Targeted Disruption of Intracellular Type I Platelet Activating Factor-acetylhydrolase Catalytic Subunits Causes Severe Impairment in Spermatogenesis*

Intracellular type I platelet activating factor-acetylhydrolase is a phospholipase that consists of a dimer of two homologous catalytic subunits α1 and α2 as well as LIS1, a product of the causative gene for type I lissencephaly. LIS1 plays an important role in neuronal migration during brain development, but thein vivo function of the catalytic subunits remains unclear. In this study, we generated α1- anda2-deficient mice by targeted disruption.α1−/− mice are indistinguishable from wild-type mice, whereas α2−/− male mice show a significant reduction in testis size. Double-mutant male mice are sterile because of severe impairment of spermatogenesis. Histological examination revealed marked degeneration at the spermatocyte stage and an increase of apoptotic cells in the seminiferous tubules. The catalytic subunits are expressed at high levels in testis as well as brain in mice. In wild-type mice, α2 is expressed in all seminiferous tubule cell types, whereas α1 is expressed only in the spermatogonia. This expression pattern parallels the finding that deletion of both subunits induces a marked loss of germ cells at an early spermatogenic stage. We also found that the LIS1 protein levels, but not the mRNA levels, were significantly reduced in α2−/− and double-mutant mice, suggesting that the catalytic subunits, especially α2, are a determinant of LIS1 expression level.

Platelet-activating factor (PAF) 1 is a potent signaling phospholipid involved in diverse physiological events, such as inflammation and anaphylaxis (1). In addition, PAF has been implicated in the central nervous system (2,3) and the reproductive system (4,5). PAF is hydrolyzed to an inactive metab-olite by a specific enzyme called PAF-acetylhydrolase (PAF-AH). At least three types of PAF-AH exist in mammals, namely the intracellular types I and II (6, 7) and a plasma type (8). Intracellular type I PAF-AH (PAF-AH (I)) is an oligomeric complex. It contains a dimer of two homologous catalytic subunits, ␣1 and ␣2, and a non-catalytic ␤ subunit (6, 9 -11). Interestingly, the ␤ subunit was later found to be identical to LIS1, the product of the causative gene for type I lissencephaly (10,12). Type I lissencephaly is a genetic brain malformation showing a smooth cerebral surface without gyri, caused by abnormal neuronal migration at early developmental stages. Mice homozygous for the Lis1 null mutation die early in embryogenesis soon after implantation (13). Heterozygous and compound heterozygous mice have expression level-dependent defects in neuronal migration (13). A series of recent studies has suggested that LIS1 interacts not only with PAF-AH (I) catalytic subunits but also with a number of proteins, including tubulin (14), cytoplasmic dynein (15,16), and NUDE (17)(18)(19)(20). Through interaction with these proteins, LIS1 plays important roles in microtubule-associated cellular functions such as mitotic cell division, chromosomal segregation, and neuronal migration. In contrast, the biological role of the catalytic subunits of PAF-AH (I) remains a complete enigma. Nothwang et al. (21) have described a case of functional hemizygosity of ␣1, possibly responsible for the resulting mental retardation, ataxia, and brain atrophy in this patient. Furthermore, Lecointe et al. (22) have proposed that deregulation of transcription of the human ␣2 gene is associated with the development of a certain lymphoma. Therefore, the PAF-AH (I) catalytic subunit is also likely to play an important role in some pathological conditions. The ␣1 and ␣2 catalytic subunits belong to a novel serine esterase family (9). These subunits, which show ϳ60% amino acid homology with each other, form homodimers and a heterodimer. Ho et al. (23) have reported the x-ray crystal structure of the ␣1 homodimer. The folding is unique among known lipases and phospholipases. The structure unexpectedly resembles those of the G-protein family such as p21 ras and G␣. To elucidate the in vivo function of the catalytic subunits of PAF-AH (I), namely ␣1 and ␣2, we generated mice lacking either one or both of these two proteins.

