Further Characterization of Reproductive Abnormalities in mCd59b Knockout Mice: A Potential New Function of mCd59 in Male Reproduction¹

Complement is an effector of the immune system that mediates both acquired and innate responses to defend against foreign Ags such as those present in bacteria or heterologous erythrocytes (1). The complement system consists of >12 soluble plasma proteins that interact with one another in three distinct enzymatic activation cascades known as the classical, alternative, and lectin pathways. The three pathways converge at the level of C3, a central complement component that is cleaved by pathway-specific convertases. Thereafter, the common terminal complement cascade leads to formation of the membrane attack complex (MAC),⁴ a macromolecular pore that inserts itself into the cell membrane causing colloidosmotic lysis of the targeted cell.

To protect “self” cells from the catastrophic effect of autologous complement damage, >10 plasma- and membrane-bound inhibitory proteins have evolved to restrict complement activation at different stages of the activation pathways. In particular, CD59 is a GPI-linked membrane protein that inhibits formation of the MAC. Absence of CD59 and other GPI-linked proteins in circulating cells, as the consequence of an acquired somatic mutation in a clone of hemopoietic-cell precursors, renders the circulating cells highly susceptible to complement damage and causes the complement-mediated hemolytic anemia known as paroxysmal nocturnal hemoglobinuria (2). This paradigmatic disease illustrates the protective effect of complement regulatory proteins against autologous complement activation; in their absence, even the basal tickover complement activity is sufficient to lyse erythrocytes and cause hemolytic anemia.

The reproductive system represents a unique situation in which cells expressing new Ags are either constantly produced (spermatogenesis) or periodically introduced (insemination and fetal development) in the reproductive tract (3). To avoid constant immune challenge to a successful pregnancy, several mechanisms of immune tolerance have evolved, including expression of complement regulators in the female reproductive tissues and the placenta, where CD59 is highly expressed throughout pregnancy (4, 5, 6). In mice, the targeted deletion of CD59 does not result in fetal loss (7). In contrast, the deletion of mouse complement-related receptor protein (Crry), a complement regulatory protein only expressed in mice and rats, results in fetal damage and interruption of pregnancy at or before day 9 (8). The infertility of the Crry^−/− mother is the consequence of a complement-mediated attack on the fetus rather than a reproduction-specific function of Crry^−/−, because the fetuses survive if the Crry^−/− mother is also deficient in C3 (8, 9). Another example that highlights the importance of complement in the pathophysiology of reproduction was provided by a recent study of a mouse model of the anti-phospholipid Abs syndrome in which these Abs produced decidual focal necrosis with C3 deposition and neutrophil infiltration in pregnant mice (10).

In the male reproductive system, spermatogenesis begins at puberty, long after the immune system has developed the ability to distinguish self from “nonself”, and sperm-surface Ags can trigger an immune/complement response that would compromise the potential of the sperms for successful fertilization. Thus, it is not surprising that, in addition to a protective blood-testis barrier, a variety of nonspecific and specific complement inhibitors, including CD59, are present at high concentrations in the seminal fluid, and that sperm cells express multiple complement regulators that prevent complement activation and functional MAC formation (3, 4, 5, 6, 11, 12). In the seminal plasma, CD59 is present in a GPI-anchored, prostasome-associated form that can “deliver” CD59 molecules to the sperm cell (13), and is also highly expressed on sperm plasma membranes (6). Incubation of sperm with neutralizing anti-CD59 Abs and human serum results in sperm immobilization and lysis, whereas sperm with functional CD59 are protected (14).

