Transition from a nucleosome-based to a protamine-based chromatin configuration during spermiogenesis in Drosophila

The switch from a nucleosome- to a protamine-based chromatin structure is a characteristic feature of sperm maturation in many vertebrates including humans. First, testis-specific linker histones appear (Catena et al., 2006; Martianov et al., 2005; Tanaka et al., 2005; Yan et al., 2003), and then histones are replaced by transition proteins (major types: TP1 and TP2), which in turn are replaced by protamines, leading to condensed chromatin with a doughnut structure (reviewed by Braun, 2001; Kimmins and Sassone-Corsi, 2005; Sassone-Corsi, 2002). Recently, we have shown that Drosophila sperm also contain protamines as well as Mst77F, representing at least one further chromatin component, whereas transition protein-like chromatin components have not been found in Drosophila so far (Jayaramaiah-Raja and Renkawitz-Pohl, 2005). In Drosophila, GFP-tagged versions of the histone variant H2AvD are degraded during the canoe stage of spermatid development when protamines and Mst77F are starting to accumulate (Jayaramaiah-Raja and Renkawitz-Pohl, 2005). This raises the question of whether the histones and histone variants are depleted from the chromosomes simultaneously. However, to date, little is known about the mechanism of histone displacement and degradation, and it is also unclear whether protamine synthesis is a signal for histone degradation. Multiple histone modifications have been analysed that mark transcriptionally active and inactive chromatin either on the level of euchromatin versus heterochromatin or on the level of individual genes (Lachner et al., 2003; Peterson and Laniel, 2004; Shilatifard, 2006). In Drosophila spermatogenesis, transcription nearly ceases with the entry into meiotic division (for review see Renkawitz-Pohl et al., 2005), which is long before the protamine-based chromatin status is established. Histone modifications in the haploid phase might play a fundamental role in keeping the genome largely transcriptionally inactive. In addition, we also expected that histones to be displaced from the chromatin would be marked by characteristic modifications, as is known to be the case for histone H4 that is acetylated in mammals. Furthermore, phosphorylation and ubiquitylation were also suggested to be important during this stage (reviewed by Braun, 2001). Here, we show for Drosophila that histones carry multiple modifications, which also characterise the transcriptionally highly active spermatocyte stage. Interestingly, we found de novo H2A mono-ubiquitylation and a tremendous increase in H4 acetylation shortly before histone degradation in the spermatid nucleus. This finding might indicate an opening of the chromatin which is accompanied by the expression of a transition protein such as chromatin component (Tpl^94D), the presence of CTCF and DNA breaks. We furthermore show that histone removal is independent of the presence of protamines. Both this histone removal and protamine accumulation are essential for transmission of the male genome to the oocyte, and therefore of fundamental importance for the persistence of species.

In Drosophila melanogaster, sperm morphogenesis, i.e. from meiosis until sperm individualisation, lasts 3.5 days. After meiosis, the nucleus initially is round and then gradually changes its shape accompanied by reorganisation of the chromatin during the canoe stage (Jayaramaiah-Raja and Renkawitz-Pohl, 2005), resulting in sperm containing a slim nucleus in which the nuclear volume is decreased by a factor of 200 (for review see Fuller, 1993; Renkawitz-Pohl et al., 2005).

In the work reported here we concentrated on post-meiotic sperm morphogenesis with particular focus on chromatin reorganisation from the nucleosomal- to the protamine-based structure, which is a dramatic switch. Previously, we have reported that the histone variant H2AvD-GFP vanishes at the canoe stage while protamines begin to accumulate simultaneously (Jayaramaiah-Raja and Renkawitz-Pohl, 2005). To analyse the timing of histone removal and protamine accumulation we brought protamine-eGFP and H2AvD-RFP (Clarkson and Saint, 1999) into one genetic background to enable a study in the same individual. We found that H2AvD-RFP disappeared before protamine-eGFP accumulation took place (data not shown). We then went on to immunostain testes of protamine-eGFP flies with an antibody recognising all histones. This antibody was raised against total histones of humans and detects all core histones and the linker histone H1 in mammals. As – in contrast to core histones – H1 is not well conserved between mammals and Drosophila, we presumably detect solely core histones with this antibody. Our findings show that core histones (Fig. 1A) are detectable up to the canoe stage whereas protamines start to be synthesised at the canoe stage (Fig. 1B) but with no apparent positional overlap. As the canoe stage is quite long, we defined the early canoe stage by the start of histone removal, and the late canoe stage by the start of protamine accumulation (see Fig. 1A-E, second and third columns). With histone H3.3, a further replacement variant is expressed in the testis and disappears in post-meiotic stages together with the bulk of histones (Akhmanova et al., 1997). To reinvestigate this, we used an H3.3-Flag carrying Drosophila strain (Loppin, 2005). Histone H3-Flag (data not shown) and its variant H3.3-Flag (Fig. 1C) were observed to vanish along with the bulk of histones during the early canoe stage. Within the genus Drosophila, the timing of histone degradation is a well conserved feature, as our analysis of distantly related Drosophila species (D. hydei, D. mojavensis, D. pseudoobsura and D. virilis) revealed a comparable timing of the removal of histones at the canoe stage (data not shown).

