The tumour suppressor brain tumour (Brat) regulates linker histone dBigH1 expression in the Drosophila female germline and the early embryo

Linker histones H1 are essential chromatin components that exist as multiple developmentally regulated variants. In metazoans, specific H1s are expressed during germline development in a tightly regulated manner. However, the mechanisms governing their stage-dependent expression are poorly understood. Here, we address this question in Drosophila, which encodes for a single germline-specific dBigH1 linker histone. We show that during female germline lineage differentiation, dBigH1 is expressed in germ stem cells and cystoblasts, becomes silenced during transit-amplifying (TA) cystocytes divisions to resume expression after proliferation stops and differentiation starts, when it progressively accumulates in the oocyte. We find that dBigH1 silencing during TA divisions is post-transcriptional and depends on the tumour suppressor Brain tumour (Brat), an essential RNA-binding protein that regulates mRNA translation and stability. Like other oocyte-specific variants, dBigH1 is maternally expressed during early embryogenesis until it is replaced by somatic dH1 at the maternal-to-zygotic transition (MZT). Brat also mediates dBigH1 silencing at MZT. Finally, we discuss the situation in testes, where Brat is not expressed, but dBigH1 is translationally silenced too.

In general, the patterns of expression of germline H1s are tightly regulated during lineage differentiation. Testis-specific variants are generally detected only after spermatogonia stop proliferation and differentiate to spermatocytes. In mammals, H1T is the first variant to be expressed in meiotic spermatocytes and, depending on the species, is also detected during spermatids differentiation [22][23][24][25]. On the other hand, expression of the other two mammalian male variants HILS1 and H1T2 is restricted to spermatids [4][5][6][7]. A similar situation is observed in Drosophila, where dBigH1 is detected in spermatocytes, but not in the proliferating spermatogonia and upon spermatids differentiation [19]. dBigH1 is also detected in the male germ stem cell (GSC). Regarding female-specific H1s, their expression is mostly restricted to the oocyte and the early stages of embryo development. In humans and mice, H1oo expression is restricted to the growing/maturing oocyte entering meiosis and, after fertilization, it rapidly decays during the first mitotic divisions, becoming undetectable at the 2-4 cells blastula stage when it is replaced by the somatic H1 variants [8,21,26,27]. In the sea urchin, the female-specific Cs-H1 variant is also replaced by somatic H1s at early cleavage stages [11,12]. However, the femalespecific variants of Drosophila (dBigH1), zebrafish (H1M) and Xenopus (B4) persist longer during embryo development, being replaced by the somatic H1s only after 14, 10 and 13 cleavages, respectively [10,16,17,19]. Translational regulatory mechanisms appear to play a central role in the regulation of germline H1s expression. In Xenopus, translation of B4 mRNA is regulated by CPEBs proteins that bind at the 3 0 UTR and, upon phosphorylation, promote polyadenylation [28][29][30]. Mammalian H1oo mRNA also contains several functional CPEs at the 3 0 UTR [31,32]. In addition, in testes, several 5'UTR regulatory elements regulate HILS1 expression [4,33], and in Drosophila the translational repressor Bam is required to silence dBigH1 expression during spermatogonia proliferation [34]. However, little else is known about the mechanisms that govern stagespecific expression of germline H1s. Here, we address this question in the Drosophila female germline.
