An Unexpected Localization of Basonuclin in the Centrosome, Mitochondria, and Acrosome of Developing Spermatids

Human epidermal keratinocytes were cultured (Rheinwald and Green, 1977), and poly(A)+ mRNAs were isolated using the procedure of Chomzcynski and Sacchi (1987). The RNA preparation was translated in a reticulocyte lysate system in the presence of [³⁵S]methionine, and proteins were resolved by SDS-PAGE and autoradiography. Proteins >200 kD were translated from the mRNAs, verifying the quality of the preparation. These mRNAs were then used to engineer a λ-zap phage library (Stratagene, La Jolla, CA), made from a mixture of cDNAs that were synthesized using oligo dT and random hexamer oligonucleotide primers.

A 688-bp EcoRI/EcoRD fragment encoding a 5′ segment of the human basonuclin 1b mRNA was used to screen a 129/sv mouse genomic library (Stratagene). Three hybridizing clones were identified and subsequently purified. Two clones, mBSN-1 and mBSN-2, were subcloned as ∼17 kb NotI restriction fragments into Bluescript KS+. These clones were then subjected to restriction map analyses and partial sequencing.

Three peptides corresponding to segments of the published basonuclin protein sequence (Tseng and Green, 1992) were synthesized, coupled to keyhole limpet hemacyanin, and used for generating polyclonal antisera in rabbits (Zymed Labs, Inc., S. San Francisco, CA). The three basonuclin peptides are: UC56, CRPPPSYPGSGEDSK (human sequence, corresponding to the published amino acid residues 449–463; Tseng and Green, 1992); Ab176, ESCGHRSASLPTPVD (mouse sequence, equivalent to human residues 208–222); and Ab372, ASPNPRLHAMNRNNR (mouse sequence, equivalent to human residues 404–419). All antisera were purified by affinity chromatography using the appropriate peptide-conjugated Sepharose columns. Antibodies were tested by immunoblot analyses on proteins extracted from the skin and testes of adult mice.

Immunoblot analyses were performed as described (Yang et al., 1996). RNA blots used for Northern analysis were purchased from Clontech (Palo Alto, CA), and hybridizations were performed as described by the manufacturer. A 576-bp radiolabeled human cDNA corresponding to sequences within exons 2–4 of basonuclin was used for hybridization. In situ hybridizations of frozen sections of mouse testes were performed using digoxygenin-labeled antisense and sense riboprobes as described (Chiang and Flanagan, 1996).

Frozen tissue sections (10 μm) were cut onto Superfrost plus slides. Sections were briefly fixed with methanol (−20°C) for 10 min and then washed 2× with PBS. Sections were preblocked with a solution containing 1% BSA, 0.1% Triton X-100, and 1% gelatin in PBS. Primary antibodies were then added to fresh solution and incubated with sections at room temperature for 1 h. Antibody concentrations used were: anti-BSN antibodies (1:20 to 1:500); anti–γ tubulin antibodies (1:200 dilution; gift of Dr. Harish C. Joshi, Emory University, Atlanta, GA; gift of Dr. Bruce Alberts, University of California at San Fransisco, San Fransisco, CA); H1 human autoantisera against centrosomes (1:250 dilution; gift of Dr. Thomas Medsgar, University of Pittsburgh, Pittsburgh, PA); and anti–PLCβ1 (purchased from Santa Cruz Biotechnology, Santa Cruz, CA). After washing the slides 3× with PBS for 10 min each, sections were incubated with fresh solution containing secondary Texas red or FITC-conjugated antibodies (1:100 dilution) for 30 min before washing as before and mounting. Nuclei were stained either with propidium iodide or with 4,6-diamidino-2-phenylindole (DAPI)¹. Sections were examined using either a confocal microscope (model LSM 410; Carl Zeiss, Inc., Thornwood, NY) or an immunofluorescence microscope (model Axiophot; Carl Zeiss, Inc.).

For regular EM, tissues were fixed at room temperature for at least 1 h with 2.5% glutaraldehyde and 4% paraformaldehyde in 0.2 M sodium cacodylate buffer, pH 7.4. Samples were trimmed and washed three times with the same buffer and then postfixed with 1% aqueous osmium tetroxide for 1 h at room temperature. These samples were further washed with sodium cacodylate buffer, followed by maleate buffer, pH 5.1, and stained en bloc with 1% uranyl acetate for 1 h at room temperature. The samples were then washed several times with maleate buffer, dehydrated with cold ascending grades of ethanol and propylene oxide, embedded with LX-112 medium, and polymerized at 70°C for 48 h before sectioning.

