G+C content

Tribolium, like Apis, has a very (A+T)-rich genome (33% and 34% G+C, respectively), but Tribolium G+C domains lack the extremes of G+C content present in Apis mellifera (Fig. 1 and Supplementary Fig. 3). Despite global G+C similarity to Apis, genes in Tribolium, as in Anopheles and Drosophila but not Apis, show a bias towards occurring in (G+C)-rich regions of the genome (Fig. 1). Whatever mechanism drives the accumulation of A+T nucleotides in Tribolium, it does not affect genes in the manner observed in the honeybee, where perhaps additional mechanisms are present.

Figure 1: Cumulative distribution of genic and genomic G+C-content domains in Apis mellifera, Anopheles gambiae, Drosophila melanogaster and Tribolium castaneum.

Cumulative distributions show the fraction of genes (thick lines) or of the entire genome (thin lines) occurring in G+C-content domains less than a given percentage G+C (<X). The more (A+T)-rich half of the T. castaneum genome contains only 30.8% of all T. castaneum genes (31.4% and 33% of A. gambiae and D. melanogaster genes, respectively), whereas the more (A+T)-rich half of the A. mellifera genome contains 77.6% of its genes. At every point on the T. castaneum, A. gambiae and D. melanogaster curves there are fewer genes present in the fraction of the genome less than a given percentage G+C than would be expected if the genes were randomly distributed. In contrast, A. mellifera exhibits the opposite distribution.

High resolution image and legend (108K)

Repetitive DNA

Fully one-third of the Tribolium genome assembly consists of repetitive DNA, which is also (A+T)-rich. Compared to other insects, there is a paucity of microsatellites (1–6-base-pair (bp) motifs) in Tribolium⁹. However, Tribolium contains a relative excess of larger satellites, including several with repeat units longer than 100 bp (2.5% of the Tribolium genome compared with 0.7% in Drosophila). Most (83%) of the microsatellites are found in intergenic regions (63%) or introns (20%), but there is strong over-representation of non-frameshift-causing repeats (3- and 6-bp motifs) due to a dearth of dinucleotide repeats (see Supplementary Information). Of 981 randomly chosen microsatellites, 509 (55.2%) are polymorphic in a sample of 11 Tribolium populations from around the world⁹, providing an extensive collection of markers for population studies. Preliminary efforts to assess global population structure show a shallow but significant correlation between geographic and genetic distance (Supplementary Fig. 4). This suggests that anthropogenic dispersal may maintain a modest level of gene flow across vast distances in this human commensal.

Transposable elements

Transposable elements and other repetitive DNA accumulate in regions along each linkage group that resemble the pericentric blocks of heterochromatin visible in HpaII-banded chromosomes¹⁰. These regions are probably composed largely of highly repetitive heterochromatic sequences, and represent most of the 44-Mb difference between the estimated genome size (0.2 pg or 204 Mb¹¹) and the current assembly (160 Mb). Indeed, as much as 17% of the Tribolium genome is composed of a 360-bp satellite¹² that constitutes only 0.3% of the assembled genome sequence. Several families of DNA transposons, as well as long terminal repeat (LTR) and non-LTR retrotransposons, constituting approximately 6% of the genome, were identified via encoded protein sequence similarity to previously identified elements using TEPipe or BLAST, and are listed in Supplementary Table 5.

Telomeres

Tribolium has a telomerase and telomeres containing TCAGG repeats¹³, a variant of the standard arthropod TTAGG telomeric repeat. Manual assembly of the proximal regions of multiple telomeres beyond the ends of the assembled scaffolds (Supplementary Information) reveals TCAGG repeats interrupted by full-length and 5'-truncated non-LTR retrotransposons belonging to the R1 clade, best known for insertions in the rDNA locus¹⁴. Tribolium telomeres range in length from 15 kilobases (kb) upwards and probably represent a stage intermediate to the loss of telomeres and telomerase in Diptera compared with the simple canonical structure of the honeybee¹⁵ or the more regular insertion of non-LTR retrotransposons into the simple repeats of the silkmoth¹⁶.

Comparative gene content analysis

To understand the consensus set of 16,404 gene models in the context of other available insect and vertebrate genomes, all genes were classified according to their degree of similarity using systematic cross-species analysis. Five insects (Drosophila melanogaster, Anopheles gambiae, Aedes aegypti, T. castaneum, A. mellifera) and five vertebrates (Homo sapiens, Mus musculus, Monodelphis domestica, Gallus gallus, Tetraodon nigroviridis) with similar phylogenetic branching orders were chosen for the comparison. We found the fractions of universal and insect-specific orthologues in Tribolium similar to other insect genomes, as expected, whereas the number of genes without similarity is considerably higher (Fig. 2), possibly attributable to less stringent gene prediction.