In Drosophila, the early stages of gametogenesis share remarkable similarities in females and males (reviewed in [35][36][37][38][39]). In both ovaries and testes, GSCs localize anterior, anchored to a niche of somatic cells, and divide asymmetrically for self-renewal and to produce daughter progenitor cells (cystoblasts (CBs) in females and gonioblasts (GBs) in males), which start the complex differentiation programme that, ultimately, leads to the production of functional gametes. Daughter cells undergo four successive rounds of transit-amplifying (TA) divisions with incomplete cytokinesis to produce a cyst of 16 sister germ cells (GCs) (spermatogonia in males and cystocytes in females) that remain interconnected and are surrounded by a somatic cells layer. Then, the pathways diverge; female cysts develop to produce a single egg, whereas male cysts differentiate to spermatocytes and undergo two meiotic divisions to produce 64 spermatids that develop to mature sperm cells. Our results show that, similar to males, dBigH1 is expressed in the female GSCs and CBs, is silenced in the proliferating TA cystocytes to resume expression upon oocyte differentiation. We report that dBigH1 silencing in cystocytes depends on the tumour suppressor Brain tumour (Brat), a post-transcriptional regulator that is expressed in cystocytes and represses translation of GSC maintenance factors [40,41]. We also show that, during embryogenesis, Brat silences dBigH1 expression at MZT. Altogether these results unveil the importance of posttranscriptional regulation in setting the patterns of expression of germline-specific H1 variants.  [43]. bamP-bam::GFP and bamP-GFP [44] were a gift from Dr M. Buszczak. brat K06028 was a gift from Dr J. Knoblich. vasa-EGFP construct [45] was a gift from Dr A. Nakamura. Transgenic lines carrying the various constructs described in figures 4 and 7 and electronic supplementary material, figure S4 were obtained by specific site-directed integration into ZH-86Fb and ZH-58A att line [46]. All Drosophila stocks were maintained at 25°C on standard media. For RNAi knockdown, crosses were set up at 25°C.

Antibodies
Rabbit polyclonal αdBigH1 is described in [19]. Rat polyclonal αdBigH1 was raised as described in [19]. Rabbit αdH1 was a gift from Dr J. Kadonaga and is described in [47]. Guinea pig polyclonal αTj was a gift from Dr D. Godt and is described in [48]. Rabbit αBrat was a gift from Dr J. Knoblich and is described in [49]. All other antibodies used in these experiments were commercially available: mouse monoclonal αadd (DSHB, 1B1), rat monoclonal αHA (Sigma, 3F10), mouse monoclonal αFasciclin III (DSHB, 7G10) and rat monoclonal αVasa (DSHB, 1ea).

Immunostaining experiments
Ovaries and testes were dissected in PBS, fixed in PBS with 4% paraformaldehyde for 20 min, and then washed with PBS three times for 10 min each. The samples were first incubated in blocking solution, 2% bovine serum albumin diluted in PBT (PBS and 0.3%Triton X-100), for 1 h and then with the appropriate primary antibodies diluted in blocking solution at 4°C overnight. The samples were washed with PBT three times for 10 min each and then incubated with blocking solution for 30 min. Then, samples were incubated with the appropriate secondary antibodies at 25°C for 2 h and washed with PBT three times for 10 min. Samples were mounted in Mowiol (Calbiochem-Navabiochem) containing 0.2 ng µl −1 DAPI (Sigma) and visualized in a confocal microscope (Leica TCS SP2-AOBS). Quantitative analyses presented in figures 1b and 4d were performed using ImageJ. In each germarium, DAPI was used to define the nuclear area of three different cells from each region (1-3). Three additional ROIs (same area and non-nuclear) were also defined as background. In figure 1b, royalsocietypublishing.org/journal/rsob Open Biol. 11: 200408 the nuclear signal of αdBigH1 was determined using the Gray mean value, and the average signal in each region was normalized against the average background signal. In figure 4d, the GFP signal was determined using the same method, but normalization was performed using the average signal of region 2, since the background levels of direct fluorescence were extremely low.
Embryos were dechorionated in bleach and fixed for 25 min in 1 : 1 solution of formaldehyde and heptane. Embryos were devitellinized in methanol followed by rehydration with PBT and blocking in PBT-BSA (2%). Samples were incubated with primary antibodies diluted in PBT-BSA at 4°C overnight. After washing three times with PBT, embryos were incubated with secondary antibody and stained with DAPI at room temperature for 2 h followed by three washes in PBT. Embryos were mounted in vectashield (Vector labs) and imaged using Zeiss LSM 780 confocal microscope.
