Patterns of codon usage bias in three dicot and four monocot plant species

Transcription

1 Genes Genet. Syst. (2003) 78, p Patterns of codon usage bias in three dicot and four monocot plant species Akira Kawabe* and Naohiko T. Miyashita Laboratory of Plant Genetics, Graduate School of Agriculture, Kyoto University, Sakyo-ku, Kyoto, Japan (Received 30 July 2003, accepted 18 October 2003) Codon usage in nuclear genes of four monocot and three dicot species was analyzed to find general patterns in codon choice of plant species. Codon bias was correlated with GC content at the third codon position. GC contents were higher in monocot species than in dicot species at all codon positions. The high GC contents of monocot species might be the result of relatively strong mutational bias that occurred in the lineage of the Poaceae species. In both dicot and monocot species, the effective number of codons (ENCs) for most genes was similar to that for the expected ENCs based on the GC content at the third codon positions. G and C ending codons were detected as the preferred codons in monocot species, as in Drosophila. Also, many preferred codons are the same in dicot species. Pyrimidine (C and T) is used more frequently than purine (G and A) in four-fold degenerate codon groups. Key words: codon bias, GC contents, preferred codon INTRODUCTION Non-random codon usage in synonymous codon families has been observed in many organisms. The biased codon usage is thought to be a result of natural selection (Sharp and Li 1986, Akashi 1994, 1995) or directional mutation pressure (Jukes and Bhushar 1986, Osawa et al. 1988, Sueoka 1988, Kano et al. 1991). Positive correlation between frequencies of abundant trnas and frequently used codons (Ikemura 1981ab, Moriyama and Powell 1997) and between gene expression level and codon bias (Grantham et al. 1981, Moriyama and Hartl 1993) have been reported. These observations are considered to be evidence of natural selection related to efficiency of translation. On the other hand, the correlation between codon usage bias and GC content in surrounding noncoding region has been taken as a support for directional mutation pressure (Ikemura 1985). In some organisms, such as human and yeast, codon usage in genes is related to locations in the genome because of the mosaic patterns of GC content (Bernardi 1993, Sharp and Lloyd 1993). In plant species, codon usage has been also investigated (Murray et al. 1989, Campbell and Gowri 1990, Carels et al. 1995, 1998, Miyashita et al. 1998, Carels and Bernardi 2000, De Amicis and Marchetti 2000). These studies Edited by Hidenori Tachida * Corresponding author. akirakawabe@hotmail.com * Present address: Laboratory of Dynamic Cell Biology, Department of Biotechnology, Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita , Osaka, Japan revealed that codon usage bias also exists in plant species and that the patterns of codon usage differ between monocot and dicot species. Monocots (mostly Poaceae species), in comparison to dicot species, have greater GC content at all positions in codons, suggesting different genomic organization and mutational pressures. Direct experimental evidence for a positive correlation between codon usage and gene expression had been reported by transforming plants with vectors expressing genes with modified codon usage (Perlak et al. 1991, Jensen et al. 1996, Iannacone et al. 1997, Rouwendal et al. 1997). In both dicot and monocot species, when the third codon position was modified to a G or C, increased expression of the modified gene was observed. This relation between GC content at the third position and expression level suggests that codon bias is related to efficiency of translation in both monocot and dicot species. In the present study, hundreds of coding sequences (CDSs) from four monocot and three dicot species were analyzed for codon usage bias. The main purpose of this study was to investigate the general patterns of codon usage bias in plants. Different patterns of codon usage between CDSs within a species and between monocot and dicot species have been reported. Herein, we report the common features of codon choice in plant species and provide possible explanations for the similarities. Analyzed data. MATERIALS AND METHODS All codon usage data were obtained

