DAZL is essential for stress granule formation implicated in germ cell survival upon heat stress

In most mammals, the testes are located in the scrotum, outside the body cavity, and are maintained at a lower temperature than the core body. Developmental defects in testicular descent often lead to germ-cell loss in the affected testis (Hutson et al., 1997). Artificially induced cryptorchidism, in which a testis is forced to stay in the body cavity, causes apoptosis of the germ cells (Shikone et al., 1994; Yin et al., 1997). Experimentally, forced heat stress on the testis also causes apoptosis of the germ cells, leading to subfertility or infertility (Lue et al., 1999; Rockett et al., 2001). However, the mechanism of cellular responses to heat stress remains largely unknown in male germ cells.

Stresses such as heat, hypoxia and oxidative conditions often trigger the inhibition of general translation and the formation of stress granules (SGs) in eukaryotic cells (reviewed by Anderson and Kedersha, 2002; Buchan and Parker, 2009). SGs are non-membranous cytoplasmic particles at which translationally inert messenger ribonucleoproteins (mRNPs) accumulate in response to stress (Anderson and Kedersha, 2002). The known components of SGs include polyadenylated mRNAs [poly-(A)⁺ RNAs], 40S ribosomal subunits, proteins involved in translation, such as the eukaryotic translation initiation factors eIF3, eIF4E, eIF4G and poly(A)-binding protein (PABP; PABPC1 – Mouse Genome Informatics), and several RNA-binding proteins including TIA1, HuR (ELAVL1 – Mouse Genome Informatics) and G3BP (Kedersha et al., 1999; Kedersha et al., 2002; Gallouzi et al., 2000; Tourrière et al., 2003). It is known that the phosphorylation of eIF2α (EIF2S1 – Mouse Genome Informatics) prevents the assembly of the translation initiation complex and induces the assembly of SGs (Kedersha et al., 1999). It is also known that the spontaneous assembly of SGs occurs when some of the SG components are overexpressed (Anderson and Kedersha, 2008). The mRNPs of SGs are in a dynamic equilibrium with polysomes (Kedersha et al., 1999; Kedersha et al., 2000). In this regard, SGs have been proposed to function as a transient place of mRNP remodeling for storage, degradation or reinitiation of translation during stress and recovery from stress (Kedersha and Anderson, 2002; Kedersha et al., 2005; Anderson and Kedersha, 2008). SGs have also been shown to regulate cell survival during stress by the sequestration of the signaling molecules implicated in apoptosis, such as TRAF2 and RACK1 (GNB2L1 – Mouse Genome Informatics) (Kim et al., 2005; Arimoto et al., 2008).

The human DAZ (‘deleted in azoospermia’) gene is located on the Y chromosome in a region in which microdeletions are frequently observed in infertile men (Reijo et al., 1995). The DAZ family proteins are crucial for germ cell development in Caenorhabditis elegans (Karashima et al., 2000), Drosophila melanogaster (Eberhart et al., 1996), Xenopus laevis (Houston and King, 2000) and mice (Mus musculus) (Ruggiu et al., 1997). Moreover, rescue experiments, in which an original gene was substituted with another gene family member, revealed evolutionarily conserved functions of the DAZ family genes in germ cell development (Houston et al., 1998; Slee et al., 1999; Xu et al., 2003). The DAZ family proteins are detected in male and female germ cells as well as in the germ plasm of early amphibian embryos (Ruggiu et al., 1997; Houston and King, 2000; Karashima et al., 2000; Reijo et al., 2000; Xu et al., 2001). DAZL is an RNA-binding protein and interacts with a specific set of mRNAs, such as the mouse vasa homolog, Mvh (Ddx4 – Mouse Genome Informatics), and the synaptonemal complex component Sycp3 (Reynolds et al., 2005; Reynolds et al., 2007). DAZL was proposed to function as an adaptor for mRNA transport and as a translational activator (Collier et al., 2005; Lee et al., 2006; Reynolds et al., 2005; Reynolds et al., 2007). In mouse oocytes and early embryos, DAZL is involved in translational activation of the preserved mRNAs to promote oocyte maturation and early embryonic development (Chen et al., 2011). DAZL is required for the formation of primordial germ cells from human embryonic stem cells (Kee et al., 2009). It was recently reported that DAZL plays a crucial role in licensing fetal germ cells to undergo gametogenesis (Gill et al., 2011).

