Abstract

In most bacteria, two tRNAs decode the four arginine CGN codons. One tRNA harboring a wobble inosine (tRNAArgICG) reads the CGU, CGC and CGA codons, whereas a second tRNA harboring a wobble cytidine (tRNAArgCCG) reads the remaining CGG codon. The reduced genomes of Mycoplasmas and other Mollicutes lack the gene encoding tRNAArgCCG. This raises the question of how these organisms decode CGG codons. Examination of 36 Mollicute genomes for genes encoding tRNAArg and the TadA enzyme, responsible for wobble inosine formation, suggested an evolutionary scenario where tadA gene mutations first occurred. This allowed the temporary accumulation of non-deaminated tRNAArgACG, capable of reading all CGN codons. This hypothesis was verified in Mycoplasma capricolum, which contains a small fraction of tRNAArgACG with a non-deaminated wobble adenosine. Subsets of Mollicutes continued to evolve by losing both the mutated tRNAArgCCG and tadA, and then acquired a new tRNAArgUCG. This permitted further tRNAArgACG mutations with tRNAArgGCG or its disappearance, leaving a single tRNAArgUCG to decode the four CGN codons. The key point of our model is that the A-to-I deamination activity had to be controlled before the loss of the tadA gene, allowing the stepwise evolution of Mollicutes toward an alternative decoding strategy.

INTRODUCTION

The genetic code is composed of 16 families of decoding boxes, each including four codons with the same first two nucleotides. Depending on the amino acid, these synonymous codons are read by one, two or at most three isoacceptor tRNA species harboring distinct anticodons. Therefore, fewer than 61 isoacceptor species (usually between 22 to a maximum of 46) are used to decode the 61 sense codons in mRNAs. These cellular tRNA repertoires are primarily responsible for the efficiency and accuracy of mRNA translation. The tRNA repertoires vary greatly from one organism and organelle to another, with most of the variability being found in the type of nucleotide present at the first ‘so-called’ wobble position of the anticodon (position 34), which is often post-transcriptionally modified. By interacting with the third base of the codon, this frequently modified nucleotide-34 plays an essential role in determining the preferred codons to be read by the mature and functional tRNA (1–5).

Transfer RNAs harboring an unmodified wobble adenosine-34 are rare; thus, they are not frequently used during translation. The reason is that during tRNA maturation, the encoded wobble A34 in the anticodon of the precursor tRNAs is generally enzymatically deaminated to inosine (6-deaminated adenosine–hypoxanthine base) by specific tRNA:A34 deaminases. The resulting I34-containing tRNA was predicted to base pair with a C-ending codon in the Watson–Crick mode and with U- and A-ending codons in a slightly different ‘wobble’ conformation (6), whereas the binding with a G-ending codon was forbidden, as reviewed previously (2,7). However, among the three codons read by I34-containing tRNA, the A-ending codon was expected to be difficult to translate, and this proposal was verified with Escherichia coli tRNAArgICG, using an in vitro translation system (8). Confirmation of this wobble hypothesis, with both bases in the anti-conformation as initially predicted by Francis Crick, was finally obtained from the crystal structure of the 30S ribosomal subunit, with the anticodon stem loop derived from E. coli tRNAArgICG bound to the CGA codon in an mRNA fragment (9). Therefore, once a cell has evolved and begun using I34-containing tRNA, the fourth remaining codon ending with G, in the corresponding four synonymous codons of the family box, has to be read by a second tRNA isoacceptor harboring a C34-containing anticodon (Figure 1). Although this is the usual decoding strategy observed in many living cells (10,11), a few remarkable exceptions exist.

Figure 1.

Quartet and duet decoding boxes of the bacterial genetic code, for decoding the 20 amino acids. In the case of arginine, the bacterial tRNAArg set usually involved in decoding Arg codons is also indicated with the respective anticodons.

Figure 1.

Quartet and duet decoding boxes of the bacterial genetic code, for decoding the 20 amino acids. In the case of arginine, the bacterial tRNAArg set usually involved in decoding Arg codons is also indicated with the respective anticodons.

For example, in fungi and animals, all cytoplasmic tRNAs harboring a wobble A34 and a purine-35 (R35) in the middle of the anticodon, as well as A34 in the cytoplasmic tRNAArgICG, have their wobble base deaminated to inosine-34 by the Tad2/Tad3 heterodimeric enzyme during tRNA maturation (12–14). These I34R35-containing tRNAs are found in the decoding family boxes using three or four synonymous codons (Leu, Ile, Val, Ser, Pro, Thr and Ala) (11,15). However, in Arabidopsis thaliana and other land plants, the same cytoplasmic Tad2/Tad3 deaminase does not deaminate the wobble A34 of cytoplasmic tRNAArgACG, but only those of the other A34R35-containing tRNAs (16). This raises the question of how the Arg-CGN codons in plant cytoplasmic mRNAs are translated into arginine. Only the chloroplastic tRNAArgICG in A. thaliana (and probably in all land plants) contains a deaminated A34, and its formation is catalyzed by the nuclear encoded chloroplastic TadA, a deaminase that is similar to the bacterial ortholog (17,18).

In contrast to cytoplasmic tRNA of eukaryotes, but similar to plant chloroplasts, inosine-34 in bacterial tRNA is found exclusively in tRNAArgICG, belonging to the CGN decoding box. Here, the wobble A34 is deaminated by a homodimeric tRNA:A34-deaminase (TadA) that is specific for only A34-containing tRNAArgACG (19). No other bacterial tRNAs harboring a wobble A34, either naturally occurring or experimentally generated by mutation, are deaminated by TadA. This property facilitated the examination of the decoding properties of A34-containing tRNAs other than tRNAArgACG. Using a mutant tRNAProAGG of Salmonella typhimurium, in which the naturally occurring wobble G34 was mutated to A34, Björk and co-workers (20) demonstrated that the C-ending proline codon was read in vivo almost as efficiently as the wild-type G34-containing tRNAProGGG. Likewise, a mutant of E. coli tRNAGlyCCC, in which the naturally occurring wobble C34 was changed to A34 by site-directed mutagenesis, read all four GGN glycine codons, although the A-ending Gly-GGA codon was decoded with the lowest efficiency (21). Osawa and co-workers (22) experimentally proved in vitro that the naturally occurring A34-containing tRNAThrAGU from the bacterium Mycoplasma capricolum translates all four threonine ACN codons, and only the Thr-ACA codon showed greatly reduced efficiency. Notably, M. capricolum has evolved a second tRNAThrUGU harboring an unmodified wobble U34 for reading the ACA codon without wobbling (23); therefore, it has naturally compensated for the difficulty of reading the Thr-ACA codon by A34-containing tRNAThrAGU.

As for the mitochondria of the fungus Saccharomyces cerevisiae and the nematode Ascaris suum, the tadA genes are missing in their nuclear genomes, and consequently, their encoded mitochondrial tRNAArgACG harbors an unmodified wobble A34 (24,25). As no other mitochondrial tRNAArg belonging to the same CGN arginine box exists, it was concluded that this unique tRNAArgACG must decode all four synonymous CGN codons. However, no experiments have been performed to verify this hypothesis.

Escherichia coli TadA and cytoplasmic S. cerevisiae Tad2/Tad3 are essential enzymes, and the deletions of the corresponding genes are lethal (13,19). Together, these examples demonstrated that, at variance with the information reported in all textbooks, the essential inosine at the first anticodon position does not ‘extend’ the decoding capability of an A34-containing tRNA. On the contrary, it ‘restricts’ the precursor tRNA harboring an unmodified wobble A34 to read only three of the four potential synonymous codons, excluding only the synonymous codon ending with G. This remaining synonymous G-ending codon of the same decoding box has to be decoded by a C34-containing tRNA. However, as aforementioned, although I34:A3 wobble pairing is possible (9), in practice it is inefficient (8), and cells usually limit the usage of codons involving I34:A3 base pairing during translation (26–28).

In this report, we identified the tRNAArg set in the 36 fully sequenced genomes of Mollicutes currently available. This repertoire was then correlated with the presence or absence of a gene encoding a TadA deaminase in the Mollicute genome. This genomic analysis revealed that Mollicutes are evolving by setting up alternative, and probably more efficient, arginine decoding systems able to read all four CGN codons, thus bypassing the requirement for the usually essential bacterial tadA gene.

MATERIALS AND METHODS

Data processing

All bacterial genomes analyzed were obtained from Genbank. They are listed in Supplementary Table S1. The genes encoding the TadA (tRNA-specific adenosine deaminase) and CDA (cytidine deaminase) protein sequences from the different Mycoplasmas analyzed were obtained from Genbank via BLASTP at NCBI, using TadA of Bacillus subtilis subsp. subtilis str. 168 (NP_387899.1) as the query sequence under the default conditions. The sequences of a few additional bacterial TadA proteins were obtained from published articles (Table 1). The tRNAArg genes with the anticodons ACG, GCG, TCG, CCG (belonging to the quartet decoding arginine box) and TCT or CCT (belonging to the duet decoding arginine box) were retrieved and listed in one file (Supplementary Figure S1). The two available tRNAArg sequences (including indications of their modified nucleotides) from Mycoplasmas, and the sequences of 35 tRNAs specific for other amino acids, were obtained from the tRNADB-CE databank (http://trna.nagahama-i-bio.ac.jp) (32) and tRNAdb (http://trnadb.bioinf.uni-leipzig.de) (15). Two additional sequences of tRNAArg from Acholeplasma laidlawii (anticodon branch only) were obtained from a published report (33). The numbers of occurrences of each Arg-codon in mRNA were counted directly from each genome sequence obtained from Genbank. The phylogenies of Mollicutes were obtained from the MolliGen 3.0 database (http://cbib1.cbib.u-bordeaux2.fr/molligen3b/SPECIES/phylo.php) (29).

