tBRD-1 and tBRD-2 regulate expression of genes necessary for spermatid differentiation

ABSTRACT Male germ cell differentiation proceeds to a large extent in the absence of active gene transcription. In Drosophila, hundreds of genes whose proteins are required during post-meiotic spermatid differentiation (spermiogenesis) are transcribed in primary spermatocytes. Transcription of these genes depends on the sequential action of the testis meiotic arrest complex (tMAC), Mediator complex, and testis-specific TFIID (tTFIID) complex. How the action of these protein complexes is coordinated and which other factors are involved in the regulation of transcription in spermatocytes is not well understood. Here, we show that the bromodomain proteins tBRD-1 and tBRD-2 regulate gene expression in primary spermatocytes and share a subset of target genes. The function of tBRD-1 was essential for the sub-cellular localization of endogenous tBRD-2 but dispensable for its protein stability. Our comparison of different microarray data sets showed that in primary spermatocytes, the expression of a defined number of genes depends on the function of the bromodomain proteins tBRD-1 and tBRD-2, the tMAC component Aly, the Mediator component Med22, and the tTAF Sa.


INTRODUCTION
In Drosophila melanogaster and mammals, the post-meiotic phase of spermatogenesis (spermiogenesis) is characterized by extensive morphological cell changes . In flies, transcription almost ceases as the cells enter meiotic division; therefore, these changes mainly rely on proteins arising from translationally repressed and stored mRNAs synthesized in primary spermatocytes (Olivieri and Olivieri, 1965;White-Cooper et al., 1998). Hence, a tightly regulated gene transcription program is required to ensure proper spermiogenesis and male fertility.
In primary spermatocytes, numerous transcripts are synthesized and translationally repressed (Fuller, 1993;White-Cooper et al., 1998). Transcription of the corresponding genes (spermiogenesisrelevant genes) depends on two testis-specific transcription complexes: the testis meiotic arrest complex (tMAC), and the testis-specific TFIID complex, which consists of testis-specific TATA box binding protein-associated factors (tTAFs) (Beall et al., 2007;Hiller et al., 2004Hiller et al., , 2001. Recruitment of tTAFs to chromatin requires the coactivator complex Mediator, and localization of Mediator subunits to chromatin depends on tMAC (Lu and Fuller, 2015). Based on these data, it has been suggested that Mediator acts as a key factor in a tTAF-and tMAC-dependent gene regulatory cascade that leads to transcriptional activation of spermiogenesisrelevant genes (Lu and Fuller, 2015).
Acetylated lysines of histone play an important role in gene transcription (Sanchez and Zhou, 2009). These histone modifications are recognized by bromodomain-containing proteins (Dhalluin et al., 1999). The bromodomain forms a wellconserved structure within functionally distinct proteins, such as histone acetyltransferases, chromatin-remodeling factors, transcriptional co-activators and mediators, and members of the bromodomain and extra-terminal (BET) family (Josling et al., 2012). Members of the BET family are characterized by having one (in plants) or two (in animals) N-terminal bromodomains and a conserved extra-terminal domain that is necessary for proteinprotein interactions (Florence and Faller, 2001;Matangkasombut et al., 2000;Platt et al., 1999). BET proteins contribute to transcription mainly by recruiting protein complexes, e.g. transcription factors and chromatin remodelers (Josling et al., 2012;Krogan et al., 2003;Matangkasombut et al., 2000). In mammals, the BET proteins BRD2, BRD3, BRD4, and BRDT are expressed in male germ cells (Klaus et al., 2016;Shang et al., 2004). BRDT is involved in gene expression during spermatogenesis, among other roles (Berkovits et al., 2012;Gaucher et al., 2012), but the functions of BRD2, BRD3, and BRD4 in male germ cells are not well understood.
Here, we show that a tbrd-1-eGFP transgene restores not only male fertility of tbrd-1 mutants but also localization of tBRD-2 to chromosomal regions. Protein-protein interaction studies demonstrated that both bromodomains are dispensable for tBRD-1 homodimer formation and that the extra-terminal domain of tBRD-2 interacts with the C-terminal region of tBRD-1. Peptide pull-down experiments indicated that tBRD-1 but not tBRD-2 preferentially recognizes acetylated histones H3 and H4. Microarray analyses revealed that several genes are significantly downregulated in tbrd-2-deficient testes. A comparison of different microarray data sets demonstrated that tBRD-1, tBRD-2, the tMAC component Aly, the Mediator component Med22, and the tTAF Sa share a subset of target genes. Finally, immunofluorescence stainings showed that the sub-cellular localization of tBRD-1 and tBRD-2 requires Aly function.
The bromodomains of tBRD-1 are dispensable for homodimer formation, and the very C-terminus of tBRD-1 interacts with the extra-terminal domain of tBRD-2 Recently, we have shown that tBRD-1 forms homodimers and also heterodimers with tBRD-2 (Theofel et al., 2014). Here, we aimed at mapping the interaction domains required for dimerization using a series of tBRD-1 and tBRD-2 truncation mutants in the yeast two-hybrid assay ( Fig. 2; Figs S2 and S3). tBRD-1 and tBRD-2 contain several conserved domains, namely the bromodomains and an extra-terminal domain, which consists of a NET domain and a SEED domain and is predicted to mediate protein-protein interactions (Florence and Faller, 2001;Matangkasombut et al., 2000;Platt et al., 1999). Accordingly, we focused our analysis on these domains. Full-length tBRD-1 formed homodimers with tBRD-1ΔN, which lacks the first bromodomain (BD1) ( Fig. 2A; Fig. S2B) and with tBRD-1Δ, which lacks both bromodomains and consists only of the spacer region that connects these two domains ( Fig. 2A; Fig. S2B). No interaction was observed between full-length tBRD-1 and tBRD-1ΔC, which contains the first bromodomain but an incomplete spacer region ( Fig. 2A; Fig. S2B). These results indicated that the spacer region between the bromodomains (amino acids 165-336) is essential for tBRD-1 homodimer formation (Fig. 2C). Next, we sought to determine which tBRD-2 sequences mediate binding to tBRD-1. We analyzed the interaction of several tBRD-2 deletion We first mapped the binding to a C-terminal region containing the NET and SEED domains. Further analysis revealed that neither of these two domains was essential for tBRD-1 binding. Instead, tBRD-1 interaction required the region connecting the NET and SEED domains (amino acids 444-580). Finally, we showed that the C-terminus (amino acids 410-514) of tBRD-1 is required for heterodimerization with tBRD-2 ( Fig. 2A; Fig. S3E,G). In summary, our results showed that the spacer region between the two bromodomains mediates tBRD-1 homodimerization (Fig. 2C) and indicated that tBRD-1 and tBRD-2 interact via the C-terminus of tBRD-1 and the region between the NET and the SEED domains of tBRD-2 (Fig. 2D).

