In Drosophila, don juan and don juan like encode proteins of the spermatid nucleus and the flagellum and both are regulated at the transcriptional level by the TAFII80 cannonball while translational repression is achieved by distinct elements


  • Leonie U. Hempel,

    1. Philipps-Universität Marburg, Fachbereich Biologie, Entwicklungsbiologie, Marburg, Germany
    Current affiliation:
    1. Laboratory of Cellular and Developmental Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892
    Search for more papers by this author
  • Christina Rathke,

    1. Philipps-Universität Marburg, Fachbereich Biologie, Entwicklungsbiologie, Marburg, Germany
    Search for more papers by this author
  • Sunil Jayaramaiah Raja,

    1. Philipps-Universität Marburg, Fachbereich Biologie, Entwicklungsbiologie, Marburg, Germany
    Search for more papers by this author
  • Renate Renkawitz-Pohl

    Corresponding author
    1. Philipps-Universität Marburg, Fachbereich Biologie, Entwicklungsbiologie, Marburg, Germany
    • Philipps-Universität Marburg, Fachbereich Biologie, Entwicklungsbiologie, Karl-von-Frisch- Str. 8, 35043 Marburg, Germany
    Search for more papers by this author


The genes don juan (dj) and don juan like (djl) encode basic proteins expressed in the male germline. Both proteins show a similar expression pattern being localized in the sperm heads during chromatin condensation and along the flagella. Prematurely expressed Don Juan–eGFP and Myc-Don Juan Like localize to the cytoplasm of spermatocytes and in mitochondrial derivatives from the nebenkern stage onward suggesting that both proteins associate with the mitochondria along the flagella in elongated spermatids. Premature expression of Myc-Don Juan Like does not impair spermatogenesis where-as Don Juan-eGFP when prematurely expressed causes male sterility as spermatids fail to individualize. In spite of the sequence identity of 72% on the nucleotide level and 42% on the protein level, the presumptive promoter regions and the untranslated regions of the mRNA are diverged. Our in vivo analysis revealed that don juan and don juan like are transcriptionally and translationally controlled by distinct short cis regulatory regions. Transcription of don juan and don juan like depends on the male germ line specific TAFII80, Cannonball (Can). Translational repression elements for both mRNAs are localized in the 5′ UTR and are capable to form distinct secondary structures in close proximity to the translational initiation codon. Developmental Dynamics 235:1053–1064, 2006. © 2006 Wiley-Liss, Inc.


During spermatogenesis, germ cells are first mitotically amplified, then they enter the meiotic prophase, pass through meiosis, and assemble the highly complex sperm within several days (for a review, see Fuller,1993). In postmeiotic stages, the nucleus is shaped to a needle-like form, which is accompanied by chromosome condensation. In a complex morphogenetic process, the mitochondria fuse and form the nebenkern, which then unravels and elongates to the major and minor mitochondrial derivative along the developing axoneme.

During sperm individualization, the male germline cyst is separated into single gametes by packaging each spermatid into its own plasma membrane. A cytoskeletal membrane complex assembles at the nuclear end of the cyst, which proceeds along the cyst as the so-called individualization complex. The region in which individualization occurs is enlarged and called cystic bulge (Tokuyasu et al.,1972).

It is a unique and characteristic feature that the postmeiotic stages lack transcription nearly completely in Drosophila, while in mammals a short postmeiotic transcriptional phase exists (for review see Fuller,1993, Schäfer et al.,1995, Renkawitz-Pohl et al.,2005). During the primary spermatocyte growth phase of meiotic prophase, the cell size increases and the nucleus is highly transcriptional active (Olivieri and Olivieri,1965). During this time, many mRNAs are synthesized that are translationally repressed until the appropriate time during sperm morphogenesis when the encoded protein is required for spermatid differentiation. The transcription of these translationally repressed mRNAs, for example dj mRNA (Santel et al.,1998), underlies a distinct control, as they are dependent on the tissue specific dTAFII80 homologue Can in the transcription initiation complex (White-Cooper et al.,1998; Hiller et al.,2001). The analysis has been pioneered by the identification of a translational control element (TCE) of 12 nucleotides in the 5′ untranslated region in the Mst(3)CGP gene family (Kuhn et al.,1988; Schäfer et al.,1990; Kempe et al.,1993). The tissue-specific transcripts of the dhod and the janus B gene contain an element with sequence similarity to the TCE of the Mst(3)CGP gene family (Yanicostas and Lepesant,1990; Yang et al.,1995). During sperm morphogenesis, the required proteins have to be synthesized in a coordinated timing. Therefore, it might be expected that individual mRNAs contain distinct translational repression elements. DJ indeed is synthesized earlier than the proteins and the functional analysis of the dj mRNA revealed a distinct translational repression element, TRE, within a stem and loop structure shortly before the translational initiation codon (Blümer et al.,2002). In addition to the cis regulatory elements, interacting proteins are required that bind to these sequences and keep them in a complex inaccessible to polysomes. Analyses of several RNA-binding proteins, such as Arrest, Tsr, Boule, and Rb97D, revealed that they are required for male fertility (for review, see Renkawitz-Pohl et al.,2005). So far, Boule is the only RNA binding protein for which a target mRNA, twine, is known (Eberhardt et al.,1996). Otherwise, there is no direct correlation between the characterized cis regulatory elements, TCE and TRE, and these RNA-binding proteins.

