A comparative analysis of mitochondrial genomes in Coleoptera (Arthropoda: Insecta) and genome descriptions of six new beetles.

Coleoptera is the most diverse group of insects with over 360,000 described species divided into four suborders: Adephaga, Archostemata, Myxophaga, and Polyphaga. In this study, we present six new complete mitochondrial genome (mtgenome) descriptions, including a representative of each suborder, and analyze the evolution of mtgenomes from a comparative framework using all available coleopteran mtgenomes. We propose a modification of atypical cox1 start codons based on sequence alignment to better reflect the conservation observed across species as well as findings of TTG start codons in other genes. We also analyze tRNA-Ser(AGN) anticodons, usually GCU in arthropods, and report a conserved UCU anticodon as a possible synapomorphy across Polyphaga. We further analyze the secondary structure of tRNA-Ser(AGN) and present a consensus structure and an updated covariance model that allows tRNAscan-SE (via the COVE software package) to locate and fold these atypical tRNAs with much greater consistency. We also report secondary structure predictions for both rRNA genes based on conserved stems. All six species of beetle have the same gene order as the ancestral insect. We report noncoding DNA regions, including a small gap region of about 20 bp between tRNA-Ser(UCN) and nad1 that is present in all six genomes, and present results of a base composition analysis.

Introduction

Animal mitochondrial genomes (mtgenomes) are small, circular DNA with length ranging from 14,000 bp to 17,000 bp (Boore 1999; Cameron, Johnson, and Whiting 2007). They usually encode 37 genes (13 protein-coding, 22 transfer RNA, and 2 ribosomal RNA genes). The number of complete mtgenomes has steadily been on the rise with the technical feasibility of sequencing their entirety (Hwang et al. 2001; Yamauchi et al. 2004). This increasing availability of mtgenome data invites comparative study. In addition to the large amount of nucleotide data that is useful for deep-level phylogenetic studies (Gray et al. 1999; Nardi et al. 2003; Cameron et al. 2004; Cameron, Barker, and Whiting 2006; Cameron, Lambkin, et al. 2007), mtgenomes possess a number of evolutionarily interesting features such as length variation (Boyce et al. 1989), altered tRNA anticodons or secondary structures (Steinberg and Cedergren 1994; Eddy 2002), atypical start codons (e.g., Lavrov et al. 2000), base compositional bias (Gibson et al. 2004; Gowri-Shankar and Rattray 2006), codon usage (Jia and Higgs 2007), and gene rearrangement (Zhang and Hewitt 1997; Shao and Barker 2003; Mueller and Boore 2005). Some of these features appear to be lineage specific (Dowton et al. 2002); however, this insight can only be obtained from comparative analysis at various taxonomic levels.

Insect order Coleoptera contains over 360,000 described species divided into four suborders: Adephaga, Archostemata, Myxophaga, and Polyphaga (Lawrence and Newton 1982). Despite the size and diversity of the group, there are only six published (Tribolium, Crioceris, Pyrocoelia, two species of Rhagophthalmus, and Pyrophorus) and one unpublished (Anoplophora) beetle mtgenomes, all of which belong to suborder Polyphaga (Friedrich and Muqim 2003; Stewart and Beckenbach 2003; Bae et al. 2004; Li et al. 2007; Arnoldi et al. 2007) (table 1). The data from these seven mtgenomes suggest that the gene arrangement of Coleoptera follows that of the ancestral insect, that they all have a derived UCU anticodon and a reduced or missing D-stem in tRNA-Ser(AGN), and that they have atypical cox1 start codons (Friedrich and Muqim 2003; Stewart and Beckenbach 2003; Bae et al. 2004; Li et al. 2007; Arnoldi et al. 2007). However, there has not been an attempt to describe the possible diversity of mtgenomes across the beetle suborders.

Table 1

Taxonomic Information and Accession Numbers for the Coleopteran Taxa Used in this Study

Species	Classification	Accession	Voucher/Reference	Location
This study
Tetraphalerus bruchi Heller	Archostemata: Ommatidae	EU877953	IGC-CO687	Argentina
Trachypachus holmbergi Mannerheim	Adephaga: Trachypachidae	EU877954	IGC-CO843	Canada
Sphaerius sp.	Myxophaga: Sphaeriusidae	EU877950	IGC-CO837	United States
Chaetosoma scaritides Westwood	Polyphaga: Cleroidea: Chaetosomatidae	EU877951	IGC-CO683	New Zealand
Cyphon sp.	Polyphaga: Scirtoidea: Scirtidae	EU877949	IGC-CO838	United States
Priasilpha obscura Broun	Polyphaga: Cucujoidea: Priasilphidae	EU877952	IGC-CO684	New Zealand
Previously reported
Tribolium castaneum	Polyphaga Tenebrionoidae: Tenebrionidae	NC_003081	Friedrich and Muqim (2003)
Pyrocoelia rufa	Polyphaga: Elateroidea: Lampyridae	NC_003970	Bae et al. (2004)
Crioceris duodecimpunctata	Polyphaga: Chrysomeloidea: Chrysomelidae	NC_003372	Stewart and Beckenbach (2003)
Rhagophthalmus lufengensis	Polyphaga: Elateroidea: Phengodidae	DQ888607	Li et al. (2007)
Rhagophthalmus ohbai	Polyphaga: Elateroidea: Phengodidae	AB267275	Li et al. (2007)
Pyrophorus divergens	Polyphaga: Elateroidae: Elateridae	NC_009964	Arnoldi et al. (2007)
Anoplophora glabripennis	Polyphaga: Chrysomeloidea: Cerambycidae	NC_008221	Not applicable