Primary antibodies used for immunostaining were: αdBigH1

RT-qPCR analysis
For qRT-PCR experiments, RNA was prepared from embryos using trizol reagent and purified using qiagen RNeasy mini kit following the manufacturer's instructions. Total RNA (1 µg) was used for complementary DNA (cDNA) synthesis. Reverse transcription was performed using oligo dT supplied in the kit. qRT-PCRs were run in triplicate in two independent experiments. Expression data were normalized to Act5C and analysed using the ΔΔCt method. Primers used were: dBigH1fw 5 0 -AATATGGGCGAAGAAGAGGA-3 0 , dBigH1rv 5 0 -GAGAT TATCTGTCTCGACCTC-3 0 , Act5cfw 5 0 -CACCAAATCTTACA AAATGTGTGAC-3 0 and Act5crv 5 0 -CATCGTCTCCGGCAA ATC-3 0 . In the germarium, intense nuclear αdBigH1 immunostaining was detected in region 1, being highly reduced to background levels during cystocytes proliferation in region 2, to reappear again in region 3 (figure 1b, centre and bottom). Cells showing αdBigH1 immunostaining were positive for vasa, a specific germline marker [51,52], and negative for Traffic jam (Tj), a marker of somatic cells [48] (electronic supplementary material, figure S1), suggesting that dBigH1 expression was restricted to germ cells. We also analysed whether, in region 1, dBigH1 was expressed in both GSCs and CBs. For this purpose, we performed co-immunostaining experiments with αadd antibodies, which mark the spectrosome, a cytoskeleton structure that occupies an anterior position in GSCs, moves posterior in CBs and, later, grows and branches out to form the fusome that keeps cysts cells interconnected [53,54] (figure 1b, top). In region 1, we detected nuclear αdBigH1 immunostaining in cells with anterior as well as posterior spectrosomes (figure 2a), indicating that dBigH1 was expressed in both GSCs and CBs. In addition, silencing of dBigH1 expression during cystocytes proliferation was confirmed in flies carrying a bamP-bam::GFP construct, which is specifically expressed in cystocytes [44], since nuclear αdBigH1 immunostaining was not detected in cells expressing the reporter royalsocietypublishing.org/journal/rsob Open Biol. 11: 200408 (figure 2b). After cystocytes stop proliferation and start to differentiate, dBigH1 expression resumed. In the budding cysts (region 3 of the germarium) and the early-developed egg chambers (stage 2), all nuclei were positive for αdBigH1 (figure 1a). Later, from stage 3 on, αdBigH1 immunostaining was progressively constrained to the oocyte nucleus (figure 1a,c), where it largely overlapped with DAPI at the condensed chromatin of the karyosome (electronic supplementary material, figure S2). A weak signal could also be detected in the nucleoplasm (electronic supplementary material, figure S2B; see also figure 3b), suggesting that a minor fraction of dBigH1 stays unbound and free in the nucleoplasm. At late developmental stages, αdBigH1 immunostaining was also detected in nuclei of the nurse cells (nc) proximal to the oocyte (figure 1a,c). This pattern of expression is very unusual and, interestingly, takes place around the stage when nc begin dumping of their content into the oocyte and, ultimately, die. In this regard, nc proximal to the oocyte are first in undergoing dumping. Though highly speculative, dBigH1 might regulate transcriptional activity in these cells during dumping. Of note, nuclear αdBigH1 immunostaining of germ cells was abolished in a null dBigH1 NSTOP CRISPR/CAS9 mutant [43], showing its specificity (electronic supplementary material, figure S3). By contrast, background αdBigH1 immunostaining observed in the cytoplasm and somatic FC was also detected in the null dBigH1 NSTOP mutant, indicating it was unspecific (electronic supplementary material, figure S3).
We also analysed the pattern of expression in ovaries of the single somatic linker histone of Drosophila dH1 [55][56][57]. In addition to the somatic FC cells, which showed strong αdH1 immunostaining, we also detected dH1 expression in germline cells (figure 3a). In the germarium, αdH1 signal was detected in the GSCs/CBs, which express dBigH1, as well as in cystocytes, which lack dBigH1 (figure 3a). In the budding cysts and stage 2 egg chambers, dH1 expression was detected in both the nc and the oocyte that also contained dBigH1 (figure 3a). Later, when dBigH1 starts to accumulate in the oocyte (stages 3-5), dH1 expression decayed in the oocyte, becoming undetectable at stage 5 (figure 3a), while it was still detected in the nc (figure 3a). In the nc, dH1 expression also decreased upon development to almost undetectable levels (figure 3b).

