|Home | About | Journals | Submit | Contact Us | Français|
Nuclear shaping is a critical event during sperm development as demonstrated by the incidence of male infertility associated with abnormal sperm ad shaping. Herein, we demonstrate that mouse and rat spermatids assemble in the subacrosomal space a cytoskeletal scaffold containing F-actin and Sak57, a keratin ortholog. The cytoskeletal plate, designated acroplaxome, anchors the developing acrosome to the nuclear envelope. The acroplaxome consists of a marginal ring containing keratin 5 10-nm-thick filaments and F-actin. The ring is closely associated with the leading edge of the acrosome and to the nuclear envelope during the elongation of the spermatid head. Anchorage of the acroplaxome to the gradually shaping nucleus is not disrupted by hypotonic treatment and brief Triton X-100 extraction. By examining spermiogenesis in the azh mutant mouse, characterized by abnormal spermatid/sperm head shaping, we have determined that a deformity of the spermatid nucleus is restricted to the acroplaxome region. These findings lead to the suggestion that the acroplaxome nucleates an F-actin–keratin-containing assembly with the purpose of stabilizing and anchoring the developing acrosome during spermatid nuclear elongation. The acroplaxome may also provide a mechanical planar scaffold modulating external clutching forces generated by a stack of Sertoli cell F-actin–containing hoops encircling the elongating spermatid nucleus.
Spermiogenesis is a prolonged cell differentiation process in which the male gamete acquires a propelling flagellum required for delivering the haploid genome to the receptive egg at the time of fertilization. Two major events of spermiogenesis are the development of the acrosome and the shaping and condensation of the future sperm nucleus from a spherical configuration into an elongated structure. The acrosome initiates its development when transporting vesicles derived from the Golgi apparatus fuse to form a single acrosomal sac associated with the nuclear envelope. As the development of the acrosome is in progress, the transient microtubule-containing manchette becomes visible when the spermatid nucleus initiates its elongation. The manchette disassembles upon completion of nuclear elongation and condensation (reviewed by Clermont et al., 1993 ). This developmental correlation has been interpreted as indicative of a possible involvement of the manchette in spermatid nuclear shaping (reviewed by Meistrich, 1993 ). However, recent experimental evidence supports an intramanchette transport mechanism for the delivery of molecules to the centrosome and developing spermatid tail (reviewed by Kierszenbaum, 2002 ). Molecular aspects of the possible contribution of the acrosome to spermatid nuclear shaping are beginning to emerge. A lack of acrosome formation in Hrb null mutant mice (Kang-Decker et al., 2001 ) and GOPC-deficient mice (Yao et al., 2002 ) is associated with globozoospermia (round-headed sperm) and infertility. Hrb is associated with the cytosolic surface of proacrosomic transporting vesicles and a lack of Hrb prevents the vesicles from fusing to form the large acrosomal sac. GOPC is localized in the trans-Golgi region in round spermatids, and a lack of GOPC precludes vesicle transport from the Golgi apparatus to the acrosomal sac. The developing acrosome embraces the apical one-third of the spermatid nucleus and its configuration gradually follows the shaping of the spermatid nucleus. Because of this relationship, we have considered the possibility of an anchoring mechanism, which prevents the dislocation of the acrosome sac and also modulates extrinsic and/or intrinsic forces operating during nuclear shaping in mammalian spermatids, including human.
Previous work has shown that spermatogenic cell/sperm-associated keratin of molecular mass 57 kDa (Sak57) provides a cortical scaffold to primary spermatocytes, is present in the core of intercellular bridges connecting members of a spermatocyte cohort (Tres et al., 1996 ), and is associated with microtubules of the manchette before becoming a component of the outer dense fibers surrounding the axoneme of the developing spermatid tail (Tres and Kierszenbaum, 1996 ). Based on these observations, we have searched for additional sites in which Sak57 may be present during spermiogenesis. Herein, we report the structural characterization of a plate present in the subacrosomal space and linking the developing acrosome to the spermatid nucleus. This plate, which we designate acroplaxome (from the Greek words akros, topmost; platys, flat; sõma, body), consists of F-actin and Sak57 and is bordered by a marginal ring. The marginal ring is made up of Sak57-containing intermediate filaments inserted into a plaque associated with the leading edge of the inner acrosomal membrane. To identify the molecular nature of Sak57, we have used cDNA cloning. We show that Sak57 is a rat ortholog to human keratin 5 (K5) present in the suprabasal layers of the epidermis. This finding led us to change the designation Sak57 to K5.