Semithin sections (0.75 μm) were stained with toluidine blue, and these were visualized by light microscopy. Ultrathin sections (∼80 nm) were cut with a regular diamond knife, collected on 200-mesh, uncoated copper grids, and double stained with 50% saturated uranyl acetate and 0.2% lead citrate. These sections were then examined with a transmission electron microscope (model CX100; JEOL U.S.A., Inc., Peabody, MA) operated at 60 kV.

For postembedding immunolabeling, samples were placed in warm fixative containing 2% fresh paraformaldehyde, 0.05% glutaraldehyde in 0.1 M PBS, pH 7.4, for 10 min. The samples were trimmed, washed several times with the same PBS at 4°C, then dehydrated at −25°C with ascending grades of ethanol, infiltrated with different concentrations of Lowicryl K4M medium, embedded in gelatin capsules with fresh 100% Lowicryl K4M medium, and polymerized at −25°C with UV light for 5–7 d.

Ultrathin sections (∼90 nm) were cut with a diamond knife, collected on Nickel grids coated with Formvar and thin carbon films, and labeled with specific antibodies according to the tested dilution, followed by incubation with 10-nm colloid gold-conjugated secondary antibodies. The sections were briefly stained with uranyl acetate and lead citrate and examined with an electron microscope (model CX100; JEOL U.S.A., Inc.) operated at 60 kV.

In the course of our studies on epidermal differentiation, we hybridized a human keratinocyte cDNA library with a 517-bp PCR fragment corresponding to the zinc finger domain of human basonuclin (nucleotides 2657–3174 of hBSN1a), a protein known to be expressed in mitotically active basal epidermal cells (Tseng and Green, 1992). One clone contained a 4,606-bp insert, which upon sequencing was shown to harbor a complete open reading frame, followed by 1,547 bp of 3′ sequence containing a polyadenylation signal at nucleotide 4583. This cDNA encoded a protein that was nearly identical to the published sequence from amino acid residues 33–993 (Tseng and Green, 1992). However, it differed in its 5′ segment by the absence of a 32–amino acid residue sequence and the replacement of a novel 166-nucleotide residue sequence.

To assess whether both 5′ basonuclin sequences were bona fide, we screened a genomic library and mapped the positions of the two 5′ upstream sequences relative to the remainder of the basonuclin gene. Each sequence was contained within individual exons that were found within a single genomic clone (data not shown). The new sequence was located in an exon 3′ to the one present in the previously published sequence. These data confirmed the existence of the two sequences within the basonuclin gene. We refer to the two predicted forms as BSN1a (Tseng and Green, 1992) and BSN1b (this report), based upon the positioning of their respective exons within the basonuclin gene. The two basonuclin sequences were characterized from both mouse and human and are provided in Fig. 2. The BSN1b exon is highly conserved and is nearly identical between mouse and human. The BSN1a exon found in the originally reported sequence (Tseng and Green, 1992) is less conserved between the two species, and different translation start sites are predicted.

Interestingly, both mouse and human forms of BSN1a but not BSN1b contain the sequence RRPEPG, which has been shown to be a mitochondrial targeting sequence for proteins (Komiya et al., 1994; Shore et al., 1995; McBride et al., 1996). While not previously noted, a computer survey of all protein sequences known indicates a 60% chance of the BSN1a form localizing to the mitochondria, a prediction that we address experimentally later in the text.

Previously, basonuclin was thought to be restricted in its expression to mitotically active cells of stratified squamous epithelia (Tseng and Green, 1994). In the course of examining basonuclin RNA expression in different mouse tissues, we were surprised to discover that PCR primers corresponding to the shared sequences of BSN1a and BSN1b detected a band in testis mRNA in addition to RNAs isolated from tissues known to contain keratinocytes (Fig. 3 A).

To assess whether our new form of basonuclin was expressed in both stratified epithelia and testis, we used reverse transcriptase and PCR on mouse skin and testis mRNAs in the presence of a 5′ oligonucleotide primer to one or the other of the 1a and 1b sequences. In each case, the 3′ primer corresponded to sequence shared by both cDNAs. As shown in Fig. 3 B, both primer sets produced a band of the expected size, indicating that both sequences are expressed by skin and testis. In our initial study, we have focused on using cRNA probes and antibodies that are shared by the two forms, and we refer generally to the properties of basonuclin (BSN).