EXPERIMENTAL PROCEDURES
Generation of ␣1, ␣2 Mutant Mice-␣1, ␣2 genomic clones were isolated from a mouse 129/SvJ genomic library in the Lambda FIXII vector (Stratagene). Targeting vectors were constructed for replacing part of exon 2 and 3 of the ␣1 and ␣2 genes, which include a translation initiation site and catalytic motif (GXSXV) with a PGKneobpA cassette (24). A PGKDTA (diphtheria toxin 〈 fragment) cassette was inserted at the 3Ј-end of the short arm for negative selection. The targeting vectors were linearized and electroporated into ES cell line RW4 (Genome Systems), which was cultured on neomycin-resistant mouse embryonic fibroblasts. G418-resistant colonies were screened for homologous recombinants by PCR. Candidates of homologous recombinants were verified by Southern analysis using fragments at the 3Ј-ends of the genes, external to the targeting vectors as probes. Chimeric mice were generated by injection of the ES cells into C57BL/6N blastocysts, followed by transfers to foster mothers, and backcrossed to C57BL/6N mice. Genotypes were determined by PCR and/or Southern analysis of the tail DNA samples.
Antibodies-Mouse monoclonal antibodies against ␣1 and ␣2 were established as follows. ␣1 Ϫ/Ϫ and ␣2 Ϫ/Ϫ female mice were immunized with each purified recombinant rat protein with Freund's complete adjuvant (DIFCO), followed by six boosters at 2-week intervals with 20 g of protein and established monoclonal antibody, producing hybridoma cell lines as previously described (11). Monoclonal antibody against LIS1 (clone 338, a kind gift from Dr. O. Reiner, Weizmann Institute, Rehovot, Israel) and ␣-tubulin (clone DM1A, Sigma) were used for a Western blot analysis. Polyclonal antibody against LIS1 (N-19, Santa Cruz Biotechnology) was used for an immunohistochemical analysis.
Histological and Immunohistochemical Analyses-Testes were dissected and fixed overnight in Bouin's fixative at 4°C. Paraffin sections (5 m) were prepared and stained with Periodic Acid-Schiff (PAS) and hematoxylin. For immunohistochemistry, mice under anesthesia were perfused with phosphate-buffered saline (PBS) and then with 4% paraformaldehyde in PBS. Testes were dissected and refixed overnight in 4% paraformaldehyde at 4°C. Paraffin sections (5 m) were boiled in a microwave oven in 10 mM sodium citrate buffer (pH 6.0) for antigen retrieval. Subsequent immunodetection was performed using a Vector M.O.M. immunodetection kit (Vector Laboratories) for ␣1 and ␣2 and Vectastain ABC kit (Vector Laboratories) for LIS1. Immunostaining was visualized using diaminobenzidine and counterstained with hematoxylin. For detection of apoptotic germ cells, Bouin's-fixed, paraffinembedded testis sections were subjected to TUNEL staining using an in situ cell death detection kit, POD (Roche Molecular Biochemicals) according to the manufacturer's instructions.
Western Blot Analysis-Tissues were homogenized in quadruple volumes (w/v) of SET buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 250 mM sucrose) with protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 mg/ml pepstatin, 10 mg/ml leupeptin, 10 mg/ml aprotinin) and phosphatase inhibitors (50 mM NaF, 10 mM Na 3 PO 4 ). After centrifugation at 1,000 ϫ g at 4°C, the supernatants were used as the total protein lysates. The protein concentrations of samples were determined by the BCA assay (PIERCE). Each total protein lysate (50 g/lane) was separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes or nitrocellulose membranes. The membranes were blocked with 5% (w/v) skim milk (Wako) in TTBS buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.05% (w/v) Tween 20) and incubated with antibodies in TTBS. Chemiluminescence (ECL kit, Amersham Biosciences) was used for analyzing levels of protein according to the manufacturer's instructions.
