Cloning of inversion breakpoints in the Anopheles gambiae complex traces a transposable element at the inversion junction

Abstract

Anopheles arabiensis, one of the two most potent malaria vectors of the gambiae complex, is characterized by the presence of chromosomal paracentric inversions. Elucidation of the nature and the dynamics of these inversions is of paramount importance for the understanding of the population genetics and evolutionary biology of this mosquito and of the impact on malaria epidemiology. We report here the cloning of the breakpoints of the naturally occurring polymorphic inversion 2Rd′ of A. arabiensis. A cDNA clone that cytologically mapped on the proximal breakpoint was the starting material for the isolation of a cosmid clone that spanned the breakpoint. Analysis of the surrounding sequences demonstrated that adjacent to the distal breakpoint lies a repetitive element that exhibits distinct distribution in different A. arabiensis strains. Sequencing analysis of that area revealed elements characteristic of transposable element terminal repeats. We called this presumed transposable element Odysseus. The presence of Odysseus at the junction of the naturally occuring inversion 2Rd′ suggests that the inversion may be the result of the transposable element’s activity. Characteristics of Odysseus’ terminal region as well as its cytological distribution in different strains may indicate a relatively recent activity of Odysseus.

The Anopheles gambiae complex is comprised of six sibling species, each one characterized by the presence of fixed paracentric inversions on their chromosomes (1). Two of these, A. gambiae sensu stricto and A. arabiensis, are the most important vectors of human malaria and are subdivided into discrete subpopulations, each carrying a unique set of polymorphic chromosomal inversions. Some of these inversion polymorphisms have been associated with differences in seasonality of breeding, adaptation to natural vs. human-disturbed habitats, microhabitat selection, and host preferences (2, 3). Of particular epidemiological significance is the association of certain chromosomal arrangements with higher levels of Plasmodium infection (4).

The elucidation of the DNA structure of the inversion breakpoint neighborhood is of special interest for two main reasons. First, it should provide information on long standing evolutionary and population biology questions such as their monophyletic origin (5), the mechanisms of their maintenance in nature (6), or the involvement of transposable elements (TEs) in their generation (7, 8). Second, from an epidemiological point of view, the nucleotide sequence around the breakpoint junction may be used to design a simple and unambiguous PCR-based assay that could overcome difficulties of the present karyotyping procedure, i.e., the direct microscopic observation of the banding pattern of polytene chromosomes (9). Though accurate, this technique may be applied only on particular mosquito tissues and life stages and requires special expertise for the interpretation of the banding pattern.

Here we report the cloning of the breakpoints of the naturally occurring 2Rd′ inversion of A. arabiensis. Analysis of the sequences surrounding the inversion breakpoints revealed the presence of a TE at one of the breakpoint junctions, which suggests the involvement of such elements in the generation of natural inversions in the A. gambiae complex.

EXPERIMENTAL PROCEDURES

Mosquito Strains.

DNA for all molecular manipulations came from the following laboratory strains of A. gambiae. The inversion formula for each strain represents the relationship of the particular strain to the standard chromosome sequence for the five polytene arms (X, 2R, 2L, 3R, and 3L). Letters appearing singly represent homozygosity for the particular inversions (2).

GASUA (Xag, 2R, 2La, 3R, and 3L) was derived from an A. gambiae sensu stricto laboratory colony established in our laboratory in 1986 from adult females collected at Suakoko (Liberia), that were polymorphic for inversions 2Rd and 2La. ARZAG (Xbcd, 2Rab, 2Rd′/+, 2La, 3Ra/+, and 3L) was derived from an A. arabiensis colony established in our laboratory in 1987 from adult females collected at Zagthouli (Burkina Faso). ARMOR (Xbcd, 2Ra/+, 2Rb, 2Rc/+, 2Rd′/+, 2La, 3Ra/+, and 3L) was derived from an A. arabiensis colony established in our laboratory in 1996 from adult females collected at Moribabougou (Mali).

Cosmid Library Construction and Screening.

