The anticancer agent, 5-aza-2′-deoxycytidine (5-azaCdR, decitabine), causes DNA hypomethylation and a robust, dose-dependent disruption of spermatogenesis. Previously, we have shown that altered testicular histology and reduced sperm production in 5-azaCdR-treated animals is associated with decreased global sperm DNA methylation and an increase in infertility and/or a decreased ability to support preimplantation embryonic development. The goal of this study was to determine potential contributors to 5-azaCdR-mediated infertility including alterations in sperm motility, fertilization ability, early embryo development, and sequence-specific DNA methylation. We find that although 5-azaCdR-treatment adversely affected sperm motility and the survival of sired embryos to the blastocyst stage, the major contributor to infertility was a marked (56–70%) decrease in fertilization ability. Sperm DNA methylation was investigated using Southern blot, restriction landmark genomic scanning, and quantitative analysis of DNA methylation by real-time polymerase chain reaction. Interestingly, hypomethylation was restricted to genomic loci that have been previously determined to acquire methylation during spermatogenesis, demonstrating that 5-azaCdR selectively inhibits de novo methylation activity. Similar to previous studies, we show that mice that are heterozygous for a nonfunctional Dnmt1 gene are partially protected against the deleterious effects of 5-azaCdR; however, methylation levels are not restored in these mice, suggesting that adverse effects are due to another mechanism(s) in addition to DNA hypomethylation. These results demonstrate that clinically relevant doses of 5-azaCdR specifically impair de novo methylation activity in male germ cells; however, genotype-specific differences in drug responses suggest that adverse reproductive outcomes are mainly mediated by the cytotoxic properties of the drug.
DNA methylation is an essential modification of mammalian DNA and occurs at 60 to 80% CpG dinucleotides in the genome. Known to play important roles in several cellular processes, DNA methylation is often associated with transcriptional repression and increased genomic stability (Bird, 2002); in cancer, deregulation of DNA methylation is often observed (Das and Singal, 2004). DNA methylation is catalyzed by a family of DNA methyltransferase enzymes (DNMTs). The principal enzyme, DNMT1, primarily maintains established methylation patterns during DNA replication, whereas patterns are established by the de novo methyltransferases DNMT3a and DNMT3b (Li et al., 1992; Okano et al., 1999).
Currently, 5-azaCdR is used clinically as an anticancer agent for the treatment of myelodysplastic syndromes and other types of cancer because of its ability to demethylate tumor-suppressor genes and cause replication-dependent cytotoxicity. After incorporation into replicating DNA, DNMTs become irreversibly bound to 5-azaCdR as covalent adducts (Gabbara and Bhagwat, 1995). Hypomethylation occurs during subsequent rounds of DNA replication because the depleted cellular pool of DNMTs is insufficient to maintain established genomic methylation patterns. Adducts are cytotoxic and induce apoptosis (Jüttermann et al., 1994) in a p53-dependent manner (Schneider-Stock et al., 2005). It is noteworthy that decreases in methylation can occur at noncytotoxic concentrations of 5-azaCdR that do not significantly impair DNA synthesis (Haaf, 1995). Use of 5-azaCdR inhibits all known DNA methyltransferases (Weisenberger et al., 2004).
Mounting evidence points to an important role for DNA methylation in the process of male germ cell development. Within the male germ line, germ cell-specific methylation patterns are initiated before birth and are completed during spermatogenesis by the pachytene phase of meiosis I (Davis et al., 1999; Oakes et al., 2007a). Methylation patterns in male germ cells are highly distinct from those found in somatic tissues (Eckhardt et al., 2006; Oakes et al., 2007b). Expression of various DNMTs is highly regulated throughout spermatogenesis (La Salle and Trasler, 2006), and inactivation of the DNMTs through gene-targeting results in male infertility (Bourc'his et al., 2001; Kaneda et al., 2004). Because of embryonic lethality or sterility in DNMT-deficient mice, the use of 5-azaCdR becomes a useful alternative to further investigate the role of DNA methylation in germ cells.
