Associations between PIWI proteins and TDRD1/MTR-1 are critical for integrated subcellular localization in murine male germ cells

Authors

  • Kanako Kojima,

    1. Department of Pathology, Medical School, Osaka University, Yamada-oka 2-2 Suita, Osaka 565-0871, Japan
    Search for more papers by this author
  • Satomi Kuramochi-Miyagawa,

    Corresponding author
    1. Department of Pathology, Medical School, Osaka University, Yamada-oka 2-2 Suita, Osaka 565-0871, Japan
    2. PREST, JST, 4-1-8 Honcho Kawaguchi, Saitama, Japan
    Search for more papers by this author
  • Shinichiro Chuma,

    1. Institute for Frontier Medical Sciences, and Graduate School of Medicine, Kyoto University, 53 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto 606-8507, Japan
    Search for more papers by this author
  • Takashi Tanaka,

    1. Institute for Frontier Medical Sciences, and Graduate School of Medicine, Kyoto University, 53 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto 606-8507, Japan
    Search for more papers by this author
  • Norio Nakatsuji,

    1. Institute for Frontier Medical Sciences, and Graduate School of Medicine, Kyoto University, 53 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto 606-8507, Japan
    Search for more papers by this author
  • Tohru Kimura,

    1. Department of Pathology, Medical School, Osaka University, Yamada-oka 2-2 Suita, Osaka 565-0871, Japan
    Search for more papers by this author
  • Toru Nakano

    Corresponding author
    1. Department of Pathology, Medical School, Osaka University, Yamada-oka 2-2 Suita, Osaka 565-0871, Japan
    Search for more papers by this author

  • Communicated by: Tetsuya Taga

* Correspondence: tnakano@patho.med.osaka-u.ac.jp or smiya@patho.med.osaka-u.ac.jp

Abstract

The mouse Piwi family proteins (MILI, MIWI and MIWI2) play pivotal roles in spermatogenesis through transcriptional and post-transcriptional gene regulation. To reveal the molecular functions of these proteins, we investigate the proteins that bind to MILI in adult mouse testes. We found that both MILI and MIWI bind to TDRD1/MTR-1, which is also an essential protein for spermatogenesis. Co-immunoprecipitation assays and subcellular localization of the proteins and mutants thereof revealed a complex formation involving MILI, MIWI and TDRD1/MTR-1. In addition, the subcellular localizations of MILI and TDRD1/MTR-1 were altered, and chromatoid body formation was impaired in the MIWI-null round spermatids. These data suggest that the formation of complexes between MILI, MIWI and TDRD1/MTR-1 is critical for the integrated subcellular localizations of these proteins, and is presumably essential for spermatogenesis.

Introduction

RNA silencing is an evolutionarily conserved molecular mechanism for the inhibition of gene expression through mRNA degradation, translational repression or transcriptional repression (Elbashir et al. 2001; Filipowicz 2005; Zamore & Haley 2005; Vaucheret 2006; Zaratiegui et al. 2007). Three classes of small RNAs, small interfering RNAs (siRNAs), microRNAs (miRNAs), and the newly identified PIWI-interacting RNAs (piRNAs), are involved in the RNA silencing pathway (Farazi et al. 2008; Hutvagner & Simard 2008). In contrast to the ubiquitous expression of miRNAs, piRNAs are expressed specifically in germ cells and, therefore, piRNAs are considered to play important roles in gametogenesis (Klattenhoff & Theurkauf 2008; Kim et al. 2009).

The Piwi (p element induced wimpy testis) gene was originally identified as being essential for the maintenance of germ stem cells in Drosophila (Cox et al. 1998). The Piwi gene family is highly conserved among various species, including the three murine Piwi family genes, i.e. Miwi (mouse piwi), Mili (miwi like) and Miwi2. Although all three murine PIWI family members are expressed in germ lineage cells, their durations of expression differ. MIWI is expressed from the spermatocyte to the round spermatid stage in the postnatal testis (Deng & Lin 2002). MILI is expressed from the primordial germ cell at embryonic day 12 to the round spermatid stage in the adult testis (Kuramochi-Miyagawa et al. 2001, 2004). MIWI2 is expressed from the gonocyte at embryonic day 15 until soon after birth (Kuramochi-Miyagawa et al. 2008). Mice that carry null mutations in these genes are sterile, although they show differences in the timing of the abnormality in spermatogenesis.

