tralG89 shows that each of the aforementioned phenotypes persist when only the germline cells (nurse cells and oocyte) are mutant, confirming the expectation that tral acts in the germline (Sup. Table 1). Finally, a genomic tral+ transgene rescues all of these phenotypes, raising the degree of confidence that they are caused by loss of tral activity.
BicC mutants have a mild dominant eggshell phenotype, with a minor fraction (5%) of eggs from BicCWC45 heterozygotes showing shortened or fused dorsal appendages. This defect is dramatically enhanced when the flies are also heterozygous for tral: almost all eggs of BicCWC45/+; tral1/+ females have defective dorsal appendages, as do a majority of the eggs with the combination of BicCWC45 and the weaker tral1Δ1 allele (Table 1). The distribution and levels of immunostaining for Tral are not detectably altered in BicC mutants, nor is BicC staining altered in tral mutants (Sup. Fig. 2). Thus, the strong genetic interaction is not simply due to the regulation of one gene by the other.
BicC and Tral proteins have a similar distribution in the cytoplasm. Tral has been described as being enriched at discrete sites on the endoplasmic reticulum (Wilhelm et al., 2005). Because the sites of Tral concentration are also highly enriched for the sponge body proteins Cup, Me31B, and Exu::GFP ((Wilhelm et al., 2005); data not shown), these sites appear to be sponge bodies. Similarly, BicC is also present in many cytoplasmic puncta, which contain the sponge body marker Exu::GFP (Sup. Fig. 2). Sponge bodies are subcellular structures initially identified as sites enriched in Exuperantia (Exu), a protein that acts in mRNA localization in the ovary (Berleth et al., 1988; St Johnston et al., 1989; Wilsch-Brauninger et al., 1997). Several other proteins are also highly enriched in sponge bodies, and all have been implicated in post-transcriptional gene regulation (Wilhelm et al., 2000; Nakamura et al., 2001; Wilhelm et al., 2003; Nakamura et al., 2004; Lin et al., 2006; Delanoue et al., 2007).
To ask if enhancement of the BicC phenotype by tral mutants is a property shared with mutants defective for other sponge body components, we examined the eggs of BicCWC45/+ females that were also heterozygous for cup, exu or me31B. exu and me31B mutants do enhance the dominant phenotype, although not to the same degree as tral mutants, while cup mutants have a very weak effect (Table 1). Thus, BicC genetically interacts with multiple sponge body genes, consistent with the notion that a combination of different sponge body genes acts in the regulation of an overlapping set of transcripts. It is difficult to interpret the different degrees of enhancement, as not all of the mutants are null alleles. For example, cupΔ212 is known to retain some degree of function (Nakamura et al., 2004). Nevertheless, the strongest interaction detected here is between tral and BicC, and the results presented below demonstrate that these two genes have closely related roles.
Dorsoventral patterning relies on spatially restricted signaling by the Grk protein. During stage 8 of oogenesis grk mRNA becomes localized anterodorsally within the oocyte. Grk protein made from the localized mRNA enters the secretory pathway and signals the overlying somatic follicle cells to adopt a dorsal fate (Schüpbach and Roth, 1994). Expression of Grk is altered in tral mutants, with multiple abnormally large puncta of Grk appearing in the oocyte (Wilhelm et al., 2005). BicC mutants share this phenotype, which is fully penetrant (Fig. 1D–F). Displacement of Grk protein from the oocyte/follicle cell boundary and into the puncta presumably blocks its normal program of secretion, and this phenotype should contribute to the dorsoventral patterning defects of the mutants.
The large Grk puncta in tral mutants are enriched in actin (Wilhelm et al., 2005). We have confirmed that Grk is closely associated with actin, and find the Grk puncta in both mutants are often contained in, or coincident with, a framework of actin. These unusual actin structures, which we call actin cages, are not found in wild type oocytes. Actin cages are present along the entire anterior region of the oocyte, although only those near the dorsal side contain Grk puncta. Other microfilament-related proteins that associate with the cages include Spectrin (Fig. 1J–L), which links the microfilament cytoskeleton to membranes (Bennett, 1990), Hts, closely related to Adducin, which organizes Spectrin/actin networks (Fig. 1P–R), Quail (Sup. Fig. 3), which bundles and stabilizes microfilaments, (Matova et al., 1999), and Rho1 (Sup. Fig. 3), which is important for microfilament organization in the ovary (Magie et al., 1999).