In this paper, we present six new beetle mtgenome descriptions, including representatives of Archostemata, Adephaga, and Myxophaga, and three additional Polyphaga mtgenomes from superfamilies not represented in previous analyses. The comparison of mtgenomes from all four suborders provides unique insights into the evolution of the mtgenome. We use the available 13 coleopteran mtgenomes to highlight unique features and shared characteristics and to point out particular parts of the mtgenome that have caused problems for annotation. We present possible solutions for such difficulties based on the comparative information now available.

Materials and Methods

mtgenome Sequencing, Annotation, and Analysis

We extracted total genomic DNA using the DNeasy Tissue kit (Qiagen, Hilden, Germany). Prior to extraction, we removed the entire abdominal segment to avoid possible contamination from gut content and to retain taxonomically important genital structures as vouchers. In all species, we used the entire body without abdomen. We followed primer walking and polymerase chain reaction protocols described in Cameron, Lambkin, et al. (2007). The species-specific primers designed for this study are available upon request from H.S. Morphological voucher specimens and remaining genomic DNA extracts were deposited in the Insect Genomic Collection of the Department of Biology and MLBM Museum, Brigham Young University. Throughout this paper, we refer to all species by their generic name. GenBank accession numbers, specimen vouchers, classifications, and collecting localities are listed in table 1.

Raw sequence files were proofread and aligned into contigs in Sequencher 4.6 (GeneCodes Corporation, Ann Arbor, MI). We used the programs tRNAscan-SE (Lowe and Eddy 1997), INFERNAL (Eddy 2002), DOGMA (Wyman et al. 2004), and our unpublished software MOSAS to annotate the genomes. We located tRNAs with tRNAscan-SE with the default tRNA covariance model (CM), and we developed a new CM for coleopteran mitochondrial tRNA-Ser(AGN). We developed the new CM by creating an alignment of all 13 tRNA-Ser(AGN)s using INFERNAL's cmalign utility and modifying alignments by hand to eliminate the D-stem, in order to create a structural alignment more consistent with what is known about the structure of this tRNA. We used the COVE utility coveb (Eddy and Durbin 1994) to create a new CM and used this model to annotate tRNA-Ser(AGN). We also used the hand-curated INFERNAL alignment to infer a consensus secondary structure for tRNA-Ser(AGN). For questionable tRNAs, we used INFERNAL and Rfam to further investigate and sometimes revise tRNA annotation (Griffiths-Jones et al. 2005). DOGMA and MOSAS facilitate the annotation of organellar genomes by utilizing BLAST against published mtgenomes (e.g., Podsiadlowski 2006). After DOGMA and MOSAS reported general locations for genes based on similarity to other species, we identified start and stop codons to complete the annotation. The end of the small subunit rRNA (12S) was assigned by alignment with the secondary structures of 12S genes of other insects (Gillespie et al. 2006; Cameron and Whiting 2008). Helices were numbered according to the naming system of Gillespie et al. (2006).

For comparison to published genomes, we downloaded the published beetle mtgenome sequences from GenBank. For many of the alignments, such as aligning gap regions and tRNAs across beetle mtgenomes, we used MUSCLE (Edgar 2004). To determine the start codon of nad1 and cox1, we made an alignment based on the translated amino acid sequences using ClustalW (Thompson et al. 1994) as implemented in MEGA version 3 (Kumar et al. 2004). In order to compare base compositional profiles of the six new species, we calculated base composition by codon position for each gene individually.

Results and Discussion

mtgenome Organization and Gene Content

The present study reports six new beetle mtgenomes, including sequences belonging to all four suborders of Coleoptera (table 2). Complete mtgenome sequences were obtained for Tetraphalerus (15,689 bp) and Cyphon (15,919 bp). Entire coding sequences with a partial control region were obtained for Sphaerius (15735 bp), Chaetosoma (15,511 bp), Priasilpha (16,887 bp), and Trachypachus (15,991 bp). A comparison of the mtgenome size across all four suborders of Coleoptera based on this study and previous studies suggests that the size of the coding region in Coleoptera is relatively stable around 14,700 bp in length (large intergenic spacers can cause deviations from this pattern). Although the length of coding region is constrained in order for the genes to function properly, the A + T–rich control region, located between the small rRNA subunit (12S) and tRNA-Ile, is free from such functional constraints, and its length variation is considerable. Despite being incomplete, the control region of Priasilpha was still longer than that of any other complete beetle mtgenome previously reported (table 3). Based on a restriction site mapping of mtDNA, Boyce et al. (1989) found that the control region of bark weevil Pissodes was extremely large (9–13 kb) and reported considerable size variation in the control region of Curculionidae. The size of the control region is therefore not consistent within beetle lineages but varies across them (Zhang and Hewitt 1997).