The GASUA cosmid library was constructed by Stratagene from mosquito DNA that was partially digested with Sau3AI and introduced into the BamHI site of vector pWE15. To screen the library, the insert of pKM13 was gel-purified and labeled by random priming. Positive clones were repurified in a secondary screening and restricted with either NotI and PstI or NotI and HindIII to identify distinct cosmid classes.

Standard and Fluorescent in Situ Hybridization (FISH).

Polytene chromosome spreads were obtained from half-gravid female mosquitoes by using standard methods (10). Standard in situ hybridization was performed according to Kumar and Collins (10) and della Torre et al. (11). FISH was performed as detailed by Gatti et al. (12) and chromosome identification was obtained by simultaneous 4′,6-diamidino-2-phenylindole (DAPI) staining. Digital images were obtained using a computer-controlled Zeiss Axioplan epifluorescence microscope equipped with cooled CCD camera (Photometrics, Tucson, AZ). Fluorescein isothiocyanate and DAPI fluorescence was detected by using specific filter combinations, recorded separately as gray scale images, and then pseudocolored and merged using the adobe photoshop 3.0 software on a Macintosh computer.

DNA Manipulations.

Large scale or single mosquito DNA preparations were performed as described in ref. 13. Manual double-stranded sequencing was performed on subcloned fragments by using the Pharmacia sequencing kit. Labeling of probes was performed by random priming purified inserts of the different clones.

PCR Amplification of the Breakpoints.

One hundredth of a single mosquito DNA preparation was used for the amplification of the breakpoint junction in a 50 μl of reaction containing 1X PCR buffer, 1.5 mM MgCl₂, 10 mM of each of the four dNTPs, 50 pmols of each primer, and 1 unit of Taq polymerase. Reaction conditions were: 10 min at 95°C; 30 cycles of 30 sec at 95°C, 1 min at 52°C, and 30 sec at 72°C; final extension of 7 min at 72°C. Primers used were: region a primer, AGG TGT ACA TAA AAC AAT ACG C; region b primer, GAT ATC ACT CCC GCT ACC AT; region c primer, CAG GGC AGT AGG CGA GTA TGT; and region d primer, GTC TGC TCT ACT AGT AGT T.

RESULTS

Isolation of Breakpoint Sequences in A. arabiensis.

From a large collection of cDNA, randomly amplified polymorphic DNA (RAPD), and cosmid clones cytogenetically characterized (11), we looked for clones that could be used as starting material for chromosomal walks toward inversion breakpoints. We chose cDNA clone pKM13 that mapped in the immediate vicinity of the proximal breakpoint of paracentric inversion 2Rd′ of A. arabiensis. Screening the GASUA cosmid library with the purified insert of clone pKM13 yielded three distinct and overlapping cosmids. FISH of the three clones on 2Rd′/+ chromosomes gave three hybridization signals at the breakpoint junctions. This situation would be expected if the clones spanned the breakpoint, as is illustrated in Fig. 2 for a subclone, pGA-cd (Middle row).

Figure 2.

Localization of 2Rd′ cd breakpoint. (A) Fluorescent in situ hybridization of probes pGA-d (Top), pGA-cd (Middle) and pGA-c (Bottom) shown in Fig. 1. (B) Schematic representation of FISH images. (C). Schematic interpretation of signals obtained on heterozygote 2Rd′/+ chromosomes in A. For the formation of the heterozygote chromosome, one chromatid of each homozygote is used. Therefore, a probe that spans the breakpoint (pGA-cd) would give one signal on the homozygote standard chromosomes, two signals on the homozygote inverted chromosomes, and three on the heterozygote chromosomes.

Subcloning of the 2Rd′ cd Proximal Breakpoint.

Cosmid cosKM13–1 was selected to identify and subclone a smaller fragment that contained the breakpoint. First, a restriction map of cosKM13–1 of endonucleases BamHI, EcoRI, EcoRV, HindIII, and PstI was generated (Fig. 1). Subsequently, using as an assay the characteristic three hybridization signals in a FISH on the 2Rd′/+chromosomes, we searched for gradually smaller subclones within larger ones. We thus identified a 1.5-kb NsiI fragment within a larger 10-kb EcoRV fragment that contained the breakpoint. Curiously, consecutive fragments EcoRV–PstI (1.3 kb) and PstI–PstI (2.1 kb) appeared inside and outside the inversion, respectively. This indicated that the breakpoint was very close to the common PstI site of the two fragments, which was later confirmed by sequencing analysis.

