The architecture and complete sequence of mitochondrial genome of an assassin bug Agriosphodrus dohrni (Hemiptera: Reduviidae).

The 16, 470 bp nucleotide sequence of the mitochondrial genome (mitogenome) of an assassin bug from the reduviid subfamily Harpactorinae, Agriosphodrus dohrni, has been revealed. The entire genome encodes for two ribosomal RNA genes (rrnL and rrnS), 22 transfer RNA (tRNA) genes, 13 protein-coding genes, and a control region. The nucleotide composition is biased toward adenine and thymine (A+T = 72.2%). Comparative analysis with two other reduviid species Triatoma dimidiata and Valentia hoffmanni, exhibited highly conserved genome architectures including genome contents, gene order, nucleotide composition, codon usage, amino acid composition, as well as genome asymmetry. All protein-coding genes use standard mitochondrial initiation codons (methionine and isoleucine), except that nad1 starts with GTG. All tRNAs have the classic clover-leaf structure, except that the dihydrouridine (DHU) arm of tRNA(Ser(AGN)) forms a simple loop. Secondary structure comparisons of the two mitochondrial ribosomal subunits among sequenced assassin bugs show that the sequence and structure of rrnL is more conservative than that of rrnS. The presence of structural elements in the control region is also discussed, with emphasis on their implications in the regulation of replication and/or transcription of the reduviid mitogenome. The phylogenetic analyses indicated that within Reduviidae, Harpactorinae is a sister group to the Salyavatinae + Triatominae clade.

Introduction

Mitogenomes have become a major resource for comparative genomics and play an important role in metabolism, apoptosis, disease and aging 1. In insects, mtDNA is typically a small double-stranded circular molecule of 14-20 kb in length. It encodes 13 protein-coding genes (PCGs), two rRNA genes, and 22 tRNA genes 1, 2. Additionally, insect mitogenome contains a major non-coding region known as the A+T-rich region that plays a role in initiation of transcription and replication 2. The length of this region is highly variable among different insects due to its high rate of nucleotide substitution, insertions/deletions, and the presence of a variable number of tandem repeats 3, 4. For the past two decades, mtDNA has been widely regarded as the molecular marker of choice for the phylogenetic analysis in metazoans because of its abundance in animal tissues, the small genome size, faster rate of evolution, low or absence of sequence recombination, and evolutionary conserved gene products 5-7, although there are some criticisms on using mtDNA for phylogenetics, especially at the deeper level 8, 9.

To date, mitogenomes of 236 insect species across 22 orders have been sequenced and deposited in the GenBank and the insect mitogenomes have been used extensively as molecular markers for population genetics, phylogeographic analyses, and the reconstruction of phylogenetic relationships at different taxonomic levels. Reduviidae, the largest family of terrestrial predatory Hemiptera, consists of approximately 7000 species 10. Both mitochondrial and nuclear ribosomal genes have been used as molecular markers to resolve the phylogenetic relationship in Reduviidae 10, 11. However, the number of subfamilies and their phylogenetic relationship are still at the point of juncture because the resolution is rather limited using specific gene markers 10. The utility of mitogenome data may provide new insights into reduviid higher-level systematics. Up to date, mitogenomes of two assassin bugs, Triatoma dimidiata from subfamily Triatominae 12 and Valentia hoffmanni from subfamily Salyavatinae 13, have been sequenced. T. dimidiata is a medically important insect pest species. It feeds on vertebrate blood and has been known as the vector of Chagas' disease in humans and other mammals.

In this report, we present the complete mitogenome of Agriosphodrus dohrni, a representative of Harpactorinae (the largest subfamily of Reduviidae) and a potential biocontrol agent in orchards and forests 14. Additionally, we compared the sequence and genome architecture of A. dohrni with blood-feeding T. dimidiata and V. hoffmanni. Finally, phylogenetic analysis of the three known reduviid species (including A. dohrni from this study) with representative heteropterans at the mitogenome level is provided to reexamine the feasibility of mitogenome data to resolve infraordinal relationships of Heteroptera.

Materials and Methods

Agriosphodrus dohrni and genomic DNA extraction

Adult A. dohrni samples were collected from Hanzhong, Shaanxi Province, China in May 2009. All collections were initially preserved in 95% ethanol in the field, and transferred to -20℃ for the long-term storage upon the arrival at the China Agricultural University (CAU). The genomic DNA was extracted from muscle tissues of the thorax using a cetyl-trimethylammonium bromide (CTAB)-based method 15. Voucher specimen (No. VHem-00201) was deposited at the Entomological Museum of CAU.