The germ cells of various organisms are characterized by the accumulation of dense fibrous materials into a cytoplasmic structure called ‘nuage’ (Eddy 1975; Ikenishi 1998). A collection of finely filamentous and lobulated perinuclear granules, termed the ‘chromatoid body’, is found in the mammalian postmeiotic round spermatids (Fawcett et al. 1970). VASA, which is an ATP-dependent DEAD box RNA helicase, and its murine homologue MVH (mouse vasa homolog) are found in Drosophila nuage (Liang et al. 1994) and murine chromatoid bodies (Fujiwara et al. 1994; Tanaka et al. 2000; Toyooka et al. 2000), respectively. The VASA protein presumably plays a central role in the translational regulation of many mRNA species, and nuage is the site for this process (Carrera et al. 2000). MIWI, MILI and MAELSTROME (MAEL) are also strongly concentrated in the chromatoid bodies, and MILI and MAEL are shown to be critical for transposon silencing in spermatogenesis (Kuramochi-Miyagawa et al. 2004; Kotaja et al. 2006a; Soper et al. 2008; Unhavaithaya et al. 2009). Considering that piRNA biogenesis is independent of Dicer (Vagin et al. 2006), these data raise the possibilities that: (i) the chromatoid body is also a site for the piRNA specific pathway, which is crucial for normal male germ cell differentiation; and (ii) that the protein complex that contains the murine PIWI proteins is important for functionality.

In this study, we analyzed the proteins that bind to MILI in postnatal testes. We identified TDRD1/MTR-1, which is also a component of the chromatoid body, as a protein that binds to MILI, and we discovered that MIWI also binds to TDRD1/MTR-1. Co-expression experiments using these three proteins and mutants thereof demonstrated that the binding of TDRD1/MTR-1 to these two murine PIWI proteins determines their subcellular localizations. We found that the cytoplasmic localizations of MILI and TDRD1/MTR-1 were abnormal in the MIWI-null round spermatids. In addition, the accumulation of MILI and TDRD1/MTR-1 in chromatoid bodies was impaired, and abnormal nuclear localization of the proteins was observed in the MIWI-null round spermatids. These data demonstrate that binding among MILI, MIWI and TDRD1/MTR-1 is crucial for the integrated subcellular localization of these proteins and is presumably essential for spermatogenesis.

Results

TDRD1/MTR-1 as a binding partner of MILI and MIWI

To characterize the molecular function of MILI, proteins that associate with MILI were identified by immunoprecipitation. Immunoprecipitated proteins from the control and MILI-null testes of 16-day-old mice were subjected to SDS–PAGE electrophoresis and silver staining (Fig. 1a). The IgG control had background signal. By comparing the lanes for the Mili+/− and Mili−/− testis samples, two bands exclusive to the Mili+/− testes were identified: (i) a protein with relative molecular mass of 110 kDa, which corresponded to MILI; and (ii) a 130-kDa protein. Immunoprecipitation of the proteins from the wild-type and MIWI-null testes was performed in conjunction with mass spec analysis (Ultraflex MALDI-TOF mass spectrometer; Bruker Daltonics, Billerica, MA, USA) analysis (Fig. 1b). The 23 peptide fragments of the 130-kDa band revealed that the protein was identical to TDRD1/MTR-1. Immunoprecipitation analysis with the anti-TDRD1/MTR-1 antibody verified that not only MILI, but also MIWI interacted with TDRD1/MTR-1 (Fig. 1c). These physical binding patterns were reciprocally confirmed by immunoprecipitation analysis using the anti-MILI and anti-MIWI antibodies (Fig. 1c). Thus, it is clear that a protein complex consisting of MILI, MIWI and TDRD1/MTR-1 exists in the adult testis. The other mouse PIWI family protein, MIWI2, was also bound to TDRD1/MTR-1 in 293T cells (Fig. S1 in Supporting Information).

Figure 1.

 TDRD1/MTR-1 binds to MILI and MIWI in adult testes. (a) Silver staining of the postnatal testis proteins that co-immunoprecipitated with the anti-MILI antibody. Testis lysates of day 16 Mili+/ and Mili/ mice were subjected to immunoprecipitation with the control normal rabbit IgG or the anti-MILI antibody, and SDS–PAGE was carried out. (b) Coomassie Brilliant Blue staining of the adult testis proteins co-immunoprecipitated with the anti-MILI antibody. Adult testis lysates of Miwi+/+ and Miwi/ mice were analyzed. The 130-kDa protein that binds specifically to MILI (black arrow) was examined by MALDI-TOF/MS. (c) Binding of MILI and MIWI to TDRD1/MTR-1 in adult testes. Immunoprecipitation (IP) of the adult testis lysates was carried out with the control rabbit IgG, anti-TDRD1/MTR-1, anti-MILI and anti-MIWI antibodies. The immunoprecipitated TDRD1/MTR-1, MILI and MIWI proteins were detected using the corresponding antibodies in Western blots. The intensities of the bands in Fig. 1c was analyzed by Image J program and the relative ratios were shown under each lane.