Table 1.

Comparative usage of Arg codons, number of tRNAArg genes and occurrence of the tadA gene in 10 bacterial and 36 parasitic Mollicute genomes

Number Species Group Number of Arg codons in ORFs
 
Anticodon and number of tRNA genes
 
Gene 
CGU CGC CGA CGG AGA AGG ACG GCG TCG CCG TCT CCT tadA 
1a E. coli str. K-12 substr. MG1655 Outer 28 485 29 996 4871 7432 2845 1651 4   1 1 1 1 
1b Nitrosomonas europaea ATCC 19718 Outer 13 425 14 553 4584 10 153 5082 3473 1   1 1 1 1 
1c A. aeolicus VF5 Outer 727 601 268 367 9229 12588 1   1 1 1 1 
1d Streptomyces avermitilis MA 4680 Outer 19 076 93 823 7656 74019 2208 9827 1   1 1 1 1 
1e Synechococcus elongatus PCC 6301 Outer 8173 24 198 8010 787 24 448 1135 1   1 1 1 1 
1f S. aureus subsp. aureus Mu50 Outer 10 775 2603 3956 388 9321 1202 2   1 1  1 
1g Bacillus cereus ATCC 14579 Outer 20 003 6523 7745 1911 13 891 3604 4   1 1 1 1 
1h B. subtilis subsp. subtilis str. 168 Outer 9150 10 389 4957 7839 13 194 4700 4   1 1 1 1 
1i Listeria monocytogenes EGDe Outer 10 836 6301 5099 2578 5899 1102 2   1 1 1 1 
1j Oenococcus oeni PSU1 Outer 5934 2698 2965 2152 3951 1353 1   1 1 1 1 

 
A. laidlawii PG-8A IV 4075 872 747 61 8639 670 1    1  1 
Aster yellows witches'-broom phytoplasma AYWB IV 1183 710 305 42 2109 222 1    1  1 
Candidatus Phytoplasma australiense IV 1332 804 505 77 2484 358 1    1  1 
Candidatus Phytoplasma mali IV 1047 122 350 33 1972 168 1    1  1 
Onion yellows phytoplasma OY-M IV 1455 843 379 58 2457 237 1    1  1 

 
Mesoplasma florum L1 996 66 127 2 5444 190 1    1  1 
M. capricolum subsp. capricolum ATCC 27343 904 100 153 6 6115 184 1    1  1 
Mycoplasma leachii PG50 931 107 147 5 6154 175 1    1  1 
10 M. mycoides subsp. mycoides SC str. PG1 1061 95 167 10 7324 272 1    1  1 
11 M. mycoides subsp. capri LC str. 95010 1048 107 157 9 7275 252 1    1  1 

 
12 Mycoplasma agalactiae III 1349 258 153 56 6250 711 1    1 1  
13 M. agalactiae PG2 III 1186 255 163 57 5296 653 1    1 1  
14 Mycoplasma arthritidis 158L3-1 III 1975 806 633 262 3233 327 1    1 1  
15 Mycoplasma bovis PG45 III 1256 289 191 65 6129 758 1    1 1  
16 Mycoplasma conjunctivae HRC/581 III 1786 718 728 175 3500 365 1    1   
17 M. crocodyli MP145 III 910 94 122 23 5564 337 2    1 1  
18 Mycoplasma hominis ATCC 23114 III 899 181 136 46 3818 413 1    1 1  
19 Mycoplasma hyopneumoniae 232 III 1485 938 1211 721 2858 745 1    1   
20 M. hyopneumoniae 7448 III 1463 938 1210 672 2852 719 1    1   
21 M. hyopneumoniaeIII 1460 933 1196 665 2881 710 1    1   
22 Mycoplasma hyorhinis HUB-1 III 913 125 340 41 4881 297 1    1   
23 M. mobile 163K III 618 77 171 26 5441 411 1    1   
24 Mycoplasma synoviae 53 III 986 136 96 60 4811 284 1    1   

 
25 M. fermentans JER III 2030 258 314 63 5371 236 1  1  1 1  
26 M. fermentans M64 III 2164 303 335 81 6439 315 1  1  1 1  
27 Mycoplasma penetrans HF-2 II 467 15 52 26 8579 492 1  1  1   
28 Ureaplasma parvum serovar 3 str. ATCC 27815 II 3098 447 946 122 1571 122 1  1  1   
29 Ureaplasma parvum serovar 3 str. ATCC 700970 II 3087 450 946 122 1592 127 1  1  1   
30 Ureaplasma urealyticum serovar 10 str. ATCC 33699 II 3671 369 1044 90 1652 77 1  1  1   

 
31 M. gallisepticum str. R(low) II 2031 616 925 498 4846 446  2 1  1   
32 Mycoplasma genitalium G37 II 1226 540 239 185 2439 812  1 1  1 1  
33 Mycoplasma pneumoniae M129 II 2340 2579 599 1200 968 679  1 1  1 1  
34 M. pulmonis UAB CTIP III 329 205 538 277 7228 2289  1 1  1   
35 Mycoplasma suis KI3806 II 109 55 306 49 6337 717  1 1  1   
36 M. suis str. Illinois II 123 71 355 61 6785 788  1 1  1   

 
37 M. haemofelis str. Langford 1 II 887 282 810 324 5889 3495   1  1   
Number Species Group Number of Arg codons in ORFs
 
Anticodon and number of tRNA genes
 
Gene 
CGU CGC CGA CGG AGA AGG ACG GCG TCG CCG TCT CCT tadA 
1a E. coli str. K-12 substr. MG1655 Outer 28 485 29 996 4871 7432 2845 1651 4   1 1 1 1 
1b Nitrosomonas europaea ATCC 19718 Outer 13 425 14 553 4584 10 153 5082 3473 1   1 1 1 1 
1c A. aeolicus VF5 Outer 727 601 268 367 9229 12588 1   1 1 1 1 
1d Streptomyces avermitilis MA 4680 Outer 19 076 93 823 7656 74019 2208 9827 1   1 1 1 1 
1e Synechococcus elongatus PCC 6301 Outer 8173 24 198 8010 787 24 448 1135 1   1 1 1 1 
1f S. aureus subsp. aureus Mu50 Outer 10 775 2603 3956 388 9321 1202 2   1 1  1 
1g Bacillus cereus ATCC 14579 Outer 20 003 6523 7745 1911 13 891 3604 4   1 1 1 1 
1h B. subtilis subsp. subtilis str. 168 Outer 9150 10 389 4957 7839 13 194 4700 4   1 1 1 1 
1i Listeria monocytogenes EGDe Outer 10 836 6301 5099 2578 5899 1102 2   1 1 1 1 
1j Oenococcus oeni PSU1 Outer 5934 2698 2965 2152 3951 1353 1   1 1 1 1 

 
A. laidlawii PG-8A IV 4075 872 747 61 8639 670 1    1  1 
Aster yellows witches'-broom phytoplasma AYWB IV 1183 710 305 42 2109 222 1    1  1 
Candidatus Phytoplasma australiense IV 1332 804 505 77 2484 358 1    1  1 
Candidatus Phytoplasma mali IV 1047 122 350 33 1972 168 1    1  1 
Onion yellows phytoplasma OY-M IV 1455 843 379 58 2457 237 1    1  1 

 
Mesoplasma florum L1 996 66 127 2 5444 190 1    1  1 
M. capricolum subsp. capricolum ATCC 27343 904 100 153 6 6115 184 1    1  1 
Mycoplasma leachii PG50 931 107 147 5 6154 175 1    1  1 
10 M. mycoides subsp. mycoides SC str. PG1 1061 95 167 10 7324 272 1    1  1 
11 M. mycoides subsp. capri LC str. 95010 1048 107 157 9 7275 252 1    1  1 

 
12 Mycoplasma agalactiae III 1349 258 153 56 6250 711 1    1 1  
13 M. agalactiae PG2 III 1186 255 163 57 5296 653 1    1 1  
14 Mycoplasma arthritidis 158L3-1 III 1975 806 633 262 3233 327 1    1 1  
15 Mycoplasma bovis PG45 III 1256 289 191 65 6129 758 1    1 1  
16 Mycoplasma conjunctivae HRC/581 III 1786 718 728 175 3500 365 1    1   
17 M. crocodyli MP145 III 910 94 122 23 5564 337 2    1 1  
18 Mycoplasma hominis ATCC 23114 III 899 181 136 46 3818 413 1    1 1  
19 Mycoplasma hyopneumoniae 232 III 1485 938 1211 721 2858 745 1    1   
20 M. hyopneumoniae 7448 III 1463 938 1210 672 2852 719 1    1   
21 M. hyopneumoniaeIII 1460 933 1196 665 2881 710 1    1   
22 Mycoplasma hyorhinis HUB-1 III 913 125 340 41 4881 297 1    1   
23 M. mobile 163K III 618 77 171 26 5441 411 1    1   
24 Mycoplasma synoviae 53 III 986 136 96 60 4811 284 1    1   