2 344 A. KAWABE and N. T. MIYASHITA from the CUTG (Codon Usage Tabulated from Genbank) database (Nakamura et al. 1997). In the present study, we analyzed nuclear CDSs of four monocot species (all Poaceae species; Triticum aestivum, Hordium vulgare, Oryza sativa, and Zea mays) and three dicot species (Arabidopsis thaliana, Nicotiana tabacum, and Pisum sativum). Only nuclear genes were analyzed. CDSs that were registered more than twice and mitochondrial or chloroplast genomic genes were eliminated from the analyses. Because there is a negative correlation between codon bias and gene length, i.e., codon usage is restricted in short CDSs (Wright 1990), CDSs of less than 100 codons were not included. The monocot species analyzed were all from the grass family (Poaceae), because sufficient sequence amounts were available for analysis. To reduce any bias due to our limited sample, we included genes from other monocot species in our analysis. Codon usage in each gene of other monocot species was compared to that in analyzed seven species. These genes included calmodulin of Lilium longiflorum (Liliaceae), histone H3 of Asparagus officinalis (Liliaceae), invertase of Allium cepa and Tulipa gesneriana (Liliaceae), ACC synthase of Dendrobium crumenatum (Orchidaceae), and RbcS, ACC synthase and ACC oxidase of Musa acuminata (Musaceae). Data analysis. The tryptophan, methionine, and three stop codons were excluded form analysis. GC contents of the entire gene, first, second, and third codon positions (GCall, GC1, GC2, and GC3, respectively) and effective number of codons (ENCs)(Wright 1990) were then calculated. GC12 was used for neutrality plot analyses and is the average of GC1 and GC2. The expected ENC values from GC3 were calculated according to the equation (4) in Wright (1990), which assumes equal use of G and C (A and T) in degenerate codon groups. To determine the preferred codon for each synonymous codon group, codon usage of high and low ENC CDSs are compared according to the method of Akashi and Schaeffer (1997) by cutting 10% codons for each case. RESULTS Codon usage of plant species. GC contents in the seven species analyzed in the present study is summarized in Table 1. The average GC3s were significantly higher in monocot species than in dicot species. GC1 and GC2 values were also significantly higher in monocot species than in dicots. These results were consistent with the higher genomic GC contents in Poaceae species reported by Salinas et al. (1988). Differences in GC content were greatest at the third codon position followed by the first and second positions. The differences in GC content at the second codon position, where changes cause amino acid substitutions, between dicot and monocot species suggested that a different GC mutation bias leads to different codon choice despite changes in protein sequences. To analyze relations among the three codon positions, neutrality plots (GC12 vs GC3) (Sueoka 1988) were analyzed for the seven species (Figure 1). As previously reported (Murray et al. 1989, Campbell and Gowri 1990), monocot species had a wide range of GC3, whereas dicot species had narrow GC3 distributions. In monocot species except T. aestivum, there were significant correlations, and the slope of the regression line was approximately The significantly positive correlation in neutrality plots of monocot species indicated that the intragenomic GC mutation bias effect the GC contents similarly among all position of codons. In contrast, there were no significant correlations in dicot species, and the slope of the regression line was near zero suggesting that there are low mutation bias or high conservation of GC contents level throughout the genome. Selection against mutational bias could cause narrow distribution of GC contents and no correlation between GC12 and GC3 in dicot species. In this study, it could not be concluded whether a higher level of directional mutation exists in monocot species, or high levels of selection against directional mutation exist in dicot species. Relation between ENC and GC3. The codon bias Table 1. Summary of codon usages in Plant species. n # codons GCall GC1 GC2 GC3 ENC T. aestivum H. vulgare O. sativa Z. mays A. thaliana N. tabacum P. sativum Average values are shown.

3 Codon bias in plant species 345 Fig. 1. Neutrality plot (GC12 against GC3). A) T. aestivum; the regression line is y = 0.063x , R2 = 0.067, OP = B) H. vulgare; the regression line is y = 0.133x , R2 = 0.200, OP = C) O. sativa; the regression line is y = 0.143x , R2 = 0.246, OP = D) Z. mays; the regression line is y = 0.140x , R2 = 0.218, OP = E) A. thaliana; the regression line is y = 0.005x , R2 = 0.000, OP = F) N. tabacum; the regression line is y = 0.011x , R2 = 0.000, OP = G) P. sativum; the regression line is y = 0.018x R2 = OP = OP (optimal point) indicates the point at which the regression line crosses the diagonal line.

