special/reports

Inversions and Gene Order Shuffling in Anopheles gambiae and A. funestus

See allHide authors and affiliations

Science  04 Oct 2002:
Vol. 298, Issue 5591, pp. 182-185
DOI: 10.1126/science.1076803

Abstract

In tropical Africa, Anopheles funestus is one of the three most important malaria vectors. We physically mapped 157 A. funestus complementary DNAs (cDNAs) to the polytene chromosomes of this species. Sequences of the cDNAs were mapped in silico to theA. gambiae genome as part of a comparative genomic study of synteny, gene order, and sequence conservation between A. funestus and A. gambiae. These species are in the same subgenus and diverged about as recently as humans and chimpanzees. Despite nearly perfect preservation of synteny, we found substantial shuffling of gene order along corresponding chromosome arms. Since the divergence of these species, at least 70 chromosomal inversions have been fixed, the highest rate of rearrangement of any eukaryote studied to date. The high incidence of paracentric inversions and limited colinearity suggests that locating genes in one anopheline species based on gene order in another may be limited to closely related taxa.

Malaria morbidity and mortality in tropical Africa remain disproportionately high relative to other malaria-endemic areas of the world, in part because of three efficient vectors in subgenus Cellia: A. gambiae, A. arabiensis, and A. funestus. These species co-occur geographically across sub-Saharan Africa and can inhabit the same villages, shelter in the same houses, and feed on the same individuals. Yet A. funestus has evolved unique breeding site preferences, mating behavior, relative seasonal abundance, and degree of specialization on humans. The genetic basis of these differences is unknown. This species has received far less attention than A. gambiae, in part because obligatory swarming behavior associated with mating has been an obstacle to establishing laboratory colonies. However, it is obvious that successful malaria control strategies for Africa must take this and other species into account. The full genome sequence of A. gambiae (1) allows detailed comparative genomic analysis of closely related anopheline species. To the extent that synteny and colinearity are conserved between model organisms such as A. gambiae and more poorly characterized species such as A. funestus, comparative gene mapping is a powerful tool for candidate positional cloning. These comparisons could locate the genes responsible for ecological adaptations, speciation, insecticide resistance, host preference, and parasite defense before their signal is lost in a background of accumulated mutational noise.

Using fluorescence in situ hybridization (FISH), we mapped A. funestus cDNA clones on the five arms of the polytene chromosome complement (2). Incorporation of different fluorescent labels allowed us to probe simultaneously with two different cDNAs (Fig. 1). The cDNAs were isolated from a library prepared from A. funestus larval, pupal, and adult mRNA. The clones chosen for mapping were a subset of a larger collection of 3233 clones, whose sequence was determined from the 5′-end as part of an ongoing expressed sequence tag (EST) project (3). Based on the results of BLASTN (Basic Local Alignment Search Tool) searches against the A. gambiae genome, we selected A. funestus cDNAs whose scores had statistical significance thresholds of less than 10−6 (2). Of 157 cDNAs used as probes, 116 mapped to single chromosomal locations on theA. funestus cytogenetic map (111 of which are given in table S1), and the remainder hybridized in multiple locations (table S2). These constitute the only available physical map of this species. No genetic linkage map exists because controlled crosses cannot be performed. The cDNAs mapped not only to euchromatic bands but also to apparent interbands [for example, 24_G09 in subdivisions 10C, 12B, and 18A (Fig. 1B)]. Four others hybridized to a diffuse region of β-heterochromatin on 3L:38C-39A that serves as an attachment site to the nuclear envelope (Fig. 1C) (4).

Figure 1

(A to C) FISH performed on the chromosomes of A. funestus. Chromosomes counterstained with the fluorophore YOYO-1 and hybridized with fluorescently labeled probes Cy5 (blue) and Cy3 (red) are shown. The chromosome arm featured in each panel is identified at bottom left; the probe and numbered/lettered subdivision to which it hybridized are given beside arrows pointing to the corresponding signal.