We investigated which part of MILI is necessary for binding to TDRD1/MTR-1 by immunoprecipitation assay and the analysis of the subcellular localization of the truncated mutants of MILI (Fig. S2 in Supporting Information). Our data suggests that PIWI domain of MILI is not required and PAZ domain is not sufficient for the interaction of MILI with TDRD1/MTR-1, and the data are well consistent with the recent report of Wang et al. (2009) and Reuter et al. (2009). The portion of TDRD1/MTR-1 necessary for binding to MILI and MIWI was similarly identified using immunoprecipitation assay (Fig. S3 in Supporting Information) and the analysis of the subcellular localization (Fig. 2). Although the full-length TDRD1/MTR-1 localized only to the cytoplasm (Fig. S4 in Supporting Information), TDRD1ΔC and TDRD1ΔN localized to both the nucleus and cytoplasm (Fig. 2b,c). When TDRD1ΔC was co-expressed with MILI and/or MIWI, its localization was restricted to the cytoplasm. The altered subcellular localization of TDRD1ΔC from nucleus to cytoplasm was due to binding to MILI or MIWI, both of which were located exclusively in the cytoplasm (Fig. S4 in Supporting Information). The localization of nuclear localized TDRD1ΔN to cytoplasm was similarly changed by co-expression with MIWI, although it was not significantly modified by co-expression with MILI (Fig. 2c). This difference may be due to the stronger binding of TDRD1ΔN to MIWI than to MILI (Fig. S3b in Supporting Information).

Figure 2.

 Association of TDRD1/MTR-1 deletion mutants with MILI and MIWI. (a) Schematic diagrams of His-tagged TDRD1/MTR-1 and deletion mutants thereof. The designation ‘T’ indicates the tudor domain. (b, c) Subcellular localizations of TDRD1ΔC and TDRD1ΔN, and alterations in localization caused by the co-expression of MILI and MIWI. The 293T cells were transfected with the plasmids that express TDRD1ΔC (b) or TDRD1ΔN (c), together with the plasmids that express Flag-tagged MILI or MIWI. TDRD1ΔC and TDRD1ΔN are stained with the Alexa Fluor 488-conjugated anti-mouse immunoglobulins (H+L) after binding with anti-His antibody (green color). MILI and MIWI are stained with the TRITC-conjugated anti-rabbit immunoglobulins (H+L) after binding with anti-MILI and anti-MIWI antibodies (red color) and DNA was visualized by DAPI.

Complex formation involving MILI, MIWI and TDRD1/MTR-1

Next, we examined the possibility that MILI, MIWI and TDRD1/MTR-1 form a protein complex. In co-immunoprecipitation experiments using the anti-MILI and MIWI antibodies, MILI and MIWI appeared to bind weakly to one another (Fig. 3a, lane 1), and TDRD1/MTR-1 co-expression significantly increased this association in 293T cells (Fig. 3a, lane 2). We confirmed that the expression levels of TDRD1/MTR-1 were essentially same when co-expressed with FLAG-MILI, FLAG-MIWI, or both (data not shown). To analyze the protein–protein interactions in more detail, we utilized MILI, MIWI and TDRD1/MTR-1 derivatives to which the nuclear localization signal (NLS) had been added. Alterations of these NLS-tagged proteins, which normally localize to the nucleus, were used as indicators of protein interactions. MILI-NLS was exclusively localized to the nucleus. However, when co-expressed with TDRD1/MTR-1, MILI-NLS localized only to the cytoplasm (Fig. 3b). In contrast, co-expression with MIWI did not have significant effects on the subcellular localization of MILI-NLS. MIWI-NLS was localized to both the cytoplasm and nucleus. In this case also, nuclear localization of MIWI-NLS was undetectable following the co-expression of TDRD1/MTR-1 but was not altered by the co-expression of MILI (Fig. 3c). These data suggest that TDRD1/MTR-1 binds to both MIWI and MILI, whereas the direct association between MIWI and MILI is weak.

Figure 3.

 Complex formation involving MILI, MIWI and TDRD1/MTR-1. (a) In vitro co-immunoprecipitation assays for MILI, MIWI and TDRD1/MTR-1. The 293T cells were transfected with the plasmids that express Flag-tagged MILI, Flag-tagged MIWI and His-tagged TDRD1/MTR-1. The lysates were immunoprecipitated with the anti-MILI or anti-MIWI antibodies, and subsequently subjected to Western blotting with the anti-Flag antibodies. (b–d) Subcellular localizations of TDRD1/MTR-1, MILI, and MIWI, and their NLS-fused mutants. The 293T cells were transfected with various combinations of the plasmids that express TDRD1/MTR-1, MILI, and MIWI, and their NLS-fused mutants. MILI-NLS, TDRD1/MTR-1, and MIWI stained with the anti-MILI, anti-His, and anti-Myc antibodies, respectively (b). MIWI-NLS, MILI, and TDRD1/MTR-1 stained with the anti-Myc, anti-MILI, and anti-TDRD1/MTR-1 antibodies, respectively (c). TDRD1-NLS, MILI-NLS, and MIWI stained with the anti-His, anti-MILI, and anti-MIWI antibodies, respectively (d). DNA is visualized by DAPI staining in all experiments.