Mass spectrometry analysis of high-performance liquid chromatography-fractionated Lys-C–digested Sak57 peptides from rat sperm tail indicated that two fragments displayed homology with the α-helical rod domain (1A and 2A regions) of human, rat, and mouse keratins (for details, see Kierszenbaum et al., 1996 ). Reverse translation of one of the sequences (KAQYEDIAQK) by using mammalian-specific frequency of nucleotide usage resulted in the DNA sequence, which was used to design nested primers (R, TTCTGGGCAATGTCCTC, and T, TCCTCGTACTGGGCCTT). We started cloning of Sak57 cDNA from a rat testis λgt11 cDNA library (BD Biosciences Clontech, Palo Alto, CA) by using nested polymerase chain reaction (PCR) with primers gt11-5′ and R followed by gt11-5′ (GACTCCTGGAGCCCG) and T. Sequencing of an obtained fragment revealed near identity to the mRNA for human keratin 5 (K5). Human K5-specific nested primers spanning the entire coding region were synthesized and used to clone rat testis K5 cDNA.
Expression vector pET-Blue2 (Novagen, Madison, WI) was used for directional cloning of a K5 cDNA fragment (encoding amino acids 166–462) into SalI-XhoI sites followed by transformation of competent NovaBlue DE3 cells (Novagen). Transformed cells were grown in the presence or absence (control) of 1 mM isopropyl β-d-thiogalactoside (IPTG). Proteins were extracted in extraction buffer (8 M urea, 1% SDS, 10 mM Na phosphate, 0.1 M dithiothreitol) and analyzed by immunoblotting. Purification of K5 fusion protein, which contains (His)6 tail at the carboxy terminus, was carried out using His.Bind Quick 900 cartridge and His.Bind buffer kit (Novagen) according to the manufacturer's protocols. Eluted protein samples were concentrated using a Centricon centrifugal filter (Millipore, Bedford, MA).
A rat pachytene cDNA expression library was used to determine the spermatogenic cell-specific expression of K5. Complete details of the construction of the rat pachytene cDNA expression library have been reported previously (Rivkin et al., 1997 ).
Polyclonal antibodies generated against the peptide KAQYEDIAQK (corresponding to a segment of the 2A region of the α-helical rod domain; Figure 1) and LEGQECRLSGEGVG (corresponding to the beginning of the tail domain) were prepared and characterized as described previously (Kierszenbaum et al., 1996 ). For affinity purification, the antigenic peptide was immobilized to an AminoLink coupling gel column (AminoLink kit; Pierce Chemical, Rockford, IL) according to the manufacturer's protocol. Binding to and elution of specific immunoglobulin from the affinity column was carried out using ImmunoPure Gentle Ag/antibody buffers (Pierce Chemical). The A280 protein peak fractions were used for immunocytochemistry and immunoblotting (see below). For competition analysis of antiserum specificity, 3 mg of KAQYEDIAQK peptide or an unrelated peptide was added to 0.1-ml aliquots of affinity-purified anti-K5 serum and incubated overnight at 4°C. These samples were diluted 1:150 and checked for recognition of the K5 fusion protein on immunoblots.