Table 2

Nucleotide Positions and Anticodons (for tRNAs) for All Genes for Six New Beetle Species

Gene	Strand	Anticodon	Tetraphalerus	Trachypachus	Sphaerius	Cyphon	Chaetosoma	Priasilpha
tRNA-I	+	GAU	1–63 (0)	1–64 (0)	1–64 (0)	1–64 (0)	1–66 (0)	1–66 (0)
tRNA-Q	−	UUG	65–133 (1)	70–138 (5)	62–130 (−3)	62–130 (−3)	64–132 (−3)	64–132 (−3)
tRNA-M	+	CAU	138–207 (4)	141–209 (2)	132–202 (1)	129–197 (−2)	132–200 (−1)	132–200 (−1)
nad2	+		208–1231 (0)	210–1238 (0)	203–1228 (0)	198–1217 (0)	201–1206 (0)	201–1214 (0)
tRNA-W	+	UCA	1232–1296 (0)	1240–1308 (1)	1229–1296 (0)	1395–1462 (177)	1207–1269 (0)	1215–1282 (0)
tRNA-C	−	GCA	1289–1351 (−8)	1354–1418 (45)	1298–1360 (1)	1455–1516 (−8)	1274–1336 (4)	1282–1343 (−1)
tRNA-Y	−	GUA	1352–1417 (0)	1431–1499 (12)	1365–1432 (4)	1516–1581 (−1)	1336–1399 (−1)	1344–1407 (0)
cox1	+		1419–2949 (1)	1501–3031 (1)	1434–2964 (1)	1583–3113 (1)	1401–2931 (1)	1611–3144 (203)
tRNA-L	+	UAA	2950–3014 (0)	3032–3097 (0)	2965–3029 (0)	3114–3177 (0)	2932–2994 (0)	3145–3209 (0)
cox2	+		3015–3687 (0)	3101–3788 (3)	3030–3711 (0)	3178–3862 (0)	2995–3682 (0)	3210–3896 (0)
tRNA-K	+	CUU	3688–3758 (0)	3789–3859 (0)	3712–3782 (0)	3863–3933 (0)	3683–3752 (0)	3898–3968 (1)
tRNA-D	+	GUC	3758–3822 (−1)	3860–3925 (0)	3799–3865 (16)	3935–4001 (1)	3752–3813 (−1)	3968–4036 (−1)
atp8	+		3823–3981 (0)	3926–4090 (0)	3866–4024 (0)	4002–4163 (0)	3897–4052 (83)	4037–4192 (0)
atp6	+		3978–4652 (−4)	4087–4761 (−4)	4021–4695 (−4)	4160–4834 (−4)	4049–4717 (−4)	4189–4860 (−4)
cox3	+		4642–5423 (−11)	4762–5553 (0)	4695–5482 (−1)	4840–5619 (5)	4717–5500 (−1)	4862–5649 (1)
tRNA-G	+	UCC	5424–5488 (0)	5560–5625 (6)	5483–5546 (0)	5620–5683 (0)	5501–5563 (0)	5650–5712 (0)
nad3	+		5489–5840 (0)	5626–5977 (0)	5547–5898 (0)	5684–6035 (0)	5564–5915 (0)	5713–6064 (0)
tRNA-A	+	UGC	5841–5904 (0)	5978–6042 (0)	5899–5964 (0)	6036–6103 (0)	5916–5980 (0)	6065–6130 (0)
tRNA-R	+	UCG	5904–5970 (−1)	6042–6106 (−1)	5973–6037 (8)	6104–6167 (0)	5980–6042 (−1)	6130–6189 (−1)
tRNA-N	+	GUU	5968–6033 (−3)	6110–6174 (3)	6040–6105 (2)	6168–6232 (0)	6042–6107 (−1)	6190–6254 (0)
tRNA-S	+	GCU/UCUa	6035–6099 (1)	6174–6242 (−1)	6107–6171 (1)	6233–6299 (0)	6108–6166 (0)	6255–6321 (0)
tRNA-E	+	UUC	6101–6163 (1)	6243–6308 (0)	6174–6240 (2)	6301–6367 (1)	6167–6229 (0)	6322–6386 (0)
tRNA-F	−	GAA	6162–6227 (−2)	6307–6373 (−2)	6239–6306 (−2)	6366–6432 (−2)	6228–6290 (−2)	6385–6449 (−2)
nad5	−		6228–7953 (0)	6374–8102 (0)	6307–8024 (0)	6433–8155 (0)	6291–8001 (0)	6450–8163 (0)
tRNA-H	−	GUG	7951–8014 (−3)	8103–8170 (0)	8025–8089 (0)	8156–8219 (0)	8002–8065 (0)	8164–8230 (0)
nad4	−		8015–9338 (0)	8171–9509 (0)	8090–9425 (0)	8220–9549 (0)	8066–9386 (0)	8231–9560 (0)
nad4l	−		9332–9625 (−7)	9503–9796 (−7)	9419–9712 (−7)	9549–9839 (−1)	9389–9670 (2)	9557–9838 (−4)
tRNA-T	+	UGU	9628–9689 (2)	9799–9863 (2)	9715–9779 (2)	9842–9906 (2)	9674–9735 (3)	9843–9907 (4)
tRNA-P	−	UGG	9690–9756 (0)	9864–9930 (0)	9780–9845 (0)	9907–9972 (0)	9736–9798 (0)	9908–9973 (0)
nad6	+		9758–10273 (1)	9932–10456 (1)	9847–10356 (1)	9974–10492 (1)	9800–10288 (1)	9975–10478 (1)
cob	+		10273–11408 (−1)	10456–11590 (−1)	10356–11490 (−1)	10492–11626 (−1)	10288–11422 (−1)	10478–11615 (−1)
tRNA-S	+	UGA	11409–11474 (0)	11591–11657 (0)	11491–11557 (0)	11627–11693 (0)	11423–11489 (0)	11616–11683 (0)
nad1	−		11493–12440 (18)	11676–12626 (18)	11580–12530 (22)	11712–12662 (18)	11507–12460 (17)	11701–12651 (17)
tRNA-L	−	UAG	12442–12506 (1)	12628–12691 (1)	12532–12594 (1)	12664–12728 (1)	12462–12523 (1)	12653–12717 (1)
rrnL	−		12507–13828 (0)	12692–14012 (0)	12595–13909 (0)	12729–14025 (0)	12524–13805 (0)	12718–14000 (0)
tRNA-V	−	UAC	13829–13898 (0)	14013–14084 (0)	13910–13980 (0)	14026–14095 (0)	13806–13870 (0)	14001–14071 (0)
RrnS	−		13899–14689 (0)	14085–14872 (0)	13981–14764 (0)	14096–14876 (0)	13871–14649 (0)	14072–14859 (0)
control	Not applicable		14690–15689 (0)	14873–15991 (0)b	14765–15735 (0)b	14877–15919 (0)	14650–15511 (0)b	14860–16887 (0)b