Here we characterize the expression of djl mRNA, which is transcribed in primary spermatocytes and translated with several days' delay in the elongated spermatid stage. Myc-DJL localizes to the sperm head during chromatin condensation and accumulates in the elongated flagellum but is not present in the mature sperm. We show that the DJL protein is 42% identical to DJ, which might suggest functional redundancy. Both proteins have a strongly overlapping expression pattern and their transcription depends on the tissue-specific TAFII80 Can. Their transcripts, however, are translationally controlled by distinct translational repression elements.


We previously characterized DJ as a protein that is expressed late in elongated spermatids and is localized in the mitochondrial derivatives of spermatids and mature sperm as well as transiently in sperm heads (Santel et al.,1998). Before the elongated spermatid stage, dj mRNA is translationally silent due to a translational repression element (TRE) in the 5′ UTR. 3′ of the dj gene, we found an open reading frame CG1984 (FlyBase), which is transcribed in the same direction as dj and encodes a basic protein showing 42% sequence identity to DJ. Therefore, we named this gene don juan like (djl). The question arose whether DJL is a somatic version of DJ or expressed in spermatids as well.

don juan and don juan like Encode Closely Related Proteins

The predicted DJL protein is 286 amino acids in length with a pI of 9.63 and a calculated molecular weight of 33 kDa. Many features are common and shared between the proteins DJ and DJL, which suggests that these genes arose by gene duplication. Both carry a mitochondrial import sequence in the N-terminal part and a cleavage site for the mitochondrial peptidase (Fig. 1A). The C-terminal regions are characterized by a high content of lysine and bipartite nuclear localization signals. Overall, the sequence homology is very high; the DJL protein comprises 38 amino acids more than DJ as it contains insertions and several additional Lysins at the C-terminus. Data base predictions give a 91.3% probability that DJL localizes to the nucleus, but only an 8.7% probability for a mitochondrial localization (PSORT II, Naika and Kanehisa,1992). Drosophila yakuba also contains dj and djl in tandem. However, dj accumulated stop codons so that only djl can encode a protein (see supplementary material and

Figure 1.

A: The primary structures of DJ and DJL bear N-terminal mitochondrial localization signals (blue) and a sequence that matches the consensus sequence of the mitochondrial peptidase (underlined in red). The predicted protein cleavage sites are marked by an arrow. Several nuclear localization signals are underlined in green. Identical amino acids are highlighted in yellow. B: RT-PCR experiments show mRNA with different 5′ primers starting −199 bp (lanes 1,2), 106 bp (lanes 3,4), 96 bp (lanes 5,6), and 74 bp (lane 7,8) upstream of the AUG. The 3′primer was chosen downstream of the intron so that priming at DNA results in a different fragment size (lanes 1,3,5,7). As a control primers in the reading frame were used (lanes 9,10) (see also Fig. 2A). Lanes with odd numbers represent PCR fragments primed on DNA, even numbers on mRNA. C: dj and djl mRNAs are compared. The AUG and the stop codons are underlined in red. The position of the introns is indicated by arrows. The putative polyadenylation signal is underlined in green.

Figure 2.

A: RT-PCR of djl from wild type testes (t), whole males (m), whole females (f), larvae (l), and embryos (e). In the control reaction (c), no RNA was used. The djl specific primers amplified a 668-bp fragment from the open reading frame of djl. Very low levels of transcript were detected in the adult female sample. A': A 397-bp fragment of the β3 tubulin gene was amplified as loading control. B: Performing in situ hybridization with a djl DNA probe, we revealed that the djl transcripts are abundant from the primary spermatocyte stage (arrow) onward until the stage of elongated spermatides (double arrow) resembling the dj transcript pattern (C). Only stem cells and spermatogonia situated at the testis tip are free of staining (arrowhead in B–E). D: Part of a the whole testis shown in E. E: A whole testis is shown in phase contrast optics to visualize the position of spermatogonia (arrowhead), spermatocytes (arrow), and elongated spermatids (double arrow).

As we are interested in transcriptional and translational regulation, we determined the transcription initiation site by RT-PCR (Fig. 1B). As −11 to +95 bp are still functional (see below for reporter gene assays), we analyzed this region for potential initiator elements. Two putative initiator elements are localized at position 95 and at position 77 relative to the AUG translational start. We chose 3′ primers downstream of the small intron (Fig. 1C, arrowhead) and 5′ primers (see Experimental Procedures section) in the promoter region to distinguish between these initiators. We obtained a RT-PCR product with a 5′ primer starting 96 bases upstream of the AUG (Fig. 1B, lane 6), but not with a 5′ primer starting 106 bases upstream of the AUG (Fig. 1B, lane 4). Therefore, we favor the initiator at position 95 upstream of the ATG as regulator of transcription for djl. This allows comparison between the complete mRNAs of dj and djl (Fig. 1C). The 5′ UTR comprises 117 bases for dj and 95 bases for djl. The polyA signal starts 308 bases after the stop codon of dj and 81 bases after the stop codon of djl. We observe considerable sequence conservation between the 5′ UTRs but not between the 3′ UTRs of D. melanogaster for djl as well as for dj.

don juan like Is Transcribed in Male Germ Cells

We aimed to clarify whether djl is expressed solely in the testes or whether this gene has a broader expression pattern. We performed RT-PCR experiments with polyA+-mRNA from male and female flies, third instar larvae, embryos, and testes. In females, djl mRNA is of very low abundance, in embryos the mRNA cannot be detected at all, while using polyA+-mRNA of larvae (male and female), male flies and testes in RT-PCR experiments resulted in an abundant PCR product (Fig. 2A). This led us to propose that djl is transcribed mainly in the testes. We next performed in situ hybridizations with djl DNA probes to testes of adult male flies. In the adult testes, djl mRNA is limited to germ cells. Transcription starts with the entry into meiotic prophase I, which is in the early spermatocyte stage (Fig. 2B, arrow) and transcripts remain detectable until the late spermatid stage (double arrow). Thus, the expression pattern is identical to that of dj mRNA (Fig. 2C). The dj mRNA is translated at the elongated spermatid stage and we asked also whether the protein distribution of DJL is identical to DJ.