Recently, we and others have demonstrated that mice have two different CD59 genes (mCd59a and mCd59b) (15, 16, 17), whereas most other rodents and humans studied have only one. mCd59b is highly expressed in testes at the level of mRNA and protein (18), a finding that suggested a possible testis/reproduction-specific function of mCd59b. To study the function of CD59 and complement in mouse models of human diseases, particularly the function of mCd59b, we generated mCd59b^−/− mice by homologous recombination. These mice express a hemolytic anemia phenotype comparable to human paroxysmal nocturnal hemoglobinuria, and progressive loss of male fertility that becomes evident at 5 mo of age, as we previously reported (7). Furthermore, we demonstrated that mCd59a protein is significantly down-regulated in mCd59b^−/− RBC, which is consistent with the reduced tissue expression of mCd59a mRNA we have reported in mCd59b^−/− (see Fig. 6 in Ref.7). This indicates that the mCd59b^−/− mouse resembles a mCd59a and mCd59b partial double knockout, explaining the hemolytic anemia phenotype of these mice (X. Qin, M. Dobarro, S. Bedford, S. Ferris, P. Miranda, W. Song, R. Bronson, P. Visconti, and J. Halperin, manuscript in preparation). In this study, we report additional studies on mCd59b^−/− that indicate that their reproductive but not their hematological abnormalities are complement independent, and that mCd59 may have a novel role in testicular function that is independent of its activity as an inhibitor of MAC formation.

The following animal studies have all been reviewed and approved by the Harvard Medical School Institutional Animal Care and Use Committee (Animal Welfare Assurance no. A3431-01; International Animal Care and Use Committee approval date: January 10, 2005).

Both mCd59b^−/− and mCd59b^+/+ littermate mice used for these experiments were offspring derived from intercrossing mCd59b^+/− in a B6/129 mixed genetic background, all characterized by PCR genotyping (7). To avoid the influence of multiple matings on sperm analysis, the experimental littermates were not mated with females.

To obtain mCd59b^−/−C3^−/− (mCd59b and C3 double knockout) and mCd59b^+/+C3^+/+ (both mCd59b and C3 wild type) littermates, we first crossed mCd59b^−/− (in B6/129 mixed genetic background) and C3^−/− (in B6 genetic background; purchased from The Jackson Laboratory) mice to generate mCd59b^+/−C3^+/− (7, 19). Both mCd59b^−/−C3^−/− and mCd59b^+/+C3^+/+ used for experiments were littermate offspring produced from intercrossing mCd59b^+/−C3^+/−, selected by genotyping as described in Refs.7 and 19 .

Blood was collected in tubes containing EDTA by cardiac puncture using precooled needles and syringes to avoid complement activation. Testes and epididymides were dissected, the cauda epididymis was separated, and sperm were collected by squeezing one cauda epididymis from each mouse into 1 ml of prewarmed HEPES-buffered saline solution containing 1% BSA (10 mM HEPES, 150 mM NaCl, 0.01 g/ml BSA (pH 7.4)). Epididymal tissue was removed and the solution was filtered through a mesh filter into a conical centrifuge tube.

LH and FSH were determined by RIA at the National Institute of Diabetes and Digestive and Kidney Disease on a fee-for-service basis as described in Ref.20 . Testosterone was measured by ELISA following the manufacturer’s recommendations (BioCheck).

Testes were fixed in Bouin’s solution, 5-mm slices were embedded in paraffin, and 6-μm sections were stained with H&E using routine procedures (21). Quantitative analyses were performed on histological sections obtained from fixed, sectioned testes using a Nikon Eclipse TE2000-E microscope equipped with a Spot digital camera. The number of multinucleated cells in seminiferous tubules was counted blindly by two independent investigators from two sections of each testis at ×20 magnification.

Eight-micrometer serial sections from paraffin-embedded tissue samples were stained with: 1) goat anti-mouse C3 Ab (ICN Pharmaceuticals) (22) or 2) rabbit anti-rat C9 Ab, which cross-reacts with mouse C9 (kindly provided by Dr. P. Morgan, University of Wales) (22). Either anti-goat for anti-C3 or goat anti-rabbit for anti-C9 conjugated to peroxidase were used as secondary Abs following established procedures (23). Two independent investigators blindly scored the samples based on the intensity differences between stainings by each specific Ab and nonimmune IgG.

Apoptotic nuclei in tissue sections were identified by a TUNEL technique using an in situ apoptosis detection kit (Apoptag; Chemicon International) according to the manufacturer’s instructions (24, 25). The sections were counterstained with toluidine blue, dehydrated with ethanol and xylene, and coverslipped. The number of TUNEL-positive nuclei in 40 ×20 microscope fields per sample was recorded blindly by different investigators.

