Altered tRNA characteristics and 30maturation in
bacterial symbionts with reduced genomes
Allison K. Hansen* and Nancy A. Moran
Department of Ecology and Evolutionary Biology, West Campus, Yale University, PO Box 27388 West Haven,
CT 06516-7388, USA
Received March 18, 2012; Revised May 1, 2012; Accepted May 5, 2012
Translational efficiency is controlled by tRNAs and
other genome-encoded mechanisms. In organelles,
translational processes are dramatically altered
becauseof genome shrinkage
acquisition of gene products. The influence of
genome reduction on translation in endosymbionts
whether divergent lineages of Buchnera aphidicola,
the reduced-genome bacterial endosymbiont of
Escherichia coli. Our RNAseq data support the
hypothesis that translation
Buchnera than in E. coli. We observed a specific,
convergent, pattern of tRNA loss in Buchnera and
other endosymbionts that have undergone genome
shrinkage. Furthermore, many modified nucleoside
pathways that are important for E. coli translation
are lost in Buchnera. Additionally, Buchnera’s
A+T compositional bias has resulted in reduced
Buchnera tRNA genes are shorter than those of
E. coli, as the majority no longer has a genome-
encoded 3’ CCA; however, all the expressed,
shortened tRNAs undergo
Moreover, expression of tRNA isoacceptors was
not correlated with the usage of corresponding
codons. Overall, our data suggest that endosymbi-
ont genome evolution alters tRNA characteristics
that are known to influence translational efficiency
in their free-living relative.
In the final step of protein synthesis, mRNA sequences
must be accurately and efficiently translated into amino
acid proteins. Reliable and efficient translation depends
critically on tRNA, which must exhibit specificity in
aminoacylation, and correct pairing of the anticodon
with its codon on the mRNA. The robust nature of the
genetic code and numerous genome-encoded mechanisms
promote translational accuracy (1,2), thus preventing
deleterious events such as the reassignment of codons
that can alter the function of thousands of genes.
Nevertheless, tRNAs and the genetic code sometimes do
change, especially in genomes undergoing size reduction
as exemplified by mitochondria and plastids (1,3,4). These
organelle genomes, which are derived from genomes of
symbiotic bacteria (5,6), exhibit the most extreme cases
of architectural alterations such as an increase in molecu-
lar evolutionary rate, inability to recombine, and massive
gene loss that sometimes leads to tRNA loss and changes
in the genetic code (7). Organelles encode a limited set
of proteins and rely on other co-occurring genomes
for enzymes and tRNAs (3,8,9).
The reduced genomes of some bacterial endosymbionts
exhibit similar but less extreme alterations in genome
sequence compared with
unlike organelles, most endosymbionts are still autono-
mous in the sense that they possess their own core
genetic machinery (11–13), including the conventional
bacterial structure of tRNAs (14,15). Most endosymbionts
retain the universal genetic code, but exceptions do exist
among the tiniest genomes, in which UGA is sometimes
recoded from Stop to Trp (16,17). In contrast, organelle
tRNAs and their translational machinery are highly diver-
gent from those of most bacteria (1,3,8,18). The question
still remains as to how endosymbiont tRNAs and transla-
free-living genomes that are not reduced. Overall, we
hypothesize that the process of genome shrinkage in endo-
symbionts results in a reduction of translational efficiency
and integrity resembling a transitional stage between
free-living ancestors and organelles.
Present day genomic features of bacterial endosymbi-
onts result from their ancient transition from a free-living
lifestyle to an obligate intracellular association (10).
*To whom correspondence should be addressed. Tel: +1 203 737 3106; Fax: +1 203 737 3109; Email: email@example.com
Nucleic Acids Research, 2012, Vol. 40, No. 16Published online 11 June 2012
? The Author(s) 2012. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/
by-nc/3.0), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Many bacteria that replicate strictly in host intracellular
environments possess reduced genomes with sequences
that are A+T biased relative to those of their free-living
ancestors (10,19,20). One such bacterium demonstrating
these genomic shifts is Buchnera aphidicola, an obligate
unculturable endosymbiont of aphids (21). Buchnera has
coevolved with its aphid hosts for 200-250 million years
(21), during which its genome shrunk to only 416–652 kbp
depending on the lineage (22–27). Based on previous gene
expression and genomic studies in Buchnera, genome
resulted in changes that are hypothesized to lower the
efficiency and accuracy of transcription and translation
(28–31) as compared with the free-living relatives. We
predict that Buchnera will also exhibit less optimal
tRNA features. Presently,
associated transcriptional mechanisms, which are key
components of efficient and accurate translation, have
not been extensively examined in Buchnera or any other
Comprehensive characterization of transcribed endo-
symbiont tRNAs has previously been difficult largely
because of the inability to isolate unculturable symbiont
tRNAs free of host contamination. However, analysis of
tRNAs beyond the level of DNA-encoded genes can
reveal the nature of tRNA maturation, including the
diversity of posttranscriptional processing that may
occur. Taking advantage of new methodologies in
high-throughput RNA sequencing (directional RNAseq),
and the availability of several divergent Buchnera genomes
(23,25,27), we investigated how genome reduction and
A+T richness affect tRNA evolution in this model
endosymbiont. This comparative framework provides us
with an understanding of the conservation of tRNA
sequences that influence specificity in aminoacylation
and secondary structure as well as conservation of nucleo-
side modification pathways that influence anticodon–
codon base pairing (1,2,32). From these data, we were
able to address how Buchnera tRNAs and associated tran-
scriptional fidelity mechanisms are altered relative to those
of free-living relatives, exemplified by Escherichia coli.
Additionally, because numerous reduced endosymbiont
investigated whether a pattern of tRNA loss was present
among reduced endosymbiont genomes.
sequence evolution has
MATERIALS AND METHODS
Four aphid species, Acyrthosiphon pisum (strains LSR1
and 5A), Acyrthosiphon kondoi (strain Ak), Schizaphis
graminum (strain Sg) and Uroleucon ambrosiae (strain
UA002, referred to as Ua), were reared in the same
growth chamber at 20?C. A. pisum was reared on seedlings
of Vicia faba, A. kondoi on Medicago sativa, U. ambrosiae
on Tithonia mexicana and S. graminum on Hordeum
For each aphid strain, B. aphidicola cells were filtered
from 3g of mixed age aphids. Filtration was done accord-
ing to the study by Moran et al. (33), with modifications as
follows. First, modified buffer A (34) was used instead of
PBS. Also, after the 1000rpm centrifugation step, the
pellet was resuspended and used for subsequent filtration
steps instead of the supernatant. After the last centrifuga-
tion step, supernatant and the protein layer were discarded
and the pellet was immediately immersed with Ambion
TRI Reagent Solution. For RNA extraction, a similar
protocol was used as in Hansen and Moran (34) except
that, after step 5, Qiagen’s miRNAeasy protocol under
appendix A from Qiagen’s miRNAeasy Mini Handbook
was used to enrich for miRNA (i.e. RNA <200bp). RNA
was DNAase treated, and quality and quantity was
checked as in Hansen and Moran (34). All filtration and
extraction materials were treated with RNAse AWAY
(Molecular BioProducts, Inc, CA, USA), and all solutions
were RNase free.
RNA sequencing, read processing, mapping, expression
The Yale Keck sequencing center carried out library
preparation and sequencing of Buchnera tRNA for all
five aphid strains. Briefly, for tRNA library preparation,
the Illumina mRNA directional sequencing protocol was
followed starting at the phosphatase treatment step. RNA
<200bp was directionally sequenced one lane per sample
with Illumina 35bp reads. The CLC Genomic Workbench
Aarhus, Denmark was used for read processing and
mapping. For all reads, small RNA adapters and reads
with ambiguous nucleotides were trimmed from reads.
Trimmed reads were then mapped to corresponding
Buchnera genomes (Table
mapping using the default settings for short reads. All
Buchnera taxa used in this study possess similar genome
Ua=615380bp; Sg=641454bp). tRNA reads that
mapped sense and anti-sense relative to the tRNA gene
were converted into Reads Per Kilobase of exon model per
million mapped reads (RPKM). Coverage per base pair
was calculated using custom perl scripts and Microsoft
Excel and was viewed in Artemis 13.0 (35) to visualize
sense and anti-sense tRNA coverage. For each Buchnera
strain, tRNA genes were annotated using genome anno-
tations in NCBI, tRNAscan-SE 1.21 (14,15) and Artemis
13.0 (35) to verify whether 30CCA was encoded in the
genome. tRNA CCA 30maturation occurs in all organ-
isms and is essential for charging tRNAs with amino
acids. To identify CCA 30maturation, the last 3020bp
of annotated tRNA’s were retrieved from all high
quality raw reads. Reads that perfectly matched the last
20bp were binned into the following three categories: (i)
reads match the 30tRNA end and no more nucleotides are
processed, (ii) reads match the 30tRNA end plus add-
itional non-CCA nucleotides are transcribed and (iii)
reads match the 30tRNA end plus CCA is added by mat-
uration. To analyse A+T richness in Buchnera and E. coli
CDS and tRNA genes the program EMBOSS (36) was
used. To calculate codon usage of 50 highly expressed
Buchnera genes (37), E-cai (38) was used.
