Co-evolution between Plasmodium species and its vectors may result in adaptive changes in genes that are crucial components of the vector's defense against the pathogen. By analyzing which genes show evidence of positive selection in malaria vectors, but not in closely related non-vectors, we can identify genes that are crucial for the mosquito's resistance against Plasmodium.
We investigated genetic variation of three anti-malarial genes; CEC1, GNBP-B1 and LRIM1, in both vector and non-vector species of the Anopheles gambiae complex. Whereas little protein differentiation was observed between species in CEC1 and GNBP-B1, McDonald-Kreitman and maximum likelihood tests of positive selection show that LRIM1 underwent adaptive evolution in a primary malaria vector; An. arabiensis. In particular, two adjacent codons show clear signs of adaptation by having accumulated three out of four replacement substitutions. Furthermore, our data indicate that this LRIM1 allele has introgressed from An. arabiensis into the other main malaria vector An. gambiae.
Although no evidence exists to link the adaptation of LRIM1 to P. falciparum infection, an adaptive response of a known anti-malarial gene in a primary malaria vector is intriguing, and may suggest that this gene could play a role in Plasmodium resistance in An. arabiensis. If so, our data also predicts that LRIM1 alleles in An. gambiae vary in their level of resistance against P. falciparum.
Citation: Slotman MA, Parmakelis A, Marshall JC, Awono-Ambene PH, Antonio-Nkondjo C, Simard F, et al. (2007) Patterns of Selection in Anti-Malarial Immune Genes in Malaria Vectors: Evidence for Adaptive Evolution in LRIM1 in Anopheles arabiensis. PLoS ONE 2(8): e793. https://doi.org/10.1371/journal.pone.0000793
Academic Editor: Leah Cowen, University of Toronto, Canada
Received: June 29, 2007; Accepted: July 18, 2007; Published: August 29, 2007
Copyright: © 2007 Slotman et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This research was supported by National Institutes of Health grant RO1 A1 046018 to JRP. Additionally, AP was supported by a Marie Curie Outgoing International Fellowship (Contract No. MOIF-CT-2006-021357).
Competing interests: The authors have declared that no competing interests exist.
Despite ongoing control efforts during the last decades, malaria remains one of the most deadly infectious diseases. The vast majority of its burden is carried by people on the African continent, where 1 to 2 million people die annually from this disease . Current malaria control efforts are hampered by the spread of insecticide and drug resistance, which has inspired research programs aimed at the development and eventual release of genetically altered mosquitoes that would be resistant to Plasmodium falciparum transmission. The need to identify refractory genes for this effort has focused much attention on the immune system of malaria's main vector in Africa, An. gambiae. The completion of the An. gambiae genome  has greatly facilitated research in this direction, and various anti-malarial immunity genes have now been identified [e.g. 3]–. Additionally, two recent studies provided many candidate anti-malarial immune genes that are up-regulated in response to Plasmodium infection , .
So far little attention has been devoted to examining polymorphism of immunity genes in natural malaria vector populations , . It is known however that molecules that are involved in interactions with pathogens, such as immune genes, are one of the major types of proteins on which positive selection has been demonstrated , . Presumably, this is because such genes are involved in co-evolution between hosts and pathogens. In the case of malaria, if Plasmodium infection affects the mosquito's fitness, we may expect the accumulation of adaptive amino acid substitutions in those anti-malarial genes that are crucial in specifically limiting Plasmodium infection in vector species, whereas such changes should not be found in closely related species that do not transmit malaria.
That An. gambiae has in fact undergone an adaptive response to P. falciparum infection is suggested by several lines of evidence. First of all, P. falciparum goes through severe bottlenecks during its life cycle in this mosquito , demonstrating that the mosquito immune system is limiting the Plasmodium infection. Furthermore, P. berghei, which is not transmitted naturally by An. gambiae, produces a much higher oocyst number in An. gambiae than its natural pathogen P. falciparum. In fact, in a review of studies estimating the fitness effect of Plasmodium infection on Anopheles species, reduced fitness was observed in 10 combinations of Plasmodium and Anopheles species that do not occur naturally, whereas in 10 natural combinations, including An. gambiae and P. falciparum, no fitness effects were observed . This is an indication that Anopheles species have evolved to limit infections of the Plasmodium species they come into contact with. This is corroborated by the fact that the immune response to P. falciparum is specific, i.e. An. gambiae up-regulates different genes in response to infection with P. falciparum vs. P. berghei , . Salivary gland infection rates of P. falciparum in An. gambiae and An. arabiensis are typically low, ranging between 3–9% . This raises the question whether selection pressures on the mosquito immune system are strong enough to result in an adaptive response to P. falciparum infection. However, it should be kept in mind that the data summarized above indicate that the rate and intensity of P. falciparum infection is likely to have been much higher when Anopheles mosquitoes first came into contact with this pathogen.
An. gambiae belongs to a complex of closely related species that includes another primary African malaria vector, An. arabiensis. Additionally it contains several species, i.e. An. melas, An. merus and An. bwambae, that occasionally transmit malaria locally, but do not have wide enough distributions to be considered important vectors. More importantly for the purpose of the present study, the An. gambiae complex also contains the highly zoophilic An. quadriannulatus A and An. quadriannulatus B, which are never or rarely exposed to the human-limited P. falciparum.
