Research Article

Genome Sequence of Aedes aegypti, a Major Arbovirus Vector

See allHide authors and affiliations

Science  22 Jun 2007:
Vol. 316, Issue 5832, pp. 1718-1723
DOI: 10.1126/science.1138878


We present a draft sequence of the genome of Aedes aegypti, the primary vector for yellow fever and dengue fever, which at ∼1376 million base pairs is about 5 times the size of the genome of the malaria vector Anopheles gambiae. Nearly 50% of the Ae. aegypti genome consists of transposable elements. These contribute to a factor of ∼4 to 6 increase in average gene length and in sizes of intergenic regions relative to An. gambiae and Drosophila melanogaster. Nonetheless, chromosomal synteny is generally maintained among all three insects, although conservation of orthologous gene order is higher (by a factor of ∼2) between the mosquito species than between either of them and the fruit fly. An increase in genes encoding odorant binding, cytochrome P450, and cuticle domains relative to An. gambiae suggests that members of these protein families underpin some of the biological differences between the two mosquito species.

Mosquitoes are vectors of many important human diseases. Transmission of arboviruses is largely associated with the subfamily Culicinae, lymphatic filarial worms with both the Culicinae and the subfamily Anophelinae, and transmission of malaria-causing parasites with the Anophelinae (1). Aedes aegypti is the best-characterized species within the Culicinae (2), primarily because of its easy transition from field to laboratory culture, and has provided much of the existing information on mosquito biology, physiology, genetics, and vector competence (3, 4). It maintains close association with human populations and is the principal vector of the etiological agents of yellow fever and dengue fever (5, 6), as well as for the recent chikungunya fever epidemics in countries in the Indian Ocean area (7). Despite an effective vaccine, yellow fever remains a disease burden in Africa and parts of South America, with ∼200,000 cases per year resulting in ∼30,000 deaths (5). About 2.5 billion people are at risk for dengue, with ∼50 million cases per year and ∼500,000 cases of dengue hemorrhagic fever, the more serious manifestation of disease. The incidence of dengue, for which mosquito management is currently the only prevention option, is on the increase (8). Thus, there is an urgent need to improve the control of these diseases and their vector.

The availability of a draft sequence of the ∼278 million base pair (Mbp) genome of Anopheles gambiae (9) has accelerated research to develop new mosquito- and malaria-control strategies. Comparisons between An. gambiae and Drosophila melanogaster (10) revealed genomic differences between the two insects that reflect their divergence ∼250 million years ago (11). Anopheles mosquitoes radiated from the Aedes and Culex lineages ∼150 million years ago (12), and Ae. aegypti and An. gambiae share similar characteristics such as anthropophily, but they exhibit variation in morphology and physiology, mating behavior, oviposition preferences, dispersal, and biting cycle (1). Both mosquito species have three pairs of chromosomes, but Ae. aegypti lacks heteromorphic sex chromosomes (13). To provide genomics platforms for research into Ae. aegypti and to harness the power of comparative genome analyses, we undertook a project to sequence the genome of this mosquito species.

Assembly of a draft genome sequence of Aedes aegypti. Whole-genome shotgun sequencing was performed on DNA purified from newly hatched larvae of an inbred substrain (LVPib12) of the Liverpool strain of Ae. aegypti, which is tolerant to inbreeding while maintaining relevant phenotypes (14). About 98% of the sequence, assembled using Arachne (15), is contained within 1257 scaffolds with an N50 scaffold size of ∼1.5 Mbp (i.e., half of the assembly resides in scaffolds this size or longer). Assembly statistics for the 1376-Mbp genome are given in table S1. Data related to the genome project have been deposited in GenBank (project accession number AAGE00000000).

The genome size of Ae. aegypti as determined by sequence analysis is larger than the original estimate, ∼813 Mbp, which was based on Cot (DNA reassociation kinetics) analysis carried out in 1991 (16). An overinflated genome size could arise from assembled sequence data as a result of allelic sequence polymorphism present in a heterogeneous population of mosquitoes being sequenced. Although the estimate of 1376 Mbp may contain some such regions, we do not believe that our estimate is out of range by a large margin for the following reasons: (i) The strain that was used for the sequencing project was highly inbred (14); (ii) assembled sequences that are potentially “undercollapsed” are <5% of the estimated genome size (fig. S1); and (iii) flow cytometry data from six isolates of Ae. aegypti, including the parent of LVPib12, indicate estimated genome sizes of 1213 to 1369 Mbp (table S2).