4 346 A. KAWABE and N. T. MIYASHITA parameter, ENC, revealed greater biased codon usage in Poaceae species than in dicot species (Table 1). The relation between GC3 and ENC (Nc plot, Wright 1990) was analyzed to determine whether the difference in ENC is related to the difference in GC contents between dicot and monocot species (Figure 2). Monocot species and dicot species showed different patterns of Nc plot. CDSs of dicot plants appeared to cluster around the expected ENCs of 30% to 50% GC3, whereas CDSs of monocot plants showed GC3s with a wide distribution of 50% to 100%. Although the Nc plot differed between dicot and monocot species, ENCs of most CDSs were close to the expected values based on their GC3s. The similarities between expected and actual ENCs suggest that most of Fige. 2. Relation between GC3 and ENCs (Nc plot). ENCs were plotted against GC content at the third codon position. The expected ENCs from GC3 (Wright 1990) are shown as a solid line. A) T. aestivum. B) H. vulgare. C) O. sativa. D) Z. mays. E) A. thaliana. F) N. tabacum. G) P. sativum.

5 Codon bias in plant species 347 CDSs contained codons predicted from the GC contents. To estimate the difference between observed and expected ENC values, we calculated (ENCexp-ENCobs)/ENCexp. The frequency distributions of (ENCexp-ENCobs)/ENCexp are shown in Figure 3. In both monocot and dicot species, there was a single peak. The shape and location of the peaks were similar among the seven monocot and dicot species. In all species, the peak located in of (ENCexp-ENCobs)/ENCexp value and most CDSs have of (ENCexp-ENCobs)/ENCexp values indicating that most CDSs have ENCs slightly smaller than expected ENC values from their GC3s. This result suggested that the difference in codon bias between monocot and dicot species is dependent on the difference in GC content at the third codon position. Despite the difference in GC contents among plant species, most CDSs in the dicot and monocot species used codons predicted from the GC content at the third codon position. The difference in ENC between dicot and monocot species might reflect the difference in GC contents at the third positions. Preferred codons in plant species. The Nc plot showed that the pattern of codon usage differs between dicot and monocot species. These differences resulted primarily from differences in GC content. However, within a species, the Nc plot pattern could not distinguish between natural selection and mutational pressure. Wright (1990) suggested two ways to distinguish selection Fig. 3. Frequency distribution of (ENCexp-ENCobs)/ENCexp. Table 2. Preferred codons in plants. DM TA HV ZM OS AT NT PS DM TA HV ZM OS AT NT PS DM TA HV ZM OS AT NT PS DM TA HV ZM OS AT NT PS Phe UUU Ser UCU * * Tyr UAU Cys UGU UUC * * * * * * UCC * * * * * UAC * * * * * * UGC * * * * * Leu UUA UCA stopuaa stopuga UUG * * * * UCG * * * * stopuag Trp UGG Leu CUU * * * Pro CCU His CAU * Arg CGU * * CUC * * * * * * CCC * * * * * CAC * * * * * * CGC * * * * CUA CCA * * * Gln CAA CGA CUG * * * * CCG * * * * CAG * * * * * * CGG Ile AUU * * * Thr ACU * * Asn AAU Ser AGU AUC * * * * * * ACC * * * * * * AAC * * * * * * AGC * * * * * AUA ACA Lys AAA Arg AGA Met AUG ACG * * * * AAG * * * * * * * AGG * * * * Val GUU * * * Ala GCU * * * Asp GAU * Gly GGU * * * GUC * * * * * * GCC * * * * * GAC * * * * * GGC * * * * * GUA GCA Glu GAA GGA * GUG * * * * * GCG * * * GAG * * * * * * GGG Species name; DM: D. melanogaster (Akashi 1995), TA: T. aestivum, HV: H. vulgare, ZM: Z. mays, OS: O. sativa, AT: A. thaliana, NT: N. tabacum, PS: P. sativum

6 348 A. KAWABE and N. T. MIYASHITA Fig. 4. PR2-bias plot [A3/(A3 + T3) against G3/(G3 + C3)]. Open circles indicate the average position for each plot. A) T. aestivum; average position is x = ± , y = ± B) H. vulgare; average position is x = ± , y = ± C) O. sativa; average position is x = ± , y = ± D) Z. mays; average position is x = ± , y = ± E) A. thaliana; average position is x = ± , y = ± F) N. tabacum; average position is x = ± , y = ± G) P. sativum; average position is x = ± , y = ±