In this study, we investigated patterns of polymorphism in three anti-malarial genes, i.e. CEC1, GNBP-B1 and LRIM1, in six species of the An. gambiae complex. CEC1 (ENSANGG00000009468, www.ensembl.org/Anopheles_gambiae) is a cecropin gene whose expression in An. gambiae is induced by infection with bacteria and Plasmodium berghei . Additionally, genetically modified An. gambiae that express CEC1 24 hours after a blood meal, showed a 60% reduction in the number of P. berghei oocysts . GNBP-B1 (ENSANGG00000015205) is a pattern recognition receptor whose expression is strongly upregulated in response to infection with both P. berghei , and P. falciparum . LRIM1 (ENSANGG00000010552) is a leucine-rich repeat immune protein that is an important plasmodium antagonist. This protein is up-regulated in response to infection with P. berghei, and silencing of this gene increases oocyst load 3.6-fold . Furthermore, this gene has been implicated in the melanization reaction of parasites . We performed various tests for positive selection on these anti-malarial genes in the two main vectors, An. gambiae and An. arabiensis, to examine if these genes show signs of an adaptive response that may implicate them in the co-evolution of the mosquito vector and Plasmodium pathogen. Whereas no evidence for positive selection was found in CEC1 and GNBP-B1, our results clearly indicate that LRIM1 underwent an adaptive response in the An. arabiensis lineage. Additionally our data also indicate that LRIM1 has introgressed from An. arabiensis into An. gambiae.
The complete CEC1 gene, consisting of 177 bp of coding sequence and two introns comprising a combined 90 bp, was amplified. A total of 186 alleles of this gene were obtained from six species of the An. gambiae complex, 66 of which were unique (Table 1, genbank accession nos EU073463–EU073527). Although several polymorphisms were shared between species, none of the alleles were. For the coding region, Dxy, the average number of nucleotide substitutions between alleles in different species, ranged from 0.829 to 2.54 (per 100 bp). Very few fixed differences were present between species, and in most comparisons no fixed non-synonymous differences were found (Table 2). Not surprisingly therefore, none of the McDonald-Kreitman tests indicated an excess of non-synonymous fixed differences between species. In particular, no fixed amino-acid changes were observed between the non-vector species An. quadriannulatus A and the two major malaria vectors An. gambiae and An. arabiensis.
For GNBP-B1 a total of 38 alleles from six species, consisting of the complete 1188 bp coding sequence and 232 bp of intron sequence, were obtained (Table 1, genbank accession nos EU073426–EU073462). All of these alleles were unique, but some polymorphisms were shared between species. Dxy ranged from 0.723 to 2.5 (per 100 bp) for the coding region. Very few fixed replacement substitutions were observed between species (Table 3). Two of the McDonald-Kreitman tests were significant. However, both indicated an excess of non-synonymous polymorphisms, and in all comparisons the ratio of non-synonymous to synonymous substitutions was higher for polymorphisms than for fixed differences. Between the non-vector species An. quadriannulatus A and the malaria vectors An. arabiensis and An. gambiae, only a single replacement substitution was observed.
For LRIM1 we sequenced 858 bp that were thought to represent a single exon constituting the entire gene. However, in the most recent release of the Ensembl An. gambiae genome (release 45) the annotation of this gene was altered, and it is now thought that these 858 bp represent about half of the coding sequence of LRIM1. We obtained a total of 138 alleles from six species, of which 108 were unique (Table 1, genbank accession nos EU073528–EU073597). As in the other two genes investigated here, polymorphisms were shared between species, but alleles were not. Dxy ranged from 1.03 to 3.06 (per 100 bp) between species. In contrast to CEC1 and GNBP-B1 however, McDonald-Kreitman tests of positive selection indicated a significant excess of fixed non-synonymous differences between An. arabiensis and An. quadriannulatus A, An. merus as well as An. bwambae (Table 4). The fact that all three comparisons involve An. arabiensis, suggests this lineage underwent more non-synonymous substitutions than expected under a neutral model.
Surprisingly no fixed differences were present between An. gambiae and three other species; An. arabiensis, An. bwambae and An. quadriannulatus A. Our phylogenetic analysis showed that alleles from An. gambiae did not form a single cluster but were interspersed across the tree (Figure S1, supporting information). In particular, several “arabiensis-like” An. gambiae alleles clustered with An. arabiensis, far removed from the majority of An. gambiae sequences. When An. gambiae was removed from the analysis (Figure 1), An. arabiensis, An. melas and An. merus formed monophyletic groups with posterior probabilities of 0.99 and higher. An. bwambae formed a paraphyletic group containing the monophyletic An. quadriannulatus A. Since the An. gambiae alleles did not form a single cluster, it was not possible to test for positive selection along a single branch leading to this species. However, we did test for positive selection along the branch leading to the other major malaria vector, An. arabiensis. To increase the power of this analysis, the length of this branch, henceforth referred to as the foreground branch, was maximized by excluding all An. gambiae alleles from the analysis.
Posterior probabilities ≥0.99 are indicated by *. Number of non-synonymous/synonymous substitutions are indicated above or on the left side of the branches. Estimated ω values are placed below or on the right side of the branches and are underlined. The foreground branch for the maximum likelihood tests of positive selection is indicated by a double line. For a more detailed phylogeny, including all posterior probabilities above 50% and sample names, see Figure S2 (supporting materials).