tBRD-1 recognizes acetylated histones H3 and H4 in vitro
Previously, we have shown that localization of tBRD-1 and tBRD-2 to the chromosomal regions in spermatocytes is acetylation dependent (Leser et al., 2012;Theofel et al., 2014). This finding implied that tBRD-1 and tBRD-2 might directly interact with acetylated histone tails. To test this hypothesis, we purified recombinant tBRD-1 and tBRD-2 using the baculovirus system and performed peptide pull-down assays with histone H3 and histone H4 peptides that were unmodified or acetylated at specific residues. Immobilized peptides were incubated with recombinant tBRD-1 or tBRD-2, and bound proteins were analyzed in western blots using tBRD-1-or tBRD-2-specific antibodies (Fig. 3A). tBRD-1 bound to all unmodified or acetylated histone H3 and H4 peptides analyzed, in keeping with the idea that histone interactions might contribute to chromatin binding of tBRD-1, but tBRD-1 preferentially bound to acetylated histone tails (Fig. 3A). Likewise, tBRD-2 bound to all unmodified or acetylated histone peptides tested. In contrast to tBRD-1, however, tBRD-2 did not preferentially bind acetylated peptides, and acetylation instead appeared to reduce binding affinity. We concluded that tBRD-1 and tBRD-2 both interact with histone tails in vitro and that this binding reaction is sensitive to histone acetylation. To investigate whether these acetylated histones are present in spermatocytes, we stained them with immunofluorescent antibodies raised against different histone H3 and H4 acetylation marks (  histones H3K14ac and H3K36ac were barely detected at the chromosomal regions in primary spermatocytes (Fig. 3C,G, arrows).