 
25 M. fermentans JER III 2030 258 314 63 5371 236 1  1  1 1  
26 M. fermentans M64 III 2164 303 335 81 6439 315 1  1  1 1  
27 Mycoplasma penetrans HF-2 II 467 15 52 26 8579 492 1  1  1   
28 Ureaplasma parvum serovar 3 str. ATCC 27815 II 3098 447 946 122 1571 122 1  1  1   
29 Ureaplasma parvum serovar 3 str. ATCC 700970 II 3087 450 946 122 1592 127 1  1  1   
30 Ureaplasma urealyticum serovar 10 str. ATCC 33699 II 3671 369 1044 90 1652 77 1  1  1   

 
31 M. gallisepticum str. R(low) II 2031 616 925 498 4846 446  2 1  1   
32 Mycoplasma genitalium G37 II 1226 540 239 185 2439 812  1 1  1 1  
33 Mycoplasma pneumoniae M129 II 2340 2579 599 1200 968 679  1 1  1 1  
34 M. pulmonis UAB CTIP III 329 205 538 277 7228 2289  1 1  1   
35 Mycoplasma suis KI3806 II 109 55 306 49 6337 717  1 1  1   
36 M. suis str. Illinois II 123 71 355 61 6785 788  1 1  1   

 
37 M. haemofelis str. Langford 1 II 887 282 810 324 5889 3495   1  1   

The frequencies of arginine codons in protein-encoding ORFs in each genome were obtained from Genbank. The information about the presence or absence of a given tRNAArg gene (the number corresponds to the number of genes encoding a tRNA with a given anticodon), as well as that about the tadA gene (always one when present), was obtained from the NCBI genome database, using BLASTN and BLASTP searches, respectively. The third base of the codon and the first wobble base of the anticodon are underlined. The accession numbers of the species, the subfamilies to which they belong and their hosts (in the cases of parasitic Mollicutes), their genome sizes, G + C% and references are provided in Supplementary Table S1. Species 2–6 correspond to Mollicutes of Group IV (Phytoplasmas), species 7–11 correspond to Mollicutes of Group I (Spiroplasmas), species 12–26 + 34 correspond to Mollicutes of Group III (Hominis) and finally species 27–33 and 35–37 correspond to Mollicutes of Group II (Pneumoniae). Descriptions of the different classes of Mollicutes are available (29–31). The CGG codon usages of Mollicute Groups IV (Phytoplasmas) and I (Spiroplasmas) are highlighted in bold letters.

Alignment of TadA amino acid sequences

As the amino acid sequences of TadA and CDA are difficult to distinguish by a simple BLAST homology search, we first aligned TadA and CDA. After identification of the genes encoding TadA, we created a second alignment of only the TadAs from the species listed in Supplementary Table S1, using Clustal X 2.0.12 (34) under the default conditions. The TadA enzyme catalyzes the deamination of wobble A34-containing tRNA, whereas the CDA enzyme catalyzes the deamination of free cytidine to produce uridine. As the TadAs are apparently derived from an ancestral CDA (35), the comparison allowed us to assess the conserved amino acids and to distinguish the ones that are ‘mechanistically’ common to all members of the deaminase superfamily (CDA and TadA) from those that are specific to TadA, such as those composing the tRNA-binding motif.

cDNA analyses of M. capricolum and B. subtilis tRNAArg

Bulk tRNA from B. subtilis strain 168 (wild-type) was obtained as described previously (36). Bulk tRNA from M. capricolum [American Type Culture Collection 27343 (kid)] at the late-log growth phase was obtained by the same procedure. Twenty micrograms of total tRNA from either M. capricolum or B. subtilis was treated with 4 U of Turbo DNase (Ambion), in the presence of 80 U of RNaseOUT (Invitrogen) for 30 min at 37°C. Following the suppliers’ protocols, the Turbo DNase was removed first, and then reverse transcription for first strand cDNA synthesis was performed, using 0.2 μg of total tRNA and 200 U of SuperScript III reverse transcriptase (Invitrogen). The primers for first strand cDNA synthesis of M. capricolum tRNAArg and B. subtilis tRNAArg were 5′-GGACT-CGAAC-CCCCA-ACCTT-TTGAT-CC-3′ (Mca-1st) and 5′-GGGAG-TCGAA-CCCCT-AACCT-TTTGA-TCC-3′ (Bsu-1st), respectively (black arrows in Figure 2A). In addition to the first strand cDNA synthesis primers, the following primers 5′-GCCCG-TAGAT-CAATT-GGATA-GATCG-CTTGA-3′ (Mca-2nd) and 5′-GCCCG-TAGCT-CAATG-GATAG-AGCGT-TTGA-3′ (Bsu-2nd) were used for further polymerase chain reaction (PCR) amplification of the cDNAs (gray arrows in Figure 2A). Aliquots (2 μl) of the aforementioned reaction mixtures, containing both types of primers, were incubated with 2.5 U of EX Taq DNA polymerase Hot Start version (TAKARA) in a 50 μl reaction solution, using a GeneAmp PCR System 9700 (Applied Biosystems, Life Technologies) thermal cycler. The final concentrations of primers and dNTPs were 400 nM and 200 μM (each), respectively. After pre-heating the PCR solution at 96°C for 4 min, 25 cycles of thermal denaturation/annealing/polymerization steps were performed (10 s at 98°C, 10 s at 50°C and 60 s at 72°C, respectively). The cDNA amplification products were analyzed by 4% agarose (MetaPhor™ Agarose, Lonza Co.) gel electrophoresis in Tris-borate-EDTA (TBE) buffer, using 100-bp size markers (New England Biolabs) to evaluate the lengths of the PCR transcripts. The recovered cDNAs were then cloned, using a TOPO-TA cloning kit for sequencing (Invitrogen). The plasmids were purified with a Montage Plasmid MiniprepHTS 96 kit (Millipore), using a Biomek 2000 (Beckman Coulter). A BigDye Terminator 3.1 kit (Applied Biosystems) was used for sequencing reactions, and a PRISM 3130xl DNA Autosequencer (Applied Biosystems) was used for sequencing. The obtained sequences were analyzed with the Geneious 5.6.5 software (Biomatters).

Figure 2.

Reverse transcriptase–PCR of tRNAArgICG from M. capricolum and B. subtilis. (A) Comparison of the nucleotide sequences of M. capricolum (Mca) and B. subtilis (Bsu) tRNAArgICG, obtained from (15). The cloverleaf structures are shown. I, 4, D, K, P, 7 and T represent inosine, 4-thio-uridine, dihydrouridine, 1-methylguanosine, pseudouridine, 7-methylguanosine and 5-methyluridine (ribosylthymine), respectively. Regions of primers for reverse transcription of the first strand (and first primers for PCR) are shown with black arrows. Regions of the second primers for PCR are shown with gray arrows. (B) Summary of sequences of cDNA clones for M. capricolum and B. subtilis tRNAArgICG. The DNA sequences of the cDNA clones, except for the PCR primer regions, are shown in brackets. The RNA sequences corresponding to the obtained DNA sequences are shown in parentheses. I (inosine) in the RNA sequence corresponds to G in the DNA sequence obtained by reverse transcription. (C) Agarose gel electrophoresis of reverse transcriptase–PCR products. Lane M: size marker (100-bp ladder, the position of 100 bp is shown with an arrow). Lanes 1–10: PCR products of various templates. Lane 1: reverse-transcribed McatRNAArgICG solution treated with DNase before reverse transcription. Lane 2: total McatRNA solution with DNase treatment. Lane 3: Reverse-transcribed McatRNAArgICG solution without DNase treatment before reverse transcription. Lane 4: total McatRNA solution without DNase treatment. Lane 6: reverse-transcribed BsutRNAArgICG solution with DNase treatment before reverse transcription. Lane 7: total BsutRNA solution with DNase treatment. Lane 8: reverse-transcribed BsutRNAArgICG solution without DNase treatment before reverse transcription. Lane 9: total BsutRNA solution without DNase treatment. Lanes 5 and 10: control (no RNA/DNA).

Figure 2.

Reverse transcriptase–PCR of tRNAArgICG from M. capricolum and B. subtilis. (A) Comparison of the nucleotide sequences of M. capricolum (Mca) and B. subtilis (Bsu) tRNAArgICG, obtained from (15). The cloverleaf structures are shown. I, 4, D, K, P, 7 and T represent inosine, 4-thio-uridine, dihydrouridine, 1-methylguanosine, pseudouridine, 7-methylguanosine and 5-methyluridine (ribosylthymine), respectively. Regions of primers for reverse transcription of the first strand (and first primers for PCR) are shown with black arrows. Regions of the second primers for PCR are shown with gray arrows. (B) Summary of sequences of cDNA clones for M. capricolum and B. subtilis tRNAArgICG. The DNA sequences of the cDNA clones, except for the PCR primer regions, are shown in brackets. The RNA sequences corresponding to the obtained DNA sequences are shown in parentheses. I (inosine) in the RNA sequence corresponds to G in the DNA sequence obtained by reverse transcription. (C) Agarose gel electrophoresis of reverse transcriptase–PCR products. Lane M: size marker (100-bp ladder, the position of 100 bp is shown with an arrow). Lanes 1–10: PCR products of various templates. Lane 1: reverse-transcribed McatRNAArgICG solution treated with DNase before reverse transcription. Lane 2: total McatRNA solution with DNase treatment. Lane 3: Reverse-transcribed McatRNAArgICG solution without DNase treatment before reverse transcription. Lane 4: total McatRNA solution without DNase treatment. Lane 6: reverse-transcribed BsutRNAArgICG solution with DNase treatment before reverse transcription. Lane 7: total BsutRNA solution with DNase treatment. Lane 8: reverse-transcribed BsutRNAArgICG solution without DNase treatment before reverse transcription. Lane 9: total BsutRNA solution without DNase treatment. Lanes 5 and 10: control (no RNA/DNA).

