NOX5 in Human Spermatozoa

Background: The identity of the reactive oxygen species (ROS)-producing enzyme(s) in human spermatozoa remains uncertain. Results: NOX5 NADPH oxidase, but not NOX1/2/4, is expressed in human spermatozoa and produces superoxide. Inhibition of NOX5 activity reduces spermatozoa motility. Conclusion: NOX5 is the main source of superoxide and is implicated in human spermatozoa motility. Significance: NOX5 might control the numerous ROS-dependent (patho)physiological processes in human spermatozoa. Physiological and pathological processes in spermatozoa involve the production of reactive oxygen species (ROS), but the identity of the ROS-producing enzyme system(s) remains a matter of speculation. We provide the first evidence that NOX5 NADPH oxidase is expressed and functions in human spermatozoa. Immunofluorescence microscopy detected NOX5 protein in both the flagella/neck region and the acrosome. Functionally, spermatozoa exposed to calcium ionophore, phorbol ester, or H2O2 exhibited superoxide anion production, which was blocked by addition of superoxide dismutase, a Ca2+ chelator, or inhibitors of either flavoprotein oxidases (diphenylene iododonium) or NOX enzymes (GKT136901). Consistent with our previous overexpression studies, we found that H2O2-induced superoxide production by primary sperm cells was mediated by the non-receptor tyrosine kinase c-Abl. Moreover, the HV1 proton channel, which was recently implicated in spermatozoa motility, was required for optimal superoxide production by spermatozoa. Immunoprecipitation experiments suggested an interaction among NOX5, c-Abl, and HV1. H2O2 treatment increased the proportion of motile sperm in a NOX5-dependent manner. Statistical analyses showed a pH-dependent correlation between superoxide production and enhanced sperm motility. Collectively, our findings show that NOX5 is a major source of ROS in human spermatozoa and indicate a role for NOX5-dependent ROS generation in human spermatozoa motility.

Physiological and pathological processes in spermatozoa involve the production of reactive oxygen species (ROS), but the identity of the ROS-producing enzyme system(s) remains a matter of speculation. We provide the first evidence that NOX5 NADPH oxidase is expressed and functions in human spermatozoa. Immunofluorescence microscopy detected NOX5 protein in both the flagella/neck region and the acrosome. Functionally, spermatozoa exposed to calcium ionophore, phorbol ester, or H 2 O 2 exhibited superoxide anion production, which was blocked by addition of superoxide dismutase, a Ca 2؉ chelator, or inhibitors of either flavoprotein oxidases (diphenylene iododonium) or NOX enzymes (GKT136901). Consistent with our previous overexpression studies, we found that H 2 O 2induced superoxide production by primary sperm cells was mediated by the non-receptor tyrosine kinase c-Abl. Moreover, the H V 1 proton channel, which was recently implicated in spermatozoa motility, was required for optimal superoxide production by spermatozoa. Immunoprecipitation experiments suggested an interaction among NOX5, c-Abl, and H V 1. H 2 O 2 treatment increased the proportion of motile sperm in a NOX5-dependent manner. Statistical analyses showed a pH-dependent correlation between superoxide production and enhanced sperm motility. Collectively, our findings show that NOX5 is a major source of ROS in human spermatozoa and indicate a role for NOX5-dependent ROS generation in human spermatozoa motility.
MacLeod (1) demonstrated as early as 1943 that spermatozoa were able to generate reactive oxygen species (ROS). 2 ROS production by spermatozoa correlates with lipid peroxidation, DNA oxidation, poor sperm function, and reduced fertility (2,3). However, recent evidence suggests that redox activity is physiologically important in promoting normal sperm function (4 -6). Generation of ROS is required for sperm capacitation, the final maturation steps associated with hyperactive motility and a physiological acrosome reaction (7,8). Despite evidence for both physiologic and pathologic effects of ROS in spermatozoa, the identity of the ROS-producing enzyme(s) remains uncertain.
Numerous studies showed similarity between the ROS-producing enzyme in spermatozoa and the enzymatic system expressed in phagocytes. In these cells, the major ROS precursor is the superoxide anion, which is generated by the singleelectron reduction of molecular oxygen by the NOX2 NADPH oxidase complex. NOX5 is a novel isoform of the NOX family of NADPH oxidases that is activated by the binding of Ca 2ϩ to its cytosolic N-terminal EF-hand domains (9,10). It is suggested that Ca 2ϩ binding induces a conformational change leading to the interaction of the N terminus of NOX5 with its C-terminal substrate-binding and flavin-containing domains, thereby inducing ROS production (9,10). Among the five splice variants identified, NOX5␣, NOX5␤, NOX5␥, and NOX5␦ (all long forms of NOX5 designated collectively as NOX5L) differ in the N-terminal sequences of their Ca 2ϩ -binding domains, whereas the truncated NOX5S (short) variant lacks the Ca 2ϩ -binding domains altogether (11). Although the NOX5 gene is widely conserved evolutionarily, it is curiously absent from rodent genomes (12). NOX5 was found in a prostate carcinoma cell line (13), hairy cell leukemia B-cells (14), endothelial cells (15), smooth muscle cells (16), stomach (17), cardiac fibroblasts (18), and Barrett esophagus epithelia (19).
