Program of Biochemistry and Department of Pharmacology and Therapeutics, Roswell Park Cancer Institute, Elm & Carlton Streets, Buffalo, New York, NY 14263, USA

Correspondence to: M F Kulesz-Martin, Program of Biochemistry and Department of Pharmacology and Therapeutics, Roswell Park Cancer Institute, Elm & Carlton Streets, Buffalo, New York, NY 14263, USA

^aCurrent address: Department of Dermatology L468, Oregon Health Sciences University, School of Medicine, 3181 SW Sam Jackson Park Road, Portland, Oregon, OR 97201-3098, USA

A novel spermatogenesis associated factor (SPAF) was found to be aberrantly expressed at the malignant conversion stage in a clonal epidermal model of chemical carcinogenesis. Sequence analysis revealed two ATPase modules, classifying this gene as a new member of the AAA-protein family (ATPase associated with diverse activities). Immunohistochemical staining of mouse testis sections with SPAF antibody localized expression to spermatogonia and early spermatocytes in the basal compartment of the seminiferous tubules. Northern and Western analysis of SPAF expression in testes of mice at different developmental stages confirmed its expression at early stages of spermatogenesis. In view of a mitochondrial-localization-like signal, sequence similarities to membrane-associated proteins, ATP binding properties, and intracellular expression patterns in testis, we speculate that SPAF protein may be involved in morphological and functional mitochondrial transformations during spermatogenesis. Ectopic expression of the SPAF gene in malignant epidermal cells may signify adoption of an early germ cell-like phenotype advantageous in malignant conversion. Oncogene (2000) 19, 1579-1588.

SPAF; AAA-protein; spermatogenesis; carcinogenesis; mitochondria

Introduction

Ectopic, inappropriate expression of cell type-specific genes is a common mechanism by which tumors may gain phenotypes that favor malignant cell growth and metastasis. One example of ectopic expression in tumor cells is mos, a germ cell-specific gene that has been documented to contribute to genetic instability in transformed cells, probably by mediating a meiosis-like phenotype in somatic cells (Fukasawa and Vande Woude, 1995). Thus, analysis of molecular events in normal tissue development provides not only basic knowledge of tissue developmental dysfunction but also insight into potential roles in tumorigenesis.

Spermatogenesis is one of the most intensively studied developmental systems. It is a dynamic process by which germ stem cells, spermatogonia, undergo complex morphological, biochemical, and physiological transitions that result in the production of fertile spermatozoa. There are three principle phases of spermatogenesis: a mitotic phase in which spermatogonia are replenished; a meiosis phase in which genetic material is recombined and segregated; and transformation phases, termed spermiogenesis, in which haploid spermatids undergo dramatic morphological changes that culminate in the release of highly polarized, flagellated spermatozoa into the lumen of the seminiferous tubule. The well-defined stages of spermatogenesis make it a good model for dissecting the regulation and function of certain oncogenes and tumor suppressor genes. The BRCA1 (Zabludoff et al., 1996) and p53 (Schwartz et al., 1993; Sjoblom and Lahdetie, 1996) tumor suppressor genes are expressed in spermatocytes during meiosis, consistent with their roles in DNA recombination and DNA repair. The c-myc, c-fos, and c-jun oncogenes (Wolfes et al., 1989) are expressed specifically in highly proliferative spermatogonia. While this information outlines potential parallels in spermatogenesis and tumorigenesis, more information is needed to elucidate the molecular events of spermatogenesis and how their deregulation might contribute to malignant transformation.

We have developed a clonal epidermal cell model of chemical carcinogenesis which includes cells at different stages of carcinogenesis: (1) normal keratinocyte progenitor (291), (2) pre-tumorigenic initiated cell (03C), (3) a benign papilloma (09RAT), and (4) a poorly differentiated squamous cell carcinoma which is invasive and metastatic (03RAT) (Kulesz-Martin et al., 1988). Analysis of known oncogene and tumor suppressor gene expression by Northern blotting in the initiated cell clones and their tumorigenic derivative cells, compared with normal clone 291, indicated deregulated expression of retrovirus-like sequence VL30 (Han et al., 1990) and tumor suppressor gene p53 (Han and Kulesz-Martin, 1992) at the point of malignant conversion. Unlike most DMBA-induced tumors in carcinogenesis models in vitro, no H-ras mutation was found in this model (Schneider et al., 1993). Although the loss of p53 is the most frequent genetic event in malignant transformation, it appears to be insufficient for tumor formation (Kemp et al., 1993). Therefore, altered gene expression was analysed by RNA differential display in an attempt to identify other oncogenic events that can contribute to malignant transformation (Liu and Kulesz-Martin, 1998).

Two differentially expressed genes at malignant conversion were identified. One cDNA fragment was decreased in the carcinoma and was essentially identical to the 3' region of ARK, an adhesion-related kinase reported to be involved in homophilic cell aggregation. Another differentially displayed cDNA fragment that is altered and overexpressed at the malignant conversion stage of carcinogenesis was the novel AAA-protein SPAF. The current report describes the cloning and initial characterization of SPAF full length cDNA and protein.

