The nuclear activity of the actin-binding Moesin protein is necessary for gene expression in Drosophila

cytoplasm by cross-linking actin ﬁlaments with plasma membrane proteins. Research has identiﬁed the nuclear localization of ERMs, as well as the involvement of a single Drosophila ERM protein, Moesin, in nuclear mRNA exports. However, the question of how important the nuclear activity of ERM proteins are for the life of an organism has so far not been explored. Here, we present the ﬁrst attempt to reveal the in vivo relevance of nuclear localization of Moesin in Drosophila . With the help of a nuclear export signal, we decreased the amount of Moesin in the nuclei of the animals. Furthermore, we observed various developmental defects, demonstrating the importance of ERM function in the nucleus for the ﬁrst time. Transcriptome analysis of the mutant ﬂies revealed that the lack of nuclear Moesin function leads to expression changes in nearly 700 genes, among them heat-shock genes. This result together with additional ﬁndings revealed that in Drosophila the expression of protein chaperones requires the nuclear functions of Moesin.


Introduction
The presence of cytoskeletal proteins in the nucleus is undisputed today, and it is now clear that they are involved in fundamental nuclear functions [1][2][3][4]. However, the biological relevance of the activity of cytoskeletal proteins in the nucleus has not yet been determined and is still debated. This is primarily due to the fact that cytoskeletal proteins have essential activities in the cytoplasm, and therefore, the elimination or inhibition of these proteins affects their activity in the cytoplasm as well. Furthermore, with the exception of actin, the mechanism of their nuclear transport has not been resolved. Therefore, the investigation of their nuclear activity separately from their cytoplasmic functions has been an enormous challenge.
The evolutionarily highly conserved family of the actin-binding, cytoskeletal ERM (ezrin, radixin, and moesin) proteins consists of three closely related paralogs: ezrin, radixin, and moesin [5,6]. They are all present in vertebrates, whereas other species such as invertebrate only have one ERM gene [7]. ERM proteins function as general cross-linkers between plasma membrane proteins and the cortical actin cytoskeleton [8]. They can assemble multiprotein complexes at the membrane-cytoskeleton interface, thereby regulating numerous signal transduction pathways, and they also function as both upstream and downstream effectors of Rho GTPases. The lack of ERM function results in serious defects or mortality in vertebrates [9] and is lethal in invertebrate species [10,11].
Recent studies indicate that the regulation of cortical actin dynamics is only a part of the function of ERMs. Similarly to their major binding partner, actin and other cytoskeletal proteins [1], ERMs have been found to localize to the cell nucleus in numerous species [12][13][14][15]. In Drosophila melanogaster, the sole ERM representative, Moesin (Moe), has been shown to be a functional component of the nucleus by participating in mRNA export [16]. However, the in vivo importance of the nuclear localization and activity of ERMs is not known at present. Here, we demonstrate that the constant removal of Moe from the nucleus, with the help of a nuclear export signal (NES), results in abnormal development, sterility, and physiological problems in Drosophila. We also show that Moe plays role in the expression of 686 genes, among them heat-shock (hsp) genes. Our study provides the first direct evidence that the activity of the actin-binding ERM proteins in the cell nucleus is essential for the normal development and life of an organism.