Silencing of dBigH1 in cystocytes is posttranscriptionally regulated
Results reported above suggest that dBigH1 expression is tightly regulated during early oogenesis, being silenced in proliferating cystocytes. This regulation is mainly posttranscriptional since expression of a ectopic dBigH1::HA construct, which carries the dBigH1 regulatory elements and royalsocietypublishing.org/journal/rsob Open Biol. 11: 200408 largely mimics expression of endogenous dBigH1 (figure 4a), was not substantially altered when the dBigH1 promoter was replaced by the germline-specific vasa promoter, which is ubiquitously active in germline cells [45] (figure 4b) (see also figure 4d). We also observed that the deletion of the dBigH1 5 0 UTR had no major effect on the pattern of dBigH1::HA expression in ovaries (figure 4c), suggesting that the dBigH1 3 0 UTR is sufficient to silence dBigH1 expression in cystocytes.
In agreement, we observed that the 3 0 UTR of dBigH1 silenced expression in cystocytes of a ubiquitously active vasa-EGFP reporter [45] (figure 4d). It must be noted that these ectopic dBigH1::HA constructs did not fully recapitulate dBigH1 silencing since we detected dBigH1::HA expression in cystocytes in approximately 25% of germaria. Noteworthy, the proportion of germaria showing ectopic dBigH1::HA expression in cystocytes tended to increase when the dBigH1 3 0 UTR was replaced by that of vasa (Fisher test, p-value: 0.227) (electronic supplementary material, figure S4A). Altogether these results suggest that elements within the dBigH1 3 0 UTR mediate post-transcriptional silencing in cystocytes.

Brat regulates dBigH1 silencing in cystocytes
We noted that the dBigH1 3'UTR sequence contains two consensus binding sites for Brat [58] (electronic supplementary material, figure S5A), an important post-transcriptional regulator that is expressed in cystocytes (electronic supplementary material, figure S6) and represses translation of stem cell maintenance factors, promoting differentiation [40,41]. Interestingly, RNA immunoprecipitation experiments (RIP-Chip) performed in embryos showed that Brat interacts with the dBigH1 mRNA [58]. Thus, we tested the possibility that Brat is involved in silencing dBigH1 expression in cystocytes. For this purpose, we performed RNAi-mediated depletion of Brat in ovaries using a nos-GAL4 driver that is specifically expressed in the germline [42]. We observed that, in agreement with its role in promoting GSCs differentiation, Brat depletion increased the number of cells in which spectrosome structures were detected (figure 5a), suggesting an accumulation of GSCs/CBs. In addition, approximately 40% of germaria showed detectable levels of dBigH1 expression in cyst cells interconnected by branched fusomes (figure 5b), suggesting that Brat is required to silence dBigH1 expression in cystocytes. It was shown earlier that, in testes, dBigH1 expression is post-transcriptionally silenced during TA spermatogonial divisions by bag-of-marbles (Bam) [34]. Bam is also expressed in ovaries [59,60], where it represses translation of GSC maintenance factors and induces differentiation [61][62][63]. In this regard, it is known that Bam is required for Brat expression   royalsocietypublishing.org/journal/rsob Open Biol. 11: 200408 in cystocytes since it represses the GSC maintenance factor nanos (nos) [63,64] that, in its turn, represses Brat [40]. Thus, we anticipated that Bam would also regulated dBigH1 expression in cystocytes. We observed that the depletion of Bam in ovaries blocked early differentiation, giving rise to tumorous germaria that contained a large number of undifferentiated GSCs/CBs expressing dBigH1 ( figure 6a). This strong phenotype, which was reported earlier in other Bam loss-of-function (LOF) mutations [59,65,66], made it challenging to determine the contribution of Bam to dBigH1 silencing in cystocytes. However, among the large number of GSCs/CBs observed upon Bam depletion, we detected dBigH1 expression in some earlydeveloped cysts containing fusome-interconnected cells ( figure 6b). These results suggest that, like in spermatogonia, Bam also regulates silencing of dBigH1 expression in cystocytes. Besides these similarities, the regulation of dBigH1 silencing in testes and ovaries shows some important differences since, in contrast with what was observed in ovaries, the dBigH1 3 0 UTR was not capable of silencing expression of vasa-EGFP in testes ( figure 7) and replacement of the dBigH1 3 0 UTR by the vasa 3 0 UTR did not affect silencing in spermatogonia of an ectopic dBigH1::HA construct (electronic supplementary material, figure S4B). In this regard, we considered the possibility that alternative polyadenylation events could give rise to different 3 0 UTRs in testes and ovaries. However, RACE experiments showed the same dBigH1 3 0 UTR in testes, ovaries and embryos (electronic supplementary material, figure S5A).

