M6 Membrane Protein Plays an Essential Role in Drosophila Oogenesis

We had previously shown that the transmembrane glycoprotein M6a, a member of the proteolipid protein (PLP) family, regulates neurite/filopodium outgrowth, hence, M6a might be involved in neuronal remodeling and differentiation. In this work we focused on M6, the only PLP family member present in Drosophila, and ortholog to M6a. Unexpectedly, we found that decreased expression of M6 leads to female sterility. M6 is expressed in the membrane of the follicular epithelium in ovarioles throughout oogenesis. Phenotypes triggered by M6 downregulation in hypomorphic mutants included egg collapse and egg permeability, thus suggesting M6 involvement in eggshell biosynthesis. In addition, RNAi-mediated M6 knockdown targeted specifically to follicle cells induced an arrest of egg chamber development, revealing that M6 is essential in oogenesis. Interestingly, M6-associated phenotypes evidenced abnormal changes of the follicle cell shape and disrupted follicular epithelium in mid- and late-stage egg chambers. Therefore, we propose that M6 plays a role in follicular epithelium maintenance involving membrane cell remodeling during oogenesis in Drosophila.


Introduction
The glycoprotein M6a was identified years ago, and despite its prominent expression in the central nervous system and other tissues [1,2], its biological function is still unknown. Previous findings established that chronic stress decreases M6a mRNA levels in the hippocampus, and this downregulation is prevented by chronic administration of antidepressants [3,4]. In addition M6a plays an important role in neurite outgrowth and filopodium/spine formation [5], as well as in filopodium motility and likely synapse formation [6], suggesting that it might be involved in the plastic changes found in the hippocampus of stressed/antidepressant-treated animals. M6a was also found to be involved in the differentiation of neurons derived from embryonic stem cells [7]. In addition to its neuronal expression, M6a is also expressed in different epithelial cell types, such as the proximal tubules of the kidney, the choroid plexus [1,8], and human lung and ovary (www.genecards.org).
M6a belongs to the myelin proteolipid protein (PLP) family. In mammals, other members of this family include the closely related M6b and the founder PLP, with its splice variant DM20. All of the PLP members have four transmembrane domains that allow their localization at the plasma membrane and are widely conserved along evolution from arthropods to mammals [9,10]. Interestingly, other members of the PLP family such as M6b and DM20, but not PLP, are also regulated by chronic stress, and were shown to be involved in neurite outgrowth and filopodium formation [11]. Members of the PLP family were proposed to interact with the actin cystoskeleton after their association with actin-enriched membranes [12]. Additionally, PLP and DM20 were proposed to act as adhesion molecules [13], and M6a as an ion channel [14], underscoring that the molecular function of the PLP family remains undefined.
While mammals contain three plp genes (M6a, M6b and Plp1), a single M6 gene is found in arthropods [10]. Not only the structural organization of the fly M6 gene resembles that of the mouse M6a gene, but also the predicted M6 protein shares amino acid similarity with the mouse M6a [15,16], suggesting that M6 might be the functional M6a homolog. However, up to date no experimental evidence for M6 expression or function has been reported. Thus we set forward to investigate M6a requirement employing Drosophila as the model system.
In this work we focused on a phenotype, namely female sterility in D. melanogaster, induced by M6 downregulation using hypomorphic alleles and RNAi lines. Such sterility was associated with egg permeability and collapse, suggesting M6 involvement in eggshell biosynthesis. We found that M6 is expressed in the follicular epithelium of egg chambers throughout oogenesis in wild type flies. Tissue-specific M6 knockdown targeted to follicle cells impaired follicle cell shape changes and triggered disruption of the follicular monolayer. Indeed, partial loss of M6 function unraveled its key role at stages where cell adhesion, cytoskeletal rearrangements and cell-cell communication enable cell shape changes and organization.

Fly strains
Flies were grown and maintained at 25uC in vials containing standard cornmeal-agar medium.
M6 mutant alleles were generated by excision. The original Pelement (EY07032) from M6 01 was removed with the transposase (D2-3). White-eyed stocks corresponding to independent excision events were kept as M6 D01 allele strains. To characterize the M6 allelic series, PCR reactions on genomic DNA from homozygous flies were performed as detailed in Table S1. PCR and sequencing reactions allowed the identification of imprecise excisions.

Fertility experiments
Female fertility was assessed crossing two females (or 4 females, if at 18uC) to 5 wild type males per vial. For male fertility experiments, 5 males were crossed to 2 wild type females. Crosses were kept in each vial with yeast paste for 2-3 days at 25uC when testing the fertility of M6 alleles, or at 18/29uC in M6 knockdown experiments. The number of adult offspring per vial resulting from each cross was quantified.

Eggshell analysis
Egg laying was analyzed in batches of 30 or 50-60 females with wild type males when incubated at 25uC or 18/29uC respectively. Females were allowed to lay eggs on yeasted agar plates for 6 hours at 25uC or for 4 hours at 18/29uC. Only eggs laid during the afternoon were scored in order to reduce variability due to rhythmic behavior [19]. The number of eggs laid in each plate was normalized by the number of tested females.
