DNA G+C content of the third codon position and codon usage biases of human genes

The human genome, as in other eukaryotes, has a wide heterogeneity in the DNA base composition. The evolutionary basis for this heterogeneity has been unknown. A previous study of the human genome (846 genes analyzed) has shown that, in the major range of the G+C content in the third codon position (0.25-0.75), biases from the Parity Rule 2 (PR2) among the synonymous codons of the four-codon amino acids are similar except in the highest G+C range (Sueoka, N., 1999. Translation-coupled violation of Parity Rule 2 in human genes is not the cause of heterogeneity of the DNA G+C content of third codon position. Gene 238, 53-58.). PR2 is an intra-strand rule where A=T and G=C are expected when there are no biases between the two complementary strands of DNA in mutation and selection rates (substitution rates). In this study, 14,026 human genes were analyzed. In addition, the third codon positions of two-codon amino acids were analyzed. New results show the following: (a) The G+C contents of the third codon position of human genes are scattered in the G+C range of 0.22-0.96 in the third codon position. (b) The PR2 biases are similar in the range of 0.25-0.75, whereas, in the high G+C range (0.75-0.96; 13% of the genes), the PR2-bias fingerprints are different from those of the major range. (c) Unlike the PR2 biases, the G+C contents of the third codon position for both four-codon and two-codon amino acids are all correlated almost perfectly with the G+C content of the third codon position over the total G+C ranges. These results support the notion that the directional mutation pressure, rather than the directional selection pressure, is mainly responsible for the heterogeneity of the G+C content of the third codon position.     

Two Aspects of DNA Base Composition: G+C Content and Translation-Coupled Deviation from Intra-Strand Rule of A=T and G=C

The relative contribution of mutation and selection to the G+C content of DNA was analyzed in bacterial species having widely different G+C contents. The analysis used two methods that were developed previously. The first method was to plot the average G+C content of a set of nucleotides against the G+C content of the third codon position for each gene. This method was used to present the G+C distribution of the third codon position and to assess the relative neutrality of a set of nucleotides to that of the G+C content of the third codon position. The second method was to plot the intrastrand bias of the third codon position from Parity Rule 2 (PR2), where A=T and G=C. It was found that whereas intragenomic distributions of the DNA G+C content of these bacteria are narrow in the majority of species, in some species the G+C content of the minor class of genes distributes over wider ranges than the major class of genes. On the other hand, ubiquitous PR2 biases are amino acid specific and independent of the G+C content of DNA, so that when averaged over the amino acids, the biases are small and not correlated with the DNA G+C content. Therefore, translation coupled PR2-biases are unlikely to explain the wide range of G+C contents among different species. Considering all data available, it was concluded that the amino acid-specific PR2 bias has only a minor effect, if any, on the average G+C content. In addition, PR2 bias patterns of different species show phylogenetic relationships, and the pattern can be as a taxal fingerprint. Received: 5 November 1998 / Accepted: 1 March 1999     

Wide intra-genomic G+C heterogeneity in human and chicken is mainly due to strand-symmetric directional mutation pressures: dGTP-oxidation and symmetric cytosine-deamination hypotheses

The intra-strand Parity Rule 2 of DNA (PR2) states that A=T and G=C within each strands. Useful corollaries of PR2 are G/(G+C)=A/(A+T)=0.5, G/(G+A)=C/(C+T)=G+C, G/(G+T)=C/(C+A)=G+C. Here. A, T, G, and C represent relative contents of the four nucleotide residues in a specific strand of DNA, so that A+T+G+C=1. Thus, deviations from the PR2 is a sign of strand-specific (or asymmetric) mutation and/or selection pressures. The present study delineates the symmetric and asymmetric effects of mutations on the intra-genomic heterogeneity of the G+C content in the human genome. The results of this study on the human genome are: (1) When both two- and four-codon amino acids were combined, only slight departures from the PR2 were observed in the total ranges of G+C content of the third-codon position. Thus, the G+C heterogeneity is likely to be caused by symmetric mutagenesis between the two strands. (2) The above result makes the deamination of cytosine due to double-strand breathing of DNA [Mol. Biol. Evol. 17 (2000) 1371] and/or incorporation of the oxidized guanine (8-oxo-guanine) opposite adenine during DNA replication (dGTP-oxidation hypothesis) as the most likely candidates for the major cause of the diversities of the G+C content. (3) Patterns of amino acid-specific PR2-biases detected by plotting PR2 corollaries against the G+C content of third codon position revealed that eight four-codon amino acids can be divided into three types by the second codon letter: (a) C₂-type (Ala, Pro, Ser4, and Thr), (b) G₂-type (Arg4 and Gly), and (c) T₂-type (Leu4 and Val). (4) Most of the asymmetric plot patterns of the above three classes in PR2 biases can be explained by C₂→T₂ deamination of C₂pG₃ of C₂-type to T₂pG₃ (T₂-type) in both human and chicken. This explains the existence of some preferred codons in human and chicken. However, these biases (asymmetric) hardly contribute to the overall G+C content diversity of the third codon position.     

