Dicer Is Required for Haploid Male Germ Cell Differentiation in Mice

Background The RNase III endonuclease Dicer is an important regulator of gene expression that processes microRNAs (miRNAs) and small interfering RNAs (siRNAs). The best-characterized function of miRNAs is gene repression at the post-transcriptional level through the pairing with mRNAs of protein-encoding genes. Small RNAs can also act at the transcriptional level by controlling the epigenetic status of chromatin. Dicer and other mediators of small RNA pathways are present in mouse male germ cells, and several miRNAs and endogenous siRNAs are expressed in the testis, suggesting that Dicer-dependent small RNAs are involved in the control of the precisely timed and highly organised process of spermatogenesis. Principal Findings Being interested in the Dicer-mediated functions during spermatogenesis, we have analysed here a male germ cell-specific Dicer1 knockout mouse model, in which the deletion of Dicer1 takes place during early postnatal development in spermatogonia. We found that Dicer1 knockout testes were reduced in size and spermatogenesis within the seminiferous tubules was disrupted. Dicer1 knockout epididymides contained very low number of mature sperm with pronounced morphological abnormalities. Spermatogonial differentiation appeared unaffected. However, the number of haploid cells was decreased in knockout testes, and an increased number of apoptotic spermatocytes was observed. The most prominent defects were found during late haploid differentiation, and Dicer was demonstrated to be critical for the normal organization of chromatin and nuclear shaping of elongating spermatids. Conclusions/Significance We demonstrate that Dicer and Dicer-dependent small RNAs are imperative regulators of haploid spermatid differentiation and essential for male fertility.


Introduction
Spermatogenesis is under strict gene control that governs the precisely timed events leading to the production of mature spermatozoa capable of fertilization [1,2]. It includes proliferation, differentiation and morphogenesis of male germ cells [3]. The process begins when diploid spermatogonia multiply by consecutive mitotic divisions and then enter the meiotic program, which involves chromosome duplication, homologous chromosome pairing, synaptonemal complex formation, meiotic recombination and meiotic divisions resulting in the formation of haploid round spermatids. Haploid germ cells then undergo a dramatic differentiation phase, spermiogenesis, which includes acrosome and flagellum formation, nuclear reshaping and massive chromatin reorganization during which histones are replaced by testisspecific proteins called protamines [4]. The histone-protamine transition sets limitations to the male germ cell-specific gene expression since protamine-bound genes are largely silenced. Therefore, post-transcriptional mRNA control is active in late spermatogenic cells to ensure the correct timing of protein expression and to provide mRNAs in transcriptionally inactive elongating spermatids.
Small non-coding RNAs are crucial gene regulators that can target gene expression both post-transcriptionally by mRNA silencing and transcriptionally by mediating changes in chromatin organization [5]. Small RNAs are also important regulators of male fertility, and distinct classes with different mechanisms of biogenesis and function have been found in the male germ line [6]. One of these classes consists of microRNAs (miRNAs), several of which are expressed in spermatogenic cells, implying that they have an important role in gene regulation during spermatogenesis [7,8]. miRNAs mostly act by destabilizing target mRNAs or inhibiting their translation [9]. Each of them may target hundreds of distinct mRNAs and thus expression of most of protein-coding genes is controlled by these small regulatory RNAs [10]. PIWIinteracting RNAs (piRNAs) are predominantly expressed in the germ cell lineage. They are synthesized in large quantities and their functions include silencing of transposon expression [11]. Processing mechanisms for piRNAs have not yet been identified but their synthesis does not involve the RNase III endonuclease Dicer that is critical for production of miRNAs and small interfering RNAs (siRNAs) [6,12]. miRNA processing from hairpin-loop-folded primary precursors requires two RNase IIIlike enzymes, Drosha and Dicer [13]. In contrast, double-stranded siRNA precursors can be processed by Dicer alone, which emphasizes the diversity of small RNA pathways. siRNA precursors are usually introduced in cells exogenously, for example by viruses, but as originally described in plants and nematodes, endogenous siRNAs (endo-siRNAs) can also be produced and can have important functions in gene silencing [6]. There is increasing evidence that endo-siRNAs can be used as a gene control mechanism also in mammals [6,[14][15][16].
Dicer is vital for mouse embryogenesis since its deletion results in an early embryonic lethal phenotype [17]. The importance of Dicer in several differentiation programs has been demonstrated, including mouse female and male germ cell maturation [18][19][20][21]. In the mouse testis, Sertoli cell-specific deletion of Dicer1 revealed its crucial importance for the normal function of these somatic nursing cells in supporting male germ cell differentiation [22,23]. The role of intrinsic miRNA pathways in male germ cells has been studied using a mouse model with a specific deletion of Dicer1 in primordial germ cells (PGCs) induced by a TNAP-Cre transgene [20,21]. These studies demonstrated the defects in PGC proliferation and spermatogenesis, thus suggesting the potential importance of Dicer-dependent pathways on postnatal male germ cell differentiation [21]. However, these studies could not address the exact role of Dicer in adult spermatogenesis since Dicer1 was already depleted in PGCs at embryonic day 10 [24], and thus the development of embryonic germ cells was interfered. Phenotypic analysis of this mouse line was also problematic due to the low penetrance of TNAP-Cre transgene. Therefore, different fully penetrant mouse lines in which Dicer1 deletion occurs in postnatal male germ cells are required to assess Dicer-mediated functions in adult spermatogenesis. We generated a knockout mouse model, in which Dicer1 was deleted specifically in spermatogonia to demonstrate the role of Dicer and Dicer-dependent small RNAs in the regulation of postnatal male germ cell development. We found that Dicer1 deletion caused a failure in haploid differentiation resulting in abnormal spermatozoa and eventual male infertility.