NOTE.—Numbers in parenthesis represent the number of intergenic nucleotides before the gene starts.

This tRNA-S has a UCU anticodon for Cyphon, Chaetosoma, and Priasilpha and a GCU anticodon for Tetraphalerus, Trachypachus, and Sphaerius.

Incomplete control region.

Table 3

AT Content Comparison by mtgenome Region in Coleoptera

	Coding Region		Ribosomal RNAs		Control Region
Taxon	Size	AT%	Size	AT%	Size	AT%
Trachypachus	14,842	79.1	2,109	81.8	1,119a	84.9
Tetraphalerus	14,689	66.2	2,113	66.4	1,000	78.4
Sphaerius	14,764	80.4	2,099	83.8	953a	89.6
Cyphon	14,876	74.5	2,078	80.8	1,043	85.2
Chaetosoma	14,649	78.3	2,061	82.2	862a	91.0
Priasilpha	14,859	75.2	2,071	81.1	2,028a	87.0
Tribolium	14,642	70.8	2,054	76.1	1,239	82.5
Pyrocoelia	16,217	76.5	2,007	81.7	1,522	87.6
Crioceris	14,660	76.4	2,081	81.4	1,220	83.3
Rhagophthalmus	14,615	78.9	2,056	82.4	1,367	86.9
Pyrophorus	14,650	68.9	2,075	83.0	1,470	74.7
Anoplophora	14,659	77.6	2,148	80.0	1,115	88.0

Incomplete control region.

The six mtgenomes had varying degrees of high A + T content, ranging from 66.2% to 80.4% in the coding region and 78.4% to 91.0% in the control region (table 3). The A + T content of the control region was consistently higher than that of the coding region, which is a well-documented pattern in insect mtgenomes (Clary and Wolstenholme 1985; Zhang and Hewitt 1997). To understand what contributed to this variation in base composition, we examined the base frequency of the protein-coding genes by codon position (fig. 1). Overall, all six beetle mtgenomes followed similar compositional profiles, but Trachypachus (Adephaga) and Sphaerius (Myxophaga) exhibit extremely low C and G content in the third codon position. The overall A + T content of Tetraphalerus (Archostemata) is the lowest of all, and in this species, the C + G content is not as biased toward first and second codon positions. When the compositional profiles of individual protein-coding genes are examined, it becomes evident that a considerable amount of gene-specific variation exists (fig. 2). For instance, in Cyphon, the frequencies of A and T in each codon position are relatively stable, whereas those of G and C vary highly across the protein-coding genes. From these observations, we can hypothesize that there is considerable variability in nucleotide content not only among different species but also among genes and codon positions.

F

Base composition for all protein-coding genes combined. Each column is divided by codon position into three categories.

F

Individual base composition for each protein-coding gene in Cyphon. Each column is divided by codon position into three categories. To improve visibility, the columns are normalized so that they show proportions rather than counts at each codon position.

The six beetle species we sequenced, like the seven previously reported, retain the inferred ancestral gene complement for insects (Boore 1999). There were no rearrangements, duplications, or deletions of any genes within these mtgenomes. This suggests that there have not been significant gene rearrangements during the diversification of Coleoptera. Given the diversity of beetles, this molecular stability is a remarkable finding because most other major insect orders exhibit diagnostic rearrangements for major taxonomic groups (Dowton and Austin 1999; Thao et al. 2004; Castro et al. 2006; Cameron and Whiting 2008). In fact, only Diptera appears to be as conservative with respect to mtgenome structure as Coleoptera (Cameron, Lambkin, et al. 2007).