Figure 1.

Restriction map of cosmid clone cKM13–1 and a few of the purified fragments and isolated subclones discussed in the text. B, BamHI; H, HindIII; I, EcoRI; V, EcoRV; N, NsiI; and P, PstI. Darkened areas at the ends of the cosmid clone indicate the beginning of vector sequences. The black arrow indicates the approximate location of the breakpoint. The c area lies to its left and the d area to its right.

Isolation of the 2Rd′ ab Distal Breakpoint.

This was achieved in a two-step procedure, first by obtaining bd breakpoint sequences from A. arabiensis and then by using the b sequences of A. arabiensis as a probe to select clones from the GASUA library that contained the ab region. In the first step, the cd breakpoint probe (the 1.5-kb NsiI fragment), used in genomic Southerns, identified a 1.6-kb fragment of the A. gambiae NsiI digestion (corresponding to the NsiI fragment of the standard orientation of 2Rd′) and three A. arabiensis fragments (one corresponding to the NsiI fragment of the standard orientation and the other two corresponding to the fragments spanning the two breakpoints of the inversion) (Fig. 3, step 1). ARZAG genomic DNA corresponding to these three fragments was gel-purified and subcloned into the PstI site of pcDNAII vector. Three positive clones were purified, two of which had a 1.5-kb insert and one a 1.15-kb insert (Fig. 3, step 3). Southern hybridization of the above three clones with probes pGA-cd, pGA-c, and pGA-d indicated that the 1.5-kb fragment corresponded to the A. arabiensis cd breakpoint region whereas the 1.15-kb fragment corresponded to the bd breakpoint.

Figure 3.

Isolation of 2Rd′ ab breakpoint. Step 1, Southern hybridization of A. gambiae (GA) and A. arabiensis (AZ) genomic DNA digested with restriction enzyme NsiI and probed with labeled pGA-cd; M, 1 kb molecular weight marker lane. The arrow to the left of the autoradiogramme indicates the 1.6-kb NsiI A. gambiae-expected fragment; the arrows to the right indicate three A. arabiensis NsiI fragments that were purified and subcloned. Step 2. Transformants were screened with the insert of pGA-cd. Step 3. Positive clones were analyzed by restriction and Southern analysis. Step 4. Clone pAZ-bd was sequenced and its sequence was compared with that of pGA-cd (also see Fig. 4). Step 5. Primers were designed to PCR amplify the region of pAZ-bd nonhomologous to clone pGA-cd (i.e., region b). The amplified fragment was labeled and used as a probe to screen the GASUA cosmid library. Step 6. Positives clones were purified and analyzed by in situ hybridization.

In the second step, clone pAZ-bd was sequenced (Fig. 3, step 4), and its sequence was compared with that of pGA-cd. The comparison clearly defined two areas, one of almost complete identity and one of no similarity at all (Fig. 4B). The inversion breakpoint was assumed to lie at the junction of these two areas. PCR primers were designed to amplify only the b side of the breakpoint, which was then used to screen the GASUA cosmid library. A very large number of positive clones were obtained, and of those tested by in situ hybridization, they all gave multiple signals, which indicated the presence of a repetitive sequence in the b probe. Only one clone, cosB14, gave an in situ signal at a position that corresponded to the distal 2Rd′ breakpoint. This clone was chosen for further analysis. A 480 bp EcoRI-EcoRV fragment was identified from cosB14 that hybridized to pAZ-bd by Southern hybridization. This fragment was subcloned into vector pcDNAII (clone pGA-ab) and fully sequenced.

Figure 4.

Dot plots of sequence comparisons of the regions of similarity of clones pAZ-bd and pGA-ab (A) and pAZ-bd and pGA-cd (B). A window of 21 bases with a stringency of 14 bases was used. Below each plot the nucleotide sequence around the breakpoints is shown in more detail. In the box are the 7 bp of unknown origin found at the breakpoint junction.