Results

Expression of SPAF in the epidermal cell model and in mouse tissues

SPAF expression in the epidermal cells was analysed with a 415 bp cDNA fragment of SPAF cDNA, first detected in 03RAT epidermal carcinoma cells by RNA differential display (Liu and Kulesz-Martin, 1998). Two mRNA species corresponding to approximately 3 and 1.7 kb were detected (Figure 1). As predicted by differential display, the expression level of the 3 kb transcript was significantly higher in the poorly differentiated squamous cell carcinoma 03RAT than in the normal clone 291 and the independently initiated epidermal tumor cells 05RAT and 09RAT. The Northern analysis of the tissue distribution of the transcripts showed predominant expression of the 3 kb transcript in testis (Figure 2), suggesting that it is a testis-specific gene despite marginal signals in spleen, skin, and 291 samples. The expression of the 1.7 kb mRNA was restricted to this poorly differentiated squamous cell carcinoma, being undetectable in the normal tissues examined. In contrast to the signal of the large transcript, the hybridization intensity of the small transcript varied along with the probes from the different segments of SPAF cDNA. The signal of the smaller transcript was comparable to that of the larger transcript when hybridized with the probe from the 3' end (Figure 2), but was attenuated with the full length SPAF cDNA as probe (Figure 6). No signal for the smaller transcript was detected with a 5' SPAF segment (224-1235 bp) as probe (data not shown), suggesting that this 1.7 kb transcript represents the 3' portion of SPAF cDNA generated by either alternative splicing or truncation/deletion of one SPAF allele in the 03RAT cells.

Since SPAF was predominantly expressed in normal testis, its normal full length cDNA was isolated by screening a mouse testis cDNA library. The library was constructed from mixed stage germ cells collected from 10-12-week-old CD1 mouse testes (Stratagen # 937308). Three positive clones were obtained by screening 1´10⁶ phage plaques, and two, being close to 3 kb in size, were selected for further characterization (SPAF1 and SPAF2 in Figure 3). The SPAF1 cDNA is 2798 bp in length, containing a putative open reading frame of 849 amino acids. SPAF2 is 122 bp shorter than SPAF1 and encodes a putative polypeptide of 820 amino acids. There is no stop codon in the upstream sequence of the putative start codon, suggesting that the cDNA clones isolated from the library were still partial ones. Therefore, two SPAF-specific anti-sense primers, RT3 (471-451 bp of SPAF) and PE3 (267-240 bp of SPAF), at the positions of the arrowheads in Figure 3, were designed for the amplification of the 5'-end of SPAF cDNA by ligation-anchored polymerase chain reaction (Troutt et al., 1992). A 267 bp DNA fragment was amplified and cloned into pBluescript plasmid digested with NotI and SmaI. The construction of the full length SPAF cDNA was achieved by taking advantage of the presence of a NotI site in the 5' anchor primer and a unique BstEII site at position 186 of SPAF cDNA that overlaps sequences of the 267 bp amplified product and the 2798 bp cDNA isolated from the library (Figure 3).

Sequence analysis of SPAF cDNA

The SPAF cDNA constructed from the cDNA library clone (106-2904) plus RACE-PCR product (1-105) totalled 2904 bp and contained a long open reading frame between nucleotide position 49 and 2725, encoding a polypeptide of 892 amino acids (Figure 4). To confirm the validity of the predicted open reading frame (ORF), the full length SPAF cDNA was expressed by in vitro transcription/translation in the presence of [³⁵S]methionine and subjected to SDS polyacrylamide gel electrophoresis. A product of 97 kDa was detected, as expected by calculation from the deduced amino acid sequence of SPAF. Since the first ATG codon (A located at nt 44) is not surrounded by a typical translation start site consensus sequence (Kozak, 1991), the authentic initiation codon of SPAF was examined. Two 5'-truncated forms of SPAF, SPAF1 and SPAF2, were made to eliminate the first and the second ATG codon (A located at nt 173), encoding putative polypeptides of 95 and 90 kD, respectively (Figure 5a). If the second/third downstream ATG in SPAF was the authentic initiation codon, the polypeptide generated from the full length SPAF cDNA would migrate at the same position as that of SPAF1 or SPAF2. Figure 5b shows that the molecular weight of the truncated forms decreases in proportion to the size of the deletion, demonstrating that the first ATG is the initiation codon of SPAF cDNA. As no termination codon in frame was found in the 5' region upstream of the presumed initiation codon, the full length ORF was confirmed by comparing the sizes of in vitro translated SPAF proteins with endogenous SPAF protein from normal adult mouse testis by immunoblotting analysis. Because the amounts of in vitro translated proteins were too low to be detected by immunoblotting, the in vitro translated SPAF proteins were enriched by immunoprecipitation with anti-SPAF antibody SpAb before immunoblotting. The protein translated from the full length SPAF cDNA migrated at the same position as its endogenous counterpart from testis lysate (Figure 5c), indicating that the cloned SPAF cDNA contains full length coding sequence of the SPAF gene. High GC content (69%) was found in the 5'-untranslated region (UTR), which may impede translation as indicated by increased translation efficiency of 5'-truncation mutants (compare the signal intensity of SPAF1 and 2 to SPAF in Figure 5b). A typical polyadenylation signal, AAUAAA, is found immediately upstream of the poly (A) tail (underlined in the cDNA sequence in Figure 4). The 3'-UTR is relatively high in AU content (70%) and contains four AUUU destabilizing elements found in many rapidly degraded mRNAs (Shaw and Kamen, 1986). These features suggest that the expression of SPAF is highly regulated both at the post-transcriptional and translational levels. A database search using BLAST (National Center for Biotechnology Information) revealed that SPAF is a new member of the AAA-protein family (ATPase Associated with diverse cellular Activities). A duplicated nucleotide binding domain of approximately 200 amino acids containing an AAA-protein family signature (VxVLxATNx[4]IDxALxR) (Kunau et al., 1993) and the Walker A and B nucleotide binding motifs (Walker et al., 1982) was found in a region from 350 a.a. to 805 a.a of SPAF protein (Figure 4). The sequence similarities between SPAF and other members containing two ATPase modules range from 50-66% (identity of 25-45%) over the whole length of the proteins (Table 1). The ATPase modules are highly conserved.