Figure 2: Insect gene orthology.

Comparison of the gene repertoire in five insect and five vertebrate genomes, ranging from the core of metazoan genes (dark blue fraction on the left) to the species-unique sequences (white band on the right). The striped boxes correspond to insect- and vertebrate-specific orthologous genes, where the darker bands correspond to all insects or vertebrates (allowing one loss). N:N:N indicates orthologues present in multiple copies in all species (allowing one loss); patchy indicates ancient orthologues (requiring at least one insect and one vertebrate gene) that have become differentially extinct in some lineages. The species tree on the left (shown in detail in Supplementary Fig. 6) was computed using the maximum-likelihood approach on concatenated sequences of 1,150 universal single-copy orthologues. It shows an accelerated rate of evolution in insects and confirms the basal position of the Hymenoptera within the Holometabola¹⁷. Aaeg, Aedes aegypti; Agam, Anopheles gambiae; Amel, Apis mellifera; Dmel, Drosophila melanogaster; Ggal, Gallus gallus; Hsap, Homo sapiens; Mdom, Monodelphis domestica; Mmus, Mus musculus; Tcas, Tribolium castaneum; Tnig, Tetraodon nigroviridis.

High resolution image and legend (142K)

Over 47% of Tribolium genes (7,579) are ancient, with traceable orthologous relations between insects and vertebrates including 15% (2,403) universal single-copy orthologues. Another 1,462 Tribolium genes (9%) constitute the core of what are currently insect-specific orthologues. In comparison, 21% (4,937) of human genes have vertebrate-specific orthologues.

Several hundred ancient genes seem to be under limited evolutionary selection and were independently lost in several species studied (the patchy fraction, defined in Fig. 2). Each new genome uncovers previously invisible ancestral relations among genes—for example, as many as 126 orthologous gene groups shared between Tribolium and humans seem to be absent from the other sequenced insect genomes (Fig. 3 and Supplementary Table 10), 44 of which are single-copy genes present in all vertebrates.

Figure 3: Orthologous genes shared between insect and human genomes.

The Venn diagram shows the number of orthologous groups of genes shared between the insect and human genomes. In addition to the majority of Urbilateria (last common ancestor of the Bilateria) genes shared by all the organisms, there are hundreds of genes that have been lost in some lineages (for example, only retained between human and Tribolium or human and honeybee, but lost in Diptera). Diptera is represented here by Anopheles gambiae, Aedes aegypti and Drosophila melanogaster (with numbers considering only D. melanogaster shown in parentheses).

High resolution image and legend (123K)

The evolutionary emergence of many predicted Tribolium genes is not clear. Thousands of genes currently appear to be species-specific as either no sequence similarity to other genes is detectable, or homology but not orthology can be determined. Reassuringly, this fraction is similar in Tribolium and Drosophila.

We quantified the species phylogeny using a maximum likelihood approach with the concatenated multiple alignment of 1,150 universal single-copy orthologues present in all the organisms studied—an ideal genome-wide data set of essential genes evolving under similar constraints (Fig. 2 and Supplementary Fig. 6). This analysis confirmed previous analyses based on expressed sequence tag (EST) sequences that the Hymenoptera are basal within the Holometabola¹⁷. The shorter branch length for Tribolium implies that the elevated rate of evolution observed in Drosophila and Anopheles occurred more recently¹⁸.

Gene family expansion, frequently associated with a particular adaptation pressure, might reveal physiologically and phenotypically unique features of beetles (Supplementary Table 9 and protein family discussions below). Many duplications shaped the gene content of Tribolium, most notably among odorant-binding proteins and the CYP450 subfamilies CYP6 and CYP9 (Supplementary Fig. 11), some of which are involved in the development of insecticide resistance in the Diptera¹⁹. Duplication of genes under copy-number selection in other species is indicative of species-specific neo-functionalization²⁰. At least 152 genes duplicated in Tribolium have single-copy status in all other insects studied, including sevenfold duplication of genes orthologous to Drosophila CG1625, encoding a putative structural constituent of cytoskeleton, and human ENSP00000269392, encoding centrosomal pre-acrosome localization protein 1.

We also analysed the phylogenetic distribution of orthologous gene group members to quantify evolutionary gene losses²¹. Although least affected, dozens of single-copy orthologues seem to be lost in each lineage. Thirty-eight such genes lost in Tribolium include rather unique genes, encoding phosphotriesterase-related protein and peroxisome assembly factor 1 (peroxin-2), compared to 59 such genes lost in Drosophila. Notably, for the less restricted fractions of orthologues (defined in Fig. 2), several hundred gene orthologues have been lost in each species.