TDRD1-NLS was localized to both the nucleus and cytoplasm (Fig. 3d). When TDRD1-NLS was co-expressed with MIWI, TDRD1-NLS was observed only in the cytoplasm. When MILI-NLS and TDRD1-NLS were expressed simultaneously, the subcellular localization of MILI-NLS, which was restricted to the nucleus alone, was similar to that of TDRD1-NLS (Fig. 3d). In addition, when MILI-NLS, MIWI, and TDRD1-NLS were expressed simultaneously, the subcellular localization of not only TDRD1-NLS but also MILI-NLS was restricted to the cytoplasm (Fig. 3d). These data strongly suggest that MILI, MIWI, and TDRD1/MTR-1 form a complex, and that the cytoplasmic localization of MIWI has a dominant effect on the subcellular localization of the complex in the cells. All the data from the subcellular localization experiments (Figs 2 and 3 and Fig. S2 in Supporting Information) are schematically summarized in Fig. S5 in Supporting Information.

Localization of MILI and TDRD1/MTR-1 in the MIWI-null mice

To examine whether the protein complex plays a role during spermatogenesis, we examined the expression and subcellular localizations of MILI and TDRD1/MTR-1 in the wild-type and the MIWI-null mice. There were no significant differences in the expression and subcellular localizations of MILI, TDRD1/MTR-1 and MVH between the control and MIWI-null spermatocytes (Fig. 4A). Although the expression levels of MILI and TDRD1/MTR-1 were not significantly altered, their subcellular localizations were quite different in the MIWI-null round spermatids, compared to that in the control cells (Fig. 4B and Fig. S6 in Supporting Information). In wild-type round spermatids, both MILI and TDRD1/MTR-1 were localized exclusively to the cytoplasm, and they were concentrated in the chromatoid bodies. In contrast, in the MIWI-null round spermatids, MILI and TDRD1/MTR-1 were localized not only to the cytoplasm but also to the nucleus. Immunostaining revealed that the localizations of these two proteins to the chromatoid bodies were less prominent, and that the chromatoid bodies were rather blurred. Electron microscopy confirmed the abnormal structures of the chromatoid bodies (Fig. 4C,d arrow). Although MVH has also been reported to bind both MIWI and MILI, its subcellular localization in the MIWI-null round spermatids did not differ from that in the wild-type cells.

Figure 4.

 Expression patterns and subcellular localizations of MILI, TDRD1/MTR-1, and MVH, and electron microscopy observations of the MIWI-null spermatocyte and round spermatid. (A, B) Immunohistochemical staining of the spermatocytes (A) and round spermatids (B) of Miwi+/+ and Miwi/ testes. Sections from Miwi+/+ (a, c, e and g), and Miwi/ (b, d, f and h) mouse testes were stained with the anti-MILI, anti-TDRD1/MTR-1, anti-MVH, and anti-MIWI antibodies, respectively. DNA was visualized by DAPI staining. (C) Electron microscopy observations of Miwi+/ and Miwi/ mice. Arrows indicate chromatoid body (c, d). Bar, 1 μm. (D) Schematic diagram of the data from Fig. 4B.

Discussion

Mechanisms of complex formation between MILI, MIWI and TDRD1/MTR-1

Immunoprecipitation of the adult testis proteins with the anti-MILI antibody and subsequent mass spec analysis revealed a physical association between MILI and TDRD1/MTR-1 (Fig. 1a,b). The binding between MILI and TDRD1/MTR-1 was also detected in the MIWI null testes by immunoprecipitation, which shows that the proteins bind under the physiological condition. Another mouse PIWI family member, MIWI, was also found to bind to TDRD1/MTR-1 (Fig. 1c). It seems likely that these associations involve direct bindings, as the experiments using 293T cells that expressed TDRD1/MTR-1 and MILI or MIWI confirmed the binding phenomena (Figs S2b and S3b in Supporting Information). Furthermore, the protein complex that contains MILI, MIWI and TDRD1/MTR-1 in round spermatids appears to be essential for spermatogenesis (Fig. 4).