Expression of Dicer1 during spermatogenesis
Dicer1 is known to be expressed in the mouse testis [25,26], but its exact expression pattern is unclear. Due to the importance of Dicer-dependent pathways in various cellular processes, Dicer is considered to be constitutively expressed in all cell types. However, gene expression patterns during spermatogenesis are usually strictly controlled both spatially and temporally [1], and we wanted to examine whether Dicer1 expression is differentially regulated during the progress of spermatogenesis. We performed quantitative RT-PCR analysis of juvenile mouse testes to show the time course of Dicer1 mRNA appearance (Fig. 1A). Total testis RNA was extracted from wild type mice sacrificed at different time points during the synchronized first wave of spermatogenesis.
These samples thus represent different phases of germ cell differentiation within the seminiferous epithelium. Testis collected at 8 days post partum (dpp) contains spermatogenic cells just prior to or at the onset of meiosis; at 14 dpp pachytene spermatocytes appear, at 20 dpp round spermatids have already been generated, and at 28 dpp the elongation of spermatids has started. Dicer1 expression was highest in testes collected at 14 dpp, and the amount of Dicer1 mRNA decreased in the testes of older animals (Fig. 1A). This indicates that Dicer1 is expressed at higher levels in early germ cell types such as spermatogonia and early spermatocytes than in late spermatocytes and spermatids as the appearance of these cell types in the testis dilutes the Dicer1 mRNA signal. Dicer1 mRNA could be detected in isolated pachytene spermatocytes and round spermatids, but at relatively low levels, further supporting the results obtained from the analysis of juvenile testes (Fig. 1A).
Germ cell-specific deletion of Dicer1 in the mouse testis To generate a mouse line with a specific deletion of Dicer1 in postnatal male germ cells, we crossed a mouse line carrying a floxed Dicer1 allele (Dcr(fx/fx)) with a transgenic mouse line expressing the Cre recombinase under the control of a Neurogenin3 (Ngn3) promoter (Ngn3Cre) [27]. In this study, we used a floxed Dicer1 allele that was generated by inserting two loxP sites around exon 24 that encodes most of the second RNase III domain (Fig. 1B,C) [28]. Cre-dependent recombination generates a mutated allele that encodes a truncated Dicer protein still recognized by an antibody made against a region at the Cterminus of the protein [28].
Ngn3 is endogenously expressed in male germ cells starting from day 5 dpp, and it has been shown to be expressed in type A spermatogonia that give rise to all differentiating germ cells [29]. Thus Ngn3-promoter-driven Cre is produced at early stage of postnatal development and induces Dicer1 deletion in spermatogonia. To verify the Cre-mediated recombination in spermatogenic cells, the Ngn3Cre line was crossed with a transgenic mouse line expressing a Cre-recombination sensitive ROSA26YFP reporter gene (Fig. 1D) [30]. Analysis of the testes from Ngn3Cre; ROSA26YFP mice demonstrated the Cre-dependent YFP expression in differentiating male germ cells throughout the meiotic and postmeiotic stages (Fig. 1D). Dcr(fx/fx); Ngn3Cre mice with germ cell-specific knockout of both Dicer1 alleles (hereafter referred to as Dicer1 knockout mice) were generated by crossing Dcr(fx/wt); Ngn3Cre and Dcr(fx/fx) animals. The Cre recombination activity on Dicer1 was demonstrated through genomic PCR using primers specifically amplifying the deleted Dicer1 allele (Fig. 1E). A deletion-specific 600 bp band was produced in the testicular DNA from 10 dpp knockout but not wild type control mouse, demonstrating that the recombination had taken place. Importantly, amplification of Dicer1 mRNA using one primer hybridizing within the deleted area confirmed a dramatic decrease of Dicer1 mRNA in adult knockout testes (Fig. 1F).

