Testicular germ cell apoptosis in Bcl6-deficient mice

The Bcl6 gene was first identified on the break-point of a chromosomal translocation involving 3q27 in human diffuse large B cell lymphomas (Kerckaert et al., 1993; Miki et al., 1994; Ye et al., 1993). Bcl6 protein contains a BTB/POZ domain in N-terminal region and six zinc-finger motifs in C-terminal region. It is located in the nucleus, where it functions as a sequence specific transcriptional repressor (Chang et al., 1996; Seyfert et al., 1996). Expression of Bcl6 is ubiquitous, but is especially high in germinal center B cells (Cattoretti et al., 1995; Onizuka et al., 1995). The germinal center formation is defective in Bcl6-deficient (Bcl6^−/−) mice (Dent et al., 1997; Fukuda et al., 1997; Ye et al., 1997). Bcl6^−/− mice also show growth retardation and eosinophilic inflammation including myocarditis. Since Bcl6 is strongly expressed in mature cardiac myocytes, it may play a role in protecting myocytes from cell death (Yoshida et al., 1999). Gene transfection experiments demonstrate that overexpression of Bcl6 enhances the viability of differentiating myocytes by preventing apoptosis (Kumagai et al., 1999); however, overexpression of Bcl6 induces apoptosis in CV-1 and HeLa cells (Yamochi et al., 1999). The role of Bcl6 in apoptosis has therefore remained controversial. Germ cell apoptosis naturally occurs in testis of rats (Billig et al., 1995) and mice (Rodriguez et al., 1997) at the prepubertal age. A peak of the physiological apoptosis, which is called the apoptotic wave, occurs around three to four weeks of age. The apoptotic cells are mainly spermatocytes and these cells express a large amount of Bax protein (Rodriguez et al., 1997). Overexpression of Bcl2 in germ cells inhibits the apoptotic wave followed by degeneration of germ cells in testes of transgenic adult mice (Furuchi et al., 1996; Rodriguez et al., 1997), indicating that the apoptotic wave is necessary for proper spermatogenesis. Many genes are expressed in germ cells during spermatogenesis (Eddy and O’Brien, 1998). Deficiency of genes such as Bax (Knudson et al., 1995), Bcl-w (Bcl2l2 – Mouse Genome Informatics; Print et al., 1998; Ross et al., 1998), Atm (Xu et al., 1996), Hsp70-2 (Dix et al., 1996), cyclin A1 (Ccna1 – Mouse Genome Informatics; Liu et al., 1998), Bsg (Igakura et al., 1998; Toyama et al., 1999) and Mlh1 (Edelmann et al., 1996) induces extensive germ cell apoptosis and arrest of spermatogenesis in mice. Thus, proper expression of these gene products is critical for spermatogenesis. Since Bcl6 mRNA is present in normal mouse testis (Fukuda et al., 1995) and Bcl6^−/− male mice have a very low fertility, Bcl6 may play a role in spermatogenesis. However, a role of Bcl6 in spermatogenesis has not been established.

Germ cell apoptosis is also induced by stress stimuli such as heat shock in adult rats (Lue et al., 1999). Stress stimuli activate mitogen-activated protein kinases (MAPKs) through phosphorylation. Several members of the MAPK family have been identified, including extracellular signal-regulated kinase (ERK) (Pelech and Sanghera, 1992; Marshall, 1994), c-JUN-N-terminal protein kinase (SAPK/JNK) (Kyriakis and Avruch, 1990; Kyriakis et al., 1994; Verheij et al., 1996) and p38 MAP kinase (Freshney et al., 1994; Lee et al., 1994; Rouse et al., 1994). ERK is principally activated by stimulation of growth factors, while both SAPK/JNK and p38 MAP kinase are activated by stress stimuli such as severe heat shock (Nagata et al, 1999), strong oxidants and UV irradiation (Derijard et al., 1994; Kyriakis et al., 1994; Raingeaud et al., 1995). In the present study, we examined spermatogenesis in testes of Bcl6^−/− mice. Although mature sperm developed in Bcl6^−/− mice, the number of spermatozoa in the cauda epididymis was lower than that in Bcl6^+/+ mice. Numerous apoptotic spermatocytes, mainly at the metaphase I stage, were evident in adult Bcl6^−/− mice. Since the activated form of p38 MAP kinase was detected in spermatocytes of adult Bcl6^−/− mice, germ cell apoptosis may be induced by stressors. We discuss the role of Bcl6 in regulating stress-induced apoptosis during spermatogenesis.