TDRD1/MTR-1, which is one of the mouse tudor-related genes, contains repeats of the Tudor domain, which was originally identified in the Drosophila tudor protein (Boswell & Mahowald 1985; Chuma et al. 2003). Targeted mutation of TDRD1/MTR-1 resulted in the apoptosis of spermatocytes and round spermatids, and consequent infertility (Chuma et al. 2006). The deletion mutant analyses showed that both the N-terminally deleted protein (TDRD1ΔN), which includes only one tudor domain, and the C-terminally deleted protein (TDRD1ΔC) bound to MILI and MIWI, as judged from the immunoprecipitation assay with 293T cells (Fig. S3 in Supporting Information). In addition, the subcellular localization analyses revealed the binding of TDRD1ΔC to both MILI and MIWI (Fig. 2b). In contrast, TDRD1ΔN binding to MIWI was noted but TDRD1ΔN binding to MILI was not detectable in similar experiments (Fig. 2c). This discrepancy may be explained by stronger binding of TDRD1ΔN to MIWI than to MILI (Fig. S3 in Supporting Information).

As shown in Fig. 1c and Fig. S3b in Supporting Information, both MILI and MIWI bound to TDRD1/MTR-1. The results of the immunoprecipitation assays suggest that MILI and MIWI bind to each other, albeit very weakly (Fig. 3a, lane 1). However, co-expression of MIWI and MILI-NLS (Fig. 3b) and co-expression of MILI and MIWI-NLS (Fig. 3c) did not significantly alter the subcellular localizations of these proteins. These results suggest that there is little or no binding between MIWI and MILI without TDRD1/MTR-1 in cells, whereas there is binding between TDRD1/MTR-1 and the PIWI proteins. The additional expression of TDRD1/MTR-1 significantly enhanced the association between MIWI and MILI (Fig. 3a, lanes 1 and 2). This enhancement is probably caused by the simultaneous binding of MIWI and MILI to TDRD1/MTR-1 and the formation of a complex that consists of these proteins. However, we cannot exclude the possibility that the binding of TDRD1/MTR-1 to MIWI and MILI strengthens the weak binding between MIWI and MILI through some conformational change(s).

Chromatoid body formation and the protein complex

MILI and TDRD1/MTR-1 were localized to the cytoplasm of both the control and MIWI-null spermatocytes. Expression of MILI and TDRD1/MTR-1 in 293T cells showed that the proteins were localized in cytoplasm as well (Fig. 4A and Fig. S4 in Supporting Information). In contrast, nuclear localization of MILI and TDRD1/MTR-1 was detected in the MIWI-null round spermatids (Fig. 4B). Although neither MILI nor TDRD1/MTR-1 carries a distinctive NLS, they have the ability to be transported to the nucleus. However, in the presence of MIWI, they are retained in the cytoplasm, mainly in the chromatoid body, in a process that would be dependent upon binding with MIWI. These data suggest that the association to MIWI should have some inhibitory roles on the nuclear transport of MILI and TDRD1/MTR-1 in round spermatids.

In the TDRD1/MTR-1 deficient round spermatids, the subcellular localization of MILI was not altered (Fig. S7 in Supporting Information), which means that MILI cannot be transported to the nucleus by itself. It is conceivable that the nuclear transport of MILI is accompanied by binding to TDRD1/MTR-1 (Fig. 4D). The loss of nuclear TDRD1/MTR-1 caused by binding to MIWI appears to be sufficient to avoid the nuclear localization of MILI in the wild-type round spermatids. However, it is very likely that the formation of a protein complex with MIWI and TDRD1/MTR-1 in the chromatoid body facilitates the appropriate localization of MILI. In other words, complex formation among MIWI, TDRD1/MTR-1, and MILI, presumably efficient in chromatoid body, plays essential roles for the appropriate subcellular localization of the proteins. TDRD1/MTR-1 produces small granular foci in the cytoplasm of 293T cells. Although MILI and MIWI were diffusely localized in the cytoplasm of 293T cell, they mainly co-localized with TDRD1/MTR-1 and showed granular foci patterns when co-expressed with TDRD1/MTR-1. Essentially similar results were obtained in the co-expression with the TDRD1-NLS. The altered subcellular localization of MILI and MIWI showed that the proteins bound to TDRD1/MTR-1. Co-localization of the proteins would imply some roles of the protein complex formation on chromatoid body formation. Meanwhile, Chuma et al. reported that TDRD1/MTR-1 were co-localized with nuclear ribonucleoproteins (snRNPs) and produced speckles in NIH3T3 fibroblasts. We assume that TDRD1-NLS efficiently bound to snRNPs and produced evident speckles.

