Diversity in G + C content at the third position of codons in vertebrate genes and its cause

Codon preferences in free-living microorganisms

A popular interpretation of the major codon preference is that it reflects the operation of a regulatory device that controls the expression of individual proteins. In this popular model, rapidly translated codons are thought to promote the accumulation of the highly expressed proteins and slowly translated codons are thought to retard the expression of poorly expressed proteins. However, this widely accepted model is not supported by kinetic theory or by experimental results. A less fashionable model in which the major codon preference has nothing to do with the expression level of the individual proteins is forwarded. In this model, the major codon preference is viewed as a global strategy to support the efficient function of the translation system and thereby to maximize the growth rates of cells under favorable conditions.

Oligonucleotide correlations between infector and host genomes hint at evolutionary relationships

The frequencies of oligonucleotides of length 3-6 were studied in 211 sequences of human DNA (659 kilobases), 22 sequences of DNA of human viruses (120 kbs), in 181 sequences of E. coli (442 kbs), and in 42 sequences of phages of E. coli (137 kbs). The sequences were obtained from Genbank(R) 48. The observed frequencies (O) were compared to the expected frequencies (E) obtained in two ways: 1) according to nucleotide composition for each series, and 2) according to first order Markow chains for triplets, second order for quadruplets, and third order for quintuplets and sextuplets. The ratio O/E was obtained for each oligonucleotide. Then, the correlation between the ratio O/E in a pair of series was calculated. Strong correlations were observed for sequences of man and human viruses, and for E. coli and its phages. Other correlations were small. For higher order Markov chains, there is indication of some correlation also between viruses and phages. It was concluded that through analysis of parallel oligonucleotide series it may be possible to infer some of the complex evolutionary relationships existing between cells and their infectors beyond the level of codon usage.

The 'effective number of codons' used in a gene

A simple measure is presented that quantifies how far the codon usage of a gene departs from equal usage of synonymous codons. This measure of synonymous codon usage bias, the 'effective number of codons used in a gene', Nc, can be easily calculated from codon usage data alone, and is independent of gene length and amino acid (aa) composition. Nc can take values from 20, in the case of extreme bias where one codon is exclusively used for each aa, to 61 when the use of alternative synonymous codons is equally likely. Nc thus provides an intuitively meaningful measure of the extent of codon preference in a gene. Codon usage patterns across genes can be investigated by the Nc-plot: a plot of Nc vs. G + C content at synonymous sites. Nc-plots are produced for Homo sapiens, Saccharomyces cerevisiae, Escherichia coli, Bacillus subtilis, Dictyostelium discoideum, and Drosophila melanogaster. A FORTRAN77 program written to calculate Nc is available on request.

Molecular evolution of the human Pgk-2 retroposon

The human phosphoglycerate kinase (Pgk) gene family includes the functional, intronless Pgk-2 gene and the intronless psi hPgk-1 pseudogene, both of which are retroposons of the intron-containing Pgk-1 gene. The divergence of the Pgk-2 retroposon from Pgk-1 is compared with that of the psi hPgk-1 retroposon from Pgk-1 to reveal nucleotide characteristics diagnostic of functional genes. A comparison of the human and mouse Pgk genes indicates that Pgk-2 has evolved more rapidly than Pgk-1 since the two genes diverged early in mammalian evolution, but that the lack of introns in Pgk-2 may have diminished inter-exon variation. The hypothesis that codon bias is related to expression level is shown not to hold for the Pgk genes; however, the idea that a deficiency of TA and CG dinucleotides and an excess of TG and CT dinucleotides contributes to codon bias is supported. Finally, the hypothesis that the Pgk-2 retroposon initially included a copy of the Pgk-1 'housekeeping' promoter and subsequently evolved a tissue-specific promoter is examined and supported. It is concluded that this process involved the loss of the 5' CpG island present in the Pgk-1 gene, and that selection for cell type-specific expression of Pgk-2 at high levels has driven the divergence of this retroposon from its progenitor, Pgk-1.