Dicer1 deletion in postnatal male germ cells results in defective spermatogenesis and infertility
Germ cell-specific Dicer1 knockout mice grew normally and did not display any visible physiological or anatomical gross abnormalities in adulthood. However, when mated with the wild type C57BL/6J females, knockout males were unable to sire any pups (data not shown). The testis size of knockout animals was reduced about 50% compared to control littermates ( Fig. 2A). Periodic acid-Schiff (PAS) and hematoxylin-eosin (HE) staining of knockout testis sections demonstrated a severe disruption of spermatogenesis (Fig. 2B, Fig. S1 and Fig. S2). Spermatogonia and early meiotic cells appeared normal without any prominent defects, and the general organization of Sertoli and germ cells in the seminiferous epithelium was unaffected (Fig. S3). However, compared to control littermates, less haploid cells were produced (Fig. 2C). Because of the decreased number of haploid cells in knockout testes, we decided to measure apoptotic activity in the seminiferous epithelium of knockout mice compared to that of control littermates by TUNEL assay. An increased number of apoptotic cells was detected in knockout testis (Fig. 2D). Apoptotic cells were identified as mostly spermatocytes based on their location and appearance in the seminiferous epithelium. Apoptotic cells within knockout tubules were normally found at stages XII-I, when spermatocytes that have failed meiosis are eliminated, and at stages IV-V, after the mid-pachytene check-point. However, the number of apoptotic cells in these stages was clearly increased.
The most obvious spermatogenic failure took place during haploid differentiation, as demonstrated by the abnormal organization, amount and morphology of elongating spermatids (Fig. 2B). The seminiferous epithelium is organized in a cyclic manner in twelve distinct stages, with each stage containing a specific association of different germ cell types at distinct phases of differentiation [3,31]. Microscopic analysis of knockout testes often revealed disorganized stages of seminiferous epithelial cycle with haploid cells from different developmental stages being mixed in the same cross sections (Fig. 2B, round spermatid acrosomes are pointed by arrows). Living cell phase contrast microscopy of stagespecific pieces of seminiferous tubules [31] supported these findings (data not shown). In addition, defects in acrosomes, such as fragmented acrosomes or abnormal acrosomal vacuoles were frequently detected (Fig. S4). Histological and sperm count analysis of the cauda epididymides confirmed the spermatogenic problems by demonstrating a drastic reduction in the number of mature spermatozoa. In addition, numerous exfoliated immature germ cells were frequently detected in the epididymal lumen ( Fig. 2E).

Meiosis in Dicer1 knockout spermatocytes
Light microscopy of knockout seminiferous tubules revealed no gross abnormalities in meiotic spermatocytes. We assessed the formation of synaptonemal complexes in knockout pachytene spermatocytes by staining with an antibody against SCP3 (Synaptonemal complex protein 3), one of the major components of the lateral elements of synaptonemal complexes. Synaptonemal complexes were normally detected, and double staining with an antibody against phosphorylated histone H2AX that is used as a marker for silenced X and Y chromosomes in the sex body, revealed the presence of sex bodies in knockout pachytene spermatocytes (Fig. 3A). Synaptonemal complexes were also detectable through electron microscopy of knockout testes (Fig.  S4). Immunostaining of phosphorylated Serine 10 of histone H3, a marker of mitotic and meiotic chromatin condensation, revealed the presence of metaphase plates in the knockout tubules (Fig. 3B).

