If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Genetics of Male Fertility Group, Unitat de Biologia Cel⋅lular (Facultat de Biociències), Universitat Autònoma de Barcelona, Bellaterra (Cerdanyola del Vallès) 08193, Spain
Genetics of Male Fertility Group, Unitat de Biologia Cel⋅lular (Facultat de Biociències), Universitat Autònoma de Barcelona, Bellaterra (Cerdanyola del Vallès) 08193, Spain
Genetics of Male Fertility Group, Unitat de Biologia Cel⋅lular (Facultat de Biociències), Universitat Autònoma de Barcelona, Bellaterra (Cerdanyola del Vallès) 08193, Spain
Departament d'Anatomia Patològica, Farmacologia i Microbiologia, Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Universitat de Barcelona, Barcelona 08036, Spain
Departament d'Anatomia Patològica, Farmacologia i Microbiologia, Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Universitat de Barcelona, Barcelona 08036, Spain
Genetics of Male Fertility Group, Unitat de Biologia Cel⋅lular (Facultat de Biociències), Universitat Autònoma de Barcelona, Bellaterra (Cerdanyola del Vallès) 08193, Spain
The influence of aberrant sperm DNA methylation on the reproductive capacity of couples has been postulated as a cause of infertility. This study compared the DNA methylation of spermatozoa of 19 fertile donors and 42 infertile patients using the Illumina 450K array. Clustering analysis of methylation data arranged fertile and infertile patients into two groups. Bivariate clustering analysis identified a differential distribution of samples according to the characteristics of seminogram and age, suggesting a possible link between these parameters and specific methylation profiles. The study identified 696 differentially methylated cytosine-guanine dinucleotides (CpG) associated with 501 genes between fertile donors and infertile patients. Ontological enrichment analysis revealed 13 processes related to spermatogenesis. Data filtering identified a set of 17 differentially methylated genes, some of which had functions relating to spermatogenesis. A significant association was identified between RPS6KA2 hypermethylation and advanced age (P = 0.016); APCS hypermethylation and oligozoospermia (P = 0.041); JAM3/NCAPD3 hypermethylation and numerical chromosome sperm anomalies (P = 0.048); and ANK2 hypermethylation and lower pregnancy rate (P = 0.040). This description of a set of differentially methylated genes provides a framework for further investigation into the influence of such variation in male fertility in larger patient cohorts.
Infertility is defined by the failure to achieve a clinical pregnancy after 12 months or more of regular unprotected sexual intercourse. Male factor infertility affects around 50% of couples and, in 40–60% of cases, the origin is unknown (
). The fact that male infertility is a complex, multifactorial and polygenic disease makes it difficult to identify the causes. It is important to consider that parameters such as sperm concentration, motility and morphology, and the physiological processes carried out by spermatozoa depend, directly or indirectly, on their genetic and epigenetic load.
The influence of sperm DNA methylation on the reproductive capacity of couples has been postulated as one of many possible explanations for infertility (
). The pattern of methylation in mature spermatozoa reflects changes in the pattern of gene expression that occurs during spermatogenesis. DNA methylation controls the transcriptional activity of genes and is involved in establishing higher order chromatin structure. A normal pattern of methylation in germ cells contributes to the progression of meiosis, culminating in the production of functional spermatozoa. Thus, abnormalities in DNA methylation may affect sperm production and could explain some cases of male infertility. Indeed, it is well documented that genomic DNA hypomethylation in germ cells is associated with increased numbers of aneuploid spermatozoa (
). Moreover, several studies have suggested a relationship between aberrant sperm DNA methylation and male infertility. In particular, there have been reports of alterations of imprinted genes (
Semen samples showing an increased rate of spermatozoa with imprinting errors have a negligible effect in the outcome of assisted reproduction techniques.
Abnormal methylation of the promoter of CREM is broadly associated with male factor infertility and poor sperm quality but is improved in sperm selected by density gradient centrifugation.