We observed previously that ectopic DAZL colocalizes with TIA1, an SG marker, in HeLa cells under oxidative stress (Lee et al., 2006). This finding led us to speculate about a novel function for DAZL upon heat stress. Here, we reveal that SGs are assembled in mouse male germ cells upon heat stress and that DAZL is an essential component of SGs. Furthermore, we provide evidence that the SGs in male germ cells have a protective function against apoptosis upon heat stress.

All mice were housed and treated under approval of Institutional Animal Care and Use Committee at Seoul National University (SNU-090320-3). Adult (8-10 weeks) or one-week-old [7 days post-partum (dpp)] mice were used for heat-treatment experiment. Male mice were anesthetized and the lower third of the body was placed in a water bath at 32, 37, 40 or 42°C for the indicated time intervals. Unless otherwise indicated, ‘heat stress’ means a treatment at 42°C for 20 minutes. Control mice were anesthetized and left at room temperature. Following treatment, the mice were sacrificed and the testes were isolated for further analyses.

C57BL/6 inbred Dazl^{Tm1Hgu/Tm1Hgu} mice were previously described (Ruggiu et al., 1997; Reynolds et al., 2007). To generate outbred C57BL/6 × CD1 lines, C57BL/6 Dazl^+/– male mice were mated with CD1 females and the resulting Dazl^+/– mice were intercrossed to produce Dazl^–/– male mice. Male Dazl^–/– and their littermates at 7 dpp were control- or heat-treated for further analyses. Genotypes were screened by PCR using DNAs isolated from tail tips as described previously (Ruggiu et al., 1997).

Some of outbred Dazl^–/– male mice lack the whole male germ cells even at 7 dpp, owing to individual variations (Reynolds et al., 2007). The presence or absence of male germ cells was determined histologically in one of the testes pair of the 7 dpp outbred Dazl^–/– mouse. Immunoblot analysis was performed with the other testis of the same animal.

Mouse testes were fixed in Bouin’s solution (Sigma-Aldrich) and paraffin embedded. All tissues were sectioned at 5 μm, and were deparaffinized and hydrated for further analyses. Antigen retrieval was conducted by microwave boiling for 15 minutes in one of three buffers: 10 mM citrate (pH 6.0), 10 mM Tris-HCl (pH 9.0) or 1 mM EDTA (pH 8.0). After cooling at room temperature, the testicular sections were blocked with 3% BSA (bovine serum albumin) in PBST (phosphate-buffered saline with 0.1% Tween 20) for 30 minutes and then incubated with the following antibodies for 1 hour: affinity-purified anti-DAZL (1:10) (Lee et al., 2006), anti-TIA-1 (1:25, Santa Cruz), affinity-purified anti-DAZAP2 (1:10) (see below), anti-phospho-eIF2α (1:50, Stressgen), anti-DCP2 (1:50) (Wang et al., 2002), anti-S6 kinase (1:50, Santa Cruz) that cross-reacts with HEDLS (Stoecklin et al., 2006), anti-eIF3 (1:25, Santa Cruz), anti-TDRD1 (1:250) (Chuma et al., 2003), anti-γ-H2AX (1:500, Millipore), anti-MVH (1:250) (Toyooka et al., 2000), anti-RACK1 (1:50, BD Biosciences or Abcam), anti-MTK1 (1:40, Sigma-Aldrich) and anti-phospho-p38 (1:50, Cell Signaling). After three washes with PBST, the sections were incubated with either biotinylated (Vector) or fluorophore-conjugated secondary antibodies (Invitrogen) for 30 minutes. Sections treated with the biotinylated antibody were then incubated with avidin:biotinylated peroxidase complex (Vector) after PBST washing and the sections were developed with 3,3′-diaminobenzidin tetrachloride (DAB) (Sigma-Aldrich). The sections were counterstained with Hematoxylin and DAPI (Sigma-Aldrich) for DAB and fluorescence staining, respectively.