PCR amplification, cloning and sequencing

The mitogenome of A. dohrni was generated by amplification of overlapping PCR fragments (. Initially, 13 fragments were amplified using the universal primer sets from 6 (Fig. . Nine perfectly matched primers were designed based on the read of these short fragments for the secondary PCRs (.

Short PCRs (< 1.5 kb) were carried out using Qiagen Taq DNA polymerase (Qiagen, Beijing, China) with the following cycling conditions: 5 min at 94℃, followed by 35 cycles of 50 s at 94℃, 50 s at 48-55℃, 1-2 min at 72℃ depending on the size of amplifications, and the subsequent final elongation step at 72℃ for 10 min. Long PCRs (> 1.5 kb) were performed using NEB Long Taq DNA polymerase (New England BioLabs, Ipswich, MA) under the following cycling conditions: 30s at 95℃, followed by 45 cycles of 10 s at 95℃, 50s at 48-55℃, 3-6 min at 68℃ depending on the size of amplicons, and the final elongation step at 68℃ for 10 min . The quality of PCR products were evaluated by spectrophotometry and agarose gel electrophoresis.

The PCR fragments were ligated into the pGEM-T Easy Vector (Promega) and resulting plasmid DNAs were isolated using the TIANprp Midi Plasmid Kit (Qiagen, Beijing, China). All fragments were sequenced in both directions using the BigDye Terminator Sequencing Kit (Applied Bio Systems) and the ABI 3730XL Genetic Analyzer (PE Applied Biosystems, San Francisco, CA, USA) with two vector-specific primers and internal primers for primer walking.

Secondary structure prediction and sequence analysis

DNA sequences were proof-read and aligned into contigs in BioEdit version 7.0.5.3 16. Protein-coding regions and rRNA genes were identified by sequence homology with published insect mitochondrial sequences from public domains (e.g., GenBank). The tRNA genes were identified by tRNAscan-SE Search Server v.1.21 17 with default setting. Some tRNA genes that could not be determined by tRNAscan-SE were identified by comparing to the tRNA coding regions in other hemipterans. Secondary structures of the small and large subunits of rRNAs were inferred using models predicted for Drosophila melanogaster and D. virilis 18, Apis mellifera 19, Manduca sexta 20 and Ruspolia dubia 21. Stem-loops were named according to the convention of Gillespie et al. (2006) 19, as well as Cameron et al. (2008) 20. Protein-coding genes were aligned with Clustal X 22. A+T content and codon usage were calculated using MEGA version 4.0 23. The putative control region was determined using the mfold web server at http://mfold.rna.albany.edu/ 24 to locate regions with potential inverted repeats or palindromes. Strand asymmetry was calculated using the formulas: AT skew= [A−T]/ [A+T] and GC skew= [G−C]/ [G+C] 25.

Phylogenetic analysis

Phylogenetic analysis was carried out based on the 33 complete or nearly complete mitogenomes of true bugs from GenBank ().Two species from Auchenorrhyncha were selected as outgroups. A DNA alignment was inferred from the amino acid alignment of the thirteen protein-coding genes using MEGA version 4.0 23, which can translate between DNA and amino acid sequences within alignments. Alignments of individual genes were then concatenated excluding the stop codon. Model selection was done with MrModeltest 2.3 26 and ModelTest 3.7 27 for Bayesian inference and ML analysis, respectively. According to the Akaike information criterion, the GTR+I+G model was optimal for analysis with nucleotide alignments. MrBayes Version 3.1.1 28 and a PHYML online web server 29 were employed to analyze this data set under the GTR+I+G model. In Bayesian inference, two simultaneous runs of 1,000,000 generations were conducted for the matrix. Each set was sampled every 200 generations with a burnin of 25%. Trees inferred prior to stationarity were discarded as burn-in, and the remaining trees were used to construct a 50% majority-rule consensus tree. In ML analysis, the parameters were estimated during analysis and the node support values were assessed by bootstrap resampling (BP) 30 calculated using 100 replicates.

Results

The complete mitogenome of A. dohrni was sequenced and determined to be 16, 470 bp in size and it contains 13 protein-coding genes, 22 tRNA genes, two rRNA genes, and a control region (Fig.. The sequence was deposited in GenBank under the accession number HM071001.