As previously reported, ultrastructural analyses showed that chromatoid body formation was impaired in the MIWI-null round spermatids (Fig. 4C; Kotaja et al. 2006b). Immunostaining showed that the localizations of MILI and TDRD1/MTR-1 were impaired (Fig. 4B). These data suggest that the formation of a protein complex that contains MILI, MIWI and TDRD1/MTR-1 is essential for the integrated subcellular localization of these proteins and for subsequent chromatoid body formation. We propose that certain protein(s) necessary for anchoring MIWI in the chromatoid body and cytoplasm are key molecule(s) for the correct subcellular localization of MIWI. In contrast, the subcellular localization of MVH was not changed by the null mutation in MIWI (Fig. 4B; Kotaja et al. 2006b). Therefore, although MVH also binds to MILI and MIWI, its localization is controlled by an unknown molecular mechanism.

Functional implications of the complex that contains MILI, MIWI and TDRD1/MTR-1

In the MILI-null mice, apoptosis occurred at the pachytene phase and, accordingly, no round spermatids developed. We have recently reported that MILI is essential for the de novo DNA methylation of retrotransposons at the fetal stage around embryonic day 17 (Kuramochi-Miyagawa et al. 2008). In that paper, we also reported that the composition of piRNA in the fetal testes was quite different to that in the adult testes after the pachytene phase, including the round spermatids. Correspondingly, two peaks of piRNA expression occur: (i) around the time of de novo DNA methylation in the fetus; and (ii) after the pachytene stage in the adult. The piRNA that is present after the pachytene stage presumably plays a role other than DNA methylation, as de novo DNA methylation does not take place at this stage. The function of MILI after the pachytene stage may be manifested by appropriate protein complex assembly and chromatoid body formation.

The MIWI-null mice exhibited male sterility due to spermatogenic arrest at the beginning of the round spermatid stage (Deng & Lin 2002). This phenotype is explained by the down-regulation of the mRNAs of the target genes of CREM, which is a master regulator of spermiogenesis, and that of ACT, which is an activator of CREM, all of which were co-immunoprecipitated with MIWI (Deng & Lin 2002). However, it remains to be determined whether these mRNAs bind directly to MIWI or via some other components of the chromatoid bodies. It is an intriguing question to consider whether impaired chromatoid body formation and inappropriate accumulations of MILI and TDRD1/MTR-1 play a role in the abnormal spermatogenesis of the MIWI-null mice. Further studies of protein complex formation in the chromatoid body should provide novel insights into the molecular mechanisms of spermatogenesis.

Experimental procedures

Plasmids

The Flag-tagged MILI, MIWI and MIWI2 expression plasmids were constructed by inserting the full-length and deletion-mutated fragments of MILI, MIWI and MIWI2 together with the Flag tag into the pcDNA4 vector (Invitrogen, Carlsbad, CA, USA). The MILIΔC and MILIΔN constructs encode amino acids (aa) 1–574 and 357–971 of MILI, respectively. The MILI-EGFP expression plasmids were constructed by inserting deletion-mutated MILI derivatives into the pEGFP-N1 vector (Clontech, Palo Alto, CA, USA). The MILIΔC-N and MILIΔC-C constructs encode aa 72–240 and 204–357 of MILI, respectively. The Flag-tagged MILI-NLS expression plasmid was constructed by ligating the NLS of MBD1 (KKRKKPSRPAKTRKR) at the C-terminus of MILI. The His-tagged, full-length TDRD1/MTR-1 and TDRD1ΔN expression plasmids were represented as pMTR-1 and pΔMTR1-T1, respectively, in a previous report (Chuma et al. 2003; Hosokawa et al. 2007), and the His-tagged TDRD1ΔC and His-tagged TDRD1ΔCΔN were constructed in the same vector. The TDRD1ΔC, TDRD1ΔCΔN and TDRD1ΔN plasmids encode aa 1–303, 304–888 and 889–1172 of TDRD1/MTR-1, respectively. The His-tagged TDRD1-NLS expression plasmid was constructed by ligating the NLS of SV40 (PKKKRKV) at the C-terminus of TDRD1/MTR-1. The Myc/His-tagged MIWI expression plasmids were constructed by inserting the full-length MIWI into the pcDNA4 vector (Invitrogen), as reported previously (Kuramochi-Miyagawa et al. 2001). The Myc/His-tagged MIWI-NLS expression plasmid was constructed by ligating the NLS of SV40 (PKKKRKV) at the C-terminus of MIWI.