Nevertheless, most of these studies used single-locus analysis methods on a limited number of loci. This has allowed the identification of a small number of epigenetic causes involving small infertile subpopulations with uncertain implications for reproductive treatments. Large-scale analysis strategies in the field of biomedicine are applicable for studying infertility. Their use, together with gene ontology tools, offers the opportunity to understand the basic mechanisms involved in male infertility. Using the HumanMethylation27 Infinium BeadChip platform (Illumina Inc., San Diego, CA, USA),
identified alterations of HDAC1, SIRT3 and DNMT3A in patients with low sperm motility. Using the same methodology, a link between aberrant sperm DNA methylation and abnormal sperm chromatin packaging has been described (
Genome-wide sperm deoxyribonucleic acid methylation is altered in some men with abnormal chromatin packaging or poor in vitro fertilization embryogenesis.
Broad DNA methylation changes of spermatogenesis, inflammation and immune response-related genes in a subgroup of sperm samples for assisted reproduction.
). Compared with the 27K array, the HumanMethylation450 Infinium BeadChip platform (450K) (Illumina Inc., San Diego, CA, USA) is substantially improved in terms of coverage and reliability. The 450K array allows the analysis of more than 485,000 cytosine-guanine dinucleotides (CpG) per sample and covers 99% of the genes described in databases. Using this approach, it has been demonstrated that human spermatozoa have a highly homogeneous methylation profile (
) has identified a set of differentially methylated sites, suggesting that some alterations would explain specific infertile phenotypes.
This study characterizes the sperm methylomes of fertile and infertile individuals using the 450K methylation array. The analysis of the results has been designed to identify genes associated with differentially methylated CpG and their possible enrichment for certain biological processes. Moreover, this study has investigated whether there is a relationship between methylation and seminal characteristics, age, chromosomal stability and assisted reproductive technology outcomes. The primary endpoint of the study is to determine the impact of aberrant sperm DNA methylation on male fertility.
Materials and methods
Study population, sample collection and assisted reproductive technology outcome
Nineteen semen samples from sperm donors (free of exposure to any genotoxic agent, and no history of chemotherapy, radiotherapy or chronic illness) were selected as a control population (coded with ‘C’-sample numbers). All controls met the following characteristics: (i) normal karyotype; (ii) proven fertility (donated spermatozoa resulted in the birth of six children in each case); (iii) total number of spermatozoa per ejaculate with progressive motility (grades a + b) above 90 × 106; (iv) more than 4% of spermatozoa being of normal form; and (v) more than 10 × 106 spermatozoa/ml with progressive motility after thawing cryosurvival test.
The average age of the control individuals was 25.95 ± 4.80 years (range: 19–36).
Forty-two semen samples from individuals submitted for IVF by intracytoplasmic sperm injection (IVF/ICSI) cycles were also analysed (coded with ‘I’-sample numbers). Conventional semen evaluation was carried out following the criteria of the World Health Organization (
) allowing the classification of the patients into the following categories: nine were normozoospermic (N), nine oligozoospermic (O), nine asthenozoospermic (A), 12 teratozoospermic (T) and three asthenoteratozoospermic (AT) (Supplementary Table S1). In the case of normozoospermic samples, patients were selected from infertile couples in which any female contribution to infertility was discarded. Samples were processed for IVF/ICSI by the standard protocols of the collaborative centres. One quarter of the remaining sample was used for this study.
The average age of the infertile individuals was 38.36 ± 5.31 years (range: 30–50). For clustering purposes, the ages of control donors and infertile patients were recoded as R1 (19–29 years), R2 (30–39 years) and R3 (40–50 years) (Supplementary Table S1). Patient data also included the result of sperm aneuploidy/diploidy screening through fluorescence in-situ hybridization (FISH) for chromosomes X, Y, 13, 18 and 21. They were classified as ‘normal’ or ‘altered’ according to the comparison of each patient with the control data (obtained through the analysis of the donor population). An altered FISH result was observed in 23 of the 42 patients (Supplementary Table S1).
In 39 of the 42 infertile individuals, couples were submitted for IVF/ICSI cycles. Two couples underwent conjugal intrauterine insemination (I-785z; I-787z) and two couples left the reproductive programme (I-731z; I-791z). The following information was compiled: fertilization rate (zygotes/mature oocytes), rate of discarded embryos, pregnancy rate per transfer and miscarriage rate per clinical pregnancy (Supplementary Table S1).
Written informed consent was obtained from all patients. The study was approved by the Ethics Commission on Human and Animal Experimentation of the Universitat Autònoma de Barcelona on 15 December 2015 (Reference 1653).