The chromosomal positions of uniquely hybridizing A. funestus cDNAs are shown in Fig. 2, along with their corresponding locations in A. gambiae(2). Not shown are three cDNAs for which no significant BLASTN match was found to the A. gambiae genome: 04_F09, a putative inhibitor of apoptosis (2R:8E); 66_C12, similar to an A. gambiae protein with a C-type lectin domain (3R:35E); and 07_D02, function unknown (2R:13C). Recognizing that inferences about synteny, gene order, and sequence relationships require comparison of orthologous genes, we also did not include in Fig. 2 sequences that were dispersed to multiple locations in either species, because the hypothesis of gene orthology (related by speciation) versus paralogy (related by gene duplication) would be less firm. The relative positions of sequences with unique map locations in both species support the hypothesized chromosome arm homologies and the reciprocal whole arm translocation between 2L and 3R, postulated previously on the basis of relative length and banding pattern (4). Correspondence between chromosome arms was contradicted by only two of the cDNAs examined in this study. Clones 11_G12 and 04_A10 (not shown in Fig. 2) hybridized to 3R:32D and 3L:45A in A. funestus, yet the corresponding sequences in the A. gambiae genome are located on 3R:29A and 2R:18A, rather than on 2L and 3L, which is consistent with a transposition event. Aside from these two exceptions, the data reveal perfect conservation of synteny at the whole-arm level.

Figure 2

Physical location of A. funestus cDNA clones and location of the putative A. gambiae orthologs, given with respect to the polytene chromosome photomaps of both species. Homologous chromosome arms are juxtaposed, with A. funestus chromosomes shown above those of A. gambiae, oriented with centromeres on the right (indicated by asterisks). Only cDNAs that localized to one cytological position in both species are shown, with their relative positions on homologous arms indicated by interconnecting lines. Also indicated are the approximate breakpoints of common polymorphic inversions (identified by lowercase italicized letters), shown in their standard (uninverted) and presumed ancestral orientation (15, 16).

Within corresponding arms, paracentric inversions have had a major impact on genome architecture since the divergence of these species. Gene order has not been preserved along the length of any chromosome arm, although there are segments with conserved gene order (hereafter, “conserved segments”). Clear examples of conserved segments are located in regions near centromeres and telomeres where the rate of meiotic recombination may be reduced (Fig. 2). However, this generalization does not hold without exception, as 2R sequences near the telomere in A. funestus (08_H11, 01_H04) are located more proximally in A. gambiae, on the opposite side of a flanking marker (21_F03). This represents one of three small inversions that can be inferred at the distal end of 2R, including a rearrangement involving the 8C region in A. gambiae that contains the major Plasmodium-refractoriness locus Pen1(5). From these data, it is apparent that inversions have involved large as well as relatively small chromosomal segments.

What has been the extent of rearrangement of gene order between these species? The number of inversion events can be estimated by considering the mean length of conserved segments, because this length decreases with each inversion fixed since the divergence of A. gambiae and A. funestus from a common ancestor. The method of Nadeau and Taylor (6) was applied to estimate mean lengths of all conserved segments in the genome, based on the nucleotide distance in A. gambiae between the outermost markers that defined the segments observed in our sample. An assumption of the method, that rearrangements fixed during evolution are randomly distributed in the genome, seems unlikely given the extraordinary concentration of polymorphic inversions on 2R in both lineages. Of eight polymorphic inversions described in A. gambiae, seven occur on chromosome 2R (7) . Similarly, 11 of 15 polymorphic inversions found in A. funestus involve 2R (8). Accordingly, we assessed each arm independently. The estimated mean lengths of all conserved segments on each arm, defined with respect to A. gambiae, were X, 2.0 ± 0.2 megabases (Mb); 2R, 0.9 ± 0.2 Mb; 2L, 2.2 ± 0.4 Mb; 3R, 2.2 ± 1.0 Mb; and 3L, 1.1 ± 0.4 Mb. In a slight departure from Nadeau and Taylor (6), each rearrangement was assumed to be an inversion requiring two disruption events. Therefore,n inversions result in 2n + 1 conserved segments. We cannot discount the possibility that some disruptions of gene order were caused by intrachromosomal transpositions rather than inversions, events that are impossible to distinguish at this level of resolution. However, the contribution of transposition events was considered negligible.

Under these assumptions, the number of inversions on each arm was 5 ± 1, 36 ± 9, 11 ± 3, 11 ± 3, and 19 ± 5, respectively. Assuming a divergence time of 5 million years (My) (2), the rate of fixation per My for each chromosome arm can be estimated as 0.5, 3.6, 1.1, 1.1, and 1.9, respectively (or 7 when estimated across the genome). When normalized to account for differences in chromosome length, the number of inversions per Mb per My for X, 2R, 2L, 3R, and 3L was estimated as 0.023, 0.057, 0.022, 0.021, and 0.044, respectively (0.031 genome-wide). This rapid rate, even more extreme than the genome-wide estimate forDrosophila (9), is the highest reported for any eukaryotic species. Moreover, these results suggest that 2R has a higher rate of rearrangement than other arms. Higher resolution comparative studies of 2R are needed to provide insights about the mechanism and dynamics of paracentric chromosomal inversions.