Elongation of spermatids is disrupted in Dicer1 knockout mice
The abnormal spermiogenesis detected in knockout testis sections (Fig. 2B) prompted us to analyze the progress of haploid cell differentiation in greater detail. Phase contrast microscopy of drying down preparations demonstrated that spermatid elongation is severely affected in knockout mouse testes (Fig. 4A). Only very few (less than ten) normal appearing step 15-16 spermatids were detected among all spermatogenic cells analyzed. Nearly all late spermatids showed abnormal head shape and chromatin condensation and a disrupted organization of tail accessory structures C) The floxed allele was genotyped using primers DcrF and DrcR, yielding a 420 bp band for the floxed (fx) and a 351 bp band for the wild type (wt) allele. D) To assess the Cre-mediated recombination efficiency, Ngn3Cre mice were crossed with ROSA26YFP reporter mice. Anti-GFP immunostaining revealed that Ngn3Cre;RO-SA26YFP testes sections show strong GFP expression in germ cells, whereas control ROSA26YFP littermates do not. Note that Leydig cells are GFP positive due to autofluorescence. E) Genomic PCR on the whole testis DNA with DcrF and DcrNull primers confirmed the deletion of Dicer1 at postnatal day 10. F) Dicer1 mRNA is dramatically decreased in knockout testes. RT-PCR analysis of the testes of 9 weeks old mice using primers for Dicer1 mRNA. The reverse primer is located in the deleted exon and no PCR product is produced from cells with deleted Dicer1 alleles. KO: Dcr(fx/fx);Ngn3Cre, WT: Dcr(fx/fx). doi:10.1371/journal.pone.0024821.g001 (Fig. 4A). The chromatin condensation state of most of the elongating spermatids resembled that of step 9-10 spermatids, but some heads with condensed chromatin were also observed. Because of the observed problems in chromatin condensation, we went on to further study the status of histone acetylation. Acetylated forms of H2A, H2B, H3 and H4 appear in step 9-11 elongating spermatids, and disappear later in condensing spermatids, thus preceding the histone-protamine transition [32]. In wild type testes, H3 became hyperacetylated in step 9 spermatids, and the acetylation signal disappeared in step 11 spermatids (Fig. 4B). Hyperacetylation was detected in knockout elongating spermatids, but it was not stage-specific and nearly all the cross-sections contained elongated spermatids with hyperacetylated H3 (Fig. 4B). In addition, elongating spermatids that were positive for protamine PRM1 were also greatly reduced in knockout testes compared to control littermates (Fig. 4C). Overall, these findings suggest an arrest in elongation prior to the histone-protamine exchange. Due to the abnormal head shape of elongating spermatids, we decided to study additional proteins that are known to be critical for chromatin architecture. H1T2/H1FNT, a testis-specific histone H1 variant, has been shown to have a polarized localization within the round and elongating spermatid nucleus, being concentrated in a cap-like structure at the inner periphery of the nuclear membrane at the apical pole. H1T2 was demonstrated to confer to the spermatid nucleus the polarity that is essential for proper DNA condensation and elongation of spermatids [33]. Interestingly, the polarized localization of H1T2 was disrupted in knockout spermatids, and H1T2 showed a bipolar localization pattern at both the apical and basal side of the nucleus (Fig. 5A). Manchette is a microtubule structure in elongating spermatids that is suggested to be involved in the shaping of sperm head. Anti-tubulin immunostaining revealed a normal manchette located in an organized fashion at the basal side of control nuclei, while this was not the case in knockout elongating spermatids, in which anti-tubulin staining was distributed more randomly in the cytoplasm of spermatids, often surrounding the whole nucleus (Fig. 5B). Furthermore, electron microscopy studies of late elongating spermatids revealed disrupted microtubule organization as well as problems in nuclear shaping (Fig. 5C).

Abnormal morphology of Dicer1 knockout spermatozoa
In order to assess the morphology of Dicer1 knockout spermatozoa, we isolated sperm from cauda epididymides of wild type and knockout mice and stained them with hematoxylin. Knockout spermatozoa were clearly abnormal with defective head structures including small heads and abnormal head shapes, as well as tail abnormalities including thin tails and disorganized accessory structures (Fig. 6A). In fact, only very few (less than ten) normal-looking spermatozoa were detected among all cells that were analyzed. Immunostaining with a fibrous sheath marker (anti-AKAP4) demonstrated that the fibrous sheath is present in some but not all of the spermatozoa (Fig. 6B). The same kind of observation was made for the mitochondrial sheath as revealed by Mitotracker staining (Fig. 6C). Furthermore, several pin-head spermatozoa with tubulin-positive thin tails but no DAPI-positive nuclei were detected (Fig. 6C). Electron microscopic examination of epididymal sperm revealed small abnormal heads that were frequently bent over the tail, and an excess of cytoplasm (Fig. 6D).
To confirm the penetrance of Cre expression and the presence of Dicer1 deletion in epididymal sperm of knockout mice, spermatozoa from the cauda epididymides were isolated and sperm DNA was examined for Cre recombination. We found that the epididymal sperm from knockout mice carried a deleted Dicer1 allele (Fig. 6E).