Previous studies on the effects of 5-azacytidine in rats (Doerksen and Trasler, 1996; Doerksen et al., 2000) and 5-azaCdR in mice (Kelly et al., 2003) have demonstrated that treatment results in a robust disruption of spermatogenesis. Treatments were of sufficient duration to expose developing germ cells from the spermatogonial stem cell stage through spermatogenesis and epididymal transit. Spermatogenesis is particularly sensitive to the effects of the drug; at these same doses, body weight and hematological parameters were unaffected. Treatment caused a dose-dependent decrease in testis weights, lowered sperm counts, and an increased level of abnormalities in testicular histology. Mating of treated mice with control females resulted in increased preimplantation loss (reduction of the number of implantation sites minus the number of oocytes ovulated); this preimplantation loss could be the result of either a failure of sperm from 5-azaCdR-treated animals to fertilize oocytes or a reduction in the survival of embryos during preimplantation development. Furthermore, global DNA methylation analysis revealed a dose-dependent decrease in DNA methylation in sperm from treated animals. Treatment of mice with constitutively reduced levels of DNMT1 results in a greater level of DNA hypomethylation and less cellular toxicity compared with wild-type mice in somatic tissues (Jüttermann et al., 1994; Laird et al., 1995). We have also previously found that in Dnmt1c/+ mice, animals that are heterozygous for a targeted mutation in the catalytic domain of DNMT1 (Lei et al., 1996), some of the adverse spermatogenic effects of 5-azaCdR are attenuated relative to Dnmt1+/+ males (Kelly et al., 2003).
In this study, we determine the cause of the 5-azaCdR-dependent preimplantation loss by performing a detailed analysis of sperm function and preimplantation development. The relationship of these effects to drug-dependent alterations in sperm DNA methylation is determined using a variety of techniques that assess sequence-specific levels of DNA methylation. Genotype-dependent responses in the extent of the hypomethylation in sperm DNA and the level of adverse spermatogenic effects between Dnmt1+/+ and Dnmt1c/+ animals allude to the contributions of cytotoxic adducts versus abnormal DNA methylation. Thus a further aim of the current study was to examine the effects of 5-aza-CdR in Dnmt1+/+ and Dnmt1c/+ mice.
Materials and Methods
Animals. Male Dnmt1+/+ and Dnmt1c/+ mice, heterozygous for a deletion within the catalytic domain of the primary mammalian DNA methyltransferase, DNMT1 (Lei et al., 1996), were bred and raised in our own facilities (McGill University–Montreal Children's Hospital Research Institute) on a C57BL/6 background. Male mice of both genotypes were obtained through crosses of Dnmt1c/+ males and C57BL/6 females; PCR genotyping of mice was done as described previously (Kelly et al., 2003). Adult virgin C57BL/6 and CD1 females were obtained from Charles River, Canada (St. Constant, QC, Canada). All mice were maintained on a 12:12-h light/dark cycle and were provided with food and water ad libitum. Animal experiments were carried out according to the principles and procedures detailed in the Guide to the Care and Use of Experimental Animals from the Canadian Council on Animal Care.
Treatment.Dnmt1+/+ and Dnmt1c/+ males (aged 7–10 weeks) were randomly assigned to one of two treatment groups (saline: Dnmt1+/+, n = 13; Dnmt1c/+, n = 15; or 5-azaCdR: Dnmt1+/+, n = 14; Dnmt1c/+, n = 15). Males were treated three times a week for 7 weeks, by i.p. injection, with either saline or 0.1 mg/kg 5-azaCdR to expose male germ cells throughout their development. During the treatment, males were weighed twice per week. After 7 weeks of treatment, males were mated with four virgin superovulated CD1 females (aged 8 weeks) and then sacrificed. The testes, epididymides, seminal vesicles, and spleen were removed, weighed, snap-frozen, and stored at –80°C. A section of liver was also removed and frozen. Spermatozoa from the caudal epididymides were isolated and purified as described previously (Alcivar et al., 1989) with modifications (Kelly et al., 2003) and were stored at –80°C.