NOX5 mRNA was initially detected in human testis in 2001 (9), and a Ca 2ϩ ionophore was shown to induce ROS generation in human spermatozoa (20). NOX5 protein was detected by immunoblot analysis in equine sperm (21). Although these observations suggest that spermatozoa might express NOX5, to date there are no reports that demonstrate NOX5 expression at the protein level in human spermatozoa. Herein, we present data documenting both the expression and function of NOX5 in human spermatozoa.

EXPERIMENTAL PROCEDURES
Semen Samples and Sperm Isolation-Human semen samples were obtained from patient donors as a part of their clinical evaluation for fertility and reproductive health status. Portions of the samples not required for clinical analyses were de-identified and made available for the research protocols described herein. Under these conditions, the Institutional Review Board determined that the study is exempt from requirements for protocol approval and informed consent. Semen was collected by masturbation into sterile plastic containers and allowed to liquefy for at least 30 min before being processed. The samples were centrifuged (500 ϫ g, 15 min) to separate the seminal plasma from spermatozoa, and the pellet washed twice in phosphate-buffered normal saline (pH 7) with 10 mM glucose (PBS-G). The spermatozoa were then counted in a hemocytometer chamber, resuspended in PBS-G at 10 9 /ml, and used for superoxide assays. Sperm samples containing leukocytes detected by microscopic examination were excluded from the study. For selected experiments, motile spermatozoa were collected using the swim-up procedure (22,23). Briefly, the spermatozoa were sedimented by centrifugation (500 ϫ g, 15 min) and the tube was then incubated upright for 60 min at 37°C. Motile spermatozoa that progressed from the pellet into the supernatant (swim-up fraction) were collected by aspiration and counted.
Immunoprecipitation-Total protein extracts of K562, K562/NOX5, or human sperm cells prepared in buffer A (250 g of protein) were pre-cleared with rabbit IgG (Sigma) and protein A/G-agarose (Santa Cruz Biotechnology), incubated overnight with anti-NOX5, anti-H V 1, or nonspecific IgG antibody as a negative control, and precipitated with protein A/Gagarose for an additional 3 h. The immune complexes were washed with lysis buffer, separated on 4 -20% SDS-polyacrylamide gradient gels, and transferred to PVDF membranes. The filters were incubated with anti-NOX5, anti-H V 1, or anti-phosphotyrosine 245 c-Abl antibodies. The antigen-antibody complexes were visualized by enhanced chemiluminescence (ECL, Amersham Biosciences).
Immunostaining-For immunohistological analyses, a drop of spermatozoa suspended in PBS-G was placed on a poly-Llysine-coated glass slide, fixed in 4% (w/v) paraformaldehyde, and permeabilized with 0.1% Triton X-100 (v/v) for 15 min at 4°C. Spermatozoa were then subjected to immunostaining using the antibodies described above against NOX5, c-Abl, and phospho-Tyr 245 -c-Abl along with Cy TM 3-or FITC-conjugated secondary antibody (Zymed Laboratories Inc., San Francisco, CA). Protein expression was visualized on a fluorescence microscope (Zeiss). For negative controls, cells were incubated with isotype-matched control antibodies with no known specificity.
Superoxide Assay-Superoxide generation was measured using a luminol-based chemiluminescence assay (Diogenes reagent, National Diagnostics, Atlanta, GA). We have previously demonstrated the specificity of this assay for superoxide versus H 2 O 2 (24). Spermatozoa were suspended at 10 9 /ml in PBS-G. A 100-l aliquot of the Diogenes reagent was mixed with 2 ϫ 10 7 spermatozoa and incubated at 37°C for 2-4 min. Superoxide generation was measured at baseline and after stimulation with ionomycin (1 M), phorbol myristate acetate (PMA, 1 g/ml), or H 2 O 2 (100 M and other concentrations as indicated). Chemiluminescence was measured every 30 -60 s for up to 10 -15 min using a Clarity TM BioTek luminometer and an integration time of 5 s.
Live Cell Imaging-Spermatozoa suspended in PBS-G were plated on poly-L-lysine-coated glass coverslips (2 ϫ 10 5 cells/ ml) in PBS-G and loaded with 6 M fluo-4AM and 300 nM dihydroethidine (Molecular Probes, Carlsbad, CA) 10 min before imaging. Ca 2ϩ influx and superoxide production were recorded using a confocal microscope (LSM 510; Carl Zeiss MicroImaging, Inc.) with a ϫ63/1.4 NA plan-apochromat objective. Fluo-4 and dihydroethidine fluorescence emissions were excited with dual laser at 488/543 nm attenuated to avoid photobleaching and saturation. Detection was through a 545-nm long-pass dichroic mirror and a band-pass filter at 500 -530 for fluo-4 and LP560 for dihydroethidine.
