Evolutionary history of a mosquito endosymbiont revealed through mitochondrial hitchhiking

Jason L Rasgon, Anthony J Cornel, Thomas W Scott


Due to cytoplasmic inheritance, spread of maternally inherited Wolbachia symbionts can result in reduction of mitochondrial variation in populations. We examined sequence diversity of the mitochondrial NADH dehydrogenase subunit 4 (ND4) gene in Wolbachia-infected (South Africa (SA), California and Thailand) and uninfected (SA) Culex pipiens complex populations. In total, we identified 12 haplotypes (A–L). In infected populations, 99% of individuals had haplotype K. In the uninfected SA population, 11 haplotypes were present, including K. Nuclear allozyme diversity was similar between infected and uninfected SA populations. Analysis of nuclear DNA sequences suggested that haplotype K presence in uninfected SA Cx. pipiens was probably due to a shared ancestral polymorphism rather than hybrid introgression. These data indicate that Wolbachia spread has resulted in drastic reduction of mitochondrial variability in widely separated Cx. pipiens complex populations. In contrast, the uninfected SA population is probably a cryptic species where Wolbachia introgression has been prevented by reproductive isolation, maintaining ancestral levels of mitochondrial diversity. Molecular clock analyses suggest that the Wolbachia sweep occurred within the last 47 000 years. The effect of Wolbachia on mitochondrial dynamics can provide insight on the potential for Wolbachia to spread transgenes into mosquito populations to control vector-borne diseases.


1. Introduction

Understanding the structure of mosquito populations is critical for addressing public health issues such as evolution and spread of insecticide resistance alleles, epidemiology of mosquito-borne pathogens and developing and testing transgenic strategies for control of vector-borne diseases (Tripet et al. 2001; Bennett et al. 2002; Fanello et al. 2003). Mitochondrial DNA sequences have been shown to be useful markers for studying the structure of medically important insect populations such as mosquitoes, tsetse flies (Glossina morsitans) and sand flies (Phlebotomus papatasi; Besansky et al. 1997; Gorrochotegui-Escalante et al. 2000; Krafsur et al. 2000; Donnell et al. 2001; Krafsur et al. 2001; Gorrochotegui-Escalante et al. 2002; Parvizi et al. 2003; Marquez et al. 2004). Patterns of mitochondrial variability can be confounded, however, by the spread of maternally inherited symbionts such as Wolbachia (Turelli & Hoffmann 1999; Hurst & Jiggins 2005). Wolbachia is associated with reproductive alterations such as cytoplasmic incompatibility (CI); i.e. reduced egg hatch when uninfected females mate with infected males. Matings between infected females and infected or uninfected males are fertile. Consequently, infected females have a reproductive advantage in a mixed population, allowing infection to spread (Turelli & Hoffmann 1999). As Wolbachia spreads, there can be a linked sweep of the associated mitochondria, resulting in reduction of mitochondrial diversity in the infected populations (Turelli et al. 1992, 1999; Hurst & Jiggins 2005). Selective mitochondrial sweeps have been observed in concert with Wolbachia invasions in a number of species (Turelli et al. 1992; Ballard et al. 1996; Baurdy et al. 2003; Jiggins 2003; Parvizi et al. 2003).

Among mosquitoes, the population biology of Wolbachia has been most thoroughly studied in the Culex pipiens species complex (Cornel et al. 2003; Rasgon & Scott 2003; 2004). The two most common and widespread members of the complex are the subspecies Culex pipiens pipiens (L.) and Culex pipiens quinquefasciatus Say (Barr 1982). Both subspecies have a global distribution and complicated population structure. For example, in North America and Asia extensive gene flow occurs between subspecies (Tabachnick & Powell 1983; Urbanelli et al. 1997; Fonseca et al. 2004), whereas in some parts of Africa gene flow between subspecies is partial (Urbanelli et al. 1985, 1995) or restricted (Jupp 1978; Cornel et al. 2003). Depending on location, members of this complex can be important vectors of nematodes that cause lymphatic filariasis and arboviruses such as St Louis encephalitis, West Nile and Rift Valley fever viruses (Hoogstraal et al. 1979; Krida et al. 1998; Day 2001; Nasci et al. 2001; Fonseca et al. 2004).