Compared with other AAA-proteins, SPAF protein showed several unique features. Firstly, the N-terminal region is approximately 130 a.a. longer than other closely related proteins. Secondly, the extreme N-terminal portion of SPAF is rich in positively charged arginine, lysine, and hydroxylated serine, implying a putative mitochondrial matrix-targeting sequence (Hartl et al., 1989). Thirdly, two putative cdc2 phosphorylation sites, SPSS and TPCK, (Nigg, 1991) lie side by side upstream of the first ATPase module. The C-terminal region of SPAF lacks a DDDLY motif which is associated with a ring-like structure found in VCP, p97, and CDC48 of the AAA-protein family and many heat shock proteins (Mian, 1993), suggesting that SPAF has a different quaternary structure.

Developmental expression pattern of SPAF in normal testis

Since testis was the predominant normal tissue examined to have SPAF expression (Figure 2), the expression pattern of normal SPAF in testis became a focus of study. The testis is composed mainly of germ cells of three categories: spermatogonia, spermatocytes, and spermatids, which represent progressively more mature phases of spermatogenesis. In mice, spermatogenesis begins shortly after birth, and the composition of the seminiferous epithelium undergoes predictable changes through development (Bellve et al., 1977). On postnatal day 8, the seminiferous tubule contains 27% spermatogonia and 73% Sertoli cells; on day 14, it contains 12% spermatogonia, 51% spermatocytes (including cells up to the pachytene stage), and 37% Sertoli cells; in the adult, it contains 4% spermatogonia, 22% spermatocytes, 71% spermatids, and 3% Sertoli cells. To examine which spermatogenic cells express SPAF, we analysed the expression of SPAF in adult epididymis (harboring primarily mature sperm) and testis from prepubertal and adult mice with full length SPAF cDNA as a probe. High levels of SPAF mRNA were detected by Northern blotting in all the RNA samples from testis but not from epididymis (Figure 6). Densitometric analysis showed 0.5-fold and twofold higher steady-state expression of SPAF RNA in 7 day and 14 day old testes compared with adult testis. The peak SPAF expression was proportional to the maximal percentages of spermatogonia and spermatocytes during development, indicating that SPAF is expressed in spermatogonia and spermatocytes, but not spermatids or sperm.

In general, the regulation of transcription is the major control of gene expression, and the presence of mRNA often indicates the expression of protein. However, it has been observed that transcription and translation are uncoupled for some genes during spermatogenesis (Braun et al., 1989). Therefore, the expression patterns of SPAF in spermatogenic cells were analysed by immunoblotting using the anti-SPAF antibodies (SpAb). The 97 kD SPAF protein was detected in both prepubertal and adult testis but not in epididymis (Figure 6). The highest expression of SPAF was observed in the prepubertal testis, consistent with the Northern analysis data. Interestingly, two bands corresponding to 97 and 70 kD were detected. The 97 kD band comigrated with in vitro translated SPAF protein and was expressed in the positive control cell 03RAT but not in negative control 291, indicating that this band represents cellular SPAF protein. Two possible explanations for detection of the smaller protein are SPAF degradation or the presence of an unrelated protein that shares one or more epitopes recognized by this antibody. It seems more likely to be antibody recognition of an unrelated protein since the 70 kD protein was present in epididymis without detectable SPAF protein or SPAF mRNA.

Cellular localization of SPAF protein

In order to determine the expression and cellular location of SPAF in cells undergoing spermatogenesis, immunohistochemical staining was performed in sections of testis from prepubertal and adult mice (Figure 7). The SPAF protein was found in germ cells in the basal compartment of seminiferous tubules containing spermatogonia and early spermatocytes up to the zygotene stage. The level of SPAF protein dramatically decreased in pachytene spermatocytes and spermatids in the adluminal compartment. No staining was observed in somatic cells of the testis, including Sertoli and interstitial cells, or in mature sperm in the epididymis. In the sections from testes of 2-week-old mice, all the spermatogenic cells were positively stained. This was consistent with results from adult testis since germ cells develop up to the pachytene stage at postnatal day 14. The intracellular staining pattern indicated that SPAF protein was localized in the cytoplasm rather than in the nucleus.

Subcellular localization of SPAF protein

To gain further insight regarding the subcellular compartmentalization of SPAF protein, sections of testis were analysed by indirect immunofluorescence staining and confocal microscopy. In addition to being diffusely distributed throughout the cytoplasm of germ cells, SPAF protein was also detected in discrete spheroidal structures (Figure 8a), suggesting that it is localized in organelles. The presence of a mitochondrial matrix targeting sequence in SPAF protein led us to examine its localization in mitochondria by co-staining with anti-hsp60 antibody, a mitochondrial-specific antibody. Instead of overlapping secondary fluorochromes, fluorochromes were mutually excluded, i.e. stained by either SpAb or anti-hsp-60, but not both (data not shown). It is conceivable that SPAF protein is absent from hsp-60 positive cells, but unlikely that SPAF-positive cells lack mitochondria. Therefore, it appears likely that SPAF and hsp60 co-localize in mitochondria and fail to co-stain due to steric hindrance of antibodies or to fluorescence quenching. To further examine this localization, sections of mouse testis were stained for SPAF protein using a pre-embedding ultrastructural immunoperoxidase technique. No immunoreactivity was detected in testis sections permeabilized by freeze-thawing followed by incubation with saponin, whereas staining was observed following permeabilization with the stronger detergent Triton X-100. The requirement for strong detergent is consistent with SPAF protein (or at least its N-terminal portion) being in internal organelles, where exposure of antigenic determinants requires stronger permeabilization not only of the cell surface membrane, but also of the organelle membranes. Examination of ultrathin sections of positively stained testis revealed immunoreactivity in mitochondria (Figure 8b). The reaction product appeared to diffuse toward the matrix and was confined within the mitochondria, with no staining diffused outward, suggesting that SPAF protein is localized in the inner membrane or matrix of mitochondria. The staining intensity varied among mitochondria, likely due to differential accessibility of the antigen. Attempts were made to localize SPAF using post-embedding techniques, in order to avoid problems of antigen accessibility. However these failed, due to destruction of the antigenicity of the protein.