Protein coding genes

All but one protein-coding genes of A. dohrni initiate with ATN as the start codon (eight with ATG, two with ATT, one with ATC and one with ATA) (Table . The only exception is the ND1 gene, which likely uses GTG as a start codon. Conventional stop codon (TAA) has been assigned to majority of the protein-coding genes of A. dohrni. COI, COIII and ND5, however, terminate with a single T residue which adjacent directly to the tRNA and non-coding region. Similar arrangement in which the termination codon is believed to be generated by the polyadenylation process has been observed for other insect species 31, 32.

Transfer RNAs

The whole set of 22 tRNAs typical of arthropod mitogenomes were found in A. dohrni, and schematic drawings of their respective secondary structures were shown in . A. dohrni tRNA genes fold into a classic clover-leaf structure, with the exception of tRNA, in which its dihydrouridine (DHU) arm simply forms a loop. Based on the secondary structure, a total of 25 unmatched base pairs are found in the A. dohrni tRNAs. Twenty-one of them are G-U pairs, which from a weak bond, locate in the AA stem (11-bp), the T stem (2-bp), and the DHU stem (8-bp). The remaining four pairs include U-U mismatches at the AA stem of tRNA and tRNA, respectively, U-C mismatche at the AA stem of tRNA, and A-A mismatch at the AC stem of tRNA ().

Ribosomal RNAs

Like other insect mitogenomes, the large and small rRNA subunits (rrnL and rrnS) in A. dohrni are located at tRNA - tRNA and tRNA - control region, respectively (Fig. . The length of rrnL and rrnS was determined to be 1, 268 bp and 796 bp, respectively. Both rrnL and rrnS are in congruent with the secondary structure models proposed for other insects 18-21, 40, 41. The secondary structure of A. dohrni rrnL consists of 6 structural domains (domain III is absent in arthropods) (Fig. . The structures of H991 region within the domain II were determined according to the models for M. sexta 20 and R. dubia 21, although they are highly variable and difficult to predict. Domain III of A. dohrni is apparently shorter than other insects. The secondary structure of rrnS contains three domains (Fig. . Helix-47 region is highly variable among different insects, and in A. dohrni can be folded into two helices, H47 and H47' (Fig. .

AT contents and codon usage

The nucleotide composition of A. dohrni mitogenome is biased toward adenine and thymine (A+T = 72.2%), ranging from 71.7% in protein-coding genes, 73.1% in rRNA genes, 73.4% in tRNA genes, to 71.9% in the control region (Table . The genome-wide bias toward AT was well documented in the codon usage table. At the third codon position, A or T are overwhelmingly overrepresented than G or C. The overall pattern was very similar among the three assassin bugs, with similar frequency of occurrences of various codons within a single codon family. There is a strong bias toward AT-rich codons with the five most prevalent codons in A. dohrni, as in order, TTA (Leu), TTT (Phe), ATT (Ile), ATA (Met) and AAT (Asn).

Non-coding region

The non-coding region of insect mitogenome consists of a control region and short intergenic spacers. In A. dohrni, a long control region (1, 643bp) and 14 intergenic spacers have been identified (Table . Majority of the intergenic spacers are only 1- to 13-bp long, however, the longest intergenic spacer located between tRNA and ND1 is 163 bp in length. It has two 58-bp tandem repeats and one 47-bp copy of the anterior portion of the repeat unit. This 163-bp intergenic spacer has no similarity to any existing sequences in the GenBank, and we suspected a similar function to the 314-bp intergenic spacer from T. dimidiata that is believed to be the other origin of replication 12, 42.

The control region of A. dohrni mitogenome is located at the conserved position between rrnS and tRNA gene cluster (Fig. . The A+T content of this region is one of the lowest in the A. dohrni mitogenome and can be divided into four parts: 1) a 410-bp region that is bordered by rrnS and a conserved region, of which the G+C content (43.4%) is higher than the average of the entire genome; 2) a short conserved sequence block (CSB) 3) a 188-bp region heavily biased toward A+T (79.2%); 4) a region composed of six long tandem repeats; and 5) the remainder of the control region (Fig. .