Sperm cell selection
To eliminate somatic cells present in the ejaculate, semen samples were processed according to the Somatic Cell Lysis (SCL) method (
). Briefly, cells were incubated on ice for 30 min in Somatic Cell Lysis Buffer (0.1% SDS and 0.5% Triton X-100 in Milli-Q® water). Optical microscopic examination was used to verify the somatic cell elimination. SCL treatment was repeated until less than one somatic cell per 10,000 spermatozoa was observed.
Sperm DNA isolation
Isolation of spermatozoa genomic DNA was performed using the commercial extraction kit Puregene (Gentra Systems, USA). Samples were centrifuged at 16,000g for 20 s, the supernatant was discarded and the pellet was incubated with the supplied lysis buffer solution supplemented with 1 mol/l 1,4-Dithiothreitol and 10 mg/ml–1 Proteinase K. Cell lysis was performed overnight at 55°C with constant vortexing. The protein fraction was extracted by centrifugation at 16,000g for 3 min after a treatment of 5 min using the kit-supplied solution. The supernatant was carefully transferred into a fresh microcentrifuge tube containing 300 µl of 100% isopropanol (2-propanol). The sample was centrifuged for 1 min at 16,000g, the supernatant discarded and the sample dried at room temperature. Finally, the sample was washed with fresh 70% ethanol, and the pellet recovered and stored at −20°C in hydration solution.
Array-based DNA methylation analysis
DNA was quantified using PicoGreen® technology using the Quant-iT PicoGreen® dsDNA Reagent (Life Technologies, USA). Bisulfite conversion was performed on the total isolated DNA using the EZ DNA Methylation-Direct™ Kit (Zymo Research, USA) according to the manufacturer's recommendation for the Illumina Infinium Assay (Illumina Inc., San Diego, CA, USA). Effective bisulfite conversion was confirmed with four controls that were converted simultaneously with the samples.
Bisulfite converted sperm DNA was hybridized onto the 450K methylation array following the Illumina Infinium HD Methylation workflow. This array measures the methylation status of 485,764 sites. Of these sites, 482,421 cytosines (99.3%) were found as part of CpG dinucleotides, and 3343 sites (0.7%) corresponded to cytosine-any base-guanine targets (CNG). Analysis was performed using the Illumina iScan SQ fluorescent scanner. Raw fluorescence intensities were extracted using GenomeStudio Methylation module software (Illumina Inc., San Diego, CA, USA), and analysed using a Minfi package available through Bioconductor (Minfi: Analyse Illumina's 450k methylation arrays. R package version 3.0.1). To exclude technical biases, an optimized pipeline with several filters developed at the IDIBAPS was used. From the initial dataset of 485,512 sites (excluding probes detecting single nucleotide polymorphisms [SNP]), those with poor detection P-values (P > 0.01) were removed. The remaining sites (n = 485,065) were used for downstream analyses. To check for batch effects the wrapper function ‘qcReportfunction’ was used to produce a PDF QC report. Single cytosine methylation values (β-values) in each sample were calculated as the ratio of the methylated signal intensity to the sum of methylated and unmethylated signals. β-values range from 0, for completely unmethylated, to 1, for completely methylated cytosines. All sites measured in the array were annotated according to the information provided by Illumina.
Statistical analyses
Control data homogeneity
To validate the reliability of the methylation results in controls, sample-to-sample correlations were performed using the Spearman rank correlation test. CpG β-value homogeneity was also checked subjecting control data to hierarchical clustering using the Euclidean distance and Ward's method. Moreover, in order to rule out the influence of age over the methylation results, the study compared the mean β-value of each CpG, grouping control samples according to age (R1 and R2). Analyses were performed using a Wilcoxon rank sum test considering CpG sites as differentially methylated if the Wilcoxon P-value and false discovery rate P-value were lower than 0.05.
Unsupervised cluster analysis
To identify inter-individual differences at the CpG β-value level, data from control and infertile individuals were subjected again to hierarchical clustering using the Euclidean distance and Ward's method. Hierarchical CpG clustering was performed for the whole population of CpG, and for the population of CpG showing a coefficient of variation greater than 0.35 (Figure 1).