To examine BSN expression in testis in more detail, we first conducted Northern blot analysis. As shown in Fig. 4 A, a single RNA band of ∼4,600 nucleotides was obtained from mouse testis. This band was comparable in size to that seen in human keratinocyte mRNA preparations (Tseng and Green, 1994), and it corresponded to the size expected for BSN1a and BSN1b mRNAs, which have a long 3′ untranslated sequence. For human testis, two bands were detected in approximately comparable levels: one was an ∼4,600-nucleotide band, as expected, and the other was an ∼3,200-nucleotide band. This smaller band is large enough to encode full-length basonuclin, although our focus for the remainder of the study was on mouse, and we have not pursued the identity of this smaller band in human testis.

To verify that the hybridizing band(s) detected in the testis corresponded to bona fide BSN mRNA and to examine the testis-specific expression of BSN mRNA during sexual maturation, we conducted RNase protection assays. In mouse germ cell development, meiosis in the seminiferous tubules begins at about 2 wk postnatally, and production of mature sperm occurs by 5 wk of age. As shown in Fig. 4 B, a single band of the expected size was protected when mRNAs were used from mouse testes taken at 2 and 5 wk after birth and adult. The overall levels of BSN RNAs appeared to be comparable. These data demonstrated unequivocally that BSN mRNAs are expressed in testis and that their expression exists before sexual maturity.

To determine where BSN mRNAs are expressed within the testis, we conducted in situ hybridizations on frozen sections of mouse testes isolated at various stages of postnatal development. A digoxygenin-labeled antisense BSN cRNA hybridized strongly in the seminiferous tubules of all testis samples examined (Fig. 5). Hybridization was detected at the periphery of the tubules and appeared to be present even at birth, before spermatogenesis (Fig. 5, A). Hybridization remained high throughout most of spermatogenesis. At 2 wk of postnatal development, hybridization was strongest in the centers of the tubules, where the primary spermatocytes are located, and weaker at the periphery, where the spermatogonia reside (Fig. 5, B). By 4–5 wk of age, spermatid formation, i.e., spermiogenesis, had begun (Rugh, 1990), and BSN RNAs were still detected throughout the tubules (Fig. 5, C–E). The persistence of BSN mRNA in late-stage spermiogenesis suggests that BSN RNAs are stable, as is the case with many mRNAs that are translated at this time (for review see Browder et al., 1991). Basonuclin cRNA hybridization was largely specific for derivatives of the germ cell population within the testis and was not detected in the interstitial Leydig cells. If present at all in Sertoli cells, the signal was reduced over that seen in germ cells. No hybridization was seen with the sense control cRNA (Fig. 5 F).

Both BSN1a and BSN1b cDNAs are predicted to encode 120-kD polypeptides; however, this has never been confirmed by immunoblot analysis. We therefore raised monospecific rabbit antibodies to three different peptide sequences present in the shared regions of these proteins (see Materials and Methods). As judged by immunoblot analysis, each of the three different affinity-purified peptide antibodies (UC56, 372, and 176) detected a single crossreacting band of 120 kD in protein extracts from mouse testis (Fig. 6,A). A band of this size was also detected in protein extracts from mouse skin and from other epithelial tissues known to contain keratinocytes (Fig. 6, A and B). Collectively, these findings: (a) establish the size of basonuclin in testis, skin, and other stratified epithelia; (b) verify the specificity of our antisera; and (c) suggest that, if other major forms of basonuclins exist, they either must be 120 kD in size or alternatively must have a most peculiar splice pattern, missing three different domains of the basonuclin protein.

To assess the location of basonuclin within differentiating male germ cells, we conducted indirect immunofluorescence on frozen sections of developing mouse testes. Basonuclin protein was first detected in testis at 2 wk postnatally (Fig. 7,A). In contrast to BSN RNAs, which are expressed in mitotic spermatogonia, protein was not detected until the cells had differentiated into primary spermatocytes located at the midregion of the 2-wk seminiferous tubules. These cells, in the first meiotic phase of differentiating male germ cells, displayed a dotlike pattern of staining with the UC56 anti-BSN (αBSN) antibody. Double immunofluorescence with DAPI to stain chromatin indicated that the labeling was located near the nucleus (Fig. 7,A). Staining was more prevalent by 4 wk (Fig. 7, B and C), when primary and secondary spermatocytes exist (Rugh, 1990). While the majority of these spermatocytes contained single dots, a few seemingly contained double dots positioned at opposing sides of the nucleus (Fig. 7 B, inset). Similar staining patterns were observed with all three affinitypurified BSN antibodies, although antibodies UC56 (shown) and 372 gave the strongest staining. The pattern was not seen with secondary antibody alone.