Cell Culture and Down-regulation of H V 1 Protein by siRNA-Human leukemia cells stably transfected with an empty vector (K562 cells) or a vector encoding for NOX5 (K562/NOX5) were grown in RPMI medium supplemented with 10% (v/v) fetal bovine serum, plus 100 units/ml of penicillin, and 100 g/ml of streptomycin and the appropriate selection antibiotic. Cells in the logarithmic phase of growth were transfected with a mixture of H V 1 siRNA (Sigma) or with scrambled siRNA (Sigma) using the Neon system (Invitrogen) according to the manufacturer's recommendations. Cells were tested 48 h after transfection for superoxide production and down-regulation of H V 1 protein was detected by immunoblot using the affinity-purified rabbit polyclonal H V 1 antibody raised against an N-terminal domain of H V 1 human protein (25).
Patch Clamp Analysis-Micropipettes were pulled in several stages using a Flaming-Brown automatic pipette puller (Sutter Instruments, San Rafael, CA) from 7052 glass (Garner Glass Co., Claremont, CA), coated with Sylgard 184 (Dow Corning Corp., Midland, MI), heat polished to a tip resistance between 2 and 10 megaohms. Electrical contact with the pipette solution was achieved by a chloride silver wire. The reference electrode was made from a Ag-AgCl wire connected to the bath through an agar bridge made with Ringer solution (160 mM NaCl, 4.5 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES adjusted to pH 7.4). Current signal from the Axopatch 200B (Axon Instruments, Foster City, CA) was recorded on a PC through a Lab View SCB-68 (National Instruments, Austin, TX). Further data analysis was made with Origin 7.5. Seals were formed in Ringer solution in the bath, and the potential was zeroed when the pipette was in contact with the cell. Most measurements were done between 21 and 25°C.
K562 cells were transferred on chips or pipette into the recording chamber. To leave internal signaling pathways undisturbed, the perforated patch configuration was used. Low access resistance was usually reached after 5-10 min. The perforated patch recording bath and pipette solution contained 50 mM NH 4 ϩ in the form of (NH 4 ) 2 SO 4 , 2 mM MgCl 2 , 5 mM Bes or Pipes buffer, and 1 mM EGTA adjusted to pH 7.0 with tetramethylammonium hydroxide. Solubilized amphotericin B (500 g/ml; ϳ45% purity, Sigma) was added to the pipette solution, after first dipping the tip into amphotericin-free solution. To investigate the proton channels "enhanced gating mode," 60 nM PMA was used to stimulate the cells. The protein kinase C (PKC) inhibitor GF109203X (GFX; 3 M) was used to revert the enhanced gating mode. Holding potential for current families measurements was Ϫ60 mV. The holding potential for test pulses applied during application of either PMA or GFX was Ϫ30 mV.
Statistical Analysis-Natural log transformations were performed for clinical measures of sperm production and fertility (e.g. total number of motile spermatozoa, days of abstinence) and superoxide production prior to analysis. Sperm progression and pH were recoded into three-level categorical variables Յ2, 3, Ͼ3, and Ͻ8, 8 -8.5, Ͼ8.5, respectively. The Kruskal-Wallis test was used to evaluate the relationship between pH (Ͻ8, 8 -8.5, and Ͼ8.5) and continuous measures of sperm characteristics including superoxide production, whereas Fisher exact tests were used when the comparison was made with categorical measures of sperm characteristics. Correlations between continuous sperm characteristics and superoxide production were assessed using Spearman rank order correlations. Regression analyses were performed with sperm characteristics as the response variable and each of the ROS measures as explanatory variables. The regression models were run controlling for pH and log abstinence. All statistical tests were twosided with a significance level of 5%. SAS software version 9.2 (SAS Institute, Cary, NC) was used throughout.
A Dunnett multiple comparison test was used for many of the in vitro experiments to adjust for multiple testing when comparing several means against the mean for a common control sample. A two-sided value of p Ͻ 0.05 was accepted as significant.

NOX5 Expression in Human
Spermatozoa-A lysate of pooled spermatozoa from several donors was used for immunoblot analysis with antibody specific for NOX5 or NOX2. The NOX5 antibody detected two main immunoreactive species with apparent molecular masses of ϳ80 and ϳ65 kDa (Fig. 1A), whereas no bands were observed using NOX2 antibody (Fig.  1A). Conversely, human neutrophil (PMN) lysates analyzed in parallel exhibited the presence of NOX2 but not NOX5. In addition, immunoblot analysis of sperm lysates using NOX1 or NOX4 antibody resulted in no detectable signals, whereas appropriate immunoreactive bands were detected in lysates of CaCo 2 and kidney cells (Fig. 1A), which are known to express NOX1 and NOX4, respectively. These results suggest that the predominant NOX enzyme expressed in human spermatozoa is NOX5.