Don Juan Like Is Transiently Localized in the Heads of Spermatids and in the Mitochondrial Derivatives of Elongated Spermatids

For DJL, mainly a nuclear localization is predicted as it is for DJ. As we observed that the majority of DJ resides within the mitochondrial derivatives (Santel et al.,1998) despite a prediction with 87% probability of being in the nucleus, we determined the subcellular localization in vivo by a Myc and an eGFP-tagged version of DJL (Fig. 3).

Figure 3.

A: Graphic of the myc-djl fusion construct used for establishing transgenic flies. Restriction sites, which were used to generate the transgene, are shown. Five myc epitopes were fused in frame to the djl N-terminus. The fusion gene is under the control of the natural djl regulatory sequence (−555 to +95). B: Schematic diagram of the djl-eGFP transgene used for P-element mediated transformation. The eGFP open reading frame was fused in frame to the djl C-terminus. This fusion gene is under the control of the natural djl regulatory sequence (−555 to +95). C,G,K: Localization of Myc-DJL in elongated spermatid bundles by Anti-Myc immunofluorescence. D,H: Hoechst staining to visualize chromatin. E,I: corresponding phase contrast. F: overlay of C,D, and E. J: Overlay of G,H and I. K: A spermatid bundle at a higher magnification. Anti-Myc immunofluorescence is visible in nuclei during chromatin condensation (C,F,K, arrow; compare to DNA staining in D), but not in nuclei of individualized sperm (G,J, double arrow; compare to DNA staining in H). J: Flagella of elongated spermatids show a high level of Myc-DJL (G,J, arrowhead), while flagella of individualized sperm lack staining (G,J, arrow).

For the design of the constructs, we considered that in general transcriptional control regions that regulate expression in the male germ line are very short (for review, see Renkawitz-Pohl et al.,2005). As the djl mRNA might be under translational control, we included the 5′ UTR. Therefore, both fusion genes were constructed with −555 to +95 bp of the djl gene. While five epitopes of the proto-oncogene c-myc were fused to the N-terminal end of the djl open reading frame (Fig. 3A), the sequence of the enhanced green fluorescent protein (eGFP) of Aequorea victoria was fused to the C-terminus (Fig. 3B). Indeed, this region was sufficient for the testis and cell type specific expression of djl. The subcellular expression pattern of the fusion proteins in testes of transgenic flies carrying the fusion constructs was analyzed. Using an anti-Myc antibody, we could detect the fusion protein Myc-DJL in elongated spermatids only. Because eGFP fused C-terminally to DJL showed no autofluorescence, we used an anti-eGFP antibody to localize the fusion protein DJL-eGFP. The two fusion proteins with an N- or C-terminal tag show no difference in their expression pattern during spermiogenesis. Both fusion proteins are detectable in the nucleus during chromatin condensation (Fig. 3C,F, arrows) (for a higher magnification see Fig. 3K). At this time, the fusion proteins are also localized along the entire flagella of the spermatid bundles (Fig. 3G,J, arrowheads). After individualization, both DJL fusion proteins do not remain in the sperm either in the nucleus or in the flagellum (Fig. 3G,J, double arrow, arrow). This is clearly a difference with DJ-eGFP, which remains in the flagella of individualized sperm. We could clearly show that there is a large discrepancy in time between the transcription and translation of djl mRNA. The transcripts appear in early spermatocytes but they are translationally repressed for several days until the elongated spermatid stage as we showed previously for dj.

Transcription of don juan like Is Controlled By Short Upstream Sequences Distinct From the don juan Regulatory Region

We showed so far that dj and djl share transcript distribution, translational repression, and the time of translation in the elongated spermatid phase. Thus, it might be anticipated that these genes are under the same transcriptional control and that their mRNAs underlie the same translational repression mechanism. Sequence comparison, however, gave no evidence for conserved elements besides AU-rich stretches.

Gradually deleting the 5′ upstream region of the djl gene, we analyzed the size of this regulatory region. We generated four promoter lacZ constructs (−555 to +95; −244 to +95; −104 to +95; −11 to +95) and established transgenic Drosophila lines. This revealed that −11 to +95 still allow a correct cell type specific transcriptional and translational regulation of the djl reporter gene construct (Fig. 4A,B). Our in situ hybridization to testes of the transgenic flies revealed that the lacZ mRNA has the same distribution in male germ cells as observed for the endogenous djl transcripts (Fig. 5B,D,F). The β-Galactosidase activity solely appears in elongated spermatid bundles (Fig. 4B, arrow) while stem cells and spermatogonia at the testis tip (double arrow) and also the following spermatocytes as well as meiotic and early postmeiotic stages, which are located in the inner curve of the testis (arrowhead), are free of staining. Thus the reporter gene expression reflects the expression of Myc-DJL in space and time, which shows that the 5′ UTR is sufficient for translational repression. Using the computer program mfold 2.3 (Jaeger et al.,1989; Walter et al.,1994; Zuker et al.,1999), we predicted that the 5′ UTR is capable to form defined secondary structures with stems and loops (Fig. 4D). It was calculated with the temperature parameter set to 24°C since this corresponds to fly culture conditions.