Generation of the moe NES mutant
To obtain more insight into the nuclear activities of Drosophila Moe and to determine the biological relevance of this nuclear localization, we created mutant animals with decreased level of nuclear Moe. We fused the protein with a NES which ensures constant removal of Moe from the nucleus (Fig. 1A). We chose the evolutionarily highly conserved, chromosomal region maintenance 1 (CRM1)-recognized NES sequence motif of the cAMP-dependent protein kinase inhibitor alpha (PKIA) [17] and tested the effect of this NES-tag on the intracellular distribution of Moe in a pilot experiment. The nuclei of Drosophila larval salivary gland cells expressing GFP-labeled MoeNES protein from a transgene showed unambiguous decrease in the nuclear Moe level (Fig. 1B).
We then tagged the Moe protein in situ with the NES sequence using CRISPR-Cas9 technology [18]. In Drosophila, the moe gene is located on the X chromosome and with the use of alternative START sites and introns it encodes 13 different transcripts that give rise to seven different protein isoforms. We inserted the NES sequence in front of the STOP codon between the 14th exon and the 3 0 UTR which are present in all 13 moe transcripts. For the selection of the mutants, the fluorescent dsRed expressing marker gene (3xP3 > dsRed) flanked by two loxP recombination sites was inserted into the last intron (Fig. 1C). From the embryos co-injected with the guide RNAs (gRNAs) and the donor construct, four independent mutant lines, hereafter referred to as moe NES , were generated. Next, the dsRed marker gene was removed with the help of Cre recombinase, leaving a single loxP sequence in the last intron of the moe gene. Sequencing of the mRNA expressed from the moe NES gene verified successful insertion of the NES sequence, the lack of the marker gene, and the normal excision of the last intron (data not shown).
We noticed that all the progenies of the females carrying at least one copy of the moe NES allele were sterile. The sterility was specific to mothers only; the progeny of hemizygous mutant males (moe NES /Y) were fertile, indicating that the phenotype is due to a dominant maternal effect. Therefore, to maintain the moe NES allele for future experiments, mutant males were crossed to females carrying attached X chromosomes (X^X/Y) to establish the stable moe NES stock. The attached X chromosome consists of two fulllength X chromosomes sharing a common centromere so that they are always inherited together. As a result, in crosses of moe NES males to females carrying an attached X, male progenies inherit their moe NES bearing X from their father and their Y from their mother, while the females never carry the moe NES allele. In this way, the problem of maternal female sterility is eliminated. The moe NES /Y mutant males were viable and fertile; however, to analyze females carrying at least one copy of the moe NES allele, moe NES males were crossed in every experiment to females carrying a wildtype (WT) X chromosome (w 1118 ) or a null allele (moe G0415 ) of moe. The resulting female progeny (moe NES /moe G0415 , referred to as moe NES mutant in the text) expressing only the Moe NES protein were used in the assays (Fig. 1D) and compared throughout the experiments to control animals carrying either the moe The cytoplasmic activity of the MoeNES protein is unaffected The complete loss of Moe functions is lethal to the fly. Null mutant animals die at the third larval stage [19,20]. In contrast, the moe NES mutant females and males were viable, suggesting that at least the main cytoplasmic functions of Moe are unaffected. In the next experiments, we aimed to confirm this notion.
In order to verify that the NES-tag does not inhibit the activation and activity of the Moe protein, salivary glands of moe NES mutant males and ovaries of moe NES mutant mothers were stained with an anti-phospho-ERM antibody specifically recognizing the activated form of Moe [21]. The immunostaining revealed that the MoeNES protein is activated through phosphorylation and, similarly to the WT protein, it primarily localizes to the cell cortex ( Fig. 2A). This demonstrates the F-actin-binding ability of MoeNES and suggests that its molecular cross-linking function is intact. The quantitation of the immunostainings showed that, in addition to localization, the nucleo-cytoplasmic distribution of P-Moe is normal in the moeNES mutant ovary ( Fig. 2A'). The measurement of the levels of P-Moe and Moe in the ovary by western blotting ( Fig. 2A") suggested that the amount of P-Moe relative to total Moe amount in the mutant differs only by 10% from that of the control (1.38 and 1.52 Moe/P-Moe ratio in the mutant and the control, respectively). The quantitation also revealed that the total amount of Moe is reduced by 30% in the moe NES mutant (142 837 vs. 99 815 intensity values in the control and the mutant, respectively).