For in situ hybridization, deparaffinized and hydrated testicular sections were placed in buffer containing 2×SSC (300 mM NaCl, 30 mM sodium citrate, pH 7.0) with 15% formamide for 5 minutes. Slides were then incubated with 50 ng of oligo-dT(45)-3′-TAMRA (5-caboxytetramethylrhodamin) (Cosmo) diluted in hybridization buffer (2×SSC, 15% formamide, 50 μg/ml heparin, 250 μg/ml tRNA, 0.1% Tween 20) at 37°C for 1 hour. After hybridization, slides were washed twice with 2×SSC, 15% formamide at 37°C for 30 minutes and were counterstained with DAPI. As a negative control, a slide was incubated in 1 mg/ml RNase at 37°C for 30 minutes before probe hybridization. For colocalization analysis, slides were processed for antigen retrieval, in situ hybridization, immunostaining with antibody and then DAPI staining.

The slides were observed with a light or fluorescence microscope (Olympus IX51), and images were acquired using a CCD camera (Qicam fast 1394, Qimaging) and processed using ImagePro 5.0 (Media Cybernetics). For measurement of p38 phosphorylation signal intensity, immunofluorescent images were processed using Adobe Photoshop (Adobe Systems). Statistical data were analyzed by a t-test using SigmaPlot (Systat Software).

The coding region of the human DAZAP2 cDNAs (Kim et al., 2008) was subcloned into pGEX-4T vector (GE Healthcare) for the generation of GST-DAZAP2 fusion protein. A rabbit polyclonal anti-DAZAP2 antibody was raised against bacterially expressed GST-DAZAP2 and affinity-purified using a method described elsewhere (Kim et al., 2009). Briefly, the antisera were incubated with a strip of nitrocellulose membrane blotted with GST-DAZAP2 fusion protein, and eluted with 100 mM glycine (pH 2.5).

HeLa cells were cultured in DMEM supplemented with 10% FBS at 37°C and 5% CO₂. Transfection was carried out using FuGENE (Roche) for the prepared truncated DAZL mutants. HeLa cells cultured on a coverslip were either control or heat (44°C for 1 hour) treated and followed by immunocytochemistry.

For immunocytochemistry, the coverslips were briefly washed with PBS and fixed with 4% paraformaldehyde for 10 minutes. After a wash with 0.3% PBST, the coverslips were blocked with 3% BSA in PBST for 30 minutes and then incubated with anti-HA (1:500, Sigma) and anti-TIA-1 (1:100) for 1 hour. After three washes with PBST, cells were incubated with secondary antibodies for 30 minutes, washed with PBST and counterstained with DAPI. Observation and image acquisition were described above.

Mouse testes were decapsulated and homogenized in 1×SDS sample buffer (50 mM Tris-HCl, pH 6.8, 100 mM dithiothreitol, 2% SDS, 0.1% Bromophenol Blue, 10% glycerol). About 20-50 μg of protein was resolved by SDS-PAGE and transferred into nitrocellulose membranes. The membranes were blocked with 5% skimmed milk in TBST (20 mM Tris, 150 mM NaCl, 0.1% Tween 20) for 30 minutes and then incubated overnight at 4°C with the following antibodies: anti-phospho-eIF2α (1:500), anti-eIF2α (1:150) (Santa Cruz), anti-DAZL (1:100), anti-β-tubulin (1:1000) (Sigma) and anti-MVH (1:2000). After three washes with TBST, the membranes were incubated with peroxidase-conjugated secondary antibodies for 30 minutes. After a further three washes with TBST, peroxidase activity was detected using ECL reagent.

For measurement of eIF2α phosphorylation levels, immunoblot images were processed using Adobe Photoshop. The fold increase was determined by the signal intensity of eIF2α phosphorylation level normalized to total eIF2α.

Full scan images of the immunoblots shown in the figures are shown in supplementary material Fig. S4.

