It is clear that the cryopreservation and/or thawing process is detrimental to sperm viability. The dead cells with high PI permeability observed may result either from simple necrosis or from necrosis secondary to apoptosis. During conventional freezing, water precipitates as ice and often leads to tissue damage, acts on cytoplasmic structures, or even effects on cytoskeleton or genome-related structures . Ice crystallization induces an unregulated and mechanical death related to necrosis. The present study tested the hypothesis that cryopreservation is associated with the activation of an apoptotic process in spermatozoa.
As the existence and origin of apoptosis in ejaculated mature spermatozoa remains controversial, we first validated each test with the human somatic cell line U937. After VP-16 apoptosis induction, changes were observed in all the tested markers along the cascade of events, including i) cell morphology, ii) decreased m, iii) caspase activation, iv) increased permeability to Yo-Pro-1, v) PS externalization, and vi) DNA fragmentation.
Our results showed that cryopreservation had a dramatic effect on spermatozoon m. The synthesis of ATP is under mitochondrial control and dependent on a high m. A m decrease can result in mitochondrial dysfunction, leading to nonrenewal of ATP. Then a lack of energy, as depletion in ATP, can be responsible for the decreased spermatozoon cell motility usually observed after cryopreservation and thawing . During programmed somatic cell death, the release of apoptotic factors located between the outer and inner membrane is another consequence of the loss of mitochondrial integrity . If male germ cells contained these profactors, a mitochondrial disruption induced by cryopreservation would be responsible for the entry of these apoptotic factors into the cytoplasmic compartment, too.
Annexin V-FITC failed to discriminate apoptotic from nonapoptotic populations objectively in bovine sperm cells. This form of labeling seems to be inappropriate for precise study of the effects of cryopreservation on PS exposure. This experiment was done three times in correlation with experiments giving exploitable results on somatic and human germ cells. Moreover, we found similar cytogram profiles using annexin V-FITC (BD Biosciences) (data not shown). These results differ from Anzar et al. , where they observed more than 31% of the ejaculate Annexin V+/ PI– after cryopreservation. Comparing the cytograms, it can be hypothesized that the green fluorescence amplification they applied for annexin V-FITC analysis was higher than ours. As we determined our amplification using somatic and human sperm controls, higher amplification would increase the problem of false positives. Our observation of a weak annexin V+/PI– population may be supported by the results of Januskauskas et al..
Yo-Pro-1 analysis in sperm cells (which is, to the best of our knowledge, the first in the literature) is inexpensive, quick, and easy to perform and gave interesting information on membrane modifications related to the apoptotic process. This experiment clearly showed an increased permeability of plasma membranes to Yo-Pro-1 in living spermatozoa after cryopreservation. This result is in line with the dramatic effects of cryopreservation usually observed in sperm cell membranes . Furthermore, cryopreservation was shown to induce premature capacitation or acrosomal reaction in sperm cells [31, 32]. Physiologically, these two membrane modifications must occur at the right time and place for successful fertilization . It could be hypothesized that the living and Yo-Pro-1-permeable cell population that we observed corresponded to living and premature reacted cells. A premature acrosomal reaction induced by cryopreservation should be partly responsible for increased cell death. The observed membrane instability could also be a manifestation of capacitation. The effects of cryopreservation on both capacitation and acrosomal reaction, in correlation with apoptotic markers, are currently under investigation. Our preliminary results, using FITC-peanut agglutinin (PNA) (Sigma) and PI, in flow cytometry, confirm that cryopreservation increased the living and acrosomal-reacted cell population (data not shown) . However, it has not been established if cells with a reacted acrosome exhibit an apoptotic-like pattern too and if these two mechanisms might be linked.
Before cryopreservation, about 10% DNA-fragmented cells were observed. Detection of DNA fragmentation using the TUNEL assay is not sufficient to discriminate whether fragmentation is internucleosomal and thus apoptosis specific or unregulated, as described for necrosis . Histone/ protamine transition defect or abortive apoptosis are two different hypotheses that could explain the observed basal DNA fragmentation rate. During spermiogenesis, histones are replaced by protamines. The result of this exchange is high DNA compaction to protect the spermatozoon DNA . Sakkas et al.  showed that incomplete replacement of histones by protamines could be partly responsible for enhanced sperm DNA sensitivity to fragmentation. The fact that the observed DNA fragmentation rate was higher in sperm cells with less condensed nuclei bears out this hypothesis. The DNA-fragmented cell population in the ejaculated sperm might also result from abortive apoptosis of spermatozoa that escaped the elimination mechanism occurring during spermatogenesis . The presence of Fas-labeled spermatozoa in human ejaculates would seem to point to abortive apoptosis. However, in the present study, cryopreservation had no effect on sperm DNA fragmentation, as previously described in human spermatozoa . In addition, DNA fragmentation was unaffected by the postthawing 30 h of incubation, as previously reported in testicular sperm from fertile men .