7 Codon bias in plant species 349 and mutation bias. If mutation bias is the cause of codon bias, GC or AT should be used proportionally among the degenerate codon groups in a gene. In contrast, natural selection for codon choice would not necessarily cause proportional use of G and C (A and T). These differences might be more obvious in highly biased-codon CDSs, because highly biased codon CDSs use a restricted set of codons. Thus, preferred codons, which are those codons used frequently in each degenerate codon group in highly biased codon usage CDSs, were determined according to the method of Akashi and Schaeffer (1997)(Table 2). In two-fold degenerate codon groups, codons with G or C in the third position were detected as the preferred codons in four monocot species. This result is consistent with our observation of a negative correlation between GC3 and the ENCs in these monocot species (Figure 2). In monocot species, CDSs with low ENCs (highly biased codon usage) had high GC3, and the reverse was true. However, in four-fold degenerate codon groups, although most preferred codons ended in either G or C in the monocot species, codons with C in the third position were detected more frequently in some four-degenerate codon groups such as Arg, and Gly. In T. aestivum, codons with G in the third position were not detected as preferred codons in the Leu, Ala, and Ser codon groups. These results suggested that not only simple GC mutational pressure but also other factors influence codon preference in highly biased CDSs in Poaceae species. In dicot species, preferred codons tended to have U in the third position, especially in N. tabacum and P. sativum. This result is consistent with the AU richness of the third codon position in these two species (Table 1, Figures 1, 2). However, the preferred codons could be determined in only about half of the synonymous codon groups in N. tabacum and P. sativum. In A. thaliana, some codons ending in U were detected as preferred codons, although codons ending in G or C were also detected as preferred codons. In A. thaliana, more than half of the detected preferred codons were the same as those detected in monocot species. In the three dicot species, codons with A in the third position were not detected as preferred codons, with the exception of CCA in Pro four-fold degenerate codon group. Relation between A and T and between C and G contents. If GC mutation pressure determined codon usage in each codon group, G and C (A and T) should be used proportionally. However, in the above analyses, CDSs with highly biased codon usage did not use GC and AT equally. To investigate whether these biased codon choices are restricted in highly biased CDSs, the relation between G and C content and between A and T content in four-fold degenerate codon families were analyzed by PR2 bias plot (Sueoka 1999a) (Figure 4). Despite the different GC contents between dicot and monocot species, a higher proportion of pyrimidines (C and T) to purines (G and A) was observed in all seven species analyzed. C and T were used more frequently than G and A in all comparisons except for A3/(A3 + T3) contents in T. aestivum and G3/(G3 + C3) in A. thaliana. Differences between C and G and between A and T contents were observed commonly in most CDSs. This observation indicated that the pattern of codon choice is similar among all plant species analyzed and among all levels of codon bias. DISCUSSION High GC contents in monocot species. The four monocot species analyzed in this study had higher GC contents at all codon positions than dicot species had (Table 1). If monocot species diverged from dicot species (Chase et al. 1993), these patterns must have developed after the split of monocot species from dicot species. Another explanation is that Poaceae species have GC contents higher than those of other plant species. Codon usages in other non-poaceae monocot species are shown in Table 3. Monocot species have higher GC3 in all of these genes than dicot species, but the difference is not the same among the families. GC3 values in Liliaceae and Orchidaceae were slightly higher than those in dicot species, but those in Musaceae were only slightly lower than those in Poaceae species. The phylogenetic study (Chase et al. 1993) showed that the most divergent family from Poaceae is Liliaceae followed by Orchidaceae and then Musaceae. These phylogenetic relations indicate that high GC contents may be present in all monocot species, but greater in Commelinidae, especially in Poaceae. The different patterns of the neutrality plots between monocot and dicot species support this idea. Reduced selection against mutational bias for