The coexistence of the nucleosome positioning code with the genetic code on eukaryotic genomes

Abstract

It is known that there are several codes residing simultaneously on the DNA double helix. The two best-characterized codes are the genetic code--the code for protein production, and the code for DNA packaging into nucleosomes. Since these codes have to coexist simultaneously on the same DNA region, both must be degenerate to allow this coexistence. A-tracts are homopolymeric stretches of several adjacent deoxyadenosines on one strand of the double helix, having unusual structural properties, which were shown to exclude nucleosomes and as such are instrumental in setting the translational positioning of DNA within nucleosomes. We observe, cross-kingdoms, a strong codon bias toward the avoidance of long A-tracts in exon regions, which enables the formation of high density of nucleosomes in these regions. Moreover, long A-tract avoidance is restricted exclusively to nucleosome-occupied exon regions. We show that this bias in codon usage is sufficient for enabling DNA organization within nucleosomes without constraints on the actual code for proteins. Thus, there is inter-dependency of the two major codes within DNA to allow their coexistence. Furthermore, we show that modulation of A-tract occurrences in exon versus non-exon regions may result in a unique alternation of the diameter of the '30-nm' fiber model.

Figure 1.

Relative frequencies for A-tracts and G-tracts occurrences in exons. Relative frequencies, defined as Log₁₀( f_DNR/P_DNR), are presented for A-tracts and G-tracts in exon sequences (blue bars) and compared to controls CS1, CS2 and CS3 (red, yellow and gray bars, respectively). X-axis is the length (in base pair) of each tract. Only tracts with significant number of occurrences are shown.

Figure 2.

Log₁₀ ( f_DNR/P_DNR) for A-tracts and G-tracts presented separately for linker (L-state) regions and nucleosome (N-state) regions within exons. Exon sequences (blue bars) are compared to controls CS1, CS2 and CS3 (red, yellow and gray bars respectively). X-axis is the length (in base pair) of each tract. Only tracts with significant number of occurrences are shown.

Figure 3.

Log₁₀(f_DNR/P_DNR) for A-tracts presented separately for regions coding for helices (3₁₀ helix, alpha helix and pi helix) and strands (beta sheet and beta bridge). Exon sequences (blue bars) are compared with controls CS1, CS2 and CS3 (red, yellow and gray bars respectively). X-axis is the length (in base pair) of the A-tracts. Shown are results for significant DNRs only.

Figure 4.

Alignments of A_n>5 with respect to exon/nonexon junctions. Protein coding genes are aligned with respect to start codon (column A), exon–intron junction (column B), intron–exon junction (column C), or stop codon (column D). Only nonexon regions free from overlapping genetic code were used. The frequencies of occurrence of nucleotides A and T (f_A and f_T, respectively) coming only from A_n>5 are displayed for natural sequences (blue lines), random sequences generated based on genome nucleotide composition (orange lines), and CS3 control sequences for exons (green lines). Since S. cerevisiae does not have sufficient number of introns for analysis, only alignments relative to start and stop codons are presented for S. cerevisiae.

Figure 5.

Alignments of nucleosome occupancy with respect to exon/nonexon junctions. Protein coding genes are aligned with respect to start codon (column A), exon–intron junction (column B), intron–exon junction (column C), or stop codon (column D). Only nonexon regions free from overlapping genetic code were used here. The frequency of N-state (well-positioned nucleosome) + F-state (delocalized nucleosome state) is calculated for each position along the aligned sequences of S. cerevisiae and C. elegans. Since S. cerevisiae does not have sufficient number of introns for analysis, only alignments relative to start and stop codons are presented for S. cerevisiae.

Figure 6.

Alignment of linker lengths with respect to start codon or stop codon. Saccharomyces cerevisiae protein coding genes, aligned with respect to start codon (A) or stop codon (B), are presented. Only nonexon regions free from overlapping genetic code were used. Each position along the sequence represents the linker size calculated by averaging the lengths of all linkers covering that position.

Figure 7.

Schematic illustration of structural modulation of the 30-nm fiber along a eukaryotic gene. In this model the diameter and packing density of the 30-nm fiber vary along the gene due to changes in the lengths of DNA linker regions. Circles represent the nucleosome core particles; lines represent DNA linkers (or nucleosome free regions).