GC content increased at CpG flanking positions of fish genes compared with sea squirt orthologs as a mechanism for reducing impact of DNA methylation

Abstract

Background: Fractional DNA methylation in sea squirts evolved to global DNA methylation in fish. The impact of global DNA methylation is reflected by more CpG depletions and/or more A/T to G/C changes at CpG flanking positions due to context-dependent mutations of methylated CpG sites.

Methods and findings: In this report, we demonstrate that the sea squirt genes have undergone more CpG to TpG/CpA substitutions than the fish orthologs using homologous fragments from orthologous genes among Ciona intestinalis, Ciona savignyi, fugufish and zebrafish. To avoid premature transcription, the TGA sites derived from CGA were largely converted to TGG in sea squirt genes. By contrast, a significant increment of GC content at CpG flanking positions was shown in fish genes. The positively selected A/T to G/C substitutions, in combination with the CpG to TpG/CpA substitutions, are the sources of the extremely low CpG observed/expected ratios in vertebrates. The nonsynonymous substitutions caused by the GC content increase have resulted in frequent amino acid replacements in the directions that were not noticed previously.

Conclusion: The increased GC content at CpG flanking positions can reduce CpG loss in fish genes and attenuate the impact of DNA methylation on CpG-containing codons, probably accounting for evolution towards vertebrates.

Figure 1. Percentage of CpG involved substitutions in pairwise comparisons.

The test detected substitutions at NCGN sites in homologous fragments. The points on solid line represent percents of four types of substitutions to all that were detected on both homologous fragments, and those on dashed line show the results of nonsynonymous substitutions. The percentage is the average of the results using the homologous fragments from all pairwise comparisons among C. intestinalis, and C. savignyi, fugufish and zebrafish. The error bars are derived from standard deviation. Four types of substitutions on NCGN patterns are shown at x axis. The substitution type of M1 is referred to substitutions of T/A to G/C detected at −1 flanking position of CpG sites; P1 means the substitutions at +1 flanking position of CpG sites. TG and CA denote CpG to TpG and CpG to CpA respectively.

Figure 2. Percent of orthologous genes falling in different CpG o/e ranges.

The abbreviated species names are as indicated in Table 1. The CpG o/e ratios at x axis represent o/e ranges in size of 0.1. The high CpG o/e ratios in sea squirt orthologous genes indicate that a large number of these genes (<40%) are not methylated.

Figure 3. Substitutions of A/T for G/C at CpG flanking sites.

CpG o/e denotes CpG observed/expected ratio. We computed the proportion (β) of T/A that changes to G/C at CpG flanking sites using an equation CpG o/e = (1−α)/((1−5α/24+β) ×(1−α/24+β)), where α is the proportion of mutated CpG sites in sea squirt genes. The prediction lines were drawn with five α values 0.15, 0.2, 0.25, 0.3 and 0.35.

Figure 4. Most frequent amino acid changes due to CpG involved substitutions.

The CpG involved substitutions point to TG, CA, M1 and P1 (see Figure 1 for details), and the results of these substitutions are shown in Figures 4A-D respectively. Amino acid changes are ranked on the basis of the percentages in all amino acid changes. We listed four to five of the highest ranked amino acid changes, and others were skipped due to low percentages. The blue columns show the results of amino acid changes occurred in fish genes, and the error bars derived from standard deviation were based on four pairwise comparisons between the fish genes and the sea squirt genes; the yellow columns represent the results from the sea squirt genes.

Figure 5. p_N plotting at NCGN sites and non-CpG sites.

Homologous fragments, both containing more than 10 CpG sites, were used to calculate proportion of nonsynonymous difference (p_N) at NCGN sites and the other sites separately. The Num in the figure means the number of homologous fragments in analyses. Nonsynonymous substitutions on NCGN sites include M1, TG, CA, and P1 (see Figure 1 for details about the substitution types). See Methods for calculation of p_N. Spearman Rank Correlation test was used to evaluate the correlation between p_N values at NCGN sites and other sites (P<0.0001).