The MHC molecules of nonmammalian vertebrates

There is very little known about the long-term evolution of the MHC and MHC-like molecules. This is because both the theory (the evolutionary questions and models) and the practice (the animals systems, functional assays and reagents to identify and characterize these molecules) have been difficult to develop. There is no molecular evidence yet to decide whether vertebrate immune systems (and particularly the MHC molecules) are evolutionarily related to invertebrate allorecognition systems, and the functional evidence can be interpreted either way. Even among the vertebrates, there is great heterogeneity in the quality and quantity of the immune response. The functional evidence for T-lymphocyte function in jawless and cartilagenous fish is poor, while the bony fish seem to have many characteristics of a mammalian immune system. The organization and sequence of fish Ig genes also indicate that important events in the evolution of the immune system and the MHC occurred in the fish, but thus far there is no molecular evidence for recognizable MHC-like molecules in any fish. There is clearly an MHC in amphibians and birds with many characteristics like the MHC of mammals (a single genetic region encoding polymorphic class I and class II molecules) and evidence for polymorphic class I and class II molecules in reptiles. However, many details differ from the mammals, and it is not clear whether these reflect historical accident or selection for different lifestyles or environment. For example, the adult frog Xenopus has a vigorous immune system with many similarities to mammals, a ubiquitous class I molecule, but a much wider class II tissue distribution than human, mouse and chicken. The Xenopus tadpole has a much more restricted immune response, no cell surface class I molecules and a mammalian class II distribution. The axolotl has a very poor immune response (as though there are no helper T cells), a wide class II distribution and, for most animals, no cell surface class I molecule. It would be enlightening to understand both the mechanisms for the regulation of the MHC molecules during ontogeny and the consequences for the immune system and survival of the animals. These animals also differ markedly in the level of MHC polymorphism. Another difference from mammals is the presence of previously uncharacterized molecules. In Xenopus and reptiles, there are two populations of class I alpha chain on the surface of erythrocytes, those in association with beta 2m and those in association with a disulfide-linked homodimer.(ABSTRACT TRUNCATED AT 400 WORDS)

Doublet preference and gene evolution

Doublet preference analysis was carried out on coding and noncoding regions of Escherichia coli, Saccharomyces cerevisiae, and human mitochondrial and nuclear DNA. The preference pattern in 1-2 and 2-3 doublets in E. coli and S. cerevisiae correlated with that in noncoding regions. The 3-1 doublet preference in E. coli genes with low optimal codon frequency and in S. cerevisiae genes also showed a correlation with each of their noncoding doublet preference. A mechanism to explain these double preference correlations in doublet preference is presented: mutational biases, the origin of the noncoding region doublet preference, evolved so as to maintain the 1-2 and 2-3 doublet preference, which is determined by codon usage. These biases then acted on the 3-1 doublet, which was almost free of coding constraints, resulting in a similar preference in this doublet.

Chloroplast genes transferred to the nuclear plant genome have adjusted to nuclear base composition and codon usage

During plant evolution, some plastid genes have been moved to the nuclear genome. These transferred genes are now correctly expressed in the nucleus, their products being transported into the chloroplast. We compared the base compositions, the distributions of some dinucleotides and codon usages of transferred, nuclear and chloroplast genes in two dicots and two monocots plant species. Our results indicate that transferred genes have adjusted to nuclear base composition and codon usage, being now more similar to the nuclear genes than to the chloroplast ones in every species analyzed.

Codon usage in higher plants, green algae, and cyanobacteria

Codon usage is the selective and nonrandom use of synonymous codons by an organism to encode the amino acids in the genes for its proteins. During the last few years, a large number of plant genes have been cloned and sequenced, which now permits a meaningful comparison of codon usage in higher plants, algae, and cyanobacteria. For the nuclear and organellar genes of these organisms, a small set of preferred codons are used for encoding proteins. Codon usage is different for each genome type with the variation mainly occurring in choices between codons ending in cytidine (C) or guanosine (G) versus those ending in adenosine (A) or uridine (U). For organellar genomes, chloroplastic and mitochrondrial proteins are encoded mainly with codons ending in A or U. In most cyanobacteria and the nuclei of green algae, proteins are encoded preferentially with codons ending in C or G. Although only a few nuclear genes of higher plants have been sequenced, a clear distinction between Magnoliopsida (dicot) and Liliopsida (monocot) codon usage is evident. Dicot genes use a set of 44 preferred codons with a slight preference for codons ending in A or U. Monocot codon usage is more restricted with an average of 38 codons preferred, which are predominantly those ending in C or G. But two classes of genes can be recognized in monocots. One set of monocot genes uses codons similar to those in dicots, while the other genes are highly biased toward codons ending in C or G with a pattern similar to nuclear genes of green algae. Codon usage is discussed in relation to evolution of plants and prospects for intergenic transfer of particular genes.