Discussion

In this study, a testis-specific gene that is altered and overexpressed in a poorly differentiated squamous cell carcinoma was identified. Sequence analysis revealed that it is a novel member of the AAA-protein family (ATPase Associated with diverse cellular Activities). The AAA-protein family is an emerging superfamily with diverse functions (Confalonieri and Duguet, 1995). The members of this family are characterized by having one or two copies of a highly conserved ATPase module of approximately 200 amino acids with consensus sequences for ATP binding and hydrolysis. Members of this gene family are present in prokaryocytes and eukaryotes, including bacteria (Confalonieri et al., 1994), yeast (Frohlich et al., 1991), plants (Feiler et al., 1995), and mammals (Egerton et al., 1992). They function in a variety of cellular processes, including vesicle-mediated transport (Eakle et al., 1988; Wilson et al., 1989; Babst et al., 1997), cell cycle regulation (Frohlich et al., 1991; Clark-Maguire and Mains, 1994), peroxisome biogenesis (Erdmann et al., 1991), transcriptional regulation (Nelbock et al., 1990), and membrane fusion (Rabouille et al., 1995; Acharya et al., 1995). In S. cerevisiae alone, 22 different proteins with AAA-protein motifs have been identified, suggesting that there are at least as many biological functions associated with this family (Beyer, 1997). SPAF belongs to the subfamily of AAA-proteins with two ATPase modules, showing overall 50-66% similarity within this group. Although the members in this group show disparate phenotypes, most of the proteins for which functions have been defined are linked to membrane fusion events of various organelles (Table 1). The first protein associated with membrane fusion is SEC18, whose mutant blocks the fusion of transport vesicles with Golgi cisternae in the protein secretory pathway (Eakle et al., 1988). Its equivalent in higher organisms is NSF (N-ethyl-maleimide-sensitive fusion protein), an essential component of SNAP/SNARE fusion machinery that mediates protein transport from the ER to the Golgi complex (Wilson et al., 1989). PAS is another member of the subfamily involved in membrane fusion events, whose mutant is incapable of forming peroxisomes from smaller precursors (Erdmann et al., 1991). Recently, two other kinds of membrane fusion events, ER fusion and Golgi cisternae formation, have been associated with AAA-proteins CDC48 (Latterich et al., 1995) and p97ATPase (Rabouille et al., 1995; Acharya et al., 1995), respectively. To date, however, none of the members in this subfamily have been associated with the membrane fusion events of mitochondria, one of the most abundant membrane-containing cellular organelles. The overall similarities between SPAF and the above mentioned proteins are less than 66%, in accord with the idea that SPAF is involved in a distinct membrane fusion event. The presence of a putative mitochondrial matrix-targeting sequence, and evidence for localization in mitochondria, suggest that SPAF may be involved in membrane fusion events of mitochondrial morphogenesis.

Thus far, we have no direct evidence for SPAF involvement in mitochondrial membrane fusion events; however its expression pattern during spermatogenesis provides clues regarding a possible role in mitochondrial morphogenesis. The immunostaining results show that SPAF protein is expressed in the mitochondria of spermatogonia and early spermatocytes in the basal compartment, but not in spermatids and late spermatocytes in the adluminal compartment. Such expression patterns in the seminiferous tubule parallel changes in mitochondrial morphology during spermatogenesis from an 'orthodox' form in the basal cells to a 'condensed' form in the adluminal cells (De Martino et al., 1979). This suggests an association of SPAF with 'orthodox' mitochondrial morphogenesis. The significance of this mitochondrial morphological transformation remains to be elucidated. However, condensed mitochondria have been associated with higher oxidative phosphorylation efficiency (De Martino et al., 1979; Hackenbrock et al., 1971). Therefore, it is possible that mitochondrial transformation might be a cellular mechanism to compensate for decreased oxygen supply to cells in the adluminal compartment due to the blood-testis barrier (Waites and Gladwell, 1982). Compared with condensed mitochondria, the architecture of orthodox mitochondria has a number of thermodynamically unfavorable features, such as ovoid shape and extended surface area of the inner membrane. SPAF, as an AAA-protein containing two ATPase modules, may function by modulating protein-protein interactions in an ATP-dependent manner to accomplish and maintain such a thermodynamically unfavored architecture. We hypothesize that SPAF is susceptible to proteolytic degradation triggered by decreased oxygen tension in the mitochondria, which may result in the remodeling of mitochondria to a condensed form. In the absence of SPAF, the remodeling of mitochondria from orthodox form to condensed form may occur spontaneously, since condensed mitochondria have several thermodynamically stable features, including spheroid shape, small size, and less surface area due to lack of cristae. Condensed mitochondria have been observed not only in spermatogenic cells in the adluminal compartment, but also in a number of rapidly growing tumors (White et al., 1974; Hruban et al., 1966, 1973). Thus, rapidly growing tumors like 03RAT may take advantage of ectopic expression of SPAF to adopt a spermatogenic cellular mechanism for undergoing rapid clonal expansion in a hypoxic environment. Generation of SPAF knock-out mice by homologous mutagenesis will be needed to provide direct evidence regarding the role of this gene product in mitochondrial morphogenesis, spermatogenesis, and tumorigenesis.