Figure 4.

A: Schematic drawing of the shortest djl reporter construct showing a specific expression pattern. Restriction sites used for the generation of the fusion gene are shown. B: Analysis of the β-Galactosidase activity in testes of this transgenic fly strain. While in premeiotic stages (double arrow) and also in meiotic stages and early spermatids (arrowhead), reporter protein activity is clearly absent, staining can be observed in elongated spermatid bundles (arrow). C: Analysis of the β-Galactosidase expression driven under the control of the djl regulatory region (−555 to + 95) in testes of flies with mutated genes for RNA-binding proteins. D: Predicted secondary structure of djl 5′ UTR including the AUG according to the structure annotation type of the computer program mfold 2.3 (Jaeger et al.,1989; Walter et al.,1994; Zucker et al.,1999) (red dots mark GC pairs, blue dots mark AU and GU pairs).

Figure 5.

In situ hybridization to whole mount can (A,C,E) and wild type (B,D,F) testes. After 4 hours of staining reaction, djl mRNA is abundant in wild type testes (B), but the level is severely reduced in mutant testes (A). LacZ mRNA is only visible in can mutants after staining reactions over night, which is shown in comparison to the lacZ mRNA in wild type background at the same staining conditions. −104 to +95 lacZ reporter construct driven mRNA is abundant in the wild type (D), but severely reduced in can mutant testes (C). The −11 to + 95 construct shows a low level of lacZ mRNA in the wild type (F), which is also severely reduced in can mutants (E). Testes of can mutants (G) contain no spermatids but only premeiotic stages. For comparison, a wild type testis (H) is shown: spermatogonia (arrowhead), spermatocytes (arrow), and elongated spermatids (double arrow).

The Level of Transcription of don juan like Is Severely Reduced in Testes of cannonball Mutants

It is suggested that the general transcription machinery is adapted for the expression of specific sets of developmentally regulated target genes by substitution of tissue-specific TAF isoforms (Hochheimer and Tjian,2003). The Drosophila gene can encodes a homolog of the dTAFII80 expressed only in primary spermatocytes (Hiller et al.,2001). Loss of function of Can results in male sterility by preventing the initiation of differentiation beyond the late spermatocyte I stage and, therefore, only premeiotic cell stages are abundant in mutant testes (Fig. 5G; compare to wild type in Fig. 5H). As a consequence, the testes of these males (Fig. 5, top row) are smaller than of wild type males (Fig. 5, bottom row). In testes of can null mutants, basal transcription of Can target genes, like mst87F, janusB, and dj, is significantly reduced (White-Cooper et al.,1998; Hiller et al.,2001). This raises the question whether transcription of djl depends on Can as well.

Performing an in situ hybridization using a Dig-labelled djl probe, we could detect only a low level of transcripts in can mutant testes (Fig. 5A) in contrast to wild type testes (Fig. 5B). We analyzed whether also the lacZ reporter constructs underlie the control by Can. We established fly strains that contain either the −104 to +95 or the −11 to +95 constructs in a can mutant background (see Experimental Procedures section). Homozygous can males were selected that contain the reporter constructs recognizable by the white+ marker. Using a Dig-labelled lacZ probe, we show that the transcripts are severely reduced in the can mutant background (Fig. 5C,E) compared to wild type (Fig. 5D,F).

Translational Repression of don juan like mRNA Is Controlled by a New Translational Repression Element, TREdjl

djl mRNA is translationally repressed in premeiotic and early postmeiotic cell stages. To identify sequences necessary for translational repression, we generated a further reporter gene construct, “djl delta 52,” in which −555 to +43 bp fused to lacZ so that 52 bp directly in front of the AUG are deleted (Fig. 6A). Testes of adult transgenic flies bearing this construct as well as testes of transgenic larvae were assayed for β-Galactosidase activity. A premature translation of the reporter gene in primary spermatocytes, and meiotic and early postmeiotic cells could be observed in testes of adult flies due to staining in the inner curves of the testes (Fig. 6B, arrow). The staining of spermatocytes in testes of third instar larvae (Fig. 6C, arrowheads) confirmed the premature expression because larval testes contain premeiotic and early spermatid stages only. Accordingly, in transgenic flies bearing the reporter gene construct “djl delta 52” translational repression is lost and thus β-Galactosidase is synthesized in premeiotic stages in a wild type background. In can mutants, premeiotic β-Galactosidase of the “djl delta 52” construct is not detectable (data not shown).

Figure 6.

A: Diagram of the djl-promoter-lacZ reporter gene construct (−555 to +43) missing 52 bp in front of the ATG. B,C: Analysis of β-Galactosidase activity in testes of adult (B) and larval (C) males revealed premature expression at the spermatocyte stage (arrow in B, arrowhead in C).