It is clear that tightly linked genes in A. gambiaeare unlikely to be similarly linked in A. funestus, particularly on 2R. The estimate of mean conserved segment length derived for each arm can be used to predict the probability of linkage in A. funestus, given the known distance between genes inA. gambiae and the assumption of random distribution of breakpoints (6). As an example, the probability that genes 1 Mb apart on 2R in A. gambiae are linked on 2R inA. funestus is only 0.31. The probability of linkage conservation across this distance estimated over all five chromosome arms was higher, but still only 0.54. Both species possess polymorphic inversions on homologous arms that have been associated with differences in Plasmodium infection rates and resting behavior after feeding. The standard arrangement of inversion 2La inA. gambiae and the inverted arrangement of 3Rab in A. funestus are associated with higher infection rates and endophily (10, 11). The limited colinearity between species leaves no doubt that the inversions associated with these behaviors have not captured identical sets of genes between the breakpoints. Only by sequencing within inversions 3Rab will it be possible to define smaller segments in common with 2La and study their function and possible contribution to bionomic heterogeneities.

Has the rapid rearrangement of gene order been accompanied by a similarly rapid pace of evolution at the nucleotide level? The extent of sequence divergence can be estimated by comparison of samples of orthologous genes. Forty A. funestus ESTs were chosen that mapped to unique sites in the A. funestus genome and for which clear orthologs were available from A. gambiae in the dBEST database (2). Levels of divergence varied considerably among ESTs (table S3), which is consistent with the well-known heterogeneity among nuclear genes in other organisms. Across a total of 5832 aligned codons, sequence difference ranged from 2.25 to 29.41% at the nucleotide level, averaging 13.03 ± 6.05%. Corresponding amino acid divergence ranged even more widely, from 0 to 32.5% (average, 7.46 ± 7.59%).

A better reflection of the pattern of mutation is obtained by consideration of sites that are synonymous (where substitutions do not result in an amino acid change) separately from those that are nonsynonymous (where substitutions do result in an amino acid change). Nonsynonymous sites may be evolving under the constraint of purifying selection or the force of positive selection, depending on the structural and functional requirements of the protein. Synonymous sites, although not evolving neutrally if there is codon usage bias (12), can be used to approximate the neutral rate. The mean numbers of substitutions per synonymous site (K s) and nonsynonymous site (K a) were 0.612 ± 0.392 and 0.041 ± 0.044, although there was considerable heterogeneity among genes for both estimates, particularly for K a, and a large degree of uncertainty surrounding the estimates (table S3). Genes involved in adaptive processes, pathogen defense, and speciation are expected to show a higher nonsynonymous rate of substitution than those involved in “housekeeping” functions. The extremeK a values for the immune-related genesgambicin and lysozyme (0.144 and 0.124, respectively), as compared with the K a values for ribosomal proteins L8 and S10 (0.008 and 0.020, respectively), are consistent with this expectation.

An average synonymous rate has been estimated for mammalian andDrosophila genes. The Drosophila rate estimate, 16 per site per 109 years, is about four to five times higher than the mammalian rate of 3.5 per site per 109years (13). Taking 5 My as the divergence time betweenA. gambiae and A. funestus lineages, the average synonymous rate of substitution in Anopheles is ∼61 per site per 109 years. Several sources of error contribute to the Anopheles estimate, including limited sample size, sequencing errors in the EST database, and, most important, uncertainty in the divergence time. Nevertheless, a reasonable conclusion is that the rate of nucleotide substitution at silent sites inAnopheles is at least as fast as in Drosophila.

Throughout mosquito evolution, chromosome number (2N = 6) has remained stable (14), unlike the many changes in chromosome number that have characterized mammalian evolution. BetweenA. gambiae and A. funestus, species that diverged from one another about as recently as humans and chimpanzees, morphological change has been relatively slight. However, since the ancestral lines of these mosquito species split about 5 My ago, both gene order and gene sequences have been evolving at rates at least as high as those estimated for Drosophila, which are the highest rates known. Our results suggest that the success of positional cloning or interspecific microarray experiments may be limited to very closely related anopheline species. The availability of the completeA. gambiae genome sequence and its use in comparative studies with other anopheline species will greatly improve our understanding of how inversions arise, how they shape variation within species, and how they reshape genome architecture between species.

Supporting Online Material

www.sciencemag.org/cgi/content/full/298/5591/182/DC1

Materials and Methods

Tables S1 to S3

References and Notes

  • * To whom correspondence should be addressed. E-mail: besansky.1{at}nd.edu

  • These authors contributed equally to this work.

REFERENCES AND NOTES

View Abstract

Navigate This Article