Genetic and physical mapping data allowed assignment, but without order or orientation, of 63, 48, 39, 43, and 45 scaffolds to Ae. aegypti chromosome 1 and chromosome arms 2p, 2q, 3p, and 3q, respectively (14). These scaffolds total ∼430 Mbp in length and represent ∼31% of the genome (table S3). Thus, the development of high-resolution physical mapping techniques and the generation of additional random or targeted sequence data are priorities for improving the quality of the current fragmented genome assembly and size estimate. Such progress would enable unambiguous differentiation between regions of segmental duplications and residual haplotype polymorphism.

The genome of Aedes aegypti is riddled with transposable elements. Transposable elements (TEs) contribute substantially to the factor of ∼5 size difference between the Ae. aegypti and An. gambiae genomes. About 47% of the Ae. aegypti genome consists of TEs (Fig. 1 and table S4; see table S4 legend for definitions of TE family, element, and copy). Aedes aegypti harbors all known types of TEs that have been reported in An. gambiae with the exception of two DNA transposons, merlin (17) and gambol (18). Simple and tandem repeats occupy ∼6% of the genome, and an additional ∼15% consists of repetitive sequences that remain to be classified.

Fig. 1.

Relative genomic content of annotated TEs and other sequences in Aedes aegypti. TEs have been deposited in TEfam, a relational database for submission, retrieval, and analysis of TEs (

Most eukaryotic TE families characterized to date (19) are presentin Ae. aegypti and more than 1000 TEs have been annotated, representing a diverse collection of TEs in a single genome (table S4). Although the majority of protein-coding TEs appear to be degenerate, more than 200 elements have at least one copy with an intact open reading frame (ORF) and other features suggesting recent transposition. About 3% of the genome is composed of ∼13,000 copies of the element Juan-A in the Jockey family of non–long terminal repeat (LTR) retrotransposons. A tRNA-related short interspersed nuclear element, Feilai-B, has the highest copy number, with ∼50,000 copies per haploid genome. Only one highly degenerate mariner element is found in Ae. aegypti, whereas at least 20 mariner elements, many with intact ORFs, were found in An. gambiae. TEs present in Ae. aegypti but missing from An. gambiae include the LOA family of non-LTR retrotransposons, the Osvaldo element of the Ty3/gypsy LTR retrotransposons (20), and a unique family, Penelope (21). Comparison of Ae. aegypti and An. gambiae TE sequences is consistent with the interpretation of an overall lack of apparent horizontal transfer events, as a single candidate for such events was identified (14); one full-length copy of the ITmD37E DNA transposon in Ae. aegypti is 93% identical at the nucleotide level to a similarly classified TE in An. gambiae.

Miniature inverted repeat transposable elements (MITEs) and MITE-like elements of non–protein-coding TEs in Ae. aegypti have terminal inverted repeat sequences and target-site duplications, features characteristic of transposition of DNA transposons. Such TEs can be mobilized to transpose in trans, by transposases encoded by DNA transposons (22). The latter TEs occupy only 3% of the Ae. aegypti genome and they are less numerous than non–protein-coding DNA elements, which occupy 16% of the genome (table S4). Thus, DNA transposons may have contributed to the expansion in size and organization of the Ae. aegypti genome through cross-mobilization of MITEs and MITE-like TEs.

Annotation of the draft genome sequence. The fragmented nature of the assembled genome sequence, an asymmetric distribution of intron lengths within genes (figs. S2 and S3), and the frequent occurrence of TE-associated ORFs close to genes and within introns complicated the process of automated gene modeling and often led to prediction of split or chimeric gene models. Thus, we developed a multistage genome masking strategy to minimize the negative effects of TEs and other repetitive elements before gene finding (resulting in masking ∼70% of the genome sequence). We also optimized gene-finding programs via iterative manual inspection of predicted gene models relative to a training set (14).