The percentage of progressively motile sperm was immediately evaluated by placing a 10-μl drop of the diluted sperm suspension between a prewarmed glass slide and a coverslip, and visually counted under a phase contrast microscope at ×200.

A total of 10 μl of the sperm suspension was diluted 1/10 into saline-buffered formaldehyde in a microcentrifuge tube. The relative distribution of morphologically normal and abnormal sperm was evaluated in a wet slide under a phase contrast microscope by counting a total of 200 sperm per sample at ×1000. The sperm abnormalities recorded included detached heads, abnormal head shapes, bent midpieces, droplets, coiled tails, other abnormalities, and immature sperm forms (round cells) (26, 27). We defined abnormal heads as the sum of detached heads and abnormal sperm heads (type IIa-IIb-IIc, IIIa-IIIb, IV, and V) (27).

A 10-μl drop of the above suspension was used to count sperm cells on a hemocytometer as described in Ref.28 .

To differentiate live from dead sperm cells we used the live/dead sperm viability kit (SYBR 14 + propidium iodide; Molecular Probes) following the manufacturer’s recommendations. Briefly, 5 μl of a 10-fold DMSO dilution of the SYBR 14 dye was mixed with the sperm suspension and incubated for 5–10 min at 37°C in the dark. Then, 5 μl of the propidium iodide solution was added and incubated for an additional 5–10 min. A total of 200 cells were counted for viability evaluation in wet samples under a fluorescence microscope where propidium-iodide stained dead cells a fluorescent red.

Sperm capacitation was evaluated by the protein tyrosine phosphorylation assay (29, 30). Cauda epididymis were dissected and placed in a petri dish containing bicarbonate-free Wittens medium to allow sperm to swim out. Total sperm recovered per mouse was recorded. Sperm were capacitated for 1.5 h in Wittens medium supplemented with 20 mM NaHCO₃ and 5 mg/ml fatty-acid-free BSA (31). Control sperm were kept in medium without any supplement. Sperm extracts were obtained by boiling for 5 min with 1% SDS and subjected to SDS-PAGE followed by immunoblot with anti-phosphotyrosine mAb (4G10; Upstate).

Acrosome reaction was induced by addition of 2 μM calcium ionophore (A23187) during the last 10 min of incubation. Acrosomal staining was performed using Alexa 488-conjugated peanut agglutinin (32) (Molecular Probes).

An aliquot of mouse blood was analyzed on an Advia 120 autoanalyzer (Abbott Laboratories) as previously described in Ref.7 .

Rabbit erythrocytes were washed in PBS, and a 2% suspension was incubated with an equal volume of mouse anti-rabbit erythrocyte antiserum (1/250 dilution in PBS; 15 min at 37°C). The Ab-sensitized rabbit erythrocytes were washed with veronal buffered saline (VBS⁺⁺) and resuspended at 1% hematocrit. Fifty microliters of the cell suspension plus 50 μl of mouse serum (10% or 20% in VBS⁺⁺) were added to 96-well plates in triplicate, and the plates were incubated for 30 min at 37°C. Nonlysed cells were removed by centrifugation, and hemoglobin in the supernatant was measured as OD_414. Percent hemolysis in each well was calculated as previously published (33).

Mean difference of multinucleated cells, apoptotic cells, total sperm, and viable sperm, as well as motility, viability, and the percentage of sperm with abnormal heads between experimental groups were compared by Student’s t test.

We compared the median levels using the Wilcoxon rank sum test and we compared the proportions in the extremes of the distributions using Fisher’s exact test. Specifically, we compared proportions with hemoglobin levels <13 g/100 ml of blood (2 SD in mCd59b^+/+C3^+/+) and proportions with reticulocyte counts greater than 4.3% (2 SD in mCd59b^+/+C3^+/+).