Comparison of the 3D structure of Staphylococcus TadA and the putative 3D structure of TadA from M. capricolum

A homology model of TadA from M. capricolum was created, based on its amino acid sequence and the crystal structure of TadA in complex with RNA from Staphylococcus aureus (PDB code: 2B3J) (37), using the SwissModel automatic modeling server from Expasy (http://swissmodel.expasy.org/). The hydrogen bonded contacts between TadA and tRNA were calculated by the LIGPLOT programs (38). Structure representations were prepared with the Pymol program (Schrödinger, LLC).

RESULTS

Decoding arginine codons in Mollicutes

Table 1 lists the frequencies of codon usage for each of the six arginine codons (4× CGN and 2× AGR, Figure 1), together with the corresponding usage of the tRNAArg isoacceptors, classified according to their anticodons (NCG and YCU) in 36 Mollicutes. This range of Mollicutes, all with reduced genome sizes (Supplementary Table S1), thoroughly covers the four major clades of the monophylogenetic phylum of this group of bacteria, i.e. Group I for Spiroplasma (items 7–11), Group II for Pneumoniae (items 27–37, except for 34 belonging to Group III), Group III for Hominis (items 34 + 12–26) and Group IV for Phytoplasma and Acholeplasma (items 2–6). For comparison, the situations in a few selected bacterial genomes outside the Mollicute family (items 1a–1j) are also shown. The table includes information about the presence or absence of a gene encoding a homolog of B. subtilis TadA (accession No. NP_387899.1), as query sequence. The E-values of the candidate protein sequences in the BLASTP search are >1e-13 (1013). No other Mollicute proteins showed E-values >1e-09 (109).

Inspection of Table 1 leads to the following conclusions:

  • In contrast to most bacteria, no gene encoding a tRNAArg harboring the same anticodon is redundant. This trend fits with the gene economization strategy used by Mollicutes, with their small genome sizes. The only exception is for tRNAArgGCG in Mycoplasma gallisepticum, which is encoded by two genes differing by only a single base at position 25 in the D-stem (C25 or A25), thus creating a mismatch G10-A25 in one of the two tRNAs (Supplementary Figure S1, and indicated in the Group II- Pneumoniae of Supplementary Figure S2). Mycoplasma crocodyli also has two genes encoding tRNAArgACG in its genome; however, these have exactly the same sequence (Supplementary Figure S1).

  • In contrast to most bacteria, none of the Mollicutes examined carries a gene encoding C34-containing tRNAArgCCG (row 13 in Table 1). This gene was obviously already lost in the genome of the common ancestor of Mollicutes. The lack of this gene is correlated with a drastic reduction, but not the complete elimination, of the CGG codons in mRNA (row 7 in Table 1), which are normally read by the missing tRNAArgCCG, especially in Spiroplasma (Group I, items 7–11) and Phytoplasma (Group IV, items 2–6, indicated in bold in Table 1). An analysis of the ORFs containing the few remaining Arg-CGG codons revealed that they are often used in genes encoding DNA and RNA modification enzymes, with only one codon in each gene, such as in Dam and DNA methylases, TruA, TruB, ThiI (indicated in bold in Supplementary Table S2) and even the tRNA-A34 deaminase TadA (indicated in bold and italics in the same Supplementary Table S2). The presence of a problematic Arg-CGG codon at the beginning (second position) of the mRNA corresponding to the tadA gene of Mycoplasma mycoides (Spiroplasma) is notable, and it suggests that the level of TadA deaminase expression in this organism may depend on the ability of the remaining single tRNAArg of the Arg-CGN decoding box to read this rare CGG codon.

  • All Mollicutes belonging to Groups III (Hominis, items 12–26 and 34) and II (Pneumoniae, items 27–33 and 35–37) lack the tadA gene, whereas in all Mollicutes of Groups IV (Phytoplasma, items 2–6) and I (Spiroplasma, items 7–11), the tadA gene is still present. The corollary is that A34, in the remaining single tRNAArgACG of the quartet decoding box, should normally be matured into I34 in all Groups I and IV Mollicutes, whereas in Groups II and III, the encoded wobble A34 will remain unmodified. Thus, the absence of the tRNA deaminase TadA in the Groups II and III Mollicutes obviously does not affect the viability of these cells, which have also adopted the strategy of preferring the arginine codon usage to mostly AGA of the duet decoding box (Table 1, compare the frequencies of codon usage in row 8 in Mollicutes—items 12–37, with those for bacteria—items 1a–1j). Groups I and IV of the Mollicutes (items 2–11) pose a more difficult problem because the cells have to read the four CGN codons with only a single I34-containing tRNAArgICG, which is normally unable to read CGG. Here, the dramatic reduction in CGG codon usage (indicated in bold in Table 1) and the preference for using the codon AGA of the duet decoding box instead is evident, especially in Spiroplasma (Group I, items 7–11). This AGA arginine codon will be read by the modified U*34-containing tRNAArgU*CU belonging to the duet decoding arginine box (see later in the text).

  • All Mollicutes of Group II (Pneumoniae), and Mycoplasma fermentans plus Mycoplasma pulmonis belonging to Group III-Hominis, have an additional tRNAArg harboring the anticodon UCG (row 12 in Table 1, items 25–37), thus alleviating the difficulty of reading both codons ending with A and G by A34- or I34-containing tRNAArg. Moreover, in most Pneumoniae with M. pulmonis (items 31–36), the A34-containing tRNAArgACG is replaced by the G34-containing tRNAArgGCG. Together with the U34-containing tRNAArgUCG, this allows all four CGN arginine codons to be easily read, in contrast to the Hominis clade (items 12–24), with only a single A34-containing tRNAArgACG. Only Mycoplasma haemofelis (Pneumoniae, item 37) remains with a single tRNAArg harboring the UCG anticodon, with the wobble U34 probably kept unmodified to enable the reading of all four CGN codons by ‘superwobbling (four-way wobbling)’ (22,39,40).

  • The only tRNAArg present in all Mollicutes analyzed is tRNAArgU*CU of the duet decoding Arg-box (Figure 1 and Table 1), where U* stands for 5-carboxymethylaminomethyluridine (cmnm5U), as demonstrated in M. capricolum tRNAArgU*CU (41). The modification of U34 in this tRNAArgU*CU is catalyzed by the multi-protein complex MnmE/MnmG present in almost all bacteria, including Mollicutes (42,43). Together with a second C34-containing tRNAArgCCU of the same duet decoding arginine box (only present in a few Mollicutes, Table 1), they translate the frequently used Arg codons AGA and AGG (AGR). From an evolutionary point of view, the existence of a second decoding box for arginine probably greatly facilitated the progressive shift in the decoding strategy within the other arginine decoding box.

In M. capricolum, the wobble A34 of a small fraction of tRNAArgACG is not deaminated

The nucleotide sequence of the naturally occurring tRNAArgICG of M. capricolum has been sequenced (41). However, no information was provided about the possibility that a small fraction of this tRNA population was not completely matured, especially at the wobble A34 position (Figure 2A). To clarify this point, we sequenced the anticodon region of cDNAArgICG, obtained after reverse transcription of the naturally occurring tRNAArgICG present in the bulk tRNA of M. capricolum (Figure 2A). As inosine behaves like G during transcription, we expected to obtain a G at the corresponding position in the cDNAArg. In contrast, if a fraction of the wobble A34 in the tRNA sample is not modified into I34, then some cDNAArg clones will now carry A at position 34, and the proportion of ‘A’-clones over ‘G’-clones will provide information about the degree of A34-to-I34 modification in the original M. capricolum tRNA sample. As shown in Figure 2B (upper part), among 86 cDNA clones analyzed, 5 clones (6%) have A at the anticodon first position, and the remaining 81 cDNA clones have G (94%). To confirm this result, several control experiments were performed. First, when the reverse-transcribed tRNA solution was used as the PCR template, only the cDNAs of M. capricolum tRNAArgICG were amplified (Figure 2C, lanes 1 and 3). Second, in the absence of reverse transcriptase, no cDNA products were PCR amplified (Figure 2C, lanes 2 and 4), confirming the absence of DNA contamination (even without DNase treatment). The results shown in Figure 2B were obtained using the cDNA shown in lane 1 of Figure 2C. The second series of control experiments involved performing the same analysis with bulk tRNA obtained from B. subtilis (Figure 2B and C). The tRNAArgICG sequence in this bacterium is similar to its M. capricolum homolog (Figure 2A) (15). The results from the analysis of 82 clones obtained from the cDNA (lane 6 in Figure 2C) indicated that, in contrast to the bulk tRNA from M. capricolum, no clone contained a cDNAArg with an A at the anticodon position 34, and only G34 was detected (100% - Figure 2B), corresponding to the fully matured I34 in the original sample of B. subtilis tRNAArgICG. These experiments demonstrated that in naturally occurring M. capricolum cells, a minor fraction of tRNAArg with unmodified wobble A34 (anticodon ACG) does exist and probably functions in translating all Arg-CGN codons (21,22).