The female fly possesses two ovaries joined with a common oviduct, and around 16 ovarioles per ovary. The ovarioles are parallel strings of 6-7 progressively older egg chambers. The egg chamber is covered with a monolayer of somatic epithelial cells, the follicle cells. In the egg chambers, 15 polyploid nurse cells are connected to the oocyte and deposit a large amount of RNA and protein into it. During stage 10 of oogenesis, nurse cells produce unipolar, filopodia-like actin cables between their cell membrane and nucleus to demonstrate that MoeNES-GFP is not present in the nucleus. Representative images of salivary glands of eight larvae examined each genotype in one experiment. Scale bars, 50 lm. (C) In vivo tagging of the moe gene with NES signal using the CRISPR-Cas9 system. The NES sequence was inserted between the protein-coding sequence and the 3 0 UTR. (D) The moe NES allele is maintained with the help of attached X females. To generate moe NES mutant females (highlighted in red), the moe NES /Y mutant males (red) are crossed in all experiments with females carrying a moe null allele (moe G0415 ).  [22]. Moe is required for the formation of these tethering actin cables in the ovary and anchors the cortical actin network to the cell membrane in the developing oocyte. Therefore, the impairment of Moe function leads to the disorganization of the microfilament network [21] and the detachment of cortical actin from the cell membrane [19]. Visualization of the actin cytoskeleton and the membranes at stage 10 of oogenesis in ovaries of control (moe G0415 /w 1118 ) and moe NES mutant females demonstrated that the intracellular position of nurse cell nuclei and the actin network are normal in the mutant (Fig. 2B). The perturbation of nurse cell filamentous actin cable organization or their integration with the cortical actin network blocks nurse cell dumping which in turn results in reduced embryonic size [23]. The regular size and shape of the embryos laid by moe NES mutant mothers confirmed further that the cytoskeletal functions of Moe are unaffected (Fig. 2C). The intact cytoskeletal network of the oocyte also ensures the correct localization of maternal factors within the cytoplasm, which is necessary for the formation of germ cell precursors at the posterior pole of the developing embryo. Moe plays essential role in this process, as demonstrated previously by the mislocalization of maternal factors Oskar, Staufen, and Vasa, and significantly reduced number of pole cells [19,21] in the embryos of moe NES mutant mothers. However, we found that in the eggs laid by moe NES mutant females, the maternal factors Oskar, Staufen, and Vasa all localize properly at the posterior pole of the oocyte or the early embryo (Fig. 2D), and at the cellularization phase, the number of embryonic pole cells is normal also (Fig. 2E), providing additional evidence that the cytoplasmic functions of MoeNES are normal.
The MoeNES protein is exported from the nucleus by the CRM1 (XPO1) export factor which recognizes the NES motif. In the next set of experiments, we investigated whether the expression of the MoeNES protein alters the normal activity of CRM1, for example, by overloading the CRM1-mediated nuclear export pathway. We thus compared the ratio of nuclear and cytoplasmic fluorescent intensity values of the GFP-tagged Medea (SMAD4) protein, a known cargo of the CRM1 exportin [24,25], in the nurse cells of WT and mutant ovaries (Fig. 2F). We expected the nuclear accumulation of Med-GFP if the export is overloaded, but the protein exhibited very similar nuclear-cytoplasmic distribution in normal and mutant nurse cells. This confirms that the CRM1 nuclear export pathway functions normally in the moe NES mutant and that reduced CRM1 pathway activity cannot be responsible for the observed phenotypes.