Mating and Embryo Culture. Adult female CD1 mice aged 8 weeks were superovulated by administration of 5 IU of pregnant mare serum gonadotropin (Sigma-Aldrich, St. Louis, MO) followed by 5 IU of human chorionic gonadotropin (Sigma-Aldrich) 48 h later. To obtain fertilized embryos, each male (n = 7–9/treatment group), after 7 weeks of treatment, was mated, overnight, with 1 superovulated virgin CD1 female per night for 4 nights, for a total of four females per male. Females were examined the next morning for the presence of a vaginal plug. One-cell embryos and unfertilized oocytes were isolated at 27 h after administration of human chorionic gonadotropin, and cumulus cells were removed by hyaluronidase treatment (1 mg/ml) (Sigma-Aldrich) in HEPES-buffered M2 medium (Sigma-Aldrich). Oocytes were examined for the presence of two pronuclei, indicating that fertilization had taken place, and oocytes were classified as fertilized, unfertilized, or fragmented; the majority of fragmented oocytes could not be evaluated as to fertilization status and, thus, were not subcategorized. Fertilized embryos were washed three times, using a mouth-controlled drawn-out glass pipette, and placed into pre-equilibrated bicarbonate-buffered KSOM medium (Erbach et al., 1994) with gentamicin, under oil, and cultured under an atmosphere of 5% O2-5% CO2, in nitrogen at 37°C in a humidified modular incubator (Billups-Rothenberg, Del Mar, CA). Embryos were examined daily on a heated stage and scored for development through to the blastocyst stage. Data are presented on a per male basis; to avoid skewing of data, males were removed from all data sets if less than 10 eggs, in total, were recovered from females mated to that male; only two males were removed, one saline-treated Dnmt1+/+ male and one 5-azaCdR-treated Dnmt1c/+ male.
Sperm Motility Analysis. Sperm motility of treated Dnmt1+/+ and Dnmt1c/+ male mice (n = 6/group) was analyzed using an IVOS semen analyzer (Hamilton-Thorne Research, Beverly, MA) with parameters determined by The Jackson Laboratory (courtesy of Hamilton-Thorne). All dishes and slides were kept at 37°C during all steps. In brief, the cauda epididymidis was tied off, both proximally and distally, removed from the epididymis and rinsed in 3 ml of warmed M199 medium with Hanks' salts (Sigma-Aldrich) supplemented with 0.5% w/v bovine serum albumin, pH 7.4 (Invitrogen, Mississauga, ON, Canada), in a 35-mm Petri dish at 37°C. The cauda epididymidis was then moved to a new Petri dish containing 3 ml of warm supplemented M199 medium (Invitrogen) and minced. Sperm were allowed to disperse for 5 min. The sperm suspension was diluted 1:10 in warm supplemented medium before motility analyses such that concentration did not impair motility. The diluted suspension (20 μl) was loaded into a prewarmed 2X-CEL Sperm Analysis Chamber (80-μm deep) (Hamilton-Thorne Research). Movement characteristics analyzed were percent motility (motile sperm divided by the sum of the motile and immotile sperm within the analysis field), percent progressive motility (progressively motile sperm divided by the sum of motile and immotile sperm within the field), average path velocity (the average velocity of the smoothed cell path), progressive/straight line velocity (the average velocity measured in a straight line from the beginning to end of the track), curvilinear velocity (the sum of the distances moved in each frame along the sampled path divided by the time taken to cover the track), amplitude of lateral head displacement, beat cross-frequency (the frequency with which the sperm track crosses the sperm path, straightness (the departure of the cell path from a straight line, and linearity (the departure of the cell track from a straight line). Tracks were digitally recorded at 60 Hz under 4× dark-field illumination. Analysis was completed using the following IVOS settings: stage temperature, 37°C; frames acquired, 30; frame rate, 60 Hz; minimal contrast, 30; minimal cell size, 4 pixels; magnification, 0.81; cell intensity, 75; static size, 0.13 to 2.43; and static intensity, 0.10 to 1.52. Five slides were analyzed for each mouse, and each slide was sampled five times such that a minimum of 300 sperm were analyzed per slide. The mean of the five slides was calculated for each mouse.
