Functional transformation of the chromatoid body in mouse spermatids requires testis-specific serine/threonine kinases

The cytoplasmic chromatoid body (CB) organizes mRNA metabolism and small regulatory RNA pathways, in relation to haploid gene expression, in mammalian round spermatids. However, little is known about functions and fate of the CB at later steps of spermatogenesis, when elongating spermatids undergo chromatin compaction and transcriptional silencing. In mouse elongating spermatids, we detected accumulation of the testis-specific serine/threonine kinases TSSK1 and TSSK2, and the substrate TSKS, in a ring-shaped structure around the base of the flagellum and in a cytoplasmic satellite, both corresponding to structures described to originate from the CB. At later steps of spermatid differentiation, the ring is found at the caudal end of the newly formed mitochondrial sheath. Targeted deletion of the tandemly arranged genes Tssk1 and Tssk2 in mouse resulted in male infertility, with loss of the CB-derived ring structure, and with elongating spermatids possessing a collapsed mitochondrial sheath. These results reveal TSSK1- and TSSK2-dependent functions of a transformed CB in post-meiotic cytodifferentiation of spermatids.


Introduction
Cytodifferentiation of mammalian spermatids, developing towards spermatozoa in the post-meiotic process named spermiogenesis, includes marked changes in the volume and structure of cytoplasm and organelles, in addition to reorganization of the nuclear chromatin. This is followed by the release of fully developed spermatids from Sertoli cells, referred to as spermiation, and the acquisition of fertilizing capacity during transit of the maturing spermatozoa through the epididymis (Clermont, 1972;Gardner, 1966). Transcriptional silencing of the haploid genome, by histoneto-protamine transition and chromatin compaction, precedes the final steps of spermiogenesis when ongoing protein synthesis depends on stability of mRNAs and developmental control of their translation (Kleene, 1993;Yang et al., 2005;Zhong et al., 1999).
Spermatogenesis is a very dynamic, but well-organized process. In mouse testis, each tubular cross-section contains a specific association of germ cells, with mitotic spermatogonia near the tubular wall, meiotic spermatocytes in the middle region of the spermatogenic epithelium, and the post-meiotic spermatids towards the lumen of the tubules. The specific associations, which follow each other in time, are indicated as stages I-XII of the spermatogenic cycle in mouse. Spermiogenesis in mouse takes nearly 14 days and is divided into 16 steps, with steps 1-8 round spermatids at stages I-VIII, start of elongation with step 9 spermatids at stage IX, and spermiation of step 16 spermatids at stage VIII (Oakberg, 1956).
Nuage (French for 'cloud') is an accumulation of dense fibrous material that occurs in the cytoplasm of germ line cells throughout the animal kingdom (Eddy, 1975). In spermatogenesis, nuage is first observed in spermatogonia, followed by the appearance of 'intermitochondrial cement' in spermatocytes in meiotic prophase (Yokota, 2008). Following the meiotic divisions, a prominent and fully developed chromatoid body (CB) represents a special form of nuage, bouncing around at the surface of the haploid nucleus of round spermatids (Parvinen and Jokelainen, 1974). Components of the CB body include proteins involved in mRNA metabolism and small regulatory RNA pathways, and the CB has been described as an RNA-processing centre, involved in control of mRNA stability and translation (Kotaja et al., 2006a;Oko et al., 1996;Toyooka et al., 2000;Tsai-Morris et al., 2004). One prominent component of the CB in mouse spermatids is MIWI, a testis-specific PIWI family member that associates with PIWI-interacting RNAs (piRNAs) (Grivna et al., 2006a). In the transition from round to elongating spermatids, the CB loses MIWI and other characteristic proteins, although MIWI is still found in the cytoplasm for some time (Grivna et al., 2006b). It seems that, in this transition, the CB loses its function as an RNA-processing centre, but recent research has not addressed the question of what happens to the CB next.
Electron microscopy (EM) studies by leading authors in the spermatogenesis field, published some 40 years ago, provide important background information about this (reviewed by Yokota, 2008). As described for various mammalian species (Fawcett et al., 1970), the CB migrates to the caudal pole of the nucleus of early elongating spermatids, where it forms a ring around the base of the developing flagellum. The ring migrates to the caudal end of the developing middle piece, moving in front of the mitochondria, which subsequently associate with the axoneme where they engage in mitochondrial sheath morphogenesis. This ring is closely associated with the annulus, a smaller and more compacted ringed barrier Functional transformation of the chromatoid body in mouse spermatids requires testis-specific serine/threonine kinases structure that bounds the middle piece, at the end of the mitochondrial sheath, where the principal piece begins. Studying rat spermiogenesis, Susi and Clermont reported that the CB first takes the form of an arc around the base of the flagellum (Susi and Clermont, 1970), in agreement with the ring described by Fawcett and colleagues (Fawcett et al., 1970), but that the bulk of the CB material then condenses into a dense sphere, which later migrates away from the nucleus and disintegrates by fragmentation. There is some disagreement between Fawcett and colleagues and Susi and Clermont concerning the emphasis on either the ring (arc) structure or the dense sphere, independent of an assignment of possible functions, which is still lacking. The position and behavior of the ring suggest a function of the structures originating from the CB in the cytodifferentiation of the middle piece.
The genes encoding the testis-specific kinases TSSK1 and TSSK2 are transcribed following completion of the meiotic divisions, and the proteins are cytoplasmic in elongating spermatids (Hao et al., 2004;Kueng et al., 1997). A protein substrate for both TSSK1 and TSSK2 has been identified (named TSKS for testisspecific kinase substrate) and is also present in the cytoplasm of elongating spermatids (Hao et al., 2004;Kueng et al., 1997;Scorilas et al., 2001). As described in this report, with newly generated antibodies we detected coinciding accumulation of TSSK1, TSSK2 and TSKS on a ring-shaped structure around the base of the flagellum and in a satellite in the cytoplasm of elongating spermatids. At later steps of spermatid differentiation, the ring, still marked by the presence of these three proteins, is found at the caudal end of the newly formed mitochondrial sheath. Furthermore, we discovered that loss of TSSK1 and TSSK2 in Tssk1/2 knockout mice (Tssk1 and Tssk2 double knockout) results in male infertility associated with production of abnormal spermatozoa. The most conspicuous abnormality concerns the mitochondrial sheath, which does not develop into a stable structure. Also, the CB-derived ring structure is prematurely lost in the knockout spermatids. From this, we propose that the CB, when it is transformed into ring and satellite structures in the transition from round spermatids to elongating spermatids, does not immediately enter a phase of functional decline in the wild type, but rather exerts important functions in the cytodifferentiation of spermatids.