Near Homogeneity of PR2-Bias Fingerprints in the Human Genome and Their Implications in Phylogenetic Analyses

Genes of a multicellular organism are heterogeneous in the G+C content, which is particularly true in the third codon position. The extent of deviation from intra-strand equality rule of A = T and G = C (Parity Rule 2, or PR2) is specific for individual amino acids and has been expressed as the PR2-bias fingerprint. Previous results suggested that the PR2-bias fingerprints tend to be similar among the genes of an organism, and the fingerprint of the organism is specific for different taxa, reflecting phylogenetic relationships of organisms. In this study, using coding sequences of a large number of human genes, we examined the intragenomic heterogeneity of their PR2-bias fingerprints in relation to the G+C content of the third codon position (P ₃ ). Result shows that the PR2-bias fingerprint is similar in the wide range of the G+C content at the third codon position (0.30–0.80). This range covers approximately 89% of the genes, and further analysis of the high G+C range (0.80–1.00), where genes with normal PR2-bias fingerprints and those with anomalous fingerprints are mixed, shows that the total of 95% of genes have the similar finger prints. The result indicates that the PR2-bias fingerprint is a unique property of an organism and represents the overall characteristics of the genome. Combined with the previous results that the evolutionary change of the PR2-bias fingerprint is a slow process, PR2-bias fingerprints may be used for the phylogenetic analyses to supplement and augment the conventional methods that use the differences of the sequences of orthologous proteins and nucleic acids. Potential advantages and disadvantages of the PR2-bias fingerprint analysis are discussed. Received: 21 December 2000 / Accepted: 16 February 2001     

Asymmetric directional mutation pressures in bacteria

Background

When there are no strand-specific biases in mutation and selection rates (that is, in the substitution rates) between the two strands of DNA, the average nucleotide composition is theoretically expected to be A = T and G = C within each strand. Deviations from these equalities are therefore evidence for an asymmetry in selection and/or mutation between the two strands. By focusing on weakly selected regions that could be oriented with respect to replication in 43 out of 51 completely sequenced bacterial chromosomes, we have been able to detect asymmetric directional mutation pressures.

Results

Most of the 43 chromosomes were found to be relatively enriched in G over C and T over A, and slightly depleted in G+C, in their weakly selected positions (intergenic regions and third codon positions) in the leading strand compared with the lagging strand. Deviations from A = T and G = C were highly correlated between third codon positions and intergenic regions, with a lower degree of deviation in intergenic regions, and were not correlated with overall genomic G+C content.

Conclusions

During the course of bacterial chromosome evolution, the effects of asymmetric directional mutation pressures are commonly observed in weakly selected positions. The degree of deviation from equality is highly variable among species, and within species is higher in third codon positions than in intergenic regions. The orientation of these effects is almost universal and is compatible in most cases with the hypothesis of an excess of cytosine deamination in the single-stranded state during DNA replication. However, the variation in G+C content between species is influenced by factors other than asymmetric mutation pressure.

     

Directional mutation pressure,selective constraints,and genetic equilibria

Summary Rates of substitution mutations in two directions, v [from an A-T or T-A nucleotide pair (AT-pair) to a G-C or C-G nucleotide pair (GC-pair)] and u [from a GC-pair to an AT-pair], are usually not the same. The net effect, v/(u + v), has previously been defined as directional mutation pressure ( _d ), which explains the wide interspecific variation and narrow intragenomic heterogeneity of DNA G+C content in bacteria. In this article, first, a theory of the evolution of DNA G+C content is presented that is based on the equilibrium among three components: directional mutation pressure, DNA G+C content, and selective constraints. According to this theory, consideration of both u and v as well as selective constraints is essential to explain the molecular evolution of the DNA base composition and sequence. Second, the theory of directional mutation pressure is applied to the analysis of the wide intragenomic heterogeneity of DNA G+C content in multicellular eukaryotes. The theory explains the extensive intragenomic heterogeneity of G+C content of higher eukaryotes primarily as the result of the intragenomic differences of directional mutation pressure and selective constraints rather than the result of positive selections for functional advantages of the DNA G+C content itself.     