Figure 1Flow chart summarizing the statistical analyses performed.
Identification of differentially methylated CpG and loci
To identify differentially methylated CpG (DM-CpG), the means of the control and infertile population β-value of each CpG were compared using a Wilcoxon rank sum test (Figure 1). CpG sites were considered differentially methylated if the Wilcoxon P-value and false discovery rate P-value were lower than 0.05. Chi-squared tests were used to assess the enrichment of hypomethylated and hypermethylated DM-CpG in functionally annotated regions, using the annotation of the 450K as background (Figure 1).
Only those loci presenting at least three consecutive DM-CpG were considered as having a possible biological significance and were categorized as differentially methylated loci (DM-loci).
Cluster analysis of differentially methylated loci
To further analyse inter-individual differences, the study assessed whether the patients were grouped according to the DM-loci. This was done using a two-step procedure: firstly, determining the β-value range of each DM-CpG that belonged to DM-loci in the control population (excluding outliers). This range was used to determine whether the β-values of these DM-CpG in the infertile patients were within or outside normal values. A patient was classified as a carrier of a DM-locus when they showed at least three consecutive DM-CpG outside the normal range (Figure 1). Secondly, hierarchical clustering was performed using the list of DM-loci for each patient (Figure 1). For this purpose, loci were recoded as: Hypermethylated loci = 1; Hypomethylated loci = 2; Normal methylated loci = 0.
Bivariate locus analysis
The study analysed whether the condition of normal/altered loci was significantly associated with age (three categories: R1, R2 or R3), seminogram (four categories: N, A, T and O) and the results of aneuploidy/diploidy sperm screening (two categories: normal or altered) using chi-squared tests (Figure 1). In the case of the variables associated with the IVF/ICSI outcome (rate, percentage of discarded embryos, pregnancy rate and miscarriage rate), data were analysed using the Kruskal-Wallis/Mann-Whitney test (Figure 1). Differences were considered to be significantly different when P < 0.05.
Clustering description
Bivariate analyses were applied, under the conditions specified in the preceding paragraph, to evaluate the influence of age, seminogram, sperm-FISH result, and assisted reproductive technology outcome in the clustering analyses (Figure 1).
Gene ontology analysis
The enrichment of biological processes and molecular functions were evaluated using DAVID Bioinformatics Resources v.6.7 (Database for Annotation, Visualization and Integrated Discovery, http://david.abcc.ncifcrf.gov) (
) (Figure 1). Analyses were performed for the following groups: (i) the population of loci with CpG with a coefficient of variation greater than 0.35; (ii) the list of loci with DM-CpG; and (iii) the DM-loci list. P-values were considered significant if P < 0.05 after Bonferroni correction.
Results
Control data homogeneity
Sample-to-sample correlation coefficients among controls were very high; rho numbers ranged from 0.919 to 0.972. Hierarchical clustering grouped all individuals into one cluster except C-758Z. Results revealed the absence of differentially methylated cytosines in the comparisons of the mean β-value of each CpG, grouping control samples according to age (R1 and R2).
Overall, these results demonstrated the homogeneity of the methylation results in controls, and its reliability as a reference value for downstream analysis.
Unsupervised cluster analysis
A global hierarchical cluster analysis was unable to split the control and infertile samples into two distinct groups. However, analysis of the 16,955 CpG with a coefficient of variation greater than 0.35 revealed the presence of two main clusters (Figure 2). Cluster 1 included one control donor and fifteen infertile patients. Seminogram characteristics were distributed as following: 3N, 3A, 6O and 4T (Supplementary Table S1). The mean age in cluster 1 was 40.2 (SD = 5.5; range 31–50). Cluster 2 included 18 control individuals and 27 infertile patients. Seminogram was distributed in 25N, 6A, 3AT, 3O and 8T (Supplementary Table S1). The mean age in cluster 2 was 32.5 (SD = 7.4; range 19–48).
Figure 2Heatmap displaying the methylation status of the 16,955 CpG with a coefficient of variation greater than 0.35. The horizontal bar above the heatmap shows the two main clusters: cluster 1 in red and cluster 2 in blue. Each column of the heatmap represents a single sample indicated at the bottom. Each cell of each line of the heatmap represents the methylation level for each CpG. Methylation levels are indicated in the colour scale on the right side of the figure as dark blue (0.0) to dark red (1.0).