C57BL/6 mice were purchased from Japan SLC (Hamamatsu, Japan). Bcl6^−/− mice described elsewhere (Yoshida et al., 1999) were maintained under specific pathogen free conditions in the animal center of Chiba University School of Medicine.

Reproductive performance of Bcl6^−/− mice was examined after eight weeks of age. Briefly, a Bcl6^−/− male mouse was mated with two wild-type female mice for 2 months. If no pregnancy was observed during the first month, these female mice were replaced with other females and pregnancy of the latter mice was monitored for the second one-month period. For short-term mating study, a Bcl6^−/− male mouse at 12 weeks of age was mated with two six-week-old C57BL/6 female mice that had been superovulated by injection of 5 IU of pregnant mare serum gonadotropin (Sigma) and 5 IU of human chorionic gonadotropin (Sigma) for a 48-hour interval, and the presence of copulatory plugs was examined the next morning. For in vitro fertilization, spermatozoa from the cauda epididymis of Bcl6^−/− mice at 12 weeks of age were prepared in 250 μl TYH (Toyoda, Yokohama and Hoshi) medium (Toyoda and Chang, 1974) containing bovine serum albumin (4 mg/ml), and sperm were allowed to disperse into the medium. Sperm were capacitated by incubation for 90 minutes at 37°C. Sperm counts and motility were determined visually by phase contrast microscopy. C57BL/6 female mice were superovulated and oocytes were isolated on the following morning. Twenty oocytes were mixed with 15,000 capacitated sperm in 100 μl TYH medium, incubated for 24 hours at 37°C, and examined for developmental progression to the two-cell stage by phase contrast microscopy.

Testes were fixed by perfusion through the heart with either 4% paraformaldehyde or 3% glutaraldehyde in Hepes buffer (10 mM, pH 7.4) for 20 minutes. Then, the testes were removed, cut into small blocks, and immersed in the fixative for an additional 12 hours. Glutaraldehyde-fixed testes were postfixed with 1% OsO₄, dehydrated and embedded in Epon 812. Semi-thin sections (1 μm thick) were stained with 0.1% Toluidine Blue and analyzed histologically by light microscopy. Ultra-thin sections (80 nm thick) were stained with uranyl acetate followed by lead citrate and examined by electron microscopy (JEOl 1200 EX, Tokyo, Japan).

For immunohistochemistry, 4% paraformaldehyde-perfused testes were immersed in 20% sucrose and embedded in OTC compound (Miles, Elkhart, IN), and frozen 10 μm sections were cut. Perfused testes were dehydrated with ethanol, embedded in paraffin, and 5 μm thick paraffin sections were cut. We used rabbit polyclonal antibodies against Bax and Bcl6 (N-3) (Santa Cruz Biotechnology, Santa Cruz, CA) on paraffin sections, and against Bcl-x_L, p53, Cdc2, cyclin B1 (Santa Cruz Biotechnology) and phospho-p38 MAP kinase (Thr 180/Tyr 182) (New England Biolabs, Beverly, MA) on frozen sections. Sections were incubated overnight with the primary antibody. Normal rabbit serum was used for a control. The biotinylated goat anti-rabbit immunoglobulin antibody (Nichirei, Tokyo, Japan) and the Vectastain Elite ABC-kit (Vector Laboratories, Burlingame, CA) were used as the second- and the third-phase reagents, respectively. Bound peroxidase activity was visualized with the diaminobenzidine detection kit (DAB kit; Nichirei). Sections were counterstained with hematoxylin.

TUNEL assay was carried out as described (Gavrieli et al., 1992) with slight modification. Briefly, 4% paraformaldehyde-perfused testes were embedded in paraffin wax. Sections (5 μm thick) were deparaffinized in xylene and treated with the proteinase K solution for 30 minutes at 37°C. The tailing reaction was then carried out in TdT buffer in the presence of dUTP-biotin for 1 hour at 37°C (The Mebstain Apoptosis kit, Medical & Biological Laboratories, Nagoya, Japan). Signals were visualized by the Vector Elite ABC-kit and the DAB kit. Sections were counterstained with hematoxylin.