DNA Methylation Analysis. DNA was extracted from purified spermatozoa using proteinase K followed by phenol extraction for Southern blot and restriction landmark genomic scanning (RLGS) analysis as described previously (Okazaki et al., 1995). Southern blots were done as described previously (Trasler et al., 1990) and visualized by autoradiography. Major and minor satellite probes were constructed by PCR amplification of genomic DNA using primers described previously (Lehnertz et al., 2003). The probe for the intracisternal A-particle (IAP) has been used previously (Michaud et al., 1994). RLGS was performed as described previously (Okazaki et al., 1995). RLGS profiles were generated for three to four individual animals in each treatment group. Each profile was visualized by autoradiography for identification of changed spots and PhosphorImager screen for spot densitometry analysis. Visual assessment of changes in spot intensity was confirmed by densitometric analysis by comparing the intensity of the spot of interest with the intensity of 8 to 10 surrounding spots of unchanged intensity. All spots showing differential intensity were observed to be changed in all replicates except one spot in a Dnmt1c/+ saline-treated profile. The genomic location of spots of interest was determined using either the mouse RLGS cloning library method (Yu et al., 2004) or a second-generation virtual RLGS resource (Smiraglia et al., unpublished data). Loci identified using virtual RLGS were confirmed by obtaining the corresponding BAC clone (Roswell Park Microarray Core Facility, Buffalo, NY) and running RLGS mixing gels. The methylation status of paternally methylated imprinted differentially methylated regions (DMRs) and RLGS spots was determined with quantitative analysis of DNA methylation by real-time PCR (qAMP) assay (Oakes et al., 2006). In brief, DNA was digested with various methylation-sensitive restriction enzymes and a methylation-dependent restriction enzyme, McrBC, followed by amplification using real-time PCR. Shifts in Ct value (ΔCt) between the sham- and enzyme-digested samples are used to calculate the percentage of methylation at the various CpG sites within the amplified region (methylation-sensitive restriction enzymes: percent methylation = 100(2–ΔCt) and McrBC percent methylation = 100(1–2–ΔCt). All ΔCt values are the means of triplicate amplifications. Primers were designed to flank CpG/restriction sites of interest and were described previously (Oakes et al., 2007a).
Statistical Analysis. Statistical analysis was done using SigmaStat 2.03 software (SPSS, Chicago, IL). Significant differences (p < 0.05) between treatment groups with respect to the various motility parameters, morphological characteristics, fertilization ability, and percent methylation were detected using two-way analysis of variance, with a post hoc Tukey test. Embryo data are expressed on a per male basis and were evaluated for significance using three-way analysis of variance, with a post hoc Tukey test.
Animal and Tissue Weights. Mice were treated with 5-aza-CdR for 7 weeks, a period sufficient for the exposure of developing germ cells throughout the entire window of spermatogenesis (from spermatogonial stem cell to spermatozoa) and epididymal transit. As observed in a previous study (Kelly et al., 2003), treatment with 5-azaCdR elicited no obvious changes in behavior and weight, although both initial and final body weights of the Dnmt1+/+ and Dnmt1c/+ genotypes were significantly different (p < 0.05) (Supplemental Table 1). Again, testis weights were significantly decreased in treated males, regardless of genotype (p < 0.001), and similar to our previous studies, the extent of reduction was considerably less in Dnmt1c/+ males than in Dnmt1+/+ males (p < 0.05). However, testis weight decline was greater for both genotypes than in our previous study (Kelly et al., 2003).
Sperm Motility and Fertilization Ability. To assess the effects of 5-azaCdR treatment on sperm, multiple motility and morphological characteristics were assayed by computer-assisted sperm analysis using a Hamilton-Thorne IVOS semen analyzer (Fig. 1a; Supplemental Table 2). 5-AzaCdR-treated males produced a greater proportion of immotile sperm, and, of those that were motile, their motility characteristics were generally reduced. Reductions in various motility parameters were more commonly observed in sperm from Dnmt1+/+ males relative to Dnmt1c/+ males. Specifically, curvilinear velocity (a measure of overall movement) and amplitude of lateral head displacement were reduced in both treated groups, whereas several other parameters were significantly reduced in treated Dnmt1+/+ mice but not in Dnmt1c/+ mice.
To determine whether the ability of sperm from 5-azaCdR-treated males to fertilize oocytes is reduced, each male was mated to four superovulated females (total matings = 132), and embryos were collected at the one-cell stage. The presence of a vaginal plug on the morning after mating indicated successful copulation, and the copulation rate was similar for all treatment groups. Only females with vaginal plugs were used for embryo collection. Approximately 500 oocytes on average were scored per treatment group. An oocyte was considered fertilized if two pronuclei were present. Whereas the incidence of fragmented oocytes/embryos was low (<5%) and was similar in all groups, the proportion of fertilized oocytes was dramatically reduced by 70 and 56% (p < 0.001) after matings with treated Dnmt1+/+ and Dnmt1c/+ males, respectively (Fig. 1b). These results show that although a significant proportion of sperm remain motile after treatment, the majority of these sperm are unable to successfully fertilize oocytes.