Statistical analysis revealed that the average methylation value of the 16,955 CpG from the individuals in cluster 1 was significantly higher than the average in cluster 2 (0.342 versus 0.287; P < 0.001). Bivariate analysis identified a significant association between clustering and the characteristics, age (P = 0.002) and seminogram (P = 0.013). The other characteristics analysed (sperm FISH, fertilization rate, percentage of discarded embryos, pregnancy rate and miscarriage rate) did not present a differential distribution between the two clusters.
Differentially methylated CpG
This study identified a total of 696 DM-CpG between control and infertile patients. Of these, 184 (26%) were hypomethylated and 512 (74%) were hypermethylated (Supplementary Table S2). From the 512 hypermethylated DM-CpG, 341 (67%) corresponded to CpG associated with genes, while 171 (33%) were in intergenic regions (Figure 3). Assessment of enriched hypomethylated and hypermethylated Cp in functionally annotated regions using the 450K annotation as background showed that the hypermethylated CpG were enriched in ‘Outside CGI’ regions (P < 0.0001); particularly distributed in intergenic regions (including 3-prime UTR) (P < 0.05) and gene bodies (exon, intron) (P < 0.05); and belonged to heterochromatin regions (P < 0.0001). Of the 184 hypomethylated DM-CpG, 166 (90%) were associated with genes, while 18 (10%) were intergenic (Figure 3). Hypomethylated CpG were primarily annotated to CpG shore (P < 0.0001) and shelf regions (P < 0.05), and were also enriched in insulators (P < 0.05) and transcribed regions (P < 0.05).
Figure 3DM-CpG methylation profiles according to chromosome state, gene annotation and CpG class. TSS = transcription start site, Txn = related to transcriptional processes.
Seventeen loci were categorized as differentially methylated (DM-loci) according to the criteria described in the ‘Materials and methods’ section (15 showed hypermethylation and two showed hypomethylation). Overall, these loci contained 76 DM-CpG (Supplementary Table S3). Specifically, the 15 hypermethylated DM-loci were ATXN7L1, APCS, PATE4, PRDM1, ANK2, RPS6KA2, RHOBTB1, C6orf118, ANKRD53, LOC100271702/LINC00940, EIF2AK3, JAM3, NCAPD3, TEX 261 and CACNA2D4. The two hypomethylated DM-loci correspond to FOXK1 and FOXK2 (Figure 4; Supplementary Table S3).
Figure 4Heatmap displaying the methylation status of the 76 DM-CpG within the 17 DM-loci. The horizontal bar above the heatmap indicates the control (light grey) and infertile (dark grey) population. Each column of the heatmap represents a single sample indicated at the bottom. Each cell of each line of the heatmap represents the methylation level for each DM-CpG and gene (indicated on the right side of the heatmap). Methylation levels are indicated in the colour scale on the left side of the figure as dark blue (0.0) to dark red (1.0).
Cluster analysis of differentially methylated loci
The β-value range of the 76 DM-CpG belonging to loci in the control population allowed the identification of patients with altered gene profiles. Thirty-four infertile patients had at least one DM-locus (Supplementary Table S4). Although DM-CpG located downstream of the EIF2AK3 gene coding region showed alteration at the population level, no patient presented values outside the range established in the controls (Supplementary Table S4). Clustering analysis classified the infertile patients into three main categories: cluster 1 was constituted by eight infertile patients without affected loci, cluster was 2 formed by eight patients showing a mean of 2.75 ± 2.38 altered genes (range: 1–8) and cluster 3 grouped 26 patients with a mean of 2.23 ± 1.2 altered genes (range: 1–5) (Supplementary Table S4). Statistical analysis revealed no significant differences among the clusters for any of the variables analysed (age, seminogram, sperm-FISH, fertilization rate, percentage of discarded embryos, pregnancy rate or miscarriage rate).