Genes encoding TSSK1 and TSSK2
First, we composed an overview of Tssk genes in mouse and TSSK genes in human, based on literature data and NCBI and Ensembl Genome Browser BLAST results (supplementary material Table  S1). All genes are autosomal, and Tssk1, Tssk2 and Tssk6 have no introns, as do the human homologs. Tssk1 and Tssk2 are located in close proximity, within a region on mouse chromosome 16 that is syntenic to the human DiGeorge Syndrome region on chromosome 22q11.21, which harbors TSSK2 (Galili et al., 1997;Gong et al., 1996). The human TSSK2 gene is positioned next to the pseudogene TSSK1A that represents a mutated version of TSSK1 (Goldmuntz et al., 1997) (supplementary material Table S1). In human, TSSK1A appears to be functionally replaced, possibly by retrotransposition, by TSSK1B located on chromosome 5q22.2 (supplementary material Table S1). Other primates also have TSSK1B substituting for the pseudogene TSSK1A, as shown in a phylogenetic tree (supplementary material Fig. S1). Evolutionary maintenance of both TSSK1 (or TSSK1B in primates) and TSSK2 indicates that there is an added value for reproductive fitness in having two similar genes. This might be related to obtaining a sufficiently high level of kinase activity, although it cannot be excluded that there is some divergence in the roles of TSSK1 and TSSK2. From data assembled in the phylogenetic tree (supplementary material Fig. S1), and because we have not detected any orthologs in non-mammalian species, it appears that TSSK1 and TSSK2 might originate from early mammalian evolution.
With regard to growth and development, Tssk1/2 knockout mice were indistinguishable from heterozygous and wild-type animals. Also, the reproductive system of the knockout mice, at the anatomical level, appeared to be normal. There was no significant difference in body weight, testis weight and epididymis weight among the age-matched adult male wild-type, heterozygous and knockout mice (Table 1). Prolonged mating (2-4 weeks) of three Tssk1/2 knockout male mice with wild-type females showed normal mating behavior, confirmed by vaginal plug, but did not result in any pregnancies. The Tssk1/2 heterozygous males are fertile (Table  1), and the Tssk1/2 knockout females also have normal fertility. These females were crossed with the +/-males, yielding +/-and -/-mice at the expected Mendelian ratio (+/-: -/- 130:126). In knockout males, the number and motility of epididymal sperm was severely reduced, which explains the infertility, but the sperm characteristics of heterozygous males seemed largely unaffected (Table 1).
Histological analysis of testes from adult Tssk1/2 knockout mice using light microscopy at low magnification did not show gross abnormalities (Fig. 1A,B). However, there was a slight disorganization towards the end of spermatogenesis, indicating a spermiation defect whereby several condensed step 16 spermatids remained present at stage IX of the cycle of the spermatogenic epithelium in the knockout testis. In wild-type animals, all condensed step 16 spermatids were released from Sertoli cells at stage VIII (Fig. 1C,D).
Using an antibody targeting the endoplasmic reticulum-Golgi intermediate compartment (ERGIC) membrane protein ERGIC53/p58, we detected some dysregulation of the cytoplasmic Journal of Cell Science 123 (3) Weights and sperm counts are mean ± s.d. for 6-9 mice per genotype, estimated for adult 9-to 11-week-old mice.
*Sperm count is the total number of sperm heads isolated from two cauda epididymides. † Fertility is the number of pregnancies per breeding test. +/+, wild-type mice; +/-, Tssk1/2 heterozygous mice; -/-, knockout mice; nd, not determined. reorganization of elongating spermatids in Tssk1/2 knockout testis, leading to abundant cytoplasm still associated with late spermatids (Fig. 1E,F). Staining of spermatozoa from cauda epididymis clearly showed marked morphological abnormalities (Fig. 1G,H), in agreement with the immotility of the knockout sperm.
Thanks to the relatively mild testicular phenotype, with neither an early spermatogenic block nor a substantial loss of advanced steps in spermatogenesis, it was possible to look at loss of protein expression in the knockout. Western blot results for cytoplasmic fragments isolated from elongating spermatids showed that the expression of TSSK1 and TSSK2 is completely lost in the Tssk1/2 knockout, whereas the heterozygous samples contain a reduced level of the two proteins, compared to wild type (supplementary material Fig. S3A). Western blotting also showed that the expression of the testis-specific kinase substrate TSKS is maintained at wild-type level in the knockout (supplementary material Fig. S3B). TSKS has been identified as an interacting protein for TSSK1 and TSSK2 in yeast two-hybrid analysis, and both kinases can phosphorylate TSKS in vitro (Hao et al., 2004;Kueng et al., 1997). With an in vitro kinase assay, following immunoprecipitation using anti-TSKS antibody from testis homogenates, we showed that loss of TSSK1 and TSSK2 results in loss of [ 32 P] incorporation into TSKS (supplementary material Fig. S3C). This indicates that TSSK1 and TSSK2 are the main, if not the only, kinases phosphorylating TSKS in wild-type testis.
It has been reported that Tssk6 knockout males are infertile, with defects in spermatogenesis that might be related to a role of TSSK6 in post-meiotic chromatin remodeling (Spiridonov et al., 2005). TSSK6 is distantly related to TSSK1 and TSSK2 (supplementary material Fig. S1), but the Tssk6 gene follows the same temporal pattern of expression as that of Tssk1 and Tssk2 in mouse spermatids (Shima et al., 2004). Hence, it could not be excluded that dysregulation of spermatogenesis in the Tssk1/2 knockout is caused by loss of Tssk6 expression. However, in situ hybridization studies demonstrate a complete loss of Tssk1 and Tssk2 mRNAs, but maintenance of expression of Tssk6 mRNA, in the Tssk1/2 knockout (supplementary material Fig. S4). Clearly, the Tssk1/2 knockout phenotype is not explained by loss of Tssk6 gene expression. This result also nicely confirms that loss of Tssk1 and Tssk2 mRNAs and proteins in the Tssk1/2 knockout is not caused by loss of the spermatids expressing these genes. This was expected, on the basis of the relatively minor histological phenotype of the knockout testis described above.