Translation of bicistronic viral mRNA in transfected cells: regulation at the level of elongation

The S1 species of mammalian reovirus mRNA, like a number of other viral but not cellular mRNAs, codes for two dissimilar polypeptides by initiation of translation at two 5'-proximal, out-of-frame AUG codons. To determine if uninfected cells can utilize bicistronic genes, a bovine papilloma virus-based vector system was used to select mouse C127 cell lines containing multiple integrated copies of the reovirus S1 gene. These cell lines produced both reovirus polypeptides from a single mRNA. In addition, studies of COS cells transfected with the S1 gene containing small changes around the first AUG suggest that bicistronic mRNA translation is regulated at the level of elongation. A model is proposed in which ribosomes engaged in translation of one reading frame interfere with movement of ribosomes in the other frame because of differences in codon usage. Expression of bicistronic genes may be similarly regulated in virus-infected cells.

Nucleotide composition bias and CpG dinucleotide content in the genomes of HIV and HTLV 1/2

Nucleotide compositions of the HIV subfamily and HTLV 1/2 genomes are strongly biased in a remarkably opposite way; HIV is adenine-rich and cytosine-poor while HTLV 1/2 is cytosine-rich and adenine-poor. In addition, the CpG dinucleotides are underrepresented in HIV but abundant in HTLV 1/2. By these two properties the genomes of HIV and HTLV 1/2 mimic an (A + T)-rich and (G + C)-rich segment of the host genome, respectively. These dramatic differences between the two human retroviruses might have evolved to direct integration of the retroviral genomes into specific segments of the human chromosomes.

Strong functional GC pressure in a light-regulated maize gene encoding subunit GAPA of chloroplast glyceraldehyde-3-phosphate dehydrogenase: implications for the evolution of GAPA pseudogenes

The light-regulated nuclear gene encoding subunit A of chloroplast glyceraldehyde-3-phosphate dehydrogenase (subunit GAPA, gene Gpa1) from maize is extremely G + C rich (67% in codons). The genomic surroundings of this gene have been characterized together with the sequences of two strongly conserved Gpa pseudogenes isolated from a genomic maize library by differential cDNA hybridization. The comparisons show that the high G + C content of the maize gene is maintained independently of the surrounding noncoding sequences, which are G + C poor (42%), and only as long as the gene encodes a functional protein. After nonfunctionalization, Gpa pseudogenes rapidly loose G + C mainly due to enhanced turnover of CpG and CpXpG methylation sites. These results suggest that the maize Gpa1 gene is under strong functional GC pressure, due to constraints (CpG island) probably exerted at the transcriptional level. They also indicate that Gpa pseudogenes are methylated and that methylation was either the cause or the immediate consequence of their nonfunctionalization. It can be concluded further that the progenitor of pseudogenes 1 and 2 was a second functional Gpa gene (Gpa'), which, after duplication, accelerated in evolutionary rate due to relaxation of selective constraints. This is in agreement with the neutral theory of evolution. Comparison of Gpa intron sequences reveals a gradient of divergence: the more 3' the position of an intron the more its sequence has diverged between the three Gpa genes. A speculative model is presented explaining these observations in terms of a homologous recombination of genes with their reverse-transcribed pre-mRNAs.

5' flanking sequence and structure of a gene encoding rat 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase

The synthesis and degradation of fructose 2,6-bisphosphate, a ubiquitous stimulator of glycolysis, are catalyzed by 6-phosphofructo-2-kinase (EC 2.7.1.105) and fructose-2,6-bisphosphatase (EC 3.1.3.46), respectively. In liver, these two activities belong to separate domains of the same 470-residue polypeptide. Various mRNAs have been described for this bifunctional enzyme, which is controlled by hormonal and metabolic signals. To understand the origin and regulation of these mRNAs, we have characterized rat genomic clones encoding the liver isozyme, which is regulated by cAMP-dependent protein kinase, and the muscle isozyme, which is not. We describe here a 55-kilobase gene that encodes these isozymes by alternative splicing from two promoters. Each of the putative promoters was sequenced over about 3 kilobases and found to include nucleotide motifs for binding regulatory factors. The two isozymes share the same 13 exons and differ only by the first exon that, in the liver but not in the muscle isozyme, contains the serine phosphorylated by cAMP-dependent protein kinase. The gene was assigned to the X chromosome. An analysis of the exon limits of 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase in relation to its functional domains and to its similarity with other proteins plus its G + C content at the third codon position suggests that this gene originates from several fusion events.

