Direct but no transgenerational effects of decitabine and vorinostat on male fertility

Abstract

Establishment and maintenance of the correct epigenetic code is essential for a plethora of physiological pathways and disturbed epigenetic patterns can provoke severe consequences, e.g. tumour formation. In recent years, epigenetic drugs altering the epigenome of tumours actively have been developed for anti-cancer therapies. However, such drugs could potentially also affect other physiological pathways and systems in which intact epigenetic patterns are essential. Amongst those, male fertility is one of the most prominent. Consequently, we addressed possible direct effects of two epigenetic drugs, decitabine and vorinostat, on both, the male germ line and fertility. In addition, we checked for putative transgenerational epigenetic effects on the germ line of subsequent generations (F1-F3). Parental adult male C57Bl/6 mice were treated with either decitabine or vorinostat and analysed as well as three subsequent untreated generations derived from these males. Treatment directly affected several reproductive parameters as testis (decitabine & vorinostat) and epididymis weight, size of accessory sex glands (vorinostat), the height of the seminiferous epithelium and sperm concentration and morphology (decitabine). Furthermore, after decitabine administration, DNA methylation of a number of loci was altered in sperm. However, when analysing fertility of treated mice (fertilisation, litter size and sex ratio), no major effect of the selected epigenetic drugs on male fertility was detected. In subsequent generations (F1-F3 generations) only subtle changes on reproductive organs, sperm parameters and DNA methylation but no overall effect on fertility was observed. Consequently, in mice, decitabine and vorinostat neither affected male fertility per se nor caused marked transgenerational effects. We therefore suggest that both drugs do not induce major adverse effects-in terms of male fertility and transgenerational epigenetic inheritance-when used in anti-cancer-therapies.

Fig 1. Study design.

62 male C57Bl/6 mice were randomly assigned to four groups receiving either drugs or DMSO or were not treated at all (decitabine, vorinostat, DMSO control, untreated control). Three times per week, mice were injected intraperitoneally for seven weeks. After treatment, 10 males from each treated group were mated simultaneously with four healthy C57Bl/6 females to produce the F1-generation. Identical mating schemes were performed for the F1- and F2-generation. After mating, all male mice of one generation were analysed. Male mice of the F3-generation were not mated but analysed directly after they reached the age of 14 weeks.

Fig 2. Direct effects of decitabine and vorinostat treatment on reproductive organs and semen parameters.

A) Bi-testes weight/body weight, B) diameter of seminiferous tubules, C) height of seminiferous epithelium, D) percentage of diploid cells (undifferentiated germ cells and somatic cells) and E) of haploid cells (spermatids) in the testes, F) weight of accessory sex glands, G) weight of epididymides, H) sperm concentration and I) morphology. One point represents one animal. A), C)—G), I) The median (± interquartile range) or B), H) mean (± SEM) are shown, *: p < 0.05, **: p < 0.01, ***: p < 0.001.

Fig 3. Direct effects of decitabine and vorinostat on DNA methylation in blood of the P-generation.

Genes: A) Snrpn, B) Tcf3, C) IAPs and D) Oct4. Statistical differences were calculated for decitabine and vorinostat in comparison to DMSO vehicle control and for DMSO vehicle control in comparison to untreated control mice. One point represents one animal. The median (± interquartile range) is shown for each group, *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Fig 4. Direct effects of decitabine and vorinostat on DNA methylation in spermatozoa of the P-generation.

Genes: A) H19, B) Dazl, C) Abt1 and D) Tcf3. Statistical differences were calculated for decitabine and vorinostat in comparison to DMSO vehicle control and for DMSO vehicle control in comparison to untreated control mice. One point represents one animal. The median (± interquartile range) is shown for each group, *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Fig 5. Effects of treatment in the F1-generation.

A) Proportion of elongated spermatozoa (cells with highly condensed DNA) in the testes, B) sperm vitality and C) DNA methylation of Abt1 in spermatozoa. Statistical differences were calculated for decitabine and vorinostat in comparison to DMSO vehicle control. One point represents one animal. A, C) The median (± interquartile range) or B) the mean (± SEM) are shown for each group, *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Fig 6. Transgenerational effects of decitabine and vorinostat treatment on reproductive organs and semen parameters.

A) Percentage of diploid cells (undifferentiated germ cells and somatic cells) and B) of “double diploid” cells (spermatocytes) in the testes, C) Bi-testes weight, D) diameter of seminiferous tubules, E) height of seminiferous epithelium, F) sperm vitality. One point represents one animal. A)—C), E) The median (± interquartile range) or D), F) the mean (± SEM) are shown, *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Fig 7. Transgenerational effects of decitabine and vorinostat treatment on DNA methylation levels.

Genes: A) Abt1 in blood, B) Mest in spermatozoa, C) Snrpn and D) Tcf3 in blood, E) Oct4 in spermatozoa, F) Mest in blood. One point represents one animal. The median (± interquartile range) is shown for each group, *: p < 0.05, **: p < 0.01.

Fig 8. Clustering dendrogram based on DNA methylation of 1.35 × 106 CpG sites (min. coverage 10 reads) present in all samples.

Analysed samples: Isolated DNA of spermatozoa of two vehicle-control animals (C8, C14) and two descendants (analysed sample from the F3-generation of C8: F3C9 and from the F3-generation of C14: F3C30) as well as of three decitabine treated animals (D10, D11, D13) and three descendants (analysed samples from the F3-generation of D10: F3D1 and F3D31 and from the F3-generation of D11: F3D7). The distance is based on the Pearson’s correlation coefficient that ranges from +1 (two samples are most similar) to-1 (two samples are negatively correlated), while a 0 indicates absence of correlation. To visualize the difference between two samples the distance is calculated by converting the similarity matrix of the Pearson’s correlation coefficient into a distance matrix (1—Pearson’s correlation coefficient) which is then used for hierarchical clustering. Due to the low distance between the samples, our analysis revealed the absence of large differences in the spermatozoal methylome of decitabine treated and untreated animals and their descendants.