Intercellular protein movement in syncytial Drosophila follicle cells

4 downloads 75448 Views 2MB Size Report
1. Journal of. Cell. Science. JCS ePress online publication date 1 December 2011 ...... Images were processed using iVision and/or Adobe Photoshop. CS4.
JCS ePress online publication date 1 December 2011 Research Article

1

Intercellular protein movement in syncytial Drosophila follicle cells Stephanie J. Airoldi1,*, Peter F. McLean1,*, Yuko Shimada1,` and Lynn Cooley1,2,3,§ 1

Department of Genetics and 2Department of Cell Biology, Yale School of Medicine, 333 Cedar St., New Haven, CT 06520, USA Department of Molecular, Cellular and Developmental Biology, Yale University, 260 Whitney Ave., New Haven, CT 06510, USA

3

*These authors contributed equally to this work ` Present address: University of Tsukuba, Tennoudai 1-1-1, Tsukuba, Ibaraki 305-8572, Japan § Author for correspondence ([email protected])

Journal of Cell Science

Accepted 18 July 2011 Journal of Cell Science 124, 1–10  2011. Published by The Company of Biologists Ltd doi: 10.1242/jcs.090456

Summary Ring canals connecting Drosophila germline, follicle and imaginal disc cells provide direct contact of cytoplasm between cells. To date, little is known about the formation, structure, or function of the somatic ring canals present in follicle and imaginal disc cells. Here, we show by confocal and electron microscopy that Pavarotti kinesin-like protein and Visgun are stable components of somatic ring canals. Using live-cell confocal microscopy, we show that somatic ring canals form from the stabilization of mitotic cleavage furrows. In contrast to germline cells, syncytial follicle cells do not divide synchronously, are not maximally branched and their ring canals do not increase in size during egg chamber development. We show for the first time that somatic ring canals permit exchange of cytoplasmic proteins between follicle cells. These results provide insight into the composition and function of ring canals in somatic cells, implying a broader functional significance for syncytial organization of cells outside the germline. Key words: Drosophila oogenesis, Follicle cell, Ring canal

Introduction The final step of cytokinesis is abscission, in which the thin intercellular bridge that remains after constriction of a cleavage furrow is severed, resulting in two unconnected daughter cells. An alternative to abscission is the reinforcement of mitotic intercellular bridges, which leads to stable and persistent connections between sibling cells. Stable intercellular bridges derived from mitotic cleavage furrows are nearly always found in developing gametes in animals, including humans. Abundant genetic evidence in Drosophila and mice shows that germline intercellular bridges, or ring canals, are required for fertility (Brill et al., 2000; Greenbaum et al., 2006; Hime et al., 1996; Robinson and Cooley, 1996). Ring canals allow proteins, mRNAs and organelles to pass between cells, facilitating synchronous mitotic and meiotic divisions in males, and oocyte growth in Drosophila females. Despite topological similarities, the behavior and components of germline ring canals differ between males and females. For example, female germline ring canals in Drosophila expand during development to reach nearly 10 mm in diameter, whereas male germline ring canals (,1 mm) and somatic ring canals (0.25 mm) do not expand (Hime et al., 1996; Tilney et al., 1996; Woodruff and Tilney, 1998). Female germline ring canals are stabilized with a robust actin cytoskeleton whereas male ring canals contain a septin-based cytoskeleton (Hime et al., 1996; Robinson and Cooley, 1996). Thus, there are at least two ways cells can stabilize ring canals. In contrast to germline ring canals, intercellular bridges connecting somatic cells are poorly understood. Somatic ring canals were detected decades ago by electron microscopy of insect follicle cells that encapsulate germline cells in egg

chambers (Fiil, 1978; Giorgi, 1978; Meola et al., 1977; Ramamurty and Engels, 1977), and of imaginal discs present in larvae (Poodry, 1970). They appear to result from the stabilization of mitotic intercellular bridges, as in the germline. The stability of follicle cell ring canals is evidenced by their persistence throughout egg chamber development. Drosophila follicle cells cease mitosis at the end of stage 6, after which the egg chamber continues to develop for another 40 hours (Lin and Spradling, 1993). Ring canals connecting follicle cells persist until late in development, showing that they are stable long after mitoses are completed. Importantly, somatic ring canals provide a bridge between cells over 100 times larger than gap junctions (250 nm vs 1.5 nm), which could allow passage of metabolites, proteins and even small vesicles. Until recently, only one molecular component of somatic ring canals was positively identified at the level of electron microscopy (EM): F-actin in follicle cell ring canals (Woodruff and Tilney, 1998). Several other proteins have been reported in follicle cell or imaginal disc ring canals based on confocal microscopy: Pavarotti kinesin-like protein (Pav-KLP) (Minestrini et al., 2002), Anillin (de Cuevas and Spradling, 1998), Mucin-D (Kramerova and Kramerov, 1999), Visgun (Vsg) (Buszczak et al., 2007; Nystul and Spradling, 2007), Nasrat and Polehole (Jimenez et al., 2002). Recent work confirmed the presence of Pav-KLP and Anillin in Drosophila somatic ring canals, as well as another protein called Cindr, using a combination of immunoEM and confocal microscopy (Haglund et al., 2010). The function of ring canals in somatic cells remains unknown. To determine whether somatic ring canals allow the exchange of cytoplasm between cells, we followed the behavior of several

2

Journal of Cell Science 124 (0)

proteins, and report evidence for selective movement between connected follicle cells. Interestingly, despite intercellular movement of proteins, follicle cell mitotic cell cycles are not fully synchronized among connected cells, as is the case in germline syncytia. We also refine the list of ring canal components by showing that Drosophila Vsg, but not Nasrat or Polehole, is located at follicle cell and imaginal disc ring canals. Although ring canals are ubiquitous in main-body follicle cells, they are absent from stalk cells and polar cells, and appear to disintegrate in migrating border cells. The movement of protein between follicle cells through ring canals raises interesting questions about the importance of syncytial organization in development of follicle cells and the egg chamber. Results Somatic cell ring-canal proteins

Journal of Cell Science

A protein-trap allele of visgun revealed that GFP-tagged Vsg localizes to puncta on follicle cell membranes (Buszczak et al., 2007; Nystul and Spradling, 2007) in a pattern very similar to that of GFP::Pav-KLP (Minestrini et al., 2002) (Fig. 1A–D). In the

Fig. 1. Proteins in somatic ring canals. (A–D) GFP::Pav-KLP (A,B) and GFP::Vsg (C,D) localize to ring canals in the germline (A,C yellow arrowheads) and somatic follicle cells from region 2a of the germarium (A,C) through stage-10 egg chambers (B,D) (white arrows). (E,F) ImmunoEM of post-mitotic follicle cells expressing GFP::Pav-KLP (E) and GFP::Vsg (F) shows specific localization of these proteins to follicle cell ring canals (circles). (G) GFP::Vsg localizes to ubiquitous puncta in imaginal wing discs. (H,I) Higher magnification confirms Vsg (H) and PavKLP (I) puncta on cell membranes of wing disc cells (arrows). Scale bars: 10 mm (A–D,H,I), 0.5 mm (E,F), 50 mm (G).