Due to the absence of synonymous substitutions the ω value, i.e. the ratio of non-synonymous to synonymous substitutions or dN/dS, along the foreground branch could not be estimated in our PAML analyses and is indicated as ∞ (Table 5). The branch test did not indicate that ω along the foreground branch is significantly larger than 1 (model 2 free ω vs. model 2 ω = 1). However, a branch test did show that it is significantly larger than ω along the background branches (p = 0.011, model 2 free ω vs. model 0). Values of ω along the branches leading to An. melas, An. merus and An. bwambae/An. quadriannulatus A were estimated to be 0.299, 0.298 and 0.147 respectively (Figure 1), indicating purifying selection along these lineages. This confirms that the excess of non-synonymous substitutions detected by the McDonald-Kreitman tests mainly occurred along the lineage leading to An. arabiensis. Interestingly, the branch separating An. arabiensis and An. melas from the other species also lacked synonymous, but not non-synonymous substitutions, possibly indicating some positive selection along this branch.
The branch tests applied above use an ω value averaged across the entire gene. This severely diminishes the power of the analysis because at least some parts of each gene are expected to be under purifying selection. A more powerful test for positive selection is provided by branch-site models, which test if certain codons are under selection in the foreground branch . When implemented on our data set (model A vs. model A1), the branch-site test provided strong support for positive selection in the An. arabiensis lineage with p = 0.005 (Table 5). According to the Bayes Empirical Bayes (BEB) analysis , amino-acid positions 108 and 236 are under positive selection (i.e. ω>1), with probabilities of 0.983 and 0.985, respectively. Position 235 has a high probability (0.933) of being under positive selection as well.
A total of four non-synonymous substitutions occurred along the foreground branch (Figure 1). Three of these non-synonymous substitutions are within two adjacent codons (235 and 236), with codon 236 having two replacement substitutions. That is, regardless of the order in which the two nucleotide substitutions in codon 236 occurred, two subsequent amino-acid changes were the result. Additionally, a replacement substitution at position 234 is fixed in An. melas, is at high frequency in An. arabiensis (0.83) and is also found in one of the “arabiensis-like” An. gambiae alleles.
An. arabiensis is fixed for nucleotide A at sites 704, 706 and 707 (codons 235 and 236). With the exception of An. gambiae, all other species are fixed for nucleotides T, G and G at these respective positions. The few “arabiensis-like” An. gambiae alleles also have the (AAA) arrangement. Positions 416 through 718, a 302 bp stretch, cluster these An. gambiae (AAA) alleles with An. arabiensis (Figure 2). Only positions below 328 and above 767 contain polymorphisms that group some of the An. gambiae (AAA) alleles with the rest of An. gambiae. Additionally, no fixed differences were found between An. gambiae and An. arabiensis anywhere in this gene.
Only sites beyond position 324 that favor the clustering of An. gambiae (AAA) alleles with either An. gambiae (TGG) or An. arabiensis are included.
LRIM1 is located inside the 2La inversion. Therefore we determined the karyotype of our An. gambiae samples with respect to the 2La arrangement. Out of 32 specimens from Cameroon, three were 2La/+ heterozygotes, all of which were also heterozygous for the (AAA)/(TGG) alleles. The karyotype of one (AAA)/(TGG) heterozygote was not clear, as it produced a second band of unexpected size. The remaining four (AAA)/(TGG) heterozygotes, as well as all (TGG) homozygotes, carried the 2L+/+ karyotype. These findings confirm that the (AAA) allele is at very high frequency or even fixed in the 2La inversion in Cameroon, and is present at very low frequency in the 2L+ arrangement (≈7%).
To examine if LRIM1 in An. arabiensis showed signs of a recent selective sweep, a HKA test was performed by comparing the polymorphism/fixed differences ratio of LRIM1 to CEC1 and GNBP-B1. A selective sweep reduces the amount of standing genetic variation within a species, as indicated by a relatively low ratio. However, no significant differences were found between the genes, and in fact, this ratio was considerably higher for LRIM1 (20/20.2) as compared to CEC1 and GNBP-B1 (17/28.1).
CEC1 and GNBP-B1 did not show any signs of positive selection, and in particular, showed little or no differentiation between malaria vectors and the non-vector species, indicating that these genes are largely subject to purifying selection. In two of the species comparisons GNBP-B1 showed a significant excess of non-synonymous polymorphisms. Some cloning error is expected to be present in the GNBP-B1 data set. Since a majority of possible mutations are non-synonymous, random errors will bias the observed number of non-synonymous polymorphisms upward. However, the number of PCR errors in the data is not nearly high enough to explain the difference. Therefore, most likely purifying selection is responsible, with numerous slightly deleterious substitutions present at low frequency in populations, but which are prevented from going to fixation. In contrast, LRIM1, a gene with no known homologue in other organisms, shows clear signs of positive selection in An. arabiensis.
As pointed out by MacDonald and Kreitman, an excess of non-synonymous fixed differences between species may result from a much smaller population size in the past . This would have allowed slightly deleterious mutations to go to fixation by drift, whereas in the current larger population most of these are removed by purifying selection. Evidence for population expansion has been reported for both An. gambiae and An. arabiensis , . It is unlikely however that this could explain the excess of non-synonymous substitutions observed in LRIM1 in An. arabiensis. First of all, a demographic explanation should affect all genes. As noted before, GNBP-B1 has a higher ratio of non-synonymous/synonymous polymorphisms than fixed differences in most or all populations, including An. arabiensis. This indicates the presence of a relatively large number of slightly deleterious alleles in this gene, few or none of which became fixed in ancestral populations. This is contrary to the demographic explanation. More importantly however, it is extremely unlikely that three out of four fixed amino acid changes would occur in two adjacent positions in a 285 amino acid protein, if the random process of genetic drift were responsible.