In addition, it will be intriguing to identify the regulatory elements responsible for SPAF tissue dependent and cell maturation stage-dependent expression in testis. In particular, whether SPAF is subject to AP1 regulation might be investigated, since SPAF is expressed at the same stages of spermatogenesis and tumorigenesis during which AP1 is also active. Finally, the overexpression of SPAF in the poorly differentiated squamous cell carcinoma must be extended to studies of SPAF expression in independently derived epidermal tumors and other tumor cell types, in order to ascertain its general relevance to tumorigenesis of testis and other epithelia.

Materials and methods

cDNA cloning

A mouse testis cDNA library constructed using the uni-Zap vector with oligo(dT)-primed cDNA inserts (Stratagene) was screened using an ³²P-dCTP-labeled partial cDNA fragment (417 bp) obtained from RNA differential display of a poorly differentiated squamous cell carcinoma and its initiated and normal precursors (Liu and Kulesz-Martin, 1998). A DNA probe was prepared by random labeling of the partial cDNA fragment using ³²P-dCTP (Amersham) following the manufacturer's protocols. SPAF cDNA fragment in pBluescript SK^- vector (pBSPAF) was excised from positive uni-Zap phage isolated from a mouse testis cDNA library by co-infection with filamentous helper phage (Stratagene). The 5'-end of SPAF was amplified by means of the ligation-anchored polymerase chain reaction (Troutt et al., 1992). Two oligonucleotide primers complementary to SPAF mRNA were synthesized: a 21-mer (I; 5'-TTCCTGAGCCTGAAGCACAGC-3') and a 28-mer (II; 5'-CATAGTGTTAAGTCCAAGATGAACAAGG-3'), representing antisense sequences corresponding to 471-451 bp and 267-240 bp of SPAF cDNA, respectively. Ligation-anchored PCR was carried out using the Marathon cDNA Amplification Kit (Clontech). Adaptor-ligated double-stranded-cDNA was prepared from adult mouse testis poly(A)⁺ mRNA, following the Clontech protocol. Nested PCR was performed with the 28-mer and an adaptor primer using pfu DNA polymerase (Stratagene) at: 1 cycle of 94°C for 5 min; 30 cycles of 94°C for 30 s, 63°C for 1 min, and 72°C for 2 min; and 1 cycle of 72°C for 5 min. The amplified products were gel purified using the QIAEX II gel extraction kit (Qiagen). The purified DNA was digested with NotI and ligated into pBluescript(KS) (Stratagene) digested with NotI and SmaI. The ligation mixture was used to transform E. coli strain XL1-Blue, and clones containing inserts were selected by blue-white color selection and confirmed by PCR using the 28-mer and a sense primer corresponding to bp 113-133. The full length SPAF cDNA was constructed by subcloning NotI and BstEII digested RACE product into pBSPAF digested with the same enzymes.

DNA sequencing and analysis

SPAF cDNA was digested with SacII and XhoI to generate 3' and 5' protruding ends which are resistant and sensitive, respectively, to digestion with exonucleaseIII, in order to generate serial deletions. The overlapping deletions thus generated were ligated into a cloning vector and sequenced with M13 reverse primer using an automatic DNA sequencer (RPCI Biopolymer facilities). DNA sequence analysis, including searches for homology, motifs, and structure predictions, was performed using GCG programs (Wisconsin Package V.8.0).

In vitro translation

For testing of protein products, the SPAF cDNAs, 1 to 2904 (SPAF), 106 to 2904 (SPAF1), and 234 to 2904 (SPAF2) were cloned into the NotI/XhoI, SacII/XhoI, and BamHI/XhoI sites, respectively, of pBluescript KS under the T₇ phage promoter. The constructs were linearized with XhoI and transcribed with T₇ RNA polymerase using the mCapping kit (Stratagene). The transcripts were translated in the present of [³⁵]S-methionine (Amersham) according to standard protocols (Promega). Five l of each of the in vitro translated proteins were fractionated on a 7.5% SDS-polyacrylamide gel and visualized by autoradiography.

Northern blot analysis

Cytoplasmic RNAs from cells at approximately 70% confluency in 150 mm dishes were extracted with Trizol Reagent (GIBCO/BRL), following the manufacturer's manual. A 10 g aliquot of total RNA was separated in a 1.2% agarose, 1 M formaldehyde gel and transferred to Nytran Plus membrane (Schleicher & Schuell), according to standard protocols (Sambrook et al., 1989). Membranes were hybridized to ³²P-dCTP-labeled probe in 50% formamide at 42°C overnight. Blots were washed to a stringency of 0.1´SSC/0.5% SDS/65°C. Autoradiography was performed for 1 h to 2 days using Kodak X-Omat AR film. The membranes were rehybridized to ³²P-dCTP-labeled 7S RNA as a loading control (Balmain et al., 1982). RNA quantification was performed by scanning the autoradiograms by means of a Computing Densitometer (Molecular Dynamics) and analysed by using ImageQuant software (Molecular Dynamics).

Production of SPAF-specific antibodies

Polyclonal antibodies were raised against the SPAF N-terminus upstream of the first ATP binding module as follows: a fragment corresponding to amino acids 69-389 was cloned in-frame to the poly-histidine tag in plasmid pProEX^TMHTb (BRL/GIBCO). The resultant plasmid was used to transform E. coli strain DH5. Expression of the fusion protein was induced by adding 3 mM IPTG (isopropyl--D-thiogalactopyranoside) to exponentially growing cultures. The bacteria were homogenized in lysis buffer (10 mM Tris-HCl [pH 8.3], 100 mM NaCl, 5 mM EDTA, 1 mM PMSF) containing 0.1 mg/ml lysozyme, and by passing through a 22G syringe needle. Inclusion bodies were solubilized in 1% sarcosol and neutralized with 5% Triton-X 100. The solubilized protein was dialyzed in the lysis buffer and subjected to nickel affinity chromatography. The bound protein was eluted with lysis buffer in the presence of 100 mM imidazol. The eluted fusion protein was dialyzed against phosphate-buffered saline (PBS) and used for the generation of polyclonal antibodies by standard protocols. Three New Zealand White female rabbits (RPCI Laboratory Animal Resources) were each injected subcutaneously with 200 g of purified protein emulsified with an equal volume of Freund's complete adjuvant. One month after the primary injection, boost injections were repeated every 2 weeks with 150 g of purified protein emulsified in incomplete Freund's adjuvant. Antisera were tested for ability to precipitate in vitro translated SPAF protein. The polyclonal antibodies were affinity purified using 6 mg of purified fusion protein coupled to an Aminolink column (Pierce), following the manufacturer's protocol.