Transformation of the chromatoid body to ring and satellite
Using immunohistochemical staining, we confirmed published data that the TSSK1, TSSK2 and TSKS proteins are present in the cytoplasm of early and late elongating spermatids (Kueng et al., 1997) (data not shown). To study this in more detail, we performed immunofluorescent analysis of the cellular localization. In crosssections of adult testis, all three proteins (TSSK1, TSSK2 and TSKS) show cytoplasmic localization in elongating spermatids towards the tubular lumen, with marked accumulation of the fluorescent signals at a conspicuous cytoplasmic focus ( Fig. 2A-C). The immunosignal for TSSK1 and TSSK2 is completely lost in the knockout (insets in Fig. 2A,B), also providing evidence for specificity of the antibodies. In the knockout, the substrate TSKS remains expressed at wild-type levels in the cytoplasm of elongating spermatids, but without the accumulation in the cytoplasmic focus (inset in Fig. 2C; supplementary material Fig. S3B).
At a higher magnification of the TSKS immunostaining, we detected a focus, which we refer to as a satellite, and a ring near the base of nucleus (Fig. 2D). For TSSK1 and TSSK2, identical immunostaining patterns were observed, which colocalized with the TSKS immunosignal (supplementary material Fig. S5A-F). At later steps of spermiogenesis, the ring gets smaller and is detected near the end of the middle piece (Fig. 2E). Further observations indicate that these ring and satellite structures have been described before, representing structures originating from the CB (Fawcett et al., 1970;Susi and Clermont, 1970). At the beginning of spermatid elongation, the CB was described as located close to the centriole, where a ring and a satellite structure are being formed but have not yet clearly separated (Fawcett et al., 1970). The staining in Fig. 2F, with anti-tubulin marking the centriole, is reminiscent of this situation. Somewhat later, when ring and satellite have developed into separate structures, neither of these structures colocalizes with the Golgi remnant, marked with anti-GM130 (Fig. 2G). The scheme in Fig. 3 indicates that the CB in early elongating spermatids has lost MIWI, but has gained TSSK1, TSSK2 and TSKS. This transformed CB is found as a ring around the tail next to the annulus and close to the centriole. Because the satellite also contains TSSK1, TSSK2 and TSKS, this indicates a common origin of ring and satellite.
The present findings might resolve the slight disagreement between Fawcett and colleagues (Fawcett et al., 1970) and Susi and Clermont (Susi and Clermont, 1970) regarding the emphasis on either the ring (arc) or the dense sphere as being the structure representing the CB in elongating spermatids. We suggest that the TSSK1/TSSK2/TSKS-positive ring and satellite represent both structures. In view of the overlapping accumulation of TSSK1, TSSK2 and TSKS, we propose that ring and satellite are separate but communicating structures.
With immunofluorescent staining, we confirmed published data (Grivna et al., 2006b;Kotaja et al., 2006b) that MIWI is expelled from the CB when step 8 round spermatids develop into early elongating step 9 spermatids (Fig. 4A,C). Other proteins associated with the RNA-processing activity of the CB also leave the CB around that developmental time-point (reviewed by Kotaja and Sassone-Corsi, 2007), meaning that the CB probably loses most, if not all, of its RNA-processing activity. In the Tssk1/2 knockout, MIWI is also lost from the CB at the same developmental time-point (Fig. 4B,D). In early elongating spermatids of the knockout, we detected some transient accumulation of TSKS on ring-and satellite-like structures (supplementary material Fig. S5G-I), which led us to conclude that the initial transformation of CB to ring and satellite might still occur to a limited extent. However, EM evaluation clearly showed that the CB-derived ring disappears prematurely in the Tssk1/2 knockout (Fig.  4F) compared to the wild type (Fig. 4E).

