Chromatin organization in sperm may be the major functional consequence of base composition variation in the human genome

Abstract

Chromatin in sperm is different from that in other cells, with most of the genome packaged by protamines not nucleosomes. Nucleosomes are, however, retained at some genomic sites, where they have the potential to transmit paternal epigenetic information. It is not understood how this retention is specified. Here we show that base composition is the major determinant of nucleosome retention in human sperm, predicting retention very well in both genic and non-genic regions of the genome. The retention of nucleosomes at GC-rich sequences with high intrinsic nucleosome affinity accounts for the previously reported retention at transcription start sites and at genes that regulate development. It also means that nucleosomes are retained at the start sites of most housekeeping genes. We also report a striking link between the retention of nucleosomes in sperm and the establishment of DNA methylation-free regions in the early embryo. Taken together, this suggests that paternal nucleosome transmission may facilitate robust gene regulation in the early embryo. We propose that chromatin organization in the male germline, rather than in somatic cells, is the major functional consequence of fine-scale base composition variation in the human genome. The selective pressure driving base composition evolution in mammals could, therefore, be the need to transmit paternal epigenetic information to the zygote.

Figure 1. Base composition predicts sites of nucleosome retention in human sperm.

Nucleosome retention sites (red) across two representative genomic regions coincide with many transcription start sites and also with local peaks of high GC-content (black). Broader retention is seen at two transcription factors that regulate development, ALX3 (A) and FOXB1 (B), and this also correlates with broader regions of high GC-content. The plots were generated using the UCSC genome browser. GC-content correlates strongly with the number of sequenced reads from mononucleosome-enriched fractions of the sperm genome (C). In comparison, there is only a very weak correlation between GC-content and the number of sequenced reads from the input genomic control (D). GC-content is an excellent predictor of regions of nucleosome retention in sperm across the human genome (E). ROC curves are shown for predictions across the genome in 150 bp windows using either GC- or CpG-content. CpG islands are also excellent predictors of sites of nucleosome retention in sperm (χ² –test, p-value<2.2×10⁻¹⁶, see also Figure S6).

Figure 2. The characteristic GC-content signature of human genes account for sperm nucleosome retention at transcription start sites.

Human genes show a characteristic base composition signature with high GC-content at their start sites (A), which correctly predicts high nucleosomes in sperm (B). In contrast, in a somatic tissue (resting T-cells), nucleosomes are positioned around a strong nucleosome free region at the start site, most likely due to transcription related processes (C). The high GC-content of transcription start sites means that they have high intrinsic nucleosome binding preferences (D), which correlates well with nucleosome retention in sperm, but not occupancy in somatic cells. The average plots were generated for the 4 kb region centered at the start site of all human protein-coding genes for the GC-content (A), the normalized nucleosome retention score (B), the predicted binding preferences (nucleosome model score from Kaplan et al) (D) and the shifted somatic nucleosome read count (C) measured in 150 bp windows.

Figure 3. GC-content predicts variation in nucleosome retention among gene classes and at distal regulatory regions.

GC-content signatures around the start sites of (A) housekeeping genes (black), (B) tissue-specific genes (genes expressed in a single tissue, blue) and (C) developmental regulators (red). Average nucleosome retention in sperm (E–G), average nucleosome occupancy in T-cells (I–K), and average nucleosome affinity around the start sites (M–O) of the same three classes of genes. Nucleosome retention in sperm, but not occupancy in T-cells, mirrors the GC-content and the intrinsic nucleosome affinity. The three gene classes contain 7,308 housekeeping genes, 1,686 tissue-specific genes and 538 transcription factors that regulate development. GC-content (D), nucleosome retention in sperm (H), nucleosome occupancy in T-cells (L), and nucleosome affinity (P) are also enriched at DNase I hypersensitive sites (HS) identified in embryonic stem (ES) cells. The average scores were calculated from 64,217 DNase I HS sites from ES cells located at least 1 kb away from any gene.

Figure 4. GC-content predicts sperm nucleosome retention at individual genes.

GC-content (A,C,E) and sperm nucleosome retention (B,D,F) around the transcription start site of housekeeping genes (A–B), tissue-specific genes (C–D), and transcription factors that regulate development (E–F). Each row of the heat map is an individual gene. Genes are clustered according to their GC-content and the same gene ordering is used in the nucleosome retention plots. In both cases values are calculated in 150 bp windows.

Figure 5. Nucleosome retention in sperm is linked to the formation of DNA methylation-free regions in the early embryo.

Most (74%, 11,264/15,237) CpG islands that remain unmethylated in ES cells overlap nucleosome retention sites in sperm. In contrast, only 29% (1774/6,127) of the CpG islands that are methylated in ES cells overlap sperm nucleosome retention sites (A). CpG islands that are unmethylated in ES cells are enriched for H3K4me3 in mature sperm (B) compared to CpG islands that are methylated in ES cells (C). H3K27me3 shows moderate enrichment in sperm around CpG islands that are unmethylated in ES cells (D) compared to around CpG islands that are DNA methylated in ES cells (E).

Figure 6. A model for nucleosome retention in human sperm.

During the histone to protamine transition nucleosomes are retained at GC-rich sites which have high intrinsic affinity for nucleosomes. This results in nucleosome retention at the start sites of many genes, especially at the start sites of housekeeping genes and master regulators, as well as at distal regulatory elements. Regions that retain nucleosomes in sperm are also frequently established as free from DNA-methylation (‘Me’) in the early embryo, further suggesting a connection between the transmission of paternal nucleosomes and the establishment of gene regulation in the early embryo.