Mosquitoes in Cx. pipiens complex are commonly infected with Wolbachia (Rasgon & Scott 2003, 2004), and have been previously shown to exhibit reduced levels of mitochondrial variability among colonized strains (Guillemaud et al. 1997) and in naturally bottlenecked populations (Fonseca et al. 2000). With few exceptions all populations are believed to be infected with Wolbachia (Hoffmann & Turelli 1997). The published records of uninfected populations occur in Culex pipiens australicus Dobrotwortsky and Drummond from Australia and Cx. p. pipiens from Rhodesia (Zimbabwe; Irving-Bell 1974). We recently described a Wolbachia-uninfected Cx. p. pipiens population in South Africa (SA) near Johannesburg that was reproductively isolated from sympatric Cx. p. quinquefasciatus infected populations (Cornel et al. 2003). Reproductive isolation between these two populations was inferred through morphological assessment of male genitalia DV/D ratios (Sundararaman 1949), fixed allozyme differences and Wolbachia infection status. If reproductive isolation of this uninfected population from other infected populations predates the sweep of Wolbachia through Cx. pipiens sensu latu, then it can be used as a baseline to directly test the influence of Wolbachia spread on mitochondrial variability in the Cx. pipiens species complex and perhaps to predict the ramification of this process in insects in general. To examine this issue, we (i) compared mitochondrial sequence diversity from the Wolbachia-uninfected SA population to infected populations from SA, California and Thailand, (ii) assessed Wolbachia frequency and Wolbachia surface protein (wsp) gene diversity in each infected population and (iii) assessed levels of nuclear diversity in infected and uninfected SA populations by sequencing nuclear DNA and by allozymes. The data suggest that Wolbachia has recently swept through Cx. pipiens complex populations and has dramatically affected patterns of mitochondrial variability in this mosquito. The observed patterns of mitochondrial variation in infected and uninfected populations provide important insights for understanding drive mechanisms for transgenic mosquito disease prevention strategies.

2. Material and methods

(a) Mosquito collections and identification

(i) South Africa

Wild gravid and recently blood-fed females were collected resting inside geese and chicken coops on the outskirt of Johannesburg (26°06′ S 27°50′ E) in March 2000. Females were allowed to oviposit and each egg raft was reared separately, allowing results for mitochondrial and nuclear sequence variation and Wolbachia-infection status to be traced to individual isofemale lines. For a priori identification of each family as either Cx. p. pipiens or Cx. p. quinquefasciatus the male genitalia from four males (older than 24 h) were dissected and slide mounted for DV/D ratio measurements (the distance between the dorsal and ventral arms of the male phallosome divided by the distance between the two dorsal arms) to determine what proportion of the sample were represented by Cx. p. pipiens, Cx. p. quinquefasciatus and hybrids between the two (Sundararaman 1949). No hybrids were detected. Study populations are denoted as: SAP, South Africa Cx. p. pipiens; SAQ, South Africa Cx. p. quinquefasciatus.

(ii) California and Thailand

Mosquitoes were collected as larvae, reared to adults in the laboratory, killed by freezing and stored at −80 °C or in 95% ethanol until processed for DNA extraction. The male genitalia from at least 40 males from each location were dissected and slide mounted for DV/D ratio measurements to determine what proportion of each sample was represented by Cx. p. pipiens, Cx. p. quinquefasciatus and hybrids. No hybrids were detected. Study populations are denoted as: CAP, California Cx. p. pipiens (Shasta Co.); CAQ, California Cx. p. quinquefasciatus (Riverside Co.); THQ, Thailand Cx. p. quinquefasciatus (Mesot, Thailand).

(b) DNA extraction

Ethanol stored specimens were re-hydrated in phosphate-buffered saline before DNA extraction. DNA from individual mosquitoes was extracted by salt extraction/ethanol precipitation as previously described (Rasgon & Scott 2003), re-constituted in deionized water and stored at −20 °C until used for PCR.

(c) Wolbachia infection

(i) Polymerase chain reaction

PCR was conducted using primers 99F and 994R (O'Neill et al. 1992), which amplify an approximately 900 bp fragment from Wolbachia 16S rDNA and are designed to be specific to Wolbachia of all strains. PCR conditions were as stated by Rasgon & Scott (2003). Known infected (LIN) and uninfected (LINT) colony mosquitoes (Rasgon & Scott 2003) were included in every reaction as positive and negative controls, respectively. Template DNA quality was assessed by successful amplification of a 400 bp fragment from insect 12S mtDNA using primers 12SA1 and 12SB1 (Simon et al. 1991) as previously described (Rasgon & Scott 2003). Amplified fragments were separated by agarose gel electrophoresis, stained with ethidium bromide (1 μl ml−1) and visualized with ultraviolet light.

(ii) Wolbachia surface protein gene sequencing

From each infected population, 3–5 infected individuals were randomly chosen and subjected to PCR amplification of the Wolbachia Surface Protein (wsp) gene using primers 81F and 691R as previously described (Zhou et al. 1998). Amplified fragments were separated by agarose gel electrophoresis, purified from the gel using Qiaquick columns (QIAGEN, Valencia, CA) and directly sequenced in both directions using an ABI Prism 377 DNA sequencer with Big Dye chemistry (Perkin-Elmer Applied Biosystems, Foster City, CA). Sequencher, DNA Sequence Analysis Software v. 4.0.5 (Gene Codes Corporation, Ann Arbor, MI) was used to align sequences.

(d) Mosquito mitochondrial variability

(i) PCR and sequencing

The primers ND4+ and ND4− were used to amplify a 389 bp fragment from the NADH dehydrogenase subunit 4 (ND4) gene from all specimens examined. Sequences were deposited in the GenBank database under accession numbers AY793688–AY793703. ND4 sequences from Culex restuans (Baltimore Co., MD) and Culex tarsalis (Kern Co., CA) were obtained as out group taxa and deposited in GenBank under accession numbers AY788866–AY788867. Primer sequences and PCR conditions were as stated by Gorrochotegui-Escalante et al. (2000). Amplified fragments were separated, purified and sequenced as described above.

(ii) Phylogenetic analysis of ND4 haplotypes

After removing the primer sequences, sequences were aligned with manual correction using Clustal X (Thompson et al. 1997). Haplotype networks were constructed using statistical parsimony criteria with a 95% cut-off (Templeton et al. 1992) using TCS v. 1.13 (Clement et al. 2000). The statistical parsimony algorithm calculates a cut-off number (parsimony limit) of mutational steps below which the haplotypes can be connected with 95% confidence. We also conducted maximum-likelihood phylogenetic analysis using PAUP* v. 4.01b 10 (Swofford 1998). For analyses, the GTR+I+G model was selected as the most appropriate evolutionary model of DNA substitution using Modeltest v. 3.06 (Posada & Crandall 1998). Tree robustness was evaluated by bootstrapping (500 replicates).

(iii) Statistical analysis of mitochondrial variability

The null hypothesis that all mutations were neutral was statistically ascertained by four tests. DnaSP (Rozas & Rozas 1999) was used to implement both Tajima's D-test (Tajima 1989) and Fu & Li's D-test; Allelix software (obtained from S. Mousset) was used to implement both Depaulis' H-and K-tests (Depaulis & Veuille 1998), where p values were calculated using 10 000 independent replicates. Partitioning of mitochondrial haplotype variation was analysed by Analysis of Molecular Variance (AMOVA; Excoffier et al. 1992) using Arlequin software v. 2.0 (http://lgb.unige.ch/arlequin), which uses a non-parametric permutation test to calculate the significance of variance components associated with each level of genetic structure. Significance levels for population differentiation pairwise comparisons were computed using a Markov chain exact-test procedure and Bonferroni correction for multiple tests. For each population, DnaSP was used to estimate haplotype diversity (Hd), the number of polymorphic sites, the average number of nucleotide differences (k) and the nucleotide diversity with a Jukes–Cantor correction (π2).

(e) Mosquito nuclear variability

(i) Nuclear gene phylogenetics

We conducted sequence analysis of Cx. pipiens spp. using internal transcribed spacer (ITS) and ribosomal DNA sequences because there are a large number of complementary sequences available for comparison from a wide geographical sample in the GenBank database (Miller et al. 1996). The entire ITS 2 region, the entire 5.8S gene and a portion of the ITS1 region were amplified from five SA pipiens and five SA quinquefasciatus specimens using primers PQ10 and CP16. Primer sequences and PCR conditions were as stated by Miller et al. (1996). Specimens were randomly chosen with the exception of specimen SAP3.54a, which was selected due to its unique haplotype/Wolbachia infection status (Wolbachia-negative, haplotype K—see §3). PCR products were purified using the MinElute clean-up kit (Qiagen, Valencia, VA), cloned into the pCR 4-TOPO vector and transformed into competent Escherichia coli cells (Invitrogen, Carlsbad, CA). One clone per individual was sequenced using the M13F primer. Sequences were deposited in the GenBank database under accession numbers DQ341106–DQ341115. Sequences were aligned with sequences from other geographical isolates (Miller et al. 1996, fig. 3) using Bioedit, for a total alignment of 748 bp including indels. The HKY+G model was selected as the most appropriate evolutionary model of DNA substitution using Modeltest v. 3.06 (Posada & Crandall 1998). These settings were used with a neighbour-joining algorithm to conduct phylogenetic analyses using PAUP* v. 4.01b 10 (Swofford 1998). Tree robustness was evaluated by bootstrapping (1000 replicates).

(ii) Estimates of nuclear variation

Allozyme analysis from previously published data (Cornel et al. 2003) was undertaken to estimate differences in nuclear diversity between infected and uninfected SA Cx. pipiens complex populations. We used Genepop v. 1.2 (Raymond & Rousset 1995) to calculate (i) average heterozygosity and proportion of polymorphic loci in each population and (ii) probability of differentiation between SA populations by a Markov-chain exact test.

3. Results

In total, we identified 12 mitochondrial haplotypes (A–L) in the Cx. pipiens s.l. member populations sampled, consisting of 27 variable sites (table 1) Phylogenetic analysis of mitochondrial sequences identified two main clades; A–J and K–L. These two clades form two unconnected networks as calculated by statistical parsimony (95% cut-off: seven steps; figure 1), but group together with strong bootstrap support (99%) by phylogenetic analysis (figure 2).

View this table:
Table 1

Alignment of 27 variable sites from Cx. pipiens spp. ND4 sequences.

Figure 1
Figure 1

Statistical parsimony haplotype network of Culex pipiens s.l. ND4 sequences. Unlabelled nodes represent inferred haplotypes. Oval size is proportional to haplotype frequency. Shading represents lack of Wolbachia infections (population SAP), no shading represents infection with Wolbachia (all other populations). 95% cut-off=7 mutational steps.

Figure 2
Figure 2

Phylogenetic analysis of Culex pipiens s.l. mitochondrial ND4 haplotypes. Phylogenetic tree was generated using maximum likelihood. Numbers at tree nodes represent bootstrap support values (500 replicates).

All individuals tested from populations CAP, CAQ, THQ and SAQ were found to be infected with Wolbachia. There were no differences in wsp sequences within or between populations; sequences were identical to those obtained from Wolbachia infecting North American and Asian Cx. pipiens s.l. (e.g. AF301010, AF216859, AF216860). In addition, recently colonized individuals from populations SAQ and CAQ were reproductively compatible when crossed in either direction (79% versus 93%, ANOVA, NS). Populations SAQ (n=50), THQ (n=10) and CAP (n=10) were fixed for mitochondrial haplotype K, which was the majority haplotype in population CAQ as well. One individual tested in population CAQ (n=10) possessed haplotype L, which differs from haplotype K by a single nucleotide substitution.

In contrast, none of the individuals assayed from population SAP contained detectable Wolbachia infections. In SAP mosquitoes, 11 haplotypes (A–K) were identified from the 21 families we examined. Population SAP exhibited an over 35 fold increase in Hd and an over 170 fold increase in nucleotide diversity (with a conservative Jukes–Cantor correction for multiple hits) compared to the combined infected populations (table 2). Neutrality of mitochondrial variation in population SAP was assessed by four statistical tests. Some tests indicated deviation from neutrality due to the large genetic distance of haplotype K from the other haplotypes present in population SAP (Tajima's D=−1.824, p<0.05; Fu and Li's D=−2.796, p<0.05). Depaulis' H- and K-tests did not suggest a deviation from neutrality and indicated that the number of haplotypes present and the Hd in population SAP did not differ from what was expected by chance (observed number of haplotypes=12, calculated 95% confidence intervals: 6–15, p=0.65; observed Hd=0.895, calculated 95% confidence intervals: 0.694–0.916, p=0.85).

View this table:
Table 2

Mitochondrial ND4 haplotype variability.

Analysis of mt haplotype variation (AMOVA) was performed (i) among populations with variance partitioned within geographical regions (South Africa, SAP and SAQ; California, CAP and CAQ; Asia, Thailand) and (ii) among populations with variance partitioned within Wolbachia infection status. In the first AMOVA, there was a large amount of negative among-region variance, indicating a lack of among-region population structure with individuals more related between rather than within geographical regions; a result consistent with a selective mt sweep. The majority of the significant variation indicative of population structure was found among populations within geographical regions, highlighting the substantial difference in Hd between populations SAP and SAQ. When haplotype variation was partitioned between Wolbachia infection status, the greatest amount of variation (more than 95%) was due to variation between infection types, suggesting differences in Hd that are correlated with presence or absence of Wolbachia (table 3). Markov-chain exact pairwise comparisons between all populations indicated that population SAP had significantly greater mitochondrial variation compared to every infected population (p<0.0001), but the infected populations did not differ significantly among themselves (p>0.17).

View this table:
Table 3

AMOVA results for mitochondrial haplotype variation.

These data indicate a significant reduction in mitochondrial variability in infected populations compared to the uninfected population SAP. An alternative explanation to a Wolbachia sweep for this result is that the infected populations experienced a bottleneck prior to a population range expansion. If this is true, we would expect that the infected populations would exhibit reduction in the diversity of nuclear loci as well as mitochondrial loci. Although there are fixed allozyme differences indicating that the two SA populations are highly differentiated from one another (Markov-chain exact test, p<0.0001), levels of diversity as calculated from 13 allozyme loci are remarkably similar between both populations (SAP: average heterozygosity=0.236, proportion polymorphic loci=0.85; SAQ: average heterozygosity=0.214, proportion polymorphic loci=0.85). Our data are inconsistent with the hypothesis that reduced mt variation is attributable to a bottleneck followed by population expansion.

Phylogenetic analysis of mosquito ribosomal and ITS DNA sequences indicates that SAP and SAQ individuals do not cluster together. SAP individuals were in the same clade and clustered with other pipiens isolates from around the world (e.g. Virginia, Colorado and Sweden). This was true even for individual SAP 3.54a, which had ND4 haplotype K. In contrast, SAQ individuals did not cluster and were more scattered on the tree, but in no case did they group closely with SAP individuals (figure 3). We suggest that the most parsimonious explanation for the presence of haplotype K in population SAP is shared ancestral polymorphism rather than hybrid introgression and that it is likely that population SAP predates the sweep of Wolbachia through Cx. pipiens s.l.

Figure 3

Phylogenetic analysis of Cx. pipiens s.l. ITS and ribosomal DNA sequences. Phylogenetic tree was generated using neighbour-joining with maximum-likelihood settings. Numbers at tree nodes represent bootstrap support values (1000 replicates). Taxa in bold represent clones from South African individuals from this study. Clone SAP05 (individual 3.54a, underlined) possessed ND4 haplotype ‘K’. Alphanumeric codes represent GenBank accession numbers from Miller et al. (1996).

Mitochondria sequence data can be used to calculate an approximate time-interval of the Wolbachia sweep through Cx. pipiens populations. Clade K–L contains no synonymous substitutions. The genera Culex and Aedes diverged approximately 38 000 000 years ago (Besansky & Fahey 1997). We calculated the nucleotide divergence between haplotype K and Aedes aegypti (GenBank accession number AF334848) and estimated the mutation rate for this gene (substitutions/site/year) at twofold (2.74×10−9) and fourfold (1.19×10−8) degenerate sites with a conservative Jukes–Cantor correction for multiple hits. Assuming a molecular clock (Likelihood ratio test, p=0.21), we calculated the maximum time-interval in which we would expect to observe at least one synonymous substitution (Rich et al. 1998). The 95% confidence interval for the time of the initial Wolbachia sweep through Cx. pipiens s.l. is 0–47 000 years ago. This date should be considered a preliminary estimate because the neutrality assumption was not supported by all tests.

4. Discussion

Our results indicate that a sweep of a microbial symbiont, Wolbachia had profound effects on the worldwide population structure of a complex of closely related insects. We observed almost no mitochondrial variation in infected Cx. pipiens populations collected at geographically distinct locations, compared to high levels of variation in a single uninfected population (SAP) that was sympatric with an infected population (SAQ). Haplotype K was shared by virtually all (98.75%) infected mosquitoes regardless of location. The only other haplotype found among infected mosquitoes (L, 1.025%) differed from K by a single substitution. Levels of nuclear diversity between population SAP and the sympatric infected population SAQ were similar, indicating that the reduced variation was due to a selective sweep of the K haplotype, not to a genetic bottleneck followed by population expansion. Haplotype K was observed at low levels (4.8%) in the uninfected population SAP. Phylogenetic analysis of nuclear DNA sequences supports the hypothesis that the presence of haplotype K in population SAP reflects an ancestral polymorphism rather than hybrid introgression. However, is should be noted that the hypothesis of introgression cannot be formally rejected since the nuclear region we sequenced exists in high copy number (Miller et al. 1996) and undergoes concerted evolution, making direct comparisons with mtDNA data somewhat problematic. Nevertheless, the observed patterns of nuclear and mitochondrial variation between infected and uninfected populations are so striking that we feel the most likely explanation is genetic hitchhiking of haplotype K with the initial Wolbachia invasion into the Cx. pipiens complex.

Our current and previous data (Cornel et al. 2003) suggest that population SAP may represent a new cryptic sibling species within the Cx. pipiens species complex. Reproductive isolation of this population from other sympatric infected populations has been confirmed by multiple nuclear and cytoplasmic markers. The role of Wolbachia in speciation events is controversial, but theoretical and empirical support for the idea is growing (Werren 1997; Rokas 2000; Bordenstein et al. 2003; Telschow et al. 2005). It is possible that Wolbachia-induced unidirectional CI can contribute to reproductive isolation between infected and uninfected populations. For instance, Drosophila recens is infected with Wolbachia, while its sister taxa Drosophila subaquinaria is uninfected. In laboratory experiments, gene flow via matings between D. subaquinaria males and D. recens females is hindered by behavioural isolation. D. recens males will readily mate with D. subaquinaria females, but gene flow in this cross is prevented by strong CI-induced sterility (Shoemaker et al. 1999). We hypothesize Wolbachia may be acting in a similar manner to prevent gene flow between sympatric infected and uninfected Cx. pipiens complex populations in SA, i.e. prezygotic isolation between Cx. p. pipiens males and Cx. p. quinquefasciatus females and CI-induced sterility between Cx. p. quinquefasciatus males and Cx. p. pipiens females. Before the initial horizontal transfer of Wolbachia into Cx. pipiens s.l., non-reciprocal gene flow might have existed between these two populations. A Wolbachia sweep would explain the pattern observed today of two sympatric but genetically isolated Culex populations, both with equal levels of nucleotide diversity, high levels of mt variability in the uninfected population and fixation of a single mt haplotype in the infected population that is shared with the uninfected population. Future studies of population mating structure are warranted to confirm or refute this hypothesis.

Worldwide, populations of Cx. pipiens exhibit what is perhaps the greatest variation in CI crossing patterns observed in any insect (Laven 1967). Our analysis of wsp sequences showed no variation, indicating that worldwide Culex Wolbachia strains are closely related. Considering that the Wolbachia sweep of the Cx. pipiens complex appears to have been recent; how could this multitude of crossing types evolve in such a short evolutionary time period? It is possible that genetic differences in mosquito host factors are partially responsible; such factors have recently been implicated in modifying CI phenotype in Culex (Sinkins et al. 2005) and in other insects (Bordenstein et al. 2003; Mercot & Charlat 2004). Alternatively or in concert with host related factors, mobile genetic entities in Wolbachia, such as phages (Sinkins et al. 2005; Duron et al. 2006) or transposable elements (Bordenstein & Wernegreen 2004; Sanogo & Dobson 2004; Wu et al. 2004; Duron et al. 2005) could facilitate rapid evolution by causing mutations from insertion/transposition and/or by acting as a mechanism for horizontal transfer of genetic material between Wolbachia strains (Sinkins et al. 2005).

In addition to an improved understanding of the mechanisms that contributed to the complicated population structure of an insect complex, spread of Wolbachia in mosquito populations is of applied interest for the control of vector-borne diseases (Turelli & Hoffmann 1999; Rasgon & Scott 2003; Sinkins 2004). For more than a decade a high-profile effort has been underway to genetically modify mosquitoes so that they no longer transmit pathogens (Beaty 2000). Transgenes that block pathogen transmission have been introduced into and expressed in mosquitoes (Ito et al. 2002). However, no empirically demonstrated method to spread or ‘drive’ these engineered genetic traits into wild mosquito populations currently exists. Strategies exploiting Wolbachia-induced CI to drive introduced transgenic traits into vector populations are being considered (Turelli & Hoffmann 1999; Rasgon & Scott 2003; Sinkins 2004). In one strategy, separate cytoplasmically inherited transgenes can be driven into a population along with Wolbachia as long as the transgene construct is transmitted to 100% of the offspring (Turelli & Hoffmann 1999).

Because mitochondria are cytoplasmically inherited and transmitted with near 100% fidelity, their dynamics can be used to predict how introduced transgenes might spread under the influence of a Wolbachia driver (Turelli & Hoffmann 1999). The essentially single Wolbachia mitochondrial haplotype and lack of variation in geographically separated wsp sequences indicate that Wolbachia invasion of the Cx. pipiens complex was recent, rapid and initiated by one or very few infected females. This implies that under proper circumstances, the number of transgenic mosquitoes that must be released may be low and that the gene of interest could spread in a relatively short period of time across an extensive geographical range.


We thank Julie Christiansen (UC Davis Mosquito Control Research Laboratory) for conducting crossing experiments between California and SA Cx. p. quinquefasciatus, Michael Petridis (Johns Hopkins University) for obtaining nuclear DNA sequences and Sylvain Mousset for helpful advice regarding Allelix software. We thank three anonymous reviewers for comments and suggestions that significantly improved the manuscript. This research was supported by the Johns Hopkins Malaria Research Institute and by NIH grant GM020092 to J.L.R., funds from the state of California to T.W.S. and from the University of California Mosquito Research Laboratory to A.J.C.


    • Received December 22, 2005.
    • Accepted January 23, 2006.


View Abstract