TUNEL (terminal deoxynucleotidyl transferase-mediated deoxy-UTP nick end labeling) assay was conducted using a FragEL DNA Fragmentation Detection Kit (Calbiochem) according to the manufacturer’s instruction. For double-staining with antibody, slides were processed with antigen retrieval, TUNEL assay, immunostaining with antibody and then DAPI staining.

DAZL was detected primarily in the cytoplasm of spermatogonia and spermatocytes (Fig. 1A) (Ruggiu et al., 1997). DAZL expression is developmental stage-specific and its expression is significantly reduced in spermatocytes near meiosis at tubular stage IX (Fig. 1A). No DAZL signal was detected in postmeiotic spermatids of all tubular stages (Fig. 1A). We examined the subcellular localization of DAZL in heat-stressed mouse testes. When adult mice were placed in a warm water bath at 42°C for 20 minutes, DAZL formed aggregates at the perinuclear region of the male germ cells at tubular stages III-VI (Fig. 1A). The DAZL granules were detected in spermatogonia and in preleptotene and early pachytene spermatocytes, but not in leptotene, zygotene and late pachytene spermatocytes, in which DAZL expression was minimal (Fig. 1B).

We performed co-staining analyses to examine the possibility that the DAZL-positive granules are SGs. TIA1 is known to form a prion-like aggregation for SG assembly in a stressed cell (Kedersha et al., 1999; Gilks et al., 2004). In normal testis, TIA1 was detected in the nuclei of spermatocytes and spermatids as well as somatic cells, such as Sertoli and Leydig cells (Fig. 2A). However, no specific TIA1 signal was detected in spermatogonia (Fig. 2A). In the heat-stressed testis, TIA1 assembled into cytoplasmic granules in spermatocytes and colocalized with the DAZL-positive granules (Fig. 2A). Poly-(A)⁺ RNAs, a major component of SGs, were found in large granules that were identified as chromatoid bodies in round spermatids (Fig. 2A) (Kotaja et al., 2006). The Poly-(A)⁺ RNAs were also detected in the cytoplasm of spermatocytes and spermatogonia (Fig. 2A). Upon heat stress, however, poly-(A)⁺ RNAs colocalized with the DAZL-positive granules in spermatocytes and spermatogonia (Fig. 2A). Another SG component, DAZAP2, is known to interact with eIF4G and to induce SG formation (Kim et al., 2008). In normal testis, DAZAP2 was detected in the nucleus and cytoplasm of male germ cells of all developmental stages except for elongated spermatids (Fig. 2A). In the heat-stressed testis, DAZAP2 colocalized with the DAZL-positive granules in spermatocytes and spermatogonia (Fig. 2A). Phospho-eIF2α, an SG component, also migrated to and colocalized with the DAZL-positive cytoplasmic granules upon heat stress (Fig. 2A) (Kedersha et al., 2005). Taken together, these data reveal that DAZL localizes to the SGs in spermatogonia as well as in pachytene spermatocytes of tubular stages III-VI. SGs were detected neither in spermatocytes at tubular stages I-II and VII-XII nor in spermatids of all stages, in which DAZL is absent.

In mammalian cells, SGs are dynamically linked to processing bodies (PBs) where mRNA degradation takes place (Kedersha et al., 2005). We performed co-immunostaining analysis to determine whether DAZL localizes to PBs or not. PBs, which were immunostained with antibodies specific to a decapping enzyme (DCP2) and a decapping factor (HEDLS; EDC4 – Mouse Genome Informatics), were detected as cytoplasmic foci of pachytene spermatocytes in control- and heat-treated testes (Fig. 2B; supplementary material Fig. S1A). The co-immunostaining analyses revealed that the DAZL-positive granules in heat-stressed spermatocytes did not overlap with the PB markers (Fig. 2B; supplementary material Fig. S1A). To confirm an exclusive localization of DAZL into SGs, we co-immunostained DAZL with eIF3, a protein found in SGs but not in PBs (Kedersha et al., 2005). The results showed that DAZL-positive granules did colocalize with eIF3 in the pachytene spermatocytes of tubular stages III-VI (supplementary material Fig. S1B). These results confirm that the DAZL-positive granules in heat-stressed animals are indeed SGs.