Northern Blot Analysis-Total RNA was extracted from mouse tissues using Isogen (Nippongene). Total RNA (10 g/lane) was separated by 1% agarose-formaldehyde gel electrophoresis and transferred to Hybond-N membranes (Amersham Biosciences) in 20ϫ SSC. The membranes were hybridized in Rapid-hyb buffer (Amersham Biosciences) at 65°C and washed with 0.5ϫ SSC, 0.1% SDS at 65°C. Probes for ␣1, ␣2, and LIS1 were obtained by RT-PCR from mouse RNA and labeled by 32 P[dCTP] (Amersham Biosciences) using the Rediprime II DNA labeling system (Amersham Biosciences). Membranes were stripped and rehybridized with human glyceraldehyde-3-phosphate dehydrogenase cDNA probe (Clontech) to ensure equal loading.

RESULTS
We established monoclonal antibodies against ␣1 and ␣2 by immunizing the respective knockout mice. A Western blot analysis of adult mouse tissue shows that ␣1, ␣2, and LIS1 were most abundantly expressed in brain and testis ( Fig. 1). Expression of ␣2 and LIS1 were observed in other tissues as well, whereas ␣1 expression was restricted to embryonic brain and adult testis (Fig. 1). The expression levels of ␣2 and LIS1 were observed to be essentially proportional to that of ␣-tubulin, a component of microtubules. Because LIS1 plays an important role in microtubule dynamics (14,16), it can be postulated that both ␣ subunits are also involved in this process.
Immunohistochemical staining of adult mouse testes revealed that ␣2 and LIS1 immunoreactivity was present in all seminiferous tubule cell types (Fig. 2, D and G). Intense staining of ␣2 and LIS1 was observed in meiotically dividing spermatocytes and elongating spermatids. In contrast, ␣1 staining was restricted to the cells lining the basal compartment of seminiferous tubules ( Fig. 2A). Magnification revealed that ␣1 was specifically localized in spermatogonia cytoplasm (Fig. 2B,  arrow), whereas ␣2 and LIS1 were expressed in the cytoplasm of all types of spermatogenic cells and Sertoli cells (Fig. 2, E and H), suggesting that ␣1 is involved specifically in proliferation and/or differentiation of spermatogonia. LIS1 was also localized at meiotic spindles of spermatocytes (Fig. 2H, arrowhead) and manchettes of elongating spermatids (Fig. 2H, arrow), both of which are specific microtubule structures. No staining of ␣1 or ␣2 was detected in the seminiferous tubules of null mutant mice (Fig. 2, C and F).
We used homologous recombination in embryonic stem cells to generate mice lacking the ␣1 (Pafah1b3) and a2 (Pafah1b2) genes. Parts of exon 2 and exon 3 of each gene, including the translation initiation site and the catalytic serine residue, were replaced with a neomycin-resistance gene (Fig. 3A). Targeted embryonic stem cell clones and subsequent germ line transmissions were detected by PCR and/or Southern blot analysis (Fig.  3B). Both ␣1 Ϫ/Ϫ and ␣2 Ϫ/Ϫ mice were born with the expected Mendelian frequencies, viable and apparently indistinguishable from their wild-type littermates. Western blot analysis of the brain and the testis homogenates showed no immunoreactive bands in either ␣1 Ϫ/Ϫ or ␣2 Ϫ/Ϫ mice (Fig. 6A). ␣1 Ϫ/Ϫ / ␣2 Ϫ/Ϫ mice were also viable and apparently indistinguishable from wild-type mice. However, ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ males were found to be infertile, whereas female fertility was not affected. Testes weights of 5-week-old ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice were significantly (ϳ35%) smaller than those of wild-type mice (Table I). Testes weights were not noticeably reduced in ␣1 Ϫ/Ϫ mice, whereas they were reduced to 60% in ␣2 Ϫ/Ϫ mice (Table I). There was no significant difference in body weight among any of the genotypic combinations (Table I).