DISCUSSION
In Drosophila, spermatocytes execute a highly active and strictly regulated transcription program to provide transcripts necessary for post-meiotic spermiogenesis. Transcription of spermiogenesisrelevant genes is based on the cooperation among tTAFs, tMAC components, and Mediator complex components (Beall et al., 2007;Chen et al., 2011;Hiller et al., 2004;Lu and Fuller, 2015). Recently, we have postulated that the testis-specific bromodomain proteins tBRD-1, tBRD-2, and tBRD-3 cooperate with the testis-specific TFIID complex in regulating transcription of a subset of spermiogenesis-relevant genes (Theofel et al., 2014). Here, we uncovered additional potential links between tBRD proteins, Mediator, and tMAC.
The function of tBRD-1 is essential for proper sub-cellular localization of endogenous tBRD-2 Previously, we have shown that in testes of transgenic flies, endogenous tBRD-1 interacts with tBRD-2-eGFP (Theofel et al., 2014). Here, we further focused on the interaction between tBRD-1 and tBRD-2 and showed that expression of tBRD-1-eGFP can restore sub-cellular localization of tBRD-2 in primary spermatocytes in a tbrd-1 mutant background. These results indicated that tBRD-1 and tBRD-2 indeed interact in Drosophila spermatocytes. The structure of tBRD-1 and tBRD-2 proteins differ from that of classical BET family members in animals, which are mainly characterized by two N-terminal bromodomains and a C-terminal extra-terminal domain consisting of a NET motif and a SEED motif (Florence and Faller, 2001). tBRD-1 contains two bromodomains but no extra-terminal domain, and tBRD-2 contains only one bromodomain but does contain a C-terminal extra-terminal domain (Theofel et al., 2014). The extra-terminal domain has been described as necessary for protein-protein interactions (Florence  , and (C) sa mutant testes (y-axes) compared to wild-type control testes (x-axes). Green and blue dots represent significantly down-regulated genes in tbrd-1 mutant testes in comparison to control testes. Red and blue dots represent transcripts of genes in bam≫tbrd-2 RNAi testes expressed significantly lower than in undriven tbrd-2 RNAi . Blue dots represent transcripts that are affected by both tBRD-1 and tBRD-2. and Faller, 2001; Matangkasombut et al., 2000;Platt et al., 1999). However, it has been shown that human BRD2 requires the first Nterminal bromodomain for dimerization (Nakamura et al., 2007). More recent results have shown that homodimer and heterodimer formation of BET proteins is mediated by a conserved motif, termed motif B, between the second bromodomain and the extra-terminal domain (Garcia-Gutierrez et al., 2012). We showed in yeast twohybrid experiments that the C-terminal part of tBRD-1 and the extra-terminal domain of tBRD-2 are essential for interaction of the two proteins. By contrast, homodimer formation of tBRD-1 proteins required the region between the two bromodomains.
Recently, it has been suggested that the interaction of tBRD-1 and tBRD-2 is required for their protein stability (Kimura and Loppin, 2015). However, we did not observe an altered tBRD-1 protein distribution or changes in protein levels in tbrd-2 knockdown testes compared to controls. tBRD-2 proteins were barely detectable in tbrd-2 knockdown testes, which allows us to assume that the knockdown was efficient. These results indicated that tBRD-1 does not require tBRD-2 function for protein stability or sub-cellular localization. By contrast, the tBRD-2 signal was strongly reduced in tbrd-1 mutant spermatocyte nuclei. However, also in this case, we did not observe lower amounts of tBRD-2 protein in tbrd-1 mutant testes in western blots. Hence, the loss of tBRD-1 seems to affect the sub-cellular localization of tBRD-2. Our results showed that the function of tBRD-1 is required for proper sub-cellular localization of tBRD-2 but not vice versa. In addition, the function of tBRD-1 seems to be dispensable for tBRD-2 protein stability. Whether this dependency is based upon direct interaction of the two proteins still has to be clarified.