The following samples were analyzed: rat and mouse testis, isolated rat spermatogenic cells, and human (plantar) and rat skin (foodtpad and back). Human skin samples were from autopsy specimens, and their use was approved by the Human Subjects Committee (assurance number M-1111-XM-4XM; protocol number H-0021). Seminiferous tubular fragments (identified with a dissecting stereomicroscope as corresponding to stage I-XIV of rat spermatogenesis according to their transillumination pattern; Mochida et al., 1999 and references therein for additional details) were placed in a drop of 3.7% paraformaldehyde (electron microscopy grade) in 0.1 M sucrose in phosphate buffer, pH 7.4, on microscope slides coated with Vectabond (Vector Laboratories, Burlingame, CA). After 15-min fixation at room temperatures, spermatogenic cells were gently extruded from the seminiferous tubule by placing a coverglass on top of the preparation. The glass coverslip was removed and used for immunocytochemistry (see below). In some experiments, cells were extracted with 0.5% Triton X-100 in 0.1 M sucrose for 30 s and then fixed for 15 min in paraformaldehyde as indicated above. Cells were immunoreacted with affinity-purified K5 sera alone or double stained with α-tubulin monoclonal antibody (working dilution 1:100; Sigma-Aldrich, St. Louis, MO), followed by anti-rabbit IgG-conjugated with fluorescein isothiocyanate or anti-mouse IgG conjugated with rhodamine (working dilution 1:200; Jackson Immunoresearch Laboratories, West Grove, PA). Specimens were mounted with Vectashield (Vector Laboratories) containing propidium iodide (to detect nucleic acids by a red emission color) or 4,6-diamidino-2-phenylindole (to detect the position of the nucleus by a blue emission color). Acrosomes were labeled with peanut agglutinin (PNA) Alexa Fluor 488 conjugate (Molecular Probes, Eugene, OR) according to the procedure described by Szász et al. (2000 ).
Testes from adult rats (Sprague-Dawley), wild-type mice (C57BL/6J), and azh mutant mice (The Jackson Laboratory, Bar Harbor, ME) were used for transmission electron microscopy and immunogold electron microscopy. For transmission electron microscopy, samples were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 6.9) and postfixed in 2% osmium tetroxide in the same buffer, and embedded in a plastic resin according to a standard procedure. For immunogold electron microscopy, samples were fixed in a mixture of 1.5% glutaraldehyde and 3.4% paraformadehyde (electron microscope grade) in 0.1 M phosphate buffer, pH 7.2, embedded in Lowicryl K4M (Polysciences, Warrington, PA), and processed for immunogold microscopy as described previously (Rivkin et al., 1997 ). Anti-K5 polyclonal affinitypurified sera (see above) and β-actin and α-tubulin monoclonal (mouse) antibodies (Sigma-Aldrich) were used at working dilutions of 1:100 in phosphated-buffered saline containing 0.1% Tween 20, 1% bovine serum albumin, and 1% goat serum. Bound antibody was detected by incubating the sample overnight at 4°C with goat anti-rabbit IgG or anti-mouse IgG conjugated with 10-nm gold particles (Amersham Biosciences, Piscataway, NJ), by using a 1:50 working dilution. Sections were stained for 5 min with 5% uranyl acetate in deionized water, and specimens were examined using a JEM-100CX transmission electron microscope operated at an accelerating voltage of 60 kV.
Human and rat skin specimens were fixed in 4% paraformadehyde in 0.1 M phosphate buffer, pH 7.2, for 1 h, embedded in paraffin, and sectioned. Sections were immunoreacted with anti-K5 serum (working dilution: 1:100 for immunofluorescence and 1:3000 for immunoperoxidase) and processed using standard procedures. A commercial affinity-purified polyclonal antiserum to mouse K5 (working dilution 1:100; Covance, Richmond, CA) was used as an additional control for antibody specificity (our unpublished data).
cDNA cloning was carried out as described in MATERIALS AND METHODS. An analysis of the deduced amino acid sequence of rat testis Sak57 shows an overall 89% homology with human K5 (Figure 1, conservative substitutions are also indicated). The rod domain is the most conserved segment of K5 with 98% homology in both human and mouse (our unpublished data) K5. It has a typical length for keratins (304 amino acids) with well-conserved start and end sequences (IKTLNNKFASF and EIATYRKLLEG, respectively). The head and tail domains of K5 have significantly lower homology. In particular, the head domain has 88% homology with both human and mouse K5, whereas the tail domain has 69 and 50% homology with human and mouse K5, respectively. These lower homologies are estimates because they were obtained by introducing gaps in the sequences. Reverse transcription-PCR analysis by using rat testis total RNA and a rat pachytene spermatocyte cDNA library demonstrates the expression of K5 transcripts during spermatogenesis (Figure 2A).