Oogenesis

Despite profound differences in ovarian architecture—telotrophic versus polytrophic—we identified Tribolium orthologues of most Drosophila genes required for stem cell maintenance, RNA localization and axis formation. Like Apis, however, Tribolium lacks a bag of marbles orthologue, which is essential for the differentiation of cystoblast versus germline stem cells in Drosophila²². Interestingly, an orthologue of the gene gld-1, which fulfils a similar function in C. elegans²², is present in Tribolium.

Anterior–posterior patterning

Analysis of the genome sequence confirmed the absence of a bcd orthologue in Tribolium. Instead, anterior patterning is synergistically organized by otd and hb (ref. 23). However, it is still unclear how the posterior gradient of Tribolium Caudal is shaped in the absence of Bicoid. Notably, Tribolium contains an orthologue of mex-3, a factor that translationally represses the C. elegans cad homologue²⁴. Although the Tribolium genome contains orthologues of the Drosophila segmentation genes, their functions are not entirely conserved^{25, 26, 27}. Furthermore, the genome reveals the unexpected polycistronic organization of a novel gap gene, mille-pattes²⁸, the transcript of which encodes several short peptides.

In contrast to the classical protostomian model organisms Drosophila and Caenorhabditis, Hox genes in Tribolium map to a single cluster of 750 kb on linkage group 2. Orthologues of all Drosophila Hox genes and the Hox-derived genes ftz and zen are transcribed from the same strand, and we find no evidence for interspersion of other protein-coding genes. Taken together, these results suggest that the evolutionary constraints preserving Hox cluster integrity still function in Tribolium.

Dorso-ventral patterning

As in Drosophila, the dorso-ventral axis of the Tribolium embryo depends on a nuclear gradient of Dl, an NF-B protein, which is established through ventral activation of a Tl receptor²⁹ (one of four in Tribolium). Factors required for localized Tl activation are also present in Tribolium (potential Tl ligands: six spz-like genes; extracellular proteases: one gd, six snk and four tandem ea genes), suggesting that, as in Drosophila, an extra-embryonic signal induces the embryonic dorso-ventral axis.

Tribolium sog inhibition of Dpp/BMP generates a patterning gradient along the dorso-ventral axis³⁰. Similar chordin/sog function in spiders and a hemichordate suggest that this may represent the ancestral bilaterian condition³⁰. Like Apis, Tribolium lacks an orthologue of Drosophila scw, but knockdown of another ligand, Tribolium gbb1, affected the embryonic Dpp/BMP gradient. Tribolium contains orthologues of all five Drosophila TGF- receptors; however, Dpp signalling moderators that have duplicated and diverged in Drosophila, such as Tol/tok and Cv/tsg, occur as a single copy in Tribolium. Most strikingly, Tribolium contains homologues of BMP10 as well as bambi, Dan and gremlin BMP inhibitors, which are all known from vertebrates, but are not found in Drosophila.

The growth zone

We identified several members of the Fgf and Wnt signalling pathways. The expression patterns of Tribolium Fgf8, Wnt1, Wnt5 and WntD/8 (refs 31, 32) highlight the dynamic organization of the growth zone and underline its role in axis elongation.

Head patterning

Orthologues of 25 out of 30 key regulators of the vertebrate anterior neural plate are specifically expressed in the Tribolium embryonic head (Supplementary Table 12). Two orthologues are not expressed in the head neuroectoderm (barH, arx) and three do not have Tribolium or Drosophila orthologues (vax, hesx1, atx). Of the canonical Drosophila head gap genes, only the late head-patterning function of otd is conserved. ems function is restricted to parts of the antennal and ocular segments, and knockdown of btd seems to have no phenotypic consequences. Thus, analysis of Tribolium genes defines a set of genes that is highly conserved in bilaterian head development, and underscores the derived mode of Drosophila head patterning.

Leg and wing development

In contrast to Drosophila, ventral appendages in Tribolium develop during embryogenesis from buds that grow continuously along the proximo-distal axis³³. Nonetheless, we identified Tribolium orthologues for most of a core set of Drosophila appendage genes (Supplementary Table 13). On the other hand, orthologues of genes not found in Drosophila, such as Wnt11, gremlin, Fgf8 and an F-Box gene, are expressed in the embryonic legs^{21, 31}. Although their exact function in Tribolium appendages is not known, Fgf8 is essential to vertebrate limb development.