The dotlike staining pattern suggested that basonuclin might be localizing to centrosomes. To explore this possibility in greater detail, we used double immunofluorescence labeling with the H1 human autoimmune serum (αH1), known to cross-react with centrosomal proteins (Shu and Joshi, 1995). As shown in Fig. 7, D–F, the two antibodies displayed staining that was superimposable at the confocal microscopy level. This was further verified by staining serial cross-sections with same-species antibodies against γ-tubulin (not shown). Interestingly, only the centrosomes near the midregion of the seminiferous tubules costained with αBSN and αH1; centrosomes at the periphery stained with αH1, but not the UC56 sera (Fig. 7 G). Based upon these data, basonuclin appeared to be a specific component of the centrosomes of postmitotic, differentiating male germ cells.

In sexually mature adult testes, anti-BSN antibodies strongly stained the spermatid heads (Fig. 8). Costaining with propidium iodide, which labels chromatin, indicated that this labeling was not nuclear. The crescent-shaped staining pattern, coupled with the appearance of this strong staining in the spermatid region of the seminiferous tubules, was reflective of that seen for acrosomal proteins in spermatids (Lepage and Roberts, 1995; Walensky and Snyder, 1995; Yoshiki et al., 1995). Interestingly, despite the fact that spermiogenesis in mouse is initiated by 4 wk, and that acrosomal caps are seen throughout the centers of the 4-wk seminiferous tubules, these caps did not stain with αBSN (not shown). The relatively late acquisition of anti-BSN staining in the acrosome suggested that basonuclin is a component of late-stage sperm acrosomes.

Finally, we observed αBSN staining within the middle piece of the tail of maturing spermatids (Fig. 8,B, arrows). This structure contains the mitochondrial sheath at the upper portion of the 9 + 2 axoneme (Fig. 1 B). This observation was surprising, given that mBSN1a was not predicted to contain mitochondrial localization signal seen at the amino end of hBSN1a.

Again, as was the case for the centrosomal staining, all three affinity-purified antibodies against basonuclin labeled the acrosomes and the middle piece. This said, the UC56 and 176 antibodies showed significantly stronger staining in the acrosome than did the 372 antibody. Since the three antibodies detected a single major 120-kD band by immunoblot analysis, we posit that these differences reflect variation in masking of the basonuclin epitopes in centrosomes and acrosomes.

The pattern of BSN antibody staining was unexpected and diverse. To verify that the staining patterns reflected multiple locations for basonuclin protein, we conducted cell fractionation studies. Although procedures for isolation of centrosomes from testis tissue have not yet been developed, it is possible to dissociate isolated sperm into tail, acrosome, and headpiece by sonication and to subsequently resolve these fractions by sucrose gradient ultracentrifugation (Walensky and Snyder, 1995). We applied this procedure to mature sperm that we removed from the epididymis of adult mice. First, we verified that mature sperm, similar to spermatids, display αBSN UC56 immunofluorescence staining in the acrosome, middle piece of the tail, and centrosome. (Fig. 9 A; sperm centrosomal staining was more readily visible with the 372 antibody, which did not stain acrosomes so brightly.)

Sperm fractions were examined by phase contrast and immunofluorescence microscopy to verify that the separation procedure was successful (not shown). Proteins from each fraction were then resolved by SDS-PAGE, and the gel was stained with Coomassie blue to visualize the proteins (Fig. 9,B). All three fractions contained different sets of proteins. As judged by immunoblot analysis, basonuclin was present in the sperm tail and acrosome fractions, but it was not present in appreciable amounts in the nuclear fraction (Fig. 9,C). The purity of fractions was confirmed by immunoblot analysis using an antibody against the established acrosomal marker PLCβ1 (Walensky and Snyder, 1995; Fig. 9 C) and a protamine nuclear antibody (not shown). These data were consistent with our immunofluorescence analysis. The absence of BSN immunoblot reaction in the nuclear fraction indicated that our failure to observe αBSN staining in sperm nuclei was due to the absence of protein, rather than the masking of BSN epitopes. We do not yet know whether basonuclin is present in nuclei of germ cells at earlier stages of spermatogenesis.