Sequence Features of 2Rd′ Breakpoint Regions.

Sequence comparisons were performed among clones pGA-cd, pGA-ab, and pAZ-bd (Fig. 4). Nucleotide alignment showed that homology between pAZ-bd and pGA-ab is interrupted at the point where homology between pAZ-bd and pGA-cd starts, except for a short 7-bp sequence (GATAGGG) present in pAZ-bd but absent in either pGA-ab or pGA-cd. Alignment between pAZ-bd and pGA-cd also revealed a 110-bp deletion in pAZ-bd that is shown in Fig. 4B as an interruption of the straight line of homology between the two clones. The same deletion also is present in clone pAZ-cd, the 1.5-kb NsiI subclone of Fig. 3 (step 3) that contained the 2Rd′ cd region in ARZAG. This deletion can also explain the observed difference of the 1.6-kb GASUA and the 1.5-kb ARZAG (presumably homologous) bands in the Southern hybridization of Fig. 3, step 1.

TE Terminal-Like Sequences at the bd Breakpoint Junction.

The aforementioned 7-bp sequence of the bd breakpoint junction as well as other characteristic sequences that follow downstream are graphically presented in Fig. 5 and discussed in the next section. The bd breakpoint also was cloned by PCR amplification from a different A. arabiensis strain (ARMOR). Its sequence comparison to that of the ARZAG strain revealed a reorganization of the characteristic sequences present at the breakpoint junction (Fig. 5). These features are reminiscent of TE terminal repeats, as is presented in the Discussion section.

Figure 5.

Characteristic features of the 2Rd′ bd breakpoint. (A) Approximate location and direction of the original cDNA clone 13 which ends ≈100 bp before the breakpoint. Also shown are the differences at the end of the b region between the two A. arabiensis strains. (B) Details of the characteristic internal inverted repeats of the 69-bp terminal sequence. Dark arrows indicate the largest inverted repeat observed. Bars above the sequence indicate shorter inverted repeats.

The characteristics of the b junction sequences, the repetitiveness of the sequences of the b region tested (about 2.5 kb) in the chromosomes, and the distinct distribution of these sequences in different A. arabiensis strains (see below) were considered as proof for the presence of a new TE in the A. gambiae complex. We will use the name Odysseus for these sequences in the remaining text.

Distribution of Odysseus in A. arabiensis.

A ≈2.5 kb subclone of cosmid cosB14 containing almost exclusively junction b sequences was biotin-labeled and used in in situ hybridization on two different A. arabiensis strains. Thus Odysseus demonstrated 19 points of insertion in the ARZAG strain and 15 in the ARMOR strain, only eight of which seemed identical between the two strains.

ORFs Around the Breakpoints.

Although we have not performed a detailed analysis of the transcriptional activity of the area around the breakpoints, the major ORF corresponds to the original pKM13 cDNA clone. This ORF has its polyadenylation site at a distance of about 100 bp from the cd breakpoint, and is therefore not directly affected by the inversion.

A PCR Diagnostic for 2Rd′.

Region b, c, and d primers were used in a PCR amplification to test for the presence or absence of 2Rd′ in individual mosquitoes. Presence of a 1.6-kb cd band (1.7 kb for A. gambiae) indicates the absence of 2Rd′, presence of a 1.7-kb bd band indicates the presence of the inversion, whereas presence of both bands indicates that the individual is heterozygote for that inversion. Fig. 6 shows the amplification results of an A. gambiae (GASUA) mosquito and six A. arabiensis (ARZAG). Results obtained were consistent with the cytological determination of the individual mosquitoes. Note also that the cd band of the GASUA mosquito is slightly larger, which confirms the presence of the insertion in the d area presented in the GASUA mosquitoes. Unlike cytological examination that can only be applied to half gravid female mosquitoes, PCR diagnosis is applicable to both sexes and all developmental stages. This can be a powerful tool for field studies of population dynamics and epidemiological correlations of this inversion. It should also be pointed out that A. gambiae sensu stricto has a parallel polymorphism represented by inversion 2Rd whose breakpoints both lie close but do not coincide to those of inversion 2Rd′. Thus, primers b and d are not expected to PCR amplify an A. gambiae sensu stricto mosquito. This was verified on a few field collected A. gambiae (2Rd/+) specimens (data not shown).