Lack of nuclear Moesin causes diverse phenotypes
The moe NES mutant males and females were viable and fertile; however, they exhibited numerous phenotypes (Fig. 3A). These defects, such as slow development, decreased lifespan, egg production and climbing ability, as well as male genitalia rotation, were of   (Fig. 3B) was 32.4% (out of 145 animals), and the frequency of this phenotype decreased to 12.3% (out of 171 animals) when the males were bearing a Y chromosome that contained one copy of the WT moe gene (Dp(1:Y)619). This suggests that an extra copy of the moe gene can rescue genitalia rotation by about 60% and that the phenotype is due to the loss of Moe function.
In contrast to the zygotic phenotypes, maternal-effect phenotypes appear in the F1 descendants of moe NES females, independently from the genotype of the progeny. The maternal phenotypes primarily affected the posterior part of the animals and caused rudimentary genitalia and, as a consequence, sterility (Fig. 3A). The maternal-effect sterility of the F1 descendants was due to the undeveloped genitalia both in males and in females (Fig. 3C). The visible phenotypes in the F1 progeny of moe NES mutant mothers were defects in tergite development and pigmentation, extra hairs, wing malformations, and dark and rough eye (Fig. 3D).
The moe NES mutant stock was weak, and thus, we analyzed the viability of the animals at every main developmental stage. Out of 100 eggs laid by the moe NES mutant mothers, only 30% hatched, in contrast to the control (moe G0415 /w 1118 ) where 98% of the embryos developed to larval stage. In addition, embryonic viability further decreased with the aging of the mutant mothers (0% in the mutant vs. 92% in the control). The phenotype, especially at younger age, was partially rescued by a WT moe allele (moe NES /   4A). DAPI staining of early embryos demonstrated that, similarly to the progeny of control females, 100% of the eggs laid by moe NES mutant mothers (n = 100) pass the syncytial blastoderm stage, indicating that lack of fertilization does not contribute to embryonic lethality (not shown). After the embryonic stage, 30% lethality was observed at both larval and pupal stages of moe NES mutants. The viability at these stages was fully rescued by one copy of the WT moe gene (moe NES /w 1118 ), indicating that the lethality during development is caused by the loss of Moe function. These data also demonstrated that the development of mutant animals is slower compared with the control. The viability experiments revealed a shortened adult life span only in the case of moe NES /Y mutant males. In these males, life span was found to be decreased by~50% compared with mutant females or control males. Since these flies displayed a reduced hatching rate, we analyzed the fecundity of moe NES mutant females. They exhibited normal egg production, the daily egg yield as well as the total number of eggs laid by one female during the examined 12-day period were almost identical to those of the control females (Fig. 4B).
In the next set of experiments, we inspected the possibility that low energy homeostasis is contributing to the phenotype. We measured the egg production of mutant and control females during starvation. Two Days after pupariation  days of food deprivation resulted in a dramatic decrease in egg production in both groups, followed by a rapid return to normal levels. The kinetics of egg yield changes were the same in both groups (Fig. 4C), indicating that moe NES mutants show normal physiological reactions to temporary food deprivation and suggesting that altered energy homeostasis is not responsible for reduced life span.
During the experiments, we also noticed that mutant flies move more slowly than the control animals; therefore, we tested the basic locomotor behavior of the mutant (Fig. 4D). For the investigation of locomotor activity, we performed a negative geotaxis assay and found that mutant males move significantly more slowly than WT males, with only 13% of them climbing to the top of the vial during 8 s, compared to 48% in case of the control animals. In contrast, the locomotor activity of mutant females was normal (not shown). We also conducted the test on 1-week-old males, but the difference between the mutant and the control flies was about the same (59% vs. 10% in WT and mutant flies, respectively) (Fig. 4D).