A major innovation driving the radiation of beetles was the evolution of a highly modified protective forewing. Expression analysis and RNAi experiments revealed a high degree of conservation between Tribolium and Drosophila wing gene networks (Supplementary Table 13), supporting the hypothesis that sclerotized elytra evolved from ancestral membranous wings mainly through new interactions between conserved patterning modules and as yet unknown downstream effector genes.

Eye development

Tribolium has orthologues of nearly all genes currently known to regulate specification and differentiation in the Drosophila retina (Supplementary Table 14). Exceptions are the linker protein Phyllopod and the lens crystallin protein Drosocrystallin, which are restricted to Diptera. Eight of fifty-seven investigated eye developmental genes are duplicated in the Drosophila genome but not Tribolium, and in four cases the Drosophila paralogues have similar function. This suggests a more dynamic evolution of Drosophila retina genes and higher genetic complexity, highlighting the value of Tribolium as a more ancestral and simply organized model of insect eye development.

Cys-loop ligand-gated ion channels

Members of this superfamily mediate chemical synaptic transmission in insects and are targets of successful pest control chemicals with animal health and crop protection applications³⁴. The Tribolium Cys-loop ligand-gated ion channel (Cys-loop LGIC) superfamily contains 24 genes, the largest known so far for insects (Drosophila and Apis superfamilies comprise 23 and 21 genes, respectively), due in part to the additional nicotinic acetylcholine receptor (nAChR) subunits in Tribolium. We also found genes for ion channels gated by -aminobutyric acid (-aminobutyric acid receptors (GABARs)), glutamate (GluCls) and histamine, as well as orthologues of the Drosophila pH-sensitive chloride channel³⁵. The molecular diversity of the Tribolium Cys-loop LGIC superfamily is broadened by alternative splicing and RNA A-to-I editing, which in some cases generates species-specific receptor isoforms³⁵. The Tribolium Cys-loop LGIC superfamily is the first complete set of genes encoding molecular targets of several insecticides—imidacloprid and other neonicotinoids (nAChRs), fipronil (GABARs) and avermectins (GluCls)—described for an agricultural pest species.

Cytochrome P450 proteins

Most insect cytochrome P450 proteins (CYPs) are thought to be involved in metabolic detoxification of host plant allelochemicals and toxicants, and several are insecticide resistance genes³⁶. Other CYPs act in the synthesis and degradation of lipid signalling molecules, such as ecdysteroids³⁷. Similarly to mosquitoes, especially Aedes, Tribolium has an independently expanded CYP gene family, particularly those involved in environmental response (Supplementary Table 16).

Within the Tribolium P450s, the CYP2 and mitochondrial clans have undergone relatively little gene expansion, lack pseudogenes, and are probably reserved for essential endogenous functions in ecdysteroid metabolism and development. In contrast, expansions via tandem duplication produced 85% of Tribolium P450s clustered in groups of 2–16 genes, with large expansions of CYP3 and CYP4 clans involved in environmental response. In comparison, Apis has only four CYP4 genes, whereas Aedes has relatively similarly sized expansions of CYP3 and CYP4 clans (Supplementary Table 16). We speculate that both mosquito larvae (which are omnivorous scavengers) and Tribolium have adapted to diverse chemical environments in part by expansion of CYP gene families involved in detoxification.

C1 cysteine peptidase genes

Tribolium castaneum has successfully exploited cereal grains in spite of the arsenal of defensive allelochemicals, including inhibitors of serine peptidase digestive enzymes. In tenebrionid beetles, cathepsins B, L and serine peptidases such as trypsins and chymotrypsins are part of the digestive peptidase complex in the larval gut³⁸.

Comparing potential digestive peptidase genes in Tribolium with those in other sequenced insects (Supplementary Fig. 12) we found more C1 cysteine peptidase genes in T. castaneum. The proliferation of Tribolium C1 cysteine peptidase genes reflects expansions into five gene families, corresponding to four major clusters. This expansion is consistent with a trend seen in some beetles relative to other insects: a shift to a more acidic gut, conducive to cysteine peptidase activity.

Tribolium castaneum C1 cysteine peptidase genes encode B and L cathepsins, and include the first-known insect genes similar to O and K cathepsins (Supplementary Table 17). Most of the cathepsin-B-like peptidases lack conserved residues in functional regions and thus may lack peptidase activity, whereas all but two Tribolium cathepsin L peptidase genes encode potentially functional enzymes. In vertebrates, O and K cathepsins are lysosomal cysteine peptidases, involved in bone remodelling and resorption. Analysis of Tribolium cathepsins may provide insight into this family of proteins whose elevated expression is associated with a significant fraction of human breast cancers and tumour invasiveness.