Variation in G + C-content and codon choice: differences among synonymous codon groups in vertebrate genes

The relationship between G + C-content and codon usage in genes of human, mus, rat, bovine and chicken nuclear genomes was investigated. Correlation and lineal regression analyses were carried out on plots that related the frequency of each codon within each synonymous codon group to the G + C-content of the coding sequence as a whole. Under GC pressure, in most of the quartet codon groups there is a preferential choice of the C-ending codon, except in leucine and valine codon groups where the choice of the G-ending codon is preferred. Among ducts, the choice of codons specifying phenylalanine and glutamate shows the strongest dependence on G + C-content. The relationship found between G + C-content and codon usage in these genomes correlate with taxonomic distance.

Codon usage and gene expression level in Dictyostelium discoideum: highly expressed genes do 'prefer' optimal codons

Codon usage patterns in the slime mould Dictyostelium discoideum have been re-examined (a total of 58 genes have been analysed). Considering the extreme A + T-richness of this genome (G + C = 22%), there is a surprising degree of codon usage variation among genes. For example, G + C content at silent sites varies from less than 10% to greater than 30%. It was previously suggested [Warrick, H.M. and Spudich, J.A. (1988) Nucleic Acids Res. 16: 6617-6635] that highly expressed genes contain fewer 'optimal' codons than genes expressed at lower levels. However, it appears that the optimal codons were misidentified. Multivariate statistical analysis shows that the greatest variation among genes is in relative usage of a particular subset of codons (about one per amino acid), many of which are C-ending. We have identified these as optimal codons, since (i) their frequency is positively correlated with gene expression level, and (ii) there is a strong mutation bias in this genome towards A and T nucleotides. Thus, codon usage in D. discoideum can be explained by a balance between the forces of mutational bias and translational selection.

Novel third-letter bias in Escherichia coli codons revealed by rigorous treatment of coding constraints

A novel bias in codon third-letter usage was found in Escherichia coli genes with low fractions of "optimal codons", by comparing intact sequences with control random sequences. Third-letter usage has been found to be biased according to preference in codon usage and to doublet preference from the following first letter. The present study examines third-letter usage in the context of the nucleotide sequence when these preferences are considered. In order to exclude any influence by these factors, the random sequences were generated such that the amino acid sequence, codon usage, and the doublet frequency in each gene were all preserved. Comparison of intact sequences with these randomly generated sequences reveals that third letters of codons show a strong preference for the purine/pyrimidine pattern of the next codons: purine (R) is preferred to pyrimidine (Y) at the third site when followed by an R-Y-R codon, and pyrimidine is preferred when followed by an R-R-Y, an R-Y-Y or a Y-R-Y codon. This bias is probably related to interactions of tRNA molecules in the ribosome.

Evolution of chromosome bands: molecular ecology of noncoding DNA

Giemsa dark bands, G-bands, are a derived chromatin character that evolved along the chromosomes of early chordates. They are facultative heterochromatin reflecting acquisition of a late replication mechanism to repress tissue-specific genes. Subsequently, R-bands, the primitive chromatin state, became directionally GC rich as evidenced by Q-banding of mammalian and avian chromosomes. Contrary to predictions from the neutral mutation theory, noncoding DNA is positionally constrained along the banding pattern with short interspersed repeats in R-bands and long interspersed repeats in G-bands. Chromosomes seem dynamically stable: the banding pattern and gene arrangement along several human and murine autosomes has remained constant for 100 million years, whereas much of the noncoding DNA, especially retroposons, has changed. Several coding sequence attributes and probably mutation rates are determined more by where a gene lives than by what it does. R-band exons in homeotherms but not G-band exons have directionally acquired GC-rich wobble bases and the corresponding codon usage: CpG islands in mammals are specific to R-band exons, exons not facultatively heterochromatinized, and are independent of the tissue expression pattern of the gene. The dynamic organization of noncoding DNA suggests a feedback loop that could influence codon usage and stabilize the chromosome's chromatin pattern: DNA sequences determine affinities of----proteins that together form----a chromatin that modulates----rate constants for DNA modification that determine----DNA sequences. Theories of hierarchical selection and molecular ecology show how selection can act on Darwinian units of noncoding DNA at the genome level thus creating positionally constrained DNA and contributing minimal genetic load at the individual level.