Two independent automated pipelines for structural annotation resulted in the prediction of 17,776 and 27,284 gene models, respectively (14). We made extensive use of a large collection of ∼265,000 Ae. aegypti expressed sequence tags (ESTs) and dipteran protein and cDNA sequences in producing and then merging the two data sets into a single high-confidence gene set, which consists of 15,419 gene models (AaegL1.1). Alternative splice forms derived from these genes are predicted to generate at least 16,789 transcripts. Table 1 lists some of the genome and protein-coding characteristics of Ae. aegypti and those of D. melanogaster and An. gambiae.

Table 1.

Comparative statistics of Ae. aegypti nuclear genome coding characteristics.

View this table:

Gene descriptions and molecular function Gene Ontology (GO) codes were assigned computationally to predicted protein sequences by means of BLASTP comparison searches with protein databases (14). The functional annotation pipeline included analyses of protein domains as well as secretion signal sequence and transmembrane motifs. A total of 8332 proteins were assigned a description, 9335 proteins were assigned GO terms, 2796 were assigned as “hypothetical proteins,” and 5027 were denoted “conserved hypothetical proteins.”

Genes encoding proteins <50 amino acid residues in length were not included in this annotation release unless they encoded known small proteins. However, these and other genes are captured in a set of 15,396 lower-confidence gene models that is available for analysis as a supplementary release (14). On the basis of transcriptional mapping data and limited manual examination, we anticipate that ∼5 to 10% of the second-tier models or modified versions of them represent “real” genes.

TEs contribute to complex protein-coding gene structures in Aedes aegypti. A striking feature of protein-coding genes in Ae. aegypti is the factor of 4 to 6 increase in the average length of a gene relative to An. gambiae and D. melanogaster, which is due to longer intron lengths rather than longer exons or an increased number of introns (Table 1). The increased length of introns is primarily due to infiltration by TEs; a plot of intron size before and after masking repeat sequences reveals a shift to shorter intron lengths (fig. S2). A more global perspective of the genome expansion was revealed by the difference in genomic span (factor of ∼4.6) of conserved gene arrangements between Ae. aegypti and An. gambiae that occupy ∼33% of each genome (table S5 and fig. S3), providing evidence that TE-mediated expansion in both genic and intergenic regions has contributed to the increased size of the Ae. aegypti genome. Long introns, in particular those in 5′ and 3′ untranslated regions, are likely to complicate in silico studies to define cis-acting transcription and translational regulatory elements, as they may be distant from coding sequences (fig. S4).

Transcriptional analyses. Data derived from three different transcript-profiling platforms—ESTs, massively parallel signature sequencing (MPSS), and 60-nucleotide oligomer–based microarrays—were used to experimentally confirm predicted protein-coding gene models and to gain insight into differential transcription profiles (14). In total, the platforms identified transcripts from 12,350 (80%) of 15,419 genes. Mapping of ∼265,000 ESTs and cDNA sequences and MPSS signature sequence tags to the genome sequence as well as gene models provided evidence for transcription of 9270 and 3984 genes, respectively, whereas microarray data identified transcripts from 9143 genes (table S6). The smaller number of genes identified by MPSS (table S7) may in part be explained by the observation that only about two-thirds of the genes can be assayed by MPSS, as this approach required the presence of a Dpn II restriction enzyme site within the transcribed region. The platforms identified a common set of 2558 genes and each platform identified a unique set of genes (fig. S5), which highlights the importance of using a multiplatform approach. The data provide empirical support for ∼76% of genes annotated as hypothetical (table S8), underscoring the validity of ab initio gene-finding programs in identifying novel genes.

