A truncated form of a transcription factor Mamo activates vasa in Drosophila embryos

Expression of the vasa gene is associated with germline establishment. Therefore, identification of vasa activator(s) should provide insights into germline development. However, the genes sufficient for vasa activation remain unknown. Previously, we showed that the BTB/POZ-Zn-finger protein Mamo is necessary for vasa expression in Drosophila. Here, we show that the truncated Mamo lacking the BTB/POZ domain (MamoAF) is a potent vasa activator. Overexpression of MamoAF was sufficient to induce vasa expression in both primordial germ cells and brain. Indeed, Mamo mRNA encoding a truncated Mamo isoform, which is similar to MamoAF, was predominantly expressed in primordial germ cells. The results of our genetic and biochemical studies showed that MamoAF, together with CBP, epigenetically activates vasa expression. Furthermore, MamoAF and the germline transcriptional activator OvoB exhibited synergy in activating vasa transcription. We propose that a Mamo-mediated network of epigenetic and transcriptional regulators activates vasa expression.

T he germline is the only cell lineage that contributes to the production of the next generation. Accordingly, germline cells exhibit properties, including stemness and longevity, that are distinct from those of somatic cells. In animals, the germline is specified by maternal determinants or inductive signals 1 . Irrespective of the mode of specification, primordial germ cells (PGCs) express conserved germline genes. vasa (vas), which encodes a DEAD-box RNA helicase 2 , is a highly conserved germline gene, and its expression is a marker of germline establishment in many species 1 . For example, Ddx4, the human ortholog of Vas, is used as a molecular biomarker for germ cells 3 . Moreover, genetic studies have demonstrated that Vas is required for gametogenesis in Drosophila, C. elegans, zebrafish and mouse [4][5][6][7] . Vas has essential roles that contribute to germline development, including translational regulation of specific mRNAs and production of Piwi-interacting RNAs 8,9 . In Drosophila, maternal factors localised in the germ plasm are partitioned into PGCs to direct germline development. For example, the mitochondrial large ribosomal RNA and the Germ cell-less protein are both involved in PGC formation 10 . The maternal translational repressor Nanos is essential for repressing somatic differentiation of PGCs 11 . The Polar granule component protein has an essential role in PGC maintenance by inhibiting the positive transcription elongation factor b in newly formed PGCs 12 . The maternal Trapped in endoderm-1 protein is required for PGC migration through the midgut wall 13 . The germ plasm is necessary and sufficient for germ cell establishment 14 , suggesting that maternal transcription factors localised in the germ plasm activate vas expression in PGCs. Indeed, RNA interference-mediated knockdown of maternal mRNAs enriched in the germ plasm has revealed that six transcription factors are required for the expression of germline genes in PGCs 15 . Among them, the transcriptional activator OvoB, which is encoded by the ovo gene, is predominantly expressed in PGCs during embryogenesis, and its maternal function is essential for germline development 16 . However, the genes sufficient for vas activation remain unknown.
Previously, we showed that the maternal factor Mamo, which contains both a BTB/POZ domain and C 2 H 2 Zn-finger domains and is enriched in PGCs, is necessary for vas expression in PGCs 17 . Mamo protein is detectable in the nuclei of PGCs at stage 15, in which vas is actively expressed 17 . Mamo can bind chromosomes when Mamo is expressed in salivary gland cells 18 . Biochemical analyses demonstrated that the C 2 H 2 Zn-finger domains of Mamo directly bind specific DNA sequences 18 . C 2 H 2 Zn-finger domains are among the most common domains in the transcription factors of metazoans 19 . The Zn-finger domains of Mamo are homologous to those of human Sp1; 20 in vertebrates, Sp1-related transcription factors regulate gene expression in germ cells [21][22][23][24] . Accordingly, we focused on Mamo and investigated its role in vas activation in Drosophila embryos.
Biochemical analysis is an essential approach to understanding the molecular basis of transcriptional activation. To date, however, biochemical studies of vas activation have been difficult owing to the small number of PGCs in embryos. Accordingly, an experimental system capable of inducing forced expression of vas could be a powerful tool for analysing the mechanism of vas activation. In this study, we found that an N-terminal truncated Mamo isoform lacking the BTB/POZ domain but retaining the C 2 H 2 Zn-finger domains was a potent vas activator. Maternal overexpression of the truncated Mamo (MamoAF) was sufficient to activate vas expression in both PGCs and brain in embryos. Indeed, Mamo mRNA encoding a truncated Mamo isoform, which is similar to MamoAF, was predominantly expressed in PGCs, in which vas is actively expressed. Moreover, MamoAF could partially rescue the mamo mutant phenotype. Therefore, through biochemical and genetic analyses in which vas expression was induced by MamoAF, we elucidated the molecular mechanisms of vas activation. MamoAF directly bound the vas locus and, together with CBP, epigenetically activated vas expression. In addition, we identified the germline transcriptional activator OvoB as a cofactor of MamoAF. MamoAF physically interacted with OvoB, and the two proteins exhibited synergy in activating vas transcription. Together, these findings demonstrate that Mamo-mediated transcriptional regulation directly activates vas expression.