During stages 7–9 the association of Grk with actin cages is not accompanied by any substantial change in the overall level of Grk. However, beginning at stage 10A the level of Grk in tral (but not BicC) mutants is substantially reduced (Fig. 2, Sup. Table 2). There is no obvious reduction in the amount of grk mRNA (Sup. Table 3). Thus, the tral mutants are defective in translation or stabilization of Grk protein, in addition to their mislocalization of Grk protein in actin cages earlier in oogenesis.
To determine if the abnormal Grk distribution might result from a change in grk mRNA localization, grk transcripts in stage 8 egg chambers were detected by in situ hybridization. grk mRNA normally borders the nucleus at the junction of anterior and lateral boundaries of the oocyte. In both BicC and tral mutants, this pattern is only partially preserved. grk mRNA is restricted primarily to the anterior of the oocyte adjacent to the nucleus oocyte, and is largely absent from the lateral side of the nucleus (Fig. 3A–C). Notably, grk mRNA is always closely aligned with the anterior oocyte margin, and is not found displaced from the immediate anterior in the pattern observed for Grk protein and the actin cages. Because essentially every BicC mutant egg chamber of the appropriate stage has multiple actin cages containing Grk protein and many of these are displaced from the anterior margin, the failure to detect any grk mRNA away from the anterior margin allows us to infer that there is no colocalization of grk mRNA and protein in the cages. Thus the accumulation of Grk in the cages cannot be due to mislocalization of grk mRNA to those sites.
Localization of grk mRNA involves movement along microtubules. Consequently, the alteration of grk mRNA distribution in BicC and tral mutants could be due to incorrect microtubule organization. In wild type stage 8 oocytes microtubule density is relatively even throughout the oocyte, with some enrichment near the anterior and periphery (Fig. 3G,J). This pattern is altered in BicC and tral mutants, with a zone along the anterior of the oocyte showing a significantly higher concentration of microtubules (Fig. 3H,I,J). In contrast, the more posterior region of the oocyte is relatively deficient in microtubules.
The abnormal microtubule organization of BicC and tral mutants raised the possibility that localization of other mRNAs, especially those positioned at the oocyte anterior, might also be defective. The htsN4 mRNA is localized to the oocyte anterior with heavy enrichment at the lateral boundaries (Fig. 3D). In BicC mutant oocytes htsN4 mRNA remains at the anterior, but dispersed rather than being concentrated laterally (Fig. 3E). A similar but weaker effect is seen in tral mutants (Fig. 3F). The effect of both mutants on the anteriorly localized bcd mRNA is similar, but less extreme (data not shown). Localization of mRNAs to the posterior of the oocyte is affected to a much lesser extent. In all tral and most BicC mutant egg chambers there is significant localization of osk mRNA to the posterior pole, with some puncta of osk mRNA elsewhere in the oocyte in BicC mutants (data not shown; (Mahone et al., 1995)).
Our results demonstrate that BicC and tral mutants are defective in microtubule organization and anterior mRNA localization. The observed loss of lateral grk mRNA localization might interfere with correct signaling, as very little of the protein would be synthesized immediately adjacent to the follicle cells that need to receive the signal. However, this effect on grk mRNA localization does not appear to be sufficient to account for the abnormal positioning of Grk protein in the actin cages.