monocot species could cause a wide distribution of GC3 values as suggested for bacterial species (Sueoka 1999b). Although reduced selection and stronger mutation bias cannot be distinguished, one or both of these events occurred in the lineage that lead to monocot species. To examine the changes in GC contents in monocot species in greater detail, further analysis of additional CDSs from many non-poaceae species is needed. Neutrality plot in plant species. Neutrality plots of plant species (Figure 1) revealed that there are significant correlations between GC12 and GC3 in monocot but not dicot species. The slopes of the regression lines in monocot species are approximately 0.14, whereas those in dicot species are nearly zero. This value for monocots is approximately half of that estimated for bacterial species (Sueoka 1999b) but is similar to that for human genes (Sueoka 1999a, 2001). When CDSs with GC3 less than 0.6 were analyzed, monocot species did not show significant correlations and the slope of the regression line was

8 350 A. KAWABE and N. T. MIYASHITA Table 3. Codon usage bias of individual genes. Monocot Dicot Product TA HV ZM OS AT NT PS Other monocot species Asparagus officinalis Histone H3 GC ENC Lilium longiflorum calmodulin GC ENC Tulipa gesneriana Allium cepa Invertase GC ENC Dendrobium crumenatum Musa acuminata ACC synthase GC ENC Musa acuminata ACC oxidase GC ENC Musa acuminata RbcS GC ENC nearly zero (data not shown). The significant positive correlations between GC12 and GC3 in monocot species were related to the wide distribution of GC3 values. Because the GC3s of dicot and monocot species show compositional correlations of homologous genes (Carels et al. 1998), GC3 levels may be related to gene functions. The width of the GC3 distribution might be related to the variation in the levels of directional mutation pressure or selection against mutational biases. Similarities in patterns of codon usage among monocot and dicot plant species. In the present study, there were three similar findings among monocot and dicot species, although GC contents of these species were different. The first similarity is that the actual ENCs in most CDSs were similar to those predicted from their GC3s. The second similarity was presence of preferred codons that were similar among species. The preferred codons in monocot species were similar to those in Drosophila melanogaster (Akashi 1994). Codons with G and C at the third position were detected as preferred codons. In four-fold degenerate codon groups, C was determined as the preferred codon more frequently, although both G and C ending codons could be detected as preferred codons. In dicot species, especially in A. thaliana, many preferred codons were the same as those in monocots and Drosophila. Highly biased CDSs, which have restricted codon usages, have the same trend in codon choices. The third finding is that pyrimidines (C and T) were used more frequently in codons than purines (G and A). This pattern was found in all CDSs. In dicot and monocot species, despite the difference in GC contents, the codon choice tendencies were similar among most CDSs. These findings suggested that there are two main factors that determine the level and pattern of codon usage in plant species. The difference in GC/AT contents caused within-genome and inter-species differences in codon usage in plant species. The second factor is CT richness, which is similar among most CDSs in all species. Because of general bias for CT in codon usage, most CDSs use codons that are biased slightly from their GC contents. The causes of different GC contents are thought to be due to differences in mutational bias or natural selection for synonymous codon choice. Because all codon positions have higher GC contents in monocot species than dicot species, difference in mutation bias and/or genome composition are responsible for variations of GC contents between species. In contrast, the cause of different GC contents within a species is not easily explained. However, the similarity in preferred codons in monocot species, Drosophila, and A. thaliana, suggests there are GC preferences in highly biased codon usage CDSs. Because A. thaliana did not show GC richness in the third codon position, GC mutation bias as well as natural selection appear to cause the within-genome differences in codon usages.