Phylogenetic relationships

ML and BI analysis performed on the nucleotide dataset generate similar tree topologies (Fig. . The five Pentatomomorpha superfamilies (15 taxa) are monophyletic with the following relationships: Aradoidea + (Pentatomoidea + (Pyrrhocoroidea + (Lygaeoidea + Coreoidea))). The sister groups' relationship of five Nepomorpha superfamilies are confirmed, but the position of Pleoidea is unstable. Two Leptopodomorpha superfamilies are monophyletic with the sister relationship to Nepomorpha. Cimicomorpha is paraphyletic consisting of two groups (Reduvioidea and (Cimicoidea and Miroidea)). Two Gerromorpha superfamilies are monophyletic in the basal position of these five infraorders. The infraordinal relationships tend to be poorly resolved with low support in ML analysis. The assassin bug subfamily Harpactorinae presents a sister position to the (Salyavatinae + Triatominae) highly supported by ML and Bayesian inferences.

Discussion

The organization and characteristics of A. dohrni mitogenome

The architecture of A. dohrni mitogenome including genome content, gene order, and genome asymmetry is consistent with two other reduviid species T. dimidiata and V. hoffmanni. The mitogenomes of the three assassin bugs share the same genome content (37 genes and 1 control region) and gene order, and have exactly the same genome asymmetry with a gene strand asymmetry (GSA) rate of 0.24 [GSA = (Number of genes on the major-strand - Number of genes on the minor-strand)/Number of total genes] 7. The size differences among reduviid mitogenomes (A. dohrni, 16, 470 bp; T. dimidiata, 17, 019 bp; and V. hoffmanni, 15, 625 bp) is mainly due to the variable number of repeats in the control regions. It is also worth noting that the evolutionary conserved genome architecture shared among assassin bugs is also considered ancestral to both insects and crustaceans 7, 33.

Among three assassin bugs, six of the protein-coding genes (COI, ATP6, COIII, ND4, ND6 and CytB) have the same Met start codons (ATG or ATA) (Table . The other five genes use either Met or Ile as the start codon (ATT, ATC, ATG and ATA). In T. dimidiata, the start codon for ND1 is ATA, whereas, A. dohrni and V. hoffmanni have GTG as their start codon. On the contrary, for ND5, T. dimidiata starts with GTG, whereas, A. dohrni and V. hoffmanni use ATG and ATT as their respective start codons (Table .

Dihydrouridine (DHU) arm of A. dohrni tRNAsimply forms a loop as seen in the two reduviid species T. dimidiata and V. hoffmanni as well as other insects 2, 13, 34-39. The anticodon (AC) stem of tRNA among three reduviid species is very conservative, with a long base pairing (9-bp instead of the common 5-bp) and a bulged nucleotide in the middle. The length of A. dohrni tRNAs ranges from 62-69 bp and is similar to the size of T. dimidiata (63-70 bp) and V. hoffmanni (59-70 bp). tRNAs from the three reduviid species possess invariable lengths for the aminoacyl (AA) stem (7-bp), the AC loop (7 nucleotides), and the AC stem (5-bp). Most of the size variations stem from the DHU and T arms, within which the loop size (3-9bp) is more variable than the stem size (3-5bp, except for tRNA. The A. dohrni anticodons are identical to those observed in T. dimidiata and V. hoffmanni.

The information regarding inferred rRNA secondary structures in assassin bugs and true bugs has been limited. Therefore, it is unclear how much variation exists in the rRNA structure and whether differences in length are due to the loss of helices or the reduction in helix sizes. Although the sizes of rrnL and rrnS in A. dohrni (1, 268 bp and 796 bp, respectively) are similar to those of T. dimidiata (1, 270 bp and 781 bp, respectively) and V. hoffmanni (1,256 bp and 777 bp, respectively), the sequence variation is too high in some regions for meaningful structural comparisons (Fig. . Overall, the 5' end before the helix H183, and domain-II and -V are the most variable regions of the rrnL, and their length and secondary structure of the loop are more conserved than the stem (Fig. . For rrnS, both 5' and 3' ends, and domain-I and -II are more variable. Three regions (Fig. of the V. hoffmanni rrnS which have lengthy base deletions were highlighted in blue and the new proposed structures were highlighted in purple (Fig. . Region-a, which was proposed to form a helix (H1047), with a 15-bp deletion, can be folded into a new helix (Fig. . In region-b, helix H1303 had a 12-bp deletion and led to the loss of its stem-loop structure (Fig. . In region-c, the 9-bp loop of helix H1399 reduced to 5-bp and formed a shorter loop (Fig. . To summarize, the size and structure of rrnL is more conserved than rrnS in three sequenced reduviids.