Mitochondrial sheath abnormalities in the Tssk1/2 knockout
Migration of the annulus to the caudal end of the middle piece occurs normally in the Tssk1/2 knockout (Fig. 4G,H). However, the mitochondria around the axoneme show a less regular and less compact appearance in knockout as compared with wild-type testicular spermatids (Fig. 4G,H). For knockout epididymal spermatozoa, the most conspicuous defect was detected using MitoTracker, and this concerns the mitochondrial sheath. This sheath, normally arranged as a regularly formed solenoid around the flagellum of the sperm cells (Fawcett, 1975;Ho and Wey, 2007;Otani et al., 1988) is severely disrupted, leaving one or two clusters of MitoTracker-stained mitochondria at the middle piece region (Fig.  5A,B).
Mitochondrial sheath morphogenesis was studied in more detail using confocal laser scanning microscopy of testicular spermatids stained with MitoTracker. For wild-type testis, we could observe how mitochondrial sheath morphogenesis proceeds from loosely arranged mitochondria surrounding the axoneme to formation of a compact sheath (Fig. 5C,E). In the Tssk1/2 knockout, the mitochondria stagger around the proximal part of the tail, but fail to form such a compact sheath. Rather, the mitochondria in Tssk1/2 knockout spermatids appear to collapse together in a few clusters Journal of Cell Science 123 (3)  The Golgi is associated with the growing acrosome. (B)In early elongating spermatids, the CB has lost MIWI but gained TSSK1, TSSK2 and TSKS, forming a ring around the tail, next to the annulus and close to the centriole. Because the satellite also contains TSSK1, TSSK2 and TSKS, this indicates a common origin of ring and satellite. The Golgi becomes disengaged from the fully developed acrosome. The mitochondria are scattered throughout the cytoplasm. (C)At later steps of spermatid elongation, the ring has become smaller and has moved down the tail with the annulus. In its slipstream, the mitochondria become associated with the axoneme. The centriole degrades, and the satellite and Golgi remnant also become less prominent. This step of spermatid elongation is followed by reduction of the cytoplasmic volume and formation of the residual body (not illustrated). (Fig. 5D,F), which are also observed with EM analysis (data not shown). We conclude that the disruption of the mitochondrial sheath as observed in epididymal Tssk1/2 knockout sperm results from incomplete assembly. As a corollary, we propose that when the CB is transformed into a ring, which moves down the tail of the middle piece area (Fig. 3), it has a function in the assembly of the mitochondrial sheath. In this process, maintenance and activity of the ring requires TSSK1 and TSSK2.