Noncoding DNA

In our annotations, many gene boundaries have been assigned to avoid the implications of noncoding intergenic spacers and gene overlaps. Mitochondrial evolution has traditionally been viewed as favoring genome size reduction (Rand 1993; Macey et al. 1997; McKnight and Shaffer 1997; Boore 1999), possibly by eliminating intergenic spacers (Burger et al. 2003). From an evolutionary perspective, it makes sense that nonfunctional intergenic spacers would be eliminated over time, especially in the highly reduced and efficient mtgenome. Sometimes intergenic spacers are reduced to the point of gene overlap. However, such cases appear to be the exception rather than the rule, due both to posttranscriptional complications (if abutting genes are encoded on the same strand) as well as the low probability that the nucleotides at the end of one gene are also useful as part of an abutting reversed gene. As such, we attempted to avoid both intergenic spacers and overlaps between genes on either strand of the genome in our annotation, but we did identify a number of intergenic spacer regions of variable size.

Although most spacers appeared to be unique to individual species (see below), a small intergenic region between the tRNA-Ser(UCN) and nad1 genes, ranging between 17 and 22 bp in length, was found in all six species. An intergenic spacer of this size at this location has been reported in other insects (e.g., Kim et al. 2006) and arthropods (e.g., Lavrov et al. 2000). Four of the six previously published beetle mtgenomes also have this intergenic spacer, which ranges between 16 and 20 bp in size, and only the two species of genus Rhagophthalmus lack it (Li et al. 2007). These latter two species have only about 4 bp in this region. However, we can attribute this disparity to insertions and deletions near the end of nad1 that may be the result of sequencing errors or correction by posttranslational modification. The Anoplophora sequence is also one nucleotide off, shifting the reading frame to avoid the conserved stop codon the other beetles use. In this case, the authors annotated the gene with a partial stop codon (T) to preserve the conserved spacer. According to Taanman (1999), this intergenic spacer region may correspond to the binding site of mtTERM, a transcription attenuation factor, as this position signifies the end of the major-strand coding region. Cameron and Whiting (2008) presented an alignment of several insect orders, highlighting a 7-bp motif (ATACTAA) conserved across Lepidoptera. When we aligned this region across all coleopteran mtgenomes, we found 5 bp (TACTA or its reverse complement TAGTA) to be conserved, and this region matches well the corresponding motif in Lepidoptera (Cameron and Whiting 2008; fig. 3).

F

An alignment of the gap region between tRNA-Ser(UCN) and nad1 in all coleopteran genomes. The box indicates a conserved pentanucleotide region (TACTA) across all beetles. The dotted line indicates the location of nad1 in Rhagophthalmus if the current annotation is correct.

In addition to small intergenic regions, there were larger spacer regions of varying A + T content found in different locations in several species (table 2). These regions had no tandem repeats, did not produce any significant BLAST results, did not fold like tRNAs, and did not include open reading frames in either direction, which suggest that they are likely noncoding and nonfunctional. Although noncoding intergenic spacer regions between coding genes have been reported for several insects (e.g., Crozier RH and YC Crozier 1993; Boore 1999; Dotson and Beard 2001; Bae et al. 2004; Cameron, Beckenbach, et al. 2006), their exact origin and function are often unclear. What is evident from this study is that these noncoding regions are lineage specific and common and not conserved at higher taxonomic levels within Coleoptera. Additional sampling will, however, be useful to determine if some of these noncoding regions are conserved across groups of closely related species.

Cox1 Start Codons

There has been much discussion of potential cox1 start codons in insects because the beginning of the open reading frame after tRNA-Tyr typically does not have the canonical ATN start codon (e.g., Nardi et al. 2003; Oliveira et al. 2005; Castro et al. 2006; Lee et al. 2006; Fenn et al. 2007; Cameron and Whiting 2008). AAA (lysine), ATT (isoleucine), CTA (leucine), and ATC (isoleucine) have all been proposed as possible start codons in Coleoptera. Without an explicit RNA expression study, it is impossible to determine exactly where cox1 starts; however, by aligning the region encompassing tRNA-Tyr and cox1 from all known beetle mtgenomes, we can more accurately determine theoretical start codons for the cox1 gene in Coleoptera (fig. 4). The possible traditional ATN start codons (isoleucine) near the beginning of cox1 lie either within tRNA-Tyr or 36 bp after the end of the tRNA-Tyr. We argue that it would be most logical to choose a start codon for cox1 that would minimize intergenic space and gene overlaps. The first nonoverlapping in-frame codon in cox1 is well conserved throughout all six divergent superfamilies within Polyphaga, and it is possible to choose asparagine (AAT or AAC) as a start codon. At the same site, Tetraphalerus and Sphaerius have glutamine (CAA) and Trachypachus has arginine (CGA). This start location is well conserved, located only a single base pair downstream from the end of the tRNA-Tyr in most species. These codons correspond to the beginning of a highly conserved region, suggesting that this region may be functionally constrained. From our finding, it is possible to hypothesize that asparagine may function as a molecular synapomorphy for Polyphaga.