Brat also regulates dBigH1 silencing at MZT
It has been shown that Brat, which is maternally expressed during early embryogenesis, interacts with and silences a large subset of maternal mRNAs at MZT, among which the dBigH1 mRNA was identified [58,67]. Thus, we tested the possibility that Brat also silences dBigH1 expression at MZT. For this purpose, we took advantage of the LOF mutation brat K06028 , a recessive lethal P-element insertion allele that shows some defects in abdominal embryo segmentation and induces tumorous overgrowth in larval brains, where Brat is highly expressed [68,69]. Despite these defects, homozygous brat K06028 mutants progress relatively normal through embryogenesis and larval development [58,67,69,70]. We observed that, like in control wild-type embryos, dBigH1 was ubiquitously expressed in homozygous brat K06028 embryos throughout blastoderm stages (electronic supplementary material, figure S7). However, while in control embryos, dBigH1 expression was constrained to the primordial germ cells (PGC) at gastrula stages (figure 8a,b, left panels), intense αdBigH1 immunostaining was detected in somatic cells in 46% (N = 120) of homozygous brat K06028 gastrula (figure 8a,b, right royalsocietypublishing.org/journal/rsob Open Biol. 11: 200408 panels). Similar results were obtained in trans-heterozygous brat K06028 /Df(2 L)TE37C-7 embryos, which carried the brat K06028 mutation over the Df(2 L)TE37C-7 deficiency that uncovers Brat (58%, N = 108) (electronic supplementary material, figure  S8). In addition to its role in translational regulation, Brat has been shown to regulate the stability of a subset of maternal transcripts at MZT, including the dBigH1 mRNA [58]. RT-qPCR experiments detected increased dBigH1 mRNA levels in brat K06028 mutant embryos (electronic supplementary material, figure S5B), confirming the contribution of Brat to dBigH1 mRNA degradation/decay at MZT. Notably, in brat K06028 mutants, intense αdH1 immunostaining was detected at gastrula stages indicating that, under these conditions, dBigH1 and dH1 are both expressed ( figure 9).

Discussion
Results reported here and elsewhere [34] show that the patterns of dBigH1 expression during the early stages of oogenesis and spermatogenesis are remarkably similar. In both cases, dBigH1 is expressed in GSCs and daughter progenitor cells, is silenced during TA divisions to resume expression when proliferation stops and differentiation begins. Moreover, translational regulation accounts for silencing of dBigH1 expression during TA divisions in both ovaries and testes. However, the actual mechanisms involved show some important differences. Our results suggest that, in ovaries, Brat mediates translational dBigH1 silencing in cystocytes by directly binding the dBigH1 3 0 UTR, which contains two Brat-binding sites. However, in testes, the situation must be different since Brat is not significantly expressed (see FlyAtlas and modENCODE tissue expression data in Flybase (https://flybase.org/reports/FBgn0010300)).
In good agreement, the dBigH1 3 0 UTR is not required to silence dBigH1 expression in spermatogonia and, along the same lines, it is not sufficient to silence vasa-EGFP expression. Instead, in testes, dBigH1 silencing in spermatogonia is mediated by Bam [34]. Our results suggest that Bam is also required for dBigH1 silencing in cystocytes. However, in this case, its contribution might be indirect, through the activation of Brat expression [40]. Altogether these results suggest that, at least in part, the mechanisms governing translational regulation of dBigH1 expression are different in ovaries and testes.
We have also shown that Brat is required for dBigH1 silencing in gastrulated embryos. It has been reported that, during embryo development, Brat acts both as a translational repressor and a factor required for degradation/decay of maternal transcripts at MZT [58]. In this regard, sustained dBigH1 expression observed at gastrula stages in brat mutant embryos supports a contribution of Brat to dBigH1 mRNA stability at MZT. Instead, in early oogenesis, Brat probably acts as a repressor of dBigH1 mRNA translation since dBigH1 is silenced only transiently during TA divisions. Brat might also regulate the translation of maternal dBigH1 transcripts in early embryogenesis. The factors that control Brat action in repression or mRNA destabilization remain to be determined.