Egg collapse was determined collecting 0-4 hours-old eggs and transferring them onto dry plates. Collapsed and non-collapsed eggs were counted every 5 minutes.
Eggshell permeability was assessed by the neutral red assay (adapted from [20,21]). Briefly, laid eggs (0-4 hours-old) were collected and incubated in 5 mg/ml neutral red in PBS for 1 h. Eggs were washed three times and mounted for visualization. Only colored eggs indicating neutral red uptake were scored as permeable. Only intact eggs were analyzed.

mRNA Isolation and Quantitative Real Time Reverse Transcription Polymerase Chain Reaction (RT-qPCR)
Ovaries were dissected from well-fed young females and homogenized in Trizol Reagent (Life Technologies, New York, USA) to isolate total RNA. Then PolyA+mRNA was purified using the PolyATract mRNA Isolation System (Promega, Madison, Wisconsin, USA). Complementary DNA was synthesized using oligo dT and SuperScript TM II Reverse Transcriptase enzyme (Life Technologies).
qPCRs were carried out in a 7500 Real-Time PCR System (Applied Biosystems, Foster City, California, USA). Quantitation of each cDNA was achieved using SYBR Green PCR Master Mix (Applied Biosystems) in triplicate. Primer sequences are detailed in Table S1. Normalization was accomplished employing Rp49 and gapdh as housekeeping genes. Values shown in all figures were calculated using Rp49 as the reference gene. Normalizations to gapdh resulted in almost identical patterns. Relative quantification was performed using a comparative CT method [26,27]. Before each experiment, the calibration curves were validated. Samples whose curves amplified out of the calibrated dynamic range were eliminated. All procedures followed the manufacturer's instructions. Values shown are relative to the control genotype (white bar) at each temperature condition.

Statistical data analysis
Graphs were generated with GraphPad Prism version 5.0 (Graphpad Software, California, USA). Statistical analysis was performed with IS (Infostat software). Group means were analyzed for overall statistical significance by unpaired Student's t test or one-way analysis of variance (ANOVA) followed by Bonferroni's multiple comparison tests. Non-parametric analysis was performed (Mann-Whitney or Kruskal-Wallis followed by Dunn's Multiple Comparison Test) when assumptions on the normal distribution and variance did not allow otherwise.

Reduced M6 mRNA levels results in female sterility
To gain insight into M6 function transposon insertions within the M6 locus were identified (at http://flybase.org/); three potential M6 mutants, renamed M6 01 , M6 02 and M6 03 (Fig. 1A), were extensively analyzed. We focused on M6 01 , which dis-played clear fertility defects as homozygotes, while M6 02 were fertile and M6 03 were lethal. The lethality phenotype was also observed with other M6 alleles (data not shown) and ubiquitous expression of M6-RNAi ( Fig. S6 and Table S2), thus revealing an essential contribution of M6 during Drosophila development.
To study the sterility phenotype, the number of offspring emerged from crosses involving homozygous M6 01 or M6 02 with wild type flies were quantified (Fig. 1B). M6 01 females were completely sterile while M6 01 males showed normal fertility. We also observed female sterility of the transheterozygote M6 01 /M6 03 (data not shown). To confirm that female sterility was associated to the insertion within the M6 locus we mobilized the P-element from M6 01 , giving rise to M6 D01-rev . Excision of M6 01 rescued fertility to wild type levels ( Fig. 1B, left panel), suggesting that altered M6 levels could be responsible for the sterility phenotype.
We anticipated that M6 should be present in the central nervous system based on information available on M6a [1,8]; in addition, the sterility phenotype suggested that M6 is expressed in the ovary. We analyzed M6 mRNA profiles by RT-qPCR ( Fig. 2A) and found that it is expressed in control ovaries but to a lower extent compared to fly heads; moreover, M6 01 exhibited a significant reduction in M6 mRNA levels ( Fig. 2B), revealing its hypomorphic nature, which were restored upon P-element excision (M6 D01-rev ). Interestingly, ovaries from M6 02 , the non-sterile strain, displayed elevated M6 levels.
The locus contiguous to M6, CG33214, is only 384 bp away from the insertion site of the P-element in M6 01 . However, CG33214 mRNA levels in ovaries showed no differences between M6 01 , M6 D01-rev and control flies by RT-qPCR (Fig. S1), ruling out CG33214 contribution to the M6 01 phenotype.
Incomplete excision of the P-element in M6 01 retrieved an allelic series, M6 D01 . We analyzed female fertility and M6 mRNA levels in the ovaries from each M6 allele. Each novel M6 allele has a remnant P-element inserted (Fig. 2C, left panel). As M6 mRNA levels decreased in the allelic series, a drastic decline in the number of progeny was evidenced (middle panel), indicating that subtle changes in M6 levels (right panel) associated with a severe decrease in female fertility. Male fertility did not show significant differences among M6 alleles (data not shown). Taken together, these results show that female sterility is triggered by M6 downregulation.