The inferred absence of grk mRNA in the actin cages suggests that Grk protein is synthesized on rough endoplasmic reticulum (rough ER) in the cytoplasm, and then delivered to the cages. The actin cages could selectively harbor Grk. Alternatively, protein trafficking might be more generally disrupted, such that other secreted proteins are also enriched in the cages. To address the latter possibility, we tested the distributions of other secreted proteins. Syntaxin is normally secreted to the plasma membrane of the oocyte (Fig. 4A–C; (Sommer et al., 2005)). In BicC and tral mutants Syntaxin is present at the plasma membrane, and is also highly enriched in the actin cages (Fig 4D–F and data not shown). A second secreted protein, Cadherin, is present in the cages, but not highly enriched there (data not shown). The vitellogenin receptor does not colocalize with actin puncta and thus appears not to be in the actin cages (Wilhelm et al., 2005). In summary, Grk is not the only secreted protein to be trapped in the actin cages, but the degree of enrichment in the cages varies for different secreted proteins and can be low or undetectable.
Because the cages contain some secreted proteins, it is possible that a defect in protein secretion underlies formation of the cages. This seems unlikely, as the general protein secretion defects of sec5 mutants are not associated with formation of actin cages (Murthy and Schwarz, 2004). Similarly, cornichon (cni) mutants have a highly penetrant defect in ER to Golgi export of Grk protein (Queenan et al., 1999; Bokel et al., 2006), yet do not have actin cages (Fig. 4G–I).
The delivery of Grk and Syntaxin (and presumably other secreted proteins) to the actin cages suggests that the cages should be positioned close to ER. Moreover, the ER could be abnormally concentrated near the cages. We tested the distributions of BiP (a chaperone located in the lumen of the ER)(Munro and Pelham, 1986), Protein Disulfide Isomerase (PDI)::GFP (PDI acts in the ER lumen) (Goldberger et al., 1963; Venetianer and Straub, 1963), KDEL-containing proteins (detected with an antibody recognizing the KDEL ER retention signal), and Lava lamp (Lva, a marker for Golgi bodies)(Sisson et al., 2000). Each of these marker proteins is present within the actin cages, but with different degrees of enrichment. Puncta of Lva staining can be found inside the cages, but only rarely (Fig. 4M). PDI::GFP (Fig. 4N) and KDEL immunostaining (data not shown) is similar both within and outside of the cages. Finally, BiP is significantly concentrated in the larger actin cages, and is present but not highly concentrated in the smaller actin cages (Fig. 4O).
In their analysis of tral mutants (Wilhelm et al., 2005), characterized the distribution of Sar1::GFP. Sar1 is a component of ER exit sites (Kuge et al., 1994), and Sar1::GFP is therefore thought to serve as a marker for that compartment of the secretory apparatus. Sar1::GFP is normally present throughout the nurse cells and oocyte in many small puncta (Wilhelm et al., 2005); Fig. 4P). In tral mutants Sar1::GFP appears in abnormally large puncta in both nurse cells and the oocyte ((Wilhelm et al., 2005); Fig. 4Q). Notably, the distribution of Sar1::GFP in BicC mutants is indistinguishable from wild type (Fig. 4R). Thus there is no correlation between mislocalization of Grk to actin cages and abnormal distribution of Sar1::GFP.
A feature shared by tral and BicC mutants is the dumpless phenotype (Fig. 5A–C), in which the normal dumping of nurse cell contents into the oocyte during stages 11–13 of oogenesis is partially or completely inhibited. As a consequence, many stage 14 eggs are smaller than wild type with shorter and broader dorsal appendages. The dumpless phenotype often occurs because nurse cell nuclei enter and block the ring canals. This type of defect is typical of mutants of the chickadee, quail and singed genes, all of which encode proteins with defined biochemical roles in actin dynamics. In each case the mutants fail to form the network of actin cables that normally restrain the nuclei (Cooley et al., 1992; Cant et al., 1994; Mahajan-Miklos and Cooley, 1994). Similarly, both tral and BicC mutant egg chambers often have nuclei that have entered the ring canals, with the phenotype being more severe for BicC (Fig. 5E,F). The BicC mutant egg chambers with unrestrained nuclei lack the required actin cables (Fig. 5F), while in tral mutants actin cables are present, but they are often abnormal in their distribution and appearance (Fig. 5E).