9 Codon bias in plant species 351 CT richness should be related to natural selection. Because CT richness was observed irrespective the level of codon usage bias, a generalized mechanism for codon choice might exist. CDSs are generally located in both DNA strands, thus the biased use of pyrimidines is not the result of simple stand-specific mutational biases as that observed in mitochondrial genomes (Anderson et al. 1981; Roe et al. 1985; Desjardins and Morais 1990). Coding sequence specific bias might occur similarly among all plant species analyzed in this study. Although the exact mechanism underlying CT biased codon usage in plants could not be determined in the present study, translation and transcription efficiencies may be involved. To clarify the mechanism underlying difference in purine and pyrimidine usage in codons at the third position, direct analyses, such as transforming plants with modified codon usage genes, are needed. We express our gratitude to Dr. T. Takano and Mr. K. Yoshida for their critical reading of and comments on early versions of the manuscript. We also thank Dr. H. Tachida and two anonymous reviewers for their comments and suggestions. This work was supported in part by a JSPS Research Fellowship for Young Scientists (granted to A. K.). This article is contribution number 575 from Laboratory of Plant Genetics, Graduate School of Agriculture, Kyoto University. REFERENCES Akashi, H. (1994) Synonymous codon usage in Drosophila melanogaster: Natural selection and translational accuracy. Genetics 136: Akashi, H. (1995) Inferring weak selection from patterns of polymorphism and divergence at silent sites in Drosophila DNA. Genetics 139: Akashi, H. and Schaeffer, S. W. (1997) Natural selection and the frequency distributions of silent DNA polymorphism in Drosophila. Genetics 146: Anderson, S., Bankier, A. T., Barrell, B. G., de Bruijn, M. H., Coulson, A. R., Drouin, J., Eperon, I. C., Nierlich, D. P., Roe, B. A., Sanger, F., Schreier, P. H., Smith, A. J., Staden, R., Young, I. G. (1981) Sequence and organization of the human mitochondrial genome. Nature 290: Bernardi, G. (1993) The isochore organization of the human genome and its evolutionary history - a review. Gene 135: Campbell, W. H. and Gowri, G. (1990) Codon usage in higher plants, green algae, and cyanobacteria. Plant Physiol. 92: Carels, N., and Bernardi, G. (2000) Two classes of genes in plants. Genetics 154: Carels, N., Barakat, A., and Bernardi, G. (1995) The gene distribution of the maize genome. Proc. Natl. Acad. USA 92: Carels, N., Hatey, P., Jabbari, K., and Bernardi, G. (1998) Compositional properties of homologous coding sequences from plants. J. Mol. Evol. 46: Chase, M. W., Soltis, D. E., Olmstead, R. G., Morgan, D., Les, D. H., Mishler, B. D., Duvall, M. R., Price, R. A., Hills, H. G., Qiu, Y.-L., Kron, K. A., Rettig, J. H., Conti, E., Palmer, J. D., Manhart, J. R., Sytsma, K. J., Michaels, H. J., Kress, W. J., Karol, K. G., Clark, W. D., Hedren, M., Gaut, B. S., Jansen, R. K., Kim, K.-J., Wimpee, C. F., Smith, J. F., Furnier, G. R., Strauss, S. H., Xiang, Q.-Y., Plunkett, G. M., Soltis, P. S., Swensen, S. M., Williams, S. E., Gadek, P. A., Quinn, C. J., Egiuiarte, L. E., Golenberg, E., Learn, Jr. G. H., Graham, S. W., Barrett, S. C. H., Dayanandan, S., and Albert, V. A. (1993) Phylogenetics of seed plants: An analysis of nucleotide sequences from the plastid gene rbcl. Ann. Missouri Bot. Gard. 80: De Amicis, F., and Marchetti, S. (2000) Intercodon dinucleotides affect codon choice in plant genes. Nucleic Acids Res. 28: Desjardins, P., and Morais, R. (1990) Sequence and gene organization of the chicken mitochondrial genome. J. Mol. Biol. 212: Grantham, R., Gautier, C., Gouy, M., Jacobzone, M., and Mercier, R. (1981) Codon catalog usage is a genome strategy modulated for gene expressivity. Nucleic Acids Res. 9: r43 