A Comparison of In Vitro Nucleosome Positioning Mapped with Chicken, Frog and a Variety of Yeast Core Histones☆

Using high-throughput sequencing, we have mapped sequence-directed nucleosome positioning in vitro on four plasmid DNAs containing DNA fragments derived from the genomes of sheep, drosophila, human and yeast. Chromatins were prepared by reconstitution using chicken, frog and yeast core histones. We also assembled yeast chromatin in which histone H3 was replaced by the centromere-specific histone variant, Cse4. The positions occupied by recombinant frog and native chicken histones were found to be very similar. In contrast, nucleosomes containing the canonical yeast octamer or, in particular, the Cse4 octamer were assembled at distinct populations of locations, a property that was more apparent on particular genomic DNA fragments. The factors that may contribute to this variation in nucleosome positioning and the implications of the behavior are discussed.


Figure 1
Nucleosomal DNA preparation. Agarose gel electrophoresis of purified, nucleosomal DNA fragments used for paired-end sequencing. Selected marker DNA sizes are indicated.

Figure 2
Size distributions of the histone octamer binding sites. The distributions of nucleosomal DNA lengths, derived from paired-end sequencing, for each genomic DNA sequence (colour coded), obtained from the five different core histone reconstitutes are shown. Numbers of molecules are presented as a fraction of the total number of molecules indicated by paired-end reads that aligned to each DNA in each reconstitute.

Figure 3
Core histone octamer positioning on genomic DNA sequences. The locations and relative abundance of frog histone octamer binding sites on each genomic DNA is presented in terms of (i) sequencing coverage (black) and (ii) calculated nucleosome dyads (red, see Methods). The maps have not been normalised in terms of total signal. Consequently the total signal for each map reflects the number of sequence reads that aligned to each sequence. This representation is intended to allow a comparison of the data in terms of the relative affinity of (frog) histone octamer for the four types of DNA.  Lack of sequence reads from the ILPR region of Phins. The number of paired-end reads that aligned to the region of the Phins DNA containing the polymorphic sequence (ILPR) is shown for DNA recovered from all five reconstitutes. The location of the ILPR sequence is indicated by the green rectangle. The transcription start site, and first exon, of the insulin gene are also shown.

Figure 7
Nucleosome formation on ILPR sequences. Agarose gel (EthBr) and Southern blot of DNA recovered from micrococcal nuclease digested chromatin formed on Phins. The plasmid DNA (Phins) was reconstituted with chicken histones and the resulting chromatin digested with micrococcal nuclease on a time course basis. Aliquots were removed at the indicated times and were fractionated on an agarose gel. For Southern blot analysis the DNA transferred to a filter was hybridised with a probe specific for the ILPR ( 32 P-5'-ACAGGGGTGTGGGG).Core histone octamer positioning on genomic DNA sequences. The locations and relative abundance of frog histone octamer binding sites on each genomic DNA is presented in terms of (i) sequencing coverage (black) and (ii) calculated nucleosome dyads (red, see Methods). The maps have not been normalised in terms of total signal. Consequently the total signal for each map reflects the number of sequence reads that aligned to each sequence. This representation is intended to allow a comparison of the data in terms of the relative affinity of (frog) histone octamer for the four types of DNA.

Figure 8a
Core histone octamer positioning on Phins DNA. The locations and relative abundance of histone binding sites on Phins genomic DNA, for each of the five types of reconstitute is presented in terms of (i) sequencing coverage (black) and (ii) calculated nucleosome dyads (red, see Methods). The maps have been adjusted so that the total signal for each map is normalised (to unity).

Figure 8b
Core histone octamer positioning on BLG DNA. The locations and relative abundance of histone binding sites on BLG genomic DNA, for each of the five types of reconstitute is presented in terms of (i) sequencing coverage (black) and (ii) calculated nucleosome dyads (red, see Methods). The maps have been adjusted so that the total signal for each map is normalised (to unity).

Figure 8c
Core histone octamer positioning on Mos1 DNA. The locations and relative abundance of histone binding sites on Mos1 genomic DNA, for each of the five types of reconstitute is presented in terms of (i) sequencing coverage (black) and (ii) calculated nucleosome dyads (red, see Methods). The maps have been adjusted so that the total signal for each map is normalised (to unity).

Figure 8d
Core histone octamer positioning on YRO DNA. The locations and relative abundance of histone binding sites on YRO genomic DNA, for each of the five types of reconstitute is presented in terms of (i) sequencing coverage (black) and (ii) calculated nucleosome dyads (red, see Methods). The maps have been adjusted so that the total signal for each map is normalised (to unity).

Figure 9
Core histone octamer positioning on genomic DNA sequences at high resolution. The locations and relative abundance of histone octamer binding sites on each genomic DNA, identified as nucleosome dyads, is presented, for all five types of reconstitute, for localised regions of the DNAs. The maps were adjusted so that the total signal for each entire map was normalised (to unity). The numbers indicating nucleosome position on the X axis are on the same scale as shown in Figure 4.

Figure 10
Relationships between histone octamer binding site affinity maps and DNA sequence.
Scatter plots of the relative free energy values (ΔG 0 ) of the binding sites identified from the dyad profiles, measured on each DNA reconstituted with the indicated type of core histone, and the frequency of the AAAA sequence motif within a 147 bp binding site, centred on the dyad of the site, are shown. For each scatter plot the correlation (R) is indicated.

Figure 11
Relationships between histone octamer binding site affinity maps and DNA sequence.
The binned data, used to illustrate the relationship between the relative free energy values (ΔG 0 ) of binding sites and their base composition ( Figure 8A), was employed to estimate, from the slopes of linear regression plots, the change in G+C content required to effect a one kcal/mol change in the average free energy (ΔG 0 ) of a binding site. Analyses for data derived from yeast, Cse4, chicken and frog histone octamer reconstitutes are shown for each of the four DNA types.

Table 1
Edited Bowtie 2 output for the alignment of paired-end reads from the sequencing of nucleosomal DNAs from 5 separate reconstitutions to the Phins reference sequence.
Some of the data (blue) was generated using scripts custom written for the purpose.

Table 2a-d
Relationships between histone octamer binding site affinity maps and DNA sequence.
The correlation between the relative free energy (ΔG 0 ) profiles of binding sites and profiles of the frequency of di, tri and tetranucleotide within the binding sites was determined for each DNA type. The data derived from yeast, Cse4, chicken and frog histone octamer reconstitutes was averaged and the standard deviation determined.
The data was then sorted to identify the top 30 positively (R) and negatively (-R) correlated sequence motifs. The ranking for G+C (yellow) and AAAA (blue) is identified.