The enzymatic deamination of A34 in tRNAArgACG in Mollicutes is probably not as efficient as in other bacteria

A small fraction of non-deaminated tRNAArgACG may also exist in other Mollicutes with genomes encoding tadA. This possibility could result from insufficient tadA gene expression and/or an abnormally inefficient (degenerate) deaminase. To examine this latter possibility, we compared the amino acid sequences of 10 TadA proteins encoded in the genomes of various bacteria (sequences 1a–1j in Figure 3), with those of 10 TadA proteins of the Mollicutes of Groups I (Spiroplasma) and IV (Phytoplasma), all encoding the tadA gene (sequences 2–11 in Figure 3). The list includes the well-characterized TadAs from E. coli (sequence 1a) (19,45), Aquifex aeolicus (sequence 1c) (44) and S. aureus (sequence 1f) (37). The amino acids with identical locations in the sequences are highlighted with black or colored backgrounds, and the systematic sequence deviations among these invariant or semi-invariant amino acids are boxed. The correspondence of these remarkable amino acids within the architecture of the TadA enzyme (indicated with black and colored backgrounds), and of the nucleotide position in tRNA (indicated in black), is depicted at the top of the figure. This information was deduced from the crystal structure of S. aureus TadA in complex with a chemically synthesized anticodon stem loop (16mer) bearing nebularine-34 as a substrate, in place of inosine-34 (Figure 4A) (37). For clarity, all other important elements of the anticodon branch in contact with the deaminase are not shown, as they are similar in the tRNAArgACG of both S. aureus and M. capricolum (Figure 4B).

Figure 3.

Amino acid sequence alignment of the genes encoding TadA. The TadA amino acid sequences from the species listed in Table 1 were retrieved from Genbank and aligned by Clustal X (34), under the default conditions. The amino acid numbers from E. coli are indicated above the alignment. The amino acid numbers from other species are indicated at the beginning and the end of the sections. The TadA-specific conserved amino acids are highlighted with a red or orange background. The conserved amino acids common among TadA and CDA are highlighted with a black or gray background. The conserved deaminase catalytic and zinc-binding sequences are highlighted in blue or light blue. Structurally and functionally important residues of TadA, inferred from the tertiary structures of the A. aeolicus and S. aureus TadAs (37,44), are indicated above the alignment. The terms ‘nnb’ and ‘stack’ mean non-bonded (hydrophobic) contacts and stacking interactions, respectively. The red boxes in Mollicutes (sequences 2–11) indicate the variations from other bacterial TadAs (sequences 1a–1j). Conserved amino acids involved in tRNA interactions, which are depicted by stick models in Figure 4, are indicated by arrows below the sequences.

Figure 3.

Amino acid sequence alignment of the genes encoding TadA. The TadA amino acid sequences from the species listed in Table 1 were retrieved from Genbank and aligned by Clustal X (34), under the default conditions. The amino acid numbers from E. coli are indicated above the alignment. The amino acid numbers from other species are indicated at the beginning and the end of the sections. The TadA-specific conserved amino acids are highlighted with a red or orange background. The conserved amino acids common among TadA and CDA are highlighted with a black or gray background. The conserved deaminase catalytic and zinc-binding sequences are highlighted in blue or light blue. Structurally and functionally important residues of TadA, inferred from the tertiary structures of the A. aeolicus and S. aureus TadAs (37,44), are indicated above the alignment. The terms ‘nnb’ and ‘stack’ mean non-bonded (hydrophobic) contacts and stacking interactions, respectively. The red boxes in Mollicutes (sequences 2–11) indicate the variations from other bacterial TadAs (sequences 1a–1j). Conserved amino acids involved in tRNA interactions, which are depicted by stick models in Figure 4, are indicated by arrows below the sequences.

Figure 4.

(A) Homology model of M. capricolum TadA, superposed on S. aureus TadA complexed with tRNAArgACG. Both TadA proteins are represented by ribbon models, colored green for M. capricolum and gray for S. aureus. The S. aureus tRNA is depicted by a stick model. Conserved amino acids involved in tRNA interactions, which are indicated by arrows in Figure 3, are shown in stick models. The amino acids specific to Mycoplasma, indicated in the red boxes in Figure 3, are circled. (B) Sequences of the anticodon branches of the tRNAArgACG from S. aureus and M. capricolum (15).

Figure 4.

(A) Homology model of M. capricolum TadA, superposed on S. aureus TadA complexed with tRNAArgACG. Both TadA proteins are represented by ribbon models, colored green for M. capricolum and gray for S. aureus. The S. aureus tRNA is depicted by a stick model. Conserved amino acids involved in tRNA interactions, which are indicated by arrows in Figure 3, are shown in stick models. The amino acids specific to Mycoplasma, indicated in the red boxes in Figure 3, are circled. (B) Sequences of the anticodon branches of the tRNAArgACG from S. aureus and M. capricolum (15).

Among the important invariant amino acids to be considered in the A-to-I deaminase TadA, some are also common within the C-to-U deaminase CDA (35), including the AE motif of the deaminase catalytic center, and PCxxC of the zinc-binding motif (Figure 3). In addition, the TadA proteins from Mollicutes (sequences 2–11) share several other identity elements in common with some selected bacterial TadA proteins (red or orange background), i.e. the EVPV and TLE motifs of the TadA-structural core, and several amino acids at conserved positions, such as His57, Lys111 and Phe149 (E. coli numbers), which is precisely the region in contact with the tRNA anticodon loop (37). More interesting are the systematic sequence deviations and the absence of certain amino acids (gaps, indicated by dashes) in the TadA sequences of Mollicutes (sequences 2–11, positions in red boxes), as compared with the TadA sequences of other bacteria.

To better visualize the implications of these different amino acids within the active site architecture of the deaminase, the sequence of TadA from M. capricolum (item 8 in Figure 3) was superposed on the 3D architecture of TadA from S. aureus (item 1f in Figure 3) in complex with a 16 nt mini substrate. As shown in Figure 4A, it is now clear that Asn71 and Lys95 in M. capricolum (indicated in green and encircled in red) replaced Arg70 and Arg94 in S. aureus (indicated in blue). Therefore, the ribose phosphate backbone of nucleotides G37 and G36 in the anticodon loop, which H-bond with these amino acids in the case of the S. aureus TadA–RNA complex, may not be well fixed, or exist in a slightly different configuration in the case of the putative complex of the same RNA with M. capricolum TadA. Moreover, in the vicinity of the essential zinc motif and nebularine-34, and thus within the catalytic center of the deaminase, Ser105 (indicated in green and encircled in red) in M. capricolum replaces the important Asp104 in S. aureus (indicated in blue), which normally H-bonds with the ribose of U at position 33, adjacent to nucleoside 34 of the anticodon loop. The absence of an interaction with the ribose of U33, together with the absence of H-bonding because of the amino acid replacements at positions 70/71 and 94/95 discussed earlier in the text, may affect the dynamics (flexibility/adaptability) of the entire anticodon branch within the active site of the deaminase. Consequently, this may limit the accessibility of the amine target of the wobble A34 for deamination, which is catalyzed by the neighboring zinc atom (in the brown background) around His-53/54.

A global inspection of the 3D architecture of S. aureus TadA in complex with its RNA mini substrate (37) revealed that the A31-U39 base pair at the beginning of the anticodon stem does not interact with any amino acids of the deaminase. Only the C32-A38 pair interacts with Lys106 and Asn123 (Supplementary Figure S3). However, Lys106 (Lys107 in M. capricolum) is conserved in all TadA proteins examined (Figure 3), whereas Asn123 (Asn122 in M. capricolum) is replaced with different amino acids among the various Mollicutes; therefore, it may not be important for the catalytic function of the deaminase. It is likely that only the mutations in the tadA* gene corresponding to the catalytic core of the deaminase, as discussed earlier in the text, contribute to the modulation of the A34-deamination efficiency and ultimately play a role in decoding all four arginine CGN codons.

DISCUSSION

During protein synthesis, tRNAs bearing the complementary anticodons read mRNA codons. However, because different types of relaxed base pairing are allowed between the often modified ‘wobble’ base at position 34 of the anticodon and the last nucleotide of the codon, some tRNA species can read two, three or even four synonymous codons. Therefore, the number of isoacceptor tRNAs with distinct anticodons needed to read all synonymous codons of a given amino acid is usually lower than the number of codons specifying that particular amino acid in the genetic code. Various organisms apply different rules to adapt their tRNA sets, attesting to the existence of distinct cellular strategies for reading the almost universal genetic code (4). Here, we focused on reading the quartet arginine codons in the quickly evolving Mollicutes with reduced genomes (0.6–1.5 Mb, Supplementary Table S1).

Reading arginine codons in M. capricolum

In M. capricolum, only two kinds of tRNAArg exist for reading the six arginine codons (four in the quartet and two in the duet family boxes). One tRNA contains an anticodon with a wobble inosine (tRNAArgICG) and the other contains an anticodon with a modified wobble uridine (cmnm5U34, tRNAArgU*CU) (41). Because of the wobble inosine-34, tRNAArgICG was expected to read only the three arginine codons ending with U, C or A of the quartet family box (8,9,16). Paradoxically, a tRNAArg harboring the anticodon CCG, needed to read the remaining fourth arginine codon CGG, as found in the majority of other bacteria (Table 1, items 1a–1j), was absent (41). Here, we demonstrated that a small fraction of the cellular A34-containing tRNAArgACG precursor is not enzymatically deaminated in M. capricolum. The key point of our report is the correlation with a few characteristic amino acid variants that exist within the active sites of the TadA’s of M. capricolum and other Mollicutes, as compared with other well characterized bacterial TadA’s considered as references. We hypothesize that these point mutations are needed for reducing the enzymatic activity of the tRNA:A34 deamination (degenerate TadA*), allowing the accumulation of a small but sufficient amount of the non-deaminated A34-containing tRNAArgACG, which is competent for reading all four arginine codons of the quartet CGN decoding box (Step 1 in Figure 5A). To use a term that was first applied in the case of unmodified U34-containing tRNAs, this decoding strategy would correspond to a sort of ‘superwobbling’, facilitating the translation of synonymous codons with a reduced set of tRNAs (40). Therefore, the useless C34-containing tRNAArgCCG can be lost (Step 2 in Figure 5A). This process was probably facilitated by limiting the usage of the problematic CGG codon (Step 2). Indeed, among 1163 CGN codons, only 6 such rare CGG codons, each in different mRNAs, were detected in the ORFs of M. capricolum.