Transcriptome analysis reveals altered gene expression in moe NES mutants
Because the progeny of mutant females exhibit various maternal-effect phenotypes (reduced viability, developmental defects, sterility), and the ovary is the place of very extensive gene expression, we performed mRNA-Seq analysis on ovaries of moe NES /moe G0415 mutant females, and compared their transcriptomes with moe NES /+ and moe G0415 /+ controls (Fig. 5). Principal component analysis (PCA) of the mRNA-Seq datasets showed 24% total variance in the first principal component (PC1) and clearly separated the mutant animals from the two controls (Fig. 5A), indicating significant differences between the transcriptomes of WT (moe G0415 /w 1118 ) and mutant samples. Compared to WT (moe G0415 /w 1118 ), approximately the same number of differentially expressed genes were upregulated (n = 371) or downregulated (n = 315) in the mutant (Fig. 5B).
To gain more biologic insight, we performed Gene Ontology (GO) term analysis on statistically significant genes [twofold (log 2 (FC) ≥ 1 and adjusted P-value of < 0.05 for upregulated genes, and log 2 (FC)≤-1 and adjusted P-value of < 0.05 for downregulated genes] differentially expressed between the genotypes moe G0415 /w 1118 , moe NES /w 1118 , and moe NES /moe G0415 by subjecting the data (summarized in Table S1) to the GO database of Flybase (https://flybase.org/). We observed that major transcriptional changes in upregulated genes related to cellular component terms involve the cell membrane and extracellular matrix, while in the case of downregulated genes, nuclear and nuclear + cytoplasmic terms were affected (Fig. 5B). The top biological process terms in both up-and downregulated gene groups are related to 'developmental process', 'response to stimuli', and 'transcription' (Fig. 5C). In addition, we found the GO terms 'chromosome organization' and 'cell cycle' in downregulated genes and 'signal transduction' and 'protein metabolic process' in the case of upregulated genes. These GO terms cover 68.3% of the downregulated genes and 51.5% of the upregulated genes.
Among the upregulated genes, three key regulators of development, vasa, Notch, and dpp, showed significant changes in gene expression activity (Fig. 5D), which may contribute to the developmental defects and sterility of moe NES mutants. Genes that might account for normal locomotor activity were also identified in the upregulated group (Fig. 5E), suggesting that abnormal gene expression may be responsible for the phenotype in adult males. These results show that the biological processes behind the phenotypes are impaired by gene programs that are directly affected in moe NES mutants.   We noticed during the transcriptomic experiments that among the best hits in genes with downregulated expression in moe NES mutant, six genes encoded heat shock-inducible protein chaperons (hsp70Aa, hsp70Ab, hsp70Ba, hsp68, hsp26, and hsp23) (Fig. 6A). The decrease in hsp gene activity was rescued with one copy of WT moe allele, strongly suggesting that the nuclear activity of Moe is necessary for the normal expression of these genes. RT-qPCR on heat-shocked mutant animals reinforced the idea that nuclear Moe participates in the transcription of hsp genes (Fig. 6B).
In addition to thermotolerance [26,27]   functioning of Drosophila hsp genes is necessary for normal life span [28], fertility and germ cell maintenance [29,30], locomotor activity [31,32], and developmental timing [27]. Since all these processes are affected in moe NES animals, we performed additional experiments to confirm the involvement of Moe in hsp gene activity. The most obvious sign for hsp malfunctioning is heat intolerance; therefore, we first analyzed the viability of the mutants at elevated temperature. At 25°C, only mutant males showed shortened life span (Fig. 4A), but at 29°C, both sexes displayed a significant decrease in life span (Fig. 6C). The phenotype was fully rescued by one copy of the WT moe gene, indicating that loss of Moe function is responsible for the decreased heat tolerance. To further demonstrate that heat intolerance causes reduced viability at high temperature in males and females, we also performed the experiment at 18°C. The result was similar to the data obtained at 25°C, only mutant males exhibited shortened life span; however, at this temperature, the mutant males lived much longer (51 vs. 21 days) than at 29°C (Fig. 6C). We also investigated whether the observed heat stress intolerance affects the negative geotaxis behavior of the mutants. The 35% decrease in the climbing ability of moe NES animals at 25°C (Fig. 4D) did not change significantly at higher temperature (data not shown), indicating that heat stress does not affect climbing activity. This in turn suggests that impaired hsp expression does not contribute to locomotor problems of the mutant.
In our previous work, we discovered that upon heat shock the amount of Moe significantly increases in the nucleus [16]. Therefore, we tested if the NES-tagged Moe isoform can also localize to the nucleus at high levels, by ectopically expressing GFP-labeled Moe and Moe NES proteins in the larval salivary gland cells (Fig. 7A). In contrast to WT Moe, MoeNES-GFP did not accumulate in the nuclei after heat shock, demonstrating that the NES-tag inhibits the increase in Moe level in the nucleus. The immunostaining of WT and moe NES larvae with an anti-Moe antibody revealed that, although the endogenous MoeNES protein is not completely eliminated from the nucleus, upon heat shock it fails to accumulate in the nucleus (Fig. 7B).
Moe has been shown previously to localize to the heat-shock puffs at cytological locations 87A and 87C [16]. To further confirm that Moe is responsible for the regulation of heat-shock gene transcription, we tested whether WT Moe localizes to the actively transcribing heat-shock genes identified in the transcriptome experiment. The activation of these genes by heat stress leads to the formation of chromosome puffs, which are special euchromatic regions of the polytene chromosomes with extremely high levels of transcription. We found that Moe accumulated in high levels to the heat-shock puff sites, as demonstrated by the immunostaining of chromosome preparations from salivary gland cells (Fig. 7C), providing additional evidence that Moe is involved in the transcription of hsp genes. Because in the cytoplasm Moe is organizing cytoskeletal structures, we wanted to exclude the possibility that Moe is needed for the formation of the puff structure. For this aim, we induced transcription with heat shock at a specific chromosomal location without the formation of the puff structure and stained for Moe. The animals used in this experiment carried a heat shock-inducible transgene at cytological location 86E in heterozygous form (Fig. 7D). The homologous chromosomes in the giant polytene chromosomes are paired in interphase and often separate from each other during the preparation at the nonhomologous sites. Without induction, no Moe staining can be observed in the corresponding region, but after heat shock, we found an extra Moe-positive band at this chromosomal location though only on one of the homologues (Fig. 7D). This suggests that Moe participates in hsp gene transcription and not the construction of the puff structure. To confirm this finding, we tested whether heat-shock puffs are formed in the moe NES mutant. Staining for polymerase II CTD phospho-serine 5 (Pol2-PS5) on heat-shocked polytenic chromosomes prepared from moe NES mutant salivary glands showed that puffs can develop after heat shock (Fig. 7E). The results of the chromosome staining experiments together with the mRNA-seq results prove that Moe participates in heat-shock gene transcription.