Discussion
The role of the CB in round spermatids as an RNA-processing centre has been highlighted in many investigations (Kotaja et al., 2006a;Oko et al., 1996;Toyooka et al., 2000;Tsai-Morris et al., 2004). However, it is probable that material composing the CB serves other functions as well. The CB in round spermatids is preceded by intermitochondrial cement in spermatocytes (Chuma et al., 2006;Chuma et al., 2009;Yokota, 2008), and is succeeded by the ring and satellite structures in elongating spermatids (Fawcett et al., 1970;Susi and Clermont, 1970) (this work). This developmental series of nuage during spermatogenesis might involve different types of nuage, as suggested by Pan and co-workers, who reasoned that RNF17 (a protein containing both RING finger and tudor domains) is a component of a novel form of germ cell nuage (Pan et al., 2005). Targeted mutation of Tdrd1, a mouse homolog of the (A-D)Immunofluorescent staining of MIWI in CBs of round spermatids at step 8 (green dots) disappears when the spermatids reach step 9, both in the wild type (A,C) and in Tssk1/2 knockout testis (B,D). Cytoplasmic staining of MIWI is observed in spermatocytes. (E-H)Cross-section of the CB ring (open arrowhead) associated with the annulus (closed arrowhead) in wild-type early elongating spermatids (E). This CB ring material is absent in Tssk1/2 knockout spermatids (F). At later steps of elongation, the annulus is present at the border of the middle piece and principal piece, both in the wild type (G) and knockout (H). The insets in G,H show that some electron-dense material remains associated with the annulus in wild-type late spermatids, but not in the Tssk1/2 knockout. +/+, wild type; -/-, knockout. Scale bars: 20m (A-D), 500 nm (E,F), 1000 nm (G,H). The mitochondrial sheath is arranged as a regularly formed solenoid around the flagellum of the wild-type sperm (arrowheads in A). This arrangement is severely disrupted in the Tssk1/2 knockout, where we find one or two droplets of MitoTracker-stained mitochondria at the middle piece region (arrowheads in B). The nuclei are stained with DAPI (blue), and the acrosome is marked by monoclonal antibody 18.6 (red). (C-F)Confocal microscopy of testicular elongating spermatids stained with MitoTracker (green). In wild-type testis, the mitochondria are first loosely arranged (C) and then assemble into a compact sheath (arrowheads in E). In the Tssk1/2 knockout, the mitochondria stagger around the proximal part of the tail (encircled with dashed line in D), but fail to form a stable sheath (arrowheads in F). +/+, wild type; -/-, knockout. Scale bars: 10m (A,B E,F), 5m (C,D).
Drosophila maternal effect gene tudor, encoding the nuage component TDRD1, results in a reduction of intermitochondrial cement, whereas the CB is structurally maintained. This leads to the suggestion that CBs probably have an origin independent of, or additional to, intermitochondrial cement (Chuma et al., 2006). The different forms of nuage might contain a number of proteins that form a 'platform' throughout spermatogenesis, on which specialized forms of nuage with differential functions are assembled at subsequent steps of germ line development. In mice lacking the tudor-related gene Tdrd6, the spermatids carry 'ghost' CBs (Vasileva et al., 2009). The CB proteins MAEL, MIWI and MVH do not localize to these ghost CBs, meaning that the function of the CB as RNA-processing centre is lost. However, the ghost CBs are still structural entities and might contain other ongoing activities. Whereas the role as an RNA-processing centre reaches its peak activity in the CBs of round spermatids, other roles of nuage in spermatogenesis can be exerted at earlier or later steps. A highly important early function is exerted by the nuage components MIWI2 and MAEL, which appear to be indispensable for repression of transposons in meiotic prophase (Carmell et al., 2007;Soper et al., 2008). An entirely different putative function was described by Haraguchi and colleagues, who detected the presence of so-called aggresomal markers in CBs of rat spermatids, including chaperones and proteins of the ubiquitin-proteasome pathway (Haraguchi et al., 2005). Possibly, at subsequent steps throughout spermatogenesis, the series of nuage and intermitochondrial cement in spermatogonia and spermatocytes, CBs in round spermatids, and ring and satellite in elongating spermatids is involved in control of post-translational events that affect protein processing, modification, sorting and degradation.
In describing the CB in elongating spermatids at the EM level, Fawcett and colleagues (Fawcett et al., 1970) noted, 'it is tempting to conjecture that there might be some interaction between the CB and the mitochondria in the process of middle-piece formation'. The annulus and the ring-shaped CB move distally along the flagellum, and behind it the mitochondria become associated with the axoneme. The ring and satellite might be involved in posttranslational modification of proteins that are required, or need to be removed, in order for the mitochondria to assemble at the axoneme and to undergo maturation and morphogenesis towards development of the stably compacted mitochondrial sheath. A role for ring and satellite in mitochondrial sheath formation might also be relatively indirect, for example by interplay of molecular pathways enclosed in ring and satellite with other cytoplasmic systems, such as the endoplasmic reticulum. In mice deficient in GOPC, a protein that plays a role in vesicle transport from the Golgi apparatus, there is weak adhesion of sperm mitochondria (Suzuki-Toyota et al., 2007), in a fashion somewhat similar to what is observed in the Tssk1/2 knockout. It is important to note that CB proteins are also found free in the cytoplasm, in addition to their accumulation in the CB. For example, MIWI is found throughout the cytoplasm in round spermatids, where it associates with the translational machinery and piRNAs (Grivna et al., 2006a), and remains present in the cytoplasm also when MIWI is lost from the CB at the beginning of spermatid elongation (Grivna et al., 2006b). Similarly, when TSSK1, TSSK2 and TSKS accumulate in the ring and satellite in elongating spermatids, a substantial immunostaining of the surrounding cytoplasm remains present. Hence, these proteins might serve functions also outside the ring and satellite.
When spermatids elongate, the annulus migrates down the developing tail, at the leading edge of the cytoplasm that exvaginates the middle-piece area of the axoneme. This occurs also in the Tssk1/2 knockout spermatids. The annulus, a ringed barrier structure between the middle piece and principal piece of the sperm cell, is composed of septin proteins that are known to generate ringed barrier structures in other cells. The annulus is not formed in Sept4 knockout mice (Ihara et al., 2005;Kissel et al., 2005). As a result, sperm generated in the absence of SEPT4 have a defective mitochondrial architecture, with mitochondrial size heterogeneity and a more open structure of the mitochondria with less cristae (Kissel et al., 2005). This appears to be somewhat similar to aspects of the Tssk1/2 knockout sperm. Distal migration of the ring-shaped CB might require distal migration of the annulus. In fact, such a dynamic association is quite likely, in view of the structural proximity of the CB ring and the annulus. In the Sept4 knockout spermatids, in the absence of an annulus migrating distally, the ringshaped CB might not reach the distal end of the middle piece, so that its molecular activities are not able to cover the whole middlepiece region, leading to incomplete maturation of the sperm mitochondria. It is expected that the nuage material of ring and satellite will remain present in elongating spermatids in the Sept4 knockout, albeit at a more random localization within the cytoplasm, and might still exert some of its activities. This would be in agreement with the less severe impairment of mitochondrial sheath formation in Sept4 knockout sperm as compared to Tssk1/2 knockout sperm.
We detected abundant cytoplasm associated with Tssk1/2 knockout late spermatids, using the ERGIC53/p58 antibody. Normally, in wild-type testis, reduction of the cytoplasmic volume of elongating spermatids finally leads to formation of the residual body, which is lost from the spermatids at spermiation (Sakai and Yamashina, 1989). In the mechanism of cytoplasmic volume reduction, tubulobulbar complexes of spermatids invaginating into surrounding Sertoli cells are probably involved (Russell, 1979;Sprando and Russell, 1987). The observed relative lack of reduction in cytoplasmic volume of elongating spermatids in the Tssk1/2 knockout is not explained, but the results indicate that TSSK1 and TSSK2 might exert multiple functions involving different aspects of the cytodifferentiation of spermatids.
Spermatids are functionally diploid for most proteins by sharing mRNAs and/or proteins through the intercellular bridges by which the developing germ cells form a functional syncytium (Braun et al., 1989;Fawcett et al., 1959;Morales et al., 2002). In Tssk1/2 heterozygous testis, with reduced levels of the two proteins, we do not see loss of TSSK1 and TSSK2 immunostaining in half of the spermatids (not shown). Hence, the spermatids are functionally diploid for TSSK1 and TSSK2, which also explains transmission of the targeted Tssk1/2 allele by the haploid spermatoza when breeding the heterozygous males. Reduced levels of TSSK1 and TSSK2, such as found in the Tssk1/2 heterozygous situation, might become problematic under stressful and selective living conditions, which could result in infertility or subfertility of the heterozygous animals, but there is no apparent haploinsufficiency. During the course of our studies, Xu and co-workers reported that targeted deletion of Tssk1/2 causes male infertility due to haploinsuffiency; knockout mice were not obtained in their study (Xu et al., 2008). This might be explained by differences in the genetic background of the mouse strains used (Threadgill et al., 1997). In studies that aimed to generate mouse models of DiGeorge Syndrome, 150-550 kb deletions of chromosome 16 including Tssk1 and Tssk2 were found to be lethal in a homozygous situation, but were transmitted by the heterozygous animals (Kimber et al., 1999;Puech et al., 2000). This is in agreement with our findings on transmission of the targeted Tssk1/2 allele.
In summary, we conclude that TSSK1 and TSSK2 play crucially important roles in transformation of the CBs in round spermatids to another form of nuage with essential functions in cytodifferentiation of late spermatids. The nuage in elongating spermatids takes the shape of a ring and satellite, which we propose represent the CB structures that were identified in 1970 by Fawcett and colleagues and Susi and Clermont (Fawcett et al., 1970;Susi and Clermont, 1970). Perhaps it is a matter of coincidence that the nuage in primary spermatocytes is named intermitochondrial cement, and the ring and satellite in elongating spermatids might exert functions in mitochondrial attachment and formation of a stably compacted mitochondrial sheath. However, there might be a real functional link between nuage and mitochondria at subsequent steps of spermatogenesis, as also suggested by Chuma and coworkers (Chuma et al., 2009). TSSK1 and TSSK2, and possibly other factors specific for elongating spermatids, appear to be required for performance of the final act of male germ line nuage in cytodifferentiation of spermatids.