Three to nine testes of each group were used for each experiment. The apoptotic cell number was calculated and averaged by counting apoptotic cells in 30-50 apoptotic tubule cross sections (TCS) within 400 TCS in each testis.

Testes in the scrotum of adult mice were immersed in a thermostatically controlled water bath for 15 minutes at 42°C. Testes were removed 6 hours later, fixed with 4% paraformaldehyde, and apoptotic cells in the testes were detected by TUNEL assay.

Testicular cells (1×10⁸) were suspended in 1 ml of IP (immunoprecipitation) buffer (Kobayashi et al, 1997) at 4°C and the clarified lysate was obtained by centrifugation. The amount of protein in cell lysates was determined using the BioRad protein assay (BioRad Laboratories). Cell lysates (80 μg) were resolved by SDS-PAGE and transferred to a polyvinyl difluoride membrane (Immobilon-P, Millipore, Bedford, MA). Blots were blocked in Blockace (Yukijirushi, Sapporo, Japan) for 1 hour, washed three times with Tris-buffered saline with 0.2% Tween 20 (TBST), and incubated with the antibody for phospho-p38 MAP kinase (Thr 180/Tyr 182) and p38 MAP kinase (New England Biolabs) in TBST for overnight at 4°C. The blots were then washed three times with TBST, incubated with affinity-purified donkey anti-rabbit antibody conjugated with horseradish peroxidase (HRP, Amersham) for 1 hour at room temperature, washed four times with TBST and developed with enhanced chemiluminescence reagents (Amersham).

Paired comparisons of testicular weight, epididymal spermatozoa numbers, in vitro fertilization rates of sperm, serum testosterone levels, and the apoptotic index in testes among Bcl6^+/+, Bcl6^+/− and Bcl6^−/− mice were performed for statistical significance by calculating means±s.e.m. and Student’s t test.

The reproductive function of Bcl6^−/− male mice was examined by mating them with wild-type female mice. Only 15% of Bcl6^−/− mice (four out of 27 mice) at 8-12 weeks of age were able to make female mice pregnant whereas all 26 Bcl6^+/− and Bcl6^+/+ mice examined were fertile. A short-term mating study was carried out to determine whether the absence of Bcl6 might affect the mating performance of Bcl6^−/− mice or not. Bcl6^−/− males produced copulatory plugs in three out of nine mates, the frequency of which is lower than that of Bcl6^+/− (9 out of 9 mates) and Bcl6^+/+ males (9 out of 9 mates). Testes of Bcl6^−/− mice weighed slightly less (79±12 mg) than those of Bcl6^+/+ mice (93±11 mg) but not significantly so (P=0.25). The weight of those from Bcl6^+/− mice (102±19 mg) was similar to testes from Bcl6^+/+ mice (P=0.41). The number of spermatozoa in the cauda epididymis of the Bcl6^−/− mice (2.2±0.4 ×10⁶/ml) was less than the number in Bcl6^+/+ mice (9.8±2.0 ×10⁶/ml; P=0.0031). The number of spermatozoa in the cauda epididymis of the Bcl6^+/− mice (8.2±3.0 ×10⁶/ml) was similar to that of the Bcl6^+/+ mice (P=0.75). In vitro fertilization of normal oocytes was carried out using equivalent numbers of motile sperm from Bcl6^−/− and Bcl6^+/+ mice. The fertilization rate of Bcl6^−/− sperm (27±6%) was less than that of Bcl6^+/+ sperm (80±5%; P<0.0001).