tBRD-1 binds to acetylated histones independently of tBRD-2
Previously, we have shown that an increased acetylation level in spermatocytes enhances the localization of tBRD-1 and tBRD-2 to the chromosomal regions (Leser et al., 2012;Theofel et al., 2014). However, it was unclear whether both proteins directly bind to acetylated histone tails. In the current study, in vitro experiments demonstrated that the double bromodomain protein tBRD-1 bind to H3 peptides acetylated at lysines 9 and 14 and to H4 peptides acetylated at lysines 5, 8, and 12. By contrast, tBRD-2 exhibited a higher affinity for non-acetylated histone peptides under the same conditions. Acetylation of N-terminal histone tails of H3 and H4 is a typical feature of transcriptional active chromatin and serves as a binding platform for epigenetic regulators, such as BET proteins (Davie and Candido, 1978;Dhalluin et al., 1999;Hebbes et al., 1988). It has been previously shown that the acetylation marks tested in this study are recognized by BET proteins (Marchand and Caflisch, 2015) and are involved in active gene expression (Morris et al., 2007;Wang et al., 2008). In addition, all tested acetylation marks except those of H3K14ac and H3K36ac were detected in spermatocyte nuclei, which indicated that tBRD-1 recognizes acetylated H3 at lysine 9 and/or 14 and acetylated H4 at lysine 5, 8, and/or 12 also in vivo.
In murine round spermatids, acetylated H3 and H4 are enriched at the transcription start sites of spermiogenesis-relevant genes and are recognized by the BET proteins BRD4 and BRDT (Bryant et al., 2015). Recently, it has been suggested that the interaction of tBRD-1 and tBRD-2 allows the two proteins to function together as a single BRDT-like BET protein (Kimura and Loppin, 2015). Therefore, it is conceivable that tBRD-2 requires tBRD-1 for efficient binding to chromatin. However, it is also possible that tBRD-2 recognizes other, not yet tested acetylation marks independently of tBRD-1. As tBRD-1 and tBRD-2 regulate both common and different sets of target genes, both scenarios could occur in spermatocytes. Our data suggest that in Drosophila, as in mice, bromodomain proteins act together to efficiently support the activation of spermiogenesis-relevant genes by binding to acetylated lysine residues.
tBRD-1 and tBRD-2 co-regulate a subset of target genes Our microarray analyses showed that tBRD-2, like tBRD-1, is involved in gene activation and repression. The comparison of transcriptome data of a tbrd-1 mutant with that of a tbrd-2 knockdown clearly indicated that the two bromodomain proteins share a subset of target genes. However, we observed that the expression of some genes was altered in tbrd-1 mutant testes but not in tbrd-2 knockdown testes and vice versa, which suggested that some genes are regulated specifically by either tBRD-1 or tBRD-2. In mice, the BET proteins BRDT and BRD4 cooperate to regulate transcription of spermiogenesis-relevant genes, although they can also act independently. Importantly, it has been demonstrated that genes co-bound by BRDT and BRD4 show a higher transcriptional activity than genes bound only by BRD4 or BRDT (Bryant et al., 2015). Further experiments are required to examine whether tBRD-1 and tBRD-2 directly bind to their target genes and whether the binding of both enhances transcription.
An overlapping set of spermiogenesis-relevant genes is regulated by tBRD-1, tBRD-2, the tMAC complex, Mediator complex, and tTAFs It has been proposed that the activation of spermiogenesis-relevant genes in Drosophila spermatocytes requires the sequential action of the tMAC complex, Mediator complex, and testis-specific TFIID (tTFIID) complex (Chen et al., 2011;Lu and Fuller, 2015). The tMAC component Topi interacts with the Mediator component Med22, but no direct interaction has been observed between Mediator and tTFIID components. However, when Med22 is knocked down, the tTAF Sa fails to localize to chromatin, which suggests that tTAFs depend on Mediator to be recruited to chromatin or stabilized there (Lu and Fuller, 2015). Previously, we have shown that the proper localization of tBRD-1 and tBRD-2 depends on tTAF function (Leser et al., 2012;Theofel et al., 2014). In addition, we have demonstrated that tBRD-1 and the tTAF Sa share a subset of target genes (Theofel et al., 2014). In our current study, immunofluorescence analyses revealed a dramatically reduced localization of tBRD-1 and tBRD-2 to chromosomal regions in homozygous aly mutant spermatocytes. We hypothesized that also tBRD-1 and tBRD-2 are involved in the gene regulatory cascade in spermatocytes recently proposed by Lu and Fuller (2015). Therefore, we compared our tbrd-1 and tbrd-2 mutant transcriptome data with that of sa, aly, and med22 mutants (Lu and Fuller, 2015;Theofel et al., 2014). Indeed, a defined subset of 31 genes were regulated by all five factors. The transcripts of most of these genes are enriched in the testes and accumulate in post-meiotic germ cells (Chintapalli et al., 2007;Vibranovski et al., 2009), which suggests that these transcripts are among the translationally repressed mRNAs required for spermatid differentiation. In contrast to Sa, Aly, and Med22, tBRD-1 and tBRD-2 are involved in the regulation of only a small number of genes. Expression of known tTAF-, tMAC-, and Mediator-dependent spermiogenesisrelevant genes, e.g., fzo, janB, gdl and CG9173, is not affected in tbrd-1 and tbrd-2 mutants. Nevertheless, our data showed that tBRD-1, tBRD-2, Sa, Aly, and Med22 regulate a common set of genes. We hypothesize that tBRD-1 and tBRD-2 act at the end of a gene regulatory cascade involving tMAC, Mediator, and tTAF functions to regulate expression of spermiogenesis-relevant genes.