The AT bias is consistent in these three assassin bugs and the nucleotide skew statistics 25 for the whole mitogenome of these three species show that 1) the entire genome and control region are moderately A/C skewed; 2) protein-coding genes and codons lack significant G or C skew but moderate T skew; 3) rRNA genes are significant T/G skewed; and 4) tRNA genes are A/G skewed. Overall, the nucleotide composition of all three reduviid species representing three different subfamilies, respectively, is consistent and likely conserved in the family Reduviidae. It is worth noting that the control regions are not the most AT-rich region in the reduviid mitogenomes, and this generalized labeling of “AT-rich region” for the control region should be reconsidered 13.

The sequence alignment of all three assassin bug control regions reveals a CSB (Fig. , including a G element which has been reported in triatomine bugs Rhodnius prolixus and T. dimidiata (referred as Gs) 12, and some dipterans (referred as G islands) 43. After the CSB region, there is an A+T rich region which potentially can form the stem-loop structure (Fig. A possible involvement of this unique motif in insect replication and transcription initiation 43-45 is one of the interests for the future research.

The presence of various numbers of tandem repeats is one of the characteristics of the insect control region 46. In the case of T. dimidiata, the 2, 165 bp control region has eight tandem repeats including one 82-bp, five 140 bp, and two 173 bp repeats (Fig. . Six tandem repeats identified in A. dohrni include two 193 bp, one 41 bp (a partial copy of the anterior repeat unit), and three 194-bp repeats. However, in V. hoffmanni, the 725 bp control region only has two 38-bp repeats, separated by a non-repetitive sequence (Fig. . Repeated sequences are common in the control region for most insects, and length variations due to the various numbers of repeats are not without precedents 12. Consequently, analysis of the repeat units among individuals from different geographical locations may shed light on the geographical structuring and phylogenetic relationships of species.

Phylogenetic relationships among Heteroptera inferred from mitogenome sequences

Based on both morphological and molecular characterizations, previous studies support the contention that Reduviidae is monophyletic, whereas Reduviinae is polyphyletic in true bugs 10, 57, 58. However the number of subfamilies and the phylogenetic relationships within Reduviidae has remained a point of discussion. As the mitogenome sequencing becoming a common practice, the utility of the mitochondrial genome data for the resolution of subfamily-level relationships within Reduviidae is promising.

The seven-infraorder classification of the Heteroptera has been accepted by most researchers 49, 50, however, the phylogenetic relationships among infraorders are still controversial 47-51. The mitogenomes of 32 species in 29 families within five infraorders of Heteroptera have already been sequenced 12, 13, 52, 53 and those data provide a new source for understanding deep-level true bug phylogeny. Based on the analysis of nine nepomorphan mitogenomes, Hua 13 suggested elevating Pleoidea to the infraorder Plemorpha. Within Cimicomorpha, Reduvioidea is paraphyletic with respect to Cimicoidea and Miroidea in this study, and it is incongruent with previous results 54-56. These discrepancies suggest that the selection of representative taxa at family-level may influence the phylogenetic relationships within the infraorder resolved from the mitogenome data.

In the present study, the sister-relationship within Pentatomomorpha, Nepomorpha, Leptopodomorpha and Gerromorpha are highly supported by BI and ML analysis. In addition, the Gerromorpha clade is stable in the basal position. This result may provide evidence for the feasibility of mitogenome data to resolve infraordinal relationships of Heteroptera, however, the prerequisite is to ensure the integrality and representative of the infraorder-level taxa. Future analyses should focus on including Enicocephalomorpha and Dipsocoromorpha mitogenome data and additional representatives for some poorly sampled clades.

Table S1: Primer sequences used in this study; Table S2: General informatics of the taxa used in this study; Figure S1: Inferred secondary structure of 22 tRNAs of the A. dohrni mitogenome.

Click here for additional data file.