Directional mutation pressure,mutator mutations,and dynamics of molecular evolution

Using a general form of the directional mutation theory, this paper analyzes the effect of mutations in mutator genes on the G + C content of DNA, the frequency of substitution mutations, and evolutionary changes (cumulative mutations) under various degrees of selective constraints. Directional mutation theory predicts that when the mutational bias between A/T and G/C nucleotide pairs is equilibrated with the base composition of a neutral set of DNA nucleotides, the mutation frequency per gene will be much lower than the frequency immediately after the mutator mutation takes place. This prediction explains the wide variation of the DNA G + C content among unicellular organisms and possibly also the wide intragenomic heterogeneity of third codon positions for the genes of multicellular eukaryotes. The present analyses lead to several predictions that are not consistent with a number of the frequently held assumptions in the field of molecular evolution, including belief in a constant rate of evolution, symmetric branching of phylogenetic trees, the generality of higher mutation frequency for neutral sets of nucleotides, the notion that mutator mutations are generally deleterious because of their high mutation rates, and teleological explanations of DNA base composition. Presented at the NATO Advanced Research Workshop onGenome Organization and Evolution, Spetsai, Greece, 16–22 September 1992     

Evolution of codon usage and base contents in kinetoplastid protozoans

In this study we analyze and compare the trends in codon usage in five representative species of kinetoplastid protozoans (Crithidia fasciculata, Leishmania donovani, L. major, Trypanosoma cruzi and T. brucei), with the purpose of investigating the processes underlying these trends. A principal component analysis shows that the G+C content at the third codon position represents the main source of codon-usage variation, both within species (among genes) and among species. The non- Trypanosoma species exhibit narrow distributions in codon usage, while both Trypanosoma species present large within-species heterogeneity. The three non-Trypanosoma species have very similar codon-usage preferences. These codon preferences are also shared by the highly expressed genes of T. cruzi and to a lesser degree by those of T. brucei. This leads to the conclusion that the codon preferences shared by these species are the ancestral ones in the kinetoplastids. On the other hand, the study of noncoding sequences shows that Trypanosoma species exhibit mutational biases toward A + T richness, while the non- Trypanosoma species present mutational pressure in the opposite direction. These data taken together allow us to infer the origin of the different codon-usage distributions observed in the five species studied. In C. fasciculata and Leishmania, both mutational biases and (translational) selection pull toward G + C richness, resulting in a narrow distribution. In Trypanosoma species the mutational pressure toward A + T richness produced a shift in their genomes that differentially affected coding and noncoding sequences. The effect of these pressures on the third codon position of genes seems to have been inversely proportional to the level of gene expression.      

Synonymous codon usage in Cryptosporidium parvum: identification of two distinct trends among genes

The usage of alternative synonymous codons in the apicomplexan Cryptosporidium parvum has been investigated. A data set of 54 genes was analysed. Overall, A- and U-ending codons predominate, as expected in an A+T-rich genome. Two trends of codon usage variation among genes were identified using correspondence analysis. The primary trend is in the extent of usage of a subset of presumably translationally optimal codons, that are used at significantly higher frequencies in genes expected to be expressed at high levels. Fifteen of the 18 codons identified as optimal are more G+C-rich than the otherwise common codons, so that codon selection associated with translation opposes the general mutation bias. Among 40 genes with lower frequencies of these optimal codons, a secondary trend in G+C content was identified. In these genes, G+C content at synonymously variable third positions of codons is correlated with that in 5' and 3' flanking sequences, indicative of regional variation in G+C content, perhaps reflecting regional variation in mutational biases.     

Codon usage in Aspergillus nidulans.