Mammalian male germ cells are known to contain an RNA granule termed nuage. Tudor domain-containing 1 (TDRD1) proteins are characteristically located at the nuage structure (Chuma et al., 2003; Chuma et al., 2006). To determine whether nuages and SGs are the same structures in heat-stressed mice, we performed co-immunostaining analysis with antibodies specific to DAZL and TDRD1. The results showed that heat stress did not affect the subcellular distribution of the TDRD1-positive granules (Fig. 2C). Furthermore, the DAZL-containing SGs did not colocalize with the TDRD1-containing nuages (Fig. 2C). These results reveal that DAZL-containing SGs are distinct from nuages in male germ cells.

We examined which domain of DAZL is responsible for the SG recruitment. DAZL, an RNA-binding protein, is 298 amino acids long with an RNA recognition motif and a conserved DAZ repeat (Fig. 2D). A series of deletion mutants were transiently transfected into HeLa cells and the subcellular distribution of the truncated DAZL mutant proteins was examined in control and heat-stressed cells. The full length and N-terminal truncated DAZL variants were recruited into SGs, whereas the C-terminal and short N-terminal truncated mutants appeared diffused throughout the cytoplasm in the heat-stressed condition (Fig. 2D). The truncated mutant with a deletion between RRM and the DAZ repeats (Δ116-166) also failed to localize in SGs, suggesting that the 116-166 region is crucial for the recruitment of DAZL to SGs (Fig. 2D).

The mammalian core body temperature (37°C) is hyperthermic for male germ cells (Hutson et al., 1997). We carefully examined SG formation in male germ cells under hyperthermic conditions. First, we performed immunoblot analysis to determine the phosphorylation status of eIF2α, a molecular signature of translational inhibition and SG assembly, in heat-stressed testes (Kedersha et al., 1999). The results showed that eIF2α began to be phosphorylated even after treatment at 37°C for 20 minutes and that the phospho-eIF2α levels increased further at higher temperatures (Fig. 3A). Next, immunohistochemical analysis was performed to confirm SG formation at 37°C. The results showed that SGs, which were immunostained with antibodies specific to DAZL and TIA1, started to be assembled at 37°C and were conspicuous at 40°C in early pachytene spermatocytes (Fig. 3B). These results reveal that the core body temperature is sufficient to induce SG formation in pre-meiotic male germ cells.

We examined further the dynamics of eIF2α phosphorylation and SG formation at 37°C. Upon prolonged exposure to heat stress at 37°C, the mouse testis showed an increase in eIF2α phosphorylation and SG formation; phosphorylation of eIF2α was induced within half an hour and it continued to increase slightly up to 2 hours (Fig. 3C). Similarly, SG started to assemble in 10% of early pachytene spermatocytes at tubular stages III-IV within half an hour and gradually increased to 29% after 2 hours (Fig. 3D). It is known that mammalian SGs are reversibly assembled and disassembled in accordance with heat stress and recovery (Kedersha et al., 1999). When the heat-stressed mice were placed at room temperature for 1 hour, eIF2α was rapidly dephosphorylated and the SGs were disassembled (Fig. 3E,F). DAZL protein levels were unchanged for either the stress or recovery conditions (Fig. 3A,C,E). Taken together, these results reveal that SG formation is an immediate and reversible response to hyperthermic stress in pre-meiotic spermatocytes.

We decided to use Dazl-knockout (Dazl-KO) mice to explore the involvement of DAZL in SG formation. It is known that adult Dazl-KO mice lack germ cells in their testes (Ruggiu et al., 1997). Careful examination revealed that the Dazl-KO male mice with an inbred C57BL/6 background lose their germ cells starting at embryonic day 15.5, but those with an outbred genetic background, C57BL/6 × CD1, can maintain their germ cells until 7 dpp (Lin and Page, 2005; Reynolds et al., 2007). Some of the testes of Dazl-KO mice lacked male germ cells at 7 dpp owing to the individual variation within a mixed genetic background providing a valuable control (Reynolds et al., 2007). Therefore, we used the testes from 7 dpp Dazl-KO mice of a mixed genetic background for our experiments.