Testes of Bcl6^−/− and Bcl6^+/+ mice at 12 weeks of age were histologically examined. Testes of Bcl6^+/+ mice revealed seminiferous tubules that contained normal spermatogenic cells at various stages of development and numerous numbers of mature spermatids (Fig. 1A,C). Although seminiferous tubules of Bcl6^−/− mice contained spermatogenic cells at various stages, including mature spermatids, many degenerating germ cells were identified (Fig. 1B,D). We then carried out a TUNEL assay to detect apoptotic cells in testes of Bcl6^−/− and Bcl6^+/+ mice at 12 weeks of age. Seminiferous tubules of Bcl6^−/− mice (Fig. 2B,D) contained larger numbers of TUNEL-positive cells than did those of Bcl6^+/+ mice (Fig. 2A,C). The majority (>90%) of TUNEL-positive cells were metaphase I spermatocytes where the chromosomes were clearly visible as a cluster in stage XII seminiferous tubules (Russell et al., 1990) (Fig. 2D). Some of TUNEL-positive cells (<10%) were pachytene spermatocytes. To explore the relationship between these apoptotic cells and modified levels of protein involved in cell death, immunohistochemistry was performed on serial sections of the testes with antibodies specific for Bax, Bcl-x_L or p53. The apoptotic cells in seminiferous tubules of both Bcl6^+/+ and Bcl6^−/− mice express a detectable amount of Bax protein (Fig. 2E,F) but not Bcl-x_L or p53 protein (data not shown).

Germ cell apoptosis in Bcl6^−/− mice at 12 weeks of age was confirmed by electron microscope analysis. A normal pachytene spermatocyte in Bcl6^+/+ mice had round nuclei with a regular contour (Fig. 3A). A spermatocyte in Bcl6^−/− mice was degenerated with a clear area in the nucleus (Fig. 3B). A normal metaphase I spermatocyte in Bcl6^+/+ mice shows condensed chromosomes without a nuclear membrane (Fig. 3C). However, a metaphase I spermatocyte in Bcl6^−/− mice had deformed mitochondria (Fig. 3D, inset) and a peripherally located endoplasmic reticulum. These results confirmed the presence of apoptotic cells in testes of adult Bcl6^−/− mice. Spermatogonia, Sertoli and Leydig cells of the Bcl6^−/− mice were normal in this examination (data not shown).

Bcl6 mRNA was detected in total RNA of normal mouse testis (Fukuda et al., 1995). To examine the relationship between loss of Bcl6 expression and the apoptosis in adult Bcl6^−/− mice, the presence of Bcl6 protein in spermatocytes was examined in testes of Bcl6^+/+ mice at 12 weeks of age by immunohistochemistry. Bcl6 was detectable in nuclei of primary spermatocytes at the pachytene, diplotene and metaphase I stage, secondary spermatocytes, and spermatids up to step 8 (Fig. 4A). Sertoli and Leydig cells did not express Bcl6 in the testes. As a control for specificity of the anti-Bcl6 antibody, testes of Bcl6^−/− mice were stained. Bcl6 protein was not detected in testes of Bcl6^−/− mice (Fig. 4B).

In order to examine the onset of germ cell apoptosis in testes of Bcl6^−/− mice, we examined testes histologically at the prepubertal age. Seminiferous tubular lumens of Bcl6^+/+ and Bcl6^−/− mice were closed at two weeks of age (Fig. 5A,B) but were open at the age of four weeks (Fig. 5C,D). Mature spermatids were detectable in testes of Bcl6^+/+ and Bcl6^−/− mice at eight weeks of age, and there was no significant histological difference in testes between them until eight weeks of age (Fig. 5E,F).

Germ cell apoptosis at prepubertal age occurred in spermatocytes (Billing et al., 1995) mainly at the pachytene and the metaphase I stage of normal mice (data not shown). The number of apoptotic cells in a seminiferous tubule during development was calculated and statistically analyzed between Bcl6^+/+ and Bcl6^−/− mice (Fig. 6). The apoptotic cell number of Bcl6^−/− mice was similar to that of Bcl6^+/+ mice until four weeks of age, suggesting that the physiological apoptotic wave occurs normally in Bcl6^−/− mice. Those apoptotic cells expressed a detectable amount of Bax protein (data not shown). The apoptotic cell number decreased in Bcl6^+/+ mice after six weeks of age. However, it continued at an increasingly high level in Bcl6^−/− mice older than eight weeks.