Polymorphisms are shared between species in all three genes we examined, and there is no doubt some of this is due to the retention of ancestral polymorphism. However, the pattern of polymorphism observed in LRIM1 provides strong evidence that the presence of “arabiensis-like” alleles in An. gambiae is caused by introgression and not by the retention of ancestral polymorphism. It is unlikely that recombination would not have broken down a linkage group of at least 352 bp, if it were maintained in the population for long time. Furthermore, the introgression hypothesis is supported by the complete absence of fixed differences between these species anywhere in the LRIM1 gene.
Based on the shared polymorphisms between An. gambiae (AAA) and An. arabiensis (positions 416, 624 and 701), introgression has occurred multiple times, after which, according to the shared polymorphism between the An. gambiae (AAA) and (TGG) alleles (positions 324 and 327), these alleles recombined between position 327 and 416. The introgression of LRIM1 from An. arabiensis into An. gambiae is also consistent with previous studies that have shown that introgression between these two species has occurred in the past , . Additionally, it has been shown through sequence analyses , as well as crossing experiments , that different chromosomes vary in their capacity for horizontal transfer between these two species. The 2nd chromosome, on which LRIM1 is located, has been shown to transfer most readily between An. gambiae and An. arabiensis. .
Mosquitoes, like other organisms, encounter numerous pathogens during their life cycle, all of which could potentially exert selection pressure on the immune system. In fact, molecules that play a role in host-parasite interactions are one of the main groups of proteins on which positive selection has been demonstrated . No direct evidence is available to show that LRIM1 in An. arabiensis has evolved in direct response to malaria infection, but our observation that this anti-malarial gene shows distinct signs of positive selection in a primary malaria vector, but not in a species that does not transmit Plasmodium, is intriguing. This suggests that the observed adaptive evolution of LRIM1 may have been the result of the infection of An. arabiensis with Plasmodium, with at least two adjacent amino acids playing a crucial role in this adaptation. If this is true, variation for Plasmodium resistance should be present at LRIM1 in An. gambiae.
LRIM1 contains a leucine-rich repeat (LRR), which in other LRR genes is crucial for the three-dimensional structure by folding it into an arc . Little is known about the structure of LRIM1, but the two neighboring adaptive amino acids (i.e. 235 and 236) are located well outside the leucine-rich repeat region (positions 30–160), suggesting that they could have a more specific function.
Although LRIM1 is known to play a role in suppressing P. berghei infection in An. gambiae , a recent RNAi study failed to show an effect of LRIM1 on P. falciparum infection in field-collected An. gambiae , while an effect on P. berghei infection was confirmed. This could be because the action of LRIM1 in An. gambiae is specific against P. berghei. However, it is also possible that only some LRIM1 alleles suppress infection with P. falciparum, and these may even be specific for certain P. falciparum strains. Another study demonstrated the existence of such genotype by genotype interactions between P. falciparum and An. gambiae, by showing that no single strain of P. falciparum was best at infecting all of a set of iso-female An.gambiae lines . Additionally, LRIM1 is located inside the 2La inversion. While An. arabiensis is fixed for this 2La arrangement, An. gambiae is 2La/+ polymorphic. Since LRIM1 alleles introgressed from An. arabiensis into An. gambiae, we may expect that these An. gambiae (AAA) alleles are mostly found in the 2La arrangement. Interestingly, the mosquitoes that failed to show an effect of LRIM1 knockdown on P. falciparum infection  all carried the standard chromosome arrangement (2L+). Our molecular karyotyping of the 2La inversion in our An. gambiae specimens shows that the (AAA) allele is indeed found at very high frequency in 2La inversions, whereas it is present in very low frequency in 2L+ (≈7%).
Pathogen-host co-evolution has mainly been considered in terms of an evolutionary arms race . Under this model, the host continuously evolves to limit infection with the pathogen, which in turns evolves to evade host defenses. This is expected to lead to repeated selective sweeps, which leave a signature in the selected genes in the form of a low level of standing genetic variation. A comparison of the polymorphism to divergence ratio in LRIM1 vs GNBP-B1/CEC1 did not show a relatively low level of genetic variation in LRIM1 in An. arabiensis. In fact, the relative level of polymorphism was higher in this gene than in CEC1 and GNBP-B1. Therefore we have no indication that LRIM1 in An. arabiensis is currently involved in an evolutionary arms race. This also implies that possible selective sweeps occurred long enough ago to allow mutation to regenerate polymorphism.
The data presented here indicate that the anti-malarial gene LRIM1 has undergone adaptive evolution in a primary malaria vector. This could be because this gene has evolved in response to P. falciparum infection in this species. If so, LRIM1 is expected to play a role in the resistance of An. arabiensis against P. falciparum. So far the immune system of this mosquito species has not yet been investigated, and our data suggest the possibility that a knockdown of LRIM1 will enhance infections of P. falciparum in An. arabiensis. If LRIM1 did indeed evolve in response to P. falciparum infection in An. arabiensis, this gene also deserves further study in An. gambiae, in particular with respect to potential variation in resistance between the two major alleles found in this species.