Differences in transcript abundance between pools of RNA from nonadult developmental stages and from 4-day-old, non–blood-fed adult females were revealed by the microarray analyses, which identified 398 and 208 preadult stage and adult female enriched transcripts, respectively (table S9). Functional categorization of these transcripts differed mainly with regard to cytoskeletal, structural, and chemosensory functions (Fig. 2). Differential transcription of genes thought to be involved in chemosensory processes between these stages was conspicuous, with 17 transcripts highly enriched in mosquito developmental stages and only 3 enriched in adult females. A larger number of immunesystem gene transcripts were also enriched in preadult stages (38 preadult versus 19 adult), which may reflect a broader microbial exposure of larvae and pupae in their aqueous environments. In addition, highly expressed genes encoding cuticle proteins in preadult stages (38 preadult versus 1 adult) are indicative of their function in cuticle metabolism and in a variety of other processes, including immunity, that are particularly dominant during development. The non–blood-fed status of the female mosquitoes did not enable discrimination of genes that carry out female-specific functions that mostly relate to blood processing and egg production.

Fig. 2.

Transcriptome analyses of Aedes aegypti. (A) Functional class distributions of genes that are enriched in preadult stages (DEV) and the adult female stage (ADULT) (table S9). (B) Proportions of functional gene classes, expressed as percentage of the total number of genes that are enriched in preadult stages (DEV), adult female stage (ADULT), and constitutively expressed genes (CONST.). (C and D) same as (A) and (B) for genes enriched in the male, in the female, and common (CONST.) for both sexes (table S18). Functional classes: IMM, immunity; RED/STE, redox and oxidoreductive stress; CSR, chemosensory reception; DIG, blood and sugar food digestive; PROT, proteolysis; CYT/STR, cytoskeletal and structural; TRP, transport; R/T/T, replication, transcription, and translation; MET, metabolism; DIV, diverse functions; UNK, unknown functions. The total number of genes in each category is indicated in parentheses.

Aedes aegypti gene families and domain composition. Consistent with evolutionary distance estimates (12), there is a higher degree of similarity between the Ae. aegypti and An. gambiae proteomes than between the mosquito and D. melanogaster proteomes. Orthologous proteins were computed among the three genomes, with 67% of the Ae. aegypti proteins having an ortholog in An. gambiae and 58% having an ortholog in D. melanogaster (Fig. 3A). Analysis of three-way, single-copy orthologs revealed average amino acid identity of 74% between the mosquito proteins, in contrast with ∼58% identity between mosquito and fruit fly proteins (fig. S7). About 2000 orthologs are shared only between the mosquitoes and may represent functions central to mosquito biology. Although most of these proteins are of unknown function, ∼250 can be assigned a predicted function, of which 28% are involved in gustatory or olfactory systems, 12% are members of the cuticular gene family, and 8% are members of the cytochrome P450 family (14).

Fig. 3.

Orthology and chromosomal synteny among Ae. aegypti, An. gambiae, and D. melanogaster. (A) Each circle represents a gene set for Ae. aegypti (Ae), An. gambiae (An), and D. melanogaster (Dm). Because a gene can be involved in several homologies, gene sets do not always have the same number of genes within intersections (e.g., in the Ae-Dm comparison, 943 Ae genes are similar to Dm and 925 Dm genes are similar to Ae). (B) Aedes aegypti chromosomes are represented in gray (not to scale). Chromosome arms are designated as “p” and “q”; chromosome 1 has no arm distinctions. Colored chromosomes represent the syntenic chromosome from An. gambiae or D. melanogaster (not to scale). Solid and dashed lines link each Ae. aegypti chromosome to its primary and secondary syntenic chromosome, respectively. The number of Ae orthologs to An and Dm chromosome arms is indicated, and the total number of orthologs on the Ae chromosome arm to Ae or Dm is shown in italics and parentheses.

Mapping of protein domains with Interpro (23) revealed an expansion of zinc fingers, insect cuticle, chitin-binding peritrophin-A, cytochrome P450, odorant binding protein (OBP) A10/OS-D, and insect allergen–related domains, among others, in Ae. aegypti relative to An. gambiae, D. melanogaster, and the honey bee Apis mellifera (table S10). Some of these constitute large Ae. aegypti gene families, as revealed by two independent clustering methods (14) (table S11). Genes containing zinc finger–like domains could be of transposon or retroviral origin, and these remain to be assessed.