F

An alignment of the 5′ region of cox1 and the abutting tRNA-Tyr. The dotted line indicates the tRNA; the solid line indicates the beginning of the cox1 gene as previously proposed. The comparative analysis indicates that the first amino acid after the tRNA (asparagines) is completely conserved across Polyphaga, suggesting a possible molecular synapomorphy for Polyphaga.

Initiation and Termination in Protein-Coding Genes

In insects, most protein-coding genes except cox1 use typical ATN (methionine or isoleucine) start codons, and we found the same pattern in all six beetle species (table 4). However, there were some genes that varied: nad1 of Trachypachus, Sphaerius, Chaetosoma, and Priasilpha and nad2 of Sphaerius. For these genes, there is no upstream possibility of ATN start codon due to in-frame stop codons, and downstream possibilities all create a considerable intergenic gap. In this study, we propose TTG (leucine) as a start codon for these genes (Okimoto et al. 1990). TTG has been proposed as a start codon for nad1 in several insects, including Anopheles quadrimaculatus (Mitchell et al. 1993), Tricholepidion gertschi (Nardi et al. 2003), and Pyrocoelia rufa (Bae et al. 2004). In justifying the use of this start codon, Bae et al. (2004) argued from the evolutionary economic perspective that it would minimize intergenic space and avoid overlap with the abutting tRNA. Even more importantly, TTG as a start codon of nad1 is positionally well conserved as inferred from an alignment of all published beetle mtgenomes (fig. 5). Although some of the previously published mtgenomes (Crioceris, Tribolium, and Anoplophora) annotated nad1 with a typical ATN start codon which created overlap with tRNA-Leu or a considerable intergenic gap, we suggest that TTG is a more conserved possibility (fig. 5). Additionally, with the revised start codons, the C-terminal end of the peptide is quite conserved with an acidic polar amino acid (D or E) at position 5, and neutral, nonpolar amino acids (I, L, M, V, or F) at positions 1–4 and 6–11 (fig. 5). The evolution of the TTG start codon does not appear to be lineage specific, however. Of the seven polyphagan species, two had the typical ATN (methionine) start codon, whereas the other five had the TTG (leucine) start codon (fig. 5). Different start codons were used in two lineages (Pyrocoelia and Rhagophthalmus) within the same superfamily (Elateroidea), suggesting that the TTG start codon has evolved multiple times within Coleoptera without much lineage-specific conservation.

Table 4

Start/Stop Codons for Protein-Coding Genes in Six New Beetle Species

Gene	Tetraphalerus	Trachypachus	Sphaerius	Cyphon	Chaetosoma	Priasilpha
nad2	ATT/T	ATG/TAA	TTG/TAA	ATA/TAA	ATA/T	ATT/TAA
cox1	CAA/T	CGA/T	CAA/T	AAT/T	AAC/T	AAC/T
cox2	ATG/T	ATG/T	ATA/T	ATC/T	ATA/T	ATT/TAA
atp8	ATG/TAA	ATT/TAA	ATT/TAA	ATT/TAA	ATT/TAA	ATT/TAG
atp6	ATA/TAA	ATA/TAA	ATA/TAA	ATA/TAA	ATA/TAA	ATA/TAA
cox3	ATG/TA	ATA/TAA	ATG/TA	ATA/TAA	ATG/T	ATG/TA
nad3	ATT/T	ATT/T	ATT/T	ATT/T	ATA/T	ATT/T
nad5	ATA/T	ATT/T	ATT/TA	ATT/T	ATT/T	ATA/T
nad4	ATG/T	ATG/T	ATG/T	ATG/T	ATA/T	ATA/T
nad4l	ATG/TAA	ATT/TAA	ATT/TAA	ATG/TAA	ATG/TAA	ATG/TAA
nad6	ATG/TAA	ATT/TAA	ATT/TAA	ATA/TAA	ATA/TAA	ATT/TAA
cob	ATG/TA	ATG/T	ATG/T	ATG/T	ATG/T	ATG/T
nad1	ATG/TAA	TTG/TAG	TTG/TAA	ATG/TAG	TTG/TAG	TTG/TAG

NOTE.—Incomplete stop codons are noted by either T or TA.

F

An alignment of the tRNA-Leu and nad1 genes. Dotted line indicates hypothetical amino acid translation of nucleotide sequence that codes for tRNA-Leu. Bold letters indicate the amino acids of the putative start codons that were previously proposed. The box indicates our proposed start codons, which shows that the TTG start codon (leucine) is more common than previously thought.