Testis histological comparison between mCd59b^+/+ and mCd59b^−/− littermate mice at 6 and 10 mo of age revealed that 33% of mCd59b^−/− mice had abnormal multinucleated cells within the walls of seminiferous tubules (mean number of abnormal multinucleated cells in mCd59b^−/− = 10 ± 8 (n = 6 per age group)); these cells were absent in all mCd59b^+/+ mice analyzed (n = 6 per age group) (Fig. 1). The TUNEL assay performed on testicular paraffin sections from 6-mo-old mCd59b^−/− and mCd59b^+/+ mice revealed the presence of a significantly greater number of apoptotic, TUNEL-positive germ cells in mCd59b^−/− mice (Fig. 1, D, F, and G) than in their age-matched mCd59b^+/+ control mice (Fig. 1, D and E). The comparison of mCd59b^+/+ and mCd59b^−/− littermates of different ages (4, 6, and 10 mo old) showed that there was no difference in the testes to total body weight ratio, and no evidence of testicular atrophy or of inflammatory cells in the seminiferous tubules (data not shown). Also, immunostaining of testicular sections with either anti-C3 or anti-C9 Abs did not show an increased complement deposition in mCd59b^−/− testes (data not shown). Thus, the presence of abnormal multinucleated cells and of an increased number of apoptotic germ cells, the main histological findings in the testes of mCd59b^−/−, do not seem to be associated with an increased complement/MAC deposition, as one would expect from the functional deletion of CD59.

Comparative analysis of sperm at 3 mo of age showed that relative to mCd59b^+/+, mCd59b^−/− mice have a significant decrease in the total number of live sperm (Fig. 2,A), and in the percentage of viable sperm (Fig. 2,C). This decrease in critical sperm parameters in mCd59b^−/− was associated with a significant increase in the percentage of abnormal sperm heads (Fig. 2 C).

Sperm quality parameters progressively worsened; at 6 mo of age, both the number of total and live sperm (Fig. 2,B), and the percentage of motile and viable sperm were all significantly decreased, whereas the percentage of abnormal sperm heads was significantly higher in mCd59b^−/− mice (Fig. 2,D). Fig. 2, E–G, show examples of abnormal sperm head morphologies frequently observed in mCd59b^−/− mice. These results together with the abnormal histology findings indicate that the progressive loss of male fertility in mCd59b^−/− mice may be due to abnormal spermatogenesis.

Fig. 3 shows that there were no significant differences in the serum levels of the critical reproduction-related hormones FSH (Fig. 3,A) and LH (Fig. 3 B); levels of testosterone were also similar in mCd59b^−/− and mCd59b^+/+ males (data not shown). Thus, the progressive loss of male fertility and testicular abnormalities in mCd59b^−/− mice do not seem to be caused by low levels of hormones that affect spermatogenesis.

Both sperm capacitation and the acrosome reaction are physiological processes required for successful fertilization. Capacitation is associated with an increase in protein tyrosine phosphorylation. Indeed, Western blot analysis with specific Abs against phosphotyrosine has become an accepted method to assess sperm capacitation (29, 30, 34). In other cell types, CD59 has been found to be physically connected to the membrane tyrosine phosphorylation machinery; if this were the case in sperm cells, CD59 could play a role in sperm capacitation (35, 36, 37). To investigate whether the absence of mCd59 could affect capacitation-dependent tyrosine phosphorylation and ionophore-induced acrosome reaction in mouse sperm, we analyzed both tyrosine phosphorylation and acrosome reaction in sperm obtained by the swim-out method from the epididymis of 3-mo-old mice. mCd59b^−/− consistently yielded a significantly lower number of sperm as compared with mCd59b^+/+ (Fig. 3,C). Analysis of the same number of motile mCd59b^+/+ and mCd59b^−/− sperm showed a similar amount of tyrosine phosphorylation (Fig. 3,D). As mentioned above, the ability of sperm to undergo the acrosome reaction is also necessary for fertilization. This process is dependent on calcium influx. Therefore, to analyze whether mCd59b^−/− are able to undergo the acrosome reaction, sperm from these mice as well as wild-type control were challenged with calcium ionophore A23187, and the acrosome reaction was assessed using Alexa 488-conjugated peanut agglutinin (32). In mCd59b^+/+ and mCd59b^−/−, the percentage of acrosome reaction was similar (Fig. 3 E). Together, both sperm capacitation and the acrosome reaction experiments suggest that absence of mCd59 does not affect sperm capacitation and/or acrosome reaction.