Expression of transposable elements and centromeric repeat transcripts in Dicer1 knockout testes
Expression of transposable elements is induced in mouse oocytes lacking Dicer [18]. In addition, human Dicer mediates the degradation of Alu/SINE transposable elements in retinal pigmented epithelium [34]. We thus decided to examine the expression of some transposable elements in the testis of knockout mice and their wild type littermates to assess weather Dicer have similar functions in male germ cells. We did not find any differences in LINE1 (Long Interspersed Nuclear Element 1), SINEB1 (Short Interspersed Nuclear Element 1), or IAP (Intracisternal A-Particle) expression in 28 days old or adult knockout testes as compared to control littermates (Fig. 7A).
Since the absence of Dicer in mouse embryonic germ cells has been shown to induce aberrant expression of centromeric transcripts [35,36], we decided to study the status of repetitive element silencing in knockout testes. We found that the transcription of centromeric major repeats was induced in our mouse model, indicating that Dicer is required for the correct silencing of these repeat sequences (Fig. 7B). Furthermore, we wished to find out if the absence of Dicer has any effects on heterochromatin organization during spermatogenesis. The localization pattern of the centromeric protein CENP-A in the nucleus of spermatogenic cells was not drastically changed in knockout testes, suggesting that there is not general widespread disorganization of the centromeric heterochromatin areas (Fig. S5). Similar results were obtained by immunostaining using antibodies against other heterochromatin markers, such as heterochromatin protein HP-1 (data not shown), and dimethylated H3K9 (Fig. S5).