Because the NOX5 antibody was generated against a GST peptide corresponding to the N terminus of NOX5, including the EF-hand domains present in the long splice variants, such as NOX5␤, it was possible that the antibody detected more than one long NOX5 variant. However, because the sequence of the peptide immunogen extended 10 amino acids into the predicted N-terminal domain of the short NOX5 variant, the more rapidly migrating band could have represented NOX5S. To test this possibility we performed immunoblot analysis using anti-NOX5 and anti-Myc on lysates of K562 cells stably overexpressing either NOX5␤ (24) or Myc-tagged NOX5S (construct kindly provided by Guangjie Cheng, University of Alabama at Birmingham). The data show that the NOX5 antibody recognized only NOX5␤, whereas anti-Myc detected only Myc-NOX5S (Fig. 1B). Moreover, Myc-NOX5S exhibited an apparent molecular mass of ϳ55 kDa, which is substantially smaller than either of the bands detected in spermatozoa. We conclude that the species recognized by anti-NOX5 in sperm lysates represent EF-hand-containing NOX5L splice variants. However, because their apparent molecular sizes do not match precisely to the NOX5␤ expressed in K562 cells, the exact sequences remain to be determined to identify the predominant NOX5L variants present in human spermatozoa.
Immunostaining experiments confirmed the presence of NOX5 in human spermatozoa and showed its localization in the sperm flagellum, neck, and acrosome regions (Fig. 1C). These immunoblotting and immunostaining results together demonstrate for the first time that NOX5 is expressed in human spermatozoa.
Ca 2ϩ -dependent Superoxide Production by Human Spermatozoa-Because the binding of Ca 2ϩ to N-terminal EF-hand domains activates NOX5L forms, we assessed whether NOX5 was functional in human sperm by testing the ability of intact spermatozoa to produce superoxide upon exposure to the Ca 2ϩ ionophore ionomycin ( Fig. 2A). Using the superoxide-specific luminol-based Diogenes reagent chemiluminescence assay (24), we found that ionomycin induced vigorous superoxide production by human spermatozoa. The chemiluminescence signal detected was blocked by addition of SOD, the flavoprotein oxidase inhibitor diphenylene iodonium (DPI), or the Ca 2ϩ chelator BAPTA ( Fig. 2A), indicating that under these conditions we were detecting generation of superoxide anion by a Ca 2ϩ -dependent flavoprotein oxidase, such as NOX5. Similar patterns of superoxide generation and blocking agent effects were observed when spermatozoa were exposed to PMA (Fig.  2B), a potent phorbol ester activator of PKC that is known to regulate NOX5 activity (28).
We also tested the effect of a newly described NOX inhibitor, GKT136901, to further investigate whether NOX5 accounts for Ca 2ϩ -dependent superoxide production. This compound is a potent inhibitor of both NOX1 and NOX4 (K i ϭ 160 Ϯ 10 and 165 Ϯ 5 nM, respectively) with a ϳ9-fold selectivity over NOX2 (K i ϭ 1530 Ϯ 90 nM) (26,27). Importantly, when tested under the same conditions as described by Laleu et al. (26) GKT136901 showed an inhibitory constant of 450 Ϯ 10 nM for NOX5. 3 Our results show that in sperm cells, GKT136901 inhibited the stimulation of superoxide production by PMA in a concentration-dependent manner. GKT136901 was ϳ10-fold more potent in inhibiting superoxide production stimulated by PMA in sperm cells as compared with neutrophils (Fig. 2C). Because human spermatozoa do not express detectable levels of NOX1 or NOX4, these results strongly suggest that the NOX5 protein expressed in human spermatozoa is the functionally active, superoxide-generating enzyme.
Effect of H 2 O 2 on Superoxide Production by Human Spermatozoa-We have previously demonstrated that exogenous H 2 O 2 induces NOX5 activity in an overexpression system (24). Testing this regulation in human primary cells, we found that H 2 O 2 induced a marked burst in superoxide production by human spermatozoa, beginning within 1-2 min after peroxide addition and achieving maximal activity levels, varying somewhat among donors, in 5-10 min (Fig. 3A). The induction of superoxide generation was dose-dependent from 1 to 500 M H 2 O 2 (Fig. 3B). Subsequent experiments were performed using a standard concentration of 100 M H 2 O 2 . As for superoxide generation induced by ionomycin or PMA, the chemiluminescence signal generated by exposure to H 2 O 2 was nearly completely abrogated by the addition of SOD, BAPTA, DPI, or 3 GenKyoTex, personal communication. GKT136901 (Fig. 3C). Because we have shown that superoxide production by phagocyte NOX2 is also regulated by H 2 O 2 , we ensured that leukocytes did not account for the superoxide signal stimulated by H 2 O 2 in sperm samples by discarding all semen samples in which leukocytes were evident by microscopic observation.