To investigate further whether the role of CD59 in spermatogenesis is dependent on an efficient complement system, we generated mCd59b^+/+C3^+/+, mCd59b^−/−C3^+/+, mCd59b^+/+C3^−/−, and mCd59b^−/−C3^−/− littermates by intercrossing mCd59b^+/−C3^+/− followed by genotyping, as described in Materials and Methods. The functional deficiency of C3 in the different combinations involving the C3^−/− genotype was confirmed by measuring the serum complement activity as described in (33). Sperm analysis at 6 mo of age revealed that in both mCd59b^−/−C3^+/+ and mCd59b^−/−C3^−/− there was a similar decrease in the total and viable sperm number, as well as in sperm motility and viability as compared with sperm from mCd59b^+/+C3^+/+ (Fig. 4, A and B). In contrast, the functional deficiency of C3 rescued the hemolytic anemia phenotype that we originally described in mCd59b^−/− mice (7), as shown in Table I. Moreover, 17% of mCd59b^−/−C3^+/+ mice had a hemoglobin value below 2 SD below the mean in wild-type mice (i.e., below 13 g/dl), a fraction significantly higher than that in the mCd59b^−/−C3^−/− (4.5%) or in the mCd59b^+/+C3^+/+ (3.9%) groups (Fig. 5,A). Similar results were obtained for reticulocyte counts: 15.8% of mCd59b^−/−C3^+/+ littermates had a reticulocyte count 2 SD above the mean of their mCd59b^+/+C3^+/+ wild-type counterparts, a fraction that is also significantly higher than that in the mCd59b^−/−C3^−/− (4.5%) or in the mCd59b^+/+C3^+/+ (3.9%) groups (Fig. 5 B).

To understand the incomplete penetrance of the hemolytic anemia of mCd59b^−/−, we assessed the complement activity among mCd59b^−/− mice. The percent lysis by each mouse serum is dependent on the activity of C in that particular serum sample. Fig. 5,C demonstrates that there is a significant variation in the complement activity between mCd59b^−/−, which may explain the incomplete penetrance of the hemolytic anemia in mCd59b^−/− (a similar variation occurs in wild-type mice; data not shown). In summary, the deficiency of C3 in mCd59b^−/−C3^−/− rescues the hemolytic anemia but not the sperm abnormalities of mCd59b^−/−, a result consistent with a complement-independent function of mCd59 in spermatogenesis. The alternative explanation that C3 could have a role in spermatogenesis that would result in sperm abnormalities in C3^−/− mice (independently of the absence of CD59) was ruled out because all parameters of sperm function and quantity were similar in 10-mo-old mCd59b^+/+C3^−/− and mCd59b^+/+C3^+/+ littermates, a result consistent with the apparently normal fertility of mCd59b^+/+C3^−/− (Fig. 6, A and B).

A critical role of complement in reproduction has been experimentally characterized by two recent studies showing the deleterious effect on the fetus of unrestricted or increased complement activation, as seen respectively in pregnant mice after either functional deletion of Crry (8, 9) or injection of anti-phospholipid Abs (10). A potential role of complement in sperm development, maturation, and fertilization (38) has long been suspected, but its formal demonstration has remained elusive. Because complement regulatory proteins, including CD59, decay accelerating factor (DAF), membrane cofactor protein (MCP), and S-protein, are highly expressed in spermatozoa and present in seminal plasma (5, 6, 39), it has been postulated that they could be critical in protecting sperm against complement attack in the female reproductive tract, and/or play physiological roles in spermatogenesis/sperm maturation. However, several molecularly engineered mice already developed to study the function of complement regulatory proteins, including two mDAF knockout mice (40), and mCd59a- (33) and S-protein-deficient mice (41) did not show any major fertility problem. Several in vitro studies suggested a role for MCP in sperm-egg interaction (42), but the targeted deletion of mouse MCP did not affect their reproductive efficacy (43). Together, these studies illustrate that in mice, DAF, mCd59a, MCP, and S-protein are not individually essential for fetal development/survival or for protecting sperm from complement activation in the male or the female reproductive system. In contrast, the targeted deletion of the mCd59b gene, resulting in the down-regulation of mCd59a and functional deficiency of mCd59b, leads to a progressive loss of male fertility, as we have reported in (7).