Generation of Tssk1/2 knockout mice
The Tssk1 and Tssk2 genes are located in close proximity to each other, separated by 3.06-kb intergenic sequence. We decided to generate a Tssk1 and Tssk2 doubleknockout, referred to as Tssk1/2 knockout (Tssk1/2 -/-). The lambda knockout shuttle (KOS) system was used to build a Tssk1/2 targeting vector (Wattler et al., 1999), by Lexicon Genetics (Lexicon Pharmaceuticals, Princeton, NJ). The targeting vector was derived from one clone representing the mouse Tssk1/Tssk2 locus, which contains a Neo selection cassette at the 3Ј end of Tssk2. The entire Tssk1/Tssk2/Neo sequence was flanked by LoxP sites. The vector was linearized with NotI for electroporation into strain 129/SvEv brd (LEX1) embryonic stem (ES) cells. G418/FIAU-resistant ES cell clones were isolated and the correct targeting was confirmed by Southern blot. One targeted ES cell clone (Tssk1/2 LoxP/+ ) was injected into blastocysts of strain C57BL/6J-Tyr<c-2J> (albino C57BL/6). The injected blastocysts were transplanted into pseudopregnant female mice. Resulting chimeric male mice were mated to albino C57BL/6 females to generate offspring with germline transmission. Male F2 heterozygotes (Tssk1/2 LoxP/+ ) were generated from a backcross of F1 heterozygotes to wild-type hybrids with 129/SvEv brd and C57BL/6 background, and were provided to us by Lexicon Genetics. Heterozygotes (Tssk1/2 LoxP/+ ) were then crossed with a CMV-Cre deleter mouse, to obtain the knockout allele by removing the region in between the LoxP sites. Mice heterozygous for the Tssk1/2 knockout allele (Tssk1/2 +/-) were inbred to obtain Tssk1/2 -/knockout mice. The Tssk1/2 mutation was backcrossed to the C57BL/6 genetic background for eight generations. Animals were housed at the Erasmus MC Laboratory Animal Science Center (EDC), and the studies were subject to review by an independent Animal Ethics Committee (Stichting DEC Consult, The Netherlands).