Bivariate locus analysis
This study identified a significant association between hypermethylation of the three consecutive DM-CpG within intron 2 of RPS6KA2 and advanced age (P = 0.016); all four patients with altered methylation in this intron belonged to the R3 group. An association was found between oligozoospermia and hypermethylation of the five consecutive DM-CpG mapped in the upstream, promoter and intron 1 regions of APCS (P = 0.041) and an association between altered FISH pattern and hypermethylation of the 13 DM-CpG in the 3'UTR of JAM/NCAPD3 (P = 0.048). Specifically, six of the 15 patients with APCS alterations (40%) were oligozoospermics, while 11 of 14 patients with altered methylation of JAM/NCAPD3 (78.6%) also showed an altered FISH result. The mean pregnancy rate was significantly lower (P = 0.040) in the seven patients with hypermethylation at the four consecutive DM-CpG identified in intron 21 of ANK2 (13.9%) compared with the mean in patients without alterations (42.6%).
Gene ontology analysis
The 16,955 CpG with a coefficient of variation greater than 0.35 were annotated to 6260 genes. The use of DAVID identified the presence of seven enriched biological processes related to cell adhesion (GO:0022610; GO:0007155; GO:0007156; GO:0016337; GO:0051056) and cell morphogenesis (GO:0000904; GO:0000902).
The 696 DM-CpG were annotated to 501 genes, 161 containing hypomethylated and 340 hypermethylated CpG (Supplementary Table S2). No significantly enriched biological process was found. However, prior to multiple comparison correction, among the 48 enriched processes, 13 CpG associated with genes potentially related to spermatogenesis were identified. In particular, hypermethylated genes were associated with cell morphogenesis (GO:0000902) whereas hypomethylated genes were related to (in alphabetical order): cell cycle (GO:0007049), cell cycle phase (GO:0022403), cell cycle process (GO:0022402), chromosome organization (GO:0051276), cytokinesis (GO:0000910), DNA repair (GO:0006281), double-strand break (DSB) repair via homologous recombination (HR) (GO:0000724), meiosis (GO:0007126), meiotic cell cycle (GO:0051321), M-Phase of meiotic cell cycle (GO:0051327), recombinational repair (GO:0000725) and response to DNA damage stimulus (GO:0006974).
Gene ontology analysis was also performed for the DM-loci list and no GO term was significantly enriched.
Discussion
This study has identified DNA methylation differences between the spermatozoa of control donors and infertile patients using the Infinium HumanMethylation 450 BeadChip. The ontological analysis of genes associated with the most variable CpG identified an enrichment of biological functions related to cell adhesion and cell morphogenesis; functions which are indispensable for ensuring spermatogenesis progression. Moreover, clustering of the most variable CpG suggests that age and seminogram are related to broad methylation differences.
Regarding age-effect, clustering analysis of the most variable CpG identified an association between age and hypermethylation. Although we are aware that mean age differences between donors and patients hamper the interpretation of the results, it should be noted that other authors have also described altered methylation patterns in spermatozoa from advanced age individuals (
) is well documented. Bivariate gene analysis also supports the association between advanced age and RPS6KA2 hypermethylation. This gene encodes a member of the ribosomal S6 kinase (RSK) family of serine/threonine kinases. The activity of this protein has been implicated in controlling cell growth and differentiation. Although the protein has been identified in mature human spermatozoa (
), its specific role and association with age and male infertility is still unknown. Intriguingly, the association with age has been also described in cumulus cells from human follicles (
Three interconnected results support the relationship between differential sperm DNA methylation and seminal parameters. First, cluster analysis of the most variable CpG revealed an enrichment of oligozoospermic samples in cluster 1 (six of the nine O samples were grouped in cluster 1) and an enrichment of normozoospermic samples in cluster 2 (25 of the 28 N samples were grouped in cluster 2). Since the average methylation value of cluster 1 was higher than that of cluster 2, the data suggests an association between hypermethylation and low sperm count. Second, six of the 17 DM-loci are involved in processes related to spermatogenesis or sperm function (Table 1). Third, bivariate gene analysis significantly associated hypermethylation of APCS with oligozoospermia. The protein encoded by this gene is a glycoprotein, belonging to the pentraxin family of proteins. It is thought to act as a chaperone controlling the degradation of chromatin during apoptosis. Although APCS has been identified in mature human spermatozoa (
) and it is described as an epididymal secretory sperm binding protein, its specific role and its association with male infertility remains elusive.
Table 1Functional annotation associated with the differentially methylated genes directly involved in spermatogenesis-related processes or sperm function.