Nucleic Acids Research, 2012,Vol.40, No. 167871
After consensus, RNAseq reads corresponding to tRNA
genes were mapped and assembled, tRNA species were
identified with tRNAscan-SE 1.21, with E. coli homology
Blast searches (39), and with verification of the presence
of signature identity elements relative to E. coli (32).
Survey of tRNA complements in small genomes
The last comprehensive survey of tRNA genes from
bacteria was conducted in 2002 and only included the
endosymbiont genome of Buchnera strain APS (40).
Because several smaller endosymbiont genomes have
been sequenced since 2002, we surveyed several more
genomes that varied drastically in genome size and
phylogenetic placement. The tRNAscan-SE Genomic
tRNA database (41) was used to characterize the
presence of tRNA gene isoacceptors (i.e. a tRNA species
that binds to one or more codons for a particular amino
acid residue) in 16 genomes.
High throughput detection of modified nucleoside bases
During library preparation, some modified bases cause the
reverse transcriptase to either fall off at the modified
position, and/or to incorporate a ‘mismatch’ relative to
modified bases and potential posttranscriptional process-
ing, we screened for mismatches in tRNA reads relative to
the reference tRNA gene similar to Iida et al. (42) and
Findeiß et al. (43). After mapping, only the sense tRNA
reads in CLC (using the same mapping parameters as
polymorphism (SNP) analyses to detect mismatches.
Threshold criteria for counting a mismatch were estab-
lished by identifying conserved mismatches in both
Ap-5A and AP-LSR1 (two different strains from the
same aphid species). These two strains shared 38
mismatches for which the mismatch rate was more than
1% per base (i.e. above Illumina’s expected error rate
per base) and the alternative variant count was at least
eight reads. This mismatch criterion was then used to
detect mismatches in other strains for a total of four
divergent Buchnera taxa (Ap, Ak, Ua and Sg).
Predicted tRNA-modified bases and their pathways
for each Buchnera tRNA were obtained from E. coli
homologs usingboth http://modomics.genesilico.pl/
pathways/ (44) and http://www.ecocyc.org/
Divergent Buchnera genomes (23,25,27) were searched
homologs using Blastp (39).
modification pathwayenzymes using E. coli
tRNA secondary structure
Infernal (46) was used to generate tRNA sequence and
secondary structure alignments among Buchnera strains
and E. coli. The covariance model, RF00005 cm, was
used, which accounts for tRNA secondary structure
constraints. Using Infernal output, 4sale (47) was used
to compute pairwise compensatory substitution tables
from stems for all tRNAs among Buchnera strains and
E. coli. Stability of tRNA secondary structure was
measured as Delta G (?G), the change in Gibbs Free
Energy (in units of kcal/mole). Thus, the more negative
?G is, the more thermodynamically stable the tRNA
secondary structure. ?G was computed for tRNAs of
each strain individually using RNAalifold (48,49) with
constraints on tRNA constraint folding generated by
tRNAscan-SE 1.21 (14,15).
All raw sense and anti-sense tRNA data were submitted
to NCBI Genbank under SRA submission: SRA049863.3,
under Bioproject #s: (i) PRJNA82811, (ii) PRJNA82809,
(iii) PRJNA82797, (iv) PRJNA82793, (v) PRJNA82789.
All paired sample-ttest
(%GC), tRNA length), correlation (pairwise RPKM com-
parisons) and regression (codon usage and tRNA expres-
sion) statistics were carried out using IBM SPSS Statistics.
2010 for Mac, standard version 19.0. New York, USA.
All Buchnera tRNAs are transcribed
For all Buchnera genomes, tRNA genes occur in the same
genomic positions (Figure 1). Based on tScan and blastn
detection of homology with E. coli the same 32 tRNA
genes and 29 anticodon types are conserved across
Buchnera taxa (Figure 1).
As expected, directional RNAseq reads map primarily
in the sense direction of tRNA genes, with antisense reads
averaging less than 1% of the sense reads (Table 1,
Figure 1). All Buchnera tRNA genes are expressed in the
sense direction, but some lack antisense expression,
depending on strain, and sense expression is always
higher than antisense expression (except for Phe GAA in
Ak and Sg) (Figure 1). tRNA sense expression is positively
correlated across divergent Buchnera taxa (Table 2). The
level of antisense expression is highly correlated across all
Buchnera taxa, but the correlation is less for Buchnera-Sg,
the most divergent taxon (Table 2). Transcriptional start
sites and coverage curves for antisense RNAs varied
widely across Buchnera taxa. Nevertheless, conserved
50transcriptional start sites and coverage curves were
identified for several antisense RNAs that occurred
on or near tRNA genes for all five Buchnera taxa
(Supplementary Table S1).
Conservation of tRNA identity elements in Buchnera
Recognition of tRNAs by tRNA synthetases is essential to
the fidelity of translation. Aminoacyl-tRNA synthetases
Table 1. Total reads mapping per sample from Illumina directional
Buchnera RNA samples Mapped to tRNAReference genome
7872 Nucleic Acids Research, 2012,Vol.40, No. 16
Figure 1. Sense and antisense expression of 32 tRNA genes from five Buchnera taxa based on RNAseq data. (a) Buchnera 5A’s chromosome.
Green and brown regions indicate forward and reverse coding regions respectively. Letters A–E specify regions where tRNA genes are coded, and
correspond to (c). (b) phylogenetic distance of Buchnera taxa; number labels (1–5) correspond to (c). tRNA expression is presented as Reads Per
Kilobase of exon model per Million mapped reads (RPKM). (c) tRNA gene expression from the five Buchnera taxa. Blue and red boxes represent
sense and antisense RNA expression, respectively, at different magnitudes. Yellow outlined boxes indicate tRNAs with genome encoded 30CCA.
tRNA gene direction is indicated by arrows above tRNA isoacceptor names.
Nucleic Acids Research, 2012,Vol.40, No. 167873
(aaRS) must recognize multiple tRNA isoacceptors
(i.e. different tRNA species that bind to alternative
codons for the same amino acid residue) but discriminate
against others. This recognition is dependent on tRNA
identity elements, consisting of evolutionarily conserved
bases at specific positions of tRNAs (Giege’ et al. 1998).
Based on RNAseq data from all taxa, unmodified identity
elements for each tRNA are identical to those in E. coli
except for base substitutions in CysGCA(G15 to U15; A13
to G13), SerGGA(G73 to A73), SerGCT(variable loop 1bp
shorter, except in Sg) and AlaGGC(G20 to U20, 5A and
Ua only; G20 to C20, Sg and Ak only). Based on blastp
analyses, all 20 cognate aaRS are encoded within each
In contrast to E. coli and most other organisms with
nonreduced genomes, Buchnera does not encode multiple
tRNA genes with matching anticodons, except for three
tRNA genes encoding the anticodon CAU. Two of these
genes encode either an initiation or elongation Met tRNA
based on tRNA identity elements and homology (Table 3).
The other tRNA gene encoding a CAU anticodon
possesses homology and identity elements corresponding
to the IleLAUanticodon (Table 3).
Selective loss of tRNA isoacceptors from Buchnera
Numerous tRNA isoacceptors are present in E. coli but
missing from all Buchnera strains. Many Buchnera tRNA
isoacceptors that belong to 4-codon family boxes and to
two-codon families (50-NNR codon type) have been lost
from Buchnera genomes (Table 3). 50-CNN anticodons
were preferentially lost in family boxes corresponding to
Leu, Gly, Ser, Thr and Pro. Only one family box, corres-
ponding to Pro, lost both 50-CNN and 50-GNN anti-
codons. For two-codon families, a 50CNN anticodon
was lost from Gln (and Leu and Arg for 6-codon families),
relative to E. coli (Table 3). Based on Watson and Crick
base-pairing and revised wobble rules (50,51), all tRNA
isoacceptors encoded and expressed in Buchnera can base
pair with the 61 possible codons (Table 3), which are all
still encoded in Buchnera’s protein-coding genes at
The pattern of tRNA gene isoacceptor loss was
examined in 16 bacterial taxa representing a wide range
of genome sizes and phylogenetic associations, including
some with extremely reduced genomes (Figure 2).