Western immunoblotting

Mouse tissues were homogenized in Dounce tissue grinders in PBS and lysed in lysis buffer (150 mM NaCl, 50 mM Tris [pH 8.0], 1% NP-40, 1 mM PMSF, 50 M aprotinin, and 10 g/ml leupeptin). Cultured cells at 70% confluence were harvested in lysis buffer. Lysates were centrifuged for 15 min at 14 000 r.p.m. Twenty g of protein quantitated by using the Protein-assay (Bio-Rad), were incubated for 5 min at 100°C in sample buffer (100 mM Tris HCl [pH 6.8], 20% glycerol, 2% SDS, and 0.1% bromophenol blue) containing 100 mM DTT and separated in a 7.5% SDS-polyacrylamide minigel. After electroblotting onto nitrocellulose (Schleicher & Schull), membranes were blocked overnight in TBST (50 mM Tris-HCl [pH 7.5], 200 mM NaCl, 0.1% Tween 20) containing 5% nonfat milk powder, then incubated with affinity purified anti-SPAF polyclonal antibody (SpAb) in TBST milk for 1 h at room temperature. After washings in TBST, horseradish peroxidase-conjugated protein-A (Amersham) was added for 1 h and immuno-complexes were visualized by chemiluminescence (Pierce).

Immunochemical staining

Five m paraffin sections of formalin-fixed mouse tissues were deparaffinized in xylene and rehydrated through a graded series of ethanol concentrations. The sections were treated in a microwave oven set at medium for 10 min to unmask antigenic determinants, then permeabilized in 0.3% Triton-X 100/PBS for 10 min. Nonspecific antibody binding was blocked with PAB (PBS, 0.1% sodium azide, 0.5% BSA) containing 10% normal goat serum (Vector Laboratories). The sections were then incubated overnight at 4°C with affinity-purified SPAF-specific antibody (0.1 g/ml). After several washes in PBS containing 0.1% Tween 20, sections were incubated with biotin-labeled goat anti-rabbit IgG (Vector Laboratories), then with strepavidin-peroxidase conjugate (Research Genetics). Color was developed with diaminobenzidine tetrahydrochloride (DAB); slides were washed in H₂O, counterstained with hematoxylin, and mounted in Permount (Fisher). Preimmune antibody at 1 g/ml was used as a control for primary antibodies.

Immunofluorescence staining

Mouse testes were fixed in freshly depolymerized 2% paraformaldehyde fixative for 2 h at 4°C. Tissues were then washed in 50 mM ammonium chloride (three times, 15 min) and cryoprotected in 30% sucrose/PBS overnight at 4°C. Tissue was embedded in Tissue Tek OCT medium (ICN Biomedicals) and frozen in 2-methylbutane/liquid nitrogen. Four m cryosections were obtained using a cryostat microtome (Reichert Jung), air-dried for 10 min, permeabilized in 0.2% Triton-X 100/PBS for 10 min, then incubated in SpAb (0.1 g/ml in PBS/Triton) in anti-hsp60 (1 g/ml in PBS/Triton) (Stressgen), or in a mixture of both antibodies, for 45 min at room temperature on a rocker platform. Following washes in PBS, sections were incubated with Texas-conjugated goat anti-rabbit secondary antibody (Amersham), in FITC-conjugated goat anti-mouse secondary antibody (Amersham), or in a mixture of both secondary antibodies for 30 min at room temperature and washed in PBS for 30 min. The sections were mounted with Aquamount (Polysciences) and observed using a Bio-Rad confocal microscope.

Pre-embedding ultrastructural immunoperoxidase technique

Mouse testes were fixed in periodate-lysine-paraformaldehyde (PLP) fixative for 5 h at 4°C, washed in 50 mM NH₄CL/PBS (three times, 15 min each), rinsed in PBS, and cryoprotected in 60% sucrose/PBS overnight at 4°C. The tissue was then frozen in 2-methyl butane cooled with liquid nitrogen, and 30-50 m cryosections were prepared using a cryostat microstat microtome. The sections were thawed in PBS/BSA (1 mg/ml) and permeabilized in 0.025% Triton X-100 for 30 min at room temperature. The free-floating sections were then incubated with anti-SPAF polyclonal antibody (diluted 1 : 100 in PBS/BSA/0.01% sodium azide) for 1 h at room temperature and overnight at 4°C, with gentle agitation. Following an additional 1 h incubation at room temperature, the sections were washed three times, 15 min each with PBS/BSA and incubated in biotinylated anti-rabbit antibody (Vector Laboratories) for 1 h at room temperature. Following three washes in PBS/BSA (15 min each), the tissue was incubated in avidin-peroxidase (Research Genetics) for 1 h at room temperature and washed three times again. The sections were then fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer containing 5% sucrose for 1 h at 4°C and washed in phosphate buffer/7.5% sucrose and overnight at 4°C. Following three washes in 50 mM Tris HCl/7.5% sucrose, the tissue was incubated in 1 mg/ml diaminobenzidine/Tris HCl/sucrose for 20 min at room temperature before adding H₂O₂ (0.01%) for an additional 15 min. Following washes in Tris (one wash for 15 min) and phosphate buffers (two washes for 15 min each), the tissue was post-fixed in 1% OsO₄ for 1 h at 4°C, dehydrated in a graded alcohol series, and embedded in Spurr's resin. Thick and ultrathin sections were viewed in a Leitz light microscope or a Siemens 101 electron microscope, respectively.