Tex261, a novel gene presumably related but distinct from steroidogenic acute regulatory (StAR) gene, is regulated during the development of germ cells.
Interestingly, this study failed to find an association between male infertility and the methylation status of repetitiveelements such as ALU and LINE-1. Hypomethylation of repetitive sequences has been attributed to retrotransposon reactivation and infertility, according to the host defence hypothesis. To investigate further, the presence of repetitive sequences in the hypomethylated DM-CpG was analysed. This allowed the identification of very few DM-CpG belonging to Short Interspersed Elements (n = 1), Long Interspersed Elements (n = 1) and Long Terminal Repeat (n = 6) distributed at different locations in the genome, suggesting a negligible effect (Supplementary Table S2). Nevertheless, it is important to note that the 450K array is not particularly enriched in CpG of repetitive elements. A more in-depth analysis of this topic would require additional experiments using methodologies that are more appropriate.
Sperm chromosome stability has been inferred in this work from screening for numerical anomalies of the chromosomes X, Y, 13, 18 and 21. We have previously demonstrated that the analysis of these chromosomes is sufficient to identify infertile patients with a higher probability of producing chromosomally abnormal spermatozoa than is found among the general population (
). The results do not support an effect of differential DNA methylation on sperm chromosome stability. Neither the clustering of the most variable CpG nor of the DM-loci was associated with the sperm FISH results. Nevertheless, as chromosome instability has been mainly associated with hypomethylation of repetitive sequences and, as mentioned above, these elements are underrepresented in the array, other approaches must be applied. Specific bivariate gene analysis revealed a significant association of JAM3/NDCAP3 hypermethylation and altered FISH results. JAM3 (Junctional adhesion molecule 3) encodes a protein known to mediate cellular polarity during spermatogenesis, and which is essential for spermatid differentiation (
). NCAPD3 (Non-SMC Condensin II Complex, Subunit) is one of three non-SMC subunits that defines the condensin II complex, which promotes proper segregation of homologous chromosomes. It has been reported that condensin II complex defects promote merotelic attachment, resulting in the failure of chromosome congression and an increased propensity for lagging chromosomes, leading to whole-chromosome gains and losses (
The results shown in this paper discard any genome-wide effect of distinctive sperm DNA methylation on assisted reproductive technology outcome; neither the clustering of the most variable CpG nor of the DM-loci was associated with assisted reproductive technology outcome. These results suggest that, although the methylation differences detected by the 450K array affect the fertility status of patients through the alteration of the seminogram, they do not seem to have any detrimental effect beyond fertilization. However, an important consideration is that the size of the analysed population is relatively small and therefore the size of the dataset is still limited in its ability to provide definite conclusive results. When considering assisted reproductive technology outcome, the gene-specific analysis revealed a significant association; pregnancy rate was significantly lower in patients with hypermethylation at the four consecutive DM-CpG identified in intron 21 of ANK2. This gene encodes a member of the ankyrin family of proteins, which play key roles in activities such as cell motility, activation, proliferation, contact and the maintenance of specialized membrane domains (
). Nevertheless, its specific role in embryogenesis remains unknown.
In high-throughput methylation analysis, the results must be carefully considered to establish a cause-effect relationship. Accordingly, it is important to consider two aspects of the study to interpret the observed alterations.
First, the methylation array technology used in this study omits deep analysis of specific regions in favour of producing a landscape methylation profile. Therefore, this study did not analyse several continuous CpG within a single region, which would facilitate linking methylation profiles with biological significance. The required DNA input of the 450K array and the scarce amount of semen sample available for the analysis do not allow further expression analysis or pyrosequencing confirmation. Given these limitations, an arbitrary but strict criterion considering only those presenting at least three consecutive DM-CpG as DM-loci was applied. The 450K array employs Infinium I and Infinium II assays (
). Thus, when selecting loci with at least three consecutively DM-CpG, it is likely that selection in this study is very strict because the flanking CpG would have the same methylation status as those at the 3' terminus of the Infinium I probes. Applying this criterion, the 696 DM-CpG hosted by 471 DM-loci were reduced to 76 DM-CpG hosted by 17 DM-loci, all of them with Gene Annotation (Supplementary Table S3).