We observed a relatively sudden switch between the presence of histones and of protamine-eGFP, when analysing the histone distribution in transgenic flies carrying protamine-eGFP (Fig. 1A,B). This sudden switch is difficult to explain without an intermediate stage. Therefore, we speculated that a functional homologue of at least one of the transition proteins must be present in Drosophila; however, sequence homology searches failed to reveal the presence of any mammalian-type transition proteins. The testis-specific and basic proteins Don Juan and Don Juan-like are transiently present in the spermatid nucleus (Hempel et al., 2006; Santel et al., 1998). Since these two proteins disappear along with core histones (data not shown), it is unlikely that they are the true homologues of mammalian transition proteins. However, we identified in FlyBase five predicted genes (CG4691, CG15510, CG32308, CG31861, CG31281) encoding very basic proteins, which are expressed in testes. For all of them we generated corresponding eGFP fusion genes and analysed their expression patterns (C.R. and R.R.-P., unpublished). With this approach, we identified only one new fusion protein (CG31281-eGFP), which is transiently expressed exactly at the time when the switch between histone- and protamine-based chromatin configuration occurs (Fig. 1E; compare with D for histone distribution in the same sample). We named this gene transition protein-like^94D (tpl^94D), because of its expression pattern and its location on the polytene chromosomes (FlyBase). We then investigated the distribution of Tpl^94D-eGFP with special focus on the canoe stage, where chromatin remodelling takes place. Tpl^94D-eGFP is first detectable at the early canoe stage, shortly before histones disappear (Fig. 1A-E, 2nd column) and most strongly expressed during the late canoe stage, when protamines are clearly visible and histones are completely gone (Fig. 1A-E, 3rd column). When protamines are fully expressed and the chromatin appears completely condensed, Tpl^94D-eGFP is no longer detectable (Fig. 1A-E, column 4). Tpl^94D is a 164 aa protein with a predicted molecular mass of approximately 18.8 kDa and a pI (isoelectric point) of 11.4. Tpl^94D shows no sequence similarity to the transition proteins of mice, but like these it has a high arginine (11.6 %) and serine (13.4 %) content. Tpl^94D contains a high mobility group (HMG)-box (underlined in Fig. 1F), which suggests that Tpl^94D is a component of the chromatin. The cell-type-specific expression in the testis shows that Tpl^94D is a chromatin component during the switch from histone- to protamine-based configuration. We suggest that this abundant high mobility group protein plays a key role in disrupting nucleosomal structure as has been suggested for recombination and transcription (Travers, 2003). Since the presence of histones and protamines is absolutely exclusive, and since Tpl^94D expression is characteristic for the time of chromatin switch, we think that Tpl^94D is a good candidate for a functional homologue of mammalian transition proteins.

We next addressed the question of whether Tpl^94D is expressed exclusively in the testes, as are transition proteins in mammals. We performed RT-PCR experiments using RNA from adult males without testes, from testes, from adult females, embryos and larvae. We see that the tpl^94D RNA is restricted to the testes in adult flies (Fig. 1G). We observe tpl^94D RNA in larvae but not in embryos, which reflects the transcription in the spermatocytes of larval testes. The testes-specific expression further supports its functional role as a transition protein.

It is well known that proteins are marked by ubiquitin for degradation in proteasomes (for a review, see von Mikecz, 2006; Nandi et al., 2006). At the stages relevant for histone degradation, proteasomes are detectable in spermatid nuclei and include subunits expressed specifically in the testis (Ma et al., 2002). Thus, we first applied an antibody against ubiquitin in general, to start the analysis of protein degradation. We observed that, in post-meiotic stages of spermatogenesis, ubiquitin is very abundant in the nuclei at the early canoe stage (Fig. 2, row A, arrowhead) and detectable up to the late canoe stage of chromatin reorganisation (Fig. 2A, double arrow). We also observed the testis-specific nuclear proteasome subunit Prosα3T-eGFP and the corresponding somatic type Prosα3-eGFP during the early and late canoe stage in the nucleus (Fig. 2B,C, arrowhead and double arrow). Prosα3T-eGFP stays in the individualising spermatids, while the corresponding somatic type, Prosα3-eGFP, vanishes during the late canoe stage [Prosα-eGFP lines see Ma et al. (Ma et al., 2002)]. This stage-specific abundance of ubiquitin and proteasomes corresponds to the time of histone and Tpl^94D degradation. As we observed ubiquitin in the nucleus during histone degradation, we next investigated male sterile mutants with mutations in ubiquitin-conjugating enzymes (E2) and ubiquitin ligases (E3). Homozygous mutants arrest spermatogenesis during spermatid development. In testes of homozygous mutants, we observed that histones disappear at the appropriate time in crossbronx, dbruce and effete⁸ (E2) and purity of essence (E3) mutants (Table 1). Therefore, we conclude that histone removal neither depends on these ubiquitin-conjugating enzymes nor on this ubiquitin ligase, although they are required for male fertility.