Antibodies

The anti-MILI polyclonal antibody PM043 (MBL Co, Ltd, Nagoya, Japan) was used for immunohistochemical assays, and PM044 (MBL) was used for Western blotting and immunoprecipitation assays. PM043 and PM044 were generated by immunization with peptides derived from MILI (amino acids 2–19. DPVRPLFRGPTPVHPSQC) and (amino acids 107–122: VRKDREEPRSSLPDPS), respectively. The anti-MIWI-C polyclonal antibody (Kuramochi-Miyagawa et al. 2004), anti-TDRD1/MTR-1 polyclonal antibody (Chuma et al. 2003), anti-MVH polyclonal antibody (ab13840; Abcam, Tokyo, Japan), anti-Flag M2 antibody (Sigma Chemical Co., St Louis, MO, USA), anti-Myc antibody (9E10; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and anti-β-actin antibody (Sigma) were used for Western blotting, immunoprecipitation and immunohistochemical assays. The anti-His antibodies were used for Western blotting (Nacalai Tesque, Tokyo, Japan), immunoprecipitation (PM032; MBL) and immunohistochemistry (Calbiochem, San Diego, CA, USA). The anti-GFP antibodies were used for Western blotting (598, MBL) and immunoprecipitation (M048-3, MBL).

Immunoprecipitation, SDS–PAGE and CBB or silver staining

The adult testes of the control C57/Bl6, Mili+/, Mili/ (Kuramochi-Miyagawa et al. 2004) and Miwi/ (Deng & Lin 2002) mice were homogenized in lysis buffer A [20 mm HEPES (pH 7.5), 0.1% NP-40, 150 mm NaCl, 2.5 mm MgCl2, 1 mmdithiothreitol (DTT), protease inhibitor tablet (Roche, Basel, Switzerland)] (1 mL per 0.1 g testis tissue). The amount of lysate was used for 1 lane is 150 μL (for silver staining) and 250 μL (for CBB staining). The homogenate was centrifuged at 17 400 g for 10 min at 4 °C, and the supernatant was subjected to preclearing with anti-rabbit immunoglobulins (DAKO A/S, Glostrup, Denmark) (1 μg per 250 μL lysate) for 1 h at 4 °C and Protein G Sepharose 4 Fast Flow (GE Healthcare, UK) (10 μL per 250 μL lysate) for 1 h at 4 °C. After preclearing, the lysates were immunoprecipitated using the anti-MILI (PM044) antibody (1 μg per 250 μL lysate) over night at 4 °C and Protein G Sepharose 4 Fast Flow (10 μL per 250 μL lysate) for 1 h at 4 °C. Immunoprecipitates were subjected to SDS–PAGE in a 7.5% gel. The gel was stained with 0.25% CBB R-250 dissolved in methanol/acetic acid/water (5 : 1 : 5) for 0.5 h, and then destained in methanol/acetic acid/water (5 : 2 : 13) for 1 h. Silver staining was performed with Silver Quest (Invitrogen). Animal care was in accordance with the guidelines of Osaka University.

Mass spectrometry

Protein spots were excised from the CBB-stained gel. The gel fragments were destained with 50% ACN(CH3CN) in 25 mm AMBC(NH4HCO3) for 20 min at room temperature, washed with 50% ACN in 25 mm AMBC and vacuum dried. After treating with 10 mm DTT in 25 mm AMBC for 1 h at 56 °C and washing with 25 mm AMBC, the solution was replaced with 55 mm iodoacetamide in 25 mm AMBC. After washing with 50% ACN in 25 mm AMBC, the dried gel pieces were swollen on ice in 50 μL of digestion buffer that contained 10 μg/mL trypsin and 50 mm AMBC, and then incubated overnight at 37 °C. Tryptic-digested peptides were extracted with 50% ACN and 5% CF3COOH on ice and vacuum dried. The extracted peptides were analyzed by mass spectrometry.

Mass spectrometry was performed with the Ultraflex MALDI-TOF mass spectrometer (Bruker Daltonics, Billerica, MA, USA), which was equipped with a nitrogen laser operating at a frequency of 25 Hz. The accelerating voltage used was 25 kV and mass spectra were measured in the positive mode using the reflector mode. For sample preparation, the collected fractions were applied to AnchorChips and allowed to dry. Mass spectra were collected by summing the spectra from different regions of each spot and processed automatically using the MASCOT software (Matrix Science, London, UK). Spectra were externally calibrated using the prespotted calibration spots of the AnchorChips. A complete list of the peptide ion masses and their corresponding retention times received from each HPLC run was generated, using the NCBI database for identification.