The actin cages of BicC and tral mutants are found in the oocyte. Structures with a similar appearance can be detected in the nurse cells of the mutants, but they are very rare and exhibit a fundamental difference that suggests a different origin: all of the nurse cell actin structures stain rapidly with FM1-43, while none of the actin cages in the oocyte have this property. FM1-43 is a styryl dye often used to monitor exocytosis and endocytosis (Betz et al., 1996). The dye partitions readily but reversibly into membranes, and fluoresces intensely only when in the membrane. Importantly, FM1-43 cannot cross the membrane. Consequently, rapid acquisition of FM1-43 fluorescence by internal structures within a cell can occur only via direct membrane connections. Therefore, the labeling of the nurse cell structures with FM1-43 suggests that they are associated with an invagination of the plasma membrane. Indeed, in all cases examined we detected FM1-43-labeled tendrils extending from the nurse cell boundary to the internal structure, supporting this view (Sup. Fig. 4).
Not all actin-based structures of tral and BicC mutants are noticeably altered. The gross morphological organization of the egg chamber is maintained. Moreover, the ring canals appear intact, with Kelch (Kel) localized normally to the ring canal inner rim and no detectable Kel in the actin cages (Fig. 5G–I).
To determine if abnormalities in actin microfilament function or assembly is a phenotype common to mutants of other genes encoding sponge body associated proteins, we examined egg chambers from cup and me31B mutants. cup mutants have a dumpless phenotype with nuclei blocking ring canals, but no actin cages form (data not shown). me31B mutants also lack actin cages. Because me31B mutants arrest oogenesis before nurse cell dumping occurs, we cannot determine if that process is defective. In sum, we conclude that the syndrome of actin-related defects in tral and BicC mutants is not a general property arising from inactivation of sponge body components.
BicC and tral mutants share a variety of defects in mRNA localization, microtubule distribution within the oocyte, and actin microfilament organization. The extensive similarities suggest that BicC and tral act together or in a common pathway. This view is strongly supported by the genetic interaction between the mutants. BicC protein is a known post-transcriptional regulator. It binds directly to RNA and recruits the CCR4 deadenylase complex, which shortens the poly(A) tail and thus inhibits translation (Chicoine et al., 2007). Tral protein also has properties consistent with a role as a post-transcriptional regulator, and can be coimmunoprecipitated with at least two mRNAs, sar1 and sec13 (Wilhelm et al., 2005). However, the details of its action are less well understood. Tral is a member of the Scd6 family of proteins, with an Sm domain at its amino terminus and a FDF domain in its central region (Anantharaman and Aravind, 2004; Wilhelm et al., 2005). This organization suggests two ways in which Tral might associate with RNAs. First, by analogy to many other Sm-containing proteins, the Sm domain could mediate formation of hetero- or homomeric rings which bind directly to RNA (Toro et al., 2001; Arluison et al., 2004; Khusial et al., 2005). This option seems unlikely, as Tral does not associate with the Sm domain-containing proteins Lsm1, Lsm4, and Lsm7, and thus may not act as an alternative subunit in an Lsm complex. Furthermore, Tral does not self-associate and therefore does not form a homomeric ring (Tritschler et al., 2008). A second option is that Tral is part of RNP complexes in which other proteins mediate binding to specific mRNAs. The binding properties of Tral are consistent with this option. The Tral Sm domain mediates protein interactions with Dcp1 and Cup, and the FDF domain binds Me31B (Tritschler et al., 2008). Thus, Tral could act as a scaffold for the assembly of RNA protein complexes, perhaps promoting the association of BicC with other proteins. We did not detect coimmunoprecipitation of BicC and Tral (data not shown), but an interaction may not have been preserved under the conditions used or the proteins may only be transiently associated.
BicC and tral mutants disrupt the localization of several different mRNAs, particularly those that are positioned at the anterior of the oocyte. For two of the mRNAs, htsN4 and bcd, the usual concentration of the mRNAs at the junction between anterior and lateral cortex is lost. Instead, the mRNAs are spread out across most or all of the anterior. In the case of grk mRNA, it is an elaboration of the anterior localization that is lost. The grk mRNA is normally positioned along both lateral and anterior surfaces of the oocyte adjacent to the oocyte nucleus, but in the mutants the mRNA is predominantly along only the anterior. Given the general requirement for microtubules in many cases of mRNA localization (Bashirullah et al., 1998), and the more specific demonstration that grk mRNA moves along microtubules (Delanoue et al., 2007), it would not be surprising if the altered microtubule distribution seen in both BicC and tral mutants underlies the mRNA localization defects.
In tral and BicC mutants the sequestration of Grk in the actin cages must obstruct signaling to the follicle cells, thereby contributing to the ventralization phenotype. Because grk mRNA is inferred to be absent from the cages, the Grk protein must traffic to that site. (Wilhelm et al., 2005) have proposed that, for tral mutants, misregulation of transcripts encoding ER exit site components is the cause underlying mislocalization of Grk. A key question is whether formation of the actin cages is simply a secondary consequence of a more primary defect in ER exit site function. Several types of evidence argue against this possibility. First, when secretion of Grk is inhibited in other mutants, as in sec5 and cni, no actin cages form. Thus, either secretion is not required for actin cage formation, or the cages only form when a specific step in secretion is blocked. Second, not all sites of inappropriate protein trafficking in tral mutants coincide with actin cages. The vitellogenin receptor also appears in abnormal foci in tral mutant oocytes, yet these foci lack actin (Wilhelm et al., 2005). Third, BicC mutants have a very strong actin cage phenotype, more extreme than for tral mutants. Nevertheless, BicC mutants do not have the abnormal pattern of Sar1::GFP distribution of tral mutants that provides evidence of an ER exit site defect. Moreover, genomic analysis of mRNAs associated with BicC showed no substantial enrichment of Sar1 mRNA, while a validated target was very highly enriched (Chicoine et al., 2007). We conclude, then, that formation of the actin cages is not a consequence of a general secretion defect. Although tral mutants have an ER exit site abnormality, there is no evidence of an equivalent defect in BicC mutants.
We propose that the simplest explanation of the data is that formation of the actin cages near the anterior of the oocyte is the cause of the subsequent sequestration of Grk in the cages. In this model the secretory pathway could function relatively normally, up to the point where proteins become trapped in the cages. Whether Grk and other secreted proteins traffic to the cages, or to another destination, would be determined stochastically. However, with grk mRNA positioned almost exclusively at the anterior in these mutants, and very little along the lateral region of the oocyte, much of the Grk protein would be synthesized near the actin cages and trafficking to the cages would be enhanced. This model does not rule out a contribution from defective ER exit sites to Grk mislocalization in tral mutants, but accommodates the apparent absence of such a defect in BicC mutants.
The overall amount of Grk appears relatively normal up to stage 9 in both tral and BicC mutants, even though some Grk is partitioned into actin cages. However, at stage 10 the tral mutants now display a substantial reduction in Grk levels with no obvious change in grk mRNA levels or anterodorsal localization. Because BicC mutants lack this phenotype, it cannot be an indirect consequence of the multiple defects shared by BicC and tral mutants. Thus tral could play a more direct role in translation of grk mRNA, or it could act in controlling the expression of a factor that directly regulates grk mRNA.
In both BicC and tral mutants the actin cages are present only near the anterior of the oocyte, appearing first at stage 7 or 8 of oogenesis. Some of the actin cages are intimately associated with anterior cortical actin, while others are positioned nearby but appear to have no connections or contact with the cortex. The organization of the oocyte microfilament network at the nurse cell/oocyte boundary (i.e. the anterior of the oocyte) is also disrupted. In wild type these actin structures are generally of a uniform thickness, but can be irregular in tral and BicC mutants (Fig. 5G–I). The restricted distribution of the cages and the alteration of anterior cortical actin points to dysfunction of a specialized microfilament network at the anterior. One actin-based structure of the oocyte that is present only at its anterior is ring canals. However, several observations suggest that the cages are not specifically related to ring canals: ring canals are also present in the nurse cells, where actin cages appear very rarely or not at all; the cages are not specifically associated with ring canals in the oocyte; and the cages do not have the same composition as ring canals, as they lack the ring canal component Kel (Fig. 5G–I).