Arrest, Boule, TSR, and Rb97D Are Not Involved in Translational Control of don juan like mRNA

It is known that translational repression of dj mRNA is independent of the RNA binding proteins Boule, TSR, and Rb97D (Santel et al.,1997; Blümer et al.,2002). We aimed to clarify the situation for djl mRNA. Therefore, we analyzed the expression of the “−555 to +95 djl lacZ” reporter gene in testes of males that are mutant for the genes tsr, boule, Rb97D, and arrest. If those proteins participate in the mechanism of translational repression potentially by interacting with sequences necessary for translational repression, we expect a premature translation of the reporter mRNA in these mutants. In tsr, RB97D, arrest, and boule mutant testes expression of the reporter protein β-Galactosidase is under translational repression (Fig. 4C). Thus, the RNA binding proteins TSR, Rb97D, Arrest, and Boule seem to be dispensable for the translational repression of the djl mRNA.

Premature Translation of Don Juan-eGFP in Spermatocytes Leads to Male Sterility Due to Failure Of Sperm Individualization, While Premature Expression of Myc-Don Juan Like Does Not Affect Spermatogenesis

So far, there are no mutants for dj and djl available. To get further insight into the biological role of DJ and DJL, we asked whether premature expression of dj or djl leads to defects in spermatogenesis.

For djl, we generated a construct that directs premature djl expression. We fused five myc-epitopes to the N-terminus of the djl open reading frame. The fusion gene was expressed under the control of the well-characterized promoter and 5′ UTR of the β2 tubulin gene (Fig. 7A), whose mRNA is translated in spermatocytes (Michiels et al., 1989,1993; Santel et al.,2000).

Figure 7.

A: The construct used for premature expression of the fusion protein Myc-DJL, which is expressed under the control of the regulatory region of the β2 tubulin gene (−511 to +156). B: Premeiotically expressed Myc-DJL is detectable by immunofluorescence analysis with anti-Myc-antibody in the cytoplasm of primary spermatocytes (arrows). C: Corresponding phase contrast micrography. D: In early postmeiotic spermatid stages, the fusion protein is localized in the nebenkern (arrow). E: Corresponding nuclear staining with Hoechst 33258; arrowhead marks the nucleus. F: Merged picture. G: In early elongating spermatid bundles, the fusion protein is detectable along the flagella (arrow). H: Corresponding phase.

With anti-Myc antibody immunofluorescence staining the fusion protein Myc-DJL is detectable in the cytoplasm of primary spermatocytes (Fig. 7B, arrow) and shortly after meiosis in the mitochondrial derivatives, the nebenkern (Fig. 7D,F, arrow). In elongating spermatids, Myc-DJL is localized along the flagella (Fig. 7G). The untimely expression of Myc-DJL prior to meiosis and in early postmeiotic stages does not lead to any obvious defects and the transgenic flies are fertile. Previously, we expressed DJ premeiotically with a nanos-Gal4 driver in early germ cells. The total amount of DJ is low in spermatocytes and postmeiotic stages, which might explain why these males are fertile (Santel et al.,1998). Thus, we aimed to express also DJ prematurely under the control of the β2 tubulin promoter. To express dj prematurely, we fused the dj open reading frame to that of Aequorea victoria enhanced green fluorescent protein (eGFP) (Fig. 8A). This allowed us to follow the expression of this fusion gene in the male germ line. The prematurely expressed DJ-eGFP is localized in the cytoplasm of primary spermatocytes (Fig. 8B–D, arrows), in the nebenkern of early spermatids (Fig. 8B–D, arrowheads), and along the flagella of elongating spermatids. As we observed no individualized sperm in these testes, we analyzed whether the individualization complex is assembled at all. In testes of flies expressing DJ-eGFP prematurely, actin assembles initially around each spermatid nucleus in the shape of a cone (Fig. 9A–D). But the actin cones do not progress from the nuclear region to the end of the tail as observed in testes of wild type flies (Fig. 9E,F). Thus, we conclude that flies expressing DJ-eGFP prematurely are sterile due to a failure to complete individualization.

Figure 8.

A: The construct used to express the fusion protein DJ-eGFP ectopically in premeiotic and early postmeiotic stages under the control of the regulatory region of β2 tubulin gene. B: eGFP fluorescence on squashed testes could be observed in the cytoplasm of primary spermatocytes (arrows) and in the nebenkern of early postmeiotic spermatid stages (arrowheads). C: Chromatin staining with Hoechst. D: Merged picture.

Figure 9.

A–D: Squash preparations of dissected testes of flies expressing DJ-eGFP prematurely. A: By the eGFP autofluorescence, we could follow the expression of DJ-eGFP, which is localized during nuclear shaping in the spermatid heads (arrow). B: The testes squash is stained with rhodamin-phalloidin to visualize actin fibers (arrowhead), Hoechst to visualize nuclei (C, arrow), or double exposed for both (D) showing only the assembly of the actin (arrowhead) around each spermatid nucleus (arrow) but no progression. E,F: Squash preparations of dissected testes of flies bearing a construct for the expression of DJ-eGFP under the control of the natural dj promoter. DJ-eGFP localizes along the flagella of elongated spermatids (E, arrowheads). The testes squash is stained with rhodamin-phalloidin (red) to visualize actin fibers (arrows) and Hoechst (blue) to stain the nuclei, respectively (arrowhead). The progression of the actin cones is visible.


dj and djl are localized in tandem on the third chromosome of Drosophila melanogaster. Both dj and djl are highly expressed in male germ cells at meiotic prophase. We did not observe expression in other tissues/developmental stages of Drosophila. This fits to the database (FlyBase), where only testis-specific ESTs are described, one for djl (CG 1984), three for dj. The encoded mRNAs are translationally repressed until the stage of elongated spermatids. DJ and DJL are basic proteins that show 42% sequence identity. Both proteins display a similar expression pattern as they localize transiently in the sperm heads. Both proteins localize to the mitochondrial derivatives of the flagella in spermatids though only DJ remains detectable in the sperm. These facts let us assume that both genes result from a duplication and are possibly completely or partially redundant in function. This view is supported by our failure to identify EMS-induced mutants so far. We performed an EMS mutagenesis and tested 1,950 mutagenized 3rd chromosomes against the deletion Df(3R)Antp7, which deletes dj as well as djl. The corresponding males were tested for fertility. We found one sterile and one semi-sterile fly line but the molecular analysis of dj and djl showed that we obtained no mutants for dj or djl. This may indicate functional redundancy between these two proteins. However, we cannot exclude that these small genes might not have been mutated in our mutagenesis. dj and djl are well conserved as neighbouring genes in D. yakuba. However, dj accumulated several stop codons in the open reading frame. As dj mRNA cannot be translated in D. yakuba, it might be assumed that DJL is sufficient in D. yakuba and that DJ and DJL indeed are functionally redundant in D. melanogaster. Here we discuss the possible biological function of both proteins, the common transcriptional regulation by the tissue-specific TAF Can, and the distinct mode of translational repression.

Don Juan Like Localizes to the Mitochondrial Derivatives Along the Flagella of Spermatids

Because of the subcellular localization of DJ, Myc-DJL, and DJL-eGFP in the sperm heads and the flagella, we suggest that DJ and DJL are involved in the maturation of elongated spermatids during spermiogenesis. DJ, Myc-DJL, and DJL-eGFP transiently localize to the nucleus of elongating spermatids. Thus, it is unlikely that these basic proteins are components of the highly condensed chromatin of the mature sperm. Abundant chromosomal proteins of the mature sperm are Protamine A, Protamine B, and Mst77F (Jayaramaiah Raja and Renkawitz-Pohl,2005).

It was previously shown that the localization of DJ along the flagella is due to its association with the mitochondrial derivatives (Santel et al.,1998). The mitochondrial localization for DJ remains in mature sperm, while Myc-DJL and DJL-eGFP are lost in individualized sperm. The N-terminal mitochondrial localization signals of both proteins are consistent with the protein distribution in vivo. A chance of locating the proteins in the mitochondrial matrix was predicted by a computer-based analysis (PSORT II, Nakai and Kanehisa,1992). The detection of Myc-DJL by immunofluorescence staining with an anti-Myc antibody in the mitochondrial derivatives of the nebenkern stage suggests that the leader sequences for the mitochondrial import need not be, generally, at the very N-terminus of the precursor protein as previously shown by Mukhopadhyay et al. (2003). We show here for prematurely expressed DJ-GFP and also for the misexpressed Myc-DJL protein that they associate with mitochondrial derivatives in the nebenkern stage. Obviously, the N-terminally fused Myc-tag does not hinder the import in those stages. As Myc-DJL was detected along the flagella of elongating spermatids, it can be assumed that the association of Myc-DJL with mitochondrial derivatives continues further on. Since so far no mutants are available, the functional relevance of the mitochondrial localization for sperm morphogenesis remains unclear.

The TAFII80 Cannonball Regulates don juan and don juan like Transcription

The regulatory region necessary for dj and djl was determined by promoter-lacZ constructs. The region −23 to +115 is sufficient for a correct expression of dj (Blümer et al.,2002), and −11 to + 95 are sufficient for the expression of djl. Despite the coordinated expression, we found no sequence conservation in the regulatory region of dj and djl besides an initiator sequence. These short promoters are a characteristic for testis specifically expressed genes (for review see Renkawitz-Pohl et al.,2005). This might indicate a specialized mode of gene regulation. Indeed, the transcription of the genes dj, mst87F, mst84Da-d, mst98Ca/b, and janusB, whose transcripts are translationally regulated, depends on always early (aly), cannonball (can), meiotic arrest (mia), and spermatocyte arrest (sa) (White-Cooper et al.,1998). It is suggested that the transcription initiation complex in spermatocytes differs from the general one (Hiller et al.,2001, 2004). The gene can encodes the testis-specific TAF-homolog, TAFII80. We could show that also the transcript level of djl is severely reduced in can mutants as has been shown previously for dj (Hiller et al.,2001). Possibly, this testis-specific transcription initiation complex does not need further transcriptional regulators so that a short regulatory region like that of dj or djl is sufficient for transcription. Thus, also the transcription of dj and djl depends on a specific transcription initiation complex that is restricted to the male germ line.