Cell culturing, transient transfection and Western blotting

The 293T cells were cultured in Dulbecco’s modified Eagle’s medium that was supplemented with 10% FCS. 293T cells (106cells) were transfected with each plasmid using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s recommendations. At 27 h post-transfection, the cells were treated with lysis buffer A (Figs S1, S2b and S3 in Supporting Information) or lysis buffer B [10 mm Tris (pH 8.0), 150 mm NaCl, 1% Triton X-100, 1% sodium deoxycholate, 1% iodoacetamide and protease inhibitor tablet] (Fig. 3a and Fig. S2c in Supporting Information) (Costa et al. 2006) (500 μL per 106 293T cells). The lysates that were precleared using Protein G Sepharose (10 μL per 250 μL lysate) for 1 h at 4 °C were incubated with the anti-Flag, anti-His, anti-GFP, anti-MILI (PM044) or anti-MIWI-C antibodies (1 μg per 250 μL lysate) for 2 h at 4 °C. After the addition of Protein G Sepharose (10 μL per 250 μL lysate), the samples were incubated for 1 h at 4 °C, and then washed five times with the lysis buffer A (as for Figs S1, S2b and S3 in Supporting Information), or washed twice with 10 mm Tris (pH 8.0), 150 mm NaCl, and 0.1% Triton X-100, and twice with 10 mm Tris (pH 8.0) and 150 mm NaCl (as for Fig. 3a and Fig. S2c in Supporting Information). The immunoprecipitates were separated by SDS–PAGE and transferred to a PVDF membrane (Millipore, Bedford, MA, USA). After blocking, the filters were incubated with the appropriate antibodies. HRP-anti-mouse IgG (Pierce, Rockford, IL, USA) or HRP-anti-rabbit IgG (Zymed Laboratories, San Francisco, CA, USA) was used as the secondary antibody, and the signal was detected using the ECL Western Blotting detection reagents (GE Healthcare). The adult testis proteins lysed in buffer B (0.025 g testis tissue per 1 lane) were immunoprecipitated using the anti-MILI (PM044), anti-TDRD1/MTR-1 or anti-MIWI-C antibodies and separated by 7.5% SDS–PAGE, followed by Western blotting with the appropriate antibodies and HRP-anti-rabbit IgG (Zymed Laboratories) as the secondary antibody (Fig. 1c). The 26-day-old mice testes of Miwi+/ and Miwi/ (Deng & Lin 2002) mice were homogenized in lysis buffer A. The lysates (0.5 mg testis tissue per 1 lane) were separated by 7.5% SDS–PAGE, followed by Western blotting with the appropriate antibodies and HRP-anti-mouse IgG (Pierce) or HRP-anti-rabbit IgG (Zymed Laboratories) as the secondary antibody (Fig. S6 in Supporting Information).

Immunohistochemical staining

293T cells (0.4 × 106 cells) were seeded onto glass coverslips and transfected with the plasmids for 27 h, as described above. The cells were fixed in 3.7% formaldehyde in PBS for 8 h, permeabilized using 0.2% Triton X/PBS, and blocked with 10% normal goat serum and 3% BSA in PBS. Testes of the control C57/Bl6, MIWI-, and TDRD1/MTR-1-deficient (Chuma et al. 2006) mice were dissected and fixed in 4% paraformaldehyde for 2 h at 4 °C. After washing in PBS that contained 20% sucrose, the testes were embedded in OCT compound. Cryosections blocked with 10% normal goat serum and 3% BSA in PBS for 0.5 h at room temperature were used for immunofluorescence staining. After treatment of the primary antibodies over night at 4 °C, Alexa Fluor 488- or Alexa Fluor 568-conjugated anti-mouse immunoglobulins (H+L), Alexa Fluor 488-conjugated anti-rabbit immunoglobulins (H+L) (Molecular Probes, Eugene, OR, USA), Polyclonal swine anti-rabbit immunoglobulins/TRITC (DAKO) were used as the secondary antibodies for 1 h at room temperature. Nuclei were counterstained with 1 μg/mL DAPI. And the cells and cryosections were examined under a confocal microscope (LSM5Pascal; Carl Zeiss, Tokyo, Japan).

Electron microscopy

Testes (Miwi+/ and Miwi/) were fixed with 2% glutaraldehyde in 0.1 m phosphate buffer (pH 7.2), postfixed with 1% OsO4 and 0.1 m sucrose in 0.1 m phosphate buffer, dehydrated with a graded series of ethanol and embedded in epoxy resin (19). Semi-thin, 1-μm-thick sections were stained with 0.1% toluidine blue for light microscopy. Sections (70–90 nm in thickness) were placed on 150-mesh copper grids, stained with uranyl acetate followed by lead citrate, and examined under an electron microscope (H7000; Hitachi, Tokyo, Japan).

Acknowledgements

We thank Dr Haifan Lin for providing MIWI-deficient mice and fruitful discussion. We also thank Ms Mizokami for secretarial work. This work was supported in part by grants from the Japan Society for the Promotion of Science, the Ministry of Education, Culture, Sports, Science and Technology of Japan, the 21st Century COE Program ‘CICET’, and the Uehara Memorial Foundation, Japan.

Ancillary