The subdomain 2A contains the amino acid sequence KAQYEDIANR, which is identical to that previously found by others by partial amino acid sequencing of rat epidermal K5 (see Figure 2 in Kierszenbaum et al., 1996 , and references therein), but differs by two carboxyl terminal amino acids from the one obtained by mass spectrometry analysis of purified Sak57 (KAQYEDIAQK). This peptide was used to raise a polyclonal anti-K5 serum and for affinity purification (Kierszenbaum et al., 1996 ). To verify that this antibody actually recognizes K5 protein, a cDNA fragment containing this sequence was cloned into the expression vector pBlue-2 and the fusion protein was expressed in Escherichia coli. Figure 2B demonstrates that anti-K5 antibody recognizes the fusion protein produced upon induction by IPTG. The specificity of anti-K5 serum was verified by competition experiments (Figure 2C). The KAQYEDIAQK peptide completely blocked recognition of the fusion protein by the affinity-purified K5 antibody. In contrast, an unrelated peptide had no effect.
We have used affinity-purified polyclonal antibodies generated in rabbit against the peptides KAQYEDIAQK (corresponding to a segment of the 2A region of the α-helical rod domain) and LEGQECRLSGEGVG (corresponding to the beginning of the tail domain) to determine specific immunoreactive sites in rat and human epidermis. Figure 2, D and F, demonstrate the localization of K5 immunoreactive sites in the suprabasal layer of the rat epidermis (Figure 2D; indirect immunofluorescence) and human epidermis (Figure 2F; immunoperoxidase), where K5 is known to coexist with keratin 14. Specific immunoreactivity was not detected in human epidermis incubated with preimmune serum (Figure 2E; negative control). Immunoblotting experiments, using testis and human skin lysates, demonstrate that the K5 antiserum detects a 57-kDa immunoreactive protein band in both rat testis and human epidermis samples (Figure 2G). A similar immunoblotting experiment, using anti-mouse K5 antibody, yielded similar results (our unpublished data).
With these results, indicating a distinct immunoreactive homology of rat testis Sak57 and rat and human epidermal K5, we focused our search of K5 immunoreactive sites in whole rat spermatids collected from spermatogenic stage-specific seminiferous tubules (Figure 3). A mild hypotonic treatment with 0.1 M sucrose (selected panels in Figure 3) and brief Triton X-100 extraction (Figure 4) were used to determine whether the K5-containing acroplaxome remained firmly attached to the spermatid nucleus. As the acrosome extends caudally along the spermatid nucleus and the manchette initiates its development (step 8 of spermiogenesis), a conspicuous K5-immunoreactive ring coinciding with the boundary of the acrosome was visualized (Figure 3, A–E). The developing manchettte, flanking the spermatid nucleus, displayed both tubulin and K5 immunoreactivity. During step 12, both the fully developed microtubular manchette and the acrosome, descending along the elongating spermatid nucleus, are K5 immunoreactive (Figure 3, F and G). The association of K5 with microtubules of the manchette was reported previously (Tres and Kierszenbaum, 1996 ). During step 17, when the manchette has fully disassembled, K5 immunoreactive sites occupy a caudal nuclear distribution (Figure 3, H–K). A β-actin immunoreactive ring was observed during step 8 of spermiogenesis at the boundary of the acrosome (Figure 3M; compare with Figure 3, B and E, showing similar K5 localization sites). During step 12, both K5 and β-actin immunoreactive sites coexisted in the acroplaxome. A difference with K5 is the less obvious β-actin immunofluorescent localization in the manchette (compare Figure 3, F and G, and O and P). A strong propidium iodide nuclear-stained region correlates with the cap-like projection of the acrosome (Figure 3N). We concluded from these observations that a K5/β-actin-containing acroplaxome is present at the acrosome-nuclear boundary and that the acroplaxome marginal ring is better visualized in round spermatids when the orientation is either frontal or slightly lateral.
Figure 4, A–D (steps 4 and 5 of spermiogenesis), and E–H (step 6) demonstrate that a combination of sucrose hypotonic treatment and a brief Triton X-100 extraction step before fixation removes most of spermatid cytoplasm, leaving relatively undisturbed the acrosomal vesicle and associated Golgi apparatus. A shallow nuclear indentation accommodates the acrosome. The boundary of shallow indentation displays a distinct propidium iodide-stained nuclear band (Figure 4G), suggesting a different degree of chromatin condensation at this site. To determine whether the marginal ring was not due to a peripheral accumulation of acrosomal enzymes, specimens were stained with PNA conjugated with Alexa Fluor 488 to selectively label the acrosome (Szász et al., 2000 ). Figure 4, I–L, demonstrates that the entire acrosome, including the marginal region of a step 6 spermatid, stains evenly, in contrast with the ring-like distribution of K5 seen in the same developmental step (Figure 4, J and K, inset).
Indirect immunofluorescence suggested that β-actin and K5-containing elements were evenly distributed along the acroplaxome and clustered at the marginal ring. The structure of the ring was analyzed further by electron microscopy. Parallel aligned and obliquely oriented bundles of filaments with respect the nucleus can be visualized at the leading edges of the acroplaxome (rat spermatid, Figure 5A). In a sagittal view (mouse spermatid; Figure 5B), the marginal ring of the acroplaxome displays cross-sectioned 10-nm-thick filaments attached to a dense plaque bound to the inner acrosome membrane. F-Actin microfilaments (7 nm in thickness) in the Sertoli cell ectoplasmic region and microtubules of the manchette (25 nm in width) in the same microscopic field provide support to the different thickness of the intermediate filaments. The K5 immunoreactive filaments can also be visualized by immunofluorescence at the expected position and angular orientation (Figure 5, C and D).
Immunogold electron microscopy demonstrates the presence of K5 immunoreactivity in the subacrosomal region where the acroplaxome and marginal ring are located (Figure 6, A and B). Similar localization sites are seen for β-actin (Figure 6C). Scattered β-actin is also seen in the manchette in agreement with a previous report (Mochida et al., 1998 ) We concluded from these combined immunofluorescence and immunogold electron microscopy observations that F-actin and K5 coexist in the acroplaxome and that these two cytoskeletal components are concentrated at the marginal region of this subacrosomal plaque. Figure 7 presents a summary of the relevant aspects of acroplaxome development in rat spermatids within a graphic and descriptive context.
Mutant mouse models in which spermatid nuclear shaping is defective can provide clues concerning the significance of the acroplaxome in the elongation of the spermatid nucleus. Previous studies have shown that, in the azh mutant mouse, a relative number of spermatids display abnormally shaped nuclei, sperm have coiled tails (Mochida et al., 1999 ; Akutsu et al., 2001 ), and the fertility rate is significantly reduced (reviewed by Meistrich, 1993 ). Figure 8A illustrates an elongating spermatid in which the acroplaxome-containing region is particularly indented. A similar feature was observed in mouse offspring produced from normal mouse oocytes injected with sperm heads from the azh mutant (see Figure 4, B and C, in Akutsu et al., 2001 ). An additional deformity during the nuclear shaping of the azh spermatid mutant is the presence of a nuclear constriction at the site where the perinuclear ring of the manchette is located (Figure 8B). These observations suggested that stress forces acting upon and/or generated by the acroplaxome were unbalanced and prone to generate focal deformities.
We report herein structural details and the partial molecular characterization of the acroplaxome, an attachment plate present in the subacrosomal space of mammalian spermatids (including mouse, rat, and human) and linking the inner acrosomal membrane to the nuclear envelope. The acroplaxome contains F-actin and K5 and presumably other actin-associated proteins and motor proteins, which may be responsible for tethering the primary proacrosomal vesicle during the early stages of acrosomal biogenesis. A specific feature of the acroplaxome is its marginal ring housed in a shallow circular indentation of the spermatid nucleus. An attribute of the marginal ring is a conspicuous bundle of 10-nm-thick intermediate filaments attached to the dense plaque associated with the inner acrosomal membrane. The finding of intermediate filaments in the acroplaxome is not surprising. In fact, intermediate filament-like structures in the subacrosomal space were reported previously (Russell et al., 1986 ). However, neither their biochemical nature nor regular organization at the marginal ring of the acroplaxome was described. Opposite to the intermediate filament–plaque complex is another thin plaque spanning across a shallow indentation in the spermatid nuclear envelope and linked to a nuclear lamina (see Figure 9 for a summary diagram). The subacroplaxome position of the nuclear lamina correlates with the propidium iodide dense chromatin staining seen in Figure 4G. The band-like edge of the chromatin density correlates with the collar-like arrangement of the plaque housing the marginal ring of the acroplaxome. The marginal ring and associated plaques are reminiscent of the zonula adherens belt conjoining the plasma membranes of adjacent epithelial cells. During spermiogenesis, the F-actin/K5-containing marginal ring trails the descending leading edge of the acrosomal sac and then seems to disassemble when nuclear elongation reaches completion (Figure 7). This steady association implies that the acroplaxome is expected to adjust its diameter and shape without compromising the relationship of the marginal ring with both the descending acrosomal edge and the elongating spermatid nucleus. In this regard, the insertion plaques, whose molecular components are presently unknown, may play a pivotal role in stabilizing the position of the acroplaxome. Because gentle hypotonic and mild Triton X-100 treatment cannot dislodge the acrosome–acroplaxome complex from its nuclear attachment site, we postulate that the acroplaxome secures the acrosome at the corresponding nuclear pole during the elongation of the spermatid head. Several observations in mouse mutants with sperm head abnormalities determined by deficient acrosome development (Hrb mutant, Kang-Decker et al., 2001 ; GOPC mutant, Yao et al., 2002 ) make this an attractive hypothesis. As a corollary, a defect in acrosome biogenesis is likely to hamper the assembly and function of the acroplaxome in nuclear shaping.