Neurohormones and G-protein-coupled receptors

Insect neurohormones (neuropeptides, protein hormones and biogenic amines) control development, reproduction, behaviour, feeding and many other physiological processes, often by signalling through G-protein-coupled receptors (GPCRs). We found 20 genes encoding biogenic amine GPCRs in Tribolium (compared to 21 in Drosophila and 19 in Apis) and 52 genes encoding neuropeptide or protein hormone GPCRs (49 in Drosophila, 37 in Apis³⁹). Moreover, we identified the likely ligands for 45 of these 72 Tribolium GPCRs. Furthermore, we annotated 39 neuropeptide and protein hormone genes. We found excellent agreement (95%) between the proposed ligands for the Tribolium neurohormone GPCRs and the independently annotated neuropeptide and protein hormone genes. Interestingly, the Tribolium genome contains a vasopressin-like neuropeptide (TC06626) and a vasopressin-like GPCR gene (TC16363; Supplementary Fig. 14), neither of which has been detected in any other sequenced insect³⁹. Vasopressin in mammals is the major neurohormone stimulating water reabsorption in the kidneys⁴⁰. Its presence in Tribolium may help the beetle to survive in very dry habitats.

Vision

Most of the 21 investigated genes that participate in the Drosophila photo-transduction network are conserved in Tribolium (Supplementary Table 14). Most notable is the lack of ninaG and inaC, which may be functionally replaced by closely related paralogues in Tribolium.

Tribolium contains only two opsin genes, representing members of the long-wavelength and ultraviolet-sensitivity-facilitating opsin subgroups. In contrast, Drosophila contains seven, and there is evidence for minimally three in most other insects. The lack of a blue-light-sensitive opsin gene in Tribolium is consistent with the unusual expression of long-wavelength opsin in all photoreceptor cells in this species⁴¹. The implied reduction in colour discrimination in Tribolium is probably a consequence of the widespread cryptic lifestyle of this species group.

Odorant and gustatory receptors

Odorant and gustatory receptors form the insect chemoreceptor superfamily. Tribolium has a major expansion of both odorant and gustatory receptors relative to Drosophila, Anopheles and Aedes mosquitoes, silkmoth and honeybee (Supplementary Table 19). We identified and annotated 265 apparently functional odorant receptors, 42 full-length pseudogenes and 34 pseudogene fragments. Most of these T. castaneum odorant receptors are in seven species-specific subfamilies, including one containing 150 genes, and most are in tandem gene arrays, created by gene duplication within the Tribolium lineage in the last 300 million years⁴².

We annotated 220 apparently functional gustatory receptors and 25 pseudogenes (gustatory receptor gene fragments were not assessed). The gustatory receptor families in fruitflies and mosquitoes, but not honeybee, contain several genes that are alternatively spliced, with multiple alternative long first exons encoding at least the amino-terminal 50% of the gustatory receptor spliced into a set of short shared exons encoding the carboxy terminus^{43, 44, 45}. Most Tribolium gustatory receptors are encoded by single genes; however, T. castaneum Gr214 is a massive alternatively spliced locus with 30 alternative long 5' exons (six of which are pseudogenic) spliced into three shared 3' exons encoding the C terminus. Three T. castaneum gustatory receptors are orthologues of highly conserved gustatory receptors in other insects^{43, 44, 45}, two of which form a heterodimeric carbon dioxide receptor⁴⁶. The remainder form many species-specific subfamilies, one of which is expanded to 88 genes (Supplementary Information and Supplementary Fig. 16).

Systemic RNAi

In Tribolium, as in C. elegans but not Drosophila, the RNAi effect spreads systemically from the site of injection to other tissues⁵ and from injected females to their offspring⁴. Surprisingly, our survey of genes involved in systemic RNA did not reveal much conservation between Tribolium and C. elegans.

The SID-1 multi-transmembrane protein, essential for double-stranded RNA (dsRNA) uptake in C. elegans, is not found in Drosophila, suggesting that the presence or absence of a sid-1 gene is the primary determinant of whether or not systemic RNAi occurs in an organism. We found three genes in Tribolium that encode proteins similar to SID-1. However, their sequences are more similar to another C. elegans protein, TAG-130 (also known as ZK721.1), which is not required for systemic RNAi in C. elegans⁴⁷. Additionally, the secondary argonaute proteins and RNA-dependent RNA polymerase (RdRP)^{48, 49}, essential for the amplification of the initial dsRNA trigger in C. elegans, are absent in Tribolium. Therefore, the molecular basis for systemic RNAi in Tribolium and other insects might differ from that in C. elegans and remains to be elucidated.