A general histone code is established whereby certain modifications indicate transcriptionally active chromatin whereas others indicate transcriptionally inactive regions of the genome. At the spermatocyte stage, histones carry multiple modifications (Table 2). We also observed all these modifications in post-meiotic stages (Table 2). Methylation of lysines 9 and 27 of histone H3 indicate repression, whereas methylation of lysine 4 of histone H3 indicates activation. Histone lysines can be mono-, di- or trimethylated (Lachner et al., 2003). Post-meiotic germ cells of Drosophila are virtually completely transcriptionally silent (for a review, see Renkawitz-Pohl et al., 2005); thus, histone methylation on lysine residues K9 or K27 might be expected in post-meiotic stages. We analysed mono-, di- and trimethylation of H3K9 as well as di- and trimethylation of H3K27 and found these modifications to be present from the nebenkern stage immediately after meiosis (data not shown) until the removal of histones (Table 2). Therefore, we conclude that the post-meiotic stages of male germ cells contain multiple histone methylations characteristic of transcriptionally inactive chromatin, in agreement with their transcriptionally silent status. At the same stages we also observed that N-terminal tails of core histones contain modifications which otherwise mark transcriptionally active chromatin, such as H3K4 or H4R3 methylation. At first glance, this appears surprising, but previously it was shown that trimethylation of H3K4 can also function in active gene repression (Shi et al., 2006). But it is also possible that H3K4 and H4R3 methylation at this stage might indicate overall opening of the chromatin as a prelude for histone removal. As a representative example, we show nuclear distribution of H3K4 trimethylation during the transition between histones and the protamine deposition in flies expressing protamine-eGFP (Fig. 3). The DNA staining in Fig. 3 shows the nuclei of three spermatid bundles with 64 synchronously developing spermatids each. Only the nuclei of the left bundle accumulated protamine-eGFP (Fig. 3C). This bundle cannot be visualised with antibodies recognising H3K4 trimethylation, but the middle and right bundles can (Fig. 3B). The level of H3K4 trimethylation during the different stages of chromatin reorganisation can be compared in the overlay presented in Fig. 3D, where the level of H3K4 decreases in early canoe stage spermatid nuclei (middle bundle) compared to young elongating spermatid nuclei with very high H3K4 methylation levels (right bundle), the stage directly before the canoe stage. The high levels of H3K4 and H4R3 methylation (see Table 2) as well as other modifications (see next subsection) shortly before histone degradation might be relevant for opening of the chromatin.

For mammals, hyper-acetylation of histone H4 is thought to cause loss of contact between histones and DNA. Therefore, to visualise protamine accumulation, histone removal and H4 acetylation in the same spermatids, we performed double immunostainings on testes squashes of a transgenic Drosophila line carrying protamine-eGFP (Fig. 4A-C). Indeed, we found a dramatic increase in H4 acetylation during the canoe stage immediately before histones are removed and degraded (see Fig. 4B for acetylated H4, and Fig. 4A for histones). In round spermatids (Fig. 4E, arrow) and young elongating spermatids (data not shown), we also observed mono-ubiquitylated histone H2A, which we solely find in these very early post-meiotic stages (Fig. 4, row E) and in late primary spermatocytes (data not shown). This is in clear contrast to H4 acetylation, which we also observed in early canoe stage nuclei (Fig. 4B) and in all pre-meiotic stages (data not shown). At these early post-meiotic stages, protamine-eGFP is not yet synthesised (Fig. 4C,F). Thus, we conclude that histone H4 hyper-acetylation and H2A mono-ubiquitylation shortly after meiosis and histone H4 hyper-acetylation at the canoe stage are indicative of the switch in chromatin architecture.