Figure 6.

PCR diagnostic for 2Rd′. Individual mosquitoes assayed for the presence of 2Rd′. The top part of the gel shows amplification with primers c and d (A). The bottom part of the gel shows amplification with primers b and d (B). Presence or absence of the corresponding bands determines the status of the individual with regard to 2Rd′.

DISCUSSION

Studies on inversion breakpoints should contribute on questions that regard the monophyletic origin of inversions, the involvement of TEs in their generation, or the mechanisms of their maintenance in nature. Here we present the cloning of the breakpoints of polymorphic inversion 2Rd′ of A. arabiensis. A cDNA clone that cytologically mapped close to the proximal breakpoint of inversion 2Rd′ was used as a probe to obtain a cosmid from an A. gambiae library. In situ hybridization demonstrated that the cosmid contained the proximal breakpoint junction. A smaller subclone used in Southern hybridizations identified fragments of A. arabiensis that contained the distal breakpoint. Sequencing and in situ analysis of these clones provided two major findings: precision at the nucleotide level of the breakpoint junctions and the presence of a TE at the distal junction.

Breakpoint Junction Precision and Inversion Breakpoints Cloned in Drosophila.

The only two examples of natural inversion breakpoints cloned in dipterans refer to the polymorphic inversion In(3L)Payne of D. melanogaster (14) and to an inversion fixed during the divergence of D. melanogaster and D. subobscura (15). In(3L)P was shown to have a precise breakpoint junction, whereas the fixed inversion of D. melanogaster and subobscura revealed a larger region of ≈600 bp that delimited the breakpoint. The breakpoints of 2Rd′ of A. arabiensis demonstrate an almost precise cut (Fig. 4), except for the 7-bp of unknown origin found at the junction, which may reflect a target-site duplication event during the insertion or excision of the observed TE.

Role of TEs in the Generation of Inversions.

In Drosophila, at least three distinct types of transposons have been implicated in the production of chromosome rearrangements: the short inverted repeat transposons P (16) and hobo (17), the long inverted repeat transposon FB (18), and the LINE-like element I (19). It is thought that ectopic recombination between two elements in which the elements provide the necessary DNA sequence homology for ectopic pairing may be responsible for the generation of the rearrangements (7). No such elements were detected in the two aforementioned Drosophila inversion breakpoints cloned. In this report, we present evidence of a TE, hereafter called Odysseus, found exactly at the distal breakpoint junction of 2Rd′. The evidence comes from four observations.

First, the abundance of Odysseus on the polytene complement (15–20 copies) and its distinct distribution in different laboratory stocks are characteristic of actively transposing elements. We still do not know what constitutes an intact Odysseus or how many of the elements observed cytologically are intact, both of which are topics of our current research.

Second, the terminal sequence at the breakpoint junction can be divided in three zones (Fig. 5B): a 69-bp zone, very rich in secondary structures (the most characteristic being a 17+7 almost perfect stem-and-loop structure from nucleotide 26 to nucleotide 67; stem-and-loop structures are thought to be regulatory regions); an 11-bp box; and a ≈200-bp long AT-rich zone. This is reminiscent of terminal sequences of several TEs as, e.g., are the five regions in the terminal 163-bp sequence of the P element (20, 21). Additionally, we find a 7-bp sequence of unknown origin at the bd junction, a sequence reminiscent of target site duplication in other elements.

Third, preliminary sequencing data from three of the 20 cosmids isolated with probe b of clone pAZ-bd identified sequences homologous to Odysseus. Very high homology is observed along the parts sequenced, whereas the 69+11 bp region is entirely conserved (data not shown). These cosmids do not contain the ab breakpoint because they do not hybridize to that area by in situ hybridization. We believe that they correspond to distinct insertion sites of Odysseus. These clones are under further investigation.