Reduced genomes show common patterns of retention
50-CNN anticodons followed by 50-GNN anticodons are
consistently eliminated from the small genomes. For
eliminated. In the most reduced genomes, only 50-UNN
anticodons remain for both family box and two-box
codons. Unmodified 50-U anticodons can wobble and
pair with all four base combinations for family box
codons (50). Therefore, for 50-NNR two-box codons,
the 50-U of anticodons must be modified to prevent mis-
translation of neighboring two-box (NNY) codons (1, 2)
(e.g. an unmodified 50U in a Gln 50-UUG anticodon can
mispair with His codons 50-CAU and CAC [Table 3]).
Table 2. Spearman rho correlations of tRNA sense expression (grey,
below) and tRNA anti-sense expression (white, above) between pairs
of Buchnera taxa*
Table 3. Buchnera tRNA isoacceptors retained and expressed in Buchnera genomes (bolded)
Dark boxes represent family boxes of 4- and 6-fold degenerate codon families. Light grey boxes represent 50-NNR(A/G) 2 codon boxes of 2- and
6-fold degenerate codon families.
atRNA is homologous to E. coli Ile LAU; L modification enzyme encoded in Buchnera genomes.
bRepresents two Met isoacceptors in Buchnera (initiation and elongation tRNAs).
cBuchnera lost this isoacceptor relative to its presence in E. coli.
7874Nucleic Acids Research, 2012,Vol.40, No. 16
Buchnera tRNAs are modified and RNA edited
Based on E. coli tRNA homologs, 26 different types of
nucleoside modifications are predicted to occur in
Buchnera tRNAs (Table 4, Supplementary Table S2,
Supplementary Dataset 1). Nine of these modifications
are important for the efficiency and fidelity of protein
synthesis and occur in N34 tRNA positions (wobble) of
E. coli (Table 4). We expect five of these N34 modifica-
tions to be retained to code for all cognate codon pairs
and prevent mistranslation of other amino acids (e.g.
mnm5u, mnm5s2U, cmnm5Um, I and K2C). An inosine
(I) modification is important in E. coli because 50-A
from anticodon ArgACGis modified into I, which can
wobble and pair with Arg codons CGA, CGU and CGC
(55). Lysidine (K2C) is an important modification in
E. coli because 50-C from anticodon IleCAUis modified
into K2C (L), which pairs with Ile codon AUA (instead
of the Met codon AUG) (59). Other expected N34
modifications (mnm5u, mnm5s2U and cmnm5Um,) are
important for modifying anticodon 50U for NNR two
codon boxes, thus preventing mistranslation (1,2). Based
on Buchnera genome annotations, entire pathways are
only present for expected wobble bases I, k2C and
cmnm5um (Table 4); however, some pathways are only
missing the last enzyme in a pathway (e.g. mnm5u
and mnm5s2U), and/or are still unknown in E. coli.
High throughput mismatch evidence (see ‘Materials and
Methods’ section) shared by multiple taxa supports the
presence of a modified nucleoside at 50-A from anticodon
ArgACGin all Buchnera strains. These data support
the presence of an inosine modification in all taxa.
For example, we found a high frequency of anticodon
50-ACG transcribed as 50-GCG, where the frequency
of 50-G/A at this wobble base position was, Ap-5A:
61/39%; Ap-LSR1=70/30%; Ak=69/31%; Ua=27/
73%; and Sg=72/28%. Presence of transcripts contain-
ing a 50-G for the ArgACGanticodon is strong indirect
evidence for an inosine modification. For example,
during the reverse transcription process, the modified nu-
cleoside inosine base pairs with C residues, and therefore
‘G’ is found in the consensus cDNA sequence instead of
‘A’ (60). Conserved high throughput mismatch evidence
for Ap-5A, Ap-LSR1 and Ak also supports the presence
of a modified base at N34 for LysTTT, suggesting that
mnm5s2U is present in these strains even though the
E. coli version of the pathway appears incomplete in
Buchnera. Error evidence was not detected for other
expected modified wobble positions relative to E. coli,
even though full pathways are retained in the genome
Other tRNA modifications that are very important for
the fidelity of protein synthesis are N37 modifications.
N37 modifications are known to stabilize weak A:U and
U:A base pairing between N36 of the anticodon and N1 of
the codon (1,2,51). Based on in vitro experiments, N37
modifications are known to increase the interaction of
the codon with the anticodon, preventing miscoding of
amino acids and frameshifts (52,53,61–63). In turn, to
maintain efficient translation, we expect these modifica-
tions to be retained. Based on modifications for the hom-
ologous tRNAs in E. coli, seven important N37 modified
nucleosides are predicted in Buchnera. Among Buchnera
genomes, four N37 nucleosides pathways are retained, two
are missing, and one has an unknown pathway in E. coli
(Table 4). High throughput mismatch evidence supports
the presence of a modified base at N37 for PheGAA,
ProTGG, LeuGAGand LeuTAGand thus suggests that
ms2i6A, m1G, xG and xG, respectively, are present in
all taxa. However, no mismatch was detected in Sg for
LeuGAG. The tRNA modifications at positions other
than N34 and N37 that are supported by mismatch
evidence are shown in Supplementary Table S2, and
Supplementary Dataset 1. Mismatch evidence was also
found at positions at which E. coli does not process
modified nucleosides, suggesting the presence of new
modified nucleoside sites and/or RNA editing of mature
tRNAs (Supplementary Dataset 1). Collectively, all
Figure 2. Pattern of isoacceptor tRNA gene loss with decreasing genome size for a variety of free-living and symbiotic/pathogenic bacteria from
multiple phyla and classes. Black dots represent the presence of a single tRNA isoacceptor. All bacteria have fully sequenced genomes, and their
tRNA presence is based on tScan predictions. Note that some bacteria do not possess all seven family box codons and/or all four NNR two-codon
sets. a=Alphaproteobacteria, b=Betaproteobacteria, g=Gammaproteobacteria, Bac=Bacteroidetes, Act=Actinobacteria, Firm=Firmicutes,
+= Gram-positive bacterium.
Nucleic Acids Research, 2012,Vol.40, No. 167875
Table 4. Mechanisms maintaining translational fidelity and efficiency
Modified base (Original base)
tRNA isoacceptors with
modification in E. coli
General role (reference number)
ArgACG, AspGUC, GlnUUG,
Prevents anticodon-codon mispairing for 50-NNC/G anticodons (2)
CysGCA, LeuUAA, PheGAA,
SerUGA, TrpCCA, TyrGUA
Prevents anticodon-codon mispairing for 50-NNA anticodons,
Prevents miscoding of Phe in E. coli, Prevents framehifting (2,
ArgUCU, AsnGUU, IleGAU,
LysUUU, MetCAU, IleCAU,
Prevents anticodon-codon mispairing for 50-NNU anticodons (2)
Prevents frameshifting (3,4,5)
Prevents anticodon-codon mispairing for 50-NNG anticodons (2)
Increases the efficiency of ThrGGUto read the codon ACC (54)
Wobble to decode three Arg codons (55)
Uridine 5-oxyacetic acid (U)
ValUAC, ThrUGT, ProUGG,
Increases efficiency of 50U wobble for family box codons (2)
Recognizes Arg codon AGA and much less efficiently codon AGG
in vivo, constricting wobble from Ser AGU and AGC (2,56)
TyrGUA, AsnGUU, AspGUC,
Recognition of 50NAU and NAC codons (57), Prevents
frameshifting for Tyr and His codons (53)
GlnUUG, GluUUC, LysUUU
MnmG, IscS, GidAe
Prevents frameshifting in Lys codons (53), Constricts 50U wobble
for NNR codons (2)
Recognizes Leu codons UUA and UUG but constricts wobble
from Phe codons UUU and UUC (2)
Prevent misreading of Ile AUA codons (58)
Negative determinate for aminoacylation by MetRS (59)
Conservation of Buchnera taxa wobble -N34 and N37 tRNA modified base pathways and detection of modified base products in 32 tRNAs.
aBuchnera strains screened for homologous E. coli pathway enzymes are: Ap (5a and LSR1), Ak, Ua, and Sg.
bDirectional RNAseq detection of modified bases among four divergent Buchnera taxa- Ap (5a and LSR1), Ak, Ua and Sg (see ‘Materials and methods’ section).
cPresent in all Buchnera except pseudogene present in Sg.
dPresent in Ap, Sg, but absent in Ua, and a pseudogene in Ak.
eEntire pathway present in all Buchnera taxa except for MnmC.
Dotted lines, Buchnera taxa where one or more enzymes are absent in the pathway.
Underlined tRNA isoacceptors indicate which tRNA shows RNAseq evidence of a modified base at either N34 or N37 for two or more divergent Buchnera taxa.