Discussion
To attain deeper knowledge about the nuclear functions of the Drosophila ERM protein, Moe, we tagged the moe gene in situ with an NES sequence and analyzed the consequences of the lack of nuclear Moe. The moe NES mutants exhibited diverse phenotypes that were of both zygotic and maternal origin. Maternal-effect phenotypes included posterior developmental malformations, dominant sterility, and lethality at embryonic and larval stages. Decreased life span, male genitalia rotation, reduced climbing ability, lethality at embryonic and larval stages, and decreased heat stress tolerance are zygotic phenotypes. Interestingly, decrease in viability at 18 and 25°C and in climbing activity was observed in moe NES mutant males only. Considering that the moe gene is on the first chromosome, these sex-specific phenotypic differences are

Moe Moe
Fluor. The X chromosome of hemizygous males presumably produces more MoeNES protein than that of the female, which therefore suggests a dominant-negative effect for the protein in the development of these phenotypes. The quantitation of Moe and P-Moe levels in the moe NES mutant ovary revealed that the phosphorylation, cytoplasmic localization, and intracellular distribution of the MoeNES protein are normal. The experiment also showed that the total Moe protein amount is reduced by about 30% in the mutant. This, in theory, could contribute to the phenotypes observed in the mutant. However, as shown earlier, the hypomorphic moe allele, EP1652, has a reduced Moe protein level [19] but moe[EP1652] homozygous animals are fertile and viable with no apparent visible phenotype, strongly suggesting that the decreased amount of Moe cannot be responsible for the phenotypes The subsequent control experiments demonstrated that the cytoplasmic functions of Moe are unaffected; therefore, the observed phenotypes are due to the perturbation of nuclear Moe function.
The mRNA-seq experiment revealed that the exclusion of Moe from the nucleus causes altered transcription of nearly 690 genes, from which 371 are significantly upregulated and 315 are downregulated in the mutant. Nuclear Moe was shown to participate in mRNA export [16] but it is conceivable that, similarly to the cytoplasm, Moe has multiple functions in the nucleus. In the case of the genes with decreased expression, Moe most likely plays a positive role in their transcription regulation, while in the case of upregulated genes, the protein is involved in their repression. Alternatively, the block of mRNA export, as a consequence of the lack of nuclear Moe, can also increase the activity of the genes affected, as suggested earlier [16], thereby accounting for the elevation of transcript levels.
The transcriptome analysis was performed on the ovary, but it is a reasonable assumption that part of these gene expression changes is not specific to the ovary. Moe is almost ubiquitously expressed in the animal at all developmental stages; therefore, the lack of nuclear Moesin should affect gene expression in other tissues as well. On the other hand, most of the genes with altered expression in the moeNES mutant are active in many other tissues throughout development. In fact, for example, the analysis of the heatshock genes provides supporting evidence for this assumption, because decreased heat stress tolerance was observed in the adults.
The mRNA-seq data showed that in moe NES mutant animals, the transcription of three key regulators of development, Notch, dpp, and vasa, is disturbed. Notch is a transmembrane receptor of a highly conserved cell signaling system which plays major role in the regulation of development. Notch mutants in Drosophila have characteristic wing margin 'notches', a phenotype also manifested in moe NES mutants (Fig. 3B). Dpp is a secreted morphogen necessary for the correct patterning and development of the early embryo. It has been shown recently that it is also essential for the maintenance of primordial germ cell identity [33]. Vasa is an evolutionarily conserved ATP-dependent RNA helicase that plays central role in germ cell determination and function [34]. Our microscopy experiments demonstrated that despite altered vasa expression in moe NES mothers and 100% sterility in their progenies, the Vasa protein localizes properly at the posterior pole of the developing embryo laid by moe NES mothers (Fig. 2D). Interestingly, we also found that the Vasa signal vanishes in the embryonic germ cells, and as a consequence, all germ cells disappear after cellularization in the embryos of moe NES mothers (C.B. unpublished result). Future experiments should shed light on the role of Dpp and Vasa in the molecular mechanism responsible for germ cell loss in moe NES progenies.
As a consequence of the changes in transcript levels, different molecular mechanisms can be responsible for the development of the observed phenotypes. For instance, the altered activity of the apoptotic  or sex-specific transcription factors can lead to genitalia rotation in males [35]. The complete loss of pole cells observed in the F1 progeny can be due to uncontrolled transposon activity [36], impaired sumoylation [37], or compromised BMP signaling [38]).
The exploration of the causes behind the various phenotypes was beyond the scope of this study but our mRNA-seq data uncovered that the transcription of a group of heat shock-inducible hsp genes is significantly reduced in the moe NES mutants. Drosophila hsp genes encode protein chaperones with versatile functions such as thermotolerance, normal life span, fertility and germline stem cell maintenance, locomotor activity, and developmental timing. Therefore, it is tempting to speculate that the low level of hsp gene transcription alone can contribute to most of the phenotypes observed in the moe NES mutant. Although the lack of nuclear Moe does not fully eliminate hsp gene expression, the combination of reduced expression of multiple hsp genes can account for the severity of the phenotypes observed. Given this possibility, as well as the findings that during heat stress Moe accumulates in the nucleus at the hsp loci and that moe NES mutants exhibit reduced thermotolerance, we propose that in the nucleus Moe is required for the normal expression of heat-shock genes. The mechanism through which Moe contributes to hsp transcription is still an open question. No DNA-binding ability of Moe has been described; therefore, the association of Moe with the hsp genes is most likely indirect. An obvious candidate for a binding partner of Moe is nuclear actin, which plays important role in transcription [1,39]. However, the exact activity Moe performs in the nucleus during hsp gene transcription has to be explored in the future. We believe that our results open up a new avenue to understanding the ERM protein family, their exact functions in growth and differentiation, and contribute to a better understanding of the transcription regulation of heat-shock genes.  , Hungary), and the oskMS2-MS-GFP and w,P{w + ,matTub4:GFP-Staufen}; stau D3 /CyO stocks were kind gifts from Daniel St Johnston (University of Cambridge, UK). The transgenic lines expressing Moe-GFP and Moe-HA have been described previously [16].