Materials and Methods
Adult females of An. gambiae were collected from the villages of Mbebé and Nyabessan, Cameroon in Dec. 2005. An. gambiae from Mali were collected from Banambani in 2000. Adult An. arabiensis females from Cameroon were collected from Kousseri in Dec 2005. Adult An. melas were collected in Ipono, Cameroon, Dec. 2005. Larvae of An. gambiae, An. arabiensis and An. bwambae from Bwamba county, Uganda (2004) were kindly provided by Ralph Harbach. DNA extractions of An. merus from Furvela, Mozambique (2001 and 2003) were kindly provided by David O'Brochta. An. quadriannulatus A from Kruger National Park, South Africa, were kindly provided by Anton Cornel. Sample sizes for each gene and species are represented in Table 1.
DNA was extracted using the DNeasy tissue kit (Qiagen). Species and molecular form diagnostics were performed following Fanello et al.  and Besansky et al. . All An. gambiae specimens belonged to the S molecular form. Molecular identification of 2La karyotypes was performed following White et al. . Primers to amplify CEC1, GNBP-B1 and LRIM1 were designed using Primer3  based on the An. gambiae genome and anneal to the flanking or non-transcribed regions of the genes . PCR of CEC-A and LRIM1 was performed using Amplitaq Gold polymerase (Perkin Elmer) using respectively the following primer pairs CECin1 (GTTAGCAGAGCCGTCGTCTT)/CECin12 (ACAGTCGGTTCAAAGCGTTC) and LRIM1in6 (AGGTAACGGACAGCAGCCTA)/LRIM1in9 (GTCCGGTACTGCTCCTTGAG). The following program was used for PCR amplification of CEC1 and LRIM1; 2 min at 94°, 35 cycles of 30 sec at 94°, 30 sec at 52° and 45 sec at 72°, followed by 20 min at 72°. PCR products were excised from an agarose gel and purified using the Gel Purification Kit (Qiagen) and submitted for direct sequencing. A subset of the sequences from individuals heterozygous for two or more positions were amplified again and cloned using the TOPO-TA cloning kit (Invitrogen). Individuals were selected for cloning such that all observed polymorphic sites were represented in the final data set. From each individual a single colony was sequenced. PCR of GNBP-B1 was performed using Platinum High Fidelity Taq (Invitrogen) with the primer pair GNBPin1 (GTTTGGTAGGGGACGAATGA) /GNBPIN20 (GCGCTTTCAGTGGTTTGTTT) using the following program: 2 min at 94°, 35 cycles of 30 sec at 94°, 30 sec at 52° and 90 sec at 72°, followed by 20 min at 72°. Direct sequencing of the PCR product of GNBP-B1 was not possible in many cases because of the presence of indels. Therefore, PCR products of this gene were cloned and sequenced as outlined above. However, nine sequences were produced through direct sequencing, allowing for an estimation of the PCR/cloning error. Sequencing was performed on an ABI 3730 Genetic Analyzer using Big Dye v 3.1 (Applied Biosystems )
Based on a comparison between direct sequencing and plasmid sequencing, the PCR/cloning error using Amplitaq Gold was estimated to be approximately 1.5 per 1000 bp. However, because all LRIM1 and CEC1 samples that were cloned were also sequenced directly, we were able to derive both alleles from each individual while removing PCR/cloning errors. The PCR error in the GNBP-B1 sequences amplified using the proof-reading polymerase was estimated to be 0.625 per 1000 bp. Therefore, each 1188 bp GNBP-B1 allele for which no direct sequence was available is expected to have an average of 0.74 errors.
All sequences were aligned using MEGA3.1  and alignments were improved manually. Introns were included in the phylogeny reconstructions. For all other analyses the coding region was used. Dxy values were calculated using DnaSP 4.0 . This software was also used to perform McDonald-Kreitman tests , using Fisher's exact test. The McDonald-Kreitman test compares the dN/dS ratio between species to within species and is based on the idea that substitutions under positive selection will go to fixation rapidly, and are therefore rarely observed as polymorphisms. However, they are present as fixed differences between species and an excess of replacement fixed differences is therefore an indication of positive selection.
Since few or no fixed differences were observed in CEC1 and GNBP-B1, subsequent analyses were limited to LRIM1. Aimed at reducing the computational effort, a reduced LRIM1 data set, containing 70 sequences, was used for phylogenetic analyses and maximum likelihood tests of positive selection. This data set was compiled in such a way that at every observed polymorphism and fixed difference, i.e. the relevant information for tests for maximum likelihood tests of positive selection, was retained. This reduced data set was used to construct 50% majority-rule consensus trees with MrBayes 3.1.2 . Modeltest 3.7  was used to determine the most appropriate nucleotide substitution model for our data set.
Several LRIM1 alleles from An. gambiae clustered within An. arabiensis (Figure S1 supp. mat.). Therefore, phylogeny reconstruction was also performed excluding An. gambiae sequences. This inferred tree was used for maximum likelihood tests of positive selection along the branch leading to An. arabiensis in PAML3.15 and to estimate ω (i.e. dN/dS) along the major branches of the tree. Under the neutral model the relative number of synonymous and non-synonymous substitutions is expected to be 1. Under positive selection, amino acid substitutions are favored and ω>1, whereas under purifying selection amino acid substitutions are prevented and ω<1. The An. arabiensis lineage was designated as the foreground branch, i.e. the branch of interest, and model 2 free ω was compared to model 0 to test if ω along the foreground branch was significantly larger compared to the ω along the background branches, i.e. all other branches. Model 2 ω = 1, with the ω value fixed at 1 along foreground branch, was compared to model 2 free ω to test if ω along the foreground branch was significantly larger than 1. Model 1 was used to estimate ω along the central branches of the tree (Figure 1) and to infer the number of substitutions along these branches of the phylogeny. As an additional test for positive selection along the foreground branch, we used the more powerful branch-site test 2 by comparing model A and Model A1 . Bayes Empirical Bayes (BEB) analysis was used to identify positively selected codons in the foreground branch .