Evaluation of fertility and sperm
Individual 8-week-old males were caged with three wild-type adult female C57BL/6 mice. Natural mating was confirmed by monitoring vaginal plugs during the first 3 days of mating. A prolonged mating (for 2-4 weeks) was performed for the knockout males. The females were monitored for pregnancies, and, if any, litter sizes. Spermatozoa were isolated from cauda epididymides and used for sperm counting and mobility evaluation. Cauda epididymides were dissected from adult wild-type, Tssk1/2 +/and Tssk1/2 -/mice, and gently torn apart while immerged in EKRB buffer (Bellve et al., 1977) at room temperature. Wild-type and heterozygous spermatozoa swam out of the epididymides, while immotile knockout sperm was released by gently squeezing the tissue. Motility evaluation of at least 200 spermatozoa was assessed by means of light microscopy (BX41, Olympus, JP), and spermatozoa were counted using a hemocytometer.

Histology
Dissected testes, and sperm from cauda epididymides smeared onto slides, were fixed in Bouin's fixative overnight at room temperature. Testis 5 m paraffin sections were stained with periodic acid Schiff (PAS)-hematoxylin, and sperm were stained with hematoxylin-eosin (HE).

Western blotting
Protein samples were separated on SDS-PAGE, then blotted to nitrocellulose membrane (Whatman) at 300 mA for 1.5 hours at 4°C. After transfer, the membrane was blocked for 30 minutes with blocking buffer containing 5% w/v skimmed milk and 1% w/v BSA in PBS-Tween (0.1% v/v Tween-20 in PBS). Then, the blot was incubated with primary antibodies in blocking buffer for 1.5 hours. Following two washings with PBS-Tween, the blot was incubated with the peroxidase-conjugated secondary antibody in blocking buffer for 45 minutes. The resultant interaction was detected by Western Lightning chemiluminescence reagent (PerkinElmer).