7876 Nucleic Acids Research, 2012,Vol.40, No. 16
mismatch frequencies (with the exception of ArgACG) were
dominated by the reference sequence base at a frequency
Mismatches were primarily not changes to a single nucleo-
tide base, but were composed of three different bases other
than the reference base.
relativeto mismatches foralltaxa.
No relationship between codon frequencies and
In many species, tRNA abundances are positively
correlated with codon usage for highly expressed genes
(64,65). Anticodons of highly expressed tRNAs corres-
pond to codons that are used frequently in these genes,
thus improving the efficiency of translation (64,65). Based
on Watson and Crick and revised wobble base-pairing
rules (50,51), each Buchnera isoacceptor was paired with
its corresponding codon pair. Met CAU, the only dupli-
cate anticodon coding for the same codon, was excluded
from analysis. The relationship between percent average
codon usage of highly expressed genes and corresponding
tRNA isoacceptor expression was examined for each
Buchnera strain. No significant relationship was found
between average codon usage of 50 highly expressed
genes in Ap-5A (on leading and lagging strands) and
cognate tRNA isoacceptor sense expression (Figure 3).
No significant relationship was found on examining the
relationship between highly expressed Buchnera genes
cognate tRNA isoacceptor sense expression for all taxa
(Supplementary Figure S1A and S1B). Examination of
codon usage and tRNA expression scatterplots reveals
that most tRNA isoacceptors, regardless of codon usage,
are expressed at similar levels (e.g. for Ap-5a in RPKM
the 75 percentile=843 950, median=309 742 and
max=4 407 138; Figure 3). TrpCCAis the highest
expressed isoacceptor in all taxa (except Ua), even
though the corresponding codon occurs at low frequency
(Figure 3 and Supplementary Figure S1).
Buchnera tRNAs maintain secondary structure with
compensatory base substitutions
As expected, Buchnera CDS are significantly more A+T
rich relative to CDS of E. coli [Figure 4 (c)]. Within each
Buchnera genome, tRNA genes are 2.2-fold more G+C
rich relative to CDS, indicating that selection conserves
higher %G+C in tRNA genes. Nevertheless, Buchnera
tRNA genes are significantly more A+T rich than
homologs in E. coli [Figure 4 (c)].
Stability of tRNA secondary structure can decrease
with a reduction in %GC, especially in stem structures.
Because Buchnera tRNAs are more A+T rich than those
of E. coli [Figure 4 (c)], we measured the stability of
Buchnera tRNA secondary structure. ?G was significantly
more negative in E. coli tRNAs relative to homologs in
Buchnera for all strains, indicating that Buchnera tRNAs
have reduced stability in vitro [Figure 4 (b)]. Whether they
have reduced stability in vivo, where stabilizing proteins
may play a role, remains to be tested. Two tRNAs with
the weakest secondary structure in all Buchnera relative to
E. coli were ValGAGand TrpCCA; both tRNAs possess
numerous compensatory and single base substitutions
in the stem regions [Figure 4 (a)].
Buchnera tRNAs are more A+T biased and display
weaker secondary structure than those of E. coli
(Figure 4). However, a high frequency of compensatory
base substitutions are expected in the stem regions as a
mechanism for maintaining functionality of these essential
molecules. Relative to E. coli, a total of 37–42 compensa-
tory basesubstitutions were
tRNA stem regions (Table 5). Many of these compensa-
tory substitutions were C/G to T/A directional changes
Buchnera tRNA gene shrinkage and compensatory
Genome reduction primarily reflects loss of coding genes,
as reduction in gene length is minor (<1%, 37), and gene
packing is similar for bacterial genomes of different sizes
(66). However, Buchnera tRNA genes are often shorter in
length than their homologs in E. coli [Figure 5 (a)].
The difference in length is typically 3bp and mostly
reflects the loss of encoded 30CCA in the Buchnera
tRNA genes. At the 30end of tRNAs, CCA is required
for amino acid activation, and must either be encoded in
the tRNA gene or added during tRNA maturation by the
CCA-adding enzyme. Although E. coli and other close
relatives of Buchnera such as Vibrio and Pseudomonas spp.
all encode 30CCA in all tRNA genes except that for
selenocysteine, only half of Buchnera tRNA genes
encode 30CCA [14-17 depending on strain, Figure 5 (b)].
The remaining Buchnera tRNA genes have lost the 30
encoded CCA. Our analysis of directional RNAseq
reads indicates that the mature transcript of these genes
possesses a CCA at the 30end [Figure 5 (b)], implying
Buchnera Ap-5A sense tRNA expression (RPKM)
Ave codon usage of 50 highly expressed genes (%)
r = 0.002
Figure 3. The relationship between percent average codon usage and
corresponding tRNA isoacceptor expression. Average codon usage
calculations are based on 50 highly expressed genes in Buchnera-Ap
(based on Charles et al. 2006).
Nucleic Acids Research, 2012,Vol.40, No. 167877
Some Buchnera tRNA genes with 30CCA encoded also
displayed CCA 30maturation [Figure 5 (b)], resulting in
double or triple CCA at the 30end of tRNAs. Recently,
it was shown that tRNAs with dual 30CCA are targeted
for degradation (67). More specifically, if a tRNA has 50
Gs on bp 1 and 2, and its acceptor stem is structurally
unstable, then the CCA-adding enzyme marks unstable
tRNAs by adding dual 30CCAs, targeting it for degrad-
ation by RnaseR (67). Such degradation also seems
possible in Buchnera strains, which encode both the
CCA-adding enzyme and RnaseR. Thus, we examined
all Buchnera tRNAs with dual and triple 30CCA matur-
ation. First, we noted that all E. coli tRNAs with a 50G at
the 1st and 2nd base position encode dual or triple CCA
on the 30end of the tRNA gene [Figure 5 (c)]. Based on
tRNAscan-SE 1.21, the penultimate CCA is always
incorporated into the 30acceptor stem, exposing a single
30CCA for activation. Most Buchnera tRNAs that display
dual or triple 30CCA maturation still retain a 50G at
the 1st and 2nd bases and are homologs to dual or triple
30CCA encoded E. coli tRNAs [Figure 5 (c)]. Three
strain-specific tRNAs with dual 30CCA maturation do
not have E. coli homologs with dual CCAs encoded.
These Buchnera tRNAs also do not encode 50Gs at the
Percent similarity among bacteria
* = All Buchnera except Ua
λ = All Buchnera except Ak
α = Only Ak
σ = Only Sg
Mean tRNA ΔG (+/- 1 SE)
ApAk Ua Sg
Mean % GC (+/- 1 SE)
P < 0.001
20 40 60 80
Figure 4. Reduction in stability of secondary structure and %GC in tRNAs of Buchnera relative to E. coli. Buchnera strains=(Ap-5A, Ak, Ua, Sg)
(a) Two tRNAs in Buchnera with the lowest secondary structure stability relative to E. coli based on ?G. Red arrows indicate Buchnera base
substitutions relative to E. coli in stem structures. White bases indicate compensatory substitutions in Buchnera and circled white bases indicate single
base substitutions relative to E. coli. Black bases indicate conserved bases among all bacteria strains. (b) Mean tRNA ?G and (c) mean % GC for all
32 Buchnera tRNAs per strain and their corresponding tRNA homologs in E. coli.
Table 5. Total number of pairwise compensatory base changes
(upper right) and percent of compensatory base changes with C/G
to T/A (lower left) for 32 homologous tRNAs of Buchnera strains
and E. coli
Ap-5AAk UaSgE. coli
7878Nucleic Acids Research, 2012,Vol.40, No. 16
1st and 2nd base. All Buchnera with dual or triple 30CCA
maturation incorporate the 2nd to last CCA into the
30acceptor stem as in E. coli, except for one case, tRNA
LeuTAAin Ak [Figure 5 (c)].
The efficiency and fidelity of translation is reinforced by
many mechanisms encoded in genomes. In reduced
genomes, mutation rates are typically high, and selection
becomes less effective in maintaining translational mech-
anisms. In this study, we found that bacterial endosymbi-
ont lineages (Buchnera) that experience relaxed selection
display less optimal tRNA characteristics relative to those
of their free-living relative E. coli. Gene loss and A+T
mutational bias in Buchnera have lead to the loss of
tRNA isoacceptors and loss of modified base pathways,
the reduction of tRNA gene length, and the accumulation
of base substitutions and indels (insertions/ deletions) in
tRNA sequences that weaken tRNA secondary structure
and possibly aminoacyl-tRNA synthetase recognition.
These tRNA characteristics are conserved across four
Buchnera lineages spanning 70 million years of divergence
and mayresultin reduced
and fidelity relative to their ancestors. However, we did
detect compensatory base substitutions in Buchnera
tRNAs, which are expected to maintain secondary struc-
ture of tRNA stem regions. Additionally, RNAseq reads
reveal novel 30maturation processes that compensate for
tRNA gene length reduction.