In Vitro Embryonic Development. To assess the ability of embryos sired by treated males to progress normally through preimplantation development, fertilized oocytes were placed into culture and scored daily for survival to advanced preimplantation stages. Embryo viability was calculated as the percentage of embryos that survived from the previous stage. As Fig. 2 illustrates, there was no change in the progression of embryos throughout preimplantation development, with the exception of survival between the morula and blastocyst stages. At this point, approximately 50% of embryos from saline-treated groups survived; however, the proportion of surviving embryos sired by treated Dnmt1+/+ males was significantly reduced by an additional 25% (p < 0.05). No such decrease was observed for embryos sired by treated Dnmt1c/+ males. These results demonstrate that the ability of the paternal genome from 5-azaCdR-treated Dnmt1+/+ animals to support normal embryonic growth is reduced; in contrast, blastocyst development was similar to that for saline-treated controls for the Dnmt1c/+ mice.
DNA Methylation Analysis of Repetitive Elements. Previous analysis of sperm from 5-azaCdR-treated animals demonstrated a dose-dependent reduction in global levels of DNA methylation in rats (Doerksen et al., 2000) and in mice (Kelly et al., 2003). The previous mouse study revealed that only a higher dose (0.15 mg/kg i.p., three times per week) than the dose used in this study (0.1 mg/kg i.p., three times per week) resulted in a significant reduction in global sperm DNA methylation, despite adverse spermatogenic effects occurring at the lower dose. The global assessment of DNA methylation may not be sensitive enough to detect sequence-specific changes in DNA methylation at the lower dose. To address the possibility that aberrant DNA methylation of sperm from treated males is associated with 5-azaCdR-dependent effects that are observed at the dose used in this study, we investigated the methylation status of three types of repetitive elements in sperm from 5-azaCdR-treated males using Southern blotting. No changes in the methylation status of sperm were detected in the major and minor satellite repeats (structural elements found mainly in centromeric regions), and the interspersed long terminal repeat-containing retroviral element, IAP (Supplemental Fig. 1).
DNA Methylation Analysis of Paternally Methylated Imprinted Regions. Imprinted genes display allele-specific patterns of DNA methylation that are acquired in a sex-specific manner in germ cells; loci displaying this property are termed DMRs. There are three imprinted genes with well characterized DMRs that are methylated in sperm. We investigated whether 5-azaCdR treatment could affect the ability to maintain these patterns. With primers that target restriction sites within the DMRs of H19-Igf2 (Tremblay et al., 1995), Dlk1-Gtl2 (Takada et al., 2002), and Rasgrf1 (Yoon et al., 2002), we used the qAMP assay to determine the percentages of methylation in these regions. No changes were observed for H19-Igf2 or Dlk1-Gtl2. Digestion of DNA with the HhaI restriction enzyme reveals a significant reduction in the percentage of methylation in sperm from 5-azaCdR-treated animals in both Dnmt1+/+ and Dnmt1c/+ groups (Fig. 3). Methylation at these sites is also reduced in the Dnmt1c/+ saline-treated group. Because the HhaI digest of the Rasgrf1 DMR contains the most methylation-sensitive restriction enzyme sites of all the digests tested (Fig. 3a) and only one of the three HhaI sites is required to be unmethylated for the enzyme to cleave the DNA strand, it is the most sensitive in detecting a reduction in methylation. Interestingly, in a previous study using the same approach to investigate the acquisition of DNA methylation at paternally methylated DMRs during spermatogenesis, the HhaI digest of Rasgrf1 was the sole digest to reveal a significant increase in methylation (Oakes et al., 2007a).
Genome-Wide Analysis of Single-Copy Sequences Using RLGS. RLGS is effective in determining genome-wide patterns of DNA methylation by separating genomic DNA digested with the methylation-sensitive restriction enzyme, NotI, followed by two-dimensional gel electrophoresis. Hypomethylated sites generate spots that are visible on RLGS profiles; a change in spot intensity is inversely proportional to the methylation status of an individual locus in the genome. In the mouse, NotI sites occur in a variety of sequence types, including unique and interspersed repetitive sequences. Previously, we found, using RLGS, that the methylation status of 19 spots is modified during spermatogenesis (Oakes et al., 2007a). To further examine the possibility that 5-azaCdR affects the acquisition of patterns of DNA methylation during spermatogenesis, RLGS profiles were generated from sperm from each treatment group (n = 3–4/group). This analysis revealed that a subset of spots was consistently hypomethylated in all profiles from both Dnmt1+/+ and Dnmt1c/+ 5-azaCdR-treated groups (Fig. 4; Table 1). A total of nine spots were observed to change with treatment, and all were hypomethylated. All changed spots in these groups were hypomethylated in all profiles investigated with the exception of one spot (belonging to a NotI site upstream of the AK032343 gene) that was observed to be hypomethylated in one of three of the profiles generated from Dnmt1c/+ saline-treated animals. The vast majority (>99%) of the total number of spots (both hyper- and hypomethylated) remained unaffected. Use of a second-generation virtual RLGS resource to assess the amount of hypermethylated NotI sites on real RLGS profiles (Smiraglia et al., unpublished data) revealed that more than 1000 hypermethylated “spots” are not hypomethylated as a result of 5-azaCdR treatment, including ∼240 and ∼60 spots originating from the IAP and early transposon interspersed repeats, respectively. These repeats are highly visible on virtual RLGS profiles (Oakes et al., 2007b) and were not observed in any profile from any group (data not shown). This result is consistent with the Southern blot analysis of IAP (Supplemental Fig. 1). We also investigated whether changes were specific to sperm or were also hypomethylated in somatic tissues from 5-azaCdR-treated animals. We found that all of the sites observed to be hypomethylated in sperm did not change in brain or liver (Table 1; Supplemental Fig. 2).