It is difficult to imagine how the chromatin with the core histone octamers surrounded by DNA can suddenly lose the nucleosomal structure all at once in the majority of the chromatin. In mammalian male germ cells, DNA breaks occur during chromatin remodelling (Laberge and Boissonneault, 2005). Local nicks or breaks in the DNA might facilitate the opening of the chromatin structure and thus lead to the removal of histones and the incorporation of Tpl^94D, protamines and Mst77F. To investigate this, we performed immuno- and TUNEL staining on testes squashes of protamine-eGFP transgenic flies, thereby allowing comparison of histone removal, protamine deposition and DNA breaks in male germ cells within one testis tube (Fig. 5A-C). At the early canoe stage when histones start to vanish (Fig. 5A, arrow), numerous nicks are present in the DNA (Fig. 5C, arrow). When histones are removed and protamine-eGFP starts to accumulate at the late canoe stage, the TUNEL signal is very strong, indicating numerous nicks in the DNA (see Fig. 5 column 2, arrowhead). Nicks are no longer detectable after the late canoe stage, an example is shown for the individualisation stages when protamine-eGFP is fully expressed (Fig. 5, column 3, double arrow in B). These observations indicate that chromatin remodelling is facilitated by local opening of the chromatin due to DNA nicks. Furthermore, we checked whether SUMO and UbcD6, the Drosophila homolog of Rad6 in yeast and HR6A and HR6B in mice, are expressed. Rad6 has multiple functions including involvement in postreplication repair (Baarends et al., 2003). HR6B knock-out mice show malformed spermatid nuclei indicating a role in post-meiotic stages in the nucleus (Roest et al., 1996). In addition, a high level of SUMO is characteristic of DNA repair (Branzei et al., 2006). Indeed, both SUMO and UbcD6 (detected with anti-HR6A/B, recognising HR6A and HR6B in mice, directed against the conserved N terminus) are highly expressed at the same time when DNA nicks are detected (Fig. 5C-E).

The switch between a nucleosome-based chromatin configuration and a protamine-based structure is a specialised form of chromatin remodelling in the male germline. The mammalian zinc finger protein CTCF is involved in many epigenetic processes (Ohlsson et al., 2001). Furthermore, paralogous variant of CTCF which is testis-specifically expressed, called BORIS, is exclusively expressed in the mammalian male germline. The function of BORIS in this context is still not clear (Loukinov et al., 2002). Recently, we identified CTCF in Drosophila and showed that Drosophila, in contrast to mammals, contains only one CTCF gene (Moon et al., 2005). We therefore asked whether Drosophila CTCF is also expressed in the testes, and performed immunostaining and anti-histone staining on testes of transgenic flies expressing protamine-eGFP. We observed CTCF expression during pre-meiotic and meiotic stages at the chromosomes (data not shown) as has been shown for mitotic cell division in mammalian cell culture (Burke et al., 2005). Shortly after meiosis, CTCF is visible in young elongating nuclei, where it co-localises with the chromatin as indicated by the histone distribution (Fig. 6, column 1). CTCF is also present in the early and late canoe stage spermatid heads. At the early canoe stage, CTCF is very diffusely distributed in comparison to histones (Fig. 6, column 2). CTCF does not co-localise with the chromatin which starts to condense at one side of the nucleus (Fig. 6A and C, arrow). This diffuse distribution is still visible at the late canoe stage (Fig. 6, column 3) when protamine-eGFP starts to be deposited to the chromatin. CTCF is no longer detectable after the canoe stage (Fig. 6C). The earlier chromatin-associated CTCF localisation might indicate a very early role in chromatin reorganisation at the switch between the nebenkern and canoe stage. Furthermore, CTCF might be associated primarily with the chromatin, which is not yet condensing during these stages [for a scheme of chromatin distribution based on EM studies, see Fuller (Fuller, 1993)]. The late canoe stage is the only post-meiotic stage where we find distinct regions of RNA polymerase II with an antibody directed against a phosphorylated subunit of active polymerase, which is indicative of transcription (Fig. 6D, double arrow). At this precise stage, only a very small set of genes is thought to be transcribed (H. White-Cooper, University of Oxford, UK, personal communication). Also in D. hydei we detected CTCF expression during chromatin reorganisation in the nucleus (data not shown).

We asked whether histone and CTCF degradation is dependent on protamine synthesis. To this end, we used a fly strain recently generated by FLP/FRT technology. This strain harbours a deletion (protΔ38.1) spanning the protamine genes, Mst35Ba and Mst35Bb, as well as three other ORFs from the genome (CG15279; CG4480; CG15278). Furthermore CG33309 is destroyed by one of the FRT sites. Mutant males of these strains show severely reduced fertility (S.J.-R. and R.R.-P., unpublished).