Divergent Buchnera taxa in this study encode and
express the same 32 tRNA genes composed of 32 different
isoacceptor types (Figure 1). In turn, no duplication of
tRNA gene isoacceptors was found. Based on a survey
of 50 eukaryotic, eubacterial, and archaeal genomes, low
tRNA gene redundancy (i.e. only one or two gene copies
of a particular isoacceptor) was only found in all
archaeans and several bacterial genomes, and was
approximately correlated with genome size (40). In
Buchnera, because of modified wobble rules (50,51), all
mature tRNAs expressed can theoretically base pair with
the 61 possible codons (Table 3, Figure 1), which are all
still encoded in Buchnera CDS. One special Buchnera
Buchnera-Ap (taxa type strain APS) is tRNA IleCAU
(40), where 50-C is modified into lysidine by the enzyme
TilS in E. coli (55), which all Buchnera strains still encode.
This special IleCAUisoacceptor codes for Ile instead of
Met due to a wobble modification, and is ubiquitous
in Eubacteria and Archaea (40).
During genome reduction, Buchnera has preferentially
lost 50-CNN, and to a lesser extent, 50-GNN anticodons in
family boxes and 50CNN anticodons from two-codon
NNR families(Table 3).
isoacceptor loss is common for many bacteria with
reduced genomes (Figure 2), and is most likely related to
gene deletion processes. Selective loss of these specific
isoacceptors in family boxes and NNR two-codon
families in Eubacteria was observed in previous studies
(1,40,68,69) but was related to A+T sequence bias not
deletion processes (1,70). We hypothesize that genome
reduction, which is correlated with A+T bias, is the
most likely explanation for this pattern of tRNA
isoacceptor loss. First, the potential for wobble in
codon–anticodon basepairing implies that some tRNA
isoacceptors are notessential
17 15 tRNA N=
CCA 3’ maturation
No CCA 3’ maturation
Encodes a G at the 1st and 2nd
bases on 5’ end
Does not encode a G at the 1st and 2nd
bases on 5’ end
= genome encoded CCA at the 3’ end
= evidence of CCA maturation at 3’ end
Second to last CCA not
incorporated in 3’ stem
Homolog not present
Mean tRNA Length (+/- 1 SE)
P < 0.001
Figure 5. Evidence of CCA 30tRNA maturation in Buchnera (strains
Ap-5A, Ap-LSR1, Ak, Ua, Sg). (a) Mean tRNA length for all 32
Buchnera tRNAs per strain and their corresponding tRNA homologs
in E. coli (b) Presence and absence of 30tRNA maturation in 86 E. coli
tRNAs and all 32 Buchnera tRNAs for each strain. Grey numbers in
CCA encoded black bars indicate the number of tRNA genes that both
encode CCA and process an additional CCA during 30
maturation based on directional RNAseq (c) Pattern of dual or triple
CCA 30transcription in tRNAs among E. coli and Buchnera strains.
tRNA isoacceptors in the left column correspond to the grey numbers
in the black bars in panel ii for Buchnera strains. In both E. coli and
Buchnera, the second to last CCA is always incorporated in the 30stem
except for one tRNA in Ak (asterisk).
Nucleic Acids Research, 2012,Vol.40, No. 167879
corresponding codons (e.g. 50-CNN, 50-GNN anticodons)
and can be eliminated through mutation and deletion.
followed by 50-UNN anticodons are the most promiscuous
isoacceptors when pairing with cognate codons; thus, it is
not surprising that 50UNN is always retained in family
box and two-box NNR codons in the most reduced
genomes. In turn, 50-UNN anticodons are probably
retained because of their ability to recognize alternative
codons rather than because of the high frequency of
cognate codons in A+T rich CDS. Typically in bacteria
and eukaryotes 50-CNN and 50-GNN anticodons of family
boxes and 50-CNN anticodons from two-codon families
along with 50
U anticodon modifications extending
increase the efficiency of translation (1,71). We predict
that the loss of tRNA isoacceptors in Buchnera as well
as other endosymbionts potentially results in less efficient
Numerous unmodified nucleotides at specific nucleotide
positions on tRNA isoacceptors are conserved phylogen-
etically and are known to play crucial roles in defining
tRNA specificity for aminoacylation (32,72). These
conserved nucleotides are called identity elements and
are required for proper recognition by the cognate aaRS
in addition to playing roles as deterrents to false recogni-
tion (32). Our results reveal that most Buchnera tRNAs
have maintained identity elements homologous to those in
E. coli, with the exceptions of CysGCA, SerGGA, SerGCT
and AlaGGC. In E. coli tRNAcys, the identity elements
G15·G48 form an unusual tertiary base pair called a
Levitt pair (73). Additionally, the E. coli identity
elements A13·A22 are important in determining the
structure of G15·G48 (74). Collectively, these E. coli
identity elements are required for CysRS recognition due
to their role in RNA tertiary structure (73). In all,
Buchnera taxa, tRNAcys
G15·G48 has mutated to
U15·G48 and A13·A22 has mutated to G13·A22. Hou
et al. (73) found that when G15·G48 is mutated to
U15·G48, its backbone configuration is similar to the
wild type tRNAcys; however, only partial aminoacylation
(46.2%) occurs relative to the wild type. How both types
of changes in identity element together affect tertiary
structure is unknown.
In Buchnera tRNA AlaGGC, the identity element G20 is
mutated to U20 in strains 5A and Ua and to C20 in Ak
and Sg. In E. coli tRNA AlaVGC, these same base changes
were shown to result in 6? and 50? reductions in alanine
charging activity, respectively, relative to native tRNA
AlaVGC(75). Buchnera AlaUGCdoes not possess this
mutation. Potentially, if this mutation is deleterious in
AlaGGCrecognition, AlaUGCcan wobble to all four
alternative codons for the family box codon family for
alanine. Interestingly, the smallest sequenced genome of
Buchnera, for the host Cinara cedri, retains the same
tRNA isoacceptors and aaRSs as other Buchnera taxa
examined in this study; however, the AlaGGCtRNA
gene has been lost, resulting in a total of only 31 tRNA
In Buchnera tRNA SerGGA, the identity element G73
(the discriminator base) has mutated to A73. Generally
a mutation in the discriminator base is known to result
in the loss of cognate aminoacyl-tRNA synthetase recog-
nition; however, Shimizu et al. (76) demonstrated that any
four bases substituted in the discriminator base of E. coli
Ser tRNA resulted in the same level of aminoacylation.
Nevertheless, G73 in Ser tRNA is phylogenetically
conserved (72) and has been shown to play minor roles
in SerRS discrimination (77,78). Additionally, in E. coli
Ser tRNA, the variable region plays a very important role
as an identity element (77,79). In all Buchnera taxa, except
Sg, the variable region length of the SerGCTisoacceptor is
1bp shorter than the E. coli SerGCTisoacceptor. In
summary, it is unknown how all these mutated identity
elements affect Buchnera translation, but the same muta-
tions in E. coli are known to significantly reduce the effi-
ciency of aminoacylation.
In addition to requiring specificity in aminoacylation,
reliable and efficient translation requires the anticodon
to correctly pair with its codon. Modified nucleosides of
tRNAs are essential mechanisms reinforcing translational
fidelity and efficiency, especially at the wobble (N34) and
30position immediately adjacent to the anticodon (P37),
(1,2,51). Based on E. coli tRNA homologs, we expect 16
different types of modified bases to be present in the
remaining 32 Buchnera tRNAs, for both N34 and N37
positions. In E. coli, 13 of these modified base pathways
are known and Buchnera encodes complete pathways for
six of these (Table 4). All Buchnera taxa have lost enzymes
responsible for encoding N37 modified bases m2A and
m6A, which are important in stabilizing 50-NNC/G anti-
codons (2) (Table 4). Enzymes that synthesize the N37
modification m6t6A are conserved in only half of
Buchnera taxa; this enzyme is known to slightly increase
the efficiency of base pairing of the anticodon ThrGGUto
the codon ACC in E. coli (54). All N37 modified base
pathways important for preventing frameshifts and
stabilizing A:U and U:A at the wobble position of the
anticodon and the first position of the codon were
retained in all Buchnera taxa (Table 4). These mechanisms
may be essential for the fidelity of translation, especially
for A+T rich genomes.