Fly stocks
To generate flies expressing NES and GFP (MoeNES-GFP)-tagged full-length Moe, the coding region was PCR amplified from the cDNA SD10366 (DGC Gold collection, BDGP) using the primer pair Moe_cDNSFw and Moe-NES_cDNSRev with gateway recombination sites. The resulting PCR product was sequence-verified and recombined into pDONR221 plasmid, then subcloned into the pPWG gateway vector (Drosophila Gateway Vector Collection) which enables P-element-based random integration into the fly genome. The construct was sequence-verified, and standard Drosophila methods were used to generate transgenic flies. Moe_cDNSFw

Drosophila ovary dissection and in vivo microscopy
Flies were dissected in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 ), and the ovaries were transferred to a microscope slide with a drop of Drosophila Schneider's medium (Lonza, Basel, Switzerland, Schneider's Drosophila Medium, Modified, 04-351Q), then covered with a glass coverslip. The samples were imaged with an Olympus confocal microscope (Olympus Europa SE & Co. KG, Hamburg, Germany, Olympus Fluoview FV1000 Confocal Microscope, 409 oil immersion objective, 1.3 NA).

Moesin antisera
GST-tagged FERM domain of the Drosophila Moe protein was expressed by the tac promoter-IPTG system using the pGEX-6P-1 vector and the BL21 E. coli strain. The protein was purified on Glutathione Sepharose 4B beads (GE Healthcare, Chicago, IL, USA, 17-0756-01); then, high level of soluble FERM-Moe protein was eluted from the beads by the PreScission protease (GE Healthcare, 27-0843-01) and used to immunize three rabbits, yielding three polyclonal sera (aFERM-Moe1, aFERM-Moe2, aFERM-Moe3). All animal experimentation protocols used in this study were performed in accordance with the European animal experimentation and ethics guidelines and have been Western blots of the lysate of cultured S2R + Drosophila cells were probed with a serial dilution of all three sera using the ECL system (Millipore, Darmstadt, Germany, Immobilon Western, WBKLS0100). To confirm specificity, V5-tagged, full-length Moe protein was detected with V5 antibody in parallel with the Moe antisera (Fig. 8A). Salivary glands expressing Moe-V5 were stained together with anti-V5 antibody and the Moe antisera to test for specificity in immunostaining (Fig. 8B). Aliquots of the serum number 1 (aFERM-Moe1) were used in the experiments.

Western blot
Cell lysates of dissected ovaries or cultured Drosophila S2R + cells were used. Before loading the samples on the gel, 1 µL beta-mercaptoethanol was added and the samples were boiled for 5 min and spun for 5 min with 18 000 g at

Physiological tests
In all of these experiments, flies were reared on standard Drosophila medium at constant temperature (25°C). In the hatching rate study, the number of hatched embryos, eclosed larvae and unhatched pharate adults was determined in a 2-week-old culture. For the statistical analysis of the data, we calculated mean values, standard deviations (SD), and performed unpaired t-tests to determine the significance of difference.

Embryonic lethality
Groups of 30 females of moe G0415 /w 1118 , moe NES /w 1118 , and moe NES /moe G0415 genotypes were collected on the same morning and crossed to w 1118 /Y males. Embryos were lined up every day in groups of 100 on black agar plate, and the hatched larvae, and unhatched eggs were counted.
Larval lethality moe G0415 /w 1118 , moe NES /w 1118 , and moe NES /moe G0415 females of the same age were crossed to w 1118 /Y males. After the synchronization of egg-laying, hatched larvae in groups of 20 were placed in separate vials containing fresh fly food. The number of 3rd-instar larvae, pupae, and hatched flies was recorded in the following 2 weeks.

Lifespan analysis
When testing the viability of adult animals, virgin males (w 1118 /Y and moe NES /Y) and virgin females (moe G0415 / w 1118 , moe NES /w 1118 , and moe NES /moe G0415 ) of the same age were tested separately in groups of five. In each experiment, 30 males and 30 females were analyzed by this way. The flies were transferred to fresh medium every day, and the

Starvation test
During the starvation test, 30 moe NES mutant and 30 control females were reared individually together with 3 male siblings. Flies were transferred to fresh medium every day, and the number of eggs deposited was counted. On the third and fourth days, sucrose starvation medium (1% agar, 0.5% propionic acidphosphoric acid (9 : 1), 5% sucrose) was used [40].

Fertility test
To determine the egg yield of moe NES mutant flies, moe G0415 /w 1118 , moe NES /w 1118 , and moe NES /moe G0415 virgin females were collected and crossed individually to w 1118 /Y. The flies were transferred to fresh medium every day, and the eggs were counted for 12 days.

Climbing assay
To test the climbing capability, w 1118 /Y and moe NES /Y virgin males were collected and kept at normal medium for 1 day or in other experiments for a week. Males were transferred to a transparent plastic vial with a centimeter scale, and the number of flies that were able to cover the 15-cm distance in 8 s was determined [41].
In situ gene tagging with CRISPR-Cas9
for eye-specific expression of the DsRed marker gene. Successful in vivo tagging was confirmed by sequencing the corresponding region in the moe gene and the moe NES mRNA.

Scanning electron microscopy
Adult flies were fixed with 2.5% (v/v) glutaraldehyde and 0.05 M cacodylate buffer pH 7.2 in PBS (pH 7.4) overnight. Samples were then washed twice with PBS and dehydrated with a graded ethanol series (30%, 50%, 70%, 80%, and 100% ethanol, each for overnight). The samples were dried with a critical point dryer, followed by 12-nm gold coating (Quorum Technologies, Laughton, East Sussex, UK) and observed under a field emission JEOL JSM-7100F/LV scanning electron microscope (JEOL Ltd., Tokyo, Japan).

Imaging of adult body parts
Adult flies were anesthetized with diethyl ether and arranged under the microscope (Lecia MZFL III). Images with different focal planes were focus stacked using the CombineZP freeware.

mRNA-Seq library preparation, sequencing, and analysis
Total RNA preparation was performed from three replicates of 3-day-old fly ovaries of 10 females/genotype (moe G0415 / w 1118 , moe NES /w 1118 , and moe NES /moe G0415 ) by using the TRIzol method as described in Ref. [16]. The mRNA-seq library was prepared using the TruSeq Stranded mRNA Library Prep Kit (Illumina, San Diego, CA, USA) and sequenced with the HiSEq 2500 sequencing platform (performed at the NYUAD Sequencing Center). All of the subsequent analyses, including quality trimming, were executed using the BioSAIL workflow execution system. The raw reads were quality trimmed using Trimmomatic (version 0.36) to trim low-quality bases, systematic base-calling errors, as well sequencing adapter contamination. FastQC was used to assess the quality of the sequenced reads pre/postquality trimming. Only the reads that passed quality trimming in pairs were retained for downstream analysis. The quality trimmed mRNA-seq reads were aligned to the D. melanogaster genome using HISAT2 (version 2.0.4) (http:// daehwankimlab.github.io/hisat2/). The resulting SAM alignment files for each sequenced sample were then converted to BAM format and sorted by coordinate using SAMTOOLS (version 0.1.19) (http://www.htslib.org/). The BAM alignment files were processed using HTseq-count using the reference annotation file to produce raw counts for each sample. The raw counts were then analyzed using the online analysis portal NASQAR (http://nasqar.abudhabi.nyu.edu/) in order to merge, normalize, and identify differentially expressed genes. Differentially expressed genes by at least twofold (log 2 (FC)≥1 and adjusted P-value of < 0.05 for upregulated genes, and log 2 (FC) ≤ À1 and adjusted P-value of < 0.05 for downregulated genes between the genotypes moe G0415 /w 1118 , moe NES /w 1118 , and moe NES /moe G0415 were subjected to GO database of Flybase (https://flybase.org/). Venn diagram was produced by using the 'Venn Diagrams' freeware developed by the Van de Peer Lab at the University of Gent, Belgium (http://bioinformatics.psb.ugent.be/webtools/Venn/). The RNA-seq data set has been deposited in the public repository GEO and can be retrieved using accession number GSE155778.

Quantification of pixel intensities and statistical analyses
Cytoplasmic-nuclear fluorescence intensity measurements were made as described in Ref [42]. In short: Cells were measured once using IMAGEJ (https://imagej/nih.gov/ij/). ROIs were drawn by hand in the entire nucleus and cytoplasm. Background fluorescence intensity was subtracted from the data prior to calculating the N/C ratio. In every measurement, 8-10 cells per salivary gland and 4-6 cells per egg chamber were examined. The GRAPHPAD PRISM 5 software (GraphPad Software Inc., San Diego, CA, USA) and unpaired t-test were used for statistical analysis. Error bar in bar charts indicates SD while in the boxplots represents minimum and maximum values. Statistical significance is marked with *P < 0.05, **P < 0.01, ***P < 0.0001, and n.s. (not significant).
For embryo size measurement, the length and width of 100 embryos of each genotype were measured with the ImageJ freeware. The length-width ratio was calculated for individual embryos, and the statistical analysis was performed with GRAPHPAD PRISM 5. For the quantitation of the protein levels, the signal intensities of individual bands on the western blot image were measured with the UVITEC Alliance software (Cambridge, UK). Data were normalized with the help of the actin loading control, and the ratio between Moe and P-Moe was calculated.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Table S1. mRNA-seq data and P-values of down-and upregulated genes.