Dynamics of nucleosome occupancy in developing male germ cells and mature spermatozoa

Nucleosomes are the building blocks of chromatin and thereby represent the first layer of genome organization in eukaryotes. Through their presence or absence, their post-translational modifications and the structural diversity provided by histone variants, nucleosomes constitute an essential epigenetic mechanism regulating countless cellular processes, including gene transcription. In the last stages of mammalian spermatogenesis, the genome of so-called elongating spermatids undergoes a profound remodelling of chromatin that results in the displacement of most nucleosomes from the genome. During this process, spermatid nuclei elongate and two small and highly basic germ cell-specific proteins, PRM1 and PRM2, ultimately take over as the main components of chromatin in sperm. Protamines bind directly to DNA and induce an extraordinary compaction of the genome safeguarding spermatozoa from the action of damaging agents.
Despite the replacement of nucleosomes by protamines is almost complete, ~1-10% of histones remain in mouse and human mature sperm chromatin, respectively. By classic MNase-seq and ChIP-seq approaches, sperm nucleosomes were initially found enriched at gene-regulatory regions characterized by CpG-rich sequences and devoid of DNA methylation. Sperm nucleosomes, which contain important regulatory histone marks such as H3K4me3 and H3K27me3, could thereby constitute a means of epigenetic inheritance between generations by supporting gene expression after fertilization. Subsequent studies using similar methodologies disagreed and proposed a preferential location of nucleosomes at intergenic loci and repetitive elements. Technical biases aggravated by the complex nuclear structure of sperm are likely at the core of these discrepancies. Among others, they originate from variability in the pre-treatments and enzymatic digestions required to solubilize sperm chromatin as well as from antibody pull-down efficiencies. As a consequence, the distributions and the relative abundances of nucleosomes in mammalian sperm genomes remain unclear. Similarly, the fraction of spermatozoa that contains nucleosomes at a given genomic region is, to date, unknown.
In my PhD thesis, I have used an orthogonal method to bypass technical biases associated to MNase-seq and ChIP-seq approaches employed in prior studies. I obtained total and relative nucleosome occupancy measurements at well-defined genomic regions of (i) mouse spermatids preceding or undergoing histone replacement and (ii) spermatozoa from three fertile human donors. Our method of choice, NOMe-seq, unveils the genomic positions of nucleosomes and other DNA-bound proteins as protection footprints generated through the methylation of cytosines within accessible GpC dinucleotides with an exogenous GpC methyltransferase. Using an amplicon-based amplification approach and a probabilistic mathematical model to infer and quantify NOMe-seq footprints present in our data, we observe that the extent of nucleosome loss during the histone-to-protamine exchange is comparable throughout the mouse genome at ~100 loci investigated, classified according to distinct sequence and gene regulatory characteristics.
These data argue that the dynamics and extent of chromatin remodelling are overall comparable at distinct genomic locations.
Mouse and human sperm chromatin were inaccessible to footprinting by NOMe-seq and required experimental decondensation of chromatin. Several published procedures were evaluated, finding that heparin treatment leads to effective displacement of sperm nucleosomes whereas detergent-based nuclear decompaction is not sufficient to grant accessibility to most sequenced fragments. Removal of protamines with Nucleoplasmin 2 did revert sperm chromatin inaccessibility without major measurable impact on nucleosome stability. NOMe-seq in nucleoplasmin-decondensed sperm from three fertile donors reveals that nucleosome footprints are only detectable in a small fraction of fragments at all amplicons investigated. Hence, we do not observe a penetrant retention of nucleosomes at any of the genomic loci analysed. We conclude that the total nucleosome occupancy levels are low at these loci, questioning any essential requirement in supporting functions post-fertilization.
Finally, owing to our probabilistic model, we were able to detect only a moderate association between a mild over-representation of nucleosome-containing fragments and the absence of DNA methylation within amplicons, both during chromatin remodelling in mouse spermatids and in decondensed human sperm. This is in line with prior studies, but the magnitude of this effect is considerably small when quantified as absolute levels in a population of spermatozoa. Together, we provide new evidence that, in spite of small variabilities and a potential weak modulatory function of DNA methylation, nucleosomes are by large only present at comparably low frequencies in genomic loci with distinct sequence and chromatin characteristics.