Unlike mammals, fly protamine mutants are not haploinsufficient. CTCF is removed in protamine loss-of-function mutants, as observed in wild type. Histone is not present in this mutant or in transheterozygous Mst77F mutants (ms(3)nc3 over Df(3L)ri-79c; data not shown). All the investigated histone modifications such as diverse lysine and arginine methylation (Table 2), H4 acetylation or H2A mono-ubiquitylation (data not shown) also occur in protamine loss-of-function mutants (protΔ38.1) as in wild-type flies. SUMO and ubiquitin also vanished in the mutant as in wild-type, and active RNA polymerase II was observed at the same time as in wild type (data not shown). We therefore conclude that histone and CTCF removal during the late canoe stage are independent of protamine synthesis involving Mst35Ba and Mst35Bb. Furthermore, we analysed the fate of histones in different known male sterile mutants (Table 3). Again, none of the mutants showed signs of impaired histone removal. In summary, we conclude that histone degradation and CTCF removal are regulated independently of the synthesis of protamines.

Sperm morphogenesis is characterised by an impressive degree of changes in cell architecture based on stored, translationally repressed mRNAs that are recruited at the appropriate time to the polysomes (for a review, see Renkawitz-Pohl et al., 2005). Among these are mRNAs that encode Tpl^94D and protamines. A dramatic switch in structure from the nucleosomal- to the protamine-based structure of chromatin takes place, and this remarkable chromatin reorganisation of the complete genome is a typical feature depending on stored mRNAs, e.g. for protamine synthesis (Jayaramaiah-Raja and Renkawitz-Pohl, 2005). This process ultimately leads to an extremely condensed state of the haploid genome in the sperm, which is essential for male fertility in mammals. Here, we focused on the switch between a nucleosomal- and a protamine-based chromatin reorganisation. We observed that the major steps in chromatin organisation take place in the canoe stage of spermatid development (summarised in Fig. 7). For the first time, we identified a candidate for a transition protein in Drosophila. The corresponding gene tpl^94D (CG31281) encodes a predicted basic high mobility group (HMG) protein of 18.8 kDa. In transgenic flies, Tpl^94D-eGFP fusion proteins are expressed solely during the switch between histones and protamines, as is typical for mammalian transition proteins. Since we observe a highly similar chain of events to those reported in mammals, we consider the Drosophila system an excellent choice to study the mechanism of chromatin remodelling during male germ cell development.

Generally, we observe that the bulk of histones, including their diverse modifications in the N-terminal tail, appear to be removed during the canoe stage. Furthermore, we show that the nucleus accumulates ubiquitin at the early canoe stage, when mono-ubiquitylation of histone H2A is no longer detectable (compare Fig. 2A and Fig. 4E). Therefore, taking into account the known presence of proteasomes in the nucleus at this stage of chromatin reorganisation (Ma et al., 2002) and the overlap of expression shown here, we hypothesise that this ubiquitylation is targeting histones for degradation. Here, we investigated several mutants having mutations in ubiquitin-conjugating enzymes or ubiquitin ligases, exhibiting arrested spermiogenesis during spermatid development and that are male sterile. However, in all investigated mutants, histone removal is indistinguishable from that of wild-type flies.

We found many histone modifications after meiosis (see Table 2 and Figs 2, 3, 4) and have categorised them into three classes.

1. Histone modifications that persist from pre-meiotic stages and keep the genome silent. The vast majority of the genome is transcriptionally silent in post-meiotic stages. This is accompanied by multiple histone modifications that persist from pre-meiotic stages and indicate silencing such as H3K9 and H3K27 methylation. These modifications do not change significantly during post-meiotic stages, which is in agreement with our hypothesis that these modifications predominantly play a role in maintaining transcriptional silencing. Previously, phosphorylation of histones have been analysed duringspermatogenesis. Phosphorylated histone H4S1 and H3S10 are present during meiotic divisions. H3S10 phosphorylation is hardly detectable after meiosis, whereas phosphorylation of H4S1 persists until chromatin compaction starts (Krishnamoorthy et al., 2006).

2. Histone modifications that persist from pre-meiotic stages and characterise transcriptionally active chromatin. The primary spermatocyte phase is characterised by a high level of transcriptional activity of housekeeping genes. In addition, genes are transcribed that are needed for the subsequent steps in spermatogenesis, as the majority of transcription ceases once meiotic division starts. Here, we investigated H4 acetylation and H3K4 and H4R3 methylation of histones (Fig. 4; Table 2). These histone modifications, which are indicative of transcriptional activity, persist until histone degradation.

3. Increasing or de novo appearance of histone modifications that decrease the affinity between histones and DNA as a prelude to histone removal. It might be that H4 hyper-acetylation, as postulated for mammals (reviewed by Braun 2001; Meistrich et al., 1992) and/or other secondary modifications of histones are the first step towards histone removal. The fact that these modifications are conserved between mammals and flies adds support to this hypothesis. Indeed, histone H4 acetylation is very pronounced at the canoe stage and we see de novo mono-ubiquitylation of histone H2A in round spermatids. Both types of histone modifications we propose to be necessary for opening the chromatin and decreasing the contact between DNA and histones. The fact that histone H2A mono-ubiquitylation vanishes before the early canoe stage, thus before the hyper-acetylation of histone H4, leads us to think about a stepwise remodelling of the chromatin. Here, we propose that these histone modifications open the chromatin, so that enzymes and regulators have access to histone-based chromatin and can induce and prepare the reorganisation of the genome in the male germline.