First, we analyzed the stress responses of wild-type testes at 7 dpp. Phosphorylation of eIF2α was observed following heat stress of 40°C or 42°C for 20 minutes (supplementary material Fig. S2A). SGs were assembled in most of the wild-type germ cells, which were identified with the γ-H2AX antibody (supplementary material Fig. S2B) (Mahadevaiah et al., 2001). We then evaluated SG formation in the testes of Dazl-KO mice by staining for SG markers in addition to γ-H2AX (H2AFX – Mouse Genome Informatics). As expected, DAZL was absent in the surviving germ cells of the Dazl^–/– mice, whereas DAZL was expressed in the Dazl^+/– testes and formed aggregates upon heat stress (Fig. 4A). The selected SG markers, namely poly-(A)⁺ RNAs, DAZAP2 and phospho-eIF2α, relocated to the aggregates upon heat stress in Dazl^+/– germ cells, but few such aggregates were formed in Dazl^–/– germ cells (Fig. 4B-D). We counted the number of germ cells with SGs in the heat-stressed Dazl^–/– animals and their littermates. For all three SG markers tested, SG formation was significantly reduced in the Dazl^–/– testes compared with the wild-type and Dazl^+/– testes (Fig. 4E). These results indicate that DAZL is essential for SG formation upon heat stress. TIA1 cannot be used as a SG marker because it is not expressed in the early germ cells at 7 dpp.

We observed previously the presence of phospho-eIF2α in the SGs of male germ cells under heat stress (Fig. 2A, Fig. 4D). It is known that the phosphorylation of eIF2α often precedes SG assembly (Kedersha et al., 1999; Anderson and Kedersha, 2008). Therefore, we investigated whether eIF2α is phosphorylated in the male germ cells of Dazl-KO mice where SGs are not assembled under heat stress. Immunoblot analysis confirmed a complete depletion of DAZL expression in the Dazl^–/– testes (Fig. 4F). Despite the absence of DAZL, eIF2α was still phosphorylated at a similar level in the heat-stressed testis (Fig. 4F, lanes 1-6). It is possible that a large portion of the phospho-eIF2α proteins in the Dazl-KO testes originates from cell types other than the germ cells. To test this possibility, we evaluated the phospho-eIF2α levels in Dazl-KO testes with and without germ cells under a heat-stress condition. The presence of male germ cells in the 7 dpp testes was determined with an MVH antibody (Fig. 4F, lanes 7-10). The results show that the phospho-eIF2α levels in Dazl-KO testes without germ cells were significantly lower than those with germ cells (Fig. 4F, lanes 8 and 10). These results suggest that the increase in phospho-eIF2α levels in heat-stressed testes is largely due to the germ cells, and also strongly suggest that eIF2α phosphorylation in germ cells upon heat stress is independent of DAZL.

It is known that mammalian SGs are implicated in cell survival during stress (Kwon et al., 2007; Arimoto et al., 2008; Eisinger-Mathason et al., 2008). It is also well known that male germ cells at specific developmental stages are sensitive to heat stress. The apoptotic rate increases in pre-meiotic spermatocytes at tubular stages IX-XII, and in early spermatids at tubular stages I-II (supplementary material Fig. S3) (Lue et al., 1999). We have observed that spermatocytes at tubular stages IX-XII are devoid of DAZL and do not assemble SGs under heat stress (Fig. 1). Therefore, we decided to test the hypothesis that the SGs of male germ cells have a protective role against heat stress. We first observed SG formation and apoptosis in the adult mouse testis after heat stress and recovery. The results showed that eIF2α was immediately phosphorylated upon heat stress and reversibly dephosphorylated 6 hours after the recovery (Fig. 5A). DAZL-containing SGs in spermatocytes at tubular stages III-IV were assembled and disassembled under the same conditions of heat stress and a recovery period (Fig. 5B). However, no DAZL-containing SGs were formed in the spermatocytes at tubular stages XI-XII upon heat stress (Fig. 5B). We performed the TUNEL (terminal deoxynucleotidyl transferase-mediated deoxy-UTP nick end labeling) assay to identify any germ cell apoptosis. The results showed that germ cell apoptosis occurred 6 hours after heat stress and recovery in the late spermatocytes, but not in the early spermatocytes in which SGs had been assembled (Fig. 5B,C). These results suggest a functional association of the DAZL-containing SGs with germ cell survival.