Immunoprecipitation and phosphorylation assay
Decapsulated testis tissue was homogenized in ice-cold immunoprecipitation buffer containing 20 mM Tris pH 7.4, 1% v/v NP40, 150 mM NaCl, 0.2 mM orthovanadate (Sigma), 1 mM EDTA, 0.2 mM dithiothreitol (Sigma), and one protease inhibitor cocktail tablet per 50 ml solution (Roche). For each sample, 200 l precleared lysate was incubated with 2 g anti-TSKS antibody for 4 hours at 4°C with agitation. The antigen-antibody complexes were isolated using protein A Sepharose beads (GE Healthcare) by incubation at 4°C overnight. The antigen-antibody bead complexes were resuspended in SDS sample buffer, and subjected to western blotting as described above. In vitro kinase assays were performed on these complexes, essentially as described (Kueng et al., 1997).

Immunohistochemistry and immunofluorescence microscopy
For immunostaining of testis, the tissue was fixed in 4% w/v paraformaldehyde, and 5 m paraffin sections were made. Rabbit anti-TSSK1, anti-TSSK2 and anti-TSKS antibodies were used at 1:2000, 1:500 and 1:1000 dilutions, respectively. Guinea pig anti-TSSK1, anti-TSSK2 and anti-TSKS were all used at 1:1000 dilution. Mouse monoclonal antibodies against -tubulin and GM130 (Sigma) were used at 1:1000 and 1:500 dilutions, respectively. MIWI antibody (Cell Signaling) was used at 1:200 dilution. Immunohistochemistry was performed as described previously (Roest et al., 1996). For the immunofluorescence, we adapted the method described by Tsuneoka et al. (Tsuneoka et al., 2006). Epididymal spermatozoa were isolated as described above, and, if applicable, MitoTracker (Invitrogen) was added to a final concentration of 20 nM. After 20 minutes incubation at room temperature, the sperm was smeared on slides, air-dried in the dark for one hour at room temperature, and then fixed in 100% methanol for 5 minutes. Following permeabilization with 1% v/v Triton X-100 in PBS at 37°C for 15 minutes, the slides were blocked with 10% v/v NGS and 2% w/v BSA at 37°C for 1 hour. Mouse monoclonal anti-acrosome antibody 18.6 (kindly provided by Harry D. Moore, The University of Sheffield, Sheffield, UK) was used on the blocked slides without dilution. DAPI-containing mounting medium (Invitrogen) was used to label nuclei. Images were taken using an Axioplan 2 (Carl Zeiss) equipped with a CoolSNAP-Pro color charge-coupled device camera (Media Cybernetics, Wokingham, UK). For immunofluorescent staining, the following secondary antibodies were used: goat anti mouse IgG FITC 1:128 and TRITC 1:128 (Sigma); goat anti rabbit IgG FITC 1:80 and TRITC 1:200 (Sigma); goat anti guinea pig IgG FITC 1:200 and TRITC 1:200 (Invitrogen).

Confocal microscopy of testicular spermatids
To prevent mechanical damage of elongating spermatids, these cells were gently expelled to the outside of tubule fragments by enzymatic manipulation of the testis. The tunica albuginea was removed, and the testis tissue was treated with enzymes (PBS with Ca 2+ and Mg 2+ , 12 mM lactate, 1 mg/ml glucose, 1 mg/ml collagenase, 0.5 mg/ml hyaluronidase) on a horizontal shaker at 34°C for 20 minutes. The obtained tubule fragments were washed, and then incubated with 20 nM MitoTracker-Green (Invitrogen) in PBS at room temperature for 15 minutes. The tubule fragments were imaged using a Zeiss LSM510NLO confocal laser scanning microscope (Carl Zeiss) with a 63ϫ 1.40 NA oil immersion lens.

RNA in situ hybridization
Digoxygenin-rUTP-labeled sense and antisense RNAs transcribed from Tssk1 and Tssk2 cDNA fragments corresponding to the amino acid residues 269-365 for TSSK1 and 268-358 for TSSK2 were used as probes for in situ hybridization on 5 m testis cross-sections. For Tssk6, the cDNA fragment contains a region encoding the amino acid residues 248-273 plus 366 bp of the 5Ј UTR. Hybridization was carried out as previously described (Wilkinson and Nieto, 1993).

Transmission electron microscopy
Testes were dissected, and fixed with 4% v/v formaldehyde and 1% v/v glutaraldehyde in PBS. After post-fixation with 1% w/v osmium tetroxide and dehydration with gradient acetone in Leica EM TP (Leica), testis tissue was embedded in epoxy resin LX-112, and uranylacetate-and lead nitrate-contrasted ultrathin sections (0.04 m) were studied using a transmission electron microscope (Morgagni Model 208S; Philips, NL) at 80 kV.
We are thankful to Esther Sleddens-Linkels, Mark Wijgerde, and Leendert Looijenga for helpful comments and advice. Harry Moore is gratefully acknowledged for providing us with the acrosome-specific antibody 18.6.