Most interestingly, of 11 spots shown in our previous study to gain methylation during the development of male germ cells from spermatogonia to sperm (Oakes et al., 2007a), eight were consistently hypomethylated in sperm from 5-azaCdR-treated animals (Table 1). The eight spots that are normally hypomethylated during spermatogenesis were unaffected by treatment. Only one of nine spots hypomethylated in response to 5-azaCdR was not previously observed to acquire methylation during spermatogenesis. The fact that the methylation status of >99% of all spots was unchanged both during spermatogenesis and in sperm from 5-azaCdR-treated animals demonstrates that maintenance methylation and demethylation processes proceed normally in the presence of the drug. The specificity of 5-azaCdR-dependent hypomethylation indicates that 5-azaCdR selectively inhibits de novo methylation activity in germ cells.
Analysis of 5-azaCdR-Responsive Single-Copy Sequences Using qAMP. The genomic locations of five of nine RLGS spots that are hypomethylated in sperm from 5-azaCdR-treated animals were determined previously (Oakes et al., 2007a). These NotI sites are found near or within genes and within nonrepetitive sequences, but have varied locations within genes and variable CpG island status (Table 2). To further define the specificity and extent of the hypomethylation effect observed in sperm from 5-azaCdR-treated animals and the relationship to the inhibition of de novo methylation, we used the qAMP assay to quantitatively measure the percentage of methylation in sperm from 5-azaCdR-treated mice. The qAMP assay allows for the examination of DNA methylation at multiple neighboring CpGs in addition to the NotI site. We chose to examine the methylation of three loci that were detected to be hypomethylated at their respective NotI sites, Tcf3, Abt1, and Ibtk (Fig. 5a); these three loci were chosen because of the varied location of the NotI site within their respective gene (3′,5′, and body regions, respectively). We found that all restriction enzyme digests detected a significant reduction in the percentage of methylation in response to treatment with 5-azaCdR (n = 5–6 mice/group) (Fig. 5, b–d). This indicates that the altered methylation status at NotI sites observed with RLGS is representative of changes present in neighboring CpG sites. The magnitude of the reduction in methylation is less in the methylation-dependent McrBC digests, because this particular digest is less sensitive to demethylation (Oakes et al., 2006). Slightly less methylation is observed in Dnmt1c/+ animals; however, this reduction is small relative to the effect of 5-azaCdR. In many cases, 5-azaCdR reduced sperm methylation to levels that are similar or close to the levels previously found in type A spermatogonia (horizontal lines in Fig. 5, b–d). Analysis of a fourth locus, a region located upstream of the AK137601 gene that is hypomethylated during spermatogenesis, showed that it is largely unaffected by treatment, demonstrating that hypomethylation proceeds normally in the presence of 5-azaCdR (Fig. 5e). Again, these data, taken together with the RLGS data, indicate that de novo methylation activity is selectively inhibited by 5-azaCdR, whereas maintenance activities and demethylation events that occur during spermatogenesis remain unaffected.
In this study, we determined several novel effects of 5-azaCdR treatment on male reproductive physiology and the epigenetic integrity of male germ cells. Although sperm from 5-azaCdR-treated animals display somewhat decreased motility and decreased preimplantation embryonic development, embryo loss appears to result from a sharply decreased ability of sperm to complete fertilization. Interestingly, a variety of quantitative techniques revealed that hypomethylation is restricted to genomic loci previously shown to acquire methylation during spermatogenesis, thus indicating that de novo methylation activity is selectively inhibited. We also show that despite partial protection from the adverse physiological effects of 5-azaCdR, Dnmt1c/+mice have levels of sperm DNA methylation similar to or lower than levels found in sperm from Dnmt1+/+ mice.