The second consideration to be taken into account while interpreting these results is that the gene list from this study is divergent from those previously published using locus-specific procedures or array-based methodologies (see ‘Introduction’). We have considered several explanations for these differences. In relation to the studies that identified methylation abnormalities using locus-specific methodologies at imprinted loci, it has been verified that most of the CpG analysed by these authors are underrepresented in the 450K array. Concerning the heterogeneous results derived from array-based methodology, most of the 27K-CpG are included in the 450K array (
). Therefore, the 27K and 450K methylation arrays results could be considered comparable from a technical point of view. However, there is great variability in the experimental design of the studies. For instance, while some papers analyse a pure fraction of spermatozoa (
Broad DNA methylation changes of spermatogenesis, inflammation and immune response-related genes in a subgroup of sperm samples for assisted reproduction.
). Moreover, the seminal alteration of the patients (from azoospermia to normozoospermia), the size of the populations, the primary data filtering, the statistical analysis of raw data and even the criteria used to consider a locus as differentially methylated are divergent among studies. This situation hampers an overall assessment of the results and emphasizes the need for consensus in the future.
Conclusions and future directions
Sperm DNA methylation affects fertility status at different levels. At the genome-wide level, DNA differences could be accompanied by a loss of the developmental capacity of germ cells causing low sperm count and infertility. At the specific gene level, this study identified a set of 17 genes potentially related to male fertility, some of them with alterations specifically associated with age (RPS6KA2), low sperm count (APCS), chromosome stability (JAM3/NCAPD3), and assisted reproductive technology outcome (ANK2).
Although further studies are required to determine the origin of the different patterns of sperm methylation, advanced paternal age could be one of the factors that increase the possibility of alterations. In turn, it could be accompanied by a loss of the developmental capacity of germ cells that causes infertility.
Considering the heterogeneity among studies, as well as variation in the methylome among infertile individuals, future genome-wide methylation studies in larger and well-defined cohorts of infertile patients are needed. These would provide better insights into the associations between sperm DNA methylation patterns and male infertility, and the identification of potential epigenetic biomarkers associated with fertility.
Acknowledgements
This work was supported by Projects PS09/00330 (Gobierno de España, Spain) and SGR2014-524 2 (Generalitat de Catalunya). The authors wish to thank Dr Francesca Vidal for her critical revision of the manuscript and the staff of The Spanish National Genotyping Centre (CeGen, www.cegen.org) (Madrid Node) for their technical assistance in the analysis of samples using the HumanMethylation450 Infinium BeadChip. This manuscript has been proofread by Proof-Reading-Service.org.
Appendix. Supplementary material
The following is the supplementary data to this article:
Genome-wide sperm deoxyribonucleic acid methylation is altered in some men with abnormal chromatin packaging or poor in vitro fertilization embryogenesis.
Semen samples showing an increased rate of spermatozoa with imprinting errors have a negligible effect in the outcome of assisted reproduction techniques.
Tex261, a novel gene presumably related but distinct from steroidogenic acute regulatory (StAR) gene, is regulated during the development of germ cells.
Abnormal methylation of the promoter of CREM is broadly associated with male factor infertility and poor sperm quality but is improved in sperm selected by density gradient centrifugation.
Broad DNA methylation changes of spermatogenesis, inflammation and immune response-related genes in a subgroup of sperm samples for assisted reproduction.
Cristina Camprubí obtained her PhD in Cell Biology at the Universitat Autònoma de Barcelona in 2005. She has expert knowledge in molecular human genetics and epigenetics. Her research is focused on investigating the relationship between epigenetics and assisted reproductive technology. She is Assistant Professor at the UAB and she is the founder and director of GenIntegral, a genetic and epigenetic advisory service for professionals and patients.
Article info
Publication history
Published online: September 15, 2016
Accepted:
September 1,
2016
Received in revised form:
August 30,
2016
Received:
December 17,
2015
Declaration: The authors report no financial or commercial conflicts of interest.
30506-5&id=rbmo1608-fig-0001.jpg){kind=link}
30506-5&id=rbmo1608-fig-0002.jpg){kind=link}
30506-5&id=rbmo1608-fig-0003.jpg){kind=link}
30506-5&id=rbmo1608-fig-0004.jpg){kind=link}