The use of incomplete stop codons (T or TA) was frequent in each of the six mtgenomes (table 4), due to ends of protein-coding genes overlapping with the abutting tRNAs. It is hypothesized that a complete stop codon (TAA) is created through posttranscriptional polyadenylation (Ojala et al. 1981). The presence of partial stop codons is well documented in insects (Beard et al. 1993; Coates et al. 2005; Castro et al. 2006). Not surprisingly, complete stop codons were more often TAA than TAG, consistent with patterns found in previously published mtgenomes.

tRNA-Ser(AGN)

In insect mtgenomes, there are typically 22 tRNAs, with tRNA-Ser and tRNA-Leu 8-fold redundant (two sets of 4-fold redundant tRNAs) (Boore 1999). The length of tRNAs ranged between 60 bp and 75 bp. When compared across all beetle species, including the previously published mtgenomes, we found that the tRNAs were highly conserved within Coleoptera and that all the anticodons were identical and completely conserved, with one exception: the tRNA-Ser(AGN). This particular tRNA was also the most difficult to locate and fold using conventional tRNA search methods such as tRNAscan-SE because it often does not fold into a normal cloverleaf structure due to the absence of stem pairings in the DHU arm (fig. 6). This missing D-stem has been reported in insects (Beard et al. 1993; Crozier RH and Crozier YC 1993; Shao and Barker 2003; Bae et al. 2004), mammals (Chimnaronk et al. 2005; Putz et al. 2007), as well as the rest of Metazoa (Steinberg and Cedergren 1994). Garey and Wolstenholme (1989) proposed that the missing D-stem in tRNA-Ser(AGN) evolved very early in the evolution of Metazoa. Despite lacking this stem, however, this tRNA is normally considered to be functional (Steinberg and Cedergren 1994; Stewart and Beckenbach 2003). In an in vitro study, Hanada et al. (2001) found that bovine tRNA-Ser(AGN) (which lacks the D-stem) is functional, although somewhat less effective than other cloverleaf-shaped tRNAs.

F

Consensus secondary cloverleaf structure for the tRNA-Ser(AGN) gene for all 13 published coleopteran genomes. Capitalized bases are conserved in at least 12 of the 13 sequences; lowercase bases are majority rule. Base pairs may not necessarily match because bases are majority rule.

tRNAscan-SE is often unable to find tRNA-Ser(AGN) because organellar genome searches in tRNAscan-SE are based on COVE (Eddy and Durbin 1994), which uses a CM to model the structure of typical tRNAs. The general model employed by default is based on a secondary structure alignment of over 1,000 tRNAs from all three domains of life. However, because mitochondrial tRNA-Ser(AGN) is often missing an entire stem, attempting to apply the default CM to this specific class of atypical tRNAs often fails. In order to better understand the consensus structure and ameliorate the problem of finding and folding this tRNA for future mtgenome studies, we constructed a new, specific CM that enables COVE to locate and fold this tRNA in particular. Using COVE with the specific model, we were able to identify and fold tRNA-Ser(AGN) for all 13 species with very good sensitivity (CM available from N.C.S.). Because it is not impressive that a model performs well on the sequences that were used to construct it, we also tested the new CM on additional mtgenomic regions both within Coleoptera (five unpublished mtgenomes encompassing three of the four suborders) and other insect orders including Diptera, Lepidoptera, Hymenoptera, Orthoptera, and Hemiptera (table 5). The new CM was able to identify and fold tRNA-Ser(AGN) in all cases, whereas the default CM often failed to locate it. In cases where the default CM found the tRNA-Ser(AGN), the location was usually slightly different and the resulting secondary structure questionable, whereas the new CM yielded boundaries in greater accordance with published results and secondary structures that match the consensus. We found no false positives in this data set with a COVE score cutoff of 15. Thus, we have shown the new CM to perform well on other insects, despite that fact that it was built using only coleopteran sequences. Perhaps, most importantly, we have demonstrated the utility of specific CMs to facilitate uniform and automated annotation of atypical tRNAs.

Table 5

Results Comparing COVE's Default CM versus Beetle-Specific CM for tRNA-Ser(AGN) Using tRNAscan-SE

	Default CM			Specific CM			Published
Organism	Start	End	Score	Start	End	Score	Start	End	Classification	Reference
Tribolium castaneum	—	—	—	6077	6135	33.69	6077	6135	Coleoptera	Friedrich and Muqim (2003)
Pyrophorus divergens	—	—	—	6048	6114	61.33	6048	6114	Coleoptera	Arnoldi et al. (2007)
Drosophila yakuba	6199	6268	9.38	6200	6267	48.97	6200	6267	Diptera	Clary and Wolstenholme (1985)
Culicoides arakawae	—	—	—	7975	8040	32.75	7985	8040	Diptera	NA
Adoxophyes honmai	—	—	—	6180	6246	50.12	6180	6246	Lepidoptera	Lee et al. (2006)
Bombyx mori	927	995	21.56	928	994	49.94	928	994	Lepidoptera	NA
Locusta migratoria	6115	6183	5.02	6116	6182	36.25	6116	6182	Orthoptera	Flook et al. (1995)
Gryllotalpa orientalis	6062	6130	8.96	6063	6129	55.45	6063	6129	Orthoptera	Kim et al. (2005)
Apis mellifera	103	166	3.5	117	177	25.1	116	178	Hymenoptera	Crozier RH and Crozier YC (1993)
Philaenus spumarius	—	—	—	5992	6054	34.77	5991	6055	Hemiptera	Stewart and Beckenbach (2005)
Additional coleopterans
Hydroscapha	6086	6154	9.27	6087	6153	65.84	NA		Coleoptera	NA
Necrophila	—	—	—	6082	6148	61.47	NA		Coleoptera	NA
Naupactus	—	—	—	6053	6121	35.52	NA		Coleoptera	NA
Calosoma	6229	6297	7.6	6230	6296	69.05	NA		Coleoptera	NA
Rhopaea	—	—	—	6081	6147	65.97	NA		Coleoptera	NA

NOTE.—Dashes (—) indicate that tRNAscan-SE with the default CM did not find tRNA-Ser(AGN). NA, not applicable.