Acknowledgements

We thank Dr C Wenner, Dr AJ Kinniburgh, and Dr GT Bowden for critical comments. We also thank Dr P Welch for discussion and assistance in immunohistochemical analysis; Mr Joseph Brachman for immunization of rabbits; the RPCI DNA polymer facility for DNA sequencing and oligonucleotide synthesis; M Vaughan for histological assistance; and D Ogden, S Dave; and E Howley for the electron microscopy. This study was supported by NIH RO1 CA31101 and RPCI Institute Core CA16056.

Acharya U, Jacobs R, Peters JM, Watson N, Farquhar MG and Malhotra V. (1995). Cell 82, 895-904. MEDLINE

Babst M, Sato TK, Banta LM and Emr SD. (1997). EMBO J 16, 1820-1831. Article MEDLINE

Balmain A, Krumlauf R, Vass JK and Birnie GD. (1982). Nucleic Acids Res 10, 4259-4277. MEDLINE

Bellve AR, Cavicchia JC, Millette CF, O'Brien DA, Bhatnagar YM and Dym M. (1977). J Cell Biol 74, 68-85. MEDLINE

Beyer A. (1997). Protein Science 6, 2043-2058. MEDLINE

Braun RE, Peschon JJ, Behringer RR, Brinster RL and Palmiter RD. (1989). Genes Dev 3, 793-802. MEDLINE

Clark-Maguire S and Mains PE. (1994). Genetics 136, 533-546. MEDLINE

Confalonieri F and Duguet M. (1995). Bioessays 17, 639-650. MEDLINE

Confalonieri F, Marsault J and Duguet M. (1994). J Mol Biol 235, 396-401. MEDLINE

De Martino C, Floridi A, Marcante ML, Malorni W, Scorza Barcellona P, Bellocci M and Silvestrini B. (1979). Cell Tissue Res 196, 1-22. MEDLINE

Eakle KA, Bernstein M and Emr SD. (1988). Mol Cell Biol 8, 4098-4109. MEDLINE

Egerton M, Ashe OR, Chen D, Druker BJ, Burgess WH and Samelson LE. (1992). EMBO J 11, 3533-3540. MEDLINE

Erdmann R, Wiebel FF, Flessau A, Rytka J, Beyer A, Frohlich KU and Kunau WH. (1991). Cell 64, 499-510. MEDLINE

Feiler HS, Desprez T, Santoni V, Kronenberger J, Caboche M and Traas J. (1995). EMBO J 14, 5626-5637. MEDLINE

Frohlich KU, Fries HW, Rudiger M, Erdmann R, Botstein D and Mecke D. (1991). J Cell Biol 114, 443-453. MEDLINE

Fukasawa K and Vande Woude GF. (1995). Proc Natl Acad Sci USA 92, 3430-3434. MEDLINE

Hackenbrock CR, Rehn TG, Weinbach EC and Lemasters JJ. (1971). J Cell Biol 51, 123-137. MEDLINE

Han KA and Kulesz-Martin MF. (1992). Cancer Res 52, 749-753. MEDLINE

Han KA, Rothberg P and Kulesz-Martin M. (1990). Mol Carcinogenesis 3, 75-82.

Hartl FU, Pfanner N, Nicholson DW and Neupert W. (1989). Biochim Biophis Acta 988, 1-45.

Hruban Z, Kirsten WH and Slesers A. (1966). Lab Invest 15, 576-588. MEDLINE

Hruban Z, Mochizuki Y, Morris HP and Slesers A. (1973). J Natl Cancer Inst 50, 1487-1495. MEDLINE

Kemp CJ, Donehower LA, Bradley A and Balmain A. (1993). Cell 74, 813-822. MEDLINE

Kozak M. (1991). J Cell Biol 115, 887-903. MEDLINE

Kulesz-Martin MF, Penetrante R and East CJ. (1988). Carcinogenesis 9, 171-174. MEDLINE

Kunau WH, Beyer A, Franken T, Gotte K, Marzioch M, Saidowsky J, Skaletz-Rorowski A and Wiebel FF. (1993). Biochimie 75, 209-224. MEDLINE

Latterich M, Frohlich KU and Schekman R. (1995). Cell 82, 885-893. MEDLINE

Liu Y and Kulesz-Martin MF. (1998). Carcinogenesis 19, 683-686. MEDLINE

Mian I. (1993). Trends Biochem Sci 18, 125-127. MEDLINE

Nelbock P, Dillon PJ, Perkins A and Rosen CA. (1990). Science 248, 1650-1653. MEDLINE

Nigg EA. (1991). Semin Cell Biol 2, 261-270. MEDLINE

Rabouille C, Levine TP, Peters JM and Warren G. (1995). Cell 82, 905-914. MEDLINE

Sambrook J, Fritsch EF and Maniatis T. (1989). Molecular Cloning a laboratory manual. Cold Spring Harbor Laboratory Press Cold Spring Harbor, New York,

Schneider BL, Bowden GT, Sutter C, Schweizer J, Han KA, Kulesz-Martin and MF. (1993). J Invest Derm 101, 595-599. MEDLINE

Schwartz D, Goldfinger N and Rotter V. (1993). Oncogene 8, 1487-1494. MEDLINE

Shaw G and Kamen R. (1986). Cell 46, 659-667. MEDLINE

Sjoblom T and Lahdetie J. (1996). Oncogene 12, 2499-2505. MEDLINE

Troutt AB, McHeyzer-Williams MG, Pulendran B and Nossal GJ. (1992). Proc Natl Acad Sci USA 89, 9823-9825. MEDLINE

