Eight novel families of miniature inverted repeat transposable elements in the African malaria mosquito, Anopheles gambiae

Abstract

Eight novel families of miniature inverted repeat transposable elements (MITEs) were discovered in the African malaria mosquito, Anopheles gambiae, by using new software designed to rapidly identify MITE-like sequences based on their structural characteristics. Divergent subfamilies have been found in two families. Past mobility was demonstrated by evidence of MITE insertions that resulted in the duplication of specific TA, TAA, or 8-bp targets. Some of these MITEs share the same target duplications and similar terminal sequences with MITEs and other DNA transposons in human and other organisms. MITEs in A. gambiae range from 40 to 1340 copies per genome, much less abundant than MITEs in the yellow fever mosquito, Aedes aegypti. Statistical analyses suggest that most A. gambiae MITEs are in highly AT-rich regions, many of which are closely associated with each other. The analyses of these novel MITEs underscored interesting questions regarding their diversity, origin, evolution, and relationships to the host genomes. The discovery of diverse families of MITEs in A. gambiae has important practical implications in light of current efforts to control malaria by replacing vector mosquitoes with genetically modified refractory mosquitoes. Finally, the systematic approach to rapidly identify novel MITEs should have broad applications for the analysis of the ever-growing sequence databases of a wide range of organisms.

Figure 1

Evidence of past mobility of some of the newly discovered MITEs in A. gambiae. The names of these MITEs are described in Table 1. The sequences were aligned by using gap of GCG (Genetics Computer Group, Madison, WI, Version 10, 1999) with gap weight = 40 and gap length weight = 0. The top sequences in the alignments contain MITE insertions that are not present in the bottom sequences. The bottom sequences were identified in the A. gambiae sequence tagged site (STS) database during blast searches using sequences flanking MITEs as queries. Two elements, one from the TA-IIα-Ag family (AL151950) and the other from the TA-III-Ag family (AL155989), are inserted in a middle repetitive sequence (37). The putative target duplications are underlined. Note that the target duplication flanking the TAA-II-Ag in AL141968 is different from the target consensus TAA.

Figure 2

Predicted secondary structure of the consensus sequence of TAA-II-Ag. Multiple sequence alignment of the full-length elements used to create the consensus sequence has been deposited in the EMBL database (accession no. DS43382). The structure was plotted by using genequest of Lasergene (DNASTAR, Madison, WI), which is identical to the structure predicted by using mfold of GCG (Genetics Computer Group, Madison, WI, Version 10, 1999).

Figure 3

(A) Pairwise comparison between the consensus sequences of the two subfamilies of TA-I-Ag: TA-Iα-Ag and TA-Iβ-Ag. Multiple sequence alignments of the full-length elements used to create the two consensus sequences were deposited in the EMBL database (accession nos. DS43384 and DS43385). The two consensus sequences were aligned by using gap of GCG (Genetics Computer Group, Madison, WI, Version 10, 1999) with gap weight = 30 and gap length weight = 1. Thick arrows mark the TIRs, and thin arrows mark the subterminal repeats. Flanking TA target duplications are not shown. D = A, G, T; H = A, C, T; K = G, T; M = A, C; N = A, C, G, T; R = A, G; S = G, C; V = G, A, C; W = A, T; Y = C, T. (B) Pairwise comparison between the consensus sequences of the two subfamilies of TA-II-Ag: TA-IIα-Ag and TA-IIβ-Ag. Multiple sequence alignments used to create the consensus sequences were deposited in the EMBL database (accession nos. DS43376 and DS43377). The two consensus sequences were aligned by using gap as described in A. All symbols are as in A.

Figure 4

Average AT contents of MITEs and their flanking sequences compared with STS and EST sequences in the A. gambiae database. AT contents of all full-length MITEs (see Table 1 for sample sizes) and their flanking sequences (STS minus MITE, indicated by the suffix “F”), and all of the 17,509 STS sequences in the A. gambiae genome database were calculated. Calculations of the Pegasus elements and their flanking regions were based on sequences reported by Besansky et al. (22). The forward and reverse sequences of the A. gambiae ESTs were analyzed separately because many of them represent pairs of sequences covering different regions of the same clone. Two hundred ESTs were randomly selected from each of the 2,990 forward ESTs and the 2,936 reverse ESTs (27). They were analyzed by using blast to remove redundancy that resulted from multiple copies of cDNAs. AT contents of 186 nonredundant forward ESTs (EST-For) and 181 nonredundant reverse ESTs (EST-Rev) were calculated and analyzed. Data points represent the mean AT contents. The error bar represents the SEM. Note that the standard errors for several data points are too small to be shown at the current scale. Mann–Whitney tests were used to compare the medians at α = 0.05. In most cases, t-tests were also used to compare the means, which gave the same conclusions. Samples in tier I have significantly higher AT contents than samples in tier II and III, whereas most samples in tier II have significantly higher AT contents than samples in tier III. One exception is TAA-I-AgF of tier II, which has a small sample size. Its AT content is neither significantly higher than samples in tier III nor significantly lower than TA-Iα-AgF, TA-IIα-AgF, TA-IV-AgF, and TA-IV-Ag of tier I. The other exception is the comparison between TAA-I-Ag and TA-IV-Ag, which is not significantly different. Samples in tier II are not significantly different from each other while EST-For is slightly more AT-rich than EST-Rev in tier III (P = 0.045). A few samples in tier I are slightly more AT-rich than others.