Modified nucleosides at the wobble base position (N34)
of the anticodon are important for encoding the right
amino acid, extending or restricting wobble, increasing
the efficiency of base pairing and preventing frameshifts
(2,53,55,56,58,59). Buchnera taxa all encode the enzyme
TilS that is essential for the synthesis of the modified
base lysidine, and is important for encoding the amino
acid Ile instead of Met (59). All Buchnera taxa also
encode the core enzymes MmmE and MnmG that are
important for synthesizing the modified bases mnm5u,
mnm5s2U, and cmnm5Um, which restrict 50U wobble in
NNR two-box codons, including Arg and Leucine
(Table 4). All of these pathways are complete except for
MmmC, which is involved in the last step for both
RNAseq mismatch evidence supports the presence of a
modified base at the expected position of mnm5s2U
(Table 4). Interestingly, the genes encoding MmmA,
MmmE, MnmG, and IscS or SufS, but not MnmC are
retained in several tiny endosymbiont genomes (10).
7880Nucleic Acids Research, 2012,Vol.40, No. 16
Conservation of these enzymes in reduced genomes indi-
cates that these enzymes or derivatives are important for
the production of the modified bases mnm5u, cmnm5Um,
and especially mnm5s2U, which is essential for preventing
frameshifts and restricting wobble in NNR two codon
boxes (Glu, Lys and Gln), thereby preventing the
miscoding of amino acids. For incomplete pathways
producing modified bases mnm5u and mnm5s2U, either a
derivative may be synthesized and/or the insect host may
import MnmC. For example, the pea aphid, A. pisum
expresses its mnmC homolog (XP_003245837) in both its
body and in the specialized aphid cells (bacteriocytes) that
contain Buchnera cells (34).
Another key enzyme that is retained in Buchnera is
TadA, which is responsible for synthesizing inosine in
E. coli (55). This wobble modification is present on
ArgACGin many bacteria and can wobble to three alter-
native codons of Arg (2,59). Rnaseq mismatch evidence
highly supports this modification, as inosine is recognized
as G during the reverse transcription process (60), and
therefore we were able to measure a high frequency of
modified ArgACGtranscripts from all Buchnera taxa.
Unfortunately, other modified bases do not appear to be
recognized as specific bases and in turn incorporate
different frequencies of any of the four bases during
reverse transcription of modified transcripts (42,43).
Collectively, Rnaseq evidence supported the presence of
five modified bases, four in which the pathways are known
and present (or near present for mnm5s2U) and one in
which the pathway is unknown (Table 4). If Buchnera
tRNAs can be isolated without host contamination,
modified base presence and identity can be confirmed.
In many bacterial species, tRNA abundances are posi-
tively correlated with codon usage for highly expressed
genes, thus increasing translational efficiency (64,65).
In addition to analysing specific tRNA characteristics
that influence the accuracy and efficiency of translation,
we examined whether codon usage correlates with tRNA
expression. We found that tRNA sense expression is
highly correlated across Buchnera taxa (Table 2), and
many tRNA isoacceptors are expressed at similar levels
A previous microarray study suggested that tRNA expres-
sion and codon usage of 50 highly expressed genes in
Buchnera-Ap were positively correlated (37), but the rela-
tionship was weak and expression of sense and antisense
tRNAs were not distinguished, possibly confounding
results. Our directional RNAseq data show no relation-
ship between tRNA expression and codon usage, for
the same set of highly expressed genes in Buchnera-Ap
under similar conditions (Figure 3). Furthermore, no
relationship was detectable in three other Buchnera taxa
(Supplementary Figure S1). Collectively, these results
suggest that selection is not maintaining codon bias
for highly expressed proteins. Interestingly, TrpCCA, is
the highest expressed isoacceptor in all Buchnera taxa
(except Ua) and has very low codon usage. In all,
Buchnera examined, isoacceptor TrpCCAdisplays one of
the lowest secondary structures relative to E. coli’s
homolog; potentially TrpCCAis highly expressed to com-
pensate for low aminoacylation efficiency related to
numerous base substitutions that weaken its secondary
structure [Figure 4(a)].
In this study, we found that Buchnera tRNAs have
maintained high %GC relative to its CDS; however, its
tRNAs are more A+T rich and less stable relative to
homologs in E. coli (Figure 4). These results are consistent
with previous findings (28) showing that 16S rRNAs of
Buchnera and other endosymbiont species are more A+T
rich and less stable than those of free-living relatives.
Similarly, mitochondrial tRNAs from animals are more
A+T rich and less stable than nuclear tRNAs (80).
Collectively, these results suggest that the accumulation
of deleterious mutations can lead to less stable secondary
structures of essential RNAs involved in translation.
Some selection for stabilization is also evident as
numerous compensatory base substitutions have been
fixed in the stem regions of both rRNAs (28) and
tRNAs (Figure 5). Alternatively, E. coli tRNAs may
possess higher %GC because its optimal growth tempera-
ture is higher than that of Buchnera (81), thus favoring
higher %GC for increased thermal stability.
During genome reduction, 72–78% of Buchnera tRNA
genes among all taxa have deleted 3bp, due to the loss of
30encoded CCA [Figure 5(a)]. Nevertheless, we found that
all mature Buchnera tRNAs process 30CCA, and therefore
they all have potential for amino acid activation [Figure
5(b)]. In all Buchnerataxa,
tRNAs process dual or triple 30CCA [Figure 5(b)].
These characteristics, in addition to 50G at the 1st and
2nd position and instability of the acceptor stem, result in
tRNA degradation (67). Interestingly, these tRNAs in
Buchnera and E. coli transcribe 50G at the first and
second base position and process dual or triple 30CCA
[Figure 5 (c)]. In these mature tRNAs in both E. coli
and Buchnera, the second to last 30CCA is always
incorporated into the 30acceptor stem. Potentially, the
retention of encoded 50G at N1 and N2 and the conser-
vation of dual and triple 30CCA maturation in these
tRNAs [Figure 5 (c)] are essential to maintain the
correct secondary structure and to police unstable
tRNAs via the tRNA degradation pathway.
In conclusion, our observations of altered tRNA
characteristics are consistent with the hypothesis that
translational fidelity is lower in Buchnera compared with
free-living relatives as represented by E. coli. First,
Buchnera genome reduction has resulted in the loss of
specific tRNA isoacceptors and modified nucleoside
pathways that may reduce translational efficiency and
fidelity. Second, Buchnera’s A+T mutational bias and
reduced selection has resulted in the reduction of tRNA
stability in vitro and specific tRNA base substitutions that
may alter the efficiency of aaRS recognition. Moreover,
reduced translational efficiency was supported by the lack
of relationship between codon usage of highly expressed
Nevertheless, purifying selection appears to be strong
enough in Buchnera genomes to maintain high %GC of
tRNA genes relative to CDS. Also, CCA 30maturation of
shortened tRNA genes, and numerous compensatory base
substitutions in tRNA stems help maintain tRNA second-
ary structure and function. Consequently, we predict that
Nucleic Acids Research, 2012,Vol.40, No. 16 7881
the translational efficiency and fidelity evident in Buchnera
are in an intermediate state between free-living bacteria
All raw sense and anti-sense tRNA data were submitted to
NCBI Genbank under SRA Submission: SRA049863.3,
(ii) PRJNA82809, (iii) PRJNA82797, (iv) PRJNA82793
and (v) PRJNA82789.
Supplementary Data are available at NAR Online:
Supplementary Tables 1 and 2, Supplementary Figure 1
and Supplementary Dataset 1.
The authors thank Kim Hammond for rearing aphids and
Dieter So ¨ ll, Jiqiang Ling, Patrick O’Donoghue and
Markus Englert for helpful discussions and feedback on
tRNA data. Also, they also thank Yogeshwar Kelkar,
Rahul Raghavanand Patrick
comments on the manuscript and thank four anonymous
reviewers for their helpful comments and suggestions.
Funding for open access charge: US Department of
Agriculture [2011-67012-30707 to A.H.].
Conflict of interest statement. None declared.
1. Osawa,S., Jukes,T.H., Watanabe,K. and Muto,A. (1992) Recent
evidence for evolution of the genetic code. Microbiol. Rev., 56,
2. Yokoyama,S. and Nishimura,S. (1995) Modified nucleosides and
codon recognition. In: So ¨ ll,D. and RajBhandary,U.L. (eds),
tRNA, Stucture, Biosynthesis and function. American Society of
Microbiology, Washington, DC, USA, pp. 207–223.
3. Martin,N.C. (1995) Organellar tRNAs: Biosynthesis and
Function. In: So ¨ ll,D. and RajBhandary,U.L. (eds), tRNA,
Stucture, Biosynthesis and function. American Society of
Microbiology, Washington, D.C., pp. 127–140.
4. Watanabe,K. and Osawa,S. (1995) tRNA Sequences and Variation
in the Genetic Code. American Society of Microbiology,
Washington, DC, USA.
5. Gray,M.W., Burger,G. and Lang,B.F. (1999) Mitochondrial
evolution. Science, 283, 1476–1481.