Female sterility is accompanied by abnormally permeable eggs
To explore the basis of the sterility phenotype, we examined the egg laying ability of M6 01 females. The number of eggs laid by M6 01 flies was significantly reduced compared to controls (Fig. 3A). Upon close inspection, mutant eggs also exhibited a collapsed structure soon after deposition. We followed in time the percentage of collapsed eggs in M6 01 and controls (Fig. 3B,C). While control eggs were completely normal, the percentage of collapsed eggs deposited by M6 01 females increased drastically over time and reached almost 100% in 20 minutes. Using light microscopy, no gross alterations in chorionic structures were observed. However, since collapsed eggs suggest that eggshellsnormally restricting permeability and preventing desiccation [20,21]-are compromised, we assessed egg permeability. Not surprisingly, all mutant eggs became fully stained with the dye neutral red, in contrast to control ones (Fig. 3D). Egg collapse and egg permeability were also evidenced in the trans-heterozygote M6 01 /M6 03 (Fig. S3A, and data not shown). Permeable eggs are indicative of defects in the integrity of the vitelline membrane, which disrupt the morphogenesis of the wax layer that surrounds it [21]. These results indicate that reduced M6 mRNA levels induce eggshell malformation, supporting a role for M6 during oogenesis.
We then focused on the structure of M6 01 ovarioles. Each ovariole includes a series of egg chambers organized in progressive stages from the germarium to stage 14 that mature into an egg (Fig. 4B). F-actin and DAPI staining did not reveal morphological defects at any stages of oogenesis in M6 01 ( Fig. S2; data not shown). We also employed general markers of oogenesis such as Armadillo/b-catenin to visualize cell-cell adhesive junctions [28], and Fasciclin-III, strongly expressed in the polar pairs, the specialized follicle cells (FC) involved in oocyte polarity. No differences between control and M6 01 were observed with either marker through early and middle oogenesis (Fig. S2A,B).
As mentioned earlier, the M6 01 phenotype is reminiscent of defects in eggshell formation [20,21,24,29,30]. So, we next  Table S1). The size of the remaining fragment is indicated. Middle panel: Female fertility was measured as in Figure 1; mean 6 s.e.m., n = 3. Right panel: M6 mRNA levels in ovaries was measured by RT-qPCR and normalized; mean 6 s.e.m., n = 3. Note that two alleles restored M6 mRNA to wild type levels and fully rescued the female phenotype. Spearman correlation coefficient between female fertility and M6 mRNA levels is r s = 0.92. doi:10.1371/journal.pone.0019715.g002 examined in M6 01 ovaries the localization of two major components of the vitelline membrane, sV17 and sV23, since sV23 mutant eggs also collapse [31]. However, we did not detect signs of sV23 or sV17 delocalization either in M6 01 (Figs. S2C and S3B-C) or in transheterozygote M6 01 /M6 03 egg chambers (Fig. S3B,C). Moreover, although it has been shown that the microvilli of the FC are involved in the biogenesis of the vitelline membrane [20], no clear differences in the structure of the FC microvilli between M6 01 and control ovaries were observed employing Cad99C as the specific marker (Fig. S2D). Finally, as alterations of various steps in yolk production are also associated with an egg collapse phenotype [32,33] we tested the localization of the vitellogenin receptor, Yolkless, in mutant egg chambers in order to assess vitellogenin endocytosis. However, we did not detect altered Yolkless localization throughout oogenesis in M6 01 and M6 01 /M6 03 mutants (Figs. S2E and S3D). Thus, further analysis will be required to define which step in eggshell assembly is affected upon reduction of M6 levels.

M6 localizes to the membrane of the follicular epithelium
In an attempt to assess M6 localization we generated antibodies against M6 which failed to efficiently recognize the endogenous protein. Instead, we took advantage of an enhancer trap line carrying a promoterless GFP sequence that had been mapped to M6, renamed M6 GFP (Fig. 4A, [17]). Expression of GFP-M6 in M6 GFP ovaries was tested by RT-PCR. DNA sequencing confirmed that GFP is fused to the N-terminal of M6 isoforms derived from P1 (Fig. 4A). Although M6 expression was slightly increased in M6 GFP ovaries, M6 GFP females did not show any fertility phenotype (Fig.  S4A, B). Thus, immunofluorescence of M6 GFP ovarioles was carried out to spatially and temporally localize GFP-M6.
M6-associated GFP signal was detected in the FC throughout oogenesis (Figs. 4C-F, S4C and S5). Specifically, M6 localized along the entire lateral and basal membranes but appeared more concentrated to the microvilli in the apical membrane of the FC (Fig. 4C d-d9); besides, it was detectable in the tricellular junctions of the basal membrane ( Fig. 4C b9-d, d90), and particularly in the open zone of contact (open-ZOC, Fig. 4C b0), recently implicated in cell adhesion [34,35].
We next costained M6 GFP ovarioles with follicular markers. Armadillo/b-catenin confirmed GFP-M6 localization to the membrane of the FC (Figs. 4C a-c and S4C). Additionally, Fasciclin III and GFP-M6 colocalized during early and mid stages (Figs. 4D and S5A), indicating that M6 is strongly expressed in both anterior and posterior polar cells, as well as in the migrating border cells. During stages 9-12 the squamous FC over the nurse cells did not show GFP-M6, implying that GFP-M6 is absent from this subset of FC. Further M6 localization to the microvilli through stages 9 to 10B was confirmed using anti-Cad99C (Figs. 4E and S5B). Interestingly, lack of co-staining with the cortex marker Yolkless [22], excluded GFP-M6 from the oocyte (Fig. 4F), which coupled with the observation that GFP-M6 is not expressed in nurse cells suggests that GFP-M6 is absent from the germline.