To confirm that spermatozoa produce superoxide when stimulated by H 2 O 2 , confocal images of spermatozoa freshly loaded with the red fluorescent superoxide probe dihydroethidine were recorded. In addition, we also loaded sperm cells with the Ca 2ϩ indicator dye fluo4-AM to detect cytosolic Ca 2ϩ responses in parallel with superoxide production. Spermatozoa exposed to H 2 O 2 exhibited increases in cytosolic Ca 2ϩ levels (green fluorescence), as well as the number and intensity of red fluorescing cells, reflecting superoxide generation (Fig. 3D). Together with the chemiluminescence data (Fig. 3, A-C), these confocal imaging results strongly suggest that superoxide production induced by H 2 O 2 was due to Ca 2ϩ -mediated activation of the NOX5L proteins expressed in human spermatozoa. The partial co-localization of Ca 2ϩ influx and superoxide produc-tion suggests partial temporal overlap of these events, as we previously found in NOX5 overexpression studies (24).

Role of c-Abl in H 2 O 2 -induced Sperm Superoxide Production-
We showed previously that c-Abl tyrosine kinase plays a central role in the induction of NOX5 activity in response to H 2 O 2 (24). To investigate the potential role of c-Abl in primary human spermatozoa, we assessed its expression and regulation by H 2 O 2 . Immunoblot analysis showed that c-Abl was expressed in spermatozoa (Fig. 4A). Immunostaining demonstrated the presence of c-Abl in the flagellum, neck, and acrosome regions, as well as co-localization with NOX5, as assessed by double immunostaining (Fig. 4B). Stimulation of spermatozoa by H 2 O 2 induced the phosphorylation of c-Abl on tyrosine 245, an indication of its activation (Fig. 4C), as demonstrated by both immunoblot and immunostaining. Pretreatment of human spermatozoa with the selective c-Abl inhibitor imatinib mesylate significantly decreased superoxide production induced by H 2 O 2 (Fig. 4D). These results suggest that H 2 O 2 -NOX5 regulation in human spermatozoa is mediated through a c-Abl-dependent signaling pathway.

Superoxide Production by NOX5 Requires H V 1 Proton
Channel-It has been recently reported that H V 1 voltage-regulated proton channels present in human spermatozoa are implicated in their motility (29). Because H V 1-dependent proton efflux has been shown to compensate membrane depolarization and intracellular acidification arising from phagocyte NOX2 enzymatic activity (30 -33), we hypothesized that the activity of this channel is associated with NOX5-dependent superoxide production in human spermatozoa.
We investigated the potential interaction of NOX5 and H V 1 using K562 cells overexpressing NOX5␤. We evaluated the impact of H V 1 function on NOX5 activity using Zn 2ϩ , a potent inhibitor of H V 1 proton currents (34), or siRNA to down-regulate the expression of H V 1. We observed that induction of superoxide production in K562/NOX5 cells stimulated by ionomycin, PMA, or H 2 O 2 was decreased by 1 mM ZnCl 2 (Fig.  5A) or H V 1 siRNA (Fig. 5B), by 50 -70 or 30 -50%, respectively. Furthermore, immunoprecipitation experiments showed the presence of NOX5 protein in H V 1 immunoprecipitates and H V 1 protein in NOX5 immunoprecipitates from lysates of K562/NOX5 cells, but not control K562 cells (Fig. 5C). These results suggest the existence of a specific interaction between H V 1 and NOX5 proteins.
Proton currents were consistently recorded in both K562 (4 of 5 patched cells) and K562/NOX5 cells (12 of 14 cells) voltageclamped in a perforated-patch configuration (Fig. 5D) (35). The current density (i.e. current/membrane area) at 40, 50, 60, and 70 mV was larger in K562 cells than in K562/NOX5 cells by a factor of 1.30 on average, although these differences were not statistically significant. The proton channels responded to PMA stimulation with a characteristic array of changes in their behavioral properties, designated enhanced gating mode (negative shift of the g H -V relationship, faster activation, slower deactivation, increased g H,max ) (35)(36)(37). Most features of the PMA response of proton channels were reversed by the addition of the PKC inhibitor GFX (Fig. 5D) (34,38). These data demonstrate that both K562 cells and K562/NOX5 cells expressed functional proton channels with the characteristic features of H V 1. Collectively, our data in K562 and K562/NOX5 cells demonstrate that a functional interaction exists between H V 1 and NOX5, wherein H V 1 is required for sustaining NOX5dependent superoxide production.