6. Timmis,J.N., Ayliffe,M.A., Huang,C.Y. and Martin,W. (2004)
Endosymbiotic gene transfer: organelle genomes forge eukaryotic
chromosomes. Nat. Rev. Genet., 5, 123–135.
7. Lynch,M. (2007) The Origins of Genome Architecture. Sinauer
Associates, Inc., Sunderland, MA.
8. Schneider,A. and Mare ´ chal-Drouard,L. (2000) Mitochondrial
tRNA import: are there distinct mechanisms? Trends Cell Biol.,
9. Mercer,T.R., Neph,S., Dinger,M.E., Crawford,J., Smith,M.A.,
Shearwood,A.M., Haugen,E., Bracken,C.P., Rackham,O.,
Stamatoyannopoulos,J.A. et al. (2011) The human mitochondrial
transcriptome. Cell, 146, 645–658.
10. McCutcheon,J.P. and Moran,N.A. (2012) Extreme
genome reduction in symbiotic bacteria. Nat. Rev. Microbiol.,
11. Theissen,U. and Martin,W. (2006) The difference between
organelles and endosymbionts. Curr. Biol., 16, R1016–R1017.
12. Keeling,P.J. and Archibald,J.M. (2008) Organelle evolution:
what’s in a name? Curr. Biol., 18, R345–R347.
13. McCutcheon,J.P. (2010) The bacterial essence of tiny symbiont
genomes. Curr. Opin. Microbiol., 13, 73–78.
14. Schattner,P., Brooks,A.N. and Lowe,T.M. (2005) The
tRNAscan-SE, snoscan and snoGPS web servers for the detection
of tRNAs and snoRNAs. Nucleic Acids Res., 33, W686–W689.
15. Lowe,T.M. and Eddy,S.R. (1997) tRNAscan-SE: a program for
improved detection of transfer RNA genes in genomic sequence.
Nucleic Acids Res., 25, 955–964.
16. McCutcheon,J.P., McDonald,B.R. and Moran,N.A. (2009)
Convergent evolution of metabolic roles in bacterial co-symbionts
of insects. Proc. Natl Acad. Sci. USA, 106, 15394–15399.
17. McCutcheon,J.P. and Moran,N.A. (2010) Functional convergence
in reduced genomes of bacterial symbionts spanning 200 million
years of evolution. Genome Biol. Evol., 2, 708–718.
18. Helm,M., Brule ´ ,H., Friede,D., Giege ´ ,R., Pu ¨ tz,D. and Florentz,C.
(2000) Search for characteristic structural features of mammalian
mitochondrial tRNAs. RNA, 6, 1356–1379.
19. Moran,N.A. (1996) Accelerated evolution and Muller’s rachet in
endosymbiotic bacteria. Proc. Natl Acad. Sci. USA, 93,
20. Mira,A. and Moran,N.A. (2002) Estimating population size and
transmission bottlenecks in maternally transmitted endosymbiotic
bacteria. Microb. Ecol., 44, 137–143.
21. Baumann,P., Baumann,L., Lai,C.-Y., Rouhbakhsh,D.,
Moran,N.A. and Clark,M.A. (1995) Genetics, physiology, and
evolutionary relationships of the genus Buchnera: intracellular
symbionts of aphids. Annu. Rev. Microbiol., 49, 55–94.
22. Shigenobu,S., Watanabe,H., Hattori,M., Sakaki,Y. and
Ishikawa,H. (2000) Genome sequence of the endocellular bacterial
symbiont of aphids Buchnera sp. APS. Nature, 407, 81–86.
23. Tamas,I., Klasson,L., Canback,B., Naslund,A.K., Eriksson,A.S.,
Wernegreen,J.J., Sandstrom,J.P., Moran,N.A. and Andersson,S.G.
(2002) 50 million years of genomic stasis in endosymbiotic
bacteria. Science, 296, 2376–2379.
24. Pe ´ rez-Brocal,V., Gil,R., Ramos,S., Lamelas,A., Postigo,M.,
Michelena,J.M., Silva,F.J., Moya,A. and Latorre,A. (2006)
A small microbial genome: The end of a long symbiotic
relationship? Science, 314, 312–313.
25. Moran,N.A., McLaughlin,H.J. and Sorek,R. (2009) The dynamics
and time scale of ongoing genomic erosion in symbiotic bacteria.
Science, 323, 379–382.
26. Lamelas,A., Gosalbes,M.J., Manzano-Marı´n,A., Pereto ´ ,J.,
Moya,A. and Latorre,A. (2011) Serratia symbiotica from the
aphid Cinara cedri: A missing link from facultative to obligate
insect endosymbiont. PLoS Genet., 7, e1002357.
27. Degnan,P.H., Ochman,H. and Moran,N.A. (2011) Sequence
conservation and functional constraint on intergenic spacers in
reduced genomes of the obligate symbiont Buchnera. PLoS
Genet., 7, e1002252.
28. Lambert,J.D. and Moran,N.A. (1998) Deleterious mutations
destabilize ribosomal RNA in endosymbiotic bacteria.
Proc. Natl Acad. Sci. USA, 95, 4458–4462.
29. Wilcox,J.L., Dunbar,H.E., Wolfinger,R.D. and Moran,N.A.
(2003) Consequences of reductive evolution for gene
expression in an obligate endosymbiont. Mol. Microbiol., 48,
30. Moran,N.A., Dunbar,H.E. and Wilcox,J.L. (2005) Regulation of
transcription in a reduced bacterial genome: nutrient-provisioning
genes of the obligate symbiont Buchnera aphidicola. J. Bacteriol.,
31. Tamas,I., Wernegreen,J.J., Nystedt,B., Kauppinen,S.N.,
Darby,A.C., Gomez-Valero,L., Lundin,D., Poole,A.M. and
Andersson,S.G. (2008) Endosymbiont gene functions impaired
and rescued by polymerase infidelity at poly(A) tracts.
Proc. Natl Acad. Sci. USA, 105, 14934–14939.
7882Nucleic Acids Research, 2012,Vol.40, No. 16
32. Giege,R., Sissler,M. and Florentz,C. (1998) Universal rules and
idiosyncratic features in tRNA identity. Nucleic Acids Res., 26,
33. Moran,N.A., Degnan,P.H., Santos,S.R., Dunbar,H.E. and
Ochman,H. (2005) The players in a mutualistic symbiosis: insects,
bacteria, viruses, and virulence genes. Proc. Natl Acad. Sci. USA,
34. Hansen,A.K. and Moran,N.A. (2011) Aphid genome expression
reveals host-symbiont cooperation in the production of amino
acids. Proc. Natl Acad. Sci. USA, 108, 2849–2854.
35. Rutherford,K., Parkhill,J., Crook,J., Horsnell,T., Rice,P.,
Rajandream,M.A. and Barrell,B. (2000) Artemis: sequence
visualization and annotation. Bioinformatics, 16, 944–945.
36. Rice,P., Longden,I. and Bleasby,A. (2000) EMBOSS: the
European Molecular Biology Open Software Suite. Trends Genet.,
37. Charles,H., Calevro,F., Vinuelas,J., Fayard,J.M. and Rahbe,Y.
(2006) Codon usage bias and tRNA over-expression in Buchnera
aphidicola after aromatic amino acid nutritional stress on its host
Acyrthosiphon pisum. Nucleic Acids Res., 34, 4583–4592.
38. Puigbo,P., Bravo,I.G. and Garcia-Vallve,S. (2008) E-CAI: a novel
server to estimate an expected value of Codon Adaptation Index
(eCAI). BMC Bioinformatics, 9, 65.
39. Altschul,S.F., Madden,T.L., Schaffer,A.A., Zhang,J., Zhang,Z.,
Miller,W. and Lipman,D.J. (1997) Gapped BLAST and
PSI-BLAST: a new generation of protein database search
programs. Nucleic Acids Res., 25, 3389–3402.
40. Marck,C. and Grosjean,H. (2002) tRNomics: analysis of tRNA
genes from 50 genomes of eukarya, archaea, and bacteria reveals
anticodon-sparing strategies and domain-specific features. RNA, 8,
41. Chan,P.P. and Lowe,T.M. (2009) GtRNAdb: a database of
transfer RNA genes detected in genomic sequence.
Nucleic Acids Res., 37, D93–D97.
42. Iida,K., Jin,H. and Zhu,J.-K. (2009) Bioinformatics analysis
suggests base modifications of tRNAs and miRNAs in
Arabidopsis thaliana. BMC Genomics, 10, 155.
43. Findeiß,S., Langenberger,D., Stadler,P. and Hoffmann,S. (2011)
Traces of post-transcriptional RNA modifications in deep
sequencing data. Biol. Chem., 392, 305–313.