It remains to be clarified whether and how these histone modifications influence the topology of the chromatin as a prelude to histone removal as well as for Tpl^94D, Mst77F and protamine deposition. A functional approach based on analysis of mutants of histone-modifying enzymes is difficult, as all characterised histone-modifying enzymes are already active during Drosophila development or at least in spermatogonia and spermatocytes. Therefore a tissue-specific knock-out mutant would most probably exhibit arrest of spermatogenesis before meiosis, rendering it useless for our purposes.

At the first glance, it might seem surprising that histones and all their modifications are removed. Instead of specifically reverting the differentially modified histones to their unmodified state, they are removed together with all histones. This might allow the paternal genome to form nucleosomes with unmodified histones after fertilisation and before zygote formation. Thus, the paternal genome starts embryogenesis with a nucleosomal chromatin lacking histone modifications.

Our data show that most of the histones are removed between the early and late canoe stage; such a process requires a loosening of contact between the histones and DNA, which in turn requires an unwinding of the chromatin structure. We propose that this unwinding process is facilitated by DNA nicks as they were widespread at this stage of chromatin reorganisation. Finally, Tpl^94D, UbcD6 and SUMO were also observed to accumulate in the chromatin during this process. DNA breaks, Tpl^94D, UbcD6 and SUMO were no longer detectable when protamines were fully expressed. Thus, we propose that all these proteins and the DNA breaks act together in an unknown manner to allow chromatin remodelling.

We observed that the CTCF protein is present during pre-meiotic stages in the nucleus and stays associated with the chromosomes during meiosis. After meiosis, however, we detected strong localisation to the nucleus during the transition from round spermatid nuclei to the early canoe stage of spermiogenesis. We speculate that CTCF might set borders in the chromatin for the histone modifications, which are characteristic of the canoe stage, such as acetylation and ubiquitylation. CTCF is visible for longer than histones and disappears together with active RNA polymerase II. CTCF might maintain chromatin accessibility to RNA polymerase II as a few genes are known to be transcribed at this time (H. White-Cooper, personal communication). In addition, transient occurrence of RNA polymerase II at the late canoe stage might require CTCF to insulate active genes from inactive ones. This idea needs to be tested in tissue-specific CTCF loss-of-function mutants; such mutants are, however, currently unavailable. CTCF expression in comparison to other features of chromatin remodelling is shown in Fig. 7.

We addressed the question of whether histone removal is dependent on a signal that monitors the start of protamine and Mst77F mRNA translation. Both histone modification and degradation are indistinguishable from the wild-type in loss-of-function mutants of Mst35Ba and Mst35Bb, the genes encoding protamine A and B, respectively. Also in nc3 mutants of Mst77F, histone removal is not disturbed. We conclude that N-terminal tail modification of histones and histone degradation, on the one hand, and protamine deposition, on the other, are controlled by different pathways in the cell.

In mammals, it is well known that after meiosis the nucleosomal conformation is lost. This is accompanied by the appearance of testis-specific linker histones (Catena et al., 2006; Martianov et al., 2005; Tanaka et al., 2005; Yan et al., 2003). So far, no linker histone variants have been identified in Drosophila, but variants of H2A (H2AvD) and H3 (H3.3) are known (Akhmanova et al., 1997; Swaminathan et al., 2005). In mammals, histones are hyper-acetylated before being displaced from the DNA, and phosphorylation and ubiquitylation have also been proposed to occur (reviewed by Braun, 2001). For Drosophila, we clearly show H2A mono-ubiquitylation and a strong increase in H4 acetylation shortly before histone removal and degradation. In mammals, histones are replaced first by transition proteins (major types: TP1 and TP2). In the investigation reported here we identified the high mobility group protein Tpl^94D, a first probable candidate for a functional homologue of mammalian transition proteins. In mammals, transition proteins are subsequently replaced by protamines leading to chromatin with a doughnut structure (reviewed by Braun, 2001; Kimmins and Sassone-Corsi, 2005; Sassone-Corsi, 2002). In Drosophila, we have recently shown that the sperm nucleus also contains protamines. Protamines A and B are encoded by two closely related protamine genes, Mst35Ba and Mst35Bb. In addition, the identification of Mst77F shows that sperm nuclei contain at least one further abundant chromatin component (Jayaramaiah-Raja and Renkawitz-Pohl, 2005). Moreover, in human sperm several new putative protamines have been identified by 2D gel electrophoresis and protein sequencing (Yoshii et al., 2005). In mammals, this chromatin reorganisation is essential for male fertility (reviewed by Kimmins and Sassone-Corsi, 2005). Male flies carrying the deletion protΔ38.1, where both protamines as well as three additional ORFs are removed, show severely reduced fertility (S.J.-R. and R.R.-Pohl, unpublished).