The TREdjl Is Specific and Essential to Repress Translation of don juan like mRNA for Several Days Independent of Arrest, Boule, TSR, and Rb97D

At least for dj mRNA, the observed translational repression is essential for spermatogenesis as premature translation leads to male infertility due to failure of individualization. The translational repression elements TREdj and TREdjl are distinct in sequence but similar in their position immediately before the translational initiation codon AUG. Both TREs are embedded in mRNA that is capable of forming a distinct stem and loop structure (Fig. 4D; Blümer et al.,2002). We propose that these secondary structures allow the binding of regulatory proteins and, thus, hinder translation as the translation initiation codon is masked. Searching for RNA-binding proteins that are involved in the translational repression of djl mRNA, we used a genetic approach. We analyzed the expression of the djl promoter-lacZ construct “−555 to +95 djl lacZ” in testes of null mutants for the genes arrest, boule, Rb97D, and tsr and could detect translational repression and activation as it was observed for the endogenous DJL protein. So all these proteins are not essential for the translational control of djl mRNA at least not when only one of them is missing. It was previously shown that neither Rb97D nor the testis-specific protein TSR is involved in the translational control of dj mRNA (Blümer et al.,2002), and that there is no premature DJ expression in boule mutants (Santel et al.,1997). Also the translational repression of mst87F is independent from the expression of TSR and Rb97D (Schäfer et al.,1995). Thus, proteins conferring translational repression until the elongated spermatid stage remain unknown.

In Drosophila Spermatogenesis, mRNAs Are Repressed in Translation by the 5′ UTR, While In Other Systems Translational Repression Depends Mainly on the 3′ UTR

Translational control is an abundant and important mechanism to regulate protein expression. Diverse mechanisms have evolved, which interfere with different steps of translational initiation (for review see Preiss and Hentze,2003). In Drosophila, this is best analyzed for oogenesis. During oogenesis, many mRNAs are stored and translationally repressed until after fertilization in dependence on the 3′ UTR often in connection with mRNA transport in RNP particles (for review see Kuersten and Goodwin,2003) (Huynh et al.,2004).

As in Drosophila, in mammals translational repression is a fundamental mechanism of spermatogenesis; for example, protamine mRNAs are stored as free mRNPs in early haploid cells and translated in late haploid cells. Here translational repression also depends on the 3′ UTR (for review see Kleene,2003), while in Drosophila the protamine mRNAs are regulated by the 5′ UTR (Jayaramaiah Raja and Renkawitz-Pohl,2005). However, it was shown that regulatory factors could also bind to specific regions in the 5′ UTR of testis-specific transcripts in mammals because of the formation of defined secondary structures (Theil,1993; Shi et al.,1997; Schlicker et al.,1997).

So far, control elements necessary for translational repression during Drosophila spermatogenesis have been exclusively found in the 5′ UTR (for review see Renkawitz-Pohl et al.,2005). In that respect, the situation is comparable to the regulation of iron metabolism in mammals. In ferritin mRNA, a stem and loop structure in front of the AUG regulates translation (for review see Hentze,1995; Gebauer and Hentze,2004). Here, a protein was identified that prevents the recruitment of the small ribosomal subunit to the ferritin mRNA (Muckenthaler et al.,1998). The differences between translational repression elements in Drosophila might suggest different mechanisms of repression. In the Mst(3)CGP mRNAs, a defined element close to the cap sites acts in a position-dependent manner, which might indicate a direct interference with the assembly of the 43S initiation complex (Kempe et al.,1993). The translational repression elements of dj and djl mRNA are located in the 5′ UTR in close proximity to the translation initiation codon with a complex stem and loop structure. Here it remains to be clarified whether this structure interferes with establishing the initiation complex or with scanning for the translation initiation codon and recruitment of the large ribosomal subunit.


Drosophila Strains and Culture

Drosophilae were maintained on standard medium at 25°C. Visible markers and chromosome balancers are as described in FlyBase unless otherwise specified. w was used as the wild type strain. boule1 and aret01284 stocks were obtained from the Bloomington Stock Center. The Drosophila stocks Rb97D2 (Karsch-Mizrachi and Haynes,1993; Heatwole and Haynes,1996) and tsr (Haynes et al.,1997) were kindly provided by S. Haynes, and can12 (Hiller et al.,2001) by M. T. Fuller. To analyze the −104 to +95 and the −11 to +95 djl promoter lacZ constructs in a can mutant background, flies bearing these constructs were crossed against w; Sp/CyO; Sb/Ubx to localize the insertion of the P-element. Flies with the constructs (marked with w+) on the second chromosome (balanced over CyO) were crossed against can mutants (balanced over TM3) to bring the lacZ-constructs in the can mutant background. From the offspring of this strain, homozygous can mutants with a lacZ reporter gene were selected.

Cloning of djl-lacZ Promoter Constructs

To analyze the djl regulatory region, promoter-lacZ constructs were generated by PCR strategy using genomic DNA from wild type flies and appropriate primers with linked EcoRI and BamHI restriction sites. The shortest fragment amplified comprised djl from −1 to +95; furthermore, we constructed −555 to +95, −244 to + 95, −104 to +95. To determine the region responsible for translational repression, we synthesized −555 to +43. The PCR products were directly inserted into the transformation vector AUG-β-Gal (Thummel et al.,1988).