Although most arthropods use a GCU anticodon in tRNA-Ser(AGN), all beetle mtgenomes published so far have the UCU anticodon for this tRNA, suggesting that this anticodon may be a molecular synapomorphy for Coleoptera. Outside of Coleoptera, there are a few arthropods that reportedly use a UCU anticodon in tRNA-Ser(AGN), including the sea firefly Vargula hilgendorfii (Ogoh and Ohmiya 2004), the hermit crab Pagurus longicarpus (Hickerson and Cunningham 2000), and all species of lice studied to date (Cameron, Johnson, and Whiting 2007). With an expanded taxon sampling including all four coleopteran suborders, we found that while all the species belonging to the Polyphaga had the UCU anticodon, Trachypachus, Sphaerius, and Tetraphalerus, representing the smaller suborders Adephaga, Myxophaga, and Archostemata, respectively, had the common GCU anticodon instead (fig. 7). Except for the single base difference, the sequences for anticodon and anticodon loop, as well as the distal three paired bases, were identical across all beetles. Given that most arthropods have the GCU anticodon in the tRNA-Ser(AGN), it is possible to speculate that the ancestral anticodon for Coleoptera was GCU, which mutated to UCU in the common ancestor of Polyphaga, thus serving as a molecular synapomorphy for this suborder.

F

An alignment of tRNA-Ser(UCN) anticodon loops (and 3 paired stem nucleotides). Among beetles, Adephaga, Archostemata, and Myxophaga have the common GCU anticodon; all polyphagan species reported to date have the uncommon UCU anticodon, which suggests that this particular anticodon might be a possible molecular synapomorphy for Polyphaga.

Ribosomal RNAs

The mitochondrial ribosomal RNA genes of beetles are largely uniform across the suborders and similar in secondary structure to those proposed for other insect orders (Gillespie et al. 2006; Cameron and Whiting 2008) (supplementary fig. S1 and S2, Supplementary Material online). The published annotations of 12S for beetles all included additional bases at the 5′ end that would play no functional role in the mature rRNA and so are likely not part of the gene. The 5′ end of the 12S molecule was made up of a short, unpaired leader sequence (4–5 bp) followed by a pseudoknot formed by stem H9 and the 5′ portion of stem H17. This pseudoknot can thus be used in the annotation of the 12S gene with the consensus sequence AAGTT-TDATYWT-DRYTT; the first and last 5 bp form the 5′ and 3′ portions, respectively, of stem H9. There was some length variation across the rest of 12S within Coleoptera, with most of variability located in the H47 stem and in the loop regions between H577 and H673. H47 is highly variable between different insect groups—it consists of a short stem and large loop in Hymenoptera (Gillespie et al. 2006) or a long stem and short loop in Lepidoptera (Cameron and Whiting 2008). Most beetles had the long stem form similar to Lepidoptera; however, the elateroid genera (Pyrocoelia, Rhagophthalmus, and Pyrophorus) had the short stem form found in Hymenoptera.

The 16S is more variable than the 12S both across insects and across beetles. The 16S is traditionally annotated as the entire region between adjacent tRNA genes (tRNA-Val and tRNA-Leu(CUN)). This results in considerable length variability in the 5′ end of the gene, approximately 150 bp upstream of the H533 stem. We were able to identify the three stems in this region (H183, H235, and H461); however, there was considerable sequence variability in these stems and length variability in the regions between them. At the 3′ end, there was some length variability between different beetle species; however, all beetles had truncated 16S genes relative to Lepidoptera and Hymenoptera, lacking the 3′ half and most of the loop region of the H2735 stem–loop. The major regions of length variation in beetles were the H837 and H2077 stems–loops as well as the bulge region in the middle of the H991 stem–loop. The large insertion regions and microsatellite regions that distinguish the 16S genes of Lepidoptera were absent, resulting in a much shorter 16S gene in beetles.

Conclusion

Our study represents the first comprehensive comparative analysis of beetle mtgenomes. We find that Coleoptera follows the ancestral insect arrangement with no deletions or duplications. There are several common features that many beetle lineages share, such as a noncoding region of about 18 bp between nad1 and tRNA-Leu(CUN) and the usage of a noncanonical TTG start codon. To cope with the atypical structure of tRNA-Ser(AGN), we present a new specific CM for use with COVE and tRNAscan-SE that allows for more consistent identification and secondary structure prediction of this tRNA. We also find that smaller beetle suborders have the common GCU anticodon for tRNA-Ser(AGN), whereas all polyphagans share a rare UCU anticodon variant. We hypothesize that this UCU anticodon of tRNA-Ser(AGN) and asparagine as a start codon for cox1 are possible molecular synapomorphies for the suborder Polyphaga. Our study demonstrates the importance of comparative analysis in understanding the evolution of mtgenome.