In summary, we propose a step-by-step scheme for chromatin reorganisation (Fig. 7). First, histone modifications lead to subsequent histone removal and degradation. Second, the exposed chromatin becomes nicked, resulting in DNA breaks. Finally, Tpl^94D deposition constitutes an intermediate stage that triggers subsequent protamine-based chromatin organisation.

Since many features concerning spermiogenesis are conserved between Drosophila and mammals, we propose that Drosophila is an ideal system to gain further insight into the mechanism of chromatin reorganisation in spermatid nuclei, a process that is crucial for male fertility.

Drosophila flies were maintained on standard medium at 25°C. w was used as the wild-type strain. Visible markers and chromosome balancers are as described in FlyBase unless otherwise specified. bel^cap-1, bet¹, bln¹, blw¹, Bsg^gel, cbx⁰⁵⁷⁰⁴, Df(3L)ri-79c, eff⁸, emm¹, esc¹, gish⁰⁴⁸⁹⁵, gor¹, goul¹, mlt¹, mod⁰⁷⁵⁷⁰, ms(2)29F⁰⁷⁷¹⁷, ms(2)42D⁰⁹⁹⁶⁷, ms(3)72D⁰³⁹⁵⁷, ms(3)80⁰³⁸¹⁷, ms(3)85D⁰³⁵⁶⁵, ms(3)98B⁰⁶³⁰², pea¹, poe⁰¹⁶⁵⁹, sdl^05090a, tho¹ and tmx¹ were obtained from the Bloomington Stock Center. protamine-eGFP and protΔ38.1 fly lines were generated by S.J.-R. (Jayaramaiah-Raja and Renkawitz-Pohl, 2005) (S.J.-R., unpublished). Prosα3-eGFP and Prosα3T-eGFP were kindly provided by J. Belote (Ma et al., 2002). ms(3)nc3 was obtained from M. T. Fuller (Fuller, 1986). H3-Flag and H3.3-Flag fly lines were provided by B. Loppin (Loppin et al., 2005), and H2Av-RFP flies by C. Wenzl (Clarkson and Saint, 1999). dbruce^E81 flies were obtained from H. Steller (Arama et al., 2003), the Drosophila hydei strain were from M. Schäfer (University Kassel), and D. pseudoobscura, D. mojevensis and D. virilis strains from U. Homberg (University Marburg).

Total RNA was prepared using TRIzol® (Invitrogen) from 80 testes, 40 carcass of males, 40 adult females, embryos (7-24 hours) and 40 larvae. We used the OneStep RT-PCR Kit (Qiagen) to amplify a 361 bp cDNA fragment from the open reading frame of tpl^94D. The chosen primer pair spans an intron of 68 bp to distinguish between RNA and DNA. As a further control for DNA contamination the same RT-PCR was performed using primers for the β3-tubulin gene spanning an intron as well. β3-tubulin is expressed in embryos, larvae, adult males and females and the testis. The same conditions were always used for the tpl^94D and the β3-tubulin primers in RT-PCRs.

To analyse the expression pattern of tpl^94D, the open reading frame (ORF) together with 446 bp upstream of the ATG start codon was cloned and fused with eGFP. Therefore, the ORF together with the promoter and 5′ UTR was amplified by PCR using genomic DNA and primers with linked KpnI and XbaI restriction sites. The PCR fragment was inserted into pKS⁺eGFP in frame with the eGFP. This clone was digested with KpnI and NotI and the resulting tpl^94D-eGFP fragment subcloned into the germline transformation vector pChabΔsal, which was opened with KpnI and NotI (Thummel et al., 1988). Prior to the last step, pChabΔsal was digested with XbaI to cut out the lacZ, and then religated.

Transgenic fly strains were established by injection of DNA as described by Michiels et al. (Michiels et al., 1993).