Figure 5.

Hypothetical scenario for the evolution of the CGN decoding system for arginine in Mollicutes. (A) Schematic view of the five sequential events leading from a ‘classical bacterial’ arginine decoding strategy involving two tRNAArg, one with a wobble inosine-34 and the other with a wobble C34, to another Arg decoding strategy involving only one tRNAArg with an unmodified wobble U34. In M. capricolum, this latter situation exists in many other quartet decoding boxes (Leu, Val, Ser, Pro, Ala and Gly), as well as in most mitochondria of eukarya. (B) The same events as in A, but depicted within the Mollicute evolutionary framework. Because of the degenerated TadA*, partial A-to-I deamination occurs at the first anticodon position of tRNAArgACG (Step 1), generating a situation where a mixture of both deaminated (in black) and non-deaminated tRNAArg (in red) molecules co-exist in the cell. In addition to the three synonymous arginine codons normally decoded by I34-containing tRNAArgICG, tRNAArgACG also decodes the CGG codon, but probably inefficiently (see text). The gene encoding tRNAArgCCG could then be lost (Step 2), along with the gene encoding tad* (Step 3). Further reorganization of the tRNA repertoire could occur by gaining an extra U34-containing tRNAArgUCG (Step 4). The original A34-containing tRNAArgACG can undergo a mutation in its anticodon to generate a G34-containing tRNAArgGCG (Step 5a), or simply be lost (Step 5b). The species of Mollicutes in which these different events occurred are indicated by numbers, corresponding to the organisms listed in Table 1. The phylogenetic relationships among the different Mollicutes were adapted from the literature (29–31).

Figure 5.

Hypothetical scenario for the evolution of the CGN decoding system for arginine in Mollicutes. (A) Schematic view of the five sequential events leading from a ‘classical bacterial’ arginine decoding strategy involving two tRNAArg, one with a wobble inosine-34 and the other with a wobble C34, to another Arg decoding strategy involving only one tRNAArg with an unmodified wobble U34. In M. capricolum, this latter situation exists in many other quartet decoding boxes (Leu, Val, Ser, Pro, Ala and Gly), as well as in most mitochondria of eukarya. (B) The same events as in A, but depicted within the Mollicute evolutionary framework. Because of the degenerated TadA*, partial A-to-I deamination occurs at the first anticodon position of tRNAArgACG (Step 1), generating a situation where a mixture of both deaminated (in black) and non-deaminated tRNAArg (in red) molecules co-exist in the cell. In addition to the three synonymous arginine codons normally decoded by I34-containing tRNAArgICG, tRNAArgACG also decodes the CGG codon, but probably inefficiently (see text). The gene encoding tRNAArgCCG could then be lost (Step 2), along with the gene encoding tad* (Step 3). Further reorganization of the tRNA repertoire could occur by gaining an extra U34-containing tRNAArgUCG (Step 4). The original A34-containing tRNAArgACG can undergo a mutation in its anticodon to generate a G34-containing tRNAArgGCG (Step 5a), or simply be lost (Step 5b). The species of Mollicutes in which these different events occurred are indicated by numbers, corresponding to the organisms listed in Table 1. The phylogenetic relationships among the different Mollicutes were adapted from the literature (29–31).

Reading arginine codons in other Mollicutes (Spiroplasma and Acholeplasma/Phytoplasma)

Combining our comparative genome analysis with information about the evolutionary origin of Mollicutes (29,30) revealed that the decoding strategy for M. capricolum is still in use in all Mollicutes of Groups I (Spiroplasmas, items 7–11) and IV (Acholeplasmas/Phytoplasmas, items 2–6), as shown in Table 1 and the green background in Figure 5B. Obviously, the two events (Steps 1 and 2 described earlier in the text) occurred early in evolution, almost at the root of the monophyletic Mollicute tree. These Mollicutes currently have the same original set of two genes: one gene encoding an A34-containing tRNAArgACG for reading a minimum number of CGN codons, and a second one harboring a U*CU anticodon (tRNAArgU*CU) for reading the other most frequently used arginine codons AGA and AGG; only the original TadA is now the mutant TadA*.

Further stepwise evolution of the decoding strategy in Hominis and Pneumoniae

To become less dependent on the activity of the hypothetical degenerate TadA*, a subset of the newly evolved Mollicutes lost the degenerated tadA* gene (Step 3). This new evolutionary event occurred before the divergence into Groups III (Hominis) and II (Pneumoniae), items 12–37—all indicated with a yellow background in Figure 5A and B. Interestingly, the usage of the earlier problematic and rare CGG codon in these newly evolved Mollicutes became more frequent again, confirming that a Mollicute lacking the tadA gene and encoding an unmodified wobble A34-containing tRNAArgACG (items 12–24 in Figure 5) is perfectly viable because of its ability to read all four Arg-CGN codons.

In a subset (items 25–37) of Groups II and III (Hominis/Pneumoniae), the reading of the four synonymous arginine CGN codons was probably improved by gaining a new U34-containing tRNAArgUCG (Step 4 in Figure 5A and B). This U34-containing tRNAArgUCG could have originated in diverse manners. It may have arisen from the duplication of the gene encoding A34-containing tRNAArgACG, followed by a few mutations, including the wobble A34-to-U34. It may also have resulted from duplication and subsequent recruitment/mutation of a gene encoding a tRNA possibly from the other duet Arg-AGR coding box, or belonging to another amino acid coding box. The mutations in the tRNAArgACG substrate itself may modulate the efficiency of A34-deamination and ultimately play a role in decoding all four arginine CGN codons (Supplementary Figure S2). Unfortunately, a phylogenetic analysis of all of the tRNA genes retrieved from the 36 Mollicutes examined did not allow us to confidently determine which one of these two alternatives prevailed because of the low-bootstrap values in constructing such phylogenetic trees with relatively short tRNAs, including many conserved and semi-conserved nucleotides and invariant regions under strong selective pressure (46,47).

Among the few species (items 25–30) of Groups II and III (Hominis/Pneumoniae), the four arginine CGN codons are read by a tRNAArg pair, one with a non-deaminated wobble A34 and the other with a wobble U34 (Figure 5, yellow background; U34 is probably not modified, see later in the text). This decoding strategy is also the one used presently for reading the four CGN codons as arginine in a few other non-Mollicute bacteria, such as Clostridium perfringens, Chlamydia trachomatis, Geobacter metalloreducens and Haloplasma contractile, the four CUN codons as leucine in Lactococcus lactis, and as mentioned in the ‘Introduction’ section, for reading the four ACN codons as threonine in M. capricolum (11,41).

Other species of Group II-Pneumoniae (items #31-36, including M. pulmonis) continued to evolve by using a slightly different decoding strategy (Step 5a). In these species, the CGN codons are now read by another type of tRNAArg set, one with a wobble U34 and the other one with G34 (Figure 5, yellow background). Because of the close sequence homology between the new G34-containing tRNAArg and the A34-containing tRNAArg in the other Pneumoniae (data not shown), this new G34-containing tRNAArg is believed to have arisen via a simple A34-to-G34 mutation and additional base mutations within the rest of the tRNAArgACG structure. This last decoding strategy is most frequently used in bacteria for decoding the sense codons of quartet synonymous codon boxes, at least in bacteria with moderate or low G + C content in their ORFs, as in Borrelia burgdorferi, Campylobacter jejuni, Helicobacter pylori, Treponema palladium, Thermotoga maritima and a few others (11).

Finally, one Mycoplasma in Group II, M. haemofelis (item 37 in Table 1), lost the ancient A34-(or G34)-containing tRNAArgACG (Step 5b); thus, it has only one U34-containing tRNAArgUCG for reading the four synonymous Arg-CGN codons. This situation corresponds to the minimal set of tRNAArg that a Mollicute can use to continue decoding all CGN codons as arginine, with no need for the enzyme TadA and probably with better efficiency than that with a single A34-containing tRNAArgACG. This decoding strategy was also used in other quartet decoding boxes corresponding to Leu, Val, Ser, Pro, Ala and Gly in M. capricolum, M. mycoides and the mitochondria of S. cerevisiae and mammals (24,39,41); reviewed in (5,48). The sequences of the corresponding tRNAs revealed the presence of an unmodified U34 in their anticodons (15).