To further examine basonuclin expression during spermiogenesis, we conducted electron and immunoelectron microscopy (Fig. 10). Fig. 10,A provides an example of the typical pair of centrioles that associates with the nuclear envelope during the acrosomal cap phase of spermiogenesis (also see diagram in Fig. 1,B). At this stage of differentiation, centrioles migrate to the nuclear pole opposing the acrosomal cap. The 9 + 2 axoneme assembly of the flagellum always initiates from the end of the distal centriole (Fig. 10, A and C, dis), which contains a prominent electron-dense satellite appendage often in close proximity to the nuclear envelope (Fig. 10,A and B, arrowheads). The proximal centriole (Fig. 10, A and C, px), laterally aligned with the nuclear envelope, is not directly involved in flagellar assembly, and its fate is unknown. Concomitant with the attachment of centrioles to the nuclear envelope, the cylinder of manchette microtubules forms around the lower half of the nucleus as it elongates (Fig. 10 B, ma).

Antibodies against basonuclin specifically labeled the satellite appendages of distal centrioles (Fig. 10, C and D). Perfect cross-sections of these appendages revealed a hollow ringlike structure (Fig. 10,C, inset). The labeling of these structures with αBSN was largely distinct from anti–γ- tubulin, which specifically labeled the pericentriolar material surrounding the end of the proximal centriole (Fig. 10 E). γ-Tubulin labeling was not detected at the end of the distal centriole, i.e., at the site of assembly of the 9 + 2 axoneme. Given that the fate of the proximal tubule seems to be variable dependent upon species, this might explain why in Xenopus sperm, γ-tubulin has not been found associated with the pericentriolar material of flagellar centrioles (Stearns et al., 1991; Felix et al., 1994; Stearns and Kirschner, 1994), whereas in mouse sperm, it has (Palacios et al., 1993).

αBSN also labeled acrosomes of late-stage spermatids that had undergone nuclear elongation (Fig. 10,F). By immunoelectron microscopy, the labeling was most dense at the inner surface of the outer acrosomal membrane. Finally, as predicted from our immunofluorescence data, the mitochondria within the middle piece of the sperm tail were specifically and uniformly labeled with anti-BSN antibodies (Fig. 10 G). Based upon the human sequence, we would have presumed that this labeling represented BSN1a rather than BSN1b. Further studies will be necessary to determine whether there are multiple forms of basonuclin that are differentially localized in germ cells.

For years, it has been known that zinc plays an important role in testis development, and a number of zinc finger proteins are expressed in male germ cells (Burke and Wolgemuth, 1992; Noce et al., 1992, 1993; Hosseini et al., 1994; Zambrowicz et al., 1994; Kundu and Rao, 1995; Passananti et al., 1995; Stassen et al., 1995; Mello et al., 1996; Supp et al., 1996). Where tested, these proteins have been found to be strictly nuclear. Our discovery that basonuclin is a testis protein adds another zinc finger protein to this growing list, but its location sets it apart from the others.

Given the prior studies of Tseng and Green (1992, 1994), we were surprised to find basonuclin expressed in testis at all, since it had been thought to be restricted to stratified squamous epithelia. However, BSN RNA expression was as high or higher in testis than in any other organ examined. BSN RNAs were detected early in the differentiative pathway of mouse germ cells, i.e., long before the animals reached sexual maturity. Despite basonuclin RNA expression in mitotically active spermatogonia, basonuclin protein was not detected until later, where antibody labeling was first seen in meiotic spermatocytes. While antibody masking is always a formal possibility, three different affinity-purified peptide antibodies failed to reveal labeling in spermatogonia. Thus, we conclude that if basonuclin protein is expressed earlier in development, it is present at reduced levels or in a very different complex than its location in spermiogenesis.

In meiotic spermatocytes, basonuclin appeared to be concentrated in centrosomes, a location that it then maintained throughout spermiogenesis. At least at later stages of spermiogenesis, basonuclin seemed to be localized to hollow ringlike appendages that were largely if not fully confined to the centriole forming the sperm flagellum. This was reminiscent of mitotic cells, where satellite structures are generally unique to the mature centriole and are not found on newly synthesized (immature) centrioles (for review see Lange and Gull, 1996). The restriction of appendages to the axonemal centriole of male germ cells has been described before (Browder et al., 1991; Lange and Gull, 1996).