44. Dunin-Horkawicz,S., Czerwoniec,A., Gajda,M.J., Feder,M.,
Grosjean,H. and Bujnicki,J.M. (2006) MODOMICS: a database
of RNA modification pathways. Nucleic Acids Res., 34,
45. Keseler,I.M., Collado-Vides,J., Santos-Zavaleta,A., Peralta-Gil,M.,
Gama-Castro,S., Muniz-Rascado,L., Bonavides-Martinez,C.,
Paley,S., Krummenacker,M., Altman,T. et al. (2011) EcoCyc: a
comprehensive database of Escherichia coli biology.
Nucleic Acids Res., 39, D583–D590.
46. Nawrocki,E.P., Kolbe,D.L. and Eddy,S.R. (2009) Infernal
1.0: inference of RNA alignments. Bioinformatics, 25, 1335–1337.
47. Seibel,P., Muller,T., Dandekar,T., Schultz,J. and Wolf,M. (2006)
4SALE–A tool for synchronous RNA sequence
and secondary structure alignment and editing.
BMC Bioinformatics, 7, 498.
48. Hofacker,I.L., Fekete,M. and Stadler,P.F. (2002) Secondary
structure prediction for aligned RNA sequences. J. Mol. Biol.,
49. Bernhart,S., Hofacker,I., Will,S., Gruber,A. and Stadler,P. (2008)
RNAalifold: improved consensus structure prediction for RNA
alignments. BMC Bioinformatics, 9, 474.
50. Crick,F.H. (1966) Codon—anticodon pairing: the wobble
hypothesis. J. Mol. Biol., 19, 548–555.
51. Nishimura,S. (1997) Transfer RNA: structure, properties and
recognition. Cold Spring Harbor Laboratory Press, Cold Spring
52. Wilson,R.K. and Roe,B.A. (1989) Presence of the hypermodified
nucleotide N6-(delta 2-isopentenyl)-2-methylthioadenosine prevents
codon misreading by Escherichia coli phenylalanyl-transfer RNA.
Proc. Natl Acad. Sci. USA, 86, 409–413.
53. Urbonavic ˇ ius,J., Qian,Q., Durand,J.M., Hagervall,T.G. and
Bjork,G.R. (2001) Improvement of reading frame maintenance is
a common function for several tRNA modifications. EMBO J.,
54. Qian,Q., Curran,J.F. and Bjo ¨ rk,G.R. (1998) The methyl group of
the N 6-methyl-N 6-threonylcarbamoyladenosine in tRNA of
Escherichia coli modestly improves the efficiency of the tRNA.
J. Bacteriol., 180, 1808–1813.
55. Wolf,J., Gerber,A.P. and Keller,W. (2002) tadA, an essential
tRNA-specific adenosine deaminase from Escherichia coli.
EMBO J., 21, 3841–3851.
56. Spanjaard,R.A., Chen,K., Walker,J.R. and van Duin,J. (1990)
Frameshift suppression at tandem AGA and AGG codons by
cloned tRNA genes: assigning a codon to argU tRNA and T4
tRNAArg. Nucleic Acids Res., 18, 5031–5036.
57. Harada,F. and Nishimura,S. (1972) Possible anticodon sequences
of tRNA His, tRNA Asn, and tRNA Asp from Escherichia coli
B. Universal presence of nucleoside Q in the first position of the
anticodon of three transfer ribonucleic acids. Biochemistry, 11,
58. Stern,L. and Schulman,L.H. (1978) The role of the minor base
N4-acetylcytidine in the function of the Escherichia coli
noninitiator methionine transfer RNA. J. Biol. Chem., 253,
59. Muramatsu,T., Nishikawa,K., Nemoto,F., Kuchino,Y.,
Nishimura,S., Miyazawa,T. and Yokoyama,S. (1988) Codon and
amino-acid specificities of a transfer RNA are both converted by
a single post-transcriptional modification. Nature, 336, 179–181.
60. Wittig,B. and Wittig,S. (1978) Reverse transcription of tRNA.
Nucleic Acids Res., 5, 1165–1178.
61. Weissenbach,J. and Grosjean,H. (1981) Effect of
threonylcarbamoyl modification (t6A) in yeast tRNAArgIII on
codon-anticodon and anticodon-anticodon interactions.
Eur. J. Biochem., 116, 207–213.
62. Bjo ¨ rk,G.R., Wikstro ¨ m,P.M. and Bystro ¨ m,A.S. (1989) Prevention
of translational frameshifting by the modified nucleoside
1-methylguanosine. Science, 244, 986–989.
63. Hagervall,T.G., Tuohy,T.M.F., Atkins,J.F. and Bjo ¨ rk,G.R. (1993)
Deficiency of 1-methylguanosine in tRNA from Salmonella
typhimurium induces frameshifting by quadruplet translocation.
J. Mol. Biol., 232, 756–765.
64. Ikemura,T. (1981) 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
translational system. J. Mol. Biol., 151, 389–409.
65. Ikemura,T. (1985) Codon usage and tRNA content in unicellular
and multicellular organisms. Mol. Biol. Evol., 2, 13–34.
66. Mira,A., Ochman,H. and Moran,N.A. (2001) Deletional bias and
the evolution of bacterial genomes. Trends Genet., 17, 589–596.
67. Wilusz,J.E., Whipple,J.M., Phizicky,E.M. and Sharp,P.A. (2011)
tRNAs marked with CCACCA are targeted for degradation.
Science, 334, 817–821.
68. Osawa,S. and Jukes,T.H. (1988) Evolution of the genetic code as
affected by anticodon content. Trends Genet., 4, 191–198.
69. Kano,A., Andachi,Y., Ohama,T. and Osawa,S. (1991) Novel
anticodon composition of transfer RNAs in Micrococcus luteus,
a bacterium with a high genomic G+C content: correlation with
codon usage. J. Mol. Biol., 221, 387–401.
70. Andachi,Y., Yamao,F., Muto,A. and Osawa,S. (1989) Codon
recognition patterns as deduced from sequences of the complete
set of transfer RNA species in Mycoplasma capricolum:
resemblance to mitochondria. J. Mol. Biol., 209, 37–54.
71. Rogalski,M., Karcher,D. and Bock,R. (2008) Superwobbling
facilitates translation with reduced tRNA sets.
Nat. Struct. Mol. Biol., 15, 192–198.
72. Sprinzl,M., Hartmann,T., Weber,J., Blank,J. and Zeidler,R. (1989)
Compilation of tRNA sequences and sequences of tRNA genes.
Nucleic Acids Res., 17, R1–R172.
73. Hou,Y.M., Westhof,E. and Giege ´ ,R. (1993) An unusual RNA
tertiary interaction has a role for the specific aminoacylation of
a transfer RNA. Proc. Natl Acad. Sci. USA, 90, 6776–6780.
74. Hamann,C.S. and Hou,Y.-M. (1997) An RNA
structural determinant for tRNA recognition. Biochemistry, 36,
75. Tamura,K., Asahara,H., Himeno,H., Hasegawa,T. and
Shimizu,M. (1991) Identity elements of Escherichia coli
tRNA(Ala). J. Mol. Recognit, 4, 129–132.
Nucleic Acids Research, 2012,Vol.40, No. 167883
76. Shimizu,M., Asahara,H., Tamura,K., Hasegawa,T. and
Himeno,H. (1992) The role of anticodon bases and the
discriminator nucleotide in the recognition of some E. coli tRNAs
by their aminoacyl-tRNA synthetases. J. Mol. Evol., 35, 436–443.
77. Himeno,H., Hasegawa,T., Ueda,T., Watanabe,K. and Shimizu,M.
(1990) Conversion of aminoacylation specificity from tRNATyr to
tRNASer in vitro. Nucleic Acids Res., 18, 6815–6819.
78. Rogers,M.J. and So ¨ ll,D. (1988) Discrimination between
glutaminyl-tRNA synthetase and seryl-tRNA synthetase
involves nucleotides in the acceptor helix of tRNA.
Proc. Natl Acad. Sci. USA, 85, 6627–6631.
79. Sampson,J.R. and Saks,M.E. (1993) Contributions of discrete
tRNASer domains to aminoacylation by E. coli seryl-tRNA
synthetase: a kinetic analysis using model RNA substrates.
Nucleic Acids Res., 21, 4467–4475.
80. Lynch,M. (1996) Mutation accumulation in transfer RNAs:
molecular evidence for Muller’s ratchet in mitochondrial genomes.
Mol. Biol. Evol., 13, 209–220.
81. Dunbar,H., Wilson,A., Ferguson,N. and Moran,N. (2007)
Aphid thermal tolerance is governed by a point mutation in
bacterial symbionts. PLoS Biol., 5, e96.
7884 Nucleic Acids Research, 2012,Vol.40, No. 16