Evidence for Natural Selection in Nucleotide Content Relationships Based on Complete Mitochondrial Genomes: Strong Effect of Guanine Content on Separation between Terrestrial and Aquatic Vertebrates.

The complete vertebrate mitochondrial genome consists of 13 coding genes. We used this genome to investigate the existence of natural selection in vertebrate evolution. From the complete mitochondrial genomes, we predicted nucleotide contents and then separated these values into coding and non-coding regions. When nucleotide contents of a coding or non-coding region were plotted against the nucleotide content of the complete mitochondrial genomes, we obtained linear regression lines only between homonucleotides and their analogs. On every plot using G or A content purine, G content in aquatic vertebrates was higher than that in terrestrial vertebrates, while A content in aquatic vertebrates was lower than that in terrestrial vertebrates. Based on these relationships, vertebrates were separated into two groups, terrestrial and aquatic. However, using C or T content pyrimidine, clear separation between these two groups was not obtained. The hagfish (Eptatretus burgeri) was further separated from both terrestrial and aquatic vertebrates. Based on these results, nucleotide content relationships predicted from the complete vertebrate mitochondrial genomes reveal the existence of natural selection based on evolutionary separation between terrestrial and aquatic vertebrate groups. In addition, we propose that separation of the two groups might be linked to ammonia detoxification based on high G and low A contents, which encode Glu rich and Lys poor proteins.

INTRODUCTION

The concept of natural selection was established by Charles Darwin and Alfred Russell Wallace 150 years ago when genetic information was as yet unavailable. This theory was derived from specific differences or similarities in the phenotypes of organisms that lived on geologically isolated islands. Many complete genomes, including that of Homo sapiens [1, 2], have now been analyzed. On the one hand, our knowledge of genomic and genetic changes is not enough to clarify all phenotypic changes in organisms related to natural selection. On the other hand, we showed that vertebrates can be classified into terrestrial and aquatic groups in phylogenetic trees based on Ward’s clustering analysis using amino acid composition or nucleotide content predicted from complete mitochondrial genomes as traits [3, 4]. This result was consistent with that obtained from 16S RNA sequences [3, 4].

Cytochrome C [5], t-RNA [6, 7], 12S RNA [8, 9], 16S RNA [3, 4, 8-10] and 18S RNA genes [11] have been used to construct phylogenetic trees by various analytical methods based on amino acid or nucleotide sequence changes. Although these methods are applicable for single gene(s) or small genome fragments, they are not suitable for whole genomes that consist of a huge number of genes. However, using normalization of amino acid compositions or nucleotide contents, we can use whole genomes to compare various organisms. Indeed, biological evolution has been investigated based on cellular amino acid composition obtained from cell hydrolysates [11], and amino acid composition [12] or nucleotide contents [14] predicted from whole genomes. The ratios of amino acids to the total amino acids, or those of nucleotides to the total nucleotides can characterize whole genomes [15], and these indices were used to construct phylogenetic trees [3, 4]. The basis of this concept is the homogenous structure of the genome, composed of small units encoding similar amino acid compositions [13, 14], despite each gene’s different nucleotide sequence. Chargaff’s parity rules [G = C, A = T and (A + G) = (T + C)] are inter and intra molecular rules, and the first and second parity rules represent double DNA strand [16] and single DNA strand [17] molecules, respectively. The first parity rule is understandable based on Watson and Crick’s DNA model [18], whereas the second parity rule is not understandable because of the structure of single strand DNA. However, this question was recently solved using a simple mathematical calculation [19]. Mitchell and Bridge showed that Chargaff’s second parity rule is applicable for inter species research using a large data set of whole genomes, and that inter species nucleotide relationships are expressed with linear regression lines [20]. We also showed that inter species nucleotide relationships were expressed with linear regression lines among not only whole genomes but also coding or non-coding regions [21]. In our previous study, we constructed phylogenetic trees based on amino acid composition or nucleotide content as traits, and we classified vertebrates into terrestrial and aquatic groups [3, 4]. Consequently, this study was designed to show that simple plotting of nucleotide content predicted from complete mitochondrial genomes can also classify vertebrates into these two groups.

MATERIALS AND METHODS

The same data set that was used in our previous studies on nucleotide relationships [22-25] and phylogenetic tree construction [3, 4, 26] were investigated in this study. Thus, these organisms were chosen according to their alphabetical order without any preconception of the organisms’ characteristics [21]. All complete mitochondrial genomes (45 species) that were available were collected in our previous study [22] from the National Center for Biotechnology Information (NCBI, http://www.ncbi.nlm.nih.gov/sites). The list of organisms examined is shown in Table . Nucleotide contents were normalized to 1. (G + C + A + T = 1). Statistical analyses were carried out using Student’s t-tests with Microsoft Excel 2010.