Summary Synonymous codon usage in genes from the ascomycete (filamentous) fungus Aspergillus nidulans has been investigated. A total of 45 gene sequences has been analysed. Multivariate statistical analysis has been used to identify a single major trend among genes. At one end of this trend are lowly expressed genes, whereas at the other extreme lie genes known or expected to be highly expressed. The major trend is from nearly random codon usage (in the lowly expressed genes) to codon usage that is highly biased towards a set of 19–20 optimal codons. The G+C content of the A. nidulans genome is close to 50%, indicating little overall mutational bias, and so the codon usage of lowly expressed genes is as expected in the absence of selection pressure at silent sites. Most of the optimal codons are C- or G-ending, making highly expressed genes more G+C-rich at silent sites.     

A general model of codon bias due to GC mutational bias

Background

In spite of extensive research on the effect of mutation and selection on codon usage, a general model of codon usage bias due to mutational bias has been lacking. Because most amino acids allow synonymous GC content changing substitutions in the third codon position, the overall GC bias of a genome or genomic region is highly correlated with GC3, a measure of third position GC content. For individual amino acids as well, G/C ending codons usage generally increases with increasing GC bias and decreases with increasing AT bias. Arginine and leucine, amino acids that allow GC-changing synonymous substitutions in the first and third codon positions, have codons which may be expected to show different usage patterns.

Principal Findings

In analyzing codon usage bias in hundreds of prokaryotic and plant genomes and in human genes, we find that two G-ending codons, AGG (arginine) and TTG (leucine), unlike all other G/C-ending codons, show overall usage that decreases with increasing GC bias, contrary to the usual expectation that G/C-ending codon usage should increase with increasing genomic GC bias. Moreover, the usage of some codons appears nonlinear, even nonmonotone, as a function of GC bias. To explain these observations, we propose a continuous-time Markov chain model of GC-biased synonymous substitution. This model correctly predicts the qualitative usage patterns of all codons, including nonlinear codon usage in isoleucine, arginine and leucine. The model accounts for 72%, 64% and 52% of the observed variability of codon usage in prokaryotes, plants and human respectively. When codons are grouped based on common GC content, 87%, 80% and 68% of the variation in usage is explained for prokaryotes, plants and human respectively.

Conclusions

The model clarifies the sometimes-counterintuitive effects that GC mutational bias can have on codon usage, quantifies the influence of GC mutational bias and provides a natural null model relative to which other influences on codon bias may be measured.     

Study of Completed Archaeal Genomes and Proteomes： Hypothesis of Strong Mutational AT Pressure Existed in Their Common Predecessor

The number of completely sequenced archaeal genomes has been sufficient for a large-scale bioinformatic study.We have conducted analyses for each coding region from 36 archaeal genomes using the original CGS algorithm by calculating the total GC content(G+C),GC content in first,second and third codon positions as well as in fourfold and twofold degenerated sites from third codon positions,levels of arginine codon usage(Arg2:AGA/G;Arg4:CGX),levels of amino acid usage and the entropy of amino acid content distribution.In archaeal genomes with strong GC pressure,arginine is coded preferably by GC-rich Arg4 codons,whereas in most of archaeal genomes with G+C0.6,arginine is coded preferably by AT-rich Arg2 codons.In the genome of Haloquadratum walsbyi,which is closely related to GC-rich archaea,GC content has decreased mostly in third codon positions,while Arg4Arg2 bias still persists.Proteomes of archaeal species carry characteristic amino acid biases:levels of isoleucine and lysine are elevated,while levels of alanine,histidine,glutamine and cytosine are relatively decreased.Numerous genomic and proteomic biases observed can be explained by the hypothesis of previously existed strong mutational AT pressure in the common predecessor of all archaea.     

Analysis of directional mutation pressure and nucleotide content in mitochondrial cytochrome b genes

We present a new approach for analyzing directional mutation pressure and nucleotide content in protein-coding genes. Directional mutation pressure, the heterogeneity in the likelihood of different nucleotide substitutions, is used to explain the increasing or decreasing guanine-cytosine content (GC%) in DNA and is represented by µ_D, in agreement with Sueoka (1962, Proc Natl Acad Sci USA 48:582–592). The new method uses simulation to facilitate identification of significant A + T or G + C pressure as well as the comparison of directional mutation pressure among genes, even when they are translated by different genetic codes. We use the method to analyze the evolution of directional mutation pressure and nucleotide content of mitochondrial cytochrome b genes. Results from a survey of 110 taxa indicate that the cytochrome b genes of most taxa are subjected to significant directional mutation pressure and that the gene is subject to A + T pressure in most cases. Only in the anseriform bird Cairina moschata is the cytochrome b gene subject to significant G + C pressure. The GC% at nonsynonymous codon sites decreases proportionately with increasing A + T pressure, and with a slope less than one, indicating a presence of selective constraints. The cytochrome b genes of insects, nematodes, and eumycotes are subject to extreme A + T pressures (µ_D = 0.123, 0.224, and 0.130) and, in parallel, the GC% of the nonsynonymous codon sites has decreased from about 0.44 in organisms that are not subjected to A + T or G + C pressure to about 0.332, 0.323, and 0.367, respectively. The distribution of taxa according to the GC% at nonsynonymous codon sites and directional mutation pressure supports the notion that variation in these parameters is a phylogenetic component.     