Fertility tests
Batches of 20 flies were tested for fertility. Adult males were crossed individually against two wild-type virgin females in separate vials at 25°C. After 6 days, the parental generation was removed. The number of offspring was counted after two weeks.

RNA isolation and microarray experiments
Total RNA was isolated from bam≫tbrd-2 RNAi , undriven tbrd-2 RNAi , and bam-Gal4 testes using TRIzol (Invitrogen). RNA quality was monitored using the Agilent Bioanalyser 2100 with the RNA 6000 Nano kit. Gene expression was analyzed using Affymetrix Drosophila Genome 2.0 arrays according to the manufacturer's recommendations. For each array, independent RNA from whole testes pooled from 25 animals was used. Three independent replicates were prepared for each experimental condition. The data were analyzed in the R statistical environment using BioConductor packages (Huber et al., 2015). Scanned data were parsed as CEL files into R using the /affy/ package (Gautier et al., 2004). Expression estimates were extracted using RMA normalization with the /rma/ function. Differentially expressed genes were identified using Limma (Ritchie et al., 2015). Genes with log2 (fold expression change) >1 or<−1 and an adjusted P-value <0.05 were selected as significantly up-or down-regulated, respectively.
For comparison with previously published data (Lu and Fuller, 2015), we downloaded CEL files from the GEO repository (GSE74784) and processed them as described above to obtain log2-transformed expression measures.
To make the different data sets comparable, initial RMA was applied to the complete data set.
The microarray data were deposited at the NCBI gene expression omnibus (GEO) under the accession number GSE81019.

Quantitative real-time PCR
Total RNA from 100 bam≫tbrd-2 RNAi testes, undriven tbrd-2 RNAi testes, and bam-Gal4 testes was extracted using TRIzol (Invitrogen). RNA was treated with RQ1 RNase-Free DNase (Promega). For cDNA synthesis, 1 µg DNasedigested RNA and the Transcriptor First Strand cDNA Synthesis Kit (Roche) were used. qPCR reactions contained 7.5 μl iTaq™ Universal SYBR ® Green Supermix (Bio-Rad), 5.2 μl ddH 2 O, 2 μl diluted cDNA, 0.3 μl (10 μM) genespecific primer 1, and 0.3 μl (10 μM) gene-specific primer 2. qPCR (three technical replicates) was performed with a Sybrgreen platform on a Bio-Rad CFX Cycler. Data were analyzed using Bio-Rad CFX Manager™ software. Values were normalized to the mRNA expression level of Rpl32. Differences between groups were determined with analyses of variance. Delta Ct values were analyzed for ANOVA using the aov function of R. For the differences between individual groups post hoc tests were calculated by Tukey's honest significant difference test (TukeyHSD function). Two groups were compared using one-way ANOVA. The corresponding P-values indicated in the figures are *P≤0.05, **P≤0.01, and ***P≤0.001.
Primers are given in Table S6.
Expression and purification of recombinant tBRD-1 and tBRD-2 tbrd-1 and tbrd-2 cDNAs were FLAG tagged at the C-terminus by PCR using specific primers and ligated into the baculovirus transfer expression vector pVL1392. Transfection of Sf9 cells, recombinant baculovirus production, and recombinant protein expression and purification essentially followed the methods described in Brehm et al. (2000).

Peptide pull-down experiments
H3 and H4 peptides were synthesized (PSL Peptide Specialty Laboratories) and coupled to SulfoLink™ coupling resin (Thermo Scientific) according to the manufacturer's instructions. One microgram of each peptide was added to 1 µl beads; 2.5 µl of the coupled beads were mixed with 17.5 µl uncoupled beads and washed in pull-down buffer (25 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2 mM EDTA, 0.5% NP-40, 1 mM DTT, proteinase inhibitors) for 5 min twice. After blocking for 1 h at 4°C in blocking buffer [1 mg/ml BSA, 1% fish skin gelatin (Sigma) in pull-down buffer], beads were incubated with 0.25 µg recombinant proteins for 2 h at 4°C. Beads were washed four times in pull-down buffer. Bound proteins were analyzed by SDS-PAGE and western blotting using tBRD-1-and tBRD-2specific antibodies; 20% of the input was loaded on the gel.