Similar to the K562/NOX5 cell studies, we found, in human spermatozoa that superoxide production induced by H 2 O 2 , PMA, or ionomycin was Zn 2ϩ -sensitive (Fig. 6A) and that H V 1 co-immunopreciptated with NOX5, when the immunoprecipitation was performed with either H V 1 or NOX5 antibodies (Fig.  6B). Interestingly, the activated form of c-Abl (p-c-Abl Tyr-245 ) implicated in NOX5 regulation was also detected in H V 1 immunoprecipitates (Fig. 6B), suggesting the existence of a trimolecular complex comprising H V 1, NOX5, and activated c-Abl. Taken together, these results provide strong evidence for  MARCH 16, 2012 • VOLUME 287 • NUMBER 12 a functional interaction between H V 1 and NOX5 in human spermatozoa.

NOX5 in Human Spermatozoa
Effect of H 2 O 2 on Sperm Motility-Because low levels of H 2 O 2 can induce sperm capacitation (8,39,40) and because H V 1 interacts with NOX5, we tested the effect of H 2 O 2 on sperm cell motility. For this we incubated sperm cells with or without 100 M H 2 O 2 for 10 min and then collected motile spermatozoa following the swim-up procedure. The incubation of spermatozoa with H 2 O 2 resulted in an increase in the number of sperm in the swim-up fraction (Fig. 7A), suggesting that H 2 O 2 enhanced spermatozoa motility. To investigate whether the effect of H 2 O 2 on motility was due to NOX5, we performed the swim-up experiment in the presence of NOX inhibitors. When spermatozoa were preincubated with DPI or GKT136901, the effect of H 2 O 2 on motility was abrogated. These results strongly suggest that the induction of spermatozoa motility by H 2 O 2 is dependent on NOX5 activity.
We observed that the effect of H 2 O 2 on motility was especially prominent in semen samples that had the lowest baseline percentages of motile sperm. In addition, when we measured superoxide production using spermatozoa collected from the upper fraction after the spontaneous (i.e. without H 2 O 2 ) swim-up procedure, these hypermotile spermatozoa did not produce superoxide when stimulated by H 2 O 2 compared with pre-swim-up cells (Fig. 7B). This finding indicated that following the hyperactivation process spermatozoa are no longer responsive to H 2 O 2 .
Semen pH-dependent Correlation between Superoxide Production and Sperm Motility-Sperm samples (n ϭ 128) were assessed according to the World Health Organization (WHO) laboratory manual (41) for the examination and processing of human semen. Abstinence, defined as the number of days since the previous ejaculation, ranged from 1 to 14 days and was significantly associated with sperm concentration (r ϭ 0.28, p ϭ 0.001), total sperm cell count (r ϭ 0.27, p ϭ 0.002), number of motile sperm cells (r ϭ 0.24, p ϭ 0.005), and concentration of immature forms (r ϭ 0.32, p Ͻ 0.001) (data not shown).
It is known that pH increases during capacitation (42)(43)(44)(45). Considering the relationship between H V 1, an important regulator of pH in human spermatozoa, and NOX5, we analyzed the sperm characteristics as a function of semen pH (Table 1), which was categorized into three levels (Ͻ8, 8 -8.5, and Ͼ8.5) for these analyses. Significant differences were found between pH levels for semen viscosity (p ϭ 0.02), total sperm count (p ϭ 0.04), total number of motile sperm (p ϭ 0.02), percent of motile sperm (p ϭ 0.009), and sperm progressive motility (progression), which is defined as spermatozoa moving actively, either linearly or in a large circle (p ϭ 0.04). Progression was described according to the WHO manual (41) using an ordered categorical scale from 1 to 4, where a higher number indicates a higher progressive motility. Superoxide production did not vary significantly with semen pH. We did observe that a significant difference in basal superoxide production was associated with higher levels of sperm progression (Յ2, 3, and Ͼ3) (p ϭ 0.02). However, superoxide production stimulated by H 2 O 2 or PMA were not significantly associated with progression (data not shown).
A regression model was built to analyze for possible correlations among semen pH, sperm motility, and superoxide production, adjusting for abstinence and pH. There appears to be a nonlinear association (inverted U function) between pH and a number of sperm metrics, so we re-categorized pH into levels more conductive to sperm activity (8.0 -8.5) and those less con-ductive to sperm activity (Ͻ8 or Ͼ8.5) for our regression analyses. The regression model (Table 2) confirmed the previous observations (Table 1), because significant increases in sperm progressive motility (OR ϭ 3.38, 95% CI: 1.25 to 9.15, p ϭ 0.02) and percent of motile sperm (mean difference ϭ 14, 95% CI: 4 to 23, p ϭ 0.004) were observed for the pH 8.0 -8.5 group, compared with the group with either high or low pH values that are less conductive to sperm activity ( Table 2). Basal and stimulated (H 2 O 2 or PMA) levels of superoxide production tended   MARCH 16, 2012 • VOLUME 287 • NUMBER 12

JOURNAL OF BIOLOGICAL CHEMISTRY 9383
to be higher in the pH 8.0 -8.5 group relative to the pH Ͻ 8 or Ͼ8.5 group, although these associations were not significant (p ϭ 0.11, 0.29, and 0.18 for superoxide production measured under basal condition or stimulated by H 2 O 2 or PMA, respectively). Because of the small number of sperm samples in the low plus high pH group, the analysis of sperm motility and superoxide production was conducted only for the median pH group (8.0 -8.5) ( Table 3). There was a significant correlation between the percent of motile sperm and basal levels of super-

TABLE 2
Odds ratio or mean difference of sperm motility and superoxide production for pH 8 -8.5 versus pH <8 and >8.5 1 Ordinal logistic regression. 2 Progression is a 3-level categorical variable (Յ2,3, and Ͼ3). 3 Linear regression of each sperm characteristic in terms of pH with an adjustment for log abstinence.