Hoechst staining was used to visualise chromatin. All antibodies were used in immunofluorescence stainings of squashed testes carried out essentially as described in Hime et al., 1996. Squashed testes treated with fluorescent antibodies were embedded in Fluoromount-G (Southern Biotech, Birmingham, AL, USA). To follow the fate of the histones, an anti-histone antibody from Chemicon (1:1200) was used. This antibody recognises H1 and all core histones of human, mouse, bovine and hamster. To study lysine methylation of histones during spermatogenesis, the following antibodies were used: anti-monomethyl-histone H3(K4), (Upstate; 1:1000; species cross-reactivity: human); anti-dimethyl-histone H3(K4), (Upstate; 1:500; species cross-reactivity: human and Tetrahymena); anti-dimethyl-histone H3(K4), (Abcam; 1:1000; species cross-reactivity: all mammals and a wide range of other species including D. melanogaster); anti-trimethyl-histone H3(K4), (Upstate; 1:1000; species cross-reactivity: human; also recognises to a lesser extent dimethylated K4); anti-trimethyl-histone H3(K4), (Abcam; 1:500; species cross-reactivity: all mammals and a wide range of other species including D. melanogaster; weak reactivity to dimethylated K4); anti-monomethyl-histone H3(K9), (Upstate; 1:250; species cross-reactivity: human); anti-dimethyl-histone H3(K9), (Upstate; 1:50; species cross-reactivity: human, fission yeast and chicken); anti-dimethyl-histone H3(K9), (Abcam; 1:200; species cross-reactivity: all mammals; recognises also di- and trimethylated K27); anti-trimethyl-histone H3(K9), (Upstate; 1:100; species cross-reactivity: human – also recognises to a lesser extent dimethylated K9); anti-trimethyl-histone H3(K9), (Abcam; 1:50), (species cross-reactivity: all mammals and a wide range of other species including D. melanogaster; cross reacts with trimethylated K27); anti-dimethyl-histone H3 (K27), (Upstate; 1:50; species cross-reactivity: human); anti-trimethyl-histone H3(K27), (Abcam; 1:100; species cross-reactivity: all mammals and a wide range of other species including D. melanogaster; cross reacts in western blot analysis to a lesser extent with trimethylated K9). To study arginine methylation of histones, anti-methyl-histone H4 (R3), (Upstate; 1:100); species cross-reactivity: human) was used (a gift from U. M. Bauer). To study acetylation of histones, anti-acetyl-histone H4 (Upstate; 1:300, a gift from M. Eilers) was used. This antibody recognises acetylated K5, K8, K12 and K16 of human and Tetrahymena. To study ubiquitylation and SUMOylation during the histone to protamine exchange, anti-ubiquitin (DAKO; 1:10; species cross-reactivity: human), anti-ubiquitylated-histone H2A (IgM, Upstate; 1:100; species cross-reactivity: human, monkey, rat, mouse and frog) and anti-SUMO-1 C-term (Abgent; 1:400; species cross-reactivity: Drosophila, human; a gift from G. Suske) was used. Active RNA polymerase II was detected with anti-RNA polymerase II (H14) (IgM, Babco; 1:50; a gift from M. Eilers; species cross-reactivity: yeast to human). UbcD6 was detected with anti-HR6A/B (1:100; recognises HR6A and HR6B in mice, directed against the conserved N terminus) (Baarends et al., 2005). Anti-CTCF antibody (C-term; 1:200) was created by Moon et al. (Moon et al., 2005). To visualise the Flag-tag, an anti-Flag antibody was used (Cell Signaling; 1:500). To visualise IgG antibodies, anti-rabbit Cy3-conjugated (Dianova; 1:100) and anti-mouse Cy5-conjugated (Dianova; 1:100) antibodies were used. To visualise IgM antibodies, a FITC-labelled goat anti-mouse IgM (Sigma; 1:100) or a Cy3-conjugated donkey anti-mouse IgM (Dianova, 1:100) was used.

Immunofluorescence and eGFP samples were examined using a Zeiss Apotome microscope equipped with appropriate fluorescence filters. Images were individually recorded and processed with Adobe Photoshop 7.0.

For TUNEL stainings the `In Situ Cell Death Detection Kit, TMR Red' from Roche was used. The stainings were performed on squashed testes together with antibody stainings. After incubation with the secondary antibody, the slides were air dried and then incubated with 5 μl enzyme solution and 45 μl labelling solution for 1.5 hours at 37°C in a humid chamber in the dark. Stainings were analysed by fluorescence microscopy.

We thank A. Grootegoed for stimulating discussions. We thank S. Awe for critical reading of the manuscript. We are grateful for antibodies from U. M. Bauer, M. Eilers and G. Suske. We are also grateful for fly strains from J. M. Belote, M. T. Fuller, B. Loppin, U. Homberg, M. Schäfer, H. Steller and C. Wenzl. We thank Nadine Müller for excellent technical assistance and Katja Gessner for competent secretarial assistance. This work was supported by the Deutsche Forschungsgemeinschaft [Forschergruppe 531: “Chromatin mediated biological decisions” (Re628/12-2) and the International Graduate School GRK767 (Marburg, Giessen, Rotterdam)] to R.R.-P.