To test for a reduction in the polymorphism of LRIM1 in An. arabiensis, an HKA test was performed in DnaSP 4.0, using eight An. arabiensis alleles for CEC1 and LRIM1, as well as all eight GNBP-B1 alleles from this species. The CEC1 and LRIM1 alleles were from the same individuals as the GNBP-B1 sequences if possible, otherwise were randomly chosen from the same population. Seven An. quadriannulatus A alleles from each gene were used to calculate inter-specific divergence.
Bayesian tree (unrooted) of LRIM1 from six species of the An.gambiae complex. Posterior probabilities are indicated along branches. An. gambiae samples and An. arabiensis samples from Uganda, Madagascar and Mali are indicated by UG, MAD, and MAL respectively, remaining samples of these two species are from Cameroon.
(1.19 MB TIF)
Bayesian tree (unrooted) of LRIM1 in five species of the An.gambiae complex. Posterior probabilities are indicated along branches. An. arabiensis samples from Uganda, Madagascar and Mali are indicated by UG, MAD, and MAL respectively, with all remaining An. arabiensis samples originating from Cameroon.
(0.72 MB TIF)
We are very grateful to Anton Cornel, Ralph Harbach and David O'Brochta for providing An. quadriannulatus A, An. bwambae and An. merus specimens, respectively. Additionally, we are grateful to the anonymous reviewers who provided many useful comments for improving earlier versions of the manuscript.
Conceived and designed the experiments: JP GC MS. Performed the experiments: MS. Analyzed the data: MS. Contributed reagents/materials/analysis tools: MS AP JM PA CN FS. Wrote the paper: MS.
- 1. Breman JC, Egan A, Teusch GT (2001) The intolerable burden of malaria: A new look at the numbers. Am J Trop Med Hyg 64s: iv–vii.JC BremanA. EganGT Teusch2001The intolerable burden of malaria: A new look at the numbers.Am J Trop Med Hyg64sivvii
- 2. Holt R, Subramanian GM, Halpern A, Sutton G, Charlab R, et al. (2002) The genome sequence of the malaria mosquito Anopheles gambiae. Science 298: 129–148.R. HoltGM SubramanianA. HalpernG. SuttonR. Charlab2002The genome sequence of the malaria mosquito Anopheles gambiae.Science298129148
- 3. Riehle MM, Markianos K, Niaré O, Xu J, Li J, et al. (2006) Natural malaria infection in Anopheles gambiae is regulated by a single genomic control region. Science 312: 577–579.MM RiehleK. MarkianosO. NiaréJ. XuJ. Li2006Natural malaria infection in Anopheles gambiae is regulated by a single genomic control region.Science312577579
- 4. Kim W, Koo H, Richman AM, Seeley D, Vizoli J, et al. (2004) Ectopic expression of a cecropin transgene in the human malaria vector mosquito Anopheles gambiae (Diptera: Culicidae): Effects on susceptibility to Plasmodium. J Med Entomol 41: 447–455.W. KimH. KooAM RichmanD. SeeleyJ. Vizoli2004Ectopic expression of a cecropin transgene in the human malaria vector mosquito Anopheles gambiae (Diptera: Culicidae): Effects on susceptibility to Plasmodium.J Med Entomol41447455
- 5. Tahar R, Boudin C, Thiery I, Bourgouin C (2002) Immune response of Anopheles gambiae to the early sporogonic stages of the human malaria parasite Plasmodium falciparum. The EMBO journal 21: 6673–6680.R. TaharC. BoudinI. ThieryC. Bourgouin2002Immune response of Anopheles gambiae to the early sporogonic stages of the human malaria parasite Plasmodium falciparum.The EMBO journal2166736680
- 6. Osta MA, Christophides GK, Kafatos FC (2004) Effects of mosquito genes on Plasmodium development. Science 303: 2030–2032.MA OstaGK ChristophidesFC Kafatos2004Effects of mosquito genes on Plasmodium development.Science30320302032
- 7. Dong Y, Aguilar R, Xi Z, Warr E, Mongin E, et al. (2006) Anopheles gambiae immune responses to human and rodent Plasmodium parasite species. PLoS Pathol 2: 513–525.Y. DongR. AguilarZ. XiE. WarrE. Mongin2006Anopheles gambiae immune responses to human and rodent Plasmodium parasite species.PLoS Pathol2513525
- 8. Vlachou D, Schlegelmilch T, Christophides GK, Kafatos FC (2005) Functional genomic analysis of midgut epithelial responses in Anopheles during Plasmodium invasion. Current Biol 15: 1185–1195.D. VlachouT. SchlegelmilchGK ChristophidesFC Kafatos2005Functional genomic analysis of midgut epithelial responses in Anopheles during Plasmodium invasion.Current Biol1511851195
- 9. Little TJ, Cobbe N (2005) The evolution of immune-related genes from disease carrying mosquitoes: diversity in a peptidoglycan- and a thioester-recognizing protein. Insect Mol Biol 14: 599–605.TJ LittleN. Cobbe2005The evolution of immune-related genes from disease carrying mosquitoes: diversity in a peptidoglycan- and a thioester-recognizing protein.Insect Mol Biol14599605