Our results challenge the usually accepted view that germline-specific H1s replace somatic variants, which implies that their patterns of expression do not generally overlap. Instead, we have shown that somatic dH1 is broadly expressed during oogenesis and, though it ends up being replaced by dBigH1 in the oocyte, the two variants largely coexist except during TA divisions, where dH1 is expressed, but dBigH1 is not. A similar situation was reported in testes, where dH1 coexists with dBigH1 in GSCs and GBs, but it is the only variant expressed in TA divisions [34]. However, in this case, once proliferation stops, dH1 expression is strongly silenced in spermatocytes, while dBigH1 is highly expressed [34]. Later, royalsocietypublishing.org/journal/rsob Open Biol. 11: 200408 dBigH1 is also silenced in spermatids [34]. Along the same lines, in brat mutant embryos, dBigH1 and dH1 also coexist at gastrula stages. Altogether, these results indicate that the patterns of expression of dH1 and dBigH1 are not necessarily exclusive. It is possible that dBigH1 and dH1 preferentially target different genomic loci since ectopic dBigH1 expression in S2 cells has shown that dBigH1 preferentially binds to and displaced dH1 from silent genomic regions with high dH1 content [71]. On the other hand, recent results suggest a more complex situation since, in null bigH1 mutants generated by CRISPR/CAS9, the lack of maternal dBigH1 is compensated by the expression of somatic dH1 from the earliest stages of embryo development [43,72], suggesting that dBigH1 represses dH1 expression in the early Drosophila embryo.
Further work is required to reach a better understanding of the actual link(s) between dBigH1 and dH1 expression and deposition.
Results reported here and elsewhere [19,34] show that the pattern of expression of dBigH1 is tightly regulated during germline lineage differentiation and embryogenesis, suggesting that dBigH1 plays specific functions in germline and embryo development. However, unveiling the functional contribution of dBigH1 is proving more difficult than anticipated. Based on defects associated with a genetic mutation generated through imperfect excision of a 5 0 UTR P-element insertion, dBigH1 was proposed to be essential during early embryogenesis, contributing to the activation of the zygotic genome [19]. In addition, RNAi-mediated depletion of dBigH1 in testes induced strong developmental defects and reduced fertility [34]. However, CRISPR/CAS9 null bigH1 mutants turned out to be viable and fertile, progressing through embryogenesis likely due to the compensatory expression of somatic dH1 [43,72]. Moreover, CRISPR/CAS9 lines in which the CDS of dBigH1 was replaced by that of somatic dH1 are also viable, though showing DNA replication defects and altered chromatin condensation during early embryogenesis [73]. These observations suggest that to a large extent, dBigH1 and dH1 are functionally redundant, leaving the question of the possible specific functions of dBigH1 open.
In summary, our results show that the tumour suppressor Brat is crucial to silence dBigH1 expression in both ovaries and embryos. This regulatory mechanism might not be constrained to the germline. In this regard, it was reported that, while dBigH1 is not detected in the normal larval brain (or any other somatic tissue), it becomes ectopically expressed in some Brat-induced brain tumours [74]. To what extent dBigH1 expression contributes to malignant growth in somatic tissues remains to be determined. It also remains to be determined if Brat orthologues in other species (such as human TRIM2,3, which are implicated in malignant glioma [75]) have similar effects in the expression of embryonic H1 linker histones.   Figure 8. Brat regulates dBigH1 silencing in embryos. Immunostainings with αdBigH1 antibodies (in red) of control wt and homozygous brat K06028 embryos at gastrula stages 9 (a) and 15 (b). Top and central stacks are presented. DNA was stained with DAPI (in blue). Scale bars correspond to 50 µm. For top stacks of homozygous brat K06028 embryos, enlarged views of merge images showing co-localization of αdBigH1 signal with DAPI are also presented. See also electronic supplementary material, figures S5, S7 and S8. royalsocietypublishing.org/journal/rsob Open Biol. 11: 200408