Constraints on mutability in a multiallelic gene family

A highly variable family of related DNA sequences was examined in order to determine the effect of local sequence environment on substitution mutation; 29 sequences from the Brassica self-incompatibility gene family, which possess a high level of nonsynonymous mutations, were aligned and grouped according to their similarity and function. The level and distribution of substitution mutations were calculated. A nonrandom distribution of sequence variation was observed along the sequences. The effect of neighbor biases and structural and thermodynamic measures were then compared in the absence of strong codon conservation. Biases were observed in the rates of substitution of the same base pair in different local sequence environments. The effect of the 5 neighbor was such that nucleotide A or C was associated with more mutations than G or T. There were significant interactions of certain dinucleotides with the frequency of mutation. Sequence-dependent measures of helical stability, intrinsic curvature, components of curvature, and stacking interactions were calculated for each sequence. Decreased helical stability was found to be associated with increased mutation. The compound measure of curvature, calculated according to the wedge model, showed little association with mutation. However, the components of increased wedge angle and decreased twist both showed an association with increased mutation. A small effect of A-type DNA stacking was found to be associated with mutated bases. Correspondence to: G.J. King     

Ultraviolet photoproducts at the ochre suppressor mutation site in the glnU gene of Escherichia coli: relevance to "mutation frequency decline"

Summary Ochre suppressor mutations induced by UV in the Escherichia coli glnU tRNA gene are CG to TA transitions at the first letter of the anticodon-encoding triplet, CAA. Premutational UV photoproducts at this site have long been known to exhibit an excision repair anomaly (mutation frequency decline or MFD), whereby post-irradiation inhibition of protein synthesis enhances their excision and reduces suppressor mutation yields ten-fold. We sought to clarify the basis of this unique repair response by determining the spectrum of UV photoproducts on both strands of a 36 by region of glnU which includes the anticodon-encoding triplet. We found that four different photolesions are produced within the 3 by sequence corresponding to the tRNA anticodon: (i) on the transcribed strand, TC (6–4) photoproducts and TC cyclobutane dimers are formed in equal numbers at the site of the C to T transition, indicating that this site is a hotspot for the usually less frequent (6–4) photoproduct; (ii) on the nontranscribed strand, TT dimers are found opposite the second and third letters of the anticodon-encoding triplet, adjacent to the mutation site; and (iii) on the nontranscribed strand, an alkali-sensitive lesion other than a (6–4) photoproduct is formed, apparently at the G in the mutation site. We suggest that mutation frequency decline may reflect excision repair activity at closely spaced UV lesions on opposite strands, resulting in double-strand breaks and the death of potential mutants.     

Mutation bias is the driving force of codon usage in the Gallus gallus genome

Synonymous codons are used with different frequencies both among species and among genes within the same genome and are controlled by neutral processes (such as mutation and drift) as well as by selection. Up to now, a systematic examination of the codon usage for the chicken genome has not been performed. Here, we carried out a whole genome analysis of the chicken genome by the use of the relative synonymous codon usage (RSCU) method and identified 11 putative optimal codons, all of them ending with uracil (U), which is significantly departing from the pattern observed in other eukaryotes. Optimal codons in the chicken genome are most likely the ones corresponding to highly expressed transfer RNA (tRNAs) or tRNA gene copy numbers in the cell. Codon bias, measured as the frequency of optimal codons (Fop), is negatively correlated with the G + C content, recombination rate, but positively correlated with gene expression, protein length, gene length and intron length. The positive correlation between codon bias and protein, gene and intron length is quite different from other multi-cellular organism, as this trend has been only found in unicellular organisms. Our data displayed that regional G + C content explains a large proportion of the variance of codon bias in chicken. Stepwise selection model analyses indicate that G + C content of coding sequence is the most important factor for codon bias. It appears that variation in the G + C content of CDSs accounts for over 60% of the variation of codon bias. This study suggests that both mutation bias and selection contribute to codon bias. However, mutation bias is the driving force of the codon usage in the Gallus gallus genome. Our data also provide evidence that the negative correlation between codon bias and recombination rates in G. gallus is determined mostly by recombination-dependent mutational patterns.     