The histology of mutant testes was examined at age 5 weeks, the time when the first wave of murine spermatogenesis is completed. The seminiferous tubules of ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice showed a 50% reduction in diameter, and spermatogenic cells were dramatically decreased (Fig. 4F) when compared with wild-type mice (Fig. 4A). Spermatocytes beyond the pachytene stage and round spermatids were significantly reduced in num-  ber. Elongated spermatids were rare, and the few remaining spermatids had deformed nuclei. Some germ cells appeared to be detached from the Sertoli cells. No spermatozoa were observed in the epididymis (data not shown). In older mice, early germ cell stages were more severely affected, leading to increased depletion of spermatocytes and spermatogonia (data not shown). In TUNEL assays, apoptotic cells were rare in wild-type testes as previously reported (25) (Fig. 5A), whereas ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ testes had a significantly larger number of apoptotic cells (Fig. 5B). The cells undergoing apoptosis in ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ testes were predominantly spermatocytes. These results indicate that in the absence of the catalytic subunits of PAF-AH (I), the differentiation of prehaploid stages of spermatogenesis fails, leading to induction of programmed cell death in the germ cell compartment.
Histologically, ␣1 Ϫ/Ϫ mice testes showed no apparent impairment of spermatogenesis (Fig. 4B). On the other hand, ␣2 Ϫ/Ϫ mice testes showed significant weight reduction and varying germ cell impairment ( Fig. 4C and Table I). When comparing ␣1 Ϫ/Ϫ /␣2 ϩ/Ϫ mice testes with ␣1 ϩ/Ϫ /␣2 Ϫ/Ϫ mice testes from both the histological and weight points of view, impairment was more severe when both ␣2 alleles were missing (Fig. 4, D and E, and Table I). These results indicate that ␣2 plays a more important role in spermatogenesis than ␣1 and that missing ␣1 alleles can in part be compensated by the presence of ␣2. However, it is evident that ␣1 also plays a role in male fertility based on the observation that the absence of both ␣2 alleles can be partly compensated by the presence of ␣1 alleles (Fig. 4, C, E, and F, and Table I). The ␣1 protein level was reduced to about 20% of the normal level in ␣2 Ϫ/Ϫ mice (Fig. 3C) even though there was no reduction of the mRNA level (data not shown), whereas the ␣2 level was not changed in ␣1 Ϫ/Ϫ mice.
Because a large portion of LIS1 forms complexes with ␣1 and/or ␣2 in the cytosolic fraction, we examined the LIS1 protein levels in ␣1 and/or ␣2 mutant adult mice. In ␣2 Ϫ/Ϫ mice, LIS1 levels in both brain and testis were reduced to ϳ30% compared with wild-type mice (Fig. 6A). In contrast, no reduction of LIS1 was observed in ␣1 Ϫ/Ϫ mice (Fig. 6A). In ␣1 Ϫ/Ϫ / ␣2 Ϫ/Ϫ mice, as in ␣2 Ϫ/Ϫ mice, LIS1 levels were reduced to about 20% of the levels in the wild type mice. Lis1 mRNA expression in ␣2 Ϫ/Ϫ mice was either the same as or slightly higher than the expression in wild type mice (Fig. 6B). On the other hand, LIS1 expression in E14.5 (embryonic day 14.5) brain of each mutant mouse was not significantly less than that in wild-type mice (Fig. 6C). DISCUSSION In this study, we demonstrated that the catalytic subunits of intracellular PAF-AH (I) are involved in murine spermatogenesis. Our study gives new insights into the in vivo function of PAF-AH (I). We showed by Western blotting that in mice the catalytic subunits ␣1 and ␣2 as well as LIS1 are present at high levels in both brain and testis. Interestingly, it was noted that PAF-AH (I) subunits exhibit expression levels proportional to those of ␣-tubulin, a major component of intracellular microtubules. The above findings and the fact that LIS1 is a microtubule-associated protein lead us to speculate that the catalytic subunits are also involved to a major extent in microtubule dynamics. Microtubule structures undergo dramatic rearrangements in the process of spermatogenesis. Processes involving microtubule rearranging include mitotic division of spermatogonia, meiotic division of spermatocytes, manchette formation, and flagellar axoneme assembly in spermatids. The most severe degeneration occurs in primary spermatocytes of ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice, but degeneration also occurs in meiotically dividing spermatocytes (increased apoptosis) and in elongating spermatids (abnormal nuclear morphogenesis). Therefore, it can be stated that PAF-AH (I) catalytic subunits are involved in the several processes of spermatogenesis and not just in a specific stage of spermatogenesis.