Analogy to a similar situation in the chloroplasts of higher plants

Gene knockout experiments in the plastids of the moss Physcomitrella patens demonstrated the dispensability of the C34-containing tRNAArgCCG, whereas the chloroplastic A34-containing tRNAArgACG and the chloroplastic TadA enzyme are encoded in the plastid and nuclear genomes, respectively (49). This situation corresponds to that of the Groups I (Spiroplasmas) and IV (Acholeplasmas/Phytoplasmas) Mollicutes (Table 1), which also lack C34-tRNAArgCCG (see earlier in the text). On the other hand, the chloroplasts of A. thaliana lack C34-tRNAArgCCG, and only two kinds of tRNAArg are encoded on the plastid genome: one with the anticodon ACG and the other one with the anticodon UCU. In this species, the inhibition of the chloroplastic tadA gene expression by RNAi (not the cytoplasmic Tad2/Tad3) allows plant survival, and only the chloroplast translation and photosynthesis activities were hindered (17,18). This situation corresponds to the one described earlier in the text for the Mollicutes of Hominis Group III. By analogy with our results in the case of M. capricolum, we anticipate that in the chloroplasts of wild-type A. thaliana, and probably in other plant plastids, a fraction of the chloroplastic A34-containing tRNAArgACG also remains naturally unmodified, allowing superwobbling for decoding all CGN codons, including the rare Arg-CGG (16).

Evolutionary scenario of the Mollicute decoding process

The scenario proposed in Figure 5 illustrates the evolvability of the decoding process. However, changing the decoding strategy during cellular evolution depends on a series of sequentially ordered events, such as point mutations in modification enzymes (probably also in the tRNA), gene loss, gene duplication and possibly the recruitment of a gene encoding a tRNA from another decoding box. The driving forces of this evolutionary process are almost certainly the efficacy and accuracy of translation. The sequence of events we have proposed, to explain the elimination of the essential deaminase TadA in Mollicutes, also applies to the essential tRNA–lysidine synthase TilS, responsible for the k2C modification at the wobble position 34 of tRNAIleCAU. Indeed, although it is encoded in the genomes of 35 Mollicutes, the tilS gene is notably absent in Mycoplasma mobile, with a concomitant change in the sequence of the minor tRNAIle that decodes AUA codons, from a CAU to a UAU anticodon (50,51). A similar cellular strategy has been experimentally verified in the case of B. subtilis, after the deletion of its essential tilS (52).

Finally, the idea of first reducing the activity of an enzyme (here, TadA or TilS) by point mutations, before its complete loss later in evolution, is reminiscent of recent work describing the progressive degeneration of aminoacyl-tRNA synthetases in M. mobile and other closely related Mycoplasmas of Group III-Hominis (53,54). In these cases, the degenerated aminoacyl-tRNA synthetases, while still performing the normal aminoacylation function, occasionally misacylate the cognate tRNA with a non-cognate amino acid. This allows the generation of a small number of cellular proteins with an incorrect amino acid substitution (statistical mutations). It was proposed that such misacylation reactions, if they are not too frequent, would provide an advantage to the Mycoplasma, which are indeed evolving faster than other extant bacteria by producing a more homogeneous proteome (55).

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online: Supplementary Tables 1–2 and Supplementary Figures 1–3.

FUNDING

Naito Foundation [2011-164 to Y.B.]; Daiichi-Sankyo Foundation of Life Science [12-039 to Y.B.]; X-ray Free Electron Laser Priority Strategy Program, from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan (to Y.B.). H.G. holds the position of Emeritus Scientist at the CNRS in Gif-sur-Yvette, France, in the laboratory of Dominique Fourmy and Satoko Yoshizawa. Funding for open access charge: Naito Foundation [2011-164 to Y.B.].

Conflict of interest statement. None declared.

ACKNOWLEDGEMENTS

The authors thank the Eminent Professors Syozo Osawa (Nagoya University) and Kimitsuna Watanabe (University of Tokyo) for valuable discussions about evolving genetic codes. The authors thank Dr Kazuyuki Takai (Ehime University) for discussions about the A34 decoding system. The authors also thank Pascal Sirand-Pugnet (INRA, Bordeaux, France) for advice on the phylogeny of Mollicutes, and Juan Alfonzo (Ohio State University, Columbus, USA) and Valérie de Crécy-Lagard (University of Florida, Gainsville, USA) for advice and manuscript editing. The authors thank Dr Chisato Ushida and Dr Akira Wada for bacterial sample preparation.

REFERENCES

1
Osawa
S
Jukes
TH
Watanabe
K
Muto
A
Recent evidence for evolution of the genetic code
Microbiol. Rev.
 , 
1992
, vol. 
56
 (pg. 
229
-
264
)
2
Agris
PF
Vendeix
FA
Graham
WD
tRNA's wobble decoding of the genome: 40 years of modification
J. Mol. Biol.
 , 
2007
, vol. 
366
 (pg. 
1
-
13
)
3
Watanabe
K
Suzuki
T
Universal genetic code and its natural variations
Encyclopedia of Life Sciences
 , 
2008
Chichester
John Wiley & Sons, Ltd.
4
Grosjean
H
de Crécy-Lagard
V
Marck
C
Deciphering synonymous codons in the three domains of life: co-evolution with specific tRNA modification enzymes
FEBS Lett.
 , 
2010
, vol. 
584
 (pg. 
252
-
264
)
5
Watanabe
K
Yokobori
S
tRNA Modification and genetic code variations in animal mitochondria
J. Nucleic Acids
 , 
2011
, vol. 
2011
 pg. 
623095
 
6
Crick
FH
Codon–anticodon pairing: the wobble hypothesis
J. Mol. Biol.
 , 
1966
, vol. 
19
 (pg. 
548
-
555
)
7
Lim
VI
Analysis of action of the wobble adenine on codon reading within the ribosome
J. Mol. Biol.
 , 
1995
, vol. 
252
 (pg. 
277
-
282
)
8
Curran
JF
Decoding with the A:I wobble pair is inefficient
Nucleic Acids Res.
 , 
1995
, vol. 
23
 (pg. 
683
-
688
)
9
Murphy
FV
Ramakrishnan
V
Structure of a purine-purine wobble base pair in the decoding center of the ribosome
Nat. Struct. Mol. Biol.
 , 
2004
, vol. 
11
 (pg. 
1251
-
1252
)
10
Chan
PP
Lowe
TM
GtRNAdb: a database of transfer RNA genes detected in genomic sequence
Nucleic Acids Res.
 , 
2009
, vol. 
37
 (pg. 
D93
-
D97
)
11
Marck
C
Grosjean
H
tRNomics: analysis of tRNA genes from 50 genomes of Eukarya, Archaea, and Bacteria reveals anticodon-sparing strategies and domain-specific features
RNA
 , 
2002
, vol. 
8
 (pg. 
1189
-
1232
)
12
Auxilien
S
Crain
PF
Trewyn
RW
Grosjean
H
Mechanism, specificity and general properties of the yeast enzyme catalysing the formation of inosine 34 in the anticodon of transfer RNA
J. Mol. Biol.
 , 
1996
, vol. 
262
 (pg. 
437
-
458
)
13
Gerber
AP
Keller
W
An adenosine deaminase that generates inosine at the wobble position of tRNAs
Science
 , 
1999
, vol. 
286
 (pg. 
1146
-
1149
)
14
Maas
S
Gerber
AP
Rich
A
Identification and characterization of a human tRNA-specific adenosine deaminase related to the ADAR family of pre-mRNA editing enzymes
Proc. Natl Acad. Sci. USA
 , 
1999
, vol. 
96
 (pg. 
8895
-
8900
)
15
Jühling
F
Mörl
M
Hartmann
RK
Sprinzl
M
Stadler
PF
Pütz
J
tRNAdb 2009: compilation of tRNA sequences and tRNA genes
Nucleic Acids Res.
 , 
2009
, vol. 
37
 (pg. 
D159
-
D162
)
16
Aldinger
CA
Leisinger
AK
Gaston
KW
Limbach
PA
Igloi
GL
The absence of A-to-I editing in the anticodon of plant cytoplasmic tRNAArg(ACG) demands a relaxation of the wobble decoding rules
RNA Biol.
 , 
2012
, vol. 
9
 (pg. 
1239
-
1246
)
17
Karcher
D
Bock
R
Identification of the chloroplast adenosine-to-inosine tRNA editing enzyme
RNA
 , 
2009
, vol. 
15
 (pg. 
1251
-
1257
)
18
Delannoy
E
Le Ret
M
Faivre-Nitschke
E
Estavillo
GM
Bergdoll
M
Taylor
NL
Pogson
BJ
Small
I
Imbault
P
Gualberto
JM
Arabidopsis tRNA adenosine deaminase arginine edits the wobble nucleotide of chloroplast tRNAArg(ACG) and is essential for efficient chloroplast translation
Plant Cell
 , 
2009
, vol. 
21
 (pg. 
2058
-
2071
)
19
Wolf
J
Gerber
AP
Keller
W
tadA, an essential tRNA-specific adenosine deaminase from Escherichia coli
EMBO J.
 , 
2002
, vol. 
21
 (pg. 
3841
-
3851
)
20
Chen
P
Qian
Q
Zhang
S
Isaksson
LA
Björk
GR
A cytosolic tRNA with an unmodified adenosine in the wobble position reads a codon ending with the non-complementary nucleoside cytidine
J. Mol. Biol.
 , 
2002
, vol. 
317
 (pg. 
481
-
492
)
21
Boren
T
Elias
P
Samuelsson
T
Claesson
C
Barciszewska
M
Gehrke
CW
Kuo
KC
Lustig
F
Undiscriminating codon reading with adenosine in the wobble position
J. Mol. Biol.
 , 
1993
, vol. 
230
 (pg. 
739
-
749
)
22
Inagaki
Y
Kojima
A
Bessho
Y
Hori
H
Ohama
T
Osawa
S
Translation of synonymous codons in family boxes by Mycoplasma capricolum tRNAs with unmodified uridine or adenosine at the first anticodon position
J. Mol. Biol.
 , 
1995
, vol. 
251
 (pg. 
486
-
492
)
23
Andachi
Y
Yamao
F
Iwami
M
Muto
A
Osawa
S
Occurrence of unmodified adenine and uracil at the first position of anticodon in threonine tRNAs in Mycoplasma capricolum
Proc. Natl Acad. Sci. USA
 , 
1987
, vol. 
84
 (pg. 
7398
-
7402
)
24
Sibler
AP
Dirheimer
G
Martin
RP
Codon reading patterns in Saccharomyces cerevisiae mitochondria based on sequences of mitochondrial tRNAs
FEBS Lett.
 , 
1986
, vol. 
194
 (pg. 
131
-
138
)
25
Watanabe
Y
Tsurui
H
Ueda
T
Furusihima-Shimogawara
R
Takamiya
S
Kita
K
Nishikawa
K
Watanabe
K
Primary sequence of mitochondrial tRNAArg of a nematode Ascaris suum: occurrence of unmodified adenosine at the first position of the anticodon
Biochim. Biophys. Acta
 , 
1997
, vol. 
1350
 (pg. 
119
-
122
)
26
Rocha
EP
Codon usage bias from tRNA's point of view: redundancy, specialization, and efficient decoding for translation optimization
Genome Res.
 , 
2004
, vol. 
14
 (pg. 
2279
-
2286
)
27
Fredrick
K
Ibba
M
How the sequence of a gene can tune its translation
Cell
 , 
2010
, vol. 
141
 (pg. 
227
-
229
)
28
Salinas
T
Duby
F
Larosa
V
Coosemans
N
Bonnefoy
N
Motte
P
Marechal-Drouard
L
Remacle
C
Co-evolution of mitochondrial tRNA import and codon usage determines translational efficiency in the green alga Chlamydomonas
PLoS Genet.
 , 
2012
, vol. 
8
 pg. 
e1002946
 