There are two aspects of this work that merit further discussion. First, the expression of classical keratins and outer dense fibers (Odfs) during spermatogenesis and spermiogenesis in particular. Second, the potential mechanical contribution of the acroplaxome to the nuclear shaping of the male gamete. Both nucleotide sequence analysis and deduced amino acid sequence demonstrate that testicular K5 has a 89% amino acid homology with K5, a type II keratin abundant in the basal layers of the epidermis (Fuchs, 1996 ). K5 forms a cortical cytoplasmic shell in primary spermatocytes, is present in the centrosomal region during late male meiotic prophase I, is a component of the cytoplamic bridges conjoining cohorts of spermatocytes (Tres et al., 1996 ), and is associated first with the manchette and later with the Odfs and fibrous sheath of the developing tail during rat (Kierszenbaum et al., 1996 ; Tres and Kierszenbaum, 1996 ) and mouse spermiogenesis (Akutsu et al., 2001 ). A role of K5 in spermatogenesis is supported by the occurrence of signifi-cant apoptosis during meiotic prophase and spermatid nuclear and tail abnormalities in selected seminiferous tubules of K5+/- mutant mice (testis samples provided by Thomas M. Magin and Lu Hong, University of Bonn, Germany). Spermiogenesis in K5 null mice cannot be evaluated because they exhibit neonatal lethality (Peters et al., 2001 ). Another keratin, K9, is one of the components of the perinuclear ring of the manchette (Mochida et al., 2000 ). K9 is a unique keratin because its expression is predominant in the epidermis of palm and sole (human) and footpad of rodents and other species. K9 and K5 are examples of type I keratin (which includes K9 to K20) and type II keratin (which includes K1 to K8), respectively, thus far detected during spermatogenesis. At least one member of each family of keratins is required to form heterodimeric intermediate filament proteins. In fact, we have found that keratins 14 and 2e, the common partners of K5 and K9, respectively, are also expressed in testis (our unpublished data).
It is significant to point out that Odfs are cytoskeletal proteins that are heavily disulfide-linked, contain repeats of the Cys-X-Promotif, and surround the axoneme of the sperm tail (Calvin and Bedford, 1971 ; Olson and Sammons, 1980 ; Vera et al., 1984 ; Oko, 1988 ). Three Odfs have been so far reported: Odf1 (27 kDa, van der Hoorn et al., 1990 ; Burfeind and Hoyer-Fender, 1991 ), Odf2 (84 kDa, Brohmann et al., 1997 ), and Odf3 (110 kDa, Petersen et al., 2002 ). An Odf2-like protein was found to be a component of the centrosome scaffold in somatic cells (Nakagawa et al., 2001 ). The widespread distribution of an Odf2-like protein as a centrosome scaffold component, together with the presence of K5 in the centrosome region of pachytene spermatocytes (Tres et al., 1996 ), and in the acroplaxome (this article) are a demonstration that keratins are used throughout spermatogenesis. Together, these observations strengthen the view that classical keratins (initially identified in the epidermis) and Odf proteins (initially ascribed as sperm-specific proteins) can coexist in the sperm tail to provide structural and mechanical support for stabilizing long flagella while minimizing the risk of kinking or breaking.