Anzar et al.  found that cryopreservation induces a decrease of the percentage of cells with a DNA fragmentation. This discrepancy in results may be due to differences in sperm cryopreservation protocols: for example, in our experiment, we used cryopreservation medium without egg yolk.
Our work demonstrates that normal, living ejaculated spermatozoa contain active caspases. Previously, Weng et al.  demonstrated that immature spermatozoa, exhibiting cytoplasmic droplets, contain active caspase-3. Following the hypothesis of Sakkas et al. , it is possible that this weak basal population is the result of abortive apoptosis. However, the significant increase in active caspase rates we observed after cryopreservation should not be related to abortive apoptosis. Our experiment, where we found that 12% of the cryopreserved ejaculate was constituted of living spermatozoa containing active caspases, is not in contradiction with Paasch et al.  or De Vries et al.  results. In the first study, the authors found 20% of cells contained active caspases regardless of whether they were dead or alive. In the second study, following detection of caspase-3 by immunofluorescence, the authors concluded that caspase-3 is only expressed by immature spermatozoa or contaminating cells. The tests we used allowed detection of a wide range of caspases: it is possible that the spermatozoon caspase composition differs from that of other cell types. It would be interesting to use Western blot or enzymatic analysis to specify which caspases are involved under cryopreservation induction .
By immunofluorescence, we localized active caspases and mitochondria exclusively in the midpiece area of the spermatozoa. This provides further evidence that the midpiece area of the spermatozoa could be considered as a region specialized for apoptosis in ejaculated spermatozoa. Nevertheless, the role that caspases play in spermatozoa is not yet fully elucidated. Peter et al.  found that blocking caspase activation with the anticaspase agent zVAD-fmk had no effect on postthaw acrosomal reaction, mobility, or cell viability in canine spermatozoa. These authors hypothesized that either the caspase inhibitor or its concentration were not adapted. While substrates of cleaved caspase are commonly implicated in apoptosis, a recent review showed that activated caspases can be involved in other essential processes, such as the cell cycle or cell differentiation .
To recapitulate, freshly ejaculated bovine sperm exhibits only a few apoptotic characteristics. Abortive apoptosis is one possible explanation for the observed apoptotic base rate. On the contrary, cryopreservation induced the main apoptotic features: i) m decrease, ii) caspase activation, and iii) permeability membrane increase. These three different manifestations of programmed cell death can, respectively, be implicated i) in slowing cell motility , ii) in cell-cycle regulation and cellular differentiation , or iii) in capacitation and acrosomal reactions [31, 32], too. In somatic cells, apoptosis usually leads to final DNA fragmentation, but in our experimental conditions, cryopreservation and thawing had no effect on sperm nuclei, even after 30 h of incubation after thawing. This phenomenon, that we named apoptosis-like, is supposed to involve at least the same mechanisms as those occurring in the execution phase and the early degrading phase of apoptotic somatic cells. For example, noncleavage of the poly(ADP-ribose) polymerase would prevent nuclear manifestations of apoptosis .
We found that, after cryopreservation, a large majority of living sperm cells showed low m; in contrast, few sperm cells presented caspase activation. This result points to the possibility of a caspase-independent apoptosis pathway, where the apoptosis-inducing factor is one of the mitochondrial compounds that can act .
A study of appropriate apoptotic markers in the various steps of the sperm freezing-thawing process would be helpful to specify the critical moment of apoptosis-like induction: liquid nitrogen vapor but also sperm dilution or equilibration steps are potential initiators of the apoptosis-like phenomenon. A better understanding of the cellular mechanisms involved in sperm cryopreservation would help improve the preservation of bovine sperm.
We are grateful to R. Touraine (Laboratoire de Génétique Moléculaire, Saint-Etienne, France) for his support and to N. Laroche (INSERM E0366, Saint-Etienne, France) for his help in the microscopy study.
1 Correspondence. FAX: 33 0 4 77 82 84 61; firstname.lastname@example.org
Received: 14 October 2003.
First decision: 5 November 2003.
Accepted: 15 January 2004.