Species-specific differences in the number of members within a multigene family often provide clues about biological adaptation to environmental challenges. In this context, cuticle proteins have been described to play diverse roles in exoskeleton formationand wound healing and are expressed in hemocytes, a major cell type that mediates innate immunity (24). Cuticular proteins also are implicated in arbovirus transmission (25). Expansion of olfactory receptors and OBPs in Ae. aegypti may contribute to an elaborate olfactory system, which in turn may be linked to the expansion in detoxification capacity. The latter and insect allergen–related genes, suggested to have a digestive function, may contribute to the relative robustness of Ae. aegypti andalsocould manifest in a higher insecticide resistance. In this context, the genome and EST data have led to the development of a specific microarray to identify candidate genes among members of multigene families (cytochrome P450, glutathione S-transferase, and carboxylesterase) associated with metabolic resistance to insecticides (26). This platform will provide a means to rapidly survey mechanisms of insecticide resistance in mosquito populations and represents an important tool in managing insecticide deployment and development programs.

G protein–coupled receptors (GPCRs) that are expected to function in signal transduction cascades in Ae. aegypti have been manually identified (14). This superfamily of proteins includes 111 nonsensory class A, B, and C GPCRs, 14 atypical class D GPCRs, and 10 opsin photoreceptors (tables S12 and S13). Aedes aegypti possesses orthologs for >85% of the An. gambiae and D. melanogaster nonsensory GPCRs, which suggests conservation of GPCR-mediated neurological processes across the Diptera. Many Ae. aegypti GPCRs have sequence similarity to known drug targets (27) and may reveal new opportunities for the development of novel insecticides.

Metabolic potential and membrane transporters. Aedes aegypti and An. gambiae are predicted to contain similar metabolic profiles as judged by assigning an Enzyme Commission (EC) number to both mosquito proteomes (table S14). Given the early stages of annotation, it is premature to draw conclusions from missing enzymes in predicted Ae. aegypti metabolic pathways. For example, assignment of EC numbers to the supplemental Ae. aegypti gene set (table S14) resulted in the identification of an additional 12 EC numbers (table S15) not present in AaegL1.1.

An automated pipeline (28) was used to predict potential membrane transporters for Ae. aegypti and An. gambiae, and their transport capacity resembles that of D. melanogaster (table S16). Similar to other multicellular eukaryotes, ∼32% of all three insect transporters code for ion channels and probably function to maintain hemolymph homeostasis under different environmental conditions by modulating the concentrations of Na+, K+, and Cl ions. Aedes aegypti encodes 52 more paralogs of voltagegated potassium ion channels, epithelial sodium channels, and ligand-gated ion channels than An. gambiae and 65 more such paralogs than D. melanogaster. These channels play important roles in the signal transduction pathway and cell communication in the central nervous system and at neuromuscular junctions. A collection of 64 putative adenosine triphosphate–binding cassette transporters was identified, including subgroups that encode multidrug efflux proteins. Aedes aegypti encodes 16 more members of four different types of amino acid transporters than An. gambiae and 13 more members than D. melanogaster. Mosquito larvae cannot synthesize de novo all the basic, neutral, or aromatic l–amino acids (3) and must rely on uptake of these essential amino acids. The richer repertoire of membrane transport systems in Ae. aegypti is likely to intersect with the apparent increase in odorant reception and detoxification capacity.

Autosomal sex determination and sex-specific gene expression. Heteromorphic sex chromosomes are absent in Ae. aegypti and other culicine mosquitoes (13). Instead, sex is controlled by an autosomal locus wherein the maledetermining allele, M, is dominant. The primary switch mechanism at the top of the mosquito sex determination cascade is different from that of D. melanogaster, where the X-chromosome/autosome ratio controls sex differentiation. However, we expect conservation of function in mosquito orthologs of Drosophila genes that are further downstream of the cascade (29). We verified the presence of a number of these in Ae. aegypti, including orthologs for doublesex, transformer-2, fruitless, dissatisfaction, and intersex (table S17).