NOX5 in Human Spermatozoa
oxide production. Although increased superoxide production stimulated by PMA or H 2 O 2 was significantly associated with higher progression (H 2 O 2 -stimulated superoxide production: OR ϭ 1.22, 95% CI: 1.0 to 1.5, p ϭ 0.05; PMA-stimulated superoxide production: OR ϭ 1.21, 95% CI: 1.01 to 1.45, p ϭ 0.04). These analyses demonstrate that semen pH and sperm superoxide production are closely associated with sperm motility and suggest that the correlation between superoxide production and sperm motility is pH-dependent (Fig. 7C).

DISCUSSION
Both physiologic and pathologic effects of ROS in spermatozoa have been described, but the identity of the ROS-producing enzyme(s) has remained unknown. Here, we report for the first time that NOX5 is expressed and functional in human spermatozoa and that its activity is implicated in spermatozoa motility.
Spermatozoa released from the testis are transcriptionally inert (46) and yet they undergo functional maturation to acquire ovum-fertilizing ability. These changes are believed to occur through the modulation of biochemical pathways and cellular properties by incoming signals from the sperm microenvironment, including ROS (47). The putative ROSproducing enzyme has been compared with the NADPH oxidase expressed in phagocytic leukocytes (48). Some studies have even suggested that spermatozoa themselves do not possess NADPH oxidase activity (48 -50), but rather that ROS generation is attributable to leukocytes that may be present in some sperm preparations (51)(52)(53). Our data show that spermatozoa express their own ROS-producing enzyme, namely one or more of the long splice variants of NOX5 NADPH oxidase, which produce superoxide in a Ca 2ϩ -dependent manner. Because a quantitative PCR approach is precluded by the fact that spermatozoa are transcriptionally inert, we tested human sperm samples for other NOX isoforms by immunoblot analyses using antibodies specific for NOX1, -2, and -4. For all donors tested, we were unable to detect any signals. Although we cannot categorically exclude the possibility that other NOX isoforms are expressed, these findings demonstrate that NOX5 is the predominant NOX family NADPH oxidase in human spermatozoa. Further studies are needed to identify the specific NOX5 splice variants expressed in human spermatozoa.
We reported previously (24) that H 2 O 2 positively regulates NOX5␤ via a Ca 2ϩ /c-Abl-dependent pathway. The current studies demonstrate that this regulation is also operative in human spermatozoa, as we observed that both the Ca 2ϩ chelator BAPTA and the c-Abl tyrosine kinase inhibitor imatinib mesylate abrogated H 2 O 2 -induced superoxide production. Furthermore, we provide the first evidence that a functional complex between NOX5 and c-Abl exist in human spermatozoa. ROS-induced tyrosine phosphorylation of several unidentified proteins appears to be essential for sperm capacitation (54). Reminiscent of our findings on the regulation of NOX5 by H 2 O 2 (24), two major spermatozoal proteins with molecular sizes similar to NOX5L and c-Abl have been reported to be regulated by PKA and H 2 O 2 in a protein-tyrosine phosphataseindependent manner (6,54,55). This observation further supports the functional importance of NOX5 and c-Abl in spermatozoa.
A recent report showed that the H V 1 proton channel is likely to be responsible for cytoplasmic alkalinization in human spermatozoa, and that enhanced proton currents correlate with sperm capacitation and hypermotility (29). The physiologic function of these channels has been characterized best in phagocytic leukocytes, where H V 1-dependent H ϩ efflux compensates for the membrane depolarization and intracellular acidification generated by the electrogenic NADPH oxidase enzymatic activity (30 -33). However, the complex relationship between H V 1 and NADPH oxidase remains incompletely understood. On one hand, robust proton currents in response to PMA are seen only in cells with a high level of NOX2 activity, such as neutrophils (35), eosinophils (56), and PLB-985 cells (57), whereas cells that do not have a high level of NOX2 activity exhibit a weaker response that differs in having a smaller shift of the g H -V relationship and little or no slowing of deactivation (channel closing) (34,58). On the other hand, some cell types lack any recordable proton current at all, irrespective of whether they express NADPH oxidase, for example, COS-7 cells (34). To date there is still no systematic correlation between the H V 1 proton current and the expression of a functional NOX. These observations may reflect the complexity of cellular systems that regulate membrane depolarization and pH homeostasis, suggesting that multiple proton generating systems can be coupled to H V 1 and that multiple proton transporters may be linked to NOX enzymes (59,60). It is also possible that the experimental conditions required to measure proton channel currents are not optimal for detecting weak amplitude proton currents that may result from the activity of NOX isoforms other than NOX2.