RESULTS AND DISCUSSION

Coding Region Nucleotide Relationships

In our previous study, plotting nucleotide contents of the coding region against nucleotide contents of the complete mitochondrial genome, linear regression lines were obtained between homonucleotides and their analogs, whereas relationships were heteroskedastic between heteronucleotides and their analogs [22]. When G content of the coding region was plotted against G content of the complete mitochondrial genome, their relationships were expressed by linear regression lines (terrestrial; y = 1.0851 × −0.0222, R2 = 0.6596 and aquatic; y = 0.8409 × + 0.0133, R2 = 0.8527). This is consistent with our previous result [22]. Vertebrates were separated into two groups, terrestrial and aquatic, based on G content of the mitochondrial coding region or that of the complete mitochondrial genome (Fig. ). The G content of the complete mitochondrial genome in terrestrial vertebrates was lower than that in aquatic vertebrates. External biosphere bias might lead to the separation of vertebrates into terrestrial and aquatic groups as a result of natural selection [3]. The hagfish (Eptatretus burgeri), whose G content of the complete mitochondrial genome and of the coding region is the lowest among vertebrates examined, was separated from aquatic and terrestrial vertebrates (Fig. ). Similarly, plotting A content of the coding region against G content of the complete mitochondrial genome, linear regression lines were obtained, although their regression coefficients were barely reduced (terrestrial; y = −2.0201 × + 0.5893, R2 = 0.63 and aquatic; y = −1.3314 × + 0.4973, R2 = 0.5873) (Fig. ). In addition, vertebrates were separated into two groups, terrestrial and aquatic, based on A content in the mitochondrial coding region and the complete mitochondrial genome, although a slight overlap between two groups was observed (Fig. and Supplementary Fig. ). The A content of the complete mitochondrial genome in terrestrial vertebrates was higher than that in aquatic vertebrates versus the G content. G and A purine contents played an important role in separating vertebrates into terrestrial and aquatic groups. When C or T content pyrimidine of the coding region was plotted against G content of the complete mitochondrial genome, their relationships were heteroskedastic (Supplementary Fig. ). The hagfish was separated from both groups in every plot (Fig. ).

Non-coding Region Nucleotide Relationships

Using nucleotide content of the non-coding region, the relationships between non-coding G content and complete mitochondrial G content were expressed by linear regression lines (terrestrial; y = 0.7203x + 0.0609, R2 = 1.488: aquatic; y = 1.3523x − 0.0295, R2 = 0.7444) (Fig. ), as observed in the coding region (Fig. ). Clear separation between terrestrial and aquatic vertebrates was observed based on both G contents of the non-coding region and of the complete mitochondrial genome. G content in aquatic vertebrates was higher than that in terrestrial vertebrates in the non-coding region as well as in the complete mitochondrial genome. By contrast, A content in aquatic vertebrates was lower than that in terrestrial vertebrates, and vertebrates were separated into two groups based on A content of the non-coding region, as observed in the coding region (Fig. and Supplementary Fig. ). Relationships between C or T nucleotide content of the non-coding region and G content of the complete mitochondrial genome were heteroskedastic, and no significant separation between the two groups was observed (Fig. ).

The hagfish was separated from both terrestrial and aquatic vertebrates even in the case of using the non-coding region nucleotide content. The characteristics of the hagfish seem to be of a primitive vertebrate [27]. In our previous phylogenetic study of the sample containing both vertebrates and high C/G-invertebrates, the hagfish belonged to the vertebrate group, while it belonged to invertebrate group, using a mixture of vertebrates, high C/G- and low C/G-invertebrates [26]. This result could be attributed to the primitive characteristics of hagfish.

Regarding nucleotide contents of the non-coding region, G and A purine contents played an important role in separating vertebrates into aquatic and terrestrial groups, whereas C and T pyrimidine did not (Fig. and Supplementary Fig. ), as observed in the coding region (Fig. and Supplementary Fig. ). These results indicate that nucleotide alternations including mutations occurred randomly over the mitochondrial genome, as reported in whole chromosomal genomes based on amino acid compositions [13] or nucleotide contents [14]. Natural selection seems to contribute to evolution after genotype alternations.

In the codon table, Gly is encoded with GGT, GGC, GGA and GGG, and Glu is encoded with GAG and GAA, while Lys is encoded with AAA and AAG. These relationships indicate that high G and low A contents produce acidic proteins rather than basic proteins, because Gly is a neutral amino acid. Indeed, Glu content in fish cell hydrolysates was higher than that in rat or human cell hydrolysates [11]. In many fish that lack urea production in nitrogen metabolism, ammonia (which is fatally toxic to organisms) is excreted directly out of the body. Terrestrial vertebrates including mammals have a urea cycle, and excrete urea as a final nitrogen metabolite. As acidic proteins contribute to neutralize ammonia in the bodies of fish, high G and low A content must be linked with aquatic vertebrate evolution. In the dark-spotted frog (Rana nigromaculata) tadpoles are aquatic and ammonotelic, whereas after metamorphosis, the adult, which has a urea cycle, is terrestrial. This amphibian was grouped with aquatic vertebrates based on Ward’s clustering analysis using amino acid composition or nucleotide contents predicted from complete mitochondrial genomes as traits [25], and its G content was much closer to that of aquatic vertebrates in this study. Thus, these results seem to be based on genomic characteristics of R. nigromaculata. We have proposed that separation between aquatic and terrestrial vertebrates might be linked with nitrogen metabolism in biological evolution.

Table 1

Organisms examined.

Aquatic Vertebrates	Terrestrial Vertebrates
Amia calva	Artibeus jamaicensis
Auxis rochei	Boa constrictor
Cyprinus carpio	Bos taurus
Danio rerio	Bubalus bubalis
Diodon holocanthus	Canis familiaris
Eptatretus burgeri	Canis lupus
Gadus morhua	Cavia porcellus
Latimeria chalumnae	Chelonia mydas
Melanogrammus aeglefinus	Chlamydosaurus kingii
Myripristis berndti	Equus caballus
Neoceratodus forsteri	Eumetopias jubatus
Oncorhynchus keta	Gallus gallus
Oncorhynchus mykiss	Gekko gecko
Oryzias latipes	Heteronotia binoei
Paralichthys olivaceus	Homo sapiens
Polyodon spathula	Lemur catta
Reinhardtius hippoglossoides	Lyciasalamandra atifi
Salmo salar	Mus musculus
Takifugu rubripes	Mus musculus domesticus
Theragra chalcogramma	Ovis aries
	Rana nigromaculata
	Rattus norvegicus
	Sus scrofa
	Taeniopygia guttata
	Xenopus laevis