To define gene expression differences between the sexes, we analyzed microarray transcription profiles of 4-day-old, non–blood-fed adult female and male mosquitoes (Fig. 2); 669 and 635 transcripts were enriched in females and males, respectively, and 6713 transcripts were expressed at similar levels in both sexes (table S18). An additional 373 and 534 transcripts generated exclusive hybridization signals (with signal intensity below the cutoff threshold level in one channel) in females and males, respectively, and may therefore represent sex-specific transcripts. Functional categorization of female and male enriched transcripts yielded similar results, with some exceptions; male mosquitoes expressed a larger number of immune system–related transcripts (40 in males versus 25 in females) and redox- or stress-related transcripts (45 in males versus 33 in females). By comparing the Ae. aegypti profiles with previously described An. gambiae sex-specific microarray analyses (30), we identified 144 orthologous genes displaying the same sex-specific transcription pattern in An. gambiae (table S19), whereas 74 orthologs showed an opposite profile (table S20), suggesting differences in certain sex-specific functions between the two mosquito species.

Conserved synteny with Anopheles gambiae and Drosophila melanogaster. The assignment of 238 Ae. aegypti scaffolds containing ∼5000 genes—about one-third of the predicted gene set—to a chromosomal location on the basis of genetic and physical mapping data (14) allowed us to compare ortholog position and to identify conserved evolutionary associations between Ae. aegypti and An. gambiae or D. melanogaster chromosomes (tables S3 and S21). Most of the Ae. aegypti chromosome arms, with the exception of 2p and 3q, exhibited a distinct one-to-one correlation with An. gambiae and D. melanogaster chromosome arms with respect to the proportion of orthologous genes conserved between chromosome arm pairs (Fig. 3B). These findings confirm and extend previous results that compared a small number (∼75) of Ae. aegypti genes with orthologs in An. gambiae and D. melanogaster (31).

Maps of conserved local gene arrangements (microsynteny) were computed by identifying blocks of at least two neighboring single-copy orthologs in each pair of genomes and allowing not more than two intervening genes (14). In line with the species divergence times, twice as many orthologs are similarly arranged between these mosquito species than between either of them and the fruit fly (table S22) (32); 1345 microsyntenic blocks were identified between Ae. aegypti and An. gambiae, containing 5265 out of a total of 6790 single-copy orthologs (tables S5 and S22). When D. melanogaster is used as an outgroup to count synteny breaks that have occurred in each mosquito lineage since their radiation, the data indicate a rate of genome shuffling in the Ae. aegypti lineage greater by a factor of ∼2.5 than that in the An. gambiae lineage (14). However, this estimate may be inflated because of the fragmented nature of the current Ae. aegypti genome assembly. Thus, the highly repetitive nature of the Ae. aegypti genome appears to have facilitated local gene rearrangements, but it does not appear to have had a gross influence on chromosomal synteny.

Concluding remarks. The draft genome sequence of Ae. aegypti will stimulate efforts to elucidate interactions at the molecular level between mosquitoes and the pathogens they transmit. This already can be seen in, for example, analysis of components of the Toll immune signaling pathway (33) and identification of genes encoding insulin-like hormone peptides (34).

We expect that the sequence data will facilitate the identification of Ae. aegypti genes encoding recently described midgut receptors for dengue virus (35). Dengue vector competence is a quantitative trait, and multiple loci determine virus midgut infection and escape barriers (36). Unfortunately, the fragmented nature of the genome sequence and its low gene density have precluded its use in the identification of a comprehensive list of candidate genes for vector competence phenotypes or sex determination. The sequence may be used to improve the resolution of the current genetic map (37) and to integrate transcriptional profiling data with genetic studies (38), but filling gaps in the assembled sequence remains a high priority, especially when exploring genetic variations between the sequenced strain and field populations of Ae. aegypti.

The ongoing genome project on Culex pipiens quinquefasciatus, a vector for lymphatic filariasis and West Nile virus, will provide additional resources to underpin studies to systematically study common and mosquito species–specific gene function. Such analyses should improve our understanding of mosquito biology and the complex role of mosquitoes in the transmission of pathogens, and may result in the development of new approaches for vector-targeted control of disease.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Tables S1 to S23


References and Notes

Stay Connected to Science

Navigate This Article