We found that the proton current in the response to PMA stimulation in K562 cells was similar in amplitude, channel opening and closing kinetics, and the position of the g H -V relationship, whether or not NOX5 was heterologously expressed. Because PMA can activate NOX5 and proton channels independently (28,61), it is difficult to determine whether there is a NOX5-dependent proton current. Unfortunately, in these cells TABLE 3 Association between superoxide production and progression or percent of motile sperm for pH 8 -8.5 Analysis was performed among those with pH levels 8 -8.5 (n ϭ 107). a Ordinal logistic regression of progression in terms of each measure of superoxide production with an adjustment for log abstinence. b Linear regression of percent of motile sperm in terms of each measure of superoxide production with an adjustment for log abstinence. MARCH 16, 2012 • VOLUME 287 • NUMBER 12 proton currents stimulated by ionomycin were too small to permit further analysis. The evidence of a functional interaction between H V 1 and NOX5 came rather from analyzing the effect of the H V 1 functional status on NOX5-dependent superoxide production. In particular, we found that H V 1 inhibition by Zn 2ϩ or down-regulation by H V 1-specific siRNA reduced NOX5 superoxide-generating activity. This result suggests that, as observed for NOX2, NOX5 superoxide generation requires H V 1 proton channel activity to compensate for depolarization and intracellular acidification to sustain ROS generation. The similar inhibitory effect of Zn 2ϩ on NOX5 activity and the co-immunoprecipitation of NOX5 and H V 1 in human spermatozoa strongly suggest that a similar functional interaction exists between NOX5 and H V 1 in these cells. The Zn 2ϩ -rich environment in the semen (62, 63) would tend to prevent premature proton flux through sperm H V 1 channels. Also in line with previous studies on human neutrophils, it is plausible that NOX5 activity precedes H V 1 channel opening. However, because both the intracellular spermatozoal and extracellular pH in the vaginal tract are acidic, it is not readily apparent how NOX5L sustains superoxide production (64). Although NOX5 and H V 1 clearly interact, it remains uncertain whether NOX5 activation is required for H V 1 channel opening or might occur subsequent to H V 1 activation. Little is known about the mechanism by which NOX5S is activated, although pathways involving acidic pH and PKA were shown to induce its activity and/or expression in certain cell types (15,65,66). Given the importance of Ca 2ϩ signaling in sperm function and the cross-regulation of Ca 2ϩ entry by PKA and pH (67,68), NOX5S, if expressed, could contribute to ROS-mediated sperm function in the acidic environment of the female genital tract (69 -71).

NOX5 in Human Spermatozoa
The intriguing relationship between H V 1 and NOX5 encouraged us to evaluate the impact of ROS on sperm motility. We found that the induction of superoxide production by H 2 O 2 stimulated the motility of sperm cells. However, this effect was observed only in spermatozoa that were not already hyperactivated. The statistical analysis of sperm characteristics versus superoxide production confirmed our experimental observation, establishing that pH and superoxide production were both important predictors of spermatozoa motility. Our data showing a functional interaction between the H V 1 proton channel and NOX5 NADPH oxidase, and the trends observed in our statistical analysis, suggest strongly that the correlation between superoxide production and motility is pH-dependent. Interestingly, the novel finding that progesterone, a proposed physiological sperm chemoattractant, is a modulator of the pHdependent Ca 2ϩ channel CatSper (72), provides a potential mechanism for linking Ca 2ϩ entry, NOX5 activation, and ROSstimulated hypermotility, thereby reinforcing the concept of a central role of NOX5 in human sperm motility.
This study provides novel and specific evidence that NOX5 is expressed at the protein level and is functionally active in human spermatozoa. Moreover, the regulation of NOX5 by H 2 O 2 is mediated by Ca 2ϩ entry and c-Abl tyrosine kinase activation, whereas the downstream ROS generation is implicated in spermatozoa motility. Taken together, our results suggest that the expression levels and/or activities of NOX5, c-Abl, and H V 1 are significant determinants of the redox status of human spermatozoa and consequently their motility and ovum-fertilizing potential. Moreover, NOX5 and the proteins regulating its activity represent new therapeutic targets for either male infertility (if NOX5 is hyperactive) or male contraception (if NOX5 is physiologically active).