Little is known about the functions or molecular complexity of satellite structures associated with centrioles. In fact, the first molecular marker for centriole maturation, cenexin (96 kD), was only recently discovered in mitotically active cells (Lange and Gull, 1995), and as yet cenexin has not been cloned. In contrast to cenexin, which is regulated with the cell cycle of mitotic cells, basonuclin may represent the first example of a centriolar appendage marker that, in male germ cells, is largely specific for spermatocytes and spermatids. The primarily distal centriole location in spermatids is particularly intriguing because: (a) This is the only centriole that nucleates microtubule assembly in the spermatid, and pericentriolar material surrounding centrioles has been implicated in orchestrating microtubule organizing activity (Gould and Borisy, 1977; Telzer and Rosenbaum, 1979; Calarco-Gilliam et al., 1983; Doxsey et al., 1994; Lange and Gull, 1995); and (b) so little is known about how microtubules organize into their unique and diverse arrays during spermiogenesis. Through its association with the distal centrioles of developing spermatids, basonuclin becomes a candidate for a protein that could be involved in tailoring the organization of the microtubules during male germ cell meiosis and spermatid differentiation. Moreover, by identifying one protein involved in these centrioles, basonuclin becomes a powerful tool for identifying additional proteins involved in centriole maturation during postmeiotic spermatogenesis. This should allow us to probe deeper into the differences that exist between the centrosomes of meiotic versus mitotic germ cells.

It is puzzling that BSN antibodies and cell fractionation studies also detect this protein in the acrosome and mitochondrial sheath of the mature sperm. Since the acrosome is a storage vessel for proteins used in fertilization, we surmise that basonuclin might perform a specialized role in this process. This notion is particularly interesting in light of the facts that: (a) Basonuclin appears to associate with the acrosome late in spermiogenesis; and (b) centrioles are absent in the oocytes of many species, including mouse (Schatten, 1994; Lange and Gull, 1996). In the future, it will be important to examine basonuclin expression during oogenesis and to track the fate of sperm basonuclin during fertilization.

While the localization of basonuclin in sperm acrosomes is consistent with the hypothesis that basonuclin performs a function in centrosomes and/or microtubule organization, we are at a loss to explain why basonuclin was also detected in the mitochondria of the flagellar midpiece. This said, BSN1a has a perfect mitochondrial targeting sequence at its amino terminus, and this sequence is evolutionarily conserved. Although mBSN1a seems to utilize a downstream ATG, the two putative forms may perform unique functions in separate compartments of the differentiating male germ cell.

To make matters more intriguing, basonuclin has what appears to be a reasonably bona fide nuclear localization signal. While computer analysis of known proteins indicates that basonuclin has only a 40% chance of being localized to the nucleus, the protein does associate with the nucleus in epidermal keratinocytes (Tseng and Green, 1994), and we have confirmed this with our antibodies (unpublished results). We did not detect basonuclin in isolated sperm nuclei, nor did we detect αBSN labeling in germ cell nuclei. Thus, it seems unlikely that basonuclin is nuclear in germ cells, although we cannot rule out the possibility that in the early stages of spermatogenesis, its antigenic determinants are masked by association with other nuclear proteins. Additionally, while evidence that basonuclin is a DNA-binding protein is lacking, its structural features predict that it has this potential.

One possibility is that basonuclin might be transiently associated with chromatin after nuclear envelope breakdown of meiotic germ cells. Since the cell cycle of meiotic mouse germ cells is so long, meiotic germ cells in the act of nuclear envelope breakdown and spindle formation are rare, making such analysis difficult. However, in this regard, it may be relevant that CP190, a recently described zinc finger protein in mitotic cells of Drosophila, is associated with centrosomes during mitosis and with chromatin during interphase (Oegema et al., 1995; Whitfield et al., 1995). Despite the lack of sequence similarity between CP190 and basonuclin, these findings suggest collectively that: (a) Zinc finger proteins may play important roles in the formation, structure, positioning, or function of centrosomes; and (b) the requirements for these proteins may differ in meiosis and mitosis. As future studies are conducted, the mysteries underlying the dynamic roles of basonuclin during spermiogenesis should become increasingly apparent.

We thank Christa Wellman for her artistic talents exerted in the design of Fig. 1 and Chuck Welleck for his artwork in preparation of the color figures. We thank Dr. Thomas Medsgar for his autoimmune sera against centriolar proteins, Dr. Bruce Alberts, and Dr. Harish Joshi for antibodies against γ-tubulin. We thank Dr. Gary Smith for his valuable discussions on spermiogenesis, Dr. Xiaoming Wang for her advice on RNase protection assays, and Dr. Satrajit Sinha for his computer analysis and identification of the mitochondrial localization signal.