The importance of DAZL in germ cell survival under a heat stress condition was examined in Dazl-KO mice. Cells undergoing apoptosis were detected more frequently in the testes of Dazl^–/– mice than in those of Dazl^+/– mice even under normal conditions (Fig. 5D). Furthermore, heat stress induced a significant increase in the number of apoptotic cells by 40% in Dazl^–/– testes, but not in the control testes (Fig. 5D). These results support the notion that DAZL protects germ cells against apoptosis upon heat stress by inducing SG formation.

A protective function of SGs against apoptosis has been demonstrated in cultured somatic cells. The sequestration of RACK1 into SGs perturbs its interaction with MTK1, a mitogen-activated protein kinase kinase kinase (MAP3K4), and, consequentially, blocks the apoptotic MAPK signaling pathway (Arimoto et al., 2008). In fact, it was reported that the p38-mediated MAPK signaling pathway is involved in male germ cell apoptosis upon heat stress (Jia et al., 2009). Therefore, we hypothesized that SGs inhibit germ cell apoptosis by sequestering RACK1 in SGs, thus blocking the p38-MAPK signaling pathway in male germ cells.

We examined initially the subcellular localization of RACK1 in heat-stressed mouse testes. RACK1 was distributed throughout the cytoplasm of male germ cells in a diffuse pattern, but was recruited to SGs upon heat stress (Fig. 6A). In Dazl^–/– germ cells, RACK1 remained in the cytoplasm, probably because SGs were not assembled upon heat stress (Fig. 6B). By contrast, MTK1, a RACK1-binding partner, was localized to the cytoplasm in a diffuse pattern even under heat stress (Fig. 6B). These results raise the possibility that the apoptotic MAPK signaling could be blocked in the germ cells by sequestration of RACK1 in SGs, resulting in MTK1 inactivation. To test this possibility, we analyzed p38 activation in heat-stressed Dazl^–/– male germ cells in which SG formation is impaired but not completely eliminated (Fig. 4E). As expected, p38 activation, as shown by the phospho-p38 signal, was higher in the germ cells without SGs than in those with SGs (Fig. 6C). Consistent with this finding, we also observed that RACK1 was localized to SGs along with DAZL in heat-stressed adult testes (Fig. 6D). The phospho-p38-positive cells were selectively detected in germ cells at tubular stages XI-XII, which are the developmental stages at which germ cell apoptosis occurs upon heat stress (Fig. 6E). These results support the hypothesis that SG formation inhibits germ cell apoptosis, at least partially, by blocking the p38-MAPK signaling pathway in male germ cells.

In this study, we report that phosphorylation of eIF2α is one of the immediate responses in male germ cells to heat stress. SGs are assembled at the same time. These changes might induce a global repression of translation, as shown by the reduction of the polysomes in heat-stressed testes (Cataldo et al., 1997). A proteomic study revealed that transient heat stress induces an immediate change in the protein profile of male germ cells (Zhu et al., 2006). Phosphorylation of eIF2α probably causes this swift inhibition of translation. SGs might store a specific set of mRNAs until the testis has recovered from heat stress.

We also observed that the cytoplasmic DAZL in male germ cells was relocated to the SGs upon heat stress. SGs can be formed only at specific developmental stages of male germ cells; SGs were detected in spermatogonia and early pachytene spermatocytes, in which DAZL is abundantly expressed, but not in late pachytene and diplotene spermatocytes, in which DAZL expression is reduced. Furthermore, SG assembly was significantly diminished in the early male germ cells of Dazl-KO mice. These results suggest that DAZL is involved in the heat-stress response as an essential component of SGs in male germ cells (Fig. 7).