Previous studies have shown that mating 5-azaCdR-treated males with untreated females results in an increase in preimplantation embryo loss (Kelly et al., 2003); this increased loss could be due to a failure of sperm to fertilize or to reduced survival of preimplantation embryos sired by 5-azaCdR-treated fathers. In this study, we observed a greater than 50% reduction in the ability of the sperm of 5-azaCdR-treated males to successfully fertilize oocytes versus that for saline-treated males. The magnitude of the reduction in fertilization ability in the present study was similar to the level of preimplantation loss observed in previous studies. In contrast, the 5-azaCdR-induced reduction in embryo survival from the two-cell stage to the blastocyst stage was minimal. Furthermore, the level of preimplantation loss and the reduction in fertilization ability was similar for both Dnmt1c/+ and Dnmt1+/+ males, whereas decreased preimplantation embryo survival was specific to only Dnmt1+/+ animals. These results suggest that the primary cause of the 5-azaCdR-dependent preimplantation loss noted in our previous study was the inability of sperm to successfully fertilize the oocyte. The slight reductions in sperm number and motility are unlikely to be the primary reason for the failure of the sperm to fertilize, as the reduced parameters are within the range sufficient to maintain normal fertility. The primary cause for the failure to fertilize is most likely an additional parameter of sperm function not addressed in these studies, which may include a failure of the acrosome reaction, capacitation, or sperm-egg recognition.
Treatment with 5-azaCdR is associated with a decreased capacity of the paternal genome to support embryonic growth several days postfertilization. These results support the idea that the sperm chromatin quality is affected as a result of 5-azaCdR treatment. Embryos sired by 5-azaCdR-treated fathers can progress into later stages of development (Doerksen and Trasler, 1996) despite the problems associated with fertilization and embryonic development. Further studies are needed to determine whether 5-azaCdR-dependent alterations in DNA methylation are present in developing embryos and whether offspring experience adverse effects. DNA methylation in the male pronucleus is erased shortly after fertilization (Oswald et al., 2000); however, the full extent of this reprogramming event is not known. Recent studies have shown that treatment of rats with the antiandrogenic fungicide, vinclozolin, results in a transgenerational disease phenotype that is associated with the transmission of epigenetic abnormalities in subsequent generations (Anway and Skinner, 2006; Chang et al., 2006). The fact that acquisition of DNA methylation in male germ cells occurs during adulthood, added to the knowledge that germ cell methylation can be influenced, raises the possibility that exogenous epigenetic insults could potentially be passed on to the progeny.
Investigation of the genome-wide methylation status of individual sequences in sperm DNA from treated animals revealed that male germ cells are able to maintain methylation patterns at repetitive elements and single-copy sequences in the presence of 5-azaCdR. This finding demonstrated that at the doses used here, not enough DNMT1 is covalently trapped to inhibit the maintenance of previously established patterns of methylation. Indeed, high levels of DNMT1 are detected in the testes of adult mice (Singer-Sam et al., 1990). Furthermore, Dnmt1c/+ germ cells are still able to maintain methylation patterns at these sequences in the presence of 5-azaCdR despite having only half of the normal amount of functional enzyme. Previously, we have established that, developing germ cells also possess de novo methylation activities (Oakes et al., 2007a). We found that the ability of germ cells to acquire methylation at CpG sites that normally gain methylation during spermatogenesis is selectively inhibited by 5-azaCdR treatment; almost all CpG sites previously demonstrated to gain methylation during spermatogenesis are hypomethylated in sperm from 5-azaCdR-treated animals.