Discussion
We have demonstrated here that Dicer is imperative for normal differentiation of haploid male germ cells in the mouse testis. To our knowledge, this is the first study of a postnatal deletion of Dicer1, which has enabled specific dissection of the affected spermatogenic steps without interfering with the embryonic functions of PGCs. Endogenous Ngn3 is expressed in type A spermatogonia, which give rise to the entire panel of differentiating male germ cells [29]. Ngn3-promoter-driven Cre expression is fully penetrant and appears to reflect the expression of endogenous Ngn3 in spermatogonia. We should note here that   Immunofluorescence of testis sections with an antibody against a testis-specific histone 1 variant, H1T2. H1T2 staining in knockout step 9 spermatids (lower panel) is no longer correctly polarized beneath the developing acrosome, and is in fact localized at two opposite edges of the nucleus. White arrows in the knockout merge picture point to the bipolarized H1T2 staining of two selected step 9 spermatids. Scale bar: 20 mm. B) Fluorescence confocal microscopy of testis sections immunostained with anti-tubulin revealed the disorganized manchette structure in knockout Ngn3 is known to be an important regulator of pancreatic development and is expressed in pancreatic islet precursor cells [27]. In addition, Ngn3-promoter-driven transgenes are also expressed in brain [37]. Here, we did not analyze the pancreas and brain phenotypes of Dcr(fx/fx); Ngn3Cre mice, but no gross abnormalities in the development or behavior of the mice were observed. miRNAs are known to remain in the cells for some time after the deletion of Dicer1, and therefore we did not include spermatogonial proliferation phase in our analysis but focused on the later steps of spermatogenesis. Our results are supported by the parallel analysis of a conditional mouse model using Ddx4promoter-driven expression of Cre to induce Dicer1 deletion in male germ cells (Romero et al., co-submitted manuscript). In the Dcr(fx/fx); Ddx4Cre mouse line, the deletion of Dicer1 takes place earlier than in our model and, accordingly, shows a much stronger abnormal meiotic phenotype. We believe that the reason for this difference is the later expression of the Ngn3Cre transgene, which might enable meiotic cells to escape the effects of Dicer1 deletion. However, the decreased number of haploid cells and the increased number of apoptotic spermatocytes in the seminiferous epithelium of Dcr(fx/fx); Ngn3Cre testis supports a role of Dicer-dependent pathways in meiosis. Importantly, the haploid phenotype of both mouse lines is very similar, highlighting the central role of Dicerdependent pathways in postmeiotic differentiation.
The most prominent spermatogenic defects in our Dicer1 knockout mice were found during the elongation phase of haploid differentiation. Elongation appeared to be hindered before nuclear condensation had started. This was demonstrated by the high number of uncondensed elongating spermatids as detected by phase contrast microscopy of spermatogenic cell spreads (Fig. 4A), the presence of hyperacetylated H3-positive elongating spermatids in all the stages of seminiferous epithelial cycle (Fig. 4B) and correspondingly the low number of protamine-positive nuclei (Fig. 4C) in knockout testes. Some of the germ cells were able to undergo condensation as evidenced by the presence of condensed nuclei among both testicular elongating spermatids and mature sperm (Figs. 4,5,6). However, the head morphology of these cells never appeared normal and nuclei were small and misshapen; defects in the tail development were also frequently observed. Interestingly, we demonstrate that the polarized localization of histone 1 variant H1T2 was disrupted in Dicer1 knockout mice, with H1T2 localizing on both the apical and basal sides of the nucleus. H1T2 is considered to be a molecular marker that can reveal chromatin disorganization in round and elongating spermatids [38]. H1t2 knockout mice show delayed nuclear condensation and aberrant elongation of spermatids [33]. A bipolar H1T2 mislocalization similar to the one observed in Dicer1 knockout spermatids, is also detected in testes devoid of Trf2, a transcriptional regulator involved in nuclear chromatin organization [38,39]. Therefore, our results suggest that Dicer-dependent pathways are involved in the control of chromatin architecture in haploid male germ cells.
The functions mediated by small RNAs are diverse. Small RNAs in animal cells can act at the post-transcriptional level by silencing or sequestering protein-coding mRNAs or transposable element transcripts, at the chromatin level by affecting gene transcription or at the genomic level through the regulation of chromosome segregation [6]. The predominant function of mammalian miRNAs has been demonstrated to be the destabi-lization of target mRNAs [9]. While this manuscript was under revision, Dicer-dependent but Drosha-independent endo-siRNAs were reported in male germ line that can function in posttranscriptional control of a wide variety of protein encoding mRNAs [16]. Therefore, it is likely that at least some of the defects in Dicer1 knockout testes are due to the absence of miRNA and/or endo-siRNA-mediated mRNA regulation. Post-transcriptional control is of central importance during haploid differentiation due to the transcriptional silencing that results from the tight packing of chromatin with protamines [2,4]. Since the most striking defects in the differentiation of Dicer-null male germ cells coincide chronologically with chromatin condensation and transcriptional silencing, it can be envisaged that Dicer is involved in this post-transcriptional control of haploid mRNAs. High throughput transcriptome analysis as well as detailed quantitative proteomics will uncover the genes that are under the regulation of Dicer-generated small RNAs during male germ cell differentiation.
piRNAs have a well-characterized role in transposon silencing in male germ cells [40], and several studies have also linked Dicerdependent pathways with the regulation of transposon expression [14,15,18,34]. The most extensive transposon derepression in male germ cells takes place during epigenetic reprogramming in PGCs and in very early postnatal cells (from embryonic day 15 to 3 dpp), which enables the establishment of novel sex-specific epigenetic marks in the genome [40]. The level of transposable element transcripts seemed unchanged in our Dicer1 knockout model (Fig. 7). In contrast, transposon expression was increased in the spermatocytes of Dcr(fx/fx); Ddx4Cre mice, which shows efficient Dicer1 recombination already at birth (Romero et al., co-submitted manuscript). This difference may be explained by the later Cre expression in Dcr(fx/fx); Ngn3Cre testes only after major epigenetic reprogramming events have been largely completed. Interestingly, we observed an increase in centromeric major repeat transcript expression in Dicer1 knockout testes (Fig. 7). Similar defects in centromeric silencing have been reported in mouse embryonic stem cells lacking functional Dicer [35,36]. Most of the known Dicer-dependent functions in mammals involve posttranscriptional regulation, but evidence of a small RNA-mediated transcriptional gene silencing and regulation of heterochromatin formation and maintenance is emerging, even though the mechanistic aspects still remain unclear [41]. Dicer has been localized in the nucleus and specifically on certain chromosomal domains suggesting that Dicer has a role in transcriptional regulation at the chromatin level [42,43]. However, further studies will be required to reveal the possible mechanistic connection of Dicer and Dicer-dependent small RNAs with heterochromatin formation, regulation of repeat-derived transcripts and control of chromatin organization in differentiating male germ cells.