Chromosome localization-dependent compositional bias of point mutations in Alu repetitive sequences

The Alu repetitive sequence family originated from a common ancestor. Its members, apparently free from functional constraints, are interspersed throughout primate genomes. We have found that base substitutions occurring during the evolution of primates caused a decrease in the average G + C content of those members of the family that are located in an A + T-rich region of the genome. The family members that are located in a G + C-rich genomic region have not changed their, already high, G + C content. This suggests that the regional differences in G + C content, which are responsible for chromosomal banding, are caused by an accumulation of mutations that, although selectively neutral in the majority, show different compositional bias in different regions of the vertebrate chromosome.

Nucleic acid composition, codon usage, and the rate of synonymous substitution in protein-coding genes

Based on the rates of synonymous substitution in 42 protein-coding gene pairs from rat and human, a correlation is shown to exist between the frequency of the nucleotides in all positions of the codon and the synonymous substitution rate. The correlation coefficients were positive for A and T and negative for C and G. This means that AT-rich genes accumulate more synonymous substitutions than GC-rich genes. Biased patterns of mutation could not account for this phenomenon. Thus, the variation in synonymous substitution rates and the resulting unequal codon usage must be the consequence of selection against A and T in synonymous positions. Most of the variation in rates of synonymous substitution can be explained by the nucleotide composition in synonymous positions. Codon-anticodon interactions, dinucleotide frequencies, and contextual factors influence neither the rates of synonymous substitution nor codon usage. Interestingly, the nucleotide in the second position of codons (always a nonsynonymous position) was found to affect the rate of synonymous substitution. This finding links the rate of nonsynonymous substitution with the synonymous rate. Consequently, highly conservative proteins are expected to be encoded by genes that evolve slowly in terms of synonymous substitutions, and are consequently highly biased in their codon usage.

Mutation rates differ among regions of the mammalian genome

In the traditional view of molecular evolution, the rate of point mutation is uniform over the genome of an organism and variation in the rate of nucleotide substitution among DNA regions reflects differential selective constraints. Here we provide evidence for significant variation in mutation rate among regions in the mammalian genome. We show first that substitutions at silent (degenerate) sites in protein-coding genes in mammals seem to be effectively neutral (or nearly so) as they do not occur significantly less frequently than substitutions in pseudogenes. We then show that the rate of silent substitution varies among genes and is correlated with the base composition of genes and their flanking DNA. This implies that the variation in both silent substitution rate and base composition can be attributed to systematic differences in the rate and pattern of mutation over regions of the genome. We propose that the differences arise because mutation patterns vary with the timing of replication of different chromosomal regions in the germline. This hypothesis can account for both the origin of isochores in mammalian genomes and the observation that silent nucleotide substitutions in different mammalian genes do not have the same molecular clock.

Why the rate of silent codon substitutions is variable within a vertebrate's genome

Different genes within the murine genome are diverging at different rates. The rate of synonymous codon substitutions in these genes is related to their base composition. It is proposed that the variabilities of rate of mutation accumulation, of codon choice and of average GC content within vertebrate genomes are caused by differences between DNA synthesis in repair and replication, as far as the frequency and compositional bias of mutations introduced by these systems are concerned. DNA repair contributes substantially to the evolution of the DNA domains, which are actively repaired in germline cells and which correspond to regions available for transcription in these cells and to Giemsa-negative bands in stained chromosomes.

Codon usage patterns in Escherichia coli, Bacillus subtilis, Saccharomyces cerevisiae, Schizosaccharomyces pombe, Drosophila melanogaster and Homo sapiens; a review of the considerable within-species diversity

The genetic code is degenerate, but alternative synonymous codons are generally not used with equal frequency. Since the pioneering work of Grantham's group it has been apparent that genes from one species often share similarities in codon frequency; under the "genome hypothesis" there is a species-specific pattern to codon usage. However, it has become clear that in most species there are also considerable differences among genes. Multivariate analyses have revealed that in each species so far examined there is a single major trend in codon usage among genes, usually from highly biased to more nearly even usage of synonymous codons. Thus, to represent the codon usage pattern of an organism it is not sufficient to sum over all genes as this conceals the underlying heterogeneity. Rather, it is necessary to describe the trend among genes seen in that species. We illustrate these trends for six species where codon usage has been examined in detail, by presenting the pooled codon usage for the 10% of genes at either end of the major trend. Closely-related organisms have similar patterns of codon usage, and so the six species in Table 1 are representative of wider groups. For example, with respect to codon usage, Salmonella typhimurium closely resembles E. coli, while all mammalian species so far examined (principally mouse, rat and cow) largely resemble humans.