Although the exact molecular mechanism and function of the , and ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ (F) mice were examined. Note severe reduction of spermatogenic cells, especially after pachytene spermatocyte stage in ␣1 ϩ/Ϫ /␣2 Ϫ/Ϫ (E) and ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ (F) mice. Elongating spermatids were almost absent in the latter. Scale bar: 100 m PAF-AH (I) catalytic subunits in spermatogenesis are unclear, we found that depletion of both catalytic subunits leads to a major decrease in LIS1 protein levels, suggesting that the catalytic subunits are associated with LIS1. It is most likely that LIS1 protein levels are post-transcriptionally influenced by the catalytic subunits, because LIS1 mRNA levels are not altered in ␣2 Ϫ/Ϫ mice. Because LIS1 levels are crucial for cortical brain development (13), we speculate that LIS1 is involved in microtubule organization of spermatogenesis and that reduced LIS1 protein levels are responsible for the testicular defects occurring in ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice. However, our studies also revealed that the more severe testicular degeneration in ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice than in ␣2 Ϫ/Ϫ mice cannot be explained solely by the amount of reduction in the level of LIS1, because the reduction of LIS1 was not very different between ␣2 Ϫ/Ϫ and ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice. Given that PAF-AH (I) closely resembles trimeric G-proteins (23), the catalytic subunits may mediate a novel intracellular signaling to LIS1 in mammals, and depletion of this signaling pathway may result in severe impairment in spermatogenesis.
Because PAF-AH (I) catalytic subunits are predominantly expressed in brain as well as in testis and because haploinsufficiency of LIS1 leads to severe brain malformation in both humans and mice (13), we expected that mice lacking the catalytic subunits would exhibit brain abnormalities. However, Nissl staining of adult ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice brain showed no obvious abnormalities in lamination of neurons in the cerebral cortex, hippocampus, or cerebellum (data not shown). To our surprise, in E14.5 brain of ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice, there was no significant reduction of LIS1 protein levels, suggesting that there is a mechanism to maintain LIS1 protein levels and the function of catalytic subunits in brain.
In ␣2 Ϫ/Ϫ mice, both ␣1 and LIS1 protein levels were significantly reduced compared with wild-type mice, whereas in ␣1 Ϫ/Ϫ mice both ␣2 and LIS1 levels were not changed markedly. In preliminary experiments, supernatants of mice testis or brain homogenates were subjected to DEAE-Sepharose ion exchange column chromatography. In the case of the ␣1 Ϫ/Ϫ homogenates, ␣2 eluted in the same fraction as LIS1, whereas in the case of the ␣2 Ϫ/Ϫ homogenates ␣1 and LIS1 eluted at the different positions. 2 These results suggest that ␣2, probably the ␣2 homodimer, has a strong affinity for LIS1 and that the ␣1 homodimer has a weak or negligible affinity for LIS1 in vivo. Considering the fact that ␣1 mRNA levels are not changed in ␣2 Ϫ/Ϫ mice (data not shown), the present results also suggest that the ␣1 protein is not stably expressed in the absence of ␣2 in vivo. These observations are consistent with our previous report (26) that ␣1/␣2 heterodimers and ␣2 homodimers are the major PAF-AH (I) catalytic units present in vivo.