The results of the work presented in this study show that mCd59b^−/− mice have a significantly lower number and compromised quality of sperm, associated with the presence of abnormal multinucleated cells, and an increased number of apoptotic cells within the seminiferous tubules. The multinucleated cells appear to be arrested spermatogonia that have divided without cytokinesis, an indication that spermatogenesis does not progress normally inside the seminiferous tubules. These results are consistent with a defect in spermatogenesis and would explain the intriguing loss of male fertility that progressively occurs in mCd59b^−/− mice. Interestingly, this spermatogenesis defect seems to result from a complement-independent function of CD59 because 1) it is not associated with an increased deposition of MAC in the seminiferous tubules and 2) the functional deficiency of C3, which for all practical purposes makes mice complement insufficient, does not rescue the abnormal testicular phenotype, although it does rescue the complement-dependent hemolytic anemia (Figs. 4–6 and Table I). Regarding the rescue of the hemolytic anemia of mCd59b^−/− by C3 deficiency, it is worth noting that this phenotype does not have a 100% penetrance in vivo, although 100% of the animals tested exhibited a significantly increased sensitivity to complement-mediated lysis assayed against a pool of complement-efficient mouse serum (7). This lower penetrance of the phenotype in vivo is not unexpected because the hemolytic anemia in mCd59b^−/− is the net result of the complement activity in the plasma of the individual mouse and the level of complement restriction due to the functional absence of Cd59. Assuming that all other complement regulatory factors are similar among individual mCd59b^−/− mice, the main remaining variable that might influence the penetrance of the hemolytic anemia phenotype is the complement activity in each mouse. It has been established that there are significant differences in the levels of complement activity in different strains of mice, between males and females and also among same gender within the same genetic background (Fig. 5 C) (33, 44, 45). The known difference in complement activity among individual mice is the most likely explanation for the observation that mCd59b^−/−C3^+/+ mice as a group exhibit anemia, i.e., a significantly lower blood hemoglobin value than normal littermate controls, but in only 17% of them, hemolysis is sufficient to bring this value below 2 SD of the mean in wild-type mice. In summary, C3 deficiency rescues the hemolytic anemia but not the abnormal reproductive phenotype of mCd59b^−/− mice, indicating that the later is most likely complement independent.

At present, we cannot completely rule out the alternative possibility of a local production of the C3 complement component that could take place in the testes of C3 knockout mice via an alternative splicing product of C3. However, this alternative explanation seems unlikely because 1) MAC deposits were absent in the testes of mCd59b^−/− mice; 2) a study designed to investigate the role of local complement production in renal transplantation using the same C3 knockout mice clearly indicates that there is no local production of C3 in the kidneys of these mice (46); and 3) the C3 knockout mice are also deficient in acylation-stimulation protein, a cleavage product of adipocyte C3 that contributes to fat tissue metabolism (47).

The successful reproduction of young mCd59b^−/− males, and the demonstration of normal capacitation-associated tyrosine phosphorylation and ionophore-induced acrosome reaction in mCd59b^−/− sperm, strongly support the interpretation that mCd59 may play a complement-independent role in spermatogenesis rather than in fertilization. This is consistent with the evidence that expression of mCd59b in spermatozoa progressively increases during their development and maturation (17, 18). Our in vivo data showing decreased motility of sperm in mCd59b^−/− is consistent with the previous demonstration by two different groups that spermatozoa ejaculated from healthy human donors lost their motility after exposure to anti-human CD59 Abs (48, 49).

Although this study does not define the complement-independent role of CD59 in reproduction, it is tempting to speculate that CD59 might protect sperm and/or testicular germ cells from attack by a yet not characterized mechanism independent of complement. One recent example of an alternative function in reproduction for a well-known protein is the finding that angiotensin converting enzyme, a well-characterized zinc peptidase, plays an unexpected biological role in reproduction cleaving GPI-anchored proteins from the plasma membrane (50). Alternatively, the role of CD59 in reproduction could be related to the localization of CD59 in the sperm lipid rafts. Recently, Cross et al. (51) have shown that, in noncapacitated sperm, CD59 localizes to lipid rafts, and that it redistributes after disruption of lipid rafts during the capacitation process. The molecular mechanism of this postulated new role of CD59 in reproduction is currently under investigation in our laboratory.