Global variation in G+C content along vertebrate genome DNA. Possible correlation with chromosome band structures

The global, rather than local, variation in G+C content along the nuclear DNA sequences of various organisms was studied using GenBank sequence data. When long DNA sequences of the genomes of Escherichia coli and Saccharomyces cerevisiae were examined, the levels of their G+C content (G+C%) were found to be within a narrow range around that of the whole genome. The G+C% levels for sequences of vertebrate genomes, however, were found to cover a wide range, showing that their genome is a mosaic of sequences with different G+C% levels, in each of which the sequence is fairly homogeneous in its G+C% for a very long distance. Through surveying a human genetic map and GenBank DNA sequences, the global variations in G+C% along the human genome DNA were found to be correlated with chromosome band structures.

The effects of guanine and cytosine variation on dinucleotide frequency and amino acid composition in the human genome

One hundred twelve human DNA sequences were analyzed with respect to dinucleotide frequency and amino acid composition. The variation in guanine and cytosine (G + C) content revealed: (1) at 2-3 and 3-1 doublet positions CG discrimination is attenuated at high G + C, but TA disfavor is enhanced, and (2) several amino acids are subject to G + C change. These findings have been reported in part for collections of sequences from various species. The present study confirms that in a single organism--the human--the G + C effects do exist. Aspects of the argument that connects G + C with protein thermal stability are also discussed.

Sequence comparisons of three wild-type Bronze-1 alleles from Zea mays

We have sequenced genomic clones of two wild-type Bronze-1 (Bz1) alleles, and a cDNA clone from a third wild-type Bz1 allele from maize. Two overlapping transcripts initiate at least 250 bp apart. The first AUG codon after the shorter and more abundant transcript cap site(s) begins the longest open reading frame. The transcript is preceded by a putative TATA box, but not a recognizable CAAT box. The bz1 gene contains a single intron, and exhibits a strong bias for codons with the highest G+C content. Sequence polymorphisms among the Bz1 alleles include deletions/additions, a transposable element insertion, and single base pair substitutions.

Compositional compartmentalization and gene composition in the genome of vertebrates

The compositional distribution of coding sequences from five vertebrates (Xenopus, chicken, mouse, rat, and human) is shifted toward higher GC values compared to that of the DNA molecules (in the 35-85-kb size range) isolated from the corresponding genomes. This shift is due to the lower GC levels of intergenic sequences compared to coding sequences. In the cold-blooded vertebrate, the two distributions are similar in that GC-poor genes and GC-poor DNA molecules are largely predominant. In contrast, in the warm-blooded vertebrates, GC-rich genes are largely predominant over GC-poor genes, whereas GC-poor DNA molecules are largely predominant over GC-rich DNA molecules. As a consequence, the genomes of warm-blooded vertebrates show a compositional gradient of gene concentration. The compositional distributions of coding sequences (as well as of DNA molecules) showed remarkable differences between chicken and mammals, and between mouse (or rat) and human. Differences were also detected in the compositional distribution of housekeeping and tissue-specific genes, the former being more abundant among GC-rich genes.

Endosymbiotic origin and codon bias of the nuclear gene for chloroplast glyceraldehyde-3-phosphate dehydrogenase from maize

The nuclei of plant cells harbor genes for two types of glyceraldehyde-3-phosphate dehydrogenases (GAPDH) displaying a sequence divergence corresponding to the prokaryote/eukaryote separation. This strongly supports the endosymbiotic theory of chloroplast evolution and in particular the gene transfer hypothesis suggesting that the gene for the chloroplast enzyme, initially located in the genome of the endosymbiotic chloroplast progenitor, was transferred during the course of evolution into the nuclear genome of the endosymbiotic host. Codon usage in the gene for chloroplast GAPDH of maize is radically different from that employed by present-day chloroplasts and from that of the cytosolic (glycolytic) enzyme from the same cell. This reveals the presence of subcellular selective pressures which appear to be involved in the optimization of gene expression in the economically important graminaceous monocots.