Table 1

Organization of the A. dohrni mitogenome

Gene	Direction	Location (bp)	Size (bp)	Anticodon	Start Codon	Stop Codon	Intergenic Nucleotidea
tRNAIle	F	1-64	64	30-32 GAT
tRNAGln	R	62-130	69	98-100 TTG			-3
tRNAMet	F	130-198	69	160-162 CAT			-1
ND2	F	199-1203	1005		ATT	TAA	0
tRNATrp	F	1202-1267	66	1232-1234 TCA			-2
tRNACys	R	1260-1327	68	1291-1293 GCA			-8
tRNATyr	R	1369-1434	66	1400-1402 GTA			41
COI	F	1446-2979	1534		ATG	T-	11
tRNALeu(UUR)	F	2980-3044	65	3009-3011 TAA			0
COII	F	3045-3728	684		ATC	TAA	0
tRNALys	F	3730-3799	70	3760-3762 CTT			1
tRNAAsp	F	3802-3867	66	3833-3835 GTC			2
ATP8	F	3869-4027	159		ATT	TAA	1
ATP6	F	4021-4704	684		ATG	TAA	-7
COIII	F	4691-5471	781		ATG	T-	-14
tRNAGly	F	5472-5534	63	5502-5504 TCC			0
ND3	F	5535-5888	354		ATA	TAA	0
tRNAAla	R	5889-5950	62	5918-5920 TGC			0
tRNAArg	F	5954-6019	66	5984-5986 TCG			3
tRNAAsn	F	6026-6093	68	6058-6060 GTT			6
tRNASer(AGN)	F	6093-6161	69	6120-6122 GCT			-1
tRNAGlu	F	6163-6227	64	6194-6196 TTC			1
tRNAPhe	R	6230-6294	65	6260-6262 GAA			2
ND5	R	6296-8006	1711		ATG	T-	1
tRNAHis	R	8008-8070	63	8038-8040 GTG			1
ND4	R	8071-9402	1332		ATG	TAA	0
ND4L	R	9396-9680	285		ATG	TAA	-7
tRNAThr	F	9687-9750	64	9718-9720 TGT			6
tRNAPro	R	9752-9817	66	9785-9787 TGG			1
ND6	F	9831-10328	498		ATG	TAA	13
CytB	F	10328-11464	1137		ATG	TAA	-1
tRNASer(UCN)	F	11474-11542	63	11504-11506TGA			9
ND1	R	11706-12629	924		GTG	TAA	163
tRNALeu(CUN)	R	12630-12694	65	12663-12665TAG			0
lrRNA	R	12695-13962	1268				0
tRNAVal	R	13963-14031	67	14000-14002 TAC			0
srRNA	R	14032-14827	796				0
Control region		14828-16470	1643				0

Table 2

Start and stop codons of the A. dohrni, T. dimidiata and V. hoffmanni mitogenomes

­­­­­­Gene	Start Codon			Stop Codon
	A. dohrni	T. dimidiata	V. hoffmanni	A. dohrni	T. dimidiata	V. hoffmanni
ND2	ATT	ATC	ATG	TAA	TAG	TAA
COI	ATG	ATG	ATG	T-	T-	T-
COII	ATC	ATA	ATA	TAA	T-	T-
ATP8	ATT	ATA	ATC	TAA	TAA	TAA
ATP6	ATG	ATG	ATG	TAA	TAA	TAG
COIII	ATG	ATG	ATG	T-	TA-	T-
ND3	ATA	ATA	ATT	TAA	TA-	T-
ND5	ATG	GTG	ATT	T-	TA-	T-
ND4	ATG	ATG	ATG	TAA	TAA	TAA
­­­­­­­­­ND4L	ATG	ATG	ATT	TAA	TAA	TAA
ND6	ATG	ATA	ATA	TAA	TAA	TAA
CytB	ATG	ATG	ATG	TAA	T-	TAG
ND1	GTG	ATA	GTG	TAA	TAA	TAA

Table 3

Nucleotide composition of the A. dohrni (A.), T. dimidiata (T.) and V. hoffmanni (V.) mitogenomes

Feature	%A+T			AT Skew			GC Skew
	A.	T.	V.	A.	T.	V.	A.	T.	V.
Whole genome	72.2	69.5	73.8	0.08	0.16	0.12	-0.12	-0.27	-0.21
protein-coding genes	71.7	68.8	73.4	-0.14	-0.17	-0.16	-0.005	-0.02	0.004
First codon position	73.5	69.2	74.6	-0.09	-0.13	-0.10	0.01	-0.03	0.01
Second codon position	70.0	68.0	72.3	-0.18	-0.24	-0.20	0.01	0.03	0.02
Third codon position	71.5	69.3	73.4	-0.16	-0.17	-0.16	-0.04	-0.08	-0.02
rRNA genes	74.8	73.1	75.7	-0.09	-0.18	-0.11	0.27	0.31	0.30
tRNA genes	73.4	74.3	76.0	0.02	0.03	0.01	0.11	0.15	0.14
Control region	71.9	66.0	69.9	0.05	0.20	0.03	-0.17	-0.29	-0.25