r74. Iannacone, R, Grieco, P. D., and Cellini, F. (1997) Specific sequence modifications of a cry3b endotoxin gene result in high levels of expression and insect resistance. Plant Mol. Biol. 34: Ikemura, T. (1981a) Correlation between the abundance of Escherichia coli transfer RNAs and the occurrence of the respective codons in its protein genes. J. Mol. Biol. 146: Ikemura, T. (1981b) Correlation between the abundance of Escherichia coli transfer RNAs and the occurrence of the respective codons in its protein genes: a proposal for a synonymous codon choice that is optimal for the E. coli transfer system. J. Mol. Biol. 151: Ikemura, T. (1985) Codon usage and trna content in unicellular and multicellular organisms. Mol. Biol. Evol. 2: Jensen, L. G., Olsen, O., Kops, O., Wolf, N. Thomsen, K. K., Von- Wettstein, D. (1996) Transgenic barley expressing a proteinengineered, thermostable (1,3-1,4)-beta-glucanase during germination. Proc. Natl. Acad. Sci. USA 93: Jukes, T. H. and Bhushar, V. (1986) Silent nucleotide substitution and G content of some mitochondria and bacterial genes. J. Mol. Evol. 24: Kano, A., Andachi, Y., Ohama, T., and Osawa, S. (1991) Novel anticodon composition of transfer RNAs in Micrococcus luteus, a bacterium with a high genomeic G content. J. Mol. Biol. 221: Miyashita, N. T., Kawabe, A., Innan, H., and Terauchi, R. (1998) Intra- and interspecific DNA variation and codon bias of the alcohol dehydrogenase (Adh) locus in Arabis and Arabidopsis species. Mol. Biol. Evol. 15: Moriyama, E. N. and Powell, J. R. (1997) Codon usage bias and trna abundance in Drosophila. J. Mol. Evol. 45: Moriyama, E. N. and Hartl, D. L. (1993) Codon usage bias and base composition of nuclear genes in Drosophila. Genetics 134: Murray, E. E., Lotzer, J., and Eberle, M. (1989) Codon usage in plant genes. Nucleic Acids Res. 17: Nakamura, Y., Gojobori, T., and Ikemura, T. (1997) Codon usage tabulated from the international DNA sequence database. Nucleic Acids Res. 25: Osawa, S., Ohama, T., Yamao, F., Muto, A., Jukes, T. H., Ozeki, H., and Umesono, K. (1988) Directional mutation pressure and transfer RNA in chice of the third nucleotide of synonymous two-codon sets. Proc. Nalt. Acad. Sci. USA 85: Perlak, F. J., Fuchs, R. L., Dean, D. A., Mcpherson, S. L., and

10 352 A. KAWABE and N. T. MIYASHITA Fishhoff, D. A. (1991) Modification of the coding sequence enhances plant expression of insect control protein genes. Proc. Natl. Acad. Sci. USA 88: Roe, B. A., Ma, D.-P., Wilson, R. K., and Wong, J. F.-H. (1985) The complete nucleotide sequence of the Xenopus laevis mitochondrial genome. J. Biol. Chem. 260: Rouwendal, G. J. A., Mendes, O., Wolbert, E. J. H., and de Boer, A. D. (1997) Enhanced expression in tobacco of the gene encoding green fluorescent protein by modification of its codon usage. Plant Mol. Biol. 33: Salinas, J., Matassi, G., Montero, L. M., and Bernardi, G. (1988) Compositional compartmentalization and compositional patterns in the nuclear genomes of plants. Nucl. Acids Res. 16: Sharp, P. M., and Li, W.-H. (1986) An evolutionary perspective on synopnymous codon usage in unicellular organisms. J. Mol. Evol. 24: Sharp, P. M., and Lloyd, A. T. (1993) Regional base composition variation along yeast chromosome III: evolution of chromosome primary structure. Nucleic Acids Res. 21: Sueoka, N. (1988) Directional mutation pressure and neutral molecular evolution. Proc. Natl. Acad. Sci. USA 85: Sueoka, N. (1999a) 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: Sueoka, N. (1999b) Two aspects of DNA base composition: G+C content and translation-coupled deviation from intra-strand rule of A = T and G = C. J. Mol. Evol. 49: Sueoka, N. (2001) Near homogeneity of PR2-bias fingerprinting in the human genome and their implications in phylogenetic analyses. J. Mol. Evol. 53: Wright, F. (1990) The effective number of codons used in a gene. Gene 87: