Reactive oxygen species generation by human spermatozoa: a continuing enigma

The importance of oxidative stress in the aetiology of male infertility has been appreciated since 1943 when the Scottish andrologist John MacLeod demonstrated that the addition of catalase could support the motility of human spermatozoa incubated under aerobic conditions (MacLeod, 1943). He concluded that human spermatozoa must be capable of generating hydrogen peroxide (H2O2) and that the presence of this oxidant impaired sperm function. The first of these assertions has proved much more difficult to demonstrate than the second, as exemplified by an important contribution to this debate published in this issue of the journal by Armstrong et al. (2002). The notion that human spermatozoa are vulnerable to oxidative stress induced by reactive oxygen species (ROS), such as H2O2, has been confirmed in a number of independent studies. For example, it has been demonstrated that the mixture of toxic oxygen intermediates generated by the xanthine oxidase system disrupts the functional competence of human spermatozoa and that this disruptive effect can be completely reversed by the addition of catalase, a highly specific H2O2 scavenger (Aitken et al., 1993; Griveau et al., 1995; Armstrong et al., 1999). This same system has also been used to demonstrate the damaging effect of ROS on nuclear DNA within spermatozoa (Lopes et al., 1998). Furthermore, the direct exposure of human spermatozoa to specific ROS, such as H2O2, disrupts the functional competence of these cells, as well as their genomic integrity (Aitken et al., 1993, 1998a; Oehninger et al., 1995; Kemal et al., 2000). While these observations make it abundantly clear that human spermatozoa are particularly susceptible to oxidative stress (Aitken, 1999), it has not yet been elucidated how such stress is induced. In theory, it might be anticipated that either a depletion of antioxidant protection, making the cells more vulnerable to normal levels of ROS, or exposure to excessive ROS production could be responsible for oxidative stress in the male germ line. Evidence for both of these assertions exist. On the one hand, environmental factors such as heavy smoking have been clearly demonstrated to induce systemic oxidative stress leading to a depletion of antioxidant vitamins from human seminal plasma and the induction of DNA damage in spermatozoa (Fraga et al., 1996). On the other hand, oxidative stress created by the excessive generation of ROS is suggested by the high levels of luminol-dependent chemiluminescence detected in sperm preparations from such patients (Aitken & Clarkson, 1987). These original observations have been confirmed many times using luminol as the probe (Aitken et al., 1989a,b; Pasqualotto et al., 2000; Kobayashi et al., 2001; Gil-Guzman et al., 2001; Ollero et al., 2001). Moreover, the subpopulation of human spermatozoa generating high levels of chemiluminescence in infertile patients can be isolated in the low density region of discontinuous Percoll gradients (Aitken & Clarkson, 1988; Aitken et al., 1989a; Gil-Guzman et al., 2001; Ollero et al., 2001). The low density of these redox active spermatozoa appears to be because of the presence of excess residual cytoplasm, one consequence of which may be to enhance the supply of substrate to the ROS generating systems in these cells through the generation of NAD(P)H (Gomez et al., 1996; Ollero et al., 2001). One of the major problems with detecting ROS in populations of human spermatozoa is that these preparations are invariably contaminated with leukocytes (Aitken & West, 1990). Phagocytic leukocytes are professional generators of ROS and give chemiluminescent signals that are significantly greater than spermatozoa on a cell-for-cell basis. This point is clearly emphasized by Armstrong et al. (2002) who were able to confirm ROS generation by leukocytes in response to phorbol ester stimulation (EPR superoxide signal, ATP depletion, chemiluminescence and nitroblue tetrazolium reduction) but could not find any such signals in populations of human spermatozoa isolated from the high-density region of discontinuous Percoll gradients. As a consequence of this intrinsic difference in ROS generating activity, it only takes a low level of leukocyte contamination to overwhelm and obscure any chemiluminescence produced by the spermatozoa (Aitken & West, 1990). In response to this problem, sensitive biochemical techniques involving the use of an N-formyl chemotactic peptide have been introduced for detecting leukocytes in populations of human spermatozoa, and protocols have been developed for the physical removal of these cellular contaminants (Krausz et al., 1992; Aitken et al., 1996). When these techniques are used to generate human sperm populations that are free of detectable leukocyte contamination, chemiluminescent signals of varying intensity can still be detected, which must presumably emanate from the spermatozoa themselves (Aitken & West, 1990; Aitken et al., 1992b, 1994). Moreover, 50% of sperm populations treated to remove all traces of leukocyte contamination will still generate a detectable lucigenin-dependent chemiluminescent signal on addition of phorbol ester (Gomez et al., 1998). Furthermore, these phorbol