Ethics statement
Animal husbandry and all animal experimentation were carried out in compliance with Finnish laws. All efforts were made to minimize the suffering of animals. Protocols for the use of animals were approved by the Committee on the Ethics of Animal Experimentation at the University of Turku in accordance with the Guide for Care and Use of Laboratory Animals (National Academy of Science) (license number: 2009-1206-Kotaja). step 9-12 elongating spermatids. Scale bar: 50 mm. C) Electron microscopy of knockout condensed elongating spermatids revealed defects in head shaping, disorganized manchettes (arrows) and abnormal bending of the head (arrowhead). Scale bar: 2 mm. doi:10.1371/journal.pone.0024821.g005 Figure 6. Abnormal morphology of knockout sperm. A) Very few spermatozoa could be isolated from knockout cauda epididymides, and all retrieved cells had abnormal morphology (arrows), as revealed by hematoxylin staining of the sperm slides. B) Immunofluorescence of knockout and wild type epididymal sperm with an anti-AKAP4 antibody (red) was performed to detect the fibrous sheath of the sperm flagellum. Many tails were lacking the detectable fibrous sheaths. An arrow points to a flagellum with anti-AKAP4-positive fibrous sheath, whereas an arrowhead shows a sperm tail without anti-AKAP4 staining. C) Immunostaining with an anti-tubulin antibody (red) combined with the Mitotracker staining (green) revealed the absence of mitochondrial sheath in some spermatozoa. An arrow points to the mitochondrial staining in the midpiece, and arrowheads indicate the absence or mislocalized mitochondrial sheath. Pin-head sperm with tubulin-positive axonemal structures but no DAPI-positive heads were also frequently observed (yellow arrow). Scale bar in (A), (B) and (C): 25 mm. D) Electron microscopy of epididymal sperm revealed small abnormally shaped nuclei (arrow), bending of tails from the neck region (arrowhead) and excess of cytoplasm (asterisk). Scale bar: 0.5 mm. E) Genomic PCR on the whole testis DNA with DcrF and DcrNull primers confirmed the deletion of Dicer1 in the epididymal sperm of knockout mice. doi:10.1371/journal.pone.0024821.g006

Animals
Mice were housed at the Animal Facility of the University of Turku, Finland, under controlled environmental conditions. The genetic background of all the mice used in this study was mixed C57Bl/6J and SV129. Floxed Dicer1 mice were kindly provided by B. Harfe and were genotyped as described [28]. Genotyping primers for Ngn3Cre mice [27] were: F(Ngn3Cre) 59-CCT-GTTTTGCACGTTCACCG-39, R(Ngn3Cre) 59-ATGCTTCT-GTCCGTTTGCCG-39, F(pTimer) 59-ACGGCTGCTTCATC-TACAAGG-39, R(pTimer) 59-TTGGTGTCCACGTAGTAGT-AG-39. To achieve selective inactivation of Dicer1 in germ cells, transgenic Ngn3Cre female mice were mated with male mice carrying two floxed Dicer1 (Dcr) alleles in order to generate Dcr(fx/ wt); Ngn3Cre and Dcr(fx/wt) mice. These animals were then intercrossed to produce Dcr(fx/fx); Ngn3Cre as well Dcr(fx/fx) and Dcr(fx/wt); Ngn3Cre control littermates. Recombination was confirmed with deletion-specific genomic PCR using primers described earlier [28]. The ROSA26-YFP reporter construct consist of a stop cassette upstream of the YFP coding sequence that is inserted in the ROSA26 gene locus; upon Cre expression the stop cassette is removed and expression of the reporter is induced [30].

RNA extraction and qRT-PCR
Total testis RNAs were extracted using TRI Reagent solution (Molecular Research Center, Inc.) and treated with TURBO DNase (Ambion). cDNA synthesis and qPCR quantification was performed by using DyNAmo TM cDNA synthesis kit (Finnzymes) and DyNAmo TM Flash SYBRH Green qPCR Kit (Finnzymes). Primers for Dicer1 were: F(Dicer) 59-CTTGACTGACTTGC-GCTCTG-39 and R(Dicer) 59-AATGGCACCAGCAAGAGAC-T-39. Primers for SINE B1 and LINE L1 RNA are described in [44] and IAP primers are described in [45]. Primers for centromeric major and minor repeat transcripts are described in [46]. Normalization was performed with L19 mRNA levels for repetitive elements and mRNA transcripts. Primers for L19 were: F(L19) 59-GGACAGAGTCTTGATGATCTC-39 and R(L19) 59-CTGAAGGTCAAAGGGAATG-TG-39. Each assay was performed in three independent technical and biological replicates.

Histology
For histological analyses, tissues were collected and directly fixed in 4% paraformaldehyde (PFA) or in Bouin's fixative (4-20 hours at room temperature). Tissues were then dehydrated in a series of ethanol washes and embedded in paraffin. Paraffinembedded tissues were cut and stained with hematoxylin and eosin (HE) or periodic acid-Schiff (PAS) according to standard protocols.

Squash preparations and drying down preparations
Testes were dissected and decapsulated, and the specific stages of the seminiferous epithelial cycle of corresponding tubule segments were identified on the basis of their transillumination patterns. Squash preparations were prepared according to the protocol in [31], and live cell microscopy was performed by using phase contrast optics (Leica DMRBE microscope). For drying down preparations, stage-specific segments of seminiferous tubules were isolated, and cells were released and fixed on slides as previously described [31]. Preparations were visualized by phase contrast microscopy or used for immunofluorescence.