Based on this characterization we conclude that GFP-M6 localizes to the plasma membrane of the follicular epithelium (FE) throughout oogenesis, implying it plays a role specific to the FE.
M6 knockdown restricted to the follicular epithelium phenocopies M6 01 ovarian phenotypes Given that GFP-M6 is specifically expressed in the follicular epithelium we examined M6 function restricted to this tissue. To do so we downregulated M6 expression with the GAL4/UAS system [36] using two independent M6-RNAi lines. Employing ubiquitous promoters, lethality became apparent during early development, when M6 localizes to the central nervous system and likely to the epithelia (Fig. S6A). Interestingly, certain FE-specific enhancer traps triggered lethality at pupal stages (Fig. S6B), supporting a role for M6 during embryogenesis and metamorphosis (Table S2).
Since M6 is essential throughout development we resorted to an inducible system [18] to obtain viable adult flies and restrict M6 downregulation to adulthood. We first confirmed that the enhancer trap PG45 [24], as M6 (Fig. 4), is expressed in the FE of egg chambers throughout oogenesis (Fig. S7D). To address M6 role during the development of the FE, M6-RNAi was induced for several days at 29uC in PG45.M6-RNAi females. Controls flies included M6-RNAi without PG45 (M6-RNAi/+) maintained at The P-element is indicated (black arrowhead). RT-PCR primers are indicated by arrows (see Table S1). All M6 isoforms derived from P1 are tagged to GFP at the N-terminus (Zappia et al., 29uC and PG45.M6-RNAi at 18uC, a non-induced condition (Fig. 5A). We then evaluated whether FE-restricted M6 downregulation induced the sterility phenotype displayed by M6 01 and the M6 D01 allelic series (Figs. 1, 2). PG45.M6-RNAi NIG flies kept at 29uC showed a drastic reduction in female fertility compared to controls (left panel, Fig. 5B), and in contrast to the groups kept at 18uC. The driver per se (PG45/+) did not exhibit female sterility defects at either temperature (data not shown). Thus, M6 knockdown in the FE triggered female sterility in a tissue specific manner. Figure 5. Tissue-specific M6 downregulation induces female sterility and egg permeability. (A) Outline of the TARGET system used to induce M6 downregulation. GAL4 activity is regulated by a temperature-sensitive allele of the GAL80 repressor, active when maintained at 18uC [18]. Upon transfer to 29uC the repression is released. Adult females were raised at 18uC and then incubated at the indicated temperature. (B-C) Female fertility and egg permeability were tested using two independent M6-RNAi lines, a stronger one from NIG-fly (B) and a weaker one from VDRC (C). Left panel: Female fertility was measured as in  We then corroborated whether the fertility decline associated with increased egg permeability. No permeable eggs were detected in PG45.M6-RNAi NIG and control groups at 18uC. However, a clear increase in the percentage of permeable eggs became evident in those expressing the M6-RNAi NIG (right panel, Fig. 5B). To examine whether the reduced fertility was the result of an acute reversible effect, or triggered by a long lasting (irreversible) condition, we transferred flies kept at 29uC for 76 hours to 18uC. Strikingly, only 144-168 hours at 18uC were necessary to completely revert defects in egg permeability, most likely as a result of a normalized oogenesis in PG45.M6-RNAi NIG females (right panel, Fig. 5B).
Next, we tested female fertility and egg permeability employing an independent RNAi line, termed M6-RNAi VDRC , which was also directed to the FE. A 50% reduction in the number of offspring in M6 knockdown flies was evidenced (left panel, Fig. 5C) and accompanied by an increase in egg permeability (right panel, Fig. 5C) that recovered upon transfer to 18uC. Thus, exploiting this inducible system we were able to turn on and off M6 downregulation in the FE, with the concomitant induction and restoration of altered permeability. Taken together these data imply a crucial role for M6 within the FE in eggshell formation.
A closer inspection of stage 14 egg chambers from the PG45.M6-RNAi VDRC line induced at 29uC for 72 hours showed a subtle round-shaped egg chamber phenotype (mean of the width/length ratio 6 SEM; 0.382 and 0.37660.006 for controls vs. 0.47160.006 for M6-RNAi induced egg chambers, Kruskal Wallis test, p,0.0001, n = 50 egg chambers per genotype and condition), suggesting a failure in egg chamber elongation.
In addition, we found that M6 downregulation driven by other follicular GAL4 lines were unable to trigger female sterility. A detailed description of the GAL4 drivers and enhancer traps used and the phenotypes displayed are summarized in Table S2 and Figure S7. Distinctive effects of specific GAL4 lines might be attributed to differences either in the strength, the spatial or the temporal GAL4 expression profiles in the FE throughout oogenesis.