One of the unresolved issues in this study is how DAZL is recruited to SGs and how it regulates their formation upon heat stress. It is known that protein-protein interactions are important for SG assembly (Anderson and Kedersha, 2008). Our truncated mutant analysis revealed that residues 116-166 of DAZL are required for SG localization in HeLa cells. This region is highly conserved among the DAZ family proteins. It is also responsible for the specific interaction with PABP, a major SG component, suggesting that DAZL is recruited to SGs through specific interaction with PABP (Kedersha et al., 1999; Collier et al., 2005). It is also noteworthy that DAZL interacts with DAZAP2, an SG component and an interacting partner of the translation initiation factor eIF4G (Tsui et al., 2000; Kim et al., 2008). At the same time, DAZL might function as a nucleator of SG assembly, as ectopic expression of DAZL induces SG formation in HeLa cells (Lee et al., 2006).

A protective role of SGs against apoptosis has been reported in cultured somatic cells (Arimoto et al., 2008). Here, we report evidence that the DAZL-containing SGs in male germ cells also have a protective role against apoptosis upon heat stress. The male germ cells in Dazl-KO mice did not assemble SGs properly upon heat stress and the number of apoptotic cells increased correspondingly. In adult mice, germ cell apoptosis occurred mostly at tubular stages IX-XII when the cellular DAZL level is reduced and SGs are absent. Our immunohistochemical results are consistent with the hypothesis that germ cell apoptosis is inhibited by the sequestration of specific signaling molecules, such as RACK1 in SGs, thereby causing the blockage of the apoptotic MAPK pathway (Arimoto et al., 2008). In fact, MAPK is known to be crucial for male germ cell apoptosis triggered by hormone deprivation or heat stress (Jia et al., 2009). We also do not rule out another possible protective role for SGs against apoptosis, specifically the salvaging of mRNAs that are crucial for developmental events, such as meiosis. Once the germ cells have recovered from the stress, these mRNAs might be released from the SGs and translated to resume germ cell development without interruption.

It was reported previously that DAZL functions as a translational regulator for a specific set of genes (Collier et al., 2005; Reynolds et al., 2005; Reynolds et al., 2007). It was also proposed that DAZL functions as an mRNA adaptor for the dynein motor complex for intracellular transport (Lee et al., 2006). Recently, DAZL was found to be an intrinsic factor that enables embryonic germ cells to initiate meiosis in response to retinoic acid (Lin et al., 2008). In addition, DAZL was shown to regulate the preserved mRNAs by translational activation to promote early embryonic development (Chen et al., 2011). In this study, we propose that DAZL is an essential component of SGs and has a protective function against heat stress (Fig. 7). We speculate that the stress-protective function of DAZL is not mutually exclusive with the other known functions of DAZL under normal conditions. DAZL might be associated with a specific set of mRNAs for transport and translational regulation. Once cells are under heat stress, DAZL becomes an essential component of SGs and might actively bring the associated mRNAs into SGs (Fig. 7).

Our study suggests that SG formation is one of the immediate protective mechanisms against heat stress. However, it remains to be investigated whether SGs also play a protective role against prolonged stresses, such as cryptorchidism. It is unlikely that heat stress is the main cause of male germ cell death in Dazl-KO mice, which occurs as early as embryonic day 15.5 (Lin and Page, 2005). Nonetheless, it is possible that the heat-protective function of DAZ might be related to male fertility in humans. There are three DAZ family genes in humans; DAZL and BOULE (BOLL – Human Gene Nomenclature Database) are located in autosomal chromosomes and four copies of DAZ are located at the AZFc locus of the Y chromosome. Because microdeletions at the AZFc locus are frequently observed in infertile males, DAZ is suspected to be a male infertility gene. However, there are AZFc-deficient males with normal fertility, indicating that the phenotypic penetration of DAZ is incomplete (Chang et al., 1999; Kühnert et al., 2004). Genetic and biochemical evidence supports the idea that the DAZ family proteins share common biological functions at the different developmental stages of male germ cells (reviewed by Reynolds and Cooke, 2005). If DAZ family proteins are important for SG formation, then DAZ-deficient male patients might be more susceptible to heat stress than normal males, which would intensify male infertility. This possibility should be examined extensively for the purpose of clinical application.

We thank M. Kiledjian, S. Chuma and T. Noce for generously providing DCP2, TDRD1 and MVH antibodies, respectively.