29
Barre
A
de Daruvar
A
Blanchard
A
MolliGen, a database dedicated to the comparative genomics of Mollicutes
Nucleic Acids Res.
 , 
2004
, vol. 
32
 (pg. 
D307
-
D310
)
30
Sirand-Pugnet
P
Citti
C
Barre
A
Blanchard
A
Evolution of mollicutes: down a bumpy road with twists and turns
Res. Microbiol.
 , 
2007
, vol. 
158
 (pg. 
754
-
766
)
31
Maniloff
J
Razin
S
Herrmann
R
Molecular Biology and Pathogenicity of Mycoplasmas
 , 
2002
New York, NY
Kluwer Academic/Plenum Publisher
(pg. 
31
-
43
)
32
Abe
T
Ikemura
T
Ohara
Y
Uehara
H
Kinouchi
M
Kanaya
S
Yamada
Y
Muto
A
Inokuchi
H
tRNADB-CE: tRNA gene database curated manually by experts
Nucleic Acids Res.
 , 
2009
, vol. 
37
 (pg. 
D163
-
D168
)
33
Tanaka
R
Andachi
Y
Muto
A
Evolution of tRNAs and tRNA genes in Acholeplasma laidlawii
Nucleic Acids Res.
 , 
1991
, vol. 
19
 (pg. 
6787
-
6792
)
34
Thompson
JD
Plewniak
F
Poch
O
A comprehensive comparison of multiple sequence alignment programs
Nucleic Acids Res.
 , 
1999
, vol. 
27
 (pg. 
2682
-
2690
)
35
Gerber
AP
Keller
W
RNA editing by base deamination: more enzymes, more targets, new mysteries
Trends Biochem. Sci.
 , 
2001
, vol. 
26
 (pg. 
376
-
384
)
36
Urbonavičius
J
Brochier-Armanet
C
Skouloubris
S
Myllykallio
H
Grosjean
H
In vitro detection of the enzymatic activity of folate-dependent tRNA (Uracil-54,-C5)-methyltransferase: evolutionary implications
Methods Enzymol.
 , 
2007
, vol. 
425
 (pg. 
103
-
119
)
37
Losey
HC
Ruthenburg
AJ
Verdine
GL
Crystal structure of Staphylococcus aureus tRNA adenosine deaminase TadA in complex with RNA
Nat. Struct. Mol. Biol.
 , 
2006
, vol. 
13
 (pg. 
153
-
159
)
38
Wallace
AC
Laskowski
RA
Thornton
JM
LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions
Protein Eng.
 , 
1995
, vol. 
8
 (pg. 
127
-
134
)
39
Claesson
C
Samuelsson
T
Lustig
F
Borén
T
Codon reading properties of an unmodified transfer RNA
FEBS Lett.
 , 
1990
, vol. 
273
 (pg. 
173
-
176
)
40
Rogalski
M
Karcher
D
Bock
R
Superwobbling facilitates translation with reduced tRNA sets
Nat. Struct. Mol. Biol.
 , 
2008
, vol. 
15
 (pg. 
192
-
198
)
41
Andachi
Y
Yamao
F
Muto
A
Osawa
S
Codon recognition patterns as deduced from sequences of the complete set of transfer RNA species in Mycoplasma capricolum
Resemblance to mitochondria. J. Mol. Biol.
 , 
1989
, vol. 
209
 (pg. 
37
-
54
)
42
de Crécy-Lagard
V
Marck
C
Brochier-Armanet
C
Grosjean
H
Comparative RNomics and modomics in Mollicutes: prediction of gene function and evolutionary implications
IUBMB Life
 , 
2007
, vol. 
59
 (pg. 
634
-
658
)
43
Bessho
Y
Yokoyama
S
Grosjean
H
DNA and RNA Modification Enzymes
 , 
2009
Austin
Landes Bioscience
(pg. 
406
-
422
)
44
Kuratani
M
Ishii
R
Bessho
Y
Fukunaga
R
Sengoku
T
Shirouzu
M
Sekine
S
Yokoyama
S
Crystal structure of tRNA adenosine deaminase (TadA) from Aquifex aeolicus
J. Biol. Chem.
 , 
2005
, vol. 
280
 (pg. 
16002
-
16008
)
45
Kim
J
Malashkevich
V
Roday
S
Lisbin
M
Schramm
VL
Almo
SC
Structural and kinetic characterization of Escherichia coli TadA, the wobble-specific tRNA deaminase
Biochemistry
 , 
2006
, vol. 
45
 (pg. 
6407
-
6416
)
46
Cedergren
RJ
Sankoff
D
LaRue
B
Grosjean
H
The evolving tRNA molecule
CRC Crit. Rev. Biochem.
 , 
1981
, vol. 
11
 (pg. 
35
-
104
)
47
Widmann
J
Harris
JK
Lozupone
C
Wolfson
A
Knight
R
Stable tRNA-based phylogenies using only 76 nucleotides
RNA
 , 
2010
, vol. 
16
 (pg. 
1469
-
1477
)
48
Suzuki
T
Nagao
A
Suzuki
T
Human mitochondrial tRNAs: biogenesis, function, structural aspects, and diseases
Annu. Rev. Genet.
 , 
2011
, vol. 
45
 (pg. 
299
-
329
)
49
Nakamura
T
Sugiura
C
Kobayashi
Y
Sugita
M
Transcript profiling in plastid arginine tRNA-CCG gene knockout moss: construction of Physcomitrella patens plastid DNA microarray
Plant Biol. (Stuttg.)
 , 
2005
, vol. 
7
 (pg. 
258
-
265
)
50
Silva
FJ
Belda
E
Talens
SE
Differential annotation of tRNA genes with anticodon CAT in bacterial genomes
Nucleic Acids Res.
 , 
2006
, vol. 
34
 (pg. 
6015
-
6022
)
51
Taniguchi
T
Miyauchi
K
Nakane
D
Miyata
M
Muto
A
Nishimura
S
Suzuki
T
Decoding system for the AUA codon by tRNAIle with the UAU anticodon in Mycoplasma mobile
Nucleic Acids Res.
 , 
2013
, vol. 
41
 (pg. 
2621
-
2631
)
52
Fabret
C
Dervyn
E
Dalmais
B
Guillot
A
Marck
C
Grosjean
H
Noirot
P
Life without the essential bacterial tRNA Ile2-lysidine synthetase TilS: a case of tRNA gene recruitment in Bacillus subtilis
Mol. Microbiol.
 , 
2011
, vol. 
80
 (pg. 
1062
-
1074
)
53
Li
L
Boniecki
MT
Jaffe
JD
Imai
BS
Yau
PM
Luthey-Schulten
ZA
Martinis
SA
Naturally occurring aminoacyl-tRNA synthetases editing-domain mutations that cause mistranslation in Mycoplasma parasites
Proc. Natl Acad. Sci. USA
 , 
2011
, vol. 
108
 (pg. 
9378
-
9383
)
54
Li
L
Palencia
A
Lukk
T
Li
Z
Luthey-Schulten
ZA
Cusack
S
Martinis
SA
Boniecki
MT
Leucyl-tRNA synthetase editing domain functions as a molecular rheostat to control codon ambiguity in Mycoplasma pathogens
Proc. Natl Acad. Sci. USA
 , 
2013
, vol. 
110
 (pg. 
3817
-
3822
)
55
Drummond
DA
Wilke
CO
The evolutionary consequences of erroneous protein synthesis
Nat. Rev. Genet.
 , 
2009
, vol. 
10
 (pg. 
715
-
724
)
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Comments

0 Comments