germarium, both proteins localized to germline ring canals (Fig. 1A,C, yellow arrowheads) and to puncta on follicle cell membranes (Fig. 1A,C, white arrows). The punctate localization in follicle cells persisted throughout oogenesis, including in postmitotic follicle cells (Fig. 1B,D, white arrows). Additionally, GFP::Vsg localized to germline ring canals in testes (not shown). ImmunoEM analysis of GFP::Pav-KLP and GFP::Vsg flies revealed that Pav-KLP and Vsg specifically localized to follicle cell ring canals (Fig. 1E,F). Gold particles in GFP::Pav-KLP samples were located on the cytoplasmic face of ring canals (Fig. 1E), with half the samples also displaying particles in the cytoplasm that were not associated with ring canals. Some nonring-canal localization was expected, because this is also observed with fluorescence microscopy of GFP::Pav-KLP. In the case of Vsg, all ring-canal-associated gold particles were located extracellularly (n517) (Fig. 1F) or on the plasma membrane (n54), and we also observed some particles in the cytoplasm (n56). A negative control stained only with secondary antibody showed no labeling, suggesting that the cytoplasmic localization of GFP::Vsg represented specific labeling of protein in vesicles. The ring-canal localization patterns were consistent with published information on these proteins; Pav-KLP has an intracellular localization and Vsg is predicted to be a transmembrane glycoprotein with an extracellular N-terminus that bears the GFP tag. GFP::Vsg also localized to puncta on the plasma membranes of imaginal disc epithelia (Fig. 1G,H), similarly to Pav-KLP (Fig. 1I), Anillin and Cindr (Haglund et al., 2010), indicating that Vsg is also a component of imaginal disc ring canals. Nasrat and Polehole were reported to localize to follicle cell puncta thought to be ring canals (Jimenez et al., 2002). However, neither Nasrat nor Polehole colocalized with GFP::Pav-KLP (Fig. 2A,B), indicating that these proteins are not components of ring canals. The availability of GFP-tagged ring canal components allowed us to examine the distribution of ring canals in the apical–basal axis of follicle cells. GFP::Pav-KLP (Fig. 2C,D) and GFP::Vsg (data not shown) were present in the apical half of lateral plasma membranes, basal to adherens junctions (Fig. 2C) and mainly apical to gap junctions (Fig. 2D) (see also Haglund et al., 2010).

Fig. 2. Localization of putative ring canal proteins and ring canal position. (A,B) Neither Nasrat-Flu puncta (A, red arrows) nor Polehole-Flu puncta (B, red arrows) colocalize with GFP::Pav-KLP-marked ring canals (green arrows). (C,D) GFP::Pav-KLP-marked ring canals (green arrows) are located basal to (above) adherens junctions (C, red arrows) and largely apical to (below) gap junctions (D). Scale bars: 10 mm.

Journal of Cell Science

Syncytial Drosophila follicle cells

Fig. 3. Mitotic spindle orientation and cyst organization. (A,B) Schematic of branched and linear divisions in a syncytium. Asymmetric segregation of existing ring canals (green bars) in mitosis results in a branch in the syncytium (A), whereas a mitotic plane (dotted lines) that separates existing ring canals generates a linear cyst (B). (C–E) Egg chambers expressing GFP::Pav-KLP (green) were stained with DAPI (blue) and antibodies against centrosomin (Cnn) and Adducin (htsM) (red, marking centrosomes and membranes, respectively). Angles of ring canals relative to the cleavage plane were measured (illustrated in insets). (F) The positions of 79 ring canals (green) from 67 dividing cells do not show a significant bias relative to the cleavage plane. Measurements were grouped into 10 ˚ bins. (G) The observed distribution of ring canals in 796 follicle cells indicates an intermediate branching pattern. Models of linear and maximally branched 512-cell cysts (resulting from nine follicle cell cycles) and the 16-cell maximally branched pattern observed in the female germline are also shown. The unnumbered segments in the maximally branched 512-cell model represent cells having between five and nine ring canals and decrease geometrically in their proportion of the total. Scale bar: 5 mm.

Follicle cells form a branched network

During the four synchronous mitotic divisions of Drosophila germline cells, existing ring canals are segregated asymmetrically between the daughter cells. One daughter cell retains all existing ring canals, whereas the other is left with only the most recently formed ring canal (Fig. 3A), resulting in a maximally branched pattern of connected cells. Unbranched, linear chains of cells would result from divisions that divide the cell on a plane between pre-existing ring canals (Fig. 3B). Each type of lineage has a characteristic distribution of ring canals per cell. The percentage of cells having one, two and three ring canals in a maximally branched lineage was 50, 25 and 12.5%, respectively, whereas all but the two terminal cells in a linear chain would

3

have just two ring canals (Fig. 3G). The follicle cell epithelium develops from the daughters of two follicle stem cells that undergo at least eight cell cycles to produce lineages of hundreds of cells (Margolis and Spradling, 1995). The majority of egg chambers are encompassed by two such lineages (Nystul and Spradling, 2010). To investigate the organization of follicle cell lineages, we examined mitotic spindle orientation with respect to the position of pre-existing ring canals. GFP::Pav-KLP egg chambers were stained with antibody against centrosomin (to mark centrosomes) and DAPI (to mark chromosomes). For each dividing cell with at least one pre-existing ring canal, we measured the angle between an existing ring canal and the cleavage plane (Fig. 3C–E). In total, 179 ring canals in 103 dividing cells were measured. However, many ring canals could not be confidently assigned to the dividing cell because they resided at the corners of three cells, and so were removed from the analysis. The remaining 79 ring canals in 67 dividing cells were distributed over the entire 90 ˚ range with no apparent bias for any position (Fig. 3F), indicating that the position of a pre-existing ring canal does not influence the orientation of the cleavage furrow. This suggests that in contrast to the female germline, the segregation of ring canals in dividing follicle cells is not tightly regulated and would probably produce an intermediately branched network. The number of ring canals per follicle cell supports the conclusion of an intermediate branching pattern. The percentages of follicle cells (n5796) with one, two, or three ring canals was 22, 44, and 27%, respectively, with most of the remaining 7% having four ring canals. Only 3 of 796 cells had zero ring canals. This distribution of ring canals is markedly different from the distributions predicted from linear or maximally branched lineages (Fig. 3G). Cleavage furrows give rise to follicle cell ring canals

Pav-KLP, Anillin and Cindr are enriched in mitotic cleavage furrows and function during mitotic cleavage furrow ingression (Haglund et al., 2010; Minestrini et al., 2003; Straight et al., 2005). The presence of these proteins in somatic ring canals is strong evidence that the ring canals are derived from cleavage furrows that arrested before the completion of cytokinesis, as is the case in germline cells. We used GFP::Pav-KLP to examine directly the formation of follicle cell ring canals. Time-lapse microscopy of GFP::Pav-KLP egg chambers revealed that follicle cell cleavage furrows constricted to form puncta identical in appearance to ring canals in post-mitotic cells (Fig. 4A and supplementary material Movie 1), further supporting the conclusion that somatic ring canals are derived from cleavage furrow arrest. Time-lapse microscopy showed no movement of established ring canals in either mitotic or non-mitotic cells, but ring canals forming in mitotic cells often shifted laterally between the resultant cells at the conclusion of mitosis (Fig. 4B and supplementary material Movie 2). Consistent with this observation, our fixed tissue data revealed that 64% of ring canals in post-mitotic follicle cells were at three-cell corners (223 of 350 ring canals) (Fig. 5A, arrows). Interestingly, we also observed that the ingressing cleavage furrow of dividing cells occasionally incorporated an existing ring canal (Fig. 4C and supplementary material Movie 3). A similar process has been observed in worms, where ring canals connecting cystocytes to a central, anucleate cytoplasmic core are bisected into two ring canals by the cleavage furrows during mitosis (Swiatek et al.,

Journal of Cell Science

4

Journal of Cell Science 124 (0)

Fig. 4. Time-lapse images showing formation of follicle cell ring canals. Egg chambers expressing Cherry::Histone2Av and GFP::Pav-KLP were cultured in vitro and imaged to capture ring canal formation (movies are provided in the supplementary material). Diagrams below the images show the positions of chromosomes and cleavage furrows. White arrowheads denote GFP::Pav-KLP accumulation at the cleavage furrow and subsequent ring canal. Time is indicated as minutes:seconds. (A) The nascent ring canal remained centered between daughter nuclei. (B) The nascent ring canal moved off-center. (C) An existing ring canal (yellow arrowhead) was drawn in by the constricting cleavage furrow and resolved into one object (T510:30).

2009). We observed that the cleavage furrow of a dividing follicle cell and the existing ring canal resolved into a single object, allowing the possibility that a structure connecting three cells would remain. Consistent with this, we observed tripartite ring canals by EM (Fig. 5B), as also reported previously (Tilney et al., 1996). However, our analysis of spindle orientation in dividing follicle cells found that only 10% (7 of 67) of cleavage furrows occurred within 10 ˚ of an existing ring canal (Fig. 3F), suggesting that only a fraction of the 64% of ring canals that

Fig. 5. Tripartite ring canals. (A) Confocal image of a stage-10 egg chamber expressing GFP::Pav-KLP and stained with htsM antibody to visualize membranes. Ring canals were located at two-cell or three-cell junctions (arrows). (B) ImmunoEM image of a tripartite ring canal that connects the cytoplasm of three cells (numbered). Membranes are indicated with dashed red lines. Scale bars: 10 mm (A), 0.5 mm (B).

appear to be localized at the corner of three cells could be tripartite. Ring canals persist in main-body follicle cells

To determine whether all follicle cell divisions produce a stable ring canal, we compared the total numbers of ring canals (obtained by counting GFP::Pav signals) to the number of mainbody follicle cells (obtained by counting DAPI-stained nuclei) encompassing the egg chambers. The youngest egg chambers outside of the germarium (n56) had between 35 and 93 cells (mean, 68) with 20–69 ring canals (mean, 46), or 70% of the expected number if all cell divisions resulted in a ring canal. In mitotic divisions outside the germarium, the prevalence of ring canals in egg chambers fit a linear trend with a slope of 0.887 and y-intercept of 211.3 (Fig. 6A), indicating that ,89% of cell divisions were represented by stable ring canals. The negative yintercept reflects the deficit of ring canals in egg chambers just emerged from the germarium. Thus, a greater percentage of divisions resulted in ring canals after egg chambers emerged from the germarium. Interestingly, we counted an average of 1077 follicle cells in stage 8 and stage 9 egg chambers that had completed follicle cell divisions. This is higher than the average of 651 reported previously (Margolis and Spradling, 1995), but agrees with older reports (King, 1970). Based on two lineages per egg chamber (Nystul and Spradling, 2010), each lineage contains over 500 cells produced by nine cell division cycles. We examined other populations of follicle cells for the presence of ring canals: stalk cells connecting egg chambers, polar cells at the anterior and posterior poles of egg chambers, and border cells that delaminate from the anterior pole and migrate between nurse cells to the oocyte. Stalk cells and polar cells are specified in the germarium, and border cells are specified and migrate during stage 9. We often observed ring canals among the cells of a developing stalk during the transition

Journal of Cell Science

Syncytial Drosophila follicle cells

Fig. 6. Prevalence of ring canals in follicle cells. (A) A comparison of the total number of ring canals and cells in stage 2–9 egg chambers. The data support a linear trend (y50.887x211.3; R250.998), indicating that 88.7% of follicle cell divisions outside the germarium form a stable ring canal. Divisions within the germarium exhibit a larger deficit of ring canals resulting in a shift in the y-intercept. The dotted line (slope51) represents the maximum possible number of ring canals. (B–D) Not all follicle cell subtypes have ring canals. Stalk cells (B,B9, arrowheads) and polar cells (C, highlighted) do not have ring canals. Polar cells were identified by elevated nuclear localization of GFP::Pav-KLP. Before their migration, presumptive border cells (C, adjacent to polar cells) do have ring canals. Once migration has begun, border cells contain fewer, smaller puncta that presumably represent ring canal remnants (D, arrowheads). Nurse cell nuclei and germline ring canals (arrows) are visible around the border cells. Scale bars: 10 mm.

out of the germarium (not shown), but interestingly, once the stalk was resolved to a chain of single cells it was completely devoid of ring canals (Fig. 6B,B9). Similarly, we inspected pairs of anterior and posterior polar cells in stage 5–8 egg chambers and found no evidence of ring canals (Fig. 6C). Precursor border cells did have ring canals and were indistinguishable from other main-body follicle cells by GFP::Pav-KLP and DAPI staining prior to the start of their migration during stage 9 (Fig. 6C). However, border cells that had delaminated from the follicular epithelium contained smaller, seemingly fragmented puncta that could be internalized ring canal remnants (Fig. 6D). Mitotic coordination in the follicular epithelium

The presence of stable ring canals between follicle cells introduces the possibility of cytoplasm exchange between cells that could coordinate their behavior. We sought evidence of cellular coordination by looking for synchronized mitoses using anti-PH3 antibodies to detect condensed chromosomes (Fig. 7A). If entry into mitosis is a cell-autonomous event, follicle cell cycles are not synchronized, and the amount of time spent in mitosis is known, we can calculate an expected frequency of mitotic cells occurring independently or in clusters of cells. From the live imaging we performed using H2A::RFP, we estimated that follicle cell mitosis lasts about 24 minutes, whereas the entire cell cycle requires approximately 9.6 hours (Margolis and Spradling, 1995). Therefore, the probability of finding a particular follicle cell in mitosis is 24/576 minutes, or 0.042. Follicle cells exist in a hexagonal array in which each cell contacts six neighboring cells. To calculate the probability of

5

Fig. 7. Ring canals in follicle cell mitosis. (A) Progression through mitosis is revealed by PH3 staining. (B–D) Mitotic cells in GFP::Pav-KLP egg chambers were evaluated for the presence of connecting ring canals. Membranes are marked with htsM. (B) A single mitotic cell is connected by ring canals (arrows) to non-mitotic cells. Clusters of mitotic cells with two (C) or three (D) mitotic cells that are connected by a ring canals (arrowheads) are also connected by ring canals to other, non-mitotic cells. Pro, prophase; Prm, prometaphase; Met, metaphase; Ana, anaphase; Tel, telophase; Cyt, cytokinesis. Scale bar: 10 mm.

finding two adjacent mitotic cells by chance, we first determined the probability that a second mitotic cell is not an immediate neighbor. This value is equal to the number of available nonneighbor cells divided by the total (T) number of cells. When considering N mitotic cells, the probability that they are all separated by at least one non-mitotic cell is [T2(7*N)]/T. By estimating that N50.042T, the probability that all mitotic cells are separated from each other simplifies to (127)*0.042, or 0.71. Conversely, the probability of finding neighboring mitotic cells is 0.29. We found a significantly higher number of adjacent mitotic follicle cells. Out of 253 PH3-positive nuclei, 133 (53%) were adjacent to one or more mitotic nucleus in 53 clusters of two or more cells. The majority of clusters contained two cells (n536), but mitotic clusters also included three (n510), four (n54) or five (n53) cells. The difference between the observed and expected numbers of adjacent mitotic cells was significant (Pearson’s chi-squared test, P,0.001), revealing that entry into mitosis is coordinated to some extent. In a subset of these cells, we could score the presence of ring canals using GFP::Pav-KLP. Nearly all of the single PH3-positive cells had ring canals connecting them to PH3-negative cells (41 of 42) (Fig. 7B). We also observed seven cases of two PH3-positive cells bridged by a ring canal, but all seven had another ring canal to a third, PH3negative cell (Fig. 7C). We found two examples of three adjacent PH3-positive cells connected by ring canals; in each case there was at least one ring canal connection to a PH3-negative cell (Fig. 7D). These results indicate that although some coordination of mitosis exists among small groups of follicle cells, it is not sufficient to maintain synchrony in a whole follicle cell lineage. Intercellular movement of proteins through follicle cell ring canals

Incomplete mitotic coordination in follicle cells connected by ring canals suggested very limited transit of cytoplasmic contents. However, the spread of Lucifer Yellow dye was documented in follicle cells (Woodruff and Tilney, 1998), showing that small molecules can pass between cells. We

Journal of Cell Science 124 (0)

Journal of Cell Science

6

Fig. 8. Movement through follicle cell ring canals. (A) Single-cell activation of PA-GFP reveals exchange of protein between follicle cells. (B–E) FLIP analysis demonstrates that some, but not all, proteins are exchanged between follicle cells. Multiple cells in each field were repeatedly bleached over 45–90 minutes. Red circles denote bleached cells; yellow flags indicate adjacent cells with significant loss of fluorescence. Oda, but not RpL30 or Yps, shows movement between cells. (C) Co-expression of GFP::Oda and GFP::Pav-KLP demonstrates that GFP movement occurs only between cells connected by ring canals (arrowheads). Asterisks highlight cells exhibiting measureable loss of fluorescence. Here, GFP::Oda is heterozygous and produces lower GFP expression compared with that in B. Minor differences in the position and intensity of ring canals between C and C9 are due to changes in the focal plane and are not significant. (F) UASpYps::mRFP displays mosaic expression in follicle cells and is expressed at different levels in cells connected by ring canals (arrows). Scale bars: 10 mm.

5 minutes. These results demonstrated that a soluble protein could move between cells. We next sought to determine whether endogenous Drosophila proteins were exchanged between interconnected follicle cells. For this, we turned to fluorescence loss in photobleaching (FLIP) experiments in which individual follicle cells in stage 10 egg chambers were repeatedly bleached over the course of 40– 50 minutes, and adjacent cells were monitored for loss of fluorescence. To validate the technique, we first conducted FLIP experiments on tissue expressing soluble GFP and observed loss of fluorescence in neighboring cells (data not shown), complementary to our PA-GFP results. We then carried out FLIP analysis of several GFP protein traps (Kelso et al., 2004; Quinones-Coello et al., 2007). Similarly to our PA-GFP results, we observed loss of fluorescence in neighboring cells expressing GFP-tagged ornithine decarboxylase antizyme (GFP::Oda) (Fig. 8B) and calmodulin (GFP::Cam) (data not shown). Importantly, we observed loss of fluorescence in one or two neighboring cells, which is consistent with our earlier observations. Analysis of images captured immediately after each bleach cycle confirmed that the region being bleached was confined to the target cell; therefore, loss of fluorescence in adjacent cells represents exchange of protein between connected cells. In addition, we conducted FLIP analysis of GFP::Oda in the presence of the ring canal marker GFP::Pav-KLP and observed a measurable loss of fluorescence only in cells that were connected by a ring canal (Fig. 8C). By contrast, FLIP analysis of proteins in macromolecular complexes revealed no movement between cells. Repeated bleaching of single cells expressing GFP-tagged ribosome protein subunits GFP::RpL30 (Fig. 8D) and GFP::Sop (data not shown), or translation factors GFP::eIF-4E and GFP::eRF1 (data not shown) showed no loss of fluorescence in neighboring cells. Similarly, FLIP experiments with the RNA-binding protein Ypsilon schachtel (Yps), a component of the oskar RNP (Armstrong et al., 2006; Wilhelm et al., 2000), revealed no evidence for its movement between cells (Fig. 8E). In a complementary approach, we looked for evidence of Yps mobility by taking advantage of ectopic expression. Follicle cells expressing genes under the control of the Gal4/UAS system display mosaic patterns of expression (Skora and Spradling, 2010), providing an opportunity to examine neighboring cells with differing levels of protein expression. We found many examples of cells connected by ring canals that had different levels of expression and, consistent with FLIP of GFP::Yps, we found no evidence of intercellular movement of UASpYps::mRFP driven by tubulin-Gal4 (Fig. 8F). These data show that some, but not all, cytoplasmic proteins can freely pass through ring canals between follicle cells. Discussion Somatic cell ring-canal components

tested whether a photoactivatable variant of GFP (PA-GFP) (Patterson and Lippincott-Schwartz, 2002) activated in individual follicle cells of stage 10 egg chambers could move to neighboring cells. In 50 of 50 successful activation events, PA-GFP unambiguously spread into neighboring, non-activated cells (Fig. 8A). In these experiments, activated PA-GFP typically spread into only one or two of the six neighboring cells, and movement into connected cells was clearly visible within

Similarly to germline ring canals in Drosophila egg chambers, follicle cell ring canals are supported by filamentous actin lining the lumen (Woodruff and Tilney, 1998). In germline ring canals, the highly dynamic actin cytoskeleton mediates ring canal expansion during egg chamber development (Hudson and Cooley, 2002). A 200 nm-thick mesh of F-actin bundles accumulates at the plasma membrane of germline ring canals, which reach an overall diameter of nearly 10 mm (Tilney et al., 1996). By contrast, follicle cell ring canals do not expand during

Journal of Cell Science

Syncytial Drosophila follicle cells development, remaining about 250 nm in diameter with a monolayer of actin filaments (Woodruff and Tilney, 1998). The persistence of cleavage furrow proteins in ring canals is a common feature of both germline and somatic ring canals. Many of the follicle cell ring-canal components identified to date are proteins with known roles in cytokinesis. Anillin and Pav-KLP accumulate in cleavage furrows as they begin to constrict, and remain associated with ring canals, as we observed in images of live dividing follicle cells (supplementary material Movies 1–3). Mutations in scraps (the gene encoding Anillin) and pav cause defects in cytokinesis and result in multinucleate cells (Adams et al., 1998; Field et al., 2005). In addition, an Anillin-binding protein called Cindr was recently identified as another component of both cleavage furrows and somatic ring canals. In germline ring canals, Anillin, Pav-KLP and Cindr are present initially in ring canals, but only Pav-KLP persists throughout oogenesis (Haglund et al., 2010). The recruitment of the robust actin cytoskeleton to ring canals in Drosophila female germline cells might displace cleavage furrow proteins. Our data allow us to refine the list of somatic ring canal proteins; we can add Vsg, and eliminate Nasrat and Polehole. GFP::Vsg was reported to be present in germline ring canals and follicle cell puncta (Buszczak et al., 2007; Nystul and Spradling, 2007). We confirmed the localization of GFP::Vsg to follicle cell ring canals using immunoEM and colocalization studies. Vsg is a predicted sialomucin protein with a single transmembrane domain and multiple predicted extracellular O-linked glycosylation sites. Its localization to somatic ring canals is highly reminiscent of Mucin-D localization (Kramerova and Kramerov, 1999); however, we were unable to confirm this directly because Mucin-D antibodies are no longer available. Vsg (and Mucin-D) are also present in ring canals of larval imaginal disc and brains. Vsg protein bears predicted localization signals, including an N-terminal plasma membrane localization signal and a C-terminal vesicular trafficking sequence. Therefore, its specific localization to ring canals could indicate a role of vesicular trafficking in the formation of stable intercellular bridges. Previous work on Vsg revealed roles in promoting cell proliferation and embryonic development, though its specific function is unclear (Zhou et al., 2006). Its vertebrate homolog, endolyn, is targeted to endosomes and lysosomes, and functions in the maintenance of hematopoietic progenitors (Forde et al., 2007) and myoblast fusion (Bae et al., 2008). The highest degree of similarity between Vsg and Endolyn is in the transmembrane and cytoplasmic domains, including the C-terminal lysosomal sorting signals, suggesting a common underlying role for this family of sialomucins. Not all follicle cells have stable ring canals

Our data indicate that 70% of follicle cell divisions within the germarium result in a stable ring canal, and this increases to 89% of divisions outside the germarium. A possible explanation for the absence of ring canals is that a percentage of cell divisions complete cytokinesis and a ring canal is never formed. In light of our data, this would suggest that follicle cells within the germarium are more likely to complete abscission, possibly because these cells are still in the early stages of differentiation and still contain factors that promote normal cell division. Alternatively, the absence of ring canals that we observe might be a result of instability or destruction of otherwise normal ring

7

canals, or to the merging of existing ring canals as seen in Fig. 4C. If so, the higher rate of ring canal loss in the germarium could be attributed to the mechanical stresses of migration as the follicle cells move to encapsulate the germline cyst and reposition for stalk formation. Once outside the germarium, the constant but smaller rate of ring canal loss could be a result of shuffling within the epithelium, cell death or eventual resolution of cytokinesis. Specialized subpopulations of follicle cells lack ring canals. We did not observe ring canals in mature stalk cells; however, ring canals were observed between cells in the region of the forming stalk as egg chambers exit the germarium (not shown). Loss of ring canals during stalk formation could be due to mechanical disruption as cells intercalate to form a single chain of cells. A model of mechanical disruption of existing ring canals is further supported by our observations of border cells. Presumptive border cells in stage 8 egg chambers have normal ring canals, but migrating border cells contain fewer, smaller, and apparently fragmented GFP::Pav-KLP puncta. We also observed no ring canals in pairs of polar cells. This is not surprising, because the polar cell precursors are specified while in the germarium and do not continue to divide (Nystul and Spradling, 2010). Furthermore, the precursor population is reduced to only two cells through programmed cell death between the germarium and stage 5 (Besse and Pret, 2003), which would further isolate the polar cells. However, additional investigation is necessary to determine whether the polar cell precursors have ring canals, either between themselves or with other main-body follicle cells. Coordination of follicle cell mitosis

Given the presence of persistent intercellular bridges within follicle cell lineages, we investigated whether follicle cell mitotic divisions are synchronized, as they are in male and female germline cells. Our results clearly indicate that some level of synchrony exists between small groups of follicle cells because adjacent mitotic cells are significantly more frequent than predicted from a random distribution (0.53 vs 0.29, P,0.001). The coordination of mitosis could be carried out by a cell nonautonomous mechanism in which mitosis-promoting factors pass through ring canals and initiate mitosis in the next cell. However, nearly all mitotic cells had ring canals connecting them to nonmitotic cells, so any propagation of mitosis-promoting signals would be limited. Furthermore, mitotic divisions within follicle cell lineages consisting of hundreds of cells are far from fully synchronized, suggesting that any influence of neighbors on entry into mitosis is weak. Alternatively, a cell-autonomous cell cycle ‘timer’ could be responsible for the apparent coordination of mitosis between sibling cells, regardless of a connecting ring canal. In this scenario, two sibling cells would enter mitosis together because their cell cycles were synchronized at the conclusion of the previous division. Of note, we found that 47% of cells were in mitosis independently of all neighboring cells, which implies that if a cell-autonomous cell cycle timer controls entry into mitosis, the length of the overall cell cycle (9.6 hours) will vary by more than 24 minutes in nearly half of all cells. Our data are insufficient to distinguish between cytoplasm sharing and timer mechanisms, because the existence of ring canals between neighboring mitotic cells fits both models. The pattern of connections between cells with ring canals is determined by spindle orientation with respect to ring canals

8

Journal of Cell Science 124 (0)

from previous mitoses. In Drosophila germline mitoses, spindle orientation and thus ring canal inheritance is controlled by the fusome, a cytoplasmic organelle containing ER membranes that extends through each intercellular bridge (de Cuevas and Spradling, 1998). After four mitotic divisions, the 16 cells are maximally branched with the original two cells having four ring canals each and the youngest eight cells from the final division each having one ring canal. By contrast, the follicle cells, which do not have a fusome, appear to divide randomly with respect to pre-existing ring canals, and their pattern of connections is intermediate between fully branched and completely linear. Other mechanisms are likely to contribute to the pattern of follicle cell divisions, such as a limitation on how many direct connections an individual cell can have to hexagonally packed neighbors.

Journal of Cell Science

Selective movement through follicle cell ring canals

The presence in follicle cells of intercellular bridges large enough for the passage of small vesicles presents the possibility of extensive flow of information between these cells. However, many aspects of follicle cell biology strongly suggest a cellautonomous system. For example, small subsets of follicle cells pattern the oocyte during development (Boyle et al., 2010; Yakoby et al., 2008), an essential function that would be compromised by unrestricted spreading of localized signals to neighboring cells. Also, genetically mosaic patches of follicle cells appear to be reliably marked by expression of various reporters. Follicle cells are also known to have highly mosaic expression of many proteins, some of which clearly maintain very different levels of expression despite a connecting ring canal (Fig. 8F). As a result, we were quite surprised to discover evidence for robust and rapid exchange of PA-GFP between cells, in patterns consistent with syncytia of follicle cells connected by ring canals. Importantly, we confirmed this result by observing movement of tagged endogenous proteins by FLIP. In both sets of experiments with fluorescent proteins, we observed exchange of protein in one or two immediate neighbors of the target cell. Not only does this argue against nonspecific activation or bleaching, it is also consistent with the number of ring canals per cell that we observed in other experiments. Co-expression of GFP::Oda and GFP::Pav-KLP confirmed these results by demonstrating that protein movement occurs only between cells connected by a ring canal. Of note, our FLIP data also show that some proteins are not exchanged through ring canals, whereas others pass freely. Components of large protein complexes, such as ribosomal proteins and those involved in mRNA biogenesis, appear restricted in their movement. By contrast, orinithine decarboxylase antizyme and calmodulin are relatively small (40–60 kDa), monomeric proteins, and exchange between cells rapidly. This suggests that large and/or highly complexed proteins do not exchange between cells at a high rate, but smaller proteins do. These results point to the possibility of extensive intercellular movement of protein among somatic cells connected by ring canals. Implications

Emerging data now reveal that stable ring canals extensively interconnect follicle cell lineages. In addition, similar ring canals connect cells of the imaginal discs and larval brain. These somatic ring canals are over 100 times larger than gap junctions

(250 nm for ring canals versus 1.5 nm for gap junctions) and thus provide a viable path for macromolecules to move between cells. Our live-imaging data show that some proteins can freely exchange between connected follicle cells, which raises interesting questions about the true autonomy of these cells. A practical consideration concerns the use of GFP as a marker for genetic mosaic analysis, because unrestricted movement of GFP between cells would compromise it as a clonal marker. For example, using GFP loss to identify a mitotic clone of homozygous mutant cells could be confounded by GFP movement from a twin-spot cell or an unrecombined heterozygous cell. However, our data suggest that for large clones, the movement of GFP is unlikely to affect the results. The majority (66%) of follicle cells have ring canals to only one or two other cells, which suggests that the number of potentially compromised cells would be small. Furthermore, GFP moving into a genetically mutant cell could be accompanied by wild-type gene product, potentially creating a wild-type phenotype. However, if small clones or single cells are being investigated, protein exchange through ring canals might pose a significant concern. Further investigation to determine the extent and control of traffic through ring canals will reveal how follicle cells use and regulate their ability to directly communicate with neighboring cells. Materials and Methods Drosophila strains

The following Drosophila lines were used: w1118; P[w+ Ub-GFP-Pav-KLP]31 (Minestrini et al., 2002), Nasrat-Flu and Polehole-Flu (Jimenez et al., 2002) and w; UASt-PA-GFP, UASt-PA-GFP; UASt-PA-GFP, UASt-PA-GFP (Datta et al., 2008). The following protein-trap lines were used: Vsg-GFP (CA7004), OdaGFP (YD0523), RpL30-GFP (YB0151), Yps-GFP (ZCL1607) (Buszczak et al., 2007; Quinones-Coello et al., 2007; Shimada et al., 2011). Protein-trap lines are available at Flytrap (http://flytrap.med.yale.edu/). Lines P[His2Av-mRFP1] (stock #23650), P[UAS-2xEGFP] (stock #6874), P[Act5C-GAL4] (stock #4414) and P[tubP-GAL4] (stock #5138) were obtained from the Bloomington Drosophila Stock Center. Construction of transgenes and generation of transgenic lines: A cDNA clone of ypsilon schachtel (LD37574) was obtained from Drosophila Genomics Resource Center and subcloned into pDONOR201 using the Gateway system (Invitrogen). Then the insert was cloned into a pUASp vector containing a C-terminal mRFP tag, generating a Yps::mRFP transgene. BestGene performed DNA injection into embryos to make transgenic stocks. PA-GFP and FLIP

Stage-10 egg chambers were dissected in Schneider’s Drosophila medium (Invitrogen) supplemented with 0.2 mg/ml insulin onto a coverslip that was gently affixed to a slide with vacuum grease. Photoactivation of PA-GFP was accomplished on an upright Zeiss LSM META 510 microscope equipped with a 405 nm Laser Diode. Activation within a defined ROI required 200–400 iterations of 405 nm light with 100% transmittance and a scan speed of 1.6 msec/pixel using a 406 water-immersion objective. Activated GFP was observed by capturing 2– 3 mm slices for 10–20 minutes with 488 nm excitation. FLIP experiments were conducted on a Zeiss LSM-510 META microscope. Individual cells were bleached with 488 nm light. To minimize the risk of bleaching neighboring cells because of sample drift, each bleach cycle was limited to 25–50 iterations, but conducted 2–3 times every ,5 minutes for up to an hour. 2 mm slices were captured before and after every bleach cycle. Images were processed with ImageJ and Adobe Photoshop CS4. Immunofluorescence

Ovaries were dissected in ionically matched Drosophila saline (IMADS) buffer (Singleton and Woodruff, 1994) and fixed as described previously (Verheyen and Cooley, 1994). Primary antibodies used were: rabbit polyclonal a-htsM 1:500 (Petrella et al., 2007), mouse monoclonal a-DE-cadherin 1:20 (Developmental Studies Hybridoma Bank), rabbit polyclonal anti-Innexin-1 1:100 (Bauer et al., 2006), rabbit polyclonal anti-Centrosomin 1:500 (Heuer et al., 1995), rabbit polyclonal anti-Phospho histone H3 1:1000 (Upstate Signaling). Secondary

Syncytial Drosophila follicle cells antibodies conjugated to Alexa Fluor 488, 568 or 633 were used at 1:500 (Molecular Probes). DAPI was used at 1:10,000 (Invitrogen). Images were taken on either a Zeiss LSM-710 Duo, a Zeiss LSM-510 META microscope or on a Zeiss Axiovert 200 equipped with a CARV II confocal imaging system (BD Bioscience) and CoolSnap HQ2 camera (Roper Scientific). Three Zeiss objectives were used: 256 NA 0.8 Neofluar, 406 1.2 W C-Apochromat or a 636 1.2 W CApochromat. Images were processed using iVision (BD Bioscience), ImageJ (NIH), LSM software (Zeiss), and/or Adobe Photoshop CS4. Adobe Photoshop CS4 was used to uniformly adjust levels in some images. Cell and ring canal counting

Cells (nuclei) and ring canals were manually counted using Imaris. 3D reconstruction of entire egg chambers allowed counting of all cells and ring canals in egg chambers up to stage 5. The thickness of stage 6–9 egg chambers prevented adequate acquisition of the entire follicle, so only half the cells were counted and the totals calculated by multiplying by two. The anterior and posterior polar cells are identifiable by elevated nuclear localization of GFP::Pav-KLP and were used to establish a midline that divided the egg chambers in half.

Journal of Cell Science

Live imaging

Egg chambers for live imaging of follicle cell mitoses were processed as previously described (Prasad et al., 2007). Ovaries were dissected in Schneider’s Drosophila medium (Invitrogen). Ovarioles were separated from the ovary and the surrounding muscle sheath removed. Early-stage egg chambers were placed in glass-bottom culture dishes (MatTek) in Schneider’s medium with insulin added to a final concentration of 0.2 mg/ml. Time-lapse microscopy was taken on a Zeiss Axiovert 200 equipped with a CARV II confocal imager (BD Bioscience) and CoolSnap HQ2 camera (Roper Scientific). Images were taken every 90 seconds for 20–30 minutes. Images were processed using iVision and/or Adobe Photoshop CS4. Immuno-electron microscopy

Post-mitotic GFP::Pav-KLP and GFP::Vsg egg chambers were fixed in 4% paraformaldehyde in 0.1 M HEPES for 1 hour, then transferred to 5% agarose. Samples were placed in 50 mM NH4Cl with 100 mM glycine and 2% sucrose for 30 minutes to quench aldehyde groups, placed in 0.5% tannic acid in 0.1 M HEPES for 1 hour, then fixed in 2% uranyl acetate in Tris 50 mM maleate for 1 hour. Samples were dehydrated in a series to 95% ethanol and embedded in LR white resin, and polymerized overnight at 4 ˚C. Ultrathin 60 nm sections were cut on a Reichert Ultra microtome and collected on formvar/carbon-coated grids. Grids were placed on 0.1 M ammonium chloride to quench untreated aldehyde groups, then blocked for nonspecific binding on 1% fish skin gelatin in PBS with 10% normal donkey serum. Grids were incubated in rabbit anti-GFP primary antibody (Molecular Probes) at 1:250 and in donkey anti-rabbit direct gold 12 nm secondary antibodies (Jackson Laboratories) at 1:250 dilution. All grids were rinsed in PBS, fixed using 1% glutaraldehyde, then rinsed and transferred to a drop of uranyl acetate methylcellulose. Samples were viewed on a FEI Tencai Biotwin TEM at 80 Kv. Images were taken using a Morada CCD and iTEM (Olympus) software.

Acknowledgements

We thank Marc Pypaert and Morven Graham (Yale University, New Haven, CT) for help with immunoEM. We thank Michael Buszczak (University of Texas Southwestern Medical Center, Dallas, TX) for GFP-Vsg flies, Richard Axel (Columbia University, New York, NY) for the PA-GFP flies, and the Bloomington Stock Center for numerous additional fly stocks. Monoclonal antibodies were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa. Funding

This work was supported by grants from the National Institutes of Health [grant numbers GM052702 and GM043301 to L.C.; grant number T32 GM04799 to S.J.A. and P.F.M.]; and a Long-term Fellowship from the Human Frontiers in Science Program to Y.S. Deposited in PMC for release after 12 months. Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.090456/-/DC1

9

References Adams, R. R., Tavares, A. A., Salzberg, A., Bellen, H. J. and Glover, D. M. (1998). pavarotti encodes a kinesin-like protein required to organize the central spindle and contractile ring for cytokinesis. Genes Dev. 12, 1483-1494. Armstrong, J. D., Texada, M. J., Munjaal, R., Baker, D. A. and Beckingham, K. M. (2006). Gravitaxis in Drosophila melanogaster: a forward genetic screen. Genes Brain Behav. 5, 222-239. Bae, G. U., Gaio, U., Yang, Y. J., Lee, H. J., Kang, J. S. and Krauss, R. S. (2008). Regulation of myoblast motility and fusion by the CXCR4-associated sialomucin, CD164. J. Biol. Chem. 283, 8301-8309. Bauer, R., Weimbs, A., Lechner, H. and Hoch, M. (2006). DE-cadherin, a core component of the adherens junction complex modifies subcellular localization of the Drosophila gap junction protein innexin2. Cell Commun. Adhes. 13, 103-114. Besse, F. and Pret, A. M. (2003). Apoptosis-mediated cell death within the ovarian polar cell lineage of Drosophila melanogaster. Development 130, 1017-1027. Boyle, M. J., French, R. L., Cosand, K. A., Dorman, J. B., Kiehart, D. P. and Berg, C. A. (2010). Division of labor: subsets of dorsal-appendage-forming cells control the shape of the entire tube. Dev. Biol. 346, 68-79. Brill, J. A., Hime, G. R., Scharer-Schuksz, M. and Fuller, M. T. (2000). A phospholipid kinase regulates actin organization and intercellular bridge formation during germline cytokinesis. Development 127, 3855-3864. Buszczak, M., Paterno, S., Lighthouse, D., Bachman, J., Planck, J., Owen, S., Skora, A. D., Nystul, T. G., Ohlstein, B., Allen, A. et al. (2007). The carnegie protein trap library: a versatile tool for Drosophila developmental studies. Genetics 175, 15051531. Datta, S. R., Vasconcelos, M. L., Ruta, V., Luo, S., Wong, A., Demir, E., Flores, J., Balonze, K., Dickson, B. J. and Axel, R. (2008). The Drosophila pheromone cVA activates a sexually dimorphic neural circuit. Nature 452, 473-477. de Cuevas, M. and Spradling, A. C. (1998). Morphogenesis of the Drosophila fusome and its implications for oocyte specification. Development 125, 2781-2789. Field, C. M., Coughlin, M., Doberstein, S., Marty, T. and Sullivan, W. (2005). Characterization of anillin mutants reveals essential roles in septin localization and plasma membrane integrity. Development 132, 2849-2860. Fiil, A. (1978). Follicle cell bridges in the mosquito ovary: syncytia formation and bridge morphology. J. Cell Sci. 31, 137-143. Forde, S., Tye, B. J., Newey, S. E., Roubelakis, M., Smythe, J., McGuckin, C. P., Pettengell, R. and Watt, S. M. (2007). Endolyn (CD164) modulates the CXCL12mediated migration of umbilical cord blood CD133+ cells. Blood 109, 1825-1833. Giorgi, F. (1978). Intercellular bridges in ovarian follicle cells of Drosophila melanogaster. Cell Tissue Res. 186, 413-422. Greenbaum, M. P., Yan, W., Wu, M. H., Lin, Y. N., Agno, J. E., Sharma, M., Braun, R. E., Rajkovic, A. and Matzuk, M. M. (2006). TEX14 is essential for intercellular bridges and fertility in male mice. Proc. Natl. Acad. Sci. USA 103, 4982-4987. Haglund, K., Nezis, I. P., Lemus, D., Grabbe, C., Wesche, J., Liestol, K., Dikic, I., Palmer, R. and Stenmark, H. (2010). Cindr interacts with anillin to control cytokinesis in Drosophila melanogaster. Curr. Biol. 20, 944-950. Heuer, J. G., Li, K. and Kaufman, T. C. (1995). The Drosophila homeotic target gene centrosomin (cnn) encodes a novel centrosomal protein with leucine zippers and maps to a genomic region required for midgut morphogenesis. Development 121, 3861-3876. Hime, G. R., Brill, J. A. and Fuller, M. T. (1996). Assembly of ring canals in the male germ line from structural components of the contractile ring. J. Cell Sci. 109 (Pt12), 2779-2788. Hudson, A. M. and Cooley, L. (2002). A subset of dynamic actin rearrangements in Drosophila requires the Arp2/3 complex. J. Cell Biol. 156, 677-687. Jimenez, G., Gonzalez-Reyes, A. and Casanova, J. (2002). Cell surface proteins Nasrat and Polehole stabilize the Torso-like extracellular determinant in Drosophila oogenesis. Genes. Dev. 16, 913-918. Kelso, R. J., Buszczak, M., Quinones, A. T., Castiblanco, C., Mazzalupo, S. and Cooley, L. (2004). Flytrap, a database documenting a GFP protein-trap insertion screen in Drosophila melanogaster. Nucleic. Acids. Res. 32, D418-D420. King, R. C. (1970). Ovarian Development in Drosophila melanogaster. New York: Academic Press. Kramerova, I. A. and Kramerov, A. A. (1999). Mucinoprotein is a universal constituent of stable intercellular bridges in Drosophila melanogaster germ line and somatic cells. Dev. Dyn. 216, 349-360. Lin, H. and Spradling, A. C. (1993). Germline stem cell division and egg chamber development in transplanted Drosophila germaria. Dev. Biol. 159, 140-152. Margolis, J. and Spradling, A. (1995). Identification and behavior of epithelial stem cells in the Drosophila ovary. Development 121, 3797-3807. Meola, S. M., Mollenhauer, H. H. and Thompson, J. M. (1977). Cytoplasmic bridges within the follicular epithelium of the ovarioles of two diptera, Aedes aegypti and Stomoxys calcitrans. J. Morphol. 153, 81-85. Minestrini, G., Mathe, E. and Glover, D. M. (2002). Domains of the Pavarotti kinesinlike protein that direct its subcellular distribution: effects of mislocalisation on the tubulin and actin cytoskeleton during Drosophila oogenesis. J. Cell Sci. 115, 725-736. Minestrini, G., Harley, A. S. and Glover, D. M. (2003). Localization of Pavarotti-KLP in living Drosophila embryos suggests roles in reorganizing the cortical cytoskeleton during the mitotic cycle. Mol. Biol. Cell 14, 4028-4038. Nystul, T. and Spradling, A. (2007). An epithelial niche in the Drosophila ovary undergoes long-range stem cell replacement. Cell Stem. Cell 1, 277-285. Nystul, T. and Spradling, A. (2010). Regulation of epithelial stem cell replacement and follicle formation in the Drosophila ovary. Genetics 184, 503-515.