Heat shock is a stress known to induce germ cell apoptosis in adult rats (Lue et al., 1999), so the pathology of germ cell apoptosis induced by heat stress in normal mice was examined. Testes of Bcl6^+/+ mice at eight weeks of age were subjected in vivo to hyperthermia (42°C) for 15 minutes, and apoptotic cells in the testes were detected by TUNEL assay 6 hours after the treatment. Testes of Bcl6^+/+ mice exhibited numerous apoptotic germ cells after heat treatment, with the majority (>90%) being metaphase I stage spermatocytes (Fig. 7A,B). These apoptotic cells expressed high levels of Bax protein (Fig. 7C). Since p38 MAP kinase is activated by stress stimuli (Nagata and Todokoro, 1999; Raingeaud et al., 1995), we examined expression of phospho-p38 MAP kinase in testes of Bcl6^+/+ mice at 12 weeks of age after heat treatment by immunoblotting. p38 MAP kinase (Fig. 8A) was activated in testes of Bcl6^+/+ mice 30 minutes to 1 hour after heat treatment. Phospho-p38 MAP kinase was also detected in leptotene, zygotene, pachytene spermatocytes 1 hour after heat treatment by immunohistochemistry (Fig. 8B). Interestingly phospho-p38 MAP kinase was obviously present in non-heat-treated testes of Bcl6^−/− mice at 12 weeks of age by immunoblotting (Fig. 8C). Additionally, the nuclei of primary spermatocytes of the Bcl6^−/− mice contained phospho-p38 MAP kinase by immunohistochemistry (Fig. 8D).

We next examined the relationship between the apoptotic cell number and the amount of Bcl6 in spermatocytes using testes of heterozygous Bcl6^+/− mice, since Bcl6^+/− mice are expected to express half the amount of Bcl6 as Bcl6^+/+ mice (Yoshida et al., 1999). The apoptotic cell number in testes of Bcl6 mice at 12-16 weeks of age was examined by TUNEL assay and the data were compared with those of age-matched Bcl6^+/+ and Bcl6^−/− mice (Fig. 9). Although the apoptotic cell number (17.3±5.4) of Bcl6^−/− mice was the highest, the number in Bcl6^+/− mice (6.8±3.7) was also higher than that in Bcl6^+/+ mice (2.5±0.4; P<0.005). Furthermore, heat sensitivity of spermatocytes of Bcl6^+/+, Bcl6^+/− and Bcl6^−/− mice at 12 weeks of age was examined by subjected testes in vivo to hyperthermia (42°C) for 15 minutes. Apoptotic cells in the testes were detected by TUNEL assay 6 hours after the treatment. The apoptotic cell numbers of Bcl6^+/+ (9.4±1.5) and Bcl6^+/− (13.5±3.6) mice increased four and twofold after heat treatment, respectively. However, the number of apoptotic cells in Bcl6^−/− mice (17.2±3.5) was not changed by heat treatment.

Bcl6^−/− male mice showed the lower reproductive function and only 15% of Bcl6^−/− mice were fertile. The frequency of copulation (plugging ability) of Bcl6^−/− mice was about one third that of Bcl6^+/+ mice, suggesting that the lower reproductive function is partly due to the abnormal reproductive behaviour of Bcl6^−/− mice. Furthermore, the number of spermatozoa in the cauda epididymis of Bcl6^−/− mice was about 20% of that of Bcl6^+/+ mice, and the in vitro fertilization rate of motile sperm from Bcl6^−/− mice was lower than that from Bcl6^+/+ mice. Thus, the lower reproductive function of Bcl6^−/− mice may be due to the abnormal reproductive behaviour, the lower number of spermatozoa and the abnormal fertilization function of spermatozoa. Since Bcl6^−/− mice show growth retardation and eosinophilic inflammation (Yoshida et al., 1999), these abnormal physical conditions may be one of the reasons for the abnormalities.