Waites GM and Gladwell RT. (1982). Physiol Rev 62, 624-671. MEDLINE

Walker JE, Saraste M, Runswick MJ and Gay NJ. (1982). EMBO J 1, 945-951. MEDLINE

White MT, Arya DV and Tewari KK. (1974). J Natl Cancer Inst 53, 553-559. MEDLINE

Wilson DW, Wilcox CA, Flynn GC, Chen E, Kuang WJ, Henzel WJ, Block MR, Ullrich A and Rothman JE. (1989). Nature 339, 355-359. MEDLINE

Wolfes H, Kogawa K, Millette CF and Cooper GM. (1989). Science 245, 740-743. MEDLINE

Zabludoff SD, Wright WW, Harshman K and Wold BJ. (1996). Oncogene 13, 649-653. MEDLINE

Figure 1 Northern analysis of SPAF expression in the cloned cell model of carcinogenesis. Ten g of total RNA from each cell type shown were separated by 1.2% agarose formaldehyde gel electrophoresis. The blot was hybridized with a random-primed ³²P-labeled SPAF cDNA fragment isolated by RNA differential display and autoradiographed at -70°C for 2 days. The blot was rehybridized with loading control 7S RNA and autoradiographed at room temperature for 2 h

Figure 2 Expression of SPAF in normal mouse tissues. Ten g of total RNA from the adult mouse tissues indicated were separated in a 1.2% agarose formaldehyde gel. The RNAs from 291 cells and 03RAT cells were included as negative and positive controls, respectively. The blot was hybridized with a random-primed ³²P-labeled SPAF cDNA fragment from RNA differential display and autoradiographed at -70°C for 3 days. For RNA loading control, the blot was rehybridized with 7S RNA and autoradiographed at room temperature for 2 h

Figure 3 SPAF cDNA was isolated by screening a mouse testis cDNA library with a partial cDNA fragment (7G2) isolated by RNA differential display as probe. SPAF1 and SPAF2 are the positive clones isolated from the screening. The 5' region of SPAF cDNA, from 1-106 bp, was extended by RACE-PCR with two SPAF-specific anti-sense primers (arrowheads). The full length SPAF cDNA was constructed by taking advantage of the presence of a unique BstEII site that overlaps the RACE product (RACE-5) and SPAF1 cDNA. A poly-histidine tagged fusion protein that represents the N-terminal region of SPAF protein (60-397 a.a.) was generated as immunogen for the production of anti-SPAF antibody

Figure 4 Predicted amino acid sequence of SPAF indicating the position of ATPase modules typical of AAA-proteins. A putative mitochondrial targeting sequence is underlined. Walker A and B motifs are in boxes. AAA-protein family signatures (S) are italicized and underlined. Bold letters indicate putative cdc2 phosphorylation sites. A putative polyadenylation signal is italicized. Genbank accession number AF049099

Figure 5 Size comparison of in vitro translated SPAF proteins with endogenous SPAF protein. (a) Plasmids containing SPAF cDNAs, 1-2904 (SPAF), 106-2904 (SPAF1), and 234-2904 (SPAF2) were constructed. The expected molecular weight of each SPAF is in parenthesis. (b) The plasmids were transcribed/translated in vitro in the presence of [³⁵S]methionine. The translated proteins were subjected to 7.5% SDS-polyacrylamide gel electrophoresis and visualized by autoradiography. (c) In vitro-translated SPAF proteins were compared with endogenous SPAF protein by immunoblotting with SpAb. Ten l of in vitro-translated SPAF proteins were enriched by immunoprecipitation with SpAb and subjected to 7.5% SDS-polyacrylamide gel electrophoresis along with 20 l of testis lysate from adult mice. The blot was incubated with SpAb followed by protein-A-conjugated horseradish peroxidase and visualized by chemiluminescence

Figure 6 (a) Northern blot analysis of SPAF expression during spermatogenesis. Ten g of total RNA from 1-, 2-week-old, and adult mouse testes, and adult epididymis were separated on a 1.2% agarose formaldehyde gel. The total RNAs from 291 and 03RAT were included as negative and positive controls respectively. The blots were hybridized with ³²P-labeled full length SPAF cDNA. A 7S RNA hybridization is shown as control for RNA loading. (b) Immunoblotting of SPAF protein expressed during spermatogenesis. Twenty micrograms of each lysate from adult epididymis and testes dissected from 1-, 2-week-old, and adult mice were fractionated in a 7.5% SDS-polyacrylamide minigel and transferred to a nitrocellulose membrane. The blot was incubated with SpAb followed with protein-A conjugated horseradish peroxidase and visualized by chemiluminescence

Figure 7 Localization of SPAF protein in mouse testis. Paraffin sections from adult mouse testis (a,c), and 2 week old testis (e), were immunostained with SpAb. Their serials sections were immunostained with preimmune (PreI) antibody as negative control (b,d,f). Letters: A, type A spermatogonia; Z, Zygotene spermatocye; D, spermatid; S, Sertoli cell; L, Leydig cell. The scale bar corresponds to 50 m

Figure 8 Subcellular localization of SPAF protein. (a) Immunofluorescence analysis of SPAF expression in mouse testis. The section was examined using a Bio-Rad MR600 scanning confocal microscope system. The positive staining of spheroidal structures in the cytoplasm is noted by arrows. The scale bar corresponds to 10 m. (b) Ultrastructural immunoperoxidase microscopic analysis of SPAF expression in mouse testis. The scale bar corresponds to 1 m

Table 1 Comparison of SPAF with other related AAA protein family members