- 10. Simard F, Licht M, Besansky NJ, Lehmann T (2006) Polymorphism at the defensin gene in the Anopheles gambiae complex: testing different selection hypotheses. Infect Genet Evol 7: 285–292.F. SimardM. LichtNJ BesanskyT. Lehmann2006Polymorphism at the defensin gene in the Anopheles gambiae complex: testing different selection hypotheses.Infect Genet Evol7285292
- 11. Yang Z (2006) Computational Molecular Evolution. New York: Oxford University Press. Z. Yang2006Computational Molecular EvolutionNew YorkOxford University Press357
- 12. Schlenke TA, Begun DJ (2003) Natural selection drives Drosophila immune system evolution. Genetics 164: 1471–1480.TA SchlenkeDJ Begun2003Natural selection drives Drosophila immune system evolution.Genetics16414711480
- 13. Collins FH, Zavala F, Graves PM, Cochrane AH, Gwadz RW, et al. (1984) First field trial of an immunoradiometric assay for the detection of malaria sporozoites in mosquitoes. Am J Trop Med Hyg 33: 538–543.FH CollinsF. ZavalaPM GravesAH CochraneRW Gwadz1984First field trial of an immunoradiometric assay for the detection of malaria sporozoites in mosquitoes.Am J Trop Med Hyg33538543
- 14. Ferguson HM, Read AF (2002) Why is the effect of malaria parasites on mosquito survival still unresolved? Trends Parasitol 18: 256–261.HM FergusonAF Read2002Why is the effect of malaria parasites on mosquito survival still unresolved?Trends Parasitol18256261
- 15. Hay SI, Rogers DJ, Toomer JF, Snow RW (2000) Annual Plasmodium falciparum entomological inoculation rates (EIR) across Africa: literature survey, Internet access and review. Trans R Soc Trop Med Hyg 94: 113–127.SI HayDJ RogersJF ToomerRW Snow2000Annual Plasmodium falciparum entomological inoculation rates (EIR) across Africa: literature survey, Internet access and review.Trans R Soc Trop Med Hyg94113127
- 16. Vizoli J, Bulet P, Charlet M, Lowenberger C, Blass C, et al. (2000) Cloning and analysis of a cecropin gene from the malaria vector mosquito, Anopheles gambiae. Insect Mol Biol 9: 75–84.J. VizoliP. BuletM. CharletC. LowenbergerC. Blass2000Cloning and analysis of a cecropin gene from the malaria vector mosquito, Anopheles gambiae.Insect Mol Biol97584
- 17. Dimopoulos G, Richman A, Müller H-M, Kafatos FC (1997) Molecular immune response of the mosquito Anopheles gambiae to bacteria and malaria parasites. Proc Natl Acad Sci U S A 94: 11508–11513.G. DimopoulosA. RichmanH-M MüllerFC Kafatos1997Molecular immune response of the mosquito Anopheles gambiae to bacteria and malaria parasites.Proc Natl Acad Sci U S A941150811513
- 18. Warr E, Lambrechts L, Koella JC, Bourgouin C, Dimopoulos G (2006) Anopheles gambiae immune response to Sephadex beads: involvement of anti-Plasmodium factors in regulating melanization. Insect Biochem Mol Biol 36: 769–778.E. WarrL. LambrechtsJC KoellaC. BourgouinG. Dimopoulos2006Anopheles gambiae immune response to Sephadex beads: involvement of anti-Plasmodium factors in regulating melanization.Insect Biochem Mol Biol36769778
- 19. Zhang J, Nielsen R, Yang Z (2005) Evaluation of an improved branch-site likelihood method for detecting positive selection at the molecular level. Mol Biol Evol 22: 2472–2479.J. ZhangR. NielsenZ. Yang2005Evaluation of an improved branch-site likelihood method for detecting positive selection at the molecular level.Mol Biol Evol2224722479
- 20. Yang Z, Wong WSW, Nielsen R (2005) Bayes empirical Bayes inference of amino acid sites under positive selection. Mol Biol Evol 22: 1107–1118.Z. YangWSW WongR. Nielsen2005Bayes empirical Bayes inference of amino acid sites under positive selection.Mol Biol Evol2211071118
- 21. MacDonald JH, Kreitman M (1991) Adaptive protein evolution at the Adh locus in Drosophila. Nature 351: 652–654.JH MacDonaldM. Kreitman1991Adaptive protein evolution at the Adh locus in Drosophila.Nature351652654
- 22. Donelly MJ, Licht MC, Lehmann T (2001) Evidence for recent population expansion in the evolutionary history of the malaria vectors Anopheles arabiensis and Anopheles gambiae. Mol Biol Evol 18: 1353–1364.MJ DonellyMC LichtT. Lehmann2001Evidence for recent population expansion in the evolutionary history of the malaria vectors Anopheles arabiensis and Anopheles gambiae.Mol Biol Evol1813531364
- 23. Onyabe DY, Conn JE (2001) Population structure of the malaria mosquito Anopheles arabiensis across Nigeria suggests range expansion. Mol Ecol 10: 2577–2591.DY OnyabeJE Conn2001Population structure of the malaria mosquito Anopheles arabiensis across Nigeria suggests range expansion.Mol Ecol1025772591