10

Journal of Cell Science 124 (0)

Journal of Cell Science

Patterson, G. H. and Lippincott-Schwartz, J. (2002). A photoactivatable GFP for selective photolabeling of proteins and cells. Science 297, 1873-1877. Petrella, L. N., Smith-Leiker, T. and Cooley, L. (2007). The Ovhts polyprotein is cleaved to produce fusome and ring canal proteins required for Drosophila oogenesis. Development 134, 703-712. Poodry, C. A. and Schneiderman, H. A. (1970). The ultrastructure of the developing leg of Drosophila melanogaster. Wilhelm Roux’ Archiv. 166, 1-44. Prasad, M., Jang, A. C., Starz-Gaiano, M., Melani, M. and Montell, D. J. (2007). A protocol for culturing Drosophila melanogaster stage 9 egg chambers for live imaging. Nat. Protoc. 2, 2467-2473. Quinones-Coello, A. T., Petrella, L. N., Ayers, K., Melillo, A., Mazzalupo, S., Hudson, A. M., Wang, S., Castiblanco, C., Buszczak, M., Hoskins, R. A. et al. (2007). Exploring strategies for protein trapping in Drosophila. Genetics 175, 10891104. Ramamurty, P. S. and Engels, W. (1977). Occurrence of intercellular bridges between follicle epithelial cells in the ovary of Apis mellifica queens. J. Cell Sci. 24, 195-202. Robinson, D. N. and Cooley, L. (1996). Stable intercellular bridges in development: the cytoskeleton lining the tunnel. Trends. Cell Biol. 6, 474-479. Shimada, Y., Burn, K. M., Niwa, R. and Cooley, L. (2011). Reversible response of protein localization and microtubule organization to nutrient stress during Drosophila early oogenesis. Dev. Biol. 355, 250-262. Singleton, K. and Woodruff, R. I. (1994). The osmolarity of adult Drosophila hemolymph and its effect on oocyte-nurse cell electrical polarity. Dev. Biol. 161, 154167.

Skora, A. D. and Spradling, A. C. (2010). Epigenetic stability increases extensively during Drosophila follicle stem cell differentiation. Proc. Natl. Acad. Sci. USA 107, 7389-7394. Straight, A. F., Field, C. M. and Mitchison, T. J. (2005). Anillin binds nonmuscle myosin II and regulates the contractile ring. Mol. Biol. Cell 16, 193-201. Swiatek, P., Kubrakiewicz, J. and Klag, J. (2009). Formation of germ-line cysts with a central cytoplasmic core is accompanied by specific orientation of mitotic spindles and partitioning of existing intercellular bridges. Cell Tissue. Res. 337, 137-148. Tilney, L. G., Tilney, M. S. and Guild, G. M. (1996). Formation of actin filament bundles in the ring canals of developing Drosophila follicles. J. Cell Biol. 133, 61-74. Verheyen, E. M. and Cooley, L. (1994). Profilin mutations disrupt multiple actindependent processes during Drosophila development. Development 120, 717-728. Wilhelm, J. E., Mansfield, J., Hom-Booher, N., Wang, S., Turck, C. W., Hazelrigg, T. and Vale, R. D. (2000). Isolation of a ribonucleoprotein complex involved in mRNA localization in Drosophila oocytes. J. Cell Biol. 148, 427-440. Woodruff, R. I. and Tilney, L. G. (1998). Intercellular bridges between epithelial cells in the Drosophila ovarian follicle: a possible aid to localized signaling. Dev. Biol. 200, 82-91. Yakoby, N., Bristow, C. A., Gong, D., Schafer, X., Lembong, J., Zartman, J. J., Halfon, M. S., Schupbach, T. and Shvartsman, S. Y. (2008). A combinatorial code for pattern formation in Drosophila oogenesis. Dev. Cell 15, 725-737. Zhou, G. Q., Zhang, Y., Ferguson, D. J., Chen, S., Rasmuson-Lestander, A., Campbell, F. C. and Watt, S. M. (2006). The Drosophila ortholog of the endolysosomal membrane protein, endolyn, regulates cell proliferation. J. Cell Biochem. 99, 1380-1396.