M6 knockdown in the follicular epithelium impairs oogenesis
To further analyze M6 function we performed immunohistochemistry in ovarioles from control and M6-RNAi NIG -induced females. Loss of M6 triggered a number of developmental defects such as disorganization and even total disruption of the FE as revealed by the presence of gaps (Figs. 6A,B and S8; white arrowheads). Other developmental defects included altered follicle cell shape changes revealed as the loss of columnar FC in midoogenesis after a 24 hour induction (Figs. 6A,B and S8B; yellow asterisk) accompanied with an impaired centripetal migration of FC at st10B (Fig. S8B-C, E; yellow arrowheads), probably leading to smaller oocytes (extending less than half of the egg chamber at stage 10; Fig. S8B-C, E; white arrows). Besides, the FC failed to flatten to accommodate the growth of the oocyte at later stages after a 48 hour induction (Figs. 6B and S8). Moreover, stretched FC were not detectable at 48 or 72 hours (Figs. 6A-B and S8). Sustained M6 downregulation for 72 hours resulted in abnormal egg chambers that could not be staged (Fig. 6C). Gaps (defined as interruption of the FE, Figs. 6 A,B and S8A, E; white arrowheads), altered FC shape changes in late st9 and st10 (cuboidal to columnar, Figs. 6A,B and S8B; yellow asterisks) and disorganized FE in st11-12 (i.e. an alteration in fairly uniform hexagonal array, Figs. 6B and S8D) were quantitated (Table 1). Most egg chambers showed the three alterations just described. In addition to FE disorganization, abnormal Armadillo accumulation at the tricellular junctions of FE in st12-13 egg chambers was detected when the M6-RNAi NIG was induced for 24 h at 29uC (Fig. S8D).
To quantify the extent of the defect the number of eggs laid per female was determined. A drastic reduction in the number of eggs and 48 hours are presented in (a, 0 mm) and (b, 11.26 mm). (a9-b9) Magnified views of the indicated regions. Supplementary information is presented in Fig. S8.

(C) Light transmission and fluorescent images of late egg chambers incubated at 18uC (upper panel) and 29uC for 72 hours (lower panel).
Note the aberrant morphology induced by M6 downregulation. Scale bar is 50 mm. (D) Egg laying was determined as in Figure 3A. PG45.M6-RNAi NIG (grey bars) and control females, without the PG45 driver (white bars), were incubated at 18uC and transferred to 29uC. After 76 hours at 29uC, flies were allowed to recover at 18uC for 144-168 hours and the phenotype was restored. Mean 6 s.e.m., n = 2; Mann Whitney test between genotypes at each condition, p,0.05 (*). (E-F) Confocal images of egg chambers from PG45.M6-RNAi VDRC (PG45; UAS-M6-RNAi VDRC ; tub-GAL80 ts ) females incubated at 18uC or 32uC for 72 h are shown. (E) Arrested ovarioles were detected in PG45.M6-RNAi VDRC incubated at 32uC for 72 h, which were never evidenced in controls. The white asterisk indicates an arrested st9 egg chamber. Scale bar is 50 mm. (F) No late st9 was observed in PG45.M6-RNAi VDRC induced at 32uC for 72 h. Yellow asterisks point to defects in follicle cell shape, which are not as columnar as control, whereas green arrowheads indicate the position of polar and border cells (st9). Note that at 32uC the cluster of border and polar cells did not migrate posteriorly in late st9 egg chambers. Scale bar is 20 mm. (G-H) M6 mRNA levels were measured in ovaries by RT-qPCR. (G) Ovaries from PG45.M6-RNAi NIG (grey bars) and control females, without the PG45 driver (white bars), were dissected from adult females incubated at 18uC or at 29uC for 24, 48 and 72 hours. Mean 6 s.e.m., n = 3; unpaired t-tests with Welch correction between genotypes, p.0.05 at 18uC; p,0.05 at 29uC for 24 hours and 72 hours (*); p,0.01 at 29uC for 48 hours (**). (H) Ovaries from PG45.M6-RNAi VDRC (dark grey bars) and control females, without the PG45 driver (white bars), were dissected from adult females incubated at 18uC or at 32uC for 72 hours. Mean 6 s.e.m., n = 3; unpaired t-test with Welch correction between genotypes, p.0.05 at 18uC, p,0.05 at 32uC (*). doi:10.1371/journal.pone.0019715.g006  a, b or c). The number of egg chambers examined is indicated between brackets. (a) Changes in the shape of the FC, i.e. columnar as opposed to cuboidal (yellow asterisk in Fig. 6Ab9), were considered. This phenotype was analyzed in egg chambers from late st9 to 10. (b) Loss of FE organization, including a variety of polygonal shapes instead of the uniformly hexagonal array characteristic of FC ( Fig. 6Bb-b9). Egg chambers from st11 to 12 were analyzed. (c) Presence of gaps within the FE (white arrowheads in Fig. 6A laid by RNAi-induced females became evident (Fig. 6D). No differences were found at 18uC. As mentioned earlier (Fig. 5B), a complete rescue of the phenotype was observed several hours upon transfer to 18uC (Fig. 6D). As female fertility in PG45.M6-RNAi VDRC was only partially reduced (,50%, Fig. 5C), we hypothesized that M6-RNAi VDRC is less efficient in reducing M6 levels, and considered a¨weaker line. To increase GAL4 activity and, concomitantly, the strength of M6 downregulation, flies were transferred to 32uC for 72 hours, which triggered a clear disruption in oogenesis ( Fig. 6E; white asterisk). In contrast, controls incubated at 32uC did not show any defect (Fig. 6E). At 32uC no egg chambers in stages from 9 to 13 could be identified (Fig. 6F). In addition, both polar and border cells in stage 9 did not migrate and remained localized to the anterior region (see Fig. 6F, green arrowheads).
