Patterns of Temperature Adaptation in Proteins from the Bacteria Deinococcus radiodurans and Thermus thermophilus John H. McDonald Department of Biological Sciences, University of Delaware Asymmetrical patterns of amino acid substitution in proteins of organisms living at moderate and high temperatures (mesophiles and thermophiles, respectively) are generally taken to indicate selection favoring different amino acids at different temperatures due to their biochemical properties. If that were the case, comparisons of different pairs of mesophilic and thermophilic taxa would exhibit similar patterns of substitutional asymmetry. A previous com- parison of mesophilic versus thermophilic Methanococcus with mesophilic versus thermophilic Bacillus revealed several pairs of amino acids for which one amino acid was favored in thermophilic Bacillus and the other was favored in thermophilic Methanococcus. Most of this could be explained by the higher G1C content of the DNA of thermophilic Bacillus, a phenomenon not seen in the Methanococcus comparison. Here, I compared the meso- Downloaded from https://academic.oup.com/mbe/article/18/5/741/1018660 by guest on 28 September 2021 philic bacterium Deinococcus radiodurans and its thermophilic relative Thermus thermophilus, which are similar in G1C content. Of the 190 pairs of amino acids, 83 exhibited signi®cant substitutional asymmetry, consistent with the pervasive effects of selection. Most of these signi®cantly asymmetrical pairs of amino acids were asymmetrical in the direction predicted from the Methanococcus data, consistent with thermal adaptation resulting from universal biochemical properties of the amino acids. However, 12 pairs of amino acids exhibited asymmetry signi®cantly different from and in the opposite direction of that found in the Methanococcus comparison, and 21 pairs of amino acids exhibited asymmetry that was signi®cantly different from that found in the Bacillus comparison and could not be explained by the greater G1C content in thermophilic Bacillus. This suggests that selection due to universal biochemical properties of the amino acids and differences in G1C content are not the only causes of substitutional asymmetry between mesophiles and thermophiles. Instead, selection on taxon-speci®c properties of amino acids, such as their metabolic cost, may play a role in causing asymmetrical patterns of substitution. Introduction Organisms live at a wide range of temperatures, Most of the protein sequence comparisons of me- from ,08Cto.1008C, and they do so despite the dra- sophiles with thermophiles have examined a small num- matic effects of temperature on the function and stability ber of proteins from a broad range of organisms (Argos of their proteins. Understanding how proteins have et al. 1979; MeneÂndez-Arias and Argos 1989; Vogt, adapted to different temperatures therefore has long Woell, and Argos 1997). Some consistent patterns have been the subject of considerable research. One approach been evident, such as arginine being preferred over ly- has been to search for asymmetrical substitution patterns sine at higher temperatures, but the broad taxonomic in protein sequences from organisms living at different samples have made it dif®cult to know whether any gen- temperatures, usually prokaryotes living at moderate eral patterns of thermal adaptation are being obscured temperatures (mesophiles) compared with those living or exaggerated by taxon-speci®c asymmetries that may at high temperatures (thermophiles) (Argos et al. 1979; or may not be thermally adaptive. Recent whole-genome MeneÂndez-Arias and Argos 1989; Vogt, Woell, and Ar- sequence projects have made it possible to examine a gos 1997; Haney et al. 1999; McDonald, Grasso, and large number of protein sequences from a single pair of Rejto 1999). If protein evolution is largely due to neutral mesophilic versus thermophilic taxa. Haney et al. (1999) processes, or due to adaptation that is speci®c to each compared 115 protein sequences from the mesophilic protein and each site, then the substitutions are expected archaea Methanococcus maripaludis, Methanococcus to be symmetrical: there will be as many aligned sites vannielii, and Methanococcus voltae with the thermo- with amino acid A in the mesophile and amino acid B philic Methanococcus jannaschii. They found 26 of the in the thermophile as there are sites with the opposite 190 pairs of amino acids to show signi®cant (P , 0.01) pattern. Signi®cant asymmetry, such as a greater number asymmetry, suggesting that adaptation to temperature of sites with A in the mesophile and B in the thermo- had an effect on a substantial proportion of amino acid phile, is usually interpreted as evidence that selection substitutions. McDonald, Grasso, and Rejto (1999) ex- favors different amino acids at different temperatures, amined a similar data set from the same species of and much effort has been put into trying to identify bio- Methanococcus and compared the results with the pat- chemical properties of the amino acids that would ex- terns of asymmetry between the mesophilic bacterium plain this adaptation. Bacillus subtilis and the thermophilic Bacillus stearo- thermophilus. The Bacillus comparison displayed sig- ni®cant (P , 0.05) substitutional asymmetry at 54 pairs Key words: protein adaptation, thermophile, Deinococcus, of amino acids, and most of the asymmetry was consis- Thermus. tent in direction with that seen in the Methanococcus comparison. However, several pairs of amino acids Address for correspondence and reprints: John H. McDonald, De- partment of Biological Sciences, University of Delaware, Newark, Del- showed patterns of asymmetry that were signi®cantly aware 19716. E-mail: [email protected]. different and opposite in direction in the Methanococcus Mol. Biol. Evol. 18(5):741±749. 2001 and Bacillus data sets, suggesting that taxon-speci®c q 2001 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038 processes were indeed important. Most of these differ- 741 742 McDonald ences consisted of an amino acid with a more G1C-rich onym of Thermus aquaticus (Degryse, Glansdorff, and codon being favored in thermophilic Bacillus but less Pierard 1978), T. thermophilus and T. aquaticus have common in thermophilic Methanococcus, consistent low similarity in genomic DNA : DNA hybridization with the higher genomic G1C content of B. stearo- (Manaia et al. 1994; Williams et al. 1995) and 16s se- thermophilus compared with B. subtilis. Because me- quences (Saul et al. 1993), and T. thermophilus can sophilic and thermophilic Methanococcus differ little in grow in media containing 3% NaCl and has a higher genomic G1C content, McDonald, Grasso, and Rejto maximum growth temperature than T. aquaticus (Man- (1999) suggested that the asymmetrical substitution pat- aia and da Costa 1991). Here, I compared only T. ther- terns seen there gave a better indication of which amino mophilus with D. radiodurans, because T. thermophilus acid substitutions were adaptive at different has a slightly higher optimum growth temperature and temperatures. has more sequences publicly available than does T. Even when mesophiles and thermophiles have the aquaticus. same genomic G1C content, it would be hasty to inter- Downloaded from https://academic.oup.com/mbe/article/18/5/741/1018660 by guest on 28 September 2021 pret all asymmetrical substitution patterns between them Materials and Methods as evidence for thermal adaptation, because there are other processes besides changes in G1C content that All available protein sequences (including frag- could cause taxon-speci®c patterns of substitutional ments) for T. thermophilus were downloaded from the asymmetry. Amino acids vary in bioenergetic cost, and SwissProt, TrEMBL, and TrEMBLNew databases in Oc- amino acids with lower costs presumably will be fa- tober 2000. Sequences less than 20 amino acids long vored over functionally equivalent amino acids (Craig were discarded. For each sequence, the most similar pro- and Weber 1998; Craig et al. 1999). The relative bio- tein sequence from the complete D. radiodurans ge- energetic costs of different amino acids may vary among nome (White et al. 1999) was identi®ed using BLAST species, depending on the availability for uptake of each (Altschul et al. 1997) servers at the Institute for Geno- amino acid in the environment, the biosynthetic path- mic Research (http://www.tigr.org/tdb/CMR/gdr/htmls/ ways used to synthesize each amino acid, the abundance SeqSearch.html) and the National Center for Biotech- of raw materials for biosynthesis, and the effect of tem- nology Information (http://www.ncbi.nlm.nih.gov/blast/ perature and other environmental variables on the bio- blast.cgi). Where more than one sequence from T. ther- synthetic pathways. Environmental variables other than mophilus matched a single sequence in D. radiodurans, temperature, such as salinity, pH, and hydrostatic pres- presumably re¯ecting a gene duplication in T. thermo- sure, might also cause adaptive substitutional asymme- philus or loss of a duplicate in D. radiodurans, only the try that is unrelated to temperature. one sequence with the greatest identity to the D. ra- Only patterns of substitutional asymmetry that are diodurans sequence was used. Pairs of amino acid se- repeatedly observed in comparisons of mesophiles quences with less than 35% sequence identity were dis- paired with related thermophiles will be robust evidence carded. The resulting data set consisted of 186 sequenc- for thermal adaptation, while inconsistent