Results
Mamo short isoform may be associated with vas activation. To evaluate the molecular function of Mamo protein, we expressed a C-terminally FLAG-tagged allele of Mamo that can rescue the mamo mutant phenotype 17 in embryos, under the control of the maternal Gal4 driver (matα4-Gal-VP16). During embryogenesis, maternally expressed FLAG-tagged Mamo (~150 kDa) was processed into shorter proteins (~50 kDa) that lacked the N-terminal BTB/POZ domain but retained the C-terminal Zn-finger DNAbinding domains (Fig. 1a). Full-length Mamo, as well as its~100-kDa and~75-kDa derivatives, was detected in early embryos, in which zygotic vas expression is not yet activated (Fig. 1a, lanes 1 and 2). By contrast, in later embryos, in which zygotic vas expression is activated, the level of full-length Mamo was considerably reduced, and the~50-kDa derivatives of Mamo became detectable (Fig. 1a, lanes 3 and 4). These observations imply that maternally provided Mamo is gradually processed into shorter derivatives through specific proteolytic processes. Moreover, in contrast to full-length Mamo, short derivatives of Mamo preferentially accumulated in the nuclear fraction (Fig. 1b). Based on these observations, we hypothesised that the short isoform of Mamo, which lacks the BTB/POZ domain but retains the Znfinger domains, (hereafter, referred to as Mamo short isoform) is a potent vas activator.
Mamo mRNA encoding Mamo short isoform is expressed in PGCs. Next, we investigated whether Mamo mRNA encoding the Mamo short isoform is expressed in PGCs. Expressed sequence tags (GM29022 and GM29051) have shown that the short Mamo mRNA is expressed in ovaries 20 . 5′-rapid amplification of cDNA ends (RACE) confirmed that short Mamo mRNA is expressed in ovaries. The sequences of 5′-RACE clones revealed that the short Mamo mRNA encodes the Mamo short isoform, which lacks the BTB/POZ domain, and arises owing to the use of alternative transcription initiation sites ( Supplementary Fig. 1). To investigate the expression patterns of short Mamo mRNA, we performed in situ hybridisation using a Mamo Zn-finger probe, which detects both short and full-length Mamo mRNAs, and a Mamo BTB probe, which specifically detects the full-length Mamo mRNA ( Supplementary Fig. 2a). Mamo mRNA signals were detected in germ cells in the ovaries when using a Mamo Zn-finger probe or Mamo BTB probe ( Supplementary Fig. 2b), confirming that these probes are available. These results are consistent with previous studies showing that full-length Mamo mRNA is expressed in the ovary 17,20 . By contrast, during embryogenesis, Mamo mRNA signals were specifically detected in PGCs only when using a Mamo Zn-finger probe, but not a Mamo BTB probe ( Supplementary Fig. 2b). In the early embryos, at stages 5-9, no Mamo mRNA signal was detected, whereas at stages 14-16, the Mamo mRNA signals were more intense in PGCs, suggesting that Mamo mRNA encoding the Mamo short isoform lacking the BTB/POZ domain is zygotically expressed in PGCs, in which vas is actively expressed. Antibodies specific for the Mamo short isoform are not yet available, because the amino-acid sequence of the Mamo short isoform is present within that of full-length Mamo. These observations indicate that Mamo short isoform is predominantly expressed in PGCs, and that this isoform may promote vas expression.
MamoAF can partially rescue the mamo mutant phenotype. Because a mamo mutant specific for the short isoform was not available, we used overexpression experiments to evaluate the function of the short isoform. Previously, we generated a UAS line (UAS-MZD-FLAG) to produce a truncated Mamo protein lacking the BTB/POZ domain 18 (hereafter, referred to as MamoAF). When it was expressed in embryos under the control of matα4-Gal-VP16, MamoAF preferentially accumulated in the nuclear fraction (Fig. 1c), just like the short derivatives of Mamo . We speculated that the low rescue activity of Mamo may be due to the expression level of Mamo driven by nos-Gal4, which is different from the endogenous mamo promoter. Next, we investigated whether MamoAF overexpression can rescue the differentiation of mamo mutant germ cells into mature eggs. Overexpression of full-length Mamo, but not MamoAF, rescued the differentiation of mamo mutant germ cells into mature eggs (Supplementary Table 2). To assess the difference in rescue ability between full-length Mamo and MamoAF, we examined the expression patterns of full-length Mamo-FLAG or MamoAF-FLAG in ovaries when these proteins were expressed under the control of nos-Gal4. Full-length Mamo was enriched both in the nuclei of germline cysts and in the nuclei of the nurse cells in egg chambers after oogenic stage 6 (Supplementary Fig. 4). In 65.3% of egg chambers at stage 6, fulllength Mamo signals were enriched in the nuclei of the nurse cells (n = 121): 84.3% at stage 7 (n = 115). By contrast, MamoAF was detected in the nuclei of germline cysts, whereas MamoAF signals were weak in the nuclei of nurse cells in egg chambers (Supplementary Fig. 4). In 25.3% of egg chambers at stage 6, MamoAF signals were enriched in the nuclei of the nurse cells (n = 87): 18.1% at stage 7 (n = 72). Together, these findings show that MamoAF could partially rescue the differentiation of mamo mutant germline clones. Accordingly, we used overexpression of MamoAF to investigate the effect of the Mamo short isoform on vas activation.
MamoAF is sufficient to activate vas expression. To determine whether MamoAF can induce vas expression, we monitored vas expression in embryos overexpressing full-length Mamo or MamoAF under the control of matα4-Gal-VP16 driver. Quantitative reverse-transcription PCR (qRT-PCR) revealed that maternal overexpression of full-length Mamo only moderately increased vas expression at stages 9 and 15, whereas MamoAF markedly increased expression of vas at stage 15 (Fig. 1d). In addition, we observed somatic vas expression, especially in brain of embryos overexpressing MamoAF (hereafter, referred to as MamoAF-OE embryos), but not in embryos overexpressing fulllength Mamo (Fig. 1e, f). In MamoAF-OE embryos, PGCs formed normally at the posterior poles, and brain cells expressing Vas appeared at stage 14 ( Supplementary Fig. 5). These results show that MamoAF overexpression is sufficient to induce vas expression in brain.
Next, we investigated whether MamoAF can promote vas expression in PGCs. Specifically, we monitored vas expression in PGCs when MamoAF was overexpressed under the control of the maternal nos-Gal4 driver 16 . MamoAF overexpression increased vas expression in PGCs at stages 14 and 15 ( Supplementary  Fig. 6). We also found that overexpression of full-length Mamo increased vas expression in PGCs at stage 15 ( Supplementary  Fig. 6). These results indicate that the Mamo short isoform is a potent inducer of vas expression in PGCs.
The mamo gene produces two types of vas activators. Fulllength Mamo could promote vas expression in PGCs, but its activity was weaker than that MamoAF ( Supplementary Fig. 6). In embryos, full-length Mamo preferentially accumulated in the cytoplasmic fraction relative to MamoAF (Fig. 1b), which may suppress its transcriptional activity. Thus, the BTB/POZ domain may regulate the nuclear localisation of full-length Mamo. Consistent with this notion, overexpression of full-length Mamo did not strongly induce vas expression in brain (Fig. 1e, f). These results suggest that full-length Mamo is a weak but specific activator of vas whose activity is restricted in PGCs. By contrast, MamoAF lacking the BTB/POZ domain preferentially accumulated in the nuclear fraction (Fig. 1c), and potently induced vas in PGCs and brain. Vas mRNA becomes detectable in PGCs at stage 9, when short Mamo mRNAs are not yet zygotically expressed in PGCs. Thus, we propose that maternal full-length Mamo establishes specific vas expression in early PGCs at stage 9. In later PGCs, after stage 14, zygotic Mamo short isoform contributes to the maintenance of robust vas expression. We found that, during embryogenesis, maternally expressed full-length Mamo was processed into shorter proteins (Fig. 1a). Thus, in addition to zygotic Mamo short isoform, the conversion from maternal fulllength Mamo into short derivatives may promote vas expression in late PGCs.
Zn-finger domain of MamoAF is required for vas activation. To investigate the mechanisms underlying vas activation, we focused subsequent experiments on MamoAF owing to its potency as a vas activator. To evaluate the functional domains of MamoAF, we examined vas expression in embryos expressing fragments of the protein. Fragments of MamoAF proteins that Fig. 1 MamoAF overexpression induces vas expression in embryos. a Developmental western blotting analysis of embryos expressing full-length Mamo-FLAG under the control of matα4-Gal-VP16 driver. Embryos were cultured at 18°C for indicated hours and subjected to western blotting with anti-FLAG antibody. Embryos cultured at 18°C for 0, 6, 16 and 24 h are corresponded to embryonic stage 4, 9, 12 and 15, respectively. b Analysis of fractionated samples extracted from embryos expressing full-length Mamo-FLAG by western blotting. Embryos expressing Mamo-FLAG under the control of matα4-Gal-VP16 driver were cultured at 25°C for 0-16 h and fractionated. Normalised load of each extract (7 μg) were analysed by western blotting with anti-FLAG, anti-Tubulin and anti-Histone H3 antibodies. Whole extract (lane 1), cytoplasmic extract (lane 2) and nuclear extract (lane 3) were loaded. c Analysis of fractionated samples extracted from embryos expressing MamoAF-FLAG by western blotting. Embryos expressing MamoAF-FLAG under the control of mat4α-Gal-VP16 driver were cultured at 25°C for 0-16 h and fractionated. Normalised load of each extract (11 μg) were analysed by western blotting with anti-FLAG, anti-Tubulin and anti-Histone H3 antibodies. Whole extract (lane 1), cytoplasmic extract (lane 2) and nuclear extract (lane 3) were loaded. Full images of the electrophoreses are shown in Supplementary Fig. 16. d qRT-PCR analyses of Vas mRNA in wild type, Mamo and MamoAF overexpressed embryos under the control of matα4-Gal-VP16 driver. Error bars are SD of the mean. **P < 0.01, *P < 0.05 (two-tailed Student's t test). e Double fluorescence in situ hybridisation for Vas and Elav mRNAs of wild type, the embryos overexpressing Mamo, MamoAF and MamoAF fragments under the control of matα4-Gal-VP16 driver. Elav is a neural marker. Confocal sections of the brain of the embryos are shown. The inset shows close-up image of wild-type PGCs. MamoAF directly binds and epigenetically activates vas. To determine whether MamoAF directly binds vas, we performed electrophoretic mobility shift assays (EMSA) and chromatin immunoprecipitation assays (ChIP) on MamoAF-OE embryos. Although the upstream region (~2 kbp) of the vas gene does not contain cis-elements containing a Mamo-binding consensus sequence 18 , the first intron of vas contains several possible ciselements, suggesting that MamoAF activates vas through previously unidentified cis-elements in the first intron. One such ciselement, hereafter referred to as the vas-A element, is conserved among several Drosophila species (Supplementary Fig. 7). Hence, we focused subsequent experiments on the vas-A element.
EMSA revealed that the Zn-finger domains of MamoAF directly bound the vas-A element in the first intron of vas (Fig. 2a,  b), and ChIP-qPCR assays confirmed that MamoAF bound to this sequence in vivo (Fig. 2c). Next, we investigated whether the vas-A element is required for vas expression. After removal of the Mamo-binding sequence by gene editing, vas expression in PGCs was considerably reduced ( Fig. 2d; Supplementary Fig. 8). Vas mRNA signals were detected in PGCs of 86% of wild-type embryos (n = 104) and 77% of embryos heterozygous for vas d10-3 (n = 220), but only 23% of embryos homozygous for vas d10-3 , which lacks the vas-A element (n = 158, P < 0.01). Together, these results demonstrate that MamoAF directly activates vas through the vas-A element.
MamoAF works with CBP to epigenetically activate vas. We hypothesised that the vas-A element acts as an enhancer. If so, acetylated H3K27 (H3K27ac), which marks active enhancers [25][26][27] , should accumulate in this region. ChIP assays revealed greater c ChIP-qPCR assays were performed with MamoAF-OE embryos and FLAG antibody. The precipitated DNA was detected by qPCR with primers targeting to vas-A. Error bars are SD of the mean. **P < 0.01 (two-tailed Student's t test). d Vas mRNA in situ hybridisation of wild type, vas d10-3 / + and vas d10-3 / vas d10-3 embryos at stage 10. The Mamo-binding consensus sequence (red box) is deleted from vas-A element in vas d10-3 / vas d10-3 embryo. Scale bar: 50 μm. e ChIP-qPCR assays were performed with MamoAF-OE embryos and H3K27ac antibody. The precipitated DNA was detected by qPCR with primers targeting to vas-A. Error bars are SD of the mean. *P < 0.05 (one-tailed Student's t test). f qRT-PCR analyses of Vas mRNA in wild type, MamoAF, CBP and both CBP and MamoAF overexpressed embryos. Error bars are SD of the mean. *P < 0.05, **P < 0.01 (two-tailed Student's t test). g Vas mRNA in situ hybridisation of wild type and CBP Membryo derived from nej 0.3 mutant germline clones at stage 9. White arrowheads indicate PGCs expressing Vas mRNA. Scale bar: 50 μm. h Percentages of embryos expressing Vas mRNA in wild type and CBP Membryos at stage 9. *P < 0.01 (Fisher's exact test) enrichment of H3K27ac at the vas-A element in MamoAF-OE than in wild-type embryos, suggesting that MamoAF activates vas-A through H3K27ac (Fig. 2e). Because MamoAF lacks a histone acetyltransferase (HAT) domain, MamoAF must collaborate with CBP, which is responsible for acetylation of H3K27 in Drosophila 25 . Accordingly, we investigated the genetic interaction between MamoAF and CBP. Because CBP is essential for embryogenesis, we examined the outcome of CBP overexpression. Overexpressing both CBP and MamoAF synergistically increased vas expression, whereas CBP overexpression alone did not (Fig. 2f). These results demonstrate that MamoAF collaborates with CBP to epigenetically activate vas.
Next, we investigated whether CBP is required for vas expression in PGCs. In Drosophila, CBP is encoded by nejire (nej) 25 . In embryos derived from nej germline clones, maternal CBP activity was required for both H3K27ac accumulation ( Supplementary Fig. 9) and vas expression in PGCs (Fig. 2g, h). These results show that vas expression in PGCs is epigenetically activated through H3K27ac. These findings are consistent with our idea that Mamo collaborates with CBP to epigenetically activate vas in PGCs.
MamoAF collaborates with OvoB. In addition to CBP, we assumed that a transcriptional activator collaborates with MamoAF for vas activation, because MamoAF lacks a transcriptional activation domain. The transcriptional activator OvoB, which is maternally expressed and maintained predominantly in PGCs, is required for vas expression 15,16 , making it a candidate cofactor of Mamo. However, because OvoB activity is essential for oogenesis 28 , loss-of-function alleles of ovo are not available for investigating the role of maternal OvoB in vas activation. To overcome this problem, we used MamoAF-induced vas expression to determine whether OvoB cooperates with MamoAF to promote vas expression. If OvoB is essential for vas expression, then MamoAF should induce OvoB expression in brain. We examined the effects of MamoAF on OvoB expression using the ovoB-Nterm-egfp knock-in allele, which enabled us to detect OvoB as GFP fluorescence 16 (Supplementary Fig. 10a). As expected, MamoAF overexpression induced GFP expression from the ovoB-Nterm-egfp allele, especially in brain expressing vas (Fig. 3a). The ovo locus produces three proteins: OvoA and OvoB, which serve, respectively, as a transcriptional repressor and activator in germ cells, and Shavenbaby (Svb), which regulates epidermal differentiation 29 . In situ hybridisation of MamoAF-OE embryos detected Ovo mRNA signals in brain when using an Ovo common probe, but not an Svb-specific probe ( Supplementary  Fig. 10b-q). Because OvoA is a transcriptional repressor, MamoAF probably induces OvoB in brain expressing vas. Although the mechanism remains elusive, these observations show that MamoAF induces OvoB expression in brain. Therefore, MamoAF may collaborate with OvoB transcriptional activator to promote vas expression in brain.
Because MamoAF induces OvoB expression in brain (Fig. 3a,  Supplementary Fig. 10b-q), OvoB may act downstream of MamoAF. Thus, we tested whether OvoB is sufficient to induce Vas expression in brain. We evaluated vas expression in embryos overexpressing OvoB. Strong Vas signals in brain were detected in MamoAF-OE embryos (58.3%, n = 127), but not in OvoB-OE embryos (n = 71). OvoB overexpression induced much weaker somatic Vas expression (35.2%, n = 71) than MamoAF-OE embryos (Fig. 3b). This observation indicates that OvoB alone is not sufficient to induce strong Vas expression in brain. Vas expression in brain was increased in embryos expressing both MamoAF and OvoB compared with embryos expressing OvoB alone. The increased Vas expression in brain appears to be owing to the enhancement of OvoB expression induced by MamoAF. However, Vas expression in brain of embryos expressing both MamoAF and OvoB (66.7%, n = 63) was not significantly increased relative to embryos expressing MamoAF alone (58.3%, n = 127) (Fig. 3b). This indicates that the promotion of vas expression in brain cannot be explained by the increased expression of OvoB alone. These results suggest that MamoAF collaborates with OvoB to promote vas expression in brain.
Given that the collaboration between MamoAF and OvoB are essential for vas activation, overexpression of both MamoAF and OvoB should activate vas expression in somatic cells, where MamoAF cannot induce OvoB. Indeed, overexpression of both MamoAF and OvoB synergistically induced Vas expression in somatic cells in the ventral nerve cord (VNC) (55.6%, n = 63), a region where no expression was observed in embryos expressing MamoAF or OvoB alone (n = 127, n = 71, respectively) (Fig. 4). Overexpression of both MamoAF and OvoB is sufficient for activating vas expression in VNC. However, other somatic cells exhibited little response to MamoAF and OvoB. Thus, vas expression in somatic cells is dependent on their cellular contexts. Although the pertinent differences in cellular context remain elusive, these observations show that both MamoAF and OvoB are essential for vas activation both in brain and VNC.
Next, we investigated whether Mamo collaborates with OvoB to promote vas expression in PGCs. Because OvoB activity is essential for oogenesis 28 , genetic approaches using loss-offunction alleles of ovo are not available for investigating the role of maternal OvoB in PGCs 16 of progeny. To circumvent this obstacle, we paternally introduced the UASp-ovoA transgene into embryos derived from female adults carrying both UASp-MamoAF and the nos-Gal4 driver. We found that OvoA suppressed MamoAF-induced vas expression in PGCs (Fig. 5a,  b), and that OvoB overexpression was sufficient to promote vas expression (Fig. 5c, d). These results show that Mamo collaborates with OvoB to promote vas in PGCs.
If MamoAF collaborates with OvoB to activate vas, Ovo enhancers 30 should promote vas expression. In addition to the vas-A element required for Mamo binding (Fig. 2b), the first intron of vas contains Ovo-binding consensus sequences (Ovo1 and Ovo2) 30 (Fig. 6a), which are conserved in several Drosophila species (Supplementary Fig. 7). To investigate whether these ciselements have enhancer activity, we performed luciferase reporter assays in Drosophila Schneider cells (S2 cells). We evaluated the effects of fragments of the first intron containing each cis-element (Ovo1, Ovo2 and vas-A) on luciferase expression, and found that none of the fragments were sufficient to promote luciferase expression in response to MamoAF and OvoB ( Supplementary  Figs 11, 12). These results suggested that, individually, none of the cis-elements has enhancer activity. Next, we examined combinations of these fragments containing the cis-elements. Although a reporter construct containing both Ovo2 and vas-A (Ovo2 + vas-A luc) did not exhibit enhancer activity ( Supplementary Fig. 12d), a reporter construct containing both Ovo1 and vas-A (Ovo1 + vas-A luc) promoted luciferase expression in response to both MamoAF and OvoB (Fig. 6b). Furthermore, mutation analyses of Ovo1 and vas-A demonstrated that both Ovo1 and the vas-A element were required to stimulate transcription (Fig. 6c-e). Consistent with this, both MamoAF and OvoB were required to promote luciferase expression mediated by Ovo1 and vas-A (Fig. 6b). We concluded that MamoAF collaborates with OvoB to activate vas through Ovo1 and the vas-A element. Furthermore, CBP enhanced luciferase expression mediated by MamoAF and OvoB, suggesting that these factors worked together to stimulate transcription (Fig. 6f).
MamoAF physically interacts with OvoB. Next, we investigated the physical interaction between MamoAF and OvoB. Coimmunoprecipitation assays with nuclear extracts of S2 cells transfected with both FLAG-tagged MamoAF and V5-tagged OvoB revealed a physical interaction between MamoAF and OvoB ( Fig. 6g; Supplementary Figs 13-15). We found that the N-terminal region encompassing residues 251-277 of MamoAF is necessary for the physical interaction with OvoB ( Supplementary Fig. 14). Conversely, the region encompassing residues 470-757 of OvoB is required for the interaction with MamoAF ( Supplementary  Fig. 15). These results, together with the data from the luciferase reporter assay, demonstrate that MamoAF and OvoB exhibit transcriptional synergy in activating vas expression.

Discussion
Although it is well recognised that maternal translational and transcriptional repressors play essential roles in establishing PGCs in Drosophila 11,12 , the genes sufficient for vas activation in PGCs remain unknown. In this study, we identified two types of vas activators: full-length Mamo, a weak but specific inducer of To clarify the molecular mechanisms of vas activation, we conducted biochemical and genetic analyses using MamoAF-induced vas expression, and revealed that two cofactors of MamoAF, CBP and OvoB, are both involved in activation of vas in PGCs. Thus, MamoAF-induced vas expression is useful for identifying cofactors of vas activation in PGCs. MamoAF can induce vas expression in both PGCs and brain. In both cellular contexts, the transcriptional activator OvoB is necessary for MamoAF-induced vas expression. Moreover, overexpression of both MamoAF and OvoB is sufficient to induce vas expression in VNC. Thus, the Mamo-OvoB axis is essential for directing vas activation. Our  b Embryos were classified into three groups depending on their strong (++), middle (+) and low (−) signal intensities of Vas mRNA in PGCs. Percentages of the embryos carrying PGCs with strong, middle and low signals in wild type, MamoAF-OE and MamoAF + OvoA OE embryos. **P < 0.01 (Significance is calculated with MamoAF OE by Fisher's exact test). c Vas mRNA in situ hybridisation of wild type and OvoB overexpressed embryos under the control of maternal nos-Gal4 driver at stage 15. White arrowheads indicate embryonic gonads. d Embryos were classified into three groups depending on their strong (++), middle (+) and low (−) signal intensities of Vas mRNA in PGCs. Percentages of the embryos carrying PGCs with strong, middle and low signals in wild type and OvoB-OE embryos. *P < 0.05 (Significance is calculated with wild type by Fisher's exact test). Scale bar: 20 μm data revealed that MamoAF functions as a molecular hub: it collaborates with CBP to epigenetically activate the vas locus, and physically interacts with OvoB to stimulate vas transcription (Fig. 6h). Consistent with this notion, a reporter assay demonstrated that these factors worked together to stimulate transcription. We conclude that the Mamo-mediated network of epigenetic and transcriptional regulators directs vas activation in Drosophila embryos.
We successfully demonstrated that MamoAF directly activates vas expression through the vas-A element in the first intron, which is essential for endogenous vas expression in PGCs. However, MamoAF could also activate vas transcription through other cis-elements that remain to be identified. We previously reported that Mamo has a role in the regulation of chromatin structure 18 . Therefore, Mamo may regulate chromatin structure, in addition to transcriptional activation, to promote vas expression.
A previous study using reporter assays showed that a 40-bp element in the 5′ flanking region of vas, the up-40 element, is sufficient to recapitulate germline-specific expression during oogenesis and embryogenesis 31 . The up-40 element does not contain consensus sequences for either Mamo 18 or Ovo 30 . This implies that different transcriptional factors must control vas expression through the up-40 element. Because removal of the vas-A element decreases endogenous vas expression in PGCs, vas-A and up-40 elements may act in a partially redundant manner to upregulate vas expression in PGCs. Therefore, multiple enhancers may act in parallel to activate vas expression in germ cells.
However, it remains unclear whether the up-40 element is necessary for endogenous vas expression.
In this study, we focused on MamoAF to investigate the mechanisms of vas activation due to its potent activity for vas activation. Thus, it remains unclear how full-length Mamo activates vas expression. Understanding the mechanism requires the identification of factors regulating the nuclear localisation of fulllength Mamo. Maternal full-length Mamo in the nuclei of early PGCs may interact with OvoB, which is maternally provided and enriched in PGC nuclei 16 . However, it will be necessary to investigate the interaction between full-length Mamo and OvoB under conditions in which full-length Mamo is not converted into short derivatives.
Both full-length and short form Mamo mRNAs are expressed in germ cells in ovaries. However, full-length Mamo rescues the differentiation of mamo mutant germline clones more efficiently than MamoAF. We found that full-length Mamo can be converted to truncated derivatives. Thus, full-length Mamo has the potential to complement the short isoform. By contrast, the short form of Mamo does not appear to complement the function of the full-length protein owing to the lack of the BTB/ POZ domain. During oogenesis, the transcriptional activity of full-length Mamo may be regulated through interactions with epigenetic regulators via the BTB/POZ domain, as reported for other BTB/Zn-finger transcription factors 32 . Full-length Mamo is enriched in the nuclei of the nurse cells in egg chambers after oogenic stage 6, but MamoAF is only weakly enriched. Thus, full-length Mamo may play a role in regulating gene expression We propose that Mamo short isoform is a potent vas activator. The C 2 H 2 Zn-finger domains of Mamo short isoform are homologous to those of human Sp1; 20 Sp1-related transcription factors regulate gene expression in germ cells in vertebrates [21][22][23] . For example, chicken Sp1 promotes vas expression in PGCs 24 . Moreover, Ovo has conserved roles in germline development in mouse and Drosophila 16,33 . Accordingly, Sp1-related Zn-finger proteins and Ovo may also be key transcriptional activators of vas in other animals, including mouse and human. We anticipate that our results will facilitate understanding of the molecular mechanisms that regulate germline development in animals.
Expression vectors for Mamo fragments in embryos. The fragment of s-MZD-FLAG was amplified from the pBluescript-mamo-FLAG plasmid 17  Developmental western blotting of Mamo OE embryos. Mamo OE embryos were collected at 25°C for 3 h, then the embryos were incubated at 18°C for 0, 6, 16 and 24 h, respectively. The embryos were dechorionated and homogenised in the sample buffer, and then the samples (40 embryos/lane) were fractionated using a 5-20% gradient sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) gel (c520L, Atto) and analysed by western blotting with anti-FLAG antibody.
Analyses of fractionated Mamo OE and MamoAF OE embryos. Mamo OE and MamoAF OE embryos were collected at 25°C for 0-16 h. Approximately 0.1 g of the embryos were dechorionated and homogenised using a Dounce tissue grinder in the sucrose solution (0.5 M sucrose, 5 mM HEPES, pH7.9, 5 mM KCl, 1.5 mM MgCl 2 , 0.25 mM DTT, 0.25 mM phenylmethylsulfonyl fluoride (PMSF) supplemented with Protease inhibitor complete EDTA-free protease inhibitor cocktail (Roche)). Debris was removed by centrifugation at 390 × g for 10 min at 4°C and the supernatant was recovered. The supernatant was centrifuged at 6300 × g for 10 min at 4°C. The resulting supernatant was used for cytoplasmic extract, the pellet was suspended in the sucrose solution and used for nuclear extract. Normalised load of each extract was fractionated using a 5-20% gradient SDS-PAGE gel, and then analysed by western blotting with anti-FLAG, anti-Tubulin and anti-Histone H3 antibodies.
Analyses of germline clones. To investigate the maternal function of CBP, we generated germline clones. We introduced nej 3 or nej 0.3 mutations into the chromosome carrying FRT (BL1844) by meiotic recombination. The nej 3 /Binsinscy and nej 0.3 /Binsinscy females were mated with w ovo D1 v 24 P{FRT}101;P{hsFLP}38 males, respectively. Their progenies were then subjected to heat shock at 37°C for 1 h in a water bath. Because the nej 3 germline clones were degenerated during midoogenesis, we used CBP Membryos derived from nej 0.3 germline clones were analysed by in situ hybridisation and immunostaining.
In situ hybridisation to detect Vas mRNA. To detect Vas mRNA in CBP Membryos and the embryos homozygous for vas d10-3 at stage 9-10, embryos were collected, dechorionated with sodium hypochlorite for 3 min, then fixed with PBS containing 4% formaldehyde/heptane for 25 min. PBS phase containing formaldehyde was discarded. The vitelline membranes were removed by adding methanol and vigorous shaking. Because In situ hybridisation to detect the short Mamo mRNA. To make a template for a Mamo BTB/POZ RNA probe, PCR was performed full-length Mamo cDNA as a template using Mamo FL probe Pst F (5′-CAGCTGCAGATGGGCAGTGAG-3′) and Mamo FL probe R (5′-CATGGTGGTGCGTGTGATGG-3′) primers. The resulted PCR product was subcloned and used as a template for producing a Mamo BTB/POZ RNA probe. To make a template for a Mamo Zn-finger RNA probe, PCR was performed MZD-FLAG cDNA 18 as a template. The resulted PCR product was subcloned and used as a template for producing a Mamo Zn-finger RNA probe. Wild type embryos were collected, dechorionated with sodium hypochlorite for 15 s, then fixed with PBS containing 4% paraformaldehyde/heptane for 20 min. PBS phase containing formaldehyde was discarded. The vitelline membranes were removed by adding methanol and vigorous shaking. In situ hybridisation was performed as described above. To detect Mamo mRNAs encoding Mamo shortisoforms, hybridised embryos were incubated with colour developing solution containing NBT-BCIP at room temperature for 3 h in the dark.
Double fluorescence in situ hybridisation (double FISH). To detect ectopic Vas and Ovo mRNA expression in MamoAF-OE embryos, we performed double FISH. To confirm whether ectopic vas and ovo were expressed in brains, the anti-sense Elav RNA probe, which can specifically detect neural cells 34 , was used. The template for Elav probe was amplified by PCR using Elav probe-F Eco (5′-CGGAATTCGCTAATGCAGAGTGCCGCTG-3′) and Elav probe-R Eco (5′-CGGAATTCCAGACCCTTGTTAACTGGCG-3′) primers from wild type genomic DNA and cloned into pBSK. Fluorescein isothiocyanate (FITC)-labelled anti-sense Elav RNA probe was synthesised from the template using the Fluorescein RNA Labeling Mix (Roche). We used both DIG-and FITC-labelled probes. The DIGlabelled probes for Vas and Ovo mRNAs were visualised by the anti-DIG alkaline phosphatase-conjugated antibody in combination with HNPPD/Fast Red TR (Roche). For FITC-labelled probe for Elav mRNA, the components and protocols of the TSA Plus Fluorescein kit (Perkin Elmer), including the anti-FITC horseradish peroxidase-conjugate antibody and FITC-tyramides as substrates, were used. For double FISH, the embryos were hybridised at the same time using DIG-and FITC-labelled anti-sense RNA probes in hybridisation solution (50% formamide, 5× SSC, 10% dextransulfate, 0.1% Tween-20, 0.1 mg ml −1 yeast RNA, 0.1 mg ml −1 heparin, 10 mM DTT) at 60°C for 16 h, washed with solution 2 at 60°C for 30 min three times, at room temperature for 10 min three times, and then with wash solution three at room temperature for 10 min three times. The embryos were washed with TNT Buffer (0.1 M Tris-HCl, pH7.5, 0.15 M NaCl, 0.05% Tween20) for 10 min three times and blocked with TNB buffer (Perkin Elmer). The embryos were reacted with anti-FITC horseradish peroxidase-conjugate antibody at 1:4000 at 4°C overnight. The signals of FITC-labelled probe were visualised by incubating the embryos first for 30 min with the amplification solution of TSA kit, and then washed with TNT Buffer (0.1 M Tris-HCl, pH7.5, 0.15 M NaCl, 0.05% Tween20) for 10 min three times. The embryos were incubated with anti-DIG alkaline phosphatase-conjugated antibody at 1:3000 for 1 h. The signals of DIG-labelled probe were visualised by incubating HNPPD/Fast Red TR for 30 min twice. The embryos were mounted with CC/Mount (Diagnostic BioSystems) and imaged with a confocal microscope (LSM 800, Zeiss) within 2 days.
Rescue experiments. mamo SVA53 and UASp-Mamo-FLAG has been described previously 17 . To investigate whether MamoAF can rescue mamo mutation, we generated germline clones of mamo SVA53 . Unfortunately, the w ovo D1 v 24 P{FRT} 101;P{hsFLP}38 strains (BL1813) in both the Bloomington Stock Center and the Genetic Resource Center in Kyoto Institute of Technology lost the hsFLP transgene, which is essential to produce germline clones. We crossed the strain with other hsFLP strain (Kyoto 108127) to obtain the strain transiently carrying the hsFLP transgene. Then, we crossed the strain with nos-Gal4 to obtain w ovo D1 v 24 P {FRT}101;P{hsFLP}; nos-Gal4 strain. We crossed w ovo D1 v 24 P{FRT}101/Y;P {hsFLP}; nos-Gal4 males with mamo SVA53 /Binsinscy, mamo SVA53 /Binsinscy carrying UASp-Mamo-FLAG and mamo SVA53 /Binsinscy carrying UASp-MamoAF-FLAG females, respectively. Their progenies were then subjected to heat shock at 37°C for 1 h in a water bath. The resulted females without Binsinscy balancer chromosome were crossed with y w males to investigate their progenies. The eggs of the females were collected for 3 h. The number of the eggs and their hatching rate were scored. We found that mamo phenotype was rescued by expressing Mamo-FLAG. But the phenotype of germline clones was different from that previously reported 17 . We could not obtain the mature eggs derived from germline clones homozygous for mamo SVA53 allele in this experimental condition. This difference is probably owing to w ovo D1 v 24 P{FRT}101; P{hsFLP}38 strain described above. FLP activity may be weaker than original BL1813. Moreover, genetic background of w ovo D1 v 24 P{FRT} 101; P{hsFLP}38 strains may affect the efficiency of germline clone production.
Immunostaining. The embryos were collected and fixed as described above. The fixed embryos were washed with PBS containing 0.1% Triton X-100 (PBTx), blocked PBTx containing 5% goat serum for 1 h and then reacted with primary and secondary antibodies. For primary antibodies, a rabbit anti-Vas antibody at 1:500 from Dr. S. Kobayashi, a mouse anti-Elav monoclonal antibody (Developmental Studies Hybridoma Bank) at 1:100, and a mouse anti-GFP monoclonal antibody (3E6, Thermo Fisher) at 1:300 were used. A mouse anti-H3K27ac monoclonal antibody at 1:500, a mouse anti-H3K9ac monoclonal antibody at 1:500, a mouse anti-H4K16ac monoclonal antibody at 1:500, from Dr. H. Kimura 35 were used. A mouse anti-Piwi antibody at 1:200 from Dr. M. Shiomi was used to detect PGCs 36 in Supplementary Fig. 5. AlexaFlour-conjugated antibodies (Molecular Probes) at 1:1000 were used for secondary detection. DNA was stained with DAPI. Stained embryos were mounted with Vectashield (Vector laboratories) and imaged with a confocal microscope (FV2000, Olympus or LSM 800, Zeiss). Fluorescent imaging was acquired with Zen 2.1 Software (Zeiss) or the software for the FV1200 (Olympus).
Immunostaining of ovaries with mamo mutant germline clone. Rescue experiments were performed as described above. To investigate the rescue ability of Mamo and MamoAF, we stained the ovaries containing mamo mutant germline clone expressing full-length Mamo-FLAG or MamoAF-FLAG. Germline clones were generated as described above. Females carrying mamo mutant germline clones were dissected. The ovaries were fixed with PBS containing 4% paraformaldehyde for 20 min. The ovaries were washed with PBS containing 0.1% Triton X-100 (PBTx), blocked PBTx containing 5% goat serum for 1 h and then reacted with primary and secondary antibodies. For primary antibodies, a rabbit anti-Vas antibody at 1:500. AlexaFlour-conjugated antibodies (Molecular Probes) at 1:1000 were used for secondary detection. DNA was stained with DAPI. Stained ovaries were mounted with Vectashield (Vector laboratories) and imaged with a confocal microscope (FV2000, Olympus). Fluorescent imaging was acquired with the software for the FV1200 (Olympus).
Detection of ectopic Vas expression. Ectopic Vas expression was detected in heterogenous cell population in embryonic brain and VNC. To gain the image, a single section image was used. Ectopic Vas signals were normalised by DAPI signals.
ChIP analyses. Dechorionized wild type embryos and MamoAF-OE (~0.15 g) were homogenised with Dounce tissue grinder in the buffer (15 mM HEPES, pH 7.5, 15 mM NaCl, 60 mM KCl, 4 mM MgCl 2 , 0.5% Triton X-100, 0.5 mM DTT, complete EDTA-free protease inhibitor cocktail (Roche)) containing 0.8% Luciferase reporter assays. The luciferase reporter constructs containing the cis-elements in the first intron of vas described above were used for luciferase reporter assays. The hRluc/TK (Promega, pGL4.74) served as an internal control. About 50 ng of a luc reporter construct and 1 ng of an internal control were cotransfected into S2 cells (3 × 10 4 cells) with either 10 ng of pAC-MamoAF-FLAG and/or 1 ng of pAC-OvoB and/or 1 ng of pAC-CBP and/or pBSK (for mocked transfected cells) using Effectene transfection reagent (Qiagen). The transfected cells were cultured in the culture medium supplemented with Penicillin-Streptomycin-Amphotericin B Suspension (Wako). Cells were lysed after 1 day of transfection with a passive lysis buffer (Promega) and luciferase activity was measured using a Dual-Luciferase reporter assay system (Promega). All experiments were carried out in triplicate.
Statistics and reproducibility. Statistical tests were employed for each experiment are indicated in the figure legends. Microsoft Excel was used for statistical analyses. For all data in these analyses, P values of < 0.05 were considered to be significant. Replicate experiments were successful.
Reporting summary. Further information on experiments and research design are available in the Nature Research Reporting Summary linked to this article.

Data availability
The data sets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. The source data underlying plots are provided in Supplementary Data 1. Full blots are shown in Supplementary Information. The sequencing data confirming vas mutants were deposited into the DNA Data Bank of Japan under the accession number LC503775.