Immunohistochemistry and immunofluorescence
Paraffin-embedded testis sections were rehydrated, antigens were retrieved by pressure cooking in 10 mM Sodium Citrate (pH 6.5) for 2 hours, and non-specific binding sites were blocked in 3% BSA, 10% normal goat serum. Endogenous peroxidase activity was Figure 7. Expression of transposable elements and centromeric repeat transcripts in Dicer1 knockout testes. A) Quantitative RT-PCR analysis of 28 days old Dcr(fx/fx) (WT) and Dcr(fx/fx); Ngn3Cre (KO) total testis RNA using LINE1, SINEB1 and IAP primers. No differences in transposon expression between control and knockout samples were revealed. B) Expression of major centromeric repeat sequences was induced in 28 days old knockout testes. A schematic diagram shows the organization of major repeats and locus-specific primers used in the study [46]. Only the 308 bp band was detected in our optimized conditions. Three experimental triplicates (1-3) are shown for both wild type and knockout reactions. As a control, reactions were performed with mock-transcribed cDNAs (-RT). doi:10.1371/journal.pone.0024821.g007 blocked in 3% H 2 O 2 . Antibody incubations were done in blocking solution for 1 hour at 37uC, or O/N at 4uC. Rabbit and mouse IgG were used as negative controls (Vector Laboratories). For histochemical detection, tissue sections were incubated with a biotinylated secondary antibody (1:750, Vector Laboratories) diluted in blocking solution for 1 hour at 37uC. Antibody localization was detected with the regular VECTASTAIN ABC (peroxidase) system (Vector Laboratories) and 3,39-diaminobenzidine (liquid DAB+, DAKO), stained with hematoxylin, dehydrated and mounted in PERTEX medium. Alternatively, AlexaFuor488 and AlexaFluor594 conjugated secondary antibodies (1:500, Invitrogen) diluted in blocking solution were used for immunofluorescence detection. Nuclei were stained with 49,6-diamidino-2phenylindole (DAPI) (1:10 000, Sigma) for 5 minutes and sections were mounted in Mowiol 4-88 medium (Polysciences, Inc.). Drying down and squash preparations were post-fixed in 4% paraformaldehyde, permeabilized in 0.2% Triton-X 100 for 15 minutes at room temperature, and immunofluorescence was performed as described above. For the detection of mitochondria, slides were incubated with 200 nM Mitotracker (Invitrogen) in PBS for 15 minutes. Zeiss AxioImager M1 microscope was used for the normal immunofluoresence analyses. Zeiss LSM510 META microscope was used for the confocal images.

TUNEL assay
TdT-mediated dUTP nick end labelling (TUNEL) was used on Bouin's fixed, paraffin embedded testis sections to detect DNA fragmentation in apoptotic cells. Paraffin was removed and sections were rehydrated and permeabilized in 10 mM Sodium Citrate (pH 6.5) for 20 minutes in a microwave. Endogenous peroxidase activity was blocked in 3% H 2 O 2 (Sigma) at room temperature for 15 minutes. Sections were incubated in 16 TdT buffer, Terminal transferase (0.6 U, Roche), biotin-16-dUTP (10 mM, Roche) and 10 mM CoCl 2 for an hour at 37uC. A negative control was incubated without an enzyme. Non-specific sites were blocked with 3% BSA and 10% Normal Goat Serum for 15 minutes at room temperatures and subsequently incubated with ExtrAvidinH-Peroxidase (1:50, Sigma) for 30 minutes at 37uC. Apoptotic cells were detected with 3,39-diaminobenzidine (liquid DAB+, DAKO), and cells were stained with hematoxylin, dehydrated and mounted with PERTEX medium.

Sperm counts and sperm slides
Epididymal sperm count was performed with sperm extracted from the cauda epididymis and ductus deferens of adult (60 dpp or 120 dpp) male mice and analyzed for its concentration as previously described [48]. Epididymal sperm released in PBS from cauda epididymides was spread on glass slides, air dried and either stained with hematoxylin for morphological analysis or used for immunofluorescence.

Electron microscopy
Testis and epididymal sperm samples were fixed in 5% glutaraldehyde and treated with a potassium ferrocyanide-osmium fixative. The samples were embedded in epoxy resin (Glycidether 100, Merck), sectioned, stained with 5% uranyl acetate and 5% lead citrate, and visualized on a JEOL 1200 EX transmission electron microscope.