The number of multinucleate and apoptotic cells was significantly increased in mCd59b^−/− as compared with mCd59b^+/+ mice (Fig. 1, A–D). The underlying mechanism of increased multinucleate and apoptotic cells in mCd59b^−/− needs further investigation. Of note, the number of apoptotic cells was increased in some mCd59b^−/− (four of six) but not all mCd59b^−/− (Fig. 1 D). This heterogeneous phenotype may be due to a different proportion of B6 and 129 genetic background in our experimental mice. It is well accepted that the genetic background largely influences phenotypic expression. To minimize the impact of difference genetic backgrounds, we used littermates (derived from crossing heterozygous pairs); however, this does not totally eliminate differences in genetic background that may affect phenotypic expression.

The progressive nature of the loss of male fertility in mCd59b^−/− is intriguing. It indicates that mCd59 does not play a role that is indispensable for sperm development/survival, in which case one would expect the male infertility to be early and absolute from the beginning of the mouse male fertile cycle, rather than progressive. Even though we observed a significant reduction in the total number of normal, motile sperm at the early age of 5 mo in mCd59b^−/− (Figs. 2 and 3,C), the density of normally motile sperm at that age seems to be above the minimal threshold required for a successful fertilization. In contrast, the density of normal sperm at the older age of 10 mo is not only further reduced (Fig. 5) but apparently below the fertilization threshold, explaining why older mice become completely infertile.

CD59 is a conserved gene universally expressed in most animal species (17). Mice seem to be unique in expressing two Cd59 genes, whereas all other species express only one (17). The presence of two CD59 genes in mice appears to be the result of gene duplication rather than loss of one copy in the genome of humans and other species. This interpretation is based on our finding of early transposon or L1 retro-transposon repeat sequences located in the 5′ flanking region of both mCd59a and mCd59b genes in the mouse genome (17). These transposon repeat sequences are now considered as motile or transportable genomic elements that have contributed to the evolution of species through different genomic pathways (52). Also, retro-transposition is believed to increase the probability of gene duplication and to generate insertional mutations responsible for some genetic diseases (53). Therefore, it seems possible that retro-transposons caused the gene duplication event that resulted in two Cd59 genes in the mouse genome (17). A combination of a complement-regulatory and a reproduction-specific function of mCd59b would have conferred a selective advantage that preserved the presence of both genes in the mouse genome. However, it is also possible that the reproduction-specific function is shared by both mCd59a and mCd59b mouse genes. The published work on mCd59a^−/− does not mention any reproduction abnormality (33, 44); this may be due to the absence of a reproduction-specific function in the mCd59a gene or to compensation by the mCd59b gene, which is highly expressed in testis. In mCd59b^−/−, residual mCd59a might not compensate for the loss of a reproductive-specific function in testes because 1) mCd59a does not have the reproduction-specific function, 2) the level of expression of residual mCd59a in testis is not high enough to compensate, or 3) there is a down-regulation of mCd59a, as we have documented at the level of mRNA (see Fig. 5 A in Ref.7), and more recently by FACS analysis of mCd59b RBC, using mCd59a- and mCd59b-specific Abs (X. Qin et al., manuscript in preparation) (18). The issue of the reproduction-specific function of the mCd59 genes will be resolved by the availability of mCd59a and mCd59b double-knockout mice followed by knockin rescue experiments with either gene.

Although mCd59b has a slightly higher homology with human CD59 than mCd59a (44 and 41%, respectively) (16), the available structural information does not allow us to ascertain whether mCd59a or mCd59b is the conserved gene in the human genome. At the functional level, it will be of great interest to determine whether CD59 also plays a physiological function in human sperm development.

We are grateful to Dr. D. C. Tosteson and Dr. M. T. Tosteson for continuous support and critical comments. We are grateful to Dr. Kenney for testes histological analysis, Dr. H. Aktas for lending his expertise in sperm functional analysis, and to Dr. L. Grubissich for assistance in the animal handling. We are also grateful to K. Hazard for careful editing of the manuscript.

The authors have no financial conflict of interest.