In sum, acute downregulation of M6 levels in the FE gives rise to a number of defects leading to disruption in oogenesis.

Discussion
We identified M6 as a protein involved in egg chamber development that when expressed at low levels gives rise to female sterility in Drosophila. M6 is expressed in different populations of follicle cells throughout oogenesis, and localizes to the plasma membrane. Such subcellular distribution was predicted from the degree of conservation with other PLP family proteins [9,10,15,16]. The mammalian ortholog, M6a, is also expressed in non-neural tissues [1,8], where it appears particularly concentrated in the apical surface along a highly dense array of microvilli [1]. M6 distribution to membrane protrusions of epithelial cells is reminiscent of that reported for mammalian M6a. M6 is also expressed in the fly embryo and is essential at later stages in development, suggesting it plays an additional role during embryogenesis and metamorphosis. Interestingly, downregulated expression of both proteins, M6 in Drosophila and M6a in mammals, correlates with pathologies: female sterility and chronic stress, respectively. Notably, in both cases, a cell remodeling process is involved.
Using different genetic strategies such as the insertional mutants and M6 knockdown in the FE we demonstrated that M6 plays an essential role during egg chamber development. Interestingly, in addition to the egg permeability phenotype displayed by M6 01 , tissue-specific knockdown disrupted oogenesis. While the P-element in M6 01 is inserted in the first exon of promoter 1 (P1, Fig. 1A), and thus potentially affects a subset of M6 splice variants, both M6-RNAi sequences were targeted to all M6 mRNAs. Remarkably, RNAi lines obtained from different collections (NIG and VDRC) led to similar phenotypes, indicating that egg permeability, altered follicular epithelium integrity and arrested egg chamber development were indeed the result of downregulated M6 expression in the FC. However, despite M6 downregulation in both M6 01 and M6-RNAi treated ovaries reached similar M6 mRNA levels (40-45%), the ovarian phenotypes displayed by M6-RNAi lines were stronger than M6 01 , implying a nonlinear relationship between the two. Moreover, M6 locus gives rise to different M6 versions as a result of the use of alternative promoters and alternative mRNA splicing (Zappia et al., unpublished results), underlying the complexity of this locus. We can envision at least two explanations to reconcile this apparent discrepancy. On the one hand, although both M6-RNAi putatively target all isoforms, their expression was limited to the PG45 domain. This might not completely describe the M6 endogenous pattern evidenced by the enhancer trap GFP-M6, which in turn reflects the pattern of isoforms driven by P1 (and not P2). Thus, it is possible that additional tissues (such as the germline) could contribute the remaining M6 levels detected by qPCR in whole ovarian extracts. On the other hand, the insertion in M6 01 could spare specific isoforms (such as those driven by P2), which could be differentially expressed in ovaries and contribute to the detected M6 levels. This topic is the focus of intense analysis and will be reported elsewhere (Zappia et al., unpublished observations).
Several evidences point to a temporal window for M6 requirement, which coincides with the time the FE undergoes tissue remodeling [37,38]. Using two alternative enhancer trap lines to drive M6 knockdown only the one that is active from stage 7 onwards triggered the sterility phenotype. In addition, morphological defects induced with M6-RNAi lines were first observed at stage 9. Thus, while oogenesis apparently proceeded relatively normally from germarium through stage 8, morphological defects triggered by downregulated M6 expression became detectable at later stages.
During the second half of oogenesis the FC have two main roles: the communication with the oocyte that is essential for patterning the egg chamber and early embryo, as well as the formation of the eggshell. The latter is characterized by a sequence of migrations and cell shape changes involving subpopulations of FC that cover the oocyte and synthesize specialized eggshell structures. During mid-oogenesis the epithelial FC cease proliferation (stage 7), and form a cuboidal epithelium surrounding the whole egg chamber (stage 8) [37]. In stage 9 the FC monolayer undergoes cell shape changes, concurrent with border cell migration [39]. In contrast, upon tissue-specific M6 knockdown the follicular monolayer was disorganized, did not achieve a columnar morphology and stayed cuboidal even at stages when elongation should be occurring.