Construction of the djl-eGFP and myc-djl Fusion Genes

−555 to +95 from the djl gene were amplified by PCR from genomic DNA using deduced primers with linked EcoRI and BamHI sites and subsequently inserted into the transformation vector AUG-β-Gal (Thummel et al.,1988). The ADH-lacZ gene fragment of the vector was eliminated by cutting with XbaI. For construction of the myc-djl fusion gene, the ADH-lacZ gene was replaced by inserting a 250-bp BamHI/XbaI fragment encompassing five myc epitopes. A PCR-generated XbaI linked PCR product comprising the djl coding region was fused C-terminal in frame to the myc-tag. For construction of the djleGFP fusion gene, a 995-bp fragment encompassing the eGFP open reading frame with linked NcoI and SpeI restriction sites was inserted into the vector pBSKS+ generating pBSKS+-eGFP. −555 to +95 of the djl regulatory region were amplified by PCR using primers with linked EcoRI and NcoI restriction sites. The PCR product was subcloned into the vector pBSKS+-eGFP. The fragment encompassing the djl coding region was obtained by PCR-supported amplification from genomic DNA using deduced primers with linked NcoI sites and fused N-terminal in frame to the eGFP coding region and C-terminal to the djl regulatory region. The fragment encompassing the djl regulatory region and the djl-eGFP fusion gene was obtained by restriction digestion with EcoRI and SpeI and cloned into the germ line transformation vector AUG-β-Gal (Thummel et al.,1988) that was opened using the restriction enzymes EcoRI and XbaI, eliminating the ADH-lacZ gene fragment.

Construction of β2 tubulin-myc-djl and β2 tubulin-dj-eGFP

The β2 tubulin regulatory region comprising the region −511 to +156 was amplified by PCR using primers with linked EcoRI and BamHI restriction sites. The PCR fragment was inserted into the transformation vector AUG-β-Gal (Thummel et al.,1988) to yield β2 tubulin-AUG-β-Gal. The ADH-lacZ gene fragment of the vector was eliminated by cutting with XbaI. Establishing the construct β2 tubulin-myc-djl, a fragment encompassing five myc epitopes with linked BamHI and EcoRI sites, was inserted in the opened vector. The PCR amplified djl open reading frame with linked XbaI sites was fused C-terminal in frame to the myc-tag. For construction of the transgene β2 tubulin-dj-eGFP, a 743-bp fragment with linked BamHI and XbaI sites comprising the eGFP open reading frame was inserted in the opened vector β2 tubulin-AUG-β-Gal replacing the ADH-lacZ gene. The dj open reading frame was amplified by PCR using primers with linked BamHI sites. The obtained fragment was inserted in the BamHI opened vector.

P-Element Mediated Germline Transformation

Transgenic fly strains were established by injection of purified DNA (Elutip-D columns, Schleicher & Schüll) as described previously (Michiels et al., 1989).

Antibodies and Immunofluorescence Staining

Hoechst staining was used to visualize chromatin. The antibodies anti-Myc (monoclonal mouse; Roche, Mannheim, Germany) and anti-GFP (polyclonal rabbit; Abcam, Cambridge, UK) were used in immunofluorescence stainings of squashed testes carried out essentially as described in Hime et al. (1996). The antibodies were diluted 1:100. Cy3-coupled secondary antibodies made in goat (anti-rabbit and anti-mouse) were obtained from Dianova, Hamburg (diluted 1:200). Squashed testes treated with fluorescent antibodies were embedded in Fluoromount-G (Southern Biotech, Birmingham, AL).

Microscopy, Fluorescence Analysis of β2 Tubulin-DJ-eGFP Strains and Actin Staining

Immunofluorescence and GFP samples were examined using a Zeiss Axiophot microscope equipped with fluorescence filters. Images were individually recorded or processed with Adobe Photoshop 6.0. Actin was visualized by counterstaining fixed testes squashes of the β2 tubulin-DJ-GFP strain with TRITC-Phalloidin (Sigma, St. Louis, MO) according to Fabrizio et al. (1998).

In Situ Hybridization and β-Galactosidase Assay

Whole mount in situ hybridization of adult testes was performed with modifications according to Tautz and Pfeifle (1989). As a DNA probe, we used the complete reading frame labelled with DIG. Fixation of the testes was carried out as described by Lantz et al. (1992). The β-Galactosidase enzyme activity was visualized by a histochemical reaction using the chromogenic substrate X-Gal.

RT-PCR of dj l

Total polyA+-mRNA was prepared using Oligotex® mRNA Mini Kit (Qiagen, Chatsworth, CA) from testes of wild type male flies, the whole bodies of wild type males and females, respectively, larvae and embryos. We used the OneStep RT-PCR Kit (Qiagen) to amplify a 668-bp cDNA fragment from the open reading frame of djl. DNA contamination was checked by RT-PCR of the gene β3-tubulin that is expressed in embryos, larvae, adult males, and adult females using primers that spanned an intron. Subsequently, 20 ng poly (A+) mRNA amplification (32 cycles) was carried out.

To narrow down the transcription start side of djl, we performed RT-PCR using polyA+-mRNA (see above) with different 5′ oligonucleotides starting 199, 106, 96, and 74 bp before the AUG. The antisense primers were chosen 3′ of the djl intron so that the primer pairs amplify 914-, 821-, 809-, and 791-bp fragments. As control, every primer pair was also checked with DNA. In all cases, the same amount of RNA and DNA was used. And also in all cases, 32 rounds of PCR were performed. PolyA+ mRNA was isolated from 50 testes as described above; for each lane, polyA+ mRNA from an equivalent of three testes was used.


We thank Ruth Hyland and Dominik Helmecke for help with the establishment of transgenic fly lines and Heike Sauer for excellent secretarial assistance. We greatly acknowledge the critical reading of the manuscript by Mireille Schäfer and Roxane Schröter. This work was supported by grants from the Deutsche Forschungsgemeinschaft (Re 628/12-2, Forschergruppe “Chromatin mediated biological decisions”, GRK 767) to R.R.-P.