Two mechanisms have been reported to mediate the effects of 5-azaCdR. First, cytotoxic adducts can induce apoptosis via a p53-dependent mechanism. Secondly, covalent trapping of a sufficient amount of DNMT proteins results in replication-dependent hypomethylation in surviving cells. Figure 6 details how treatment with 5-azaCdR could potentially result in germ cell cytotoxicity and selective inhibition of de novo methylation, whereas in the Dnmt1c/+ background, some adverse effects are attenuated but de novo methylation activity is not restored. In spermatogonia, both maintenance and de novo methylation occur during mitotic divisions (Fig. 6a). In the presence of 5-azaCdR, some cytosine residues are replaced, creating potential sites for the formation of DNA-protein adducts (Fig. 6b). At the dose used in this study, enough adducts are formed to result in the reduction of approximately half of the germ cell population (Kelly et al., 2003), whereas the other half-survive but contain adducts in their DNA causing the observed adverse effects to reproductive physiology. Because of the fact that the level of Dnmt1 expression is 5- to 10-fold higher compared with that of Dnmt3a or Dnmt3b in spermatogonia (Shima et al., 2004) and 5-azaCdR has a more prominent effect on Dnmt3a or Dnmt3b compared with Dnmt1 (Oka et al., 2005), maintenance activity is maintained in the presence of 5-azaCdR, whereas insufficient de novo methylation activity causes a loss of methylation at CpG sites that normally receive methylation during spermatogenesis. In Dnmt1c/+ animals, 50% of DNMT proteins have a mutation in their catalytic domain that prevents the association with incorporated 5-azaCdR (Fig. 6c). Because of the higher levels of Dnmt1 expression in germ cells, DNMT1 proteins are the major contributors to adduct formation. Fewer adducts are formed in the Dnmt1c/+ background, leading to less germ cell death and less adverse effects in surviving cells. Sufficient functional DNMT1 remains to maintain methylation patterns. Because de novo methylases are not mutated, their potential for adduct formation is not changed, causing the same level of sequestration and loss of de novo methylation activity.
To our knowledge, this is the first evidence of selective inhibition of de novo methylation activity by 5-azaCdR. The examination of a developmental process in which known hypermethylation events occur at distinct CpG sites in the genome has allowed for this novel property of the drug to be identified. Extrapolation of the number of CpG sites shown to be gaining methylation during spermatogenesis to the total number of CpG sites located in similar sequences (12 × 106 nonrepetitive, non-CpG island CpGs) (Fazzari and Greally, 2004) suggests the possibility of greater than 1 million CpGs potentially being affected by 5-azaCdR treatment. As 5-azaCdR is used for the treatment of myelodysplastic syndromes, it would be interesting to know whether this property of the drug is restricted to germ cells or whether the drug causes a similar effect elsewhere, namely in developing hematopoietic cells.
Is DNA hypomethylation the cause of the 5-azaCdR-dependent adverse effects on sperm function and embryonic development? Interestingly, some of the adverse effects of 5-azaCdR examined previously, such as decreased testis weight and sperm counts and abnormal seminiferous tubule morphology, as well as effects examined in this study, such as impaired sperm motility and ability to support embryonic development are improved in Dnmt1c/+ animals. Despite the protective effects of the Dnmt1c/+ genotype, losses of DNA methylation are not attenuated in the germ cells of these animals; rather, DNA methylation is equally diminished or slightly lower at some CpG sites examined. This finding suggests that mechanisms, such as DNA-protein adduct formation, are more prominent in the mediation of these effects than DNA hypomethylation. However, some parameters, such as fertilization ability, are equally affected by 5-azaCdR in Dnmt1c/+ and Dnmt1+/+ animals, indicating that hypomethylation may potentially have a role in mediating some of the effects.
We thank En Li for the gift of Dnmt1c/+ mice, Tara Barton and Mary Gregory for assistance with the computer-assisted sperm analysis system, and Keith Latham and Patricia Françon for help with the details of embryo culture. We also thank Stephanie Grénon for superb technical assistance.
This work was supported by grants from the Canadian Institutes of Health Research (CIHR).
C.C.O. and T.L.J.K. are recipients of CIHR Doctoral Awards and Montreal Children's Hospital Research Institute Studentships. B.R. and J.M.T. are James McGill Professors of McGill University.
C.C.O. and T.L.J.K. contributed equally to this work.
J.M.T. is a Scholar of the Fonds de la Recherche en Santé.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
ABBREVIATIONS: DNMT, DNA methyltransferase; 5-azaCdR, 5-aza-2′-deoxycytidine; PCR, polymerase chain reaction; RLGS, restriction landmark genomic scanning; IAP, intracisternal-A particle; DMR, differentially methylated region; qAMP, quantitative analysis of DNA methylation using real-time PCR; McrBc, methylation-sensitive restriction enzyme.
- Received February 20, 2007.
- Accepted June 20, 2007.
- The American Society for Pharmacology and Experimental Therapeutics