The second aspect of the discussion relates to the mechanism of assembly of the acroplaxome and its potential function during spermiogenesis. Concerning the assembly of the acroplaxome, F-actin precedes the appearance of K5 (Figure 7), thus suggesting that it may provide a template for its association with keratin filaments, an event known to occur in vitro (Weber and Bement, 2002 ). Because the diameter of the acroplaxome increases in parallel with the corresponding spreading of the acrosome over the elongating spermatid nucleus (Figure 7), it is conceivable that this process may occur at the expense of actin polymerization. F-Actin is present in the subacrosomal space during most of spermiogenesis before depolymerizing to G-actin in late spermatids and sperm (reviewed by Vogl, 1989 ).
A significant question is whether F-actin in the acroplaxome can provide a binding scaffold for motor proteins and contain actin regulatory proteins. The motor protein myosin-Va is present in the acroplaxome (our unpublished observation), and there are precedents for actin regulatory proteins. It is known that binding of the actin-depolymerizing factor cofilin to actin promotes the transition of polymeric F-actin into oligomeric or monomeric G-actin (reviewed in Ayscough, 1998 ). Phosphorylation of cofilin by LIMK (LIM kinase) prevents cofilin binding and depolymerization of actin, leading to the accumulation of filamentous actin. In this respect, the testicular isoform of LIMK2 is associated with spermatids and targeted Limk2 gene disruption affects the progression of spermatogenesis (Takahashi et al., 2002 ). Furthermore, a number of actin-binding proteins have been reported at the acrosomal region (for example, actin-capping proteins, Hurst et al., 1998 ; calicin, Lécuyer et al., 2000 ; profilin-3, Braun et al., 2002 ; Arc, Maier et al., 2003 ). A variant of Wiskott-Aldrich syndrome protein-interacting src homology 3 protein (designated WISH) was exclusively found in testis (Fukuoka et al., 2001 ). WISH strongly enhances Wiskott-Aldrich syndrome protein-induced actin-related protein 2/3 (Arp2/3) complex activation, resulting in rapid actin nucleation and polymerization. Given the existence of a number of testis-specific actin binding proteins, an analysis of their role in the regulation of actin assembly and disassembly in the acroplaxome would contribute greatly to advance an understanding of the mechanistic complexity of this structure. An experimental approach to addressing these questions is to sequester actin monomers resulting in rapid depolymerization of F-actin. Unfortunately, the physiological consequences of impairing the function of the acroplaxome by disruptors of the actin cytoskeleton (such as latrunculin A) cannot be assessed in cultured spermatids. First, the developmental time of the acrosome-acroplaxome complex is long (days; Clermont et al., 1993 ). Second, actin-disrupting agents can target the supporting Sertoli cells, which are required for spermatogenic cell viability and differentiation in vivo and in vitro (Kierszenbaum, 1994 ). Because of these reasons, mouse mutants provide at present a more feasible approach to determine how the acroplaxome might work.
How can the acroplaxome contribute to spermatid nuclear shaping? It is likely that a stress-resistant acroplaxome can transmit gentle and sustained clutching forces generated by Sertoli cell ectoplasmic F-actin bundles to the elongating spermatid nucleus (Figure 9). The spatial arrangement of stacked F-actin hoops embracing the apical one-third of the spermatid nucleus seems to be kept in place by the actin-linked afadin–nectin complex, which may ensure constant Sertoli-spermatid contact (Ozaki-Kuroda et al., 2002 ). In addition to the force exerted by Sertoli cell F-actin hoops, the perinuclear ring of the manchette can provide an endogenous clutch applied at the caudal two-thirds of the spermatid nucleus. As the manchette descends along the elongating spermatid nucleus, its perinuclear ring gradually reduces its diameter and, in a sleeve-like manner, may apply forces steering the elongation of the spermatid nucleus. In fact, fractionated manchette perinuclear rings of various diameters have been demonstrated (Mochida et al., 1998 ), and a nuclear deforming constriction at the level of the perinuclear ring of azh mutant mice spermatids can be seen (Figure 8B). In summary, we have shown that the F-actin–keratin-containing acroplaxome provides a planar scaffold, which maintains the acrosome at the nuclear anchoring site during nuclear elongation. This study provides a novel model for spermatid nuclear shaping, which can be extended by the analysis of a number of mutant mice in which the development of the acrosome and integrity of the acroplaxome are defective.
This work was supported by U.S. Public Health Service grant HD37282.