The chicken c-fos gene: cloning and nucleotide sequence analysis

Using a DNA probe from an avian transforming virus, NK24, that contains the fos gene and that was newly isolated and characterized (M. Nishizawa, N. Goto, and S. Kawai, J. Virol. 61:3733-3740, 1987), we cloned and sequenced the chicken c-fos gene. Results showed that this gene, like the mouse or human c-fos gene, contains four exons encoding a predicted gene product of 367 amino acids which is about 79% homologous to p55c-fos (mouse), excluding several small insertions or gaps. The predicted gene product, however, contains two nonhomologous regions, as suggested by sequence analysis of the NK24 genome. About half the identical amino acids of these gene products are encoded by different codons, and the guanine-plus-cytosine content in the third letter of each codon used in the entire chicken c-fos coding region is high (93%). The 5'-flanking region of the TATA box contains a sequence that is homologous to a putative transcriptional regulatory sequence in the mouse or human c-fos gene. The 3'-noncoding region of chicken c-fos is strikingly homologous to that of mouse or human c-fos and covers a 67-base-pair adenine-plus-thymine-rich stretch which was previously reported to be essential for an inhibitory effect on the expression of the mouse c-fos gene.

Synonymous codon usage in Bacillus subtilis reflects both translational selection and mutational biases

Codon usage data for 56 Bacillus subtilis genes show that synonymous codon usage in B. subtilis is less biased than in Escherichia coli, or in Saccharomyces cerevisiae. Nevertheless, certain genes with a high codon bias can be identified by correspondence analysis, and also by various indices of codon bias. These genes are very highly expressed, and a general trend (a decrease) in codon bias across genes seems to correspond to decreasing expression level. This, then, may be a general phenomenon in unicellular organisms. The unusually small effect of translational selection on the pattern of codon usage in lowly expressed genes in B. subtilis yields similar dinucleotide frequencies among different codon positions, and on complementary strands. These patterns could arise through selection on DNA structure, but more probably are largely determined by mutation. This prevalence of mutational bias could lead to difficulties in assessing whether open reading frames encode proteins.

Structure and expression of the human creatine kinase B gene

Various cDNAs for creatine kinase type B (CK-B) were isolated from human cDNA libraries using a 26-oligonucleotide guess-mer probe. One of the cDNAs appeared to be almost full-length and contained an open reading frame coding for the 381 amino acid residues of the human CK-B polypeptide. The nucleotide sequences of the translated region as well as the primary protein structure show a high degree of homology with known CK-B and CK-M sequences of other vertebrates. The level of CK-B RNA as a measure of CK-B gene activity was determined in various human tissues and cultured cells. Our results confirm that CK-B is expressed in a tissue-specific manner and give support to the previously proposed relation between CK-B gene activity and cell proliferation. Screening of genomic DNA with various cDNA regions as probes revealed that there is only one CK-B gene per haploid genome. Gene cloning and sequencing indicated that CK-B is coded for by a relatively small gene of 3.2 kb in size, which is partially overlapped by an HTF island (A. P. Bird (1986) Nature (London) 321, 557-558) with an extremely high G + C content at its 5' end.

Isolation and sequence of complementary DNA encoding human extracellular superoxide dismutase

A complementary DNA (cDNA) clone from a human placenta cDNA library encoding extracellular superoxide dismutase (EC-SOD; superoxide:superoxide oxidoreductase, EC 1.15.1.1) has been isolated and the nucleotide sequence determined. The cDNA has a very high G+C content. EC-SOD is synthesized with a putative 18-amino acid signal peptide, preceding the 222 amino acids in the mature enzyme, indicating that the enzyme is a secretory protein. The first 95 amino acids of the mature enzyme show no sequence homology with other sequenced proteins and there is one possible N-glycosylation site (Asn-89). The amino acid sequence from residues 96-193 shows strong homology (approximately 50%) with the final two-thirds of the sequences of all known eukaryotic CuZn SODs, whereas the homology with the P. leiognathi CuZn SOD is clearly lower. The ligands to Cu and Zn, the cysteines forming the intrasubunit disulfide bridge in the CuZn SODs, and the arginine found in all CuZn SODs in the entrance to the active site can all be identified in EC-SOD. A comparison with bovine CuZn SOD, the three-dimensional structure of which is known, reveals that the homologies occur in the active site and the divergences are in the part constituting the subunit contact area in CuZn SOD. Amino acid sequence 194-222 in the carboxyl-terminal end of EC-SOD is strongly hydrophilic and contains nine amino acids with a positive charge. This sequence probably confers the affinity of EC-SOD for heparin and heparan sulfate. An analysis of the amino acid sequence homologies with CuZn SODs from various species indicates that the EC-SODs may have evolved from the CuZn SODs before the evolution of fungi and plants.