Later on, as a result of both mechanical stretching and movement, the FC flatten through the expansion of their basal and apical membrane domains, which becomes evident at stage 10B. The membrane required for the expansion of the apical side (facing the oocyte) is probably stored in microvilli. However, there is no membrane reservoir in the basal one; consequently, a portion of the lateral membrane previously engaged in cell-cell interactions called zone of contact (ZOC) is converted into basal membrane (open ZOC) and faces the basal extracellular matrix. Such plasma membrane remodeling strengthens adhesion and ensures epithelial integrity at the next stages of development [34,35]. Since GFP-M6 localizes at the ZOC and at the microvilli of FC, and its downregulation induces epithelial monolayer disruption, M6 may contribute to membrane remodeling processes such as cell-cell adhesion, changes in cell shape, signaling and membrane trafficking, and thus be essential for maintaining the FE integrity, although additional experiments are required to confirm this hypothesis. Work from different laboratories has focused on the maintenance of the FE integrity during egg chamber development and showed the requirement of the cadherin-catenin complex [40], the serine/threonine kinase dPak [41], and integrins [42]. Additionally, M6 downregulation in FC triggered a dumpless-like phenotype. Although it usually implies structural defects in ring canals or actin-organization, it could also arise as a result of altered centripetal migration as observed at st10 late (Fig. S8C) and suggested by Dobbens and col. [39], pointing out that cell-cell communication between the follicular epithelium and the germ cells coordinate FC morphogenesis.
During late oogenesis, the structural components of the eggshell are secreted by the FC surrounding the oocyte in a precisely regulated temporal and spatial manner. In normal eggs the vitelline membrane and the wax layer prevent dehydration. Abnormalities in the deposition of either one of these structures or, alternatively, a failure to completely cover the anterior end of the oocyte upon migration of the anterior FC might cause eggs to loose water and show a collapsed phenotype [43]. M6 expression in FC secreting the eggshell material is also consistent with a role in this process.
In Drosophila gut and wing epithelia, integrins are implicated in the maintenance of columnar morphology [44,45]. Integrins could be playing a similar role in the FE overlying the oocyte. Indeed, distinct integrins alpha chains showed both temporal and spatial pattern expression differences in FC during mid-and late oogenesis [46] and newly synthesized basolaterally localized integrins are involved in the transition from columnar to cuboidal morphology in the follicular epithelium [34]. Our data suggests that both processes, FC shape changes and vitelline membrane assembly, are closely related and dependent on M6 function.
Mutations that cause spherical egg phenotype are known to disrupt proteins mediating interactions between the actin cytoskeleton and the extracellular matrix [47,48,49,50]. Since the round egg chamber phenotype triggered by M6 downregulation is subtle, additional studies are needed to investigate whether M6 plays a role during the planar-polarized orientation of basal Factin bundles in the follicular epithelium.
The failure in the maintenance of epithelial architecture induced by M6 downregulation may be due to defects in cell adhesion leading to alterations in cell shape, which in turn disrupt the epithelial organization. Preliminary studies supporting this hypothesis suggest altered distribution of DE-cadherin at the tricellular junctions of M6 01 late stages egg chambers (Zappia and col., unpublished results). A potential role of M6 in cell adhesion is further supported by the fact that PLP was proposed to act as an adhesion molecule [13] and to form a complex with integrins in oligodendrocytes [51]. In addition, mammalian M6a localization in primary cultures of hippocampal neurons [5] and other epithelia [1,8], and PLP/DM20 in HeLa cells [52], suggest that a common feature in the PLP family is their ability to associate with membrane structures involved in movement and growth [12]. In summary, we have uncovered an essential role of M6 in Drosophila oogenesis, which is particularly relevant in light of the conservation of this protein between flies and mammals. Figure S1 The P element insertion in M6 01 does not affect expression in the nearby CG33214 locus. The upstream locus contiguous to M6, CG33214, which transcription initiation is only 384 pb away from the insertion site in M6 01 , was not affected by the P element insertion. Expression of CG33214 was measured in ovaries from wild type, M6 01 and M6 D01-rev  , as well as in other tissues such as the epithelium (lower panel). Therefore, M6 knockdown in the embryo might be responsible for early lethality (see Table S2 for details). Scale bar is 50 mm. (B) When employing various enhancer traps reported to be expressed in the follicle cells to induce M6-RNAi, such as c204 and T155 (Bateman et al., 2001), an arrested pupae development was observed, suggesting that in addition to driving expression in follicle cells, those GAL4 drivers are also expressed at an unspecified tissue during larval/pupae formation. Thus, M6 deprivation during larval or pupae development stages abrogates metamorphosis (see also Table  S2). Images of day 10 (upper panel) and 15 (lower panel) of Drosophila development at 25uC from flies bearing the enhancer traps c204 and the strong allele of M6-RNAi (C204.M6-RNAi NIG ) and control flies (M6-RNAi NIG with no driver). Pupae arrest was detected during metamorphosis while control flies developed normally. Similar results were observed using T155. Scale bar is 2 mm. (TIF) ovarioles. GAL4 expression is revealed by CD8-GFP labels (green) in follicle cells. Nuclei were stained with propidium iodide or DAPI (red). PG45 expressed GAL4 in all follicle