Immunostaining studies revealed that ␣2 is expressed in all spermatogenic cells, whereas ␣1 is expressed only in spermatogonia. This expression pattern parallels the finding that deletion of both subunits induces a marked loss of germ cells, even at an early spermatogenic stage. We have previously shown that ␣1 is specifically expressed in migrating neurons in the embryonic and postnatal stages, whereas the ␣2 expression level is almost constant from the fetal stages through adulthood (26,27). As a result, the catalytic subunits change from the ␣1/␣2 heterodimer to the ␣2/␣2 homodimer in neurons during brain development. It is likely that the same type of alteration in the catalytic dimer occurs during differentiation from spermatogonia to spermatocytes. Interestingly, it has been shown that undifferentiated spermatogonia move to specific sites within the seminiferous tubule and spread their progeny laterally along the base of the tubule (28). Transplantation experiments demonstrated that spermatogonia are capable of moving along the length of the seminiferous tubule at a rate of more than 50 m/day (29,30). Although the biological significance of the change in the catalytic subunit combination is not known, it is interesting to speculate that ␣1 mediates a common signaling pathway in migrating neurons and spermatogonia.
The cellular function of the enzyme activity and the physiological substrate of this enzyme are largely unknown. PAF has been detected in sperm from several mammalian species and has been shown to affect sperm motility and fertility (4). Highfertility spermatozoa, for example, have a substantially greater PAF content than low-fertility spermatozoa (31,32). Exogenously added PAF increases the motility of human spermato-FIG. 6. LIS1 protein expression is reduced in ␣2 ؊/؊ and ␣1 ؊/؊ / ␣2 ؊/؊ adult mice. A, Western blot analysis of PAF-AH (I) subunits in adult brains and testes of the wild-type mice and various ␣ mutant mice. Note the ␣1 expression is reduced in both brain and testis of ␣2 Ϫ/Ϫ mice compared with wild-type mice (*), and LIS1 expression is reduced in both brain and testis of ␣2 Ϫ/Ϫ and ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice compared with wild-type mice (**). B, Northern blot analysis of mRNA expression of Lis1 in brains of the wild-type, ␣2 ϩ/Ϫ and ␣2 Ϫ/Ϫ mice. Glyceraldehyde-3-phosphate dehydrogenase is included as a control for equal RNA loading. C, Western blot analysis of LIS1 protein in embryonic (E14.5) brain of wild-type and various ␣ mutant mice. zoa (33). Our study gives hints at the possibility that PAF is not only involved in spermatozoal maturation and penetration but is also involved in spermatogenesis itself. On the other hand, PAF-AH (I) shows striking substrate specificity for an acetyl group but hydrolyzes other types of acetyl-containing esters in vitro (34). Studies on the tertiary structure of the catalytic dimer suggest that the substrate of this enzyme is not necessarily a lipophilic substance (23). Because PAF-AH (I) shows similarities to trimeric G-proteins (23), the PAF-AH (I)-mediated novel intracellular signaling is likely operating in mammals, with PAF or a related substance as a GTP-like switch.
When measuring cytosolic PAF-AH activity of ␣1 Ϫ/Ϫ /␣2 Ϫ/Ϫ mice in brain and testis, enzymatic reduction to ϳ65% of wildtype mice was seen in both tissues (data not shown). This phenomenon is likely because type I PAF-AH is the only affected subtype, whereas enzymatic activities of type II PAF-AH and further not yet identified PAF-AH subtypes are probably responsible for the remaining activity.
In conclusion, we found that the depletion of the PAF-AH (I) catalytic subunits induces reduction of LIS1 protein on the cellular level and severe testicular malformation on the phenotypic level. The next question to be considered is whether the catalytic activity of PAF-AH (I) is required for LIS1 protein stability and spermatogenesis. To answer these questions, we are planning to insert the catalytically inactive ␣1 and ␣2 subunit genes into our double-knockout mouse model.