The Effect of Multiple Evolutionary Selections on Synonymous Codon Usage of Genes in the Mycoplasma bovis Genome

Mycoplasma bovis is a major pathogen causing arthritis, respiratory disease and mastitis in cattle. A better understanding of its genetic features and evolution might represent evidences of surviving host environments. In this study, multiple factors influencing synonymous codon usage patterns in M. bovis (three strains’ genomes) were analyzed. The overall nucleotide content of genes in the M. bovis genome is AT-rich. Although the G and C contents at the third codon position of genes in the leading strand differ from those in the lagging strand (p<0.05), the 59 synonymous codon usage patterns of genes in the leading strand are highly similar to those in the lagging strand. The over-represented codons and the under-represented codons were identified. A comparison of the synonymous codon usage pattern of M. bovis and cattle (susceptible host) indicated the independent formation of synonymous codon usage of M. bovis. Principal component analysis revealed that (i) strand-specific mutational bias fails to affect the synonymous codon usage pattern in the leading and lagging strands, (ii) mutation pressure from nucleotide content plays a role in shaping the overall codon usage, and (iii) the major trend of synonymous codon usage has a significant correlation with the gene expression level that is estimated by the codon adaptation index. The plot of the effective number of codons against the G+C content at the third codon position also reveals that mutation pressure undoubtedly contributes to the synonymous codon usage pattern of M. bovis. Additionally, the formation of the overall codon usage is determined by certain evolutionary selections for gene function classification (30S protein, 50S protein, transposase, membrane protein, and lipoprotein) and translation elongation region of genes in M. bovis. The information could be helpful in further investigations of evolutionary mechanisms of the Mycoplasma family and heterologous expression of its functionally important proteins.     

Nonrandom Intragenic Variations in Patterns of Codon Bias Implicate a Sequential Interplay Between Transitional Genetic Drift and Functional Amino Acid Selection

Although most codon third bases appear to be functionless, the synonymous codons so defined exhibit a strikingly nonrandom distribution (codon bias) within human and other genes. To examine this phenomenon further, we generated a database of DNA sequences encoding human transmembrane cell-surface receptor proteins. Using this database we show here that the guanine and cytosine content of codon third bases (GC3) varies intragenically with the nature of the specified receptor domains (transmembrane > extracellular > intracellular domains; p < 0.001), the phenotype of the encoded amino acids (hydrophobic > hydrophilic > neutral amino acids; p < 0.001), and the receptor affiliation of the transmembrane domain superfamily (G-protein- coupled receptors > receptor tyrosine kinases; p < 0.001). Within gene regions specifying transmembrane domains, GC3 declines as domain functionality becomes redundant with increasing hydrophobicity (p < 0.001). Codons containing the second-base cytosine (XCZ, which encodes neutral amino acids) are selectively depleted of third-base adenine content (A3: XCA codons) when encoding transmembrane domain residues, consistent with positive selection for transitional mutation of XCG to XTG (which encodes hydrophobic amino acids) rather than to the synonymous XCA. Supporting this XCG XTG mechanism of codon bias, the G3:A3 ratio of codons specifying the transmembrane amino acid glycine (GGZ) is intermediate between that of its functional homolog alanine (GCZ) and that of hydrophobic valine (GTZ), even though the C3:T3 ratios are similar. Conversely, nearest-neighbor analysis of third bases 5 to codons specifying valine and leucine (CTZ) confirms a significant difference in C3:T3 but not G3:A3 ratios (i.e., C3/G1 T3/G1 > C3/A1; p < 0.001), consistent with the functionally advantageous retention of hydrophobic residues. These data raise the possibility that patterns of intragenic codon bias reflect a balance between negative and positive selection, suggesting in turn that analysis of codon third-base usage may help to predict the functional significance of encoded products. Supplementary information: Current address: (K. Lin) College of Life Sciences, Beijing Normal University, Beijing 100875, China