The lower number of spermatozoa from Bcl6^−/− mice may be due to germ cell apoptosis since numerous apoptotic spermatocytes mainly at the metaphase I stage were evident in adult Bcl6^−/− testes. We suspected that the germ cell apoptosis in adult Bcl6^−/− mice might be related with the abnormal physical conditions. However, apoptosis of metaphase I spermatocytes was also evident in heterozygous adult Bcl6^+/− mice under conditions without growth retardation and eosinophilic inflammation, suggesting little relation between the germ cell apoptosis and the abnormal physical conditions. When levels of follicle stimulating hormone/luteinizing hormone decrease in normal rats at three to four weeks of age, testicular germ cell apoptosis is induced (Billig et al., 1995) and the germ cell apoptosis is mostly detected in early primary spermatocytes and spermatids in rats (Hikim et al., 1995). The spermatogenic stages of the apoptotic cells are different from those in adult Bcl6^−/− mice. Furthermore, the hormonal change has no effect on spermatogenesis in adult rats (Billig et al., 1995) and mice (Kumar et al., 1997), suggesting that germ cell apoptosis in adult Bcl6^−/− mice is not due to decreased levels of the gonadotropins. Although Sertoli and Leydig cells are critical for spermatogenesis (Griswold, 1995; Lue et al., 1999; Ross et al., 1998), Sertoli and Leydig cells did not express Bcl6 in normal mice (Fig. 4A), and these cells of adult Bcl6^−/− mice were histologically normal. These results suggest that germ cell apoptosis in adult Bcl6^+/− and Bcl6^−/− mice is induced under conditions where a number of cytological and hormonal networks appeared normal.

Testicular germ cell apoptosis occurs in rats (Billig et al., 1995) and mice (Rodriguez et al., 1997) at the prepubertal age. A large amount of Bax protein is detected in testicular apoptotic cells during this physiological apoptotic wave (Schwartz et al., 1993; data not shown) that is due to a decrease in testosterone level (Rodriguez et al., 1997). p53, which is highly expressed in pachytene spermatocytes during the prepubertal age (Schwartz et al., 1993), may function as a Bax promoter (Miyashita and Reed, 1995; Selvakumaran et al., 1994). The apoptotic cells in adult Bcl6^−/− mice did express a detectable amount of Bax but not p53. Furthermore, adult Bcl6^−/− mice had normal levels of testosterone in their sera (data not shown). Thus, the mechanism of germ cell apoptosis in adult Bcl6^−/− mice appears different from the physiological apoptotic wave. Since the apoptotic cell number of Bcl6^−/− mice was similar to that of Bcl6^+/+ mice at prepubertal age, but increased as animals became adults and the activated form of p38 MAP kinase was detected in primary spermatocytes of adult Bcl6^−/− mice, the germ cell apoptosis may be stress induced.

Sporadic apoptotic cell death occurs at every stage of the spermatogenic cycle in normal adult rats (Blanco-Rodriguez and Martinez-Garcia, 1996). However, in adult Bcl6^−/− mice, germ cell apoptosis preferentially occurred in metaphase I spermatocytes, which suggests that a specific mechanism governs the apoptosis in spermatocytes. Spermatocytes at the metaphase I stage are most sensitive to heat stress in rats (Lue et al., 1999) and mice (Fig. 7), though the mechanism of germ cell apoptosis induced by testicular hyperthermia is unknown. We found a detectable amount of Bax protein and the activated form of p38 MAP kinase in spermatocytes of heat-treated Bcl6^+/+ mice. Thus, Bax and p38 MAP kinase are associated with heat-induced germ cell apoptosis, and both were also detected in spermatocytes of adult Bcl6^−/− mice without heat treatment. These results suggest that the mechanism of germ cell apoptosis in adult Bcl6^−/− mice is related to that in Bcl6^+/+ mice induced by heat stress. Since germ cell apoptosis is induced in rats even when testes are exposed to abdominal temperature (Yin et al., 1997), metaphase I spermatocytes without Bcl6 may die when subjected to mild heat stress.

A member of the heat shock protein gene family, the Hsp70-2 gene, is expressed in primary spermatocytes (Allen et al., 1988; Zakeri et al., 1988) and spermatogenesis fails to proceed to meiosis in mice with null mutation of the gene (Dix et al., 1996; Dix et al., 1997; Zhu et al., 1997). Thus, dysregulated expression of the Hsp70-2 gene by deficiency of Bcl6, a transcriptional repressor (Chang et al., 1996; Seyfert et al., 1996), might induce germ cell apoptosis in Bcl6^−/− mice. Hsp70-2 is required for Cdc2/cyclin B complex formation (Zhu et al., 1997), and kinase activity of Cdc2/cyclin B complexes is critical for synaptonemal complex desynapsis (Chapman and Wolgemuth, 1994; Zhu et al., 1997). However, these proteins are detected in spermatocytes of both Bcl6^+/+ and Bcl6^−/− mice without obvious differences in levels (data not shown), suggesting that the mechanism of germ cell apoptosis in adult Bcl6^−/− mice is not related to the process of synaptonemal complex desynapsis.