The possibility of another actin specialization at the anterior of the oocyte is suggested by the distribution of the htsN4 mRNA. This mRNA is one of several alternatively spliced products of the hts gene, which encodes a protein with significant similarity to Adducin. Adducin promotes the formation of spectrin/actin networks on membranes (Gardner and Bennett, 1987). The htsN4 mRNA is positioned at the anterior of the oocyte, where it is largely restricted to the junctions of anterior and lateral cortex (Yue and Spradling, 1992; Whittaker et al., 1999). This distribution suggests a special requirement for Hts at those sites, which could indicate some specialization of an actin based structure. Both Spectrin and Hts are associated with the actin cages (Fig. 1J–L,P–R), as would be consistent with a role for Hts in their formation.
Two patterns of mislocalization of htsN4 mRNA have been described. As reported here, in BicC and tral mutants the mRNA is retained at the anterior but fails to be restricted to the junctions of anterior and lateral cortex. Localization is more extensively disrupted in swallow (swa) mutants, with the htsN4 mRNA becoming dispersed throughout the oocyte (Ding et al., 1993). It is unlikely that the swa phenotype is a more severe version of that seen in the BicC and tral mutants, as htsN4 mislocalization appears later in the swa mutant. Furthermore, we did not detect a genetic interaction between swa and BicC (Table 1).
Each type of disruption of htsN4 mRNA localization is accompanied by the appearance of ectopic actin structures in the oocyte. Notably, the distribution of the novel actin structures parallels the altered distribution of htsN4 mRNA. For swa mutants, in which htsN4 mRNA is dispersed throughout the oocyte, the ectopic actin structures adopt a similar distribution and are present throughout the oocyte (Ding et al., 1993; Meng and Stephenson, 2002). In BicC and tral mutants the improperly localized htsN4 mRNA remains concentrated at the anterior of the oocyte, and this is where the actin cages form. The appearance of ectopic actin structures in swa mutants has been suggested to be a consequence of the htsN4 mislocalization (Zaccai and Lipshitz, 1996; Meng and Stephenson, 2002), and the same explanation could apply for the BicC and tral mutants. There is no indication that either BicC or Tral binds to htsN4 mRNA: we were unable to detect htsN4 mRNA in Tral coimmunoprecipitates (data not shown), and immunoprecipitation analysis of mRNAs bound to BicC did not identify htsN4 as a target (Chicoine et al., 2007). However, the alterations of microtubule and microfilament organization in BicC and tral mutants would accommodate an indirect effect on mRNA localization.
We thank Lynn Cooley, Trent Munro, Akira Nakamura, Mani Ramaswami, Trudi Schüpbach, Jim Wilhelm, and the Bloomington stock center for fly stocks, Bill Chia for generating the g89 and g140 GFP trap lines, and Mary Lilly, Paul Lasko, John Sisson and Jim Wilhelm for antibodies. We thank the Developmental Studies Hybridoma Bank (developed under the auspices of the National Institute for Child Health and Human Development and maintained by The University of Iowa, Department of Biological Sciences) for the following monoclonal antibodies: E7 (M. Klymkowsky), 8C3 (S. Benzer), 1B1 (H. Lipshitz), 1D12 (T. Schüpbach), 4H8 (P. Schedl), 1D9 (S. Parkhurst), 3A9 (R. Dubreuil), htsRC, kel1B and 6B9 (L. Cooley). The Berkley Drosophila Genome Project provided the plasmid containing the tral cDNA. Thanks to members of the Macdonald lab for discussions and review of the manuscript, and to the anonymous reviewers for valuable suggestions. This work was supported by grants GM42612 and GM54409 from the NIH.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.