Fourth, sequencing comparison of the bd breakpoint between two A. arabiensis strains revealed a precise multiplication of Odysseus’ terminal sequences. Although the ARZAG strain ends in an “element-11–69-7 bp target” motif, the ARMOR strain ends in an “element-11–69-69–11-69–7 bp target” motif. An interesting explanation could be that this multiplication reflects the remains of a nearly precise excision, as is the case in Tn10, for example (22), or some other activity that took place subsequently. This notion is particularly intriguing because it can argue for a recent activity of Odysseus.

Recent TE Activities and Natural Inversions.

Several TEs are thought to have only invaded their host’s genomes in recent times. The P family of TEs, for example, appears to have spread throughout the cosmopolitan D. melanogaster during the third quarter of this century (23) following horizontal transfer from D. willistoni (24). Despite the initial reduction in fitness caused by hybrid dysgenesis (25), TEs are thought to rapidly spread due to both their high replicative ability and the passive transport of Drosophila around the world associated with human activities. From the many elements capable of inducing rearrangements in laboratory populations of Drosophila, the most likely candidate responsible for regularly doing so in natural populations, is hobo. This element has been cytologically associated with both breakpoints in three of four endemic Hawaiian inversions and with one of the breakpoints in a fourth one. Interestingly, there was no association observed with any of the cosmopolitan inversions (8). In fact, it is thought that the frequency of endemic inversions may have been lower before the arrival of TEs. Cosmopolitan inversions predate the introduction of hobo, P, and I and appear to show no association with them. Endemic inversions apparently have short life spans in D. melanogaster populations, as do TE insertion variants (26).

There is still no complete study that has addressed the age of A. gambiae inversions. Clearly there are indications that inversions like 2La (fixed in A. arabiensis and A. merus and very frequent and widespread in A. gambiae) are among the most ancient in the complex (M.C., unpublished observations; ref. 27). Others, like 2Rbc are considered more recent and associated with human activities in rice field irrigation in West Africa (3). Inversion 2Rd′ is widespread throughout Africa west of the Great Rift Valley, but it is apparently missing in all the samples southeast of the Valley studied so far. Its range goes from Sudan and Eritrea, where its frequency is ≈2% (28, 29) to any of the studied West African sites, where its mean frequency is ≈8%, with peaks of 14 to 20% in localities of Nigeria, Niger, Cameroon, and Senegal (30). Assuming that the origin of the 2Rd′ is monophyletic, that it rose in West Africa (where its frequency is higher) and spread eastwards (where its frequency is lower or zero), and, finally, assuming that the range of distribution of a given inversion is as wide as it is ancient, then 2Rd′ should be comparatively more recent than other A. arabiensis inversions like 2Rb and 3Ra, which are widespread over the range of distribution of the species. These observations support the hypothesis of Odysseus having been responsible for the generation of 2Rd′ in relatively recent times and having remained active since.

The picture that arises emphasizes the important role of TEs in generating natural inversions. Being mobile, these elements excise and are often lost from the breakpoint junctions. The older the inversion, the less likely it is to find the element at the breakpoints, which would be the case for most cosmopolitan inversions. Newer inversions, on the contrary, are more likely to have at least traces of these elements at one or both of the breakpoints. This may be the case of the endemic Hawaiian inversions in Drosophila and of 2Rd′ of A. arabiensis.

Apart from the effects on the biology of their hosts, TEs are of interest for their potential as vehicles for introducing exogenous DNA into the germline, or as means of driving genes of interest into natural populations. To date, at least six families of TEs have been characterized in A. gambiae: four retrotransposons without long terminal repeats, T1 (31), RT1 and RT2 (32), and Q (33); and two inverted-repeat families, mariner (34) and Pegasus (35). However, none of these has been proven capable of introducing exogenous DNA in the mosquito. The effort to discover such a vector is of particular importance in A. gambiae because it could open the possibility of introducing genes (such as Plasmodium resistance or selective pesticide sensitivity genes) that could be used in the fight against malaria transmission.

ABBREVIATIONS

TE

transposable element

FISH

fluorescent in situ hybridization