Apoptosis of metaphase I spermatocytes was induced in adult Bcl6^−/− and Bcl6^+/− mice without heat treatment and the incidence of germ cell apoptosis was roughly correlated with the Bcl6 dose. After heat treatment, the number of apoptotic spermatocytes was not increased in adult Bcl6^−/− mice. It is suggesting that the germ cells that are most sensitive to heat stress were dead in adult Bcl6^−/− mice without heat treatment, although it is still possible that Bcl6 is somehow required for the spermatocytes to initiate heat-induced germ cell apoptosis. Furthermore, the spermatogenic stages of phospho-p38 MAP kinase expression in spermatocytes of Bcl6^−/− mice were overlapped with those of Bcl6 expression of Bcl6^+/+ mice. Thus, Bcl6 may play a role in protection of spermatocytes mainly at the metaphase I stage from apoptosis induced by stressors. Any relationship between Bcl6 function and p38 MAP kinase activation is presently unclear. Additional work is required to elucidate specific target genes for Bcl6 to protect apoptosis during spermatogenesis.

Many genes are expressed in germ cells during spermatogenesis (Eddy and O’Brien, 1998). Deficiency of the genes such as Atm (Xu et al., 1996), Hsp70-2 (Dix et al., 1996), cyclin A1 (Liu et al., 1998) and Mlh1 (Edelmann et al., 1996) induces arrest of spermatogenesis at the pachytene stage, indicating that these gene products are essential for spermatogenesis. Abnormal expression of some apoptosis-related gene products in germ cells also perturbs spermatogenesis. Bax-deficient mice (Knudson et al., 1995) show arrest of spermatogenesis with apoptotic spermatocytes. Overexpression of Bcl2 in spermatogonia (Furuchi et al., 1996) induces massive accumulation of spermatogonia leading to degeneration of seminiferous tubules. In Bcl-w-deficient mice (Print et al., 1998), the seminiferous tubules are disorganized and no mature sperm is produced. In contrast, Bcl6^−/− mice showed normal development of testes and normal physiological apoptotic wave at the prepubertal age. Although many spermatocytes go to apoptosis in adult Bcl6^−/− testes, mature sperm are developed in adult Bcl6^−/− mice. These data indicate that Bcl6 is not essential for spermatogenesis but necessary for stabilization of spermatocytes to protect against the germ cell apoptosis induced by stressors.

This protective role of Bcl6 in stress-induced apoptosis has also been reported in cardiac myocytes (Yoshida et al., 1999) and in skeletal muscle cells (Kumagai et al., 1999), which may explain a function of Bcl6 as an oncogene to produce B cell lymphoma (Ye et al., 1993). Since the BTB/POZ domain of Bcl6 interacts with the silencing mediator of retinoid and thyroid receptor protein (SMRT) (Dhordain et al., 1997) and SMRT forms a repressive complex with the histone deacetylase 1 (Nagy et al., 1997), Bcl6 may repress transcription of target genes through mechanisms involving SMRT recruitment and histone deacetylation. Thus, Bcl6^−/− mice can serve as a useful model for studying transcriptional regulatory mechanisms that allow for adaptive response to various kinds of stress.

We are grateful to Dr T. Yoshida, S. Okabe, K. Ishibashi and L. A. Herzenberg for discussion. We also thank H. Satake for skilful technical assistance, N. Fujita and K. Ujiie for secretarial services, and M. Ohara for comments on the manuscript. This work was supported in part by Grants-in-Aid from the Ministry of Education, Science, Sports and Culture of Japan, Special coordination funds for the promotion of science and technology from the Science and Technology Agency of Japan, and the Grants from the Ministry of Health and Welfare of Japan.