cells throughout oogenesis (D). Note GAL4 expression in subpopulation of follicle cells with different drivers (A,B,C and H). Interestingly, those follicular drivers (55B, da.Gal4, slbo.Gal4 and c329b) did not trigger any female sterility in the context of M6-RNAi, whereas PG45 did (see also  Table S2). GAL4 is expressed in all follicle cells from st7 onwards (c355, (E)) and from st8/9 onwards (c204, (F)). The specific spatial or temporal expression profiles or GAL4 expression strength may account for the differences observed in female sterility phenotype when crossed to the M6-RNAi (see also Table S2). Representative egg chambers are shown. Scale bar is 50 mm. (TIF) Figure S8 M6 downregulation in follicle cells induces a variety of defects in mid-and late oogenesis. Confocal images of representative egg chambers at different stages of oogenesis. Ovaries were dissected and stained with phalloidin (red), DAPI (blue) and Armadillo (green). PG45.M6-RNAi NIG (PG45; tub-GAL80 ts ; UAS-M6-RNAi NIG ) flies were raised at 18uC and then adult females were transferred to 29uC for 24, 48 or 72 hours to induce M6 interference in the follicular epithelium or kept at 18uC as the non-induced control. Control females, M6-RNAi NIG (without PG45), shifted to 29uC for 72 hours were undistinguishable from non-induced controls (data not shown). Egg chambers corresponding to stages 4-7 and 9 (A), late stage 9, stage 10A and 10B (B), stages 10 and 11 (C), stages 12 and 13 (D) and stage 12 (E) are shown. (A) No morphological defects were observed at early stages of oogenesis. The presence of gaps in the follicular epithelium after 48 h of M6-RNAi induction became detectable from stage 8 onwards (indicated by white arrowheads), and from stage 7 onwards after 72 h of induction. Note defects in the FC shape. (B) The first defects observed after a 24 hour induction involved follicle cell morphology in egg chambers from mid-oogenesis, including a deficit in the thickness gradient of the posterior follicle cells in late stage 9 along with a failure in the columnar shape of the follicle cells in egg chambers reaching stage 10A. Defects in the overall shape of the follicle cells are indicated by yellow asterisks. Magnified views of the indicated areas are shown at the bottom of each egg chamber. White arrows highlight morphological hallmarks of the dumpless-like phenotype in st10B (i.e., disproportioned oocyte to nurse cell cellular size). (C) After a 24 hour induction, defects in centripetal migration were detected in stage 10-11 egg chambers as indicated by yellow arrowheads. This alteration might cause the dumpless-like phenotype(white arrow). (D) Disorganization of the follicular epithelium induced in stage 11-13 egg chambers after 24 hours associated with abnormal Armadillo accumulation at the tricellular junctions. (E) At later stages of oogenesis, such as stage 12, gaps were evidenced after 48 h of induction (indicated by white arrowheads). The dumplesslike phenotype (white arrow) is detected likely as a result of altered centripetal FC migration (yellow arrowhead). Scale bar is 50 mm.

(TIF)
Table S1 Primer sequences used in PCR. To characterize the M6 allelic series, PCR reactions on genomic DNA obtained from homozygous flies were carried out with primers directed to the 59UTR of M6A flanking the original P element insertion. The E-GFP fusion to M6 was assessed by a Hemi-Nested PCR experiment employing cDNA from M6 GFP ovaries. Two rounds of PCR were performed in order to amplify isoforms with low representation. A second PCR experiment was carried out employing different pairs of primers. M6 mRNA isoforms tagged to the C-terminus of E-GFP were determined by sequence alignments (KAlign algorithm; www.ebi.ac.uk). The sequences of the expressed GFP::M6 proteins were predicted in silico. Primer sequences used in qPCR were designed using Primer Express 3.0 software (Applied Biosystems) and targeted to the 39UTR of all M6 mRNA isoforms, glyceraldehyde phosphate dehydrogenase (gapdh) and ribosomal protein 49 (Rp49). To confirm M6 levels in the M6 allelic series a second pair of primers for M6 39UTR was used. (DOC) Table S2 M6 knockdown triggers lethality during development and sterility in adult females. The table summarizes both the developmental arrest phenotype (middle column) and the female sterility phenotype (middle column) induced by M6-RNAi. Different GAL4 drivers and enhancer trap lines (left column) were used to express two independent UAS-M6-RNAi lines from the NIG-fly (strong allele) and VDRC (weak allele) stocks centers. Interference was amplified including UAS-Dicer2 (from VDRC). Early development arrest is triggered upon M6 downregulation. When pleitropic GAL4 drivers such as tubP-GAL4 [24], or follicular-specific GAL4 drivers such as the enhancer trap GAL4-daG32 [20] and PG45 [24], which are also expressed in the embryo, were used to trigger M6 interference, early lethality was observed. In every genetic combination (GAL4 driver.M6-RNAi) lethality became apparent at early stages during development, suggesting an essential role for M6 early on. This was also evidenced by the lethality of the M6 03 mutant and a number of M6 alleles generated by P element excision of the M6 01 mutant (data not shown). Some of the drivers required the addition of an allele of UAS-Dicer2 to trigger the same interference with both RNAi stocks (data not shown). Only complete female sterility was scored as female sterility. Although differences in the strength of the GAL4 drivers cannot be ruled out, M6 downregulation induced female sterility only when M6-RNAi was expressed in all follicle cells from st7 onwards (PG45, c355 and tubP-GAL4). Follicular GAL4 drivers were crossed to UAS-CD8GFP to check GAL4 expression in follicle cells (right column and Fig S7). Discrepancies from the literature were detected. (*) employing tub-GAL80 TS to repress M6-RNAi through development at 18uC. Then, M6-RNAi was induced in young adult females at 29uC. (**) tested with M6-RNAi NIG .