Co-evolution of base composition and codon usage in Xenopus laevis and human globin genes with long-range DNA organization of their genome

Eucaryotic DNA is punctuated by many A+T-rich segments that we named A+T-rich linkers. Two types of these A+T-rich linkers can be distinguished: (i) isolated A+T-rich linkers, and (ii) A+T-rich linkers crowded in clusters. We have analysed the distribution of A+T-rich linker across the alpha- and beta-globin gene domain in Xenopus laevis and human genomes using isodenaturation and electron microscopy. Comparison of our data with those previously obtained for the avian globin genes leads us to conclude that genes can be harboured indifferently in either domain. A correlation is established between the presence of A+T-rich linker inside introns and flanking regions and the A+T content of the coding sequence. For the coding sequence, a high A+T content is strongly correlated with high A+T content in the codon's third position and weakly in the first position.

Long range genome structure around the human alpha-globin complex analysed by PFGE

A map encompassing 300 kilobases (kb) in and around the human alpha-globin gene complex shows features with important implications for understanding the structure and function of the human genome. In contrast to other segments of the mammalian genome that have been analysed by pulsed field gradient electrophoresis (PFGE), this region contains an unusually high density of sites for infrequently cutting restriction enzymes that recognise GC rich motifs including the under-represented CpG doublet. This suggests that the 26 kilobase (kb) stretch of DNA containing the alpha-globin gene family, which is known from sequence analysis to be 60% GC rich, is itself embedded within a region of high GC content. This long-range structure, identified by PFGE, corresponds to a class of GC rich isochores that are thought to represent early replicating DNA present in Giemsa negative chromosomal bands. The identification of such regions by PFGE will be of value in understanding the organisation of human chromosomes and will influence the strategies used to construct a physical map of the genome.

Correlation between molecular clock ticking, codon usage fidelity of DNA repair, chromosome banding and chromatin compactness in germline cells

The vertebrate genome is built of long DNA regions, relatively homogeneous in GC content, which likely correspond to bands on stained chromosomes. Large differences in composition have been found among DNA regions belonging to the same genome. They are paralleled by differences in codon usage in genes differently localized. The hypothesis presented here asserts that these differences in composition are caused by different mutational bias of alpha and beta DNA polymerases, these polymerases being involved to different extents in the repair of DNA lesions in compact and relaxed chromatin, respectively, in germline cells.

Analysis of sequences from the extremely A + T-rich genome of Plasmodium falciparum

The genome of the human malaria parasite Plasmodium falciparum has an A + T content of about 82%, higher than any other organism whose DNA has been characterized. Computer analysis of 36 kb of available nucleotide sequences from this species showed that the coding regions, with an A + T content of 69.0%, are flanked by more A + T-rich regions of 86.0% A + T. Within the coding sequences, the A/T ratio was 1.68 in the mRNA sense strand, and overall A + T content in the three codon positions increased in the order 1st-2nd-3rd position. Codons with T or especially A in the third position were strongly preferred. Codon usage among individual parasite genes was very similar compared to genes from other species. Dinucleotide frequencies for the parasite DNA were close to those expected for a random sequence with the known base composition, except that the CpG frequency in the coding sequences was low.

Compositional constraints and genome evolution

Nucleotide sequences of all genomes are subject to compositional constraints that affect, to about the same extent, both coding and noncoding sequences; influence not only the structure and function of the genome, but also those of transcripts and proteins; are the result of environmental pressures; and largely control the fixation of mutations. These findings indicate that noncoding sequences are associated with biological functions; that the organismal phenotype comprises two components, the classical phenotype, corresponding to the "gene products," and a "genome phenotype," which is defined by the compositional constraints; and that natural selection plays a more important role in genome evolution than do random events.