ester-induced responses show powerful negative correlations with the quality of the original semen sample (Gomez et al., 1998). Thus, as Armstrong et al. (2002) point out, their inability to detect chemiluminescence signals does not necessarily indicate that spermatozoa are incapable of generating ROS. The lack of detectable activity may simply be a function of the fertility of the donors combined with the insensitivity of the techniques used to detect ROS. If poorer quality spermatozoa had been used, intense chemiluminescence signals would have been anticipated. The linkage between defective sperm function and high levels of luminol- or lucigenin-dependent chemiluminescence on the part of the spermatozoa has been described in so many independent studies that it is hard to ignore (Aitken & Clarkson, 1987; Aitken et al., 1989b, 1992b; D'Agata et al., 1990; Zalata et al., 1995; Henkel et al., 1997; Gil-Guzman et al., 2001; Ollero et al., 2001). There even seems to be consensus concerning the mechanism by which ROS generation is elevated in such defective sperm populations. The key to this activity appears to be the retention of excess residual cytoplasm by immature spermatozoa, which are released prematurely from the germinal epithelium (Gomez et al., 1996; Zini et al., 2000; Gil-Guzman et al., 2001; Ollero et al., 2001). The presence of this excess cytoplasm is hypothesized to increase the capacity of these immature cells to generate the NAD(P)H that serves as a source of electrons for ROS production. An increase in the availability of membrane bound enzymes to generate ROS and, possibly, a failure to elaborate intrinsic inhibitors of plasma membrane redox activity (Vernet et al., 2001), may also contribute to the oxidative stress experienced by these immature human sperm subpopulations. Chemiluminescence data are usually interpreted in terms of the excessive generation of reactive oxygen intermediates such as H2O2 or superoxide anion (O2–). The papers by Armstrong et al. (2002) and Richer & Ford (2001) cast doubt on this interpretation, because neither group could detect a superoxide signal by EPR. The latter is able to detect superoxide generation with high specificity but low sensitivity. As emphasized by Armstrong et al. (2002), the results emphasize the relative paucity of O2– generation by spermatozoa in comparison with leukocytes; however, they do not indicate that spermatozoa are absolutely incapable of generating this, or any other species, of reactive oxygen intermediate. Interestingly, Zhang & Zheng (1996) were able to detect O2– using EPR, but only in populations of capacitated spermatozoa. Kumar et al. (1991) also used EPR techniques to detect O2– generation in populations of maturing epididymal spermatozoa. Peroxidase-mediated reactions have also provided evidence for H2O2 production by spermatozoa from various species (Aitken et al., 1992a; Ball et al., 2001). Moreover, both superoxide dismutase and catalase have been shown to suppress various aspects of sperm function, emphasizing the importance of low-level ROS generation to the fundamental physiology of these cells (Bize et al., 1991; De Lamirande & Gagnon, 1993; Griveau et al., 1994, 1995; Aitken et al., 1998b). At the very least, the data generated to date reveal that spermatozoa are redox active and that this activity is both pathologically and physiologically significant. Redox active enzymes have been shown to reside in the sperm plasma membrane and can use both NADH and NADPH as electron donors (Vernet et al., 2001). The constituents of the leukocyte NADPH oxidase (gp91phox) are definitely not present in spermatozoa according to polymerase chain reaction analysis. Vernet et al. (2001) and Armstrong et al. (2002) have confirmed Richer & Ford's (2001) original observation that a cytochrome b558 spectrum cannot be obtained with human spermatozoa. While Richer & Ford (2001) interpreted these data as indicating that spermatozoa do not contain an NADPH oxidase, their conclusion appears to be incorrect, given the recent discovery of an NADPH oxidase (NOX 5) in the male germ line by Banfi et al. (2001). This oxidase is sensitive to calcium and presumably accounts for the ability of A23187 to excite luminol-dependent chemiluminescence by human spermatozoa (Aitken & Clarkson, 1987). Armstrong et al. (2002) indicate that, unlike gp91phox, NOX 5 cannot be controlled by protein kinase C (PKC) because phorbol esters do not elicit a ROS response from human spermatozoa. Whether NOX 5 is the oxidase controlling capacitation, whether it is over-expressed in the spermatozoa of patients exhibiting infertility associated with oxidative stress, whether it is responsible for DNA damage in the germ line, and whether it generates ROS or orchestrates electron transfer to redox centres in the plasma membrane, are all important questions awaiting resolution. Investigation of this area will not be easy because as Armstrong et al. (2002) emphasize, the level of ROS generation observed in normal human spermatozoa is extremely low. Nevertheless, because the redox system(s) present in human spermatozoa appear to be central to the functional competence and pathological impairment of these cells, further resolution of their biochemical features will be awaited with interest.