- 24. Besansky NJ, Krzywinski J, Lehman T, Simard F, Kern M, et al. (2003) Semipermeable species boundaries between Anopheles gambiae and Anopheles arabiensis: Evidence from multilocus DNA sequence variation. Proc Natl Acad Sci U S A 19: 10818–10823.NJ BesanskyJ. KrzywinskiT. LehmanF. SimardM. Kern2003Semipermeable species boundaries between Anopheles gambiae and Anopheles arabiensis: Evidence from multilocus DNA sequence variation.Proc Natl Acad Sci U S A191081810823
- 25. Besansky NJ, Powell JR, Caccone A, Hamm DM, Scott JA, et al. (1994) Molecular phylogeny of the Anopheles gambiae complex suggests genetic introgression between principle malaria vectors. Proc Natl Acad Sci U S A 91: 6885–6888.NJ BesanskyJR PowellA. CacconeDM HammJA Scott1994Molecular phylogeny of the Anopheles gambiae complex suggests genetic introgression between principle malaria vectors.Proc Natl Acad Sci U S A9168856888
- 26. Slotman MA, della Torre A, Calzetta M, Powell JR (2005) Differential introgression of chromosomal regions between Anopheles gambiae and An. arabiensis. Am J Trop Med Hyg 73: 326–335.MA SlotmanA. della TorreM. CalzettaJR Powell2005Differential introgression of chromosomal regions between Anopheles gambiae and An. arabiensis.Am J Trop Med Hyg73326335
- 27. Enkhbayar P, Kamiya M, Osaki M, Matsumoto T, Matsushima N (2004) Structural principles of Leucine-Rich Repeat (LRR) Proteins. Proteins 54: 394–403.P. EnkhbayarM. KamiyaM. OsakiT. MatsumotoN. Matsushima2004Structural principles of Leucine-Rich Repeat (LRR) Proteins.Proteins54394403
- 28. Cohuet A, Osta M, Morlais I, Awono-Ambene PH, Michel K, et al. (2006) Anopheles and Plasmodium: from laboratory models to natural systems in the field. EMBO Reports 7: 1285–1289.A. CohuetM. OstaI. MorlaisPH Awono-AmbeneK. Michel2006Anopheles and Plasmodium: from laboratory models to natural systems in the field.EMBO Reports712851289
- 29. Lambrechts L, Halbert J, Durand P, Gouagna LC, Koella JC (2005) Host genotype by parasite genotype interactions underlying the resistance of anopheline mosquitoes to Plasmodium falciparum. Malar J 4: 3.L. LambrechtsJ. HalbertP. DurandLC GouagnaJC Koella2005Host genotype by parasite genotype interactions underlying the resistance of anopheline mosquitoes to Plasmodium falciparum.Malar J43
- 30. Dawkins R, Krebs JR (1979) Arms races between and within species. Proc R Soc Lond B Biol Sci 205: 489–511.R. DawkinsJR Krebs1979Arms races between and within species.Proc R Soc Lond B Biol Sci205489511
- 31. Fanello C, Santolamazza F, della Torre A (2002) Simultaneous identification of species and molecular forms of the Anopheles gambiae comples by PCR-RFLP. Med Vet Entomol 16: 461–464.C. FanelloF. SantolamazzaA. della Torre2002Simultaneous identification of species and molecular forms of the Anopheles gambiae comples by PCR-RFLP.Med Vet Entomol16461464
- 32. Besansky NJ, Collins FH, Towson H (2006) A species-specific PCR for the identification of the malaria vector Anopheles bambae. Ann Trop Med Parasitol 100: 277–280.NJ BesanskyFH CollinsH. Towson2006A species-specific PCR for the identification of the malaria vector Anopheles bambae.Ann Trop Med Parasitol100277280
- 33. White BJ, Santolamazza F, Kamau L, Pombi M, Grushko O, et al. (2007) Molecular karyotyping of the 2La inversion in Anopheles gambiae. Am J Trop Med Hyg 76: 334–339.BJ WhiteF. SantolamazzaL. KamauM. PombiO. Grushko2007Molecular karyotyping of the 2La inversion in Anopheles gambiae.Am J Trop Med Hyg76334339
- 34. Rozen S, Skaletsky HJ (2000) Primer3 on the WWW for general users and for biologist programmers. In: Krawetz S, Misener S, editors. Bioinformatics Methods and Protocols: Methods in Molecular Biology. Totowa: Humana Press. pp. 365–38.S. RozenHJ Skaletsky2000Primer3 on the WWW for general users and for biologist programmers.S. KrawetzS. MisenerBioinformatics Methods and Protocols: Methods in Molecular BiologyTotowaHumana Press36538
- 35. Kumar S, Tamura K, Nei M (2004) MEGA3: integrated software for molecular evolutionary genetics analysis and sequence alignment. Briefings in Bioinformatics 5: 150–163.S. KumarK. TamuraM. Nei2004MEGA3: integrated software for molecular evolutionary genetics analysis and sequence alignment.Briefings in Bioinformatics5150163
- 36. Rozas J, Sánchez-DelBarrio JC, Messeguer X, Rozas R (2003) DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 19: 2496–2497.J. RozasJC Sánchez-DelBarrioX. MesseguerR. Rozas2003DnaSP, DNA polymorphism analyses by the coalescent and other methods.Bioinformatics1924962497
- 37. Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574.F. RonquistJP Huelsenbeck2003MrBayes 3: Bayesian phylogenetic inference under mixed models.Bioinformatics1915721574
- 38. Posada D, Crandall KA (1998) Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817–818.D. PosadaKA Crandall1998Modeltest: testing the model of DNA substitution.Bioinformatics14817818