Skip to main content

Population age structure and age structure modification via Wolbachia in Anopheles gambiae

Final Report Summary - ANOPOPAGE (Population age structure and age structure modification via Wolbachia in Anopheles gambiae)

Executive summary:

Mosquitoes of the anopheles gambiae species complex (an. gambiae and an. arabiensis in particular) are the major vectors of malaria in Africa. The project explored the potential for novel malaria control strategies based on the use of wolbachia inherited symbionts. It also examined methodologies for investigating population age structure in natural populations of these species, because some strains of wolbachia can shorten lifespan and mean age reached by females age is a very important parameter in malaria transmission given plasmodium parasite incubation periods.

We have introduced a number of strains of wolbachia, together with several other strains of the bacterium, into various anopheles cell lines, creating stable transinfections. Using quantitative polymerase chain reaction assays together with fluorescence microscopy we have observed steady increases in density over time and observed that some strains form much more stable infections than others in anopheles. As well as providing a useful way to determine which strains should be taken forward for use in embryo microinjections of an. gambiae, the cell-line adapted wolbachia have also been purified and used for embryo, ovary and adult thoracic an. gambiae injections.

Somatic infections using injection of purified wolbachia into adult female an. gambiae have demonstrated a direct inhibition of plasmodium parasites in the presence of wolbachia. We have shown that in both anopheles cell lines and adult aedes aegypti and anopheles gambiae mosquitoes infected with wolbachia, there is a significant and chronic upregulation of a large number of immune genes. Using gene knockdown experiments we have shown that a major contributory factor to the plasmodium inhibition is the upregulation of a particular immune gene, TEP1, in the presence of wolbachia. Immune upregulation has been shown to be induced by recombinant WSPand has a direct effect on the wolbachia itself, acting to limit density and thus potentially to limit virulence.

We have set up an age estimation assay for the an. gambiae complex based on transcriptional profiling using quantitative reverse transcription polymerase chain reaction (PCR) profiles of four genes we identified in a microarray study. This assay successfully predicts age to within acceptable parameters for lab-reared an. gambiae specimens. A series of collections of an. gambiae s.I. have been conducted at various field sites in Burkina Faso and the captive cohort method (CCM) has been used to estimate the population age structure at different times of the year. A continuous population of an. arabiensis contained in a large field cage (screen house) has been set up in Kenya.

A suite of population dynamic mathematical models have been produced that examine the spread of life-shortening wolbachia through seasonal anopheles gambiae populations and effects on vectorial capacity and reveal that direct inhibition of plasmodium parasite development is more efficient strategy than use of virulent life-shortening wMelPop strain. These models emphasise the importance of beginning releases at the beginning of the wet season if success is to be achieved in initiating wolbachia spread while minimising overall release numbers. In addition they show how highly male-biased releases can be used to initiate wolbachia spread, combining temporary population reduction through the incompatible males mating with wild females with the longer term benefits of the spreading wolbachia infection in reducing vectorial capacity. The models provide a tool for selecting and optimising the best strategies for wolbachia-based disease control.

Project context and objectives:

There are estimated to be 500 million cases of malaria annually and over a million deaths, the latter primarily associated with plasmodium falciparum infections in sub-Saharan Africa. Mosquitoes of the anopheles gambiae complex (an. gambiae and an. arabiensis in particular) are the major vectors in Africa and any reductions in their ability to transmit the disease would thus provide considerable health benefits. Control programmes targeting the vector have been the most successful method of controlling malaria, but the effectiveness of pesticide-based strategies has been compromised by the repeated emergence of insecticide resistance and novel methods are much needed.

Some of the most sensitive parameters contributing to the vectorial capacity of malaria vector species are those determining the number of long-lived adult females. Long-lived females are significant because after picking up the plasmodium pathogen there is an incubation period before transmission can occur. On entry into the vector exflagination occurs followed by fertilisation and zygote elongation. The ookinete stage in the midgut develops into the oocyct in the midgut wall, where multiplication results in the release and migration of thousands of sporozoites to the salivary glands. Only then is the mosquito infectious and able to transmit malaria the next time it takes a blood meal.

Although timing of the extrinsic incubation period depends on temperature and species of plasmodium / mosquito, typically at least 9 to 15 days are needed (around 11 days for p. falciparum in an. gambiae at 24 °C). Adult mosquitoes experience a high daily mortality rate such that only a small percentage of the total population survive long enough to transmit plasmodium. However, despite the critical importance of estimating the fraction of old mosquitoes in a population this is one of the least understood components of vectorial capacity. The reason for this is that current direct measures of insect age in wild populations are relatively inaccurate, while indirect estimates (by mark-release-recapture experiments for example) are labour-intensive and logistically difficult. Furthermore, it is often not fully appreciated that reduction in mean lifespan is a major reason why residual insecticides can be so effective (at least prior to the development of insecticide resistance). Rather than simply reducing mosquito abundance uniformly, they also, through a cumulative uptake effect, preferentially affect the class of older individuals responsible for transmission.

Therefore there is considerable motivation to attempt to identify better methods to estimate population age structure in the field. Molecular methods based on changes in transcription of certain genes over time have been developed for aedes aegypti mosquitoes and an aim here was to develop a similar transcriptional profiling methodology for an. gambiae. An additional aim was to examine the application to an. gambiae of the CCM, developed by Carey, Müller and colleagues and used successfully to examine the population age structure in Mediterranean fruit flies. This simple, low-tech and inexpensive method only requires the establishment of life table assays for field-collected mosquitoes of unknown age and for reference cohorts of known age raised under identical conditions. Data on the remaining lifespans of live-captured individuals and from reference cohorts are then used to estimate population age structure using a deconvolution model based on Carey's equality; in a stationary population, age composition and the distribution of remaining lifespans are identical and can be used to estimate age structure parameters in wild populations. This method was adapted and used to explore age structure properties and variation in wild an. gambiae population from study area in Burkina Faso and the results compared with the transcriptional profiling method.

Wolbachia pipientis is a maternally inherited bacterial symbiont first reported in mosquitoes but also found in many insects, other arthropods and filarial nematodes. In arthropods it manipulates host reproduction to its own advantage, by skewing sex ratios towards females (the only sex that transmit the bacterium), or by inducing crossing sterilities known as cytoplasmic incompatibility (CI). In its basic unidirectional form, CI is expressed as offspring early embryonic arrest when uninfected females mate with infected males - in other words wolbachia modifies the sperm of infected males. However a rescue function when wolbachia is present in the egg allows the offspring of infected females to develop normally, regardless of the infection status of the males with which they mate. Thus in a mixed population infected females will produce a greater mean number of offspring. Once a threshold population frequency has been exceeded, which depends on the maternal transmission efficiency, level of CI and any fitness costs, the reproductive advantage of infection will steadily increase with generation and infection frequency will rise with increasing rapidity. Population invasion by wolbachia has been directly observed in nature, with drosophila simulans in California the best documented example.

A virulent wolbachia strain has been identified that, as well as being capable of inducing CI, preferentially accelerates late-life mortality (shortens adult lifespan) in drosophila melanogaster. It appears to over-replicate and causes a pathology in nervous tissues similar to expanding popcorn kernels, resulting in it being given the name popcorn. It was afterwards designated wMelPop due to its close genetic similarity to the wMel strain, the genome of which has been sequenced.

If a life-shortening phenotype can be successfully transferred into anopheles mosquitoes it could reduce their ability to transmit pathogens, by removing the sector of the mosquito population responsible for the majority of disease transmission. It has already been successfully demonstrated that the lifespan-shortening effect of this strain is maintained following transfer to aedes aegypti mosquitoes, the main vectors of dengue fever. In wild mosquito populations the high daily mortality rate means that most of the next generation are the offspring of younger females. However the fitness cost imposed by a virulent wolbachia would impact its ability to spread through populations, especially in the case of subdivided rather than continuous populations and raise the threshold frequency that must be exceeded before population spread can occur. Thus one of the aims of the project was to assess the degree of virulence of wMelPop and various other wolbachia strains, for their degree of virulence and effects on anopheles fitness and lifespan. It was also an aim to produce mathematical models that would allow the effects of lifespan and fitness parameters on spread dynamics to be assessed.

A separate impact of the successful introduction of wolbachia into anopheles could be on research aiming to develop strategies for reducing their susceptibility to plasmodium transmission. Natural an. gambiae populations show variability with respect to the degree of plasmodium susceptibility. Naturally occurring alleles of certain an. gambiae immune genes, such as TEP1, have been shown to have strong inhibitory effects on plasmodium development. Thus the possibility of wolbachia effects on plasmodium development in the mosquito was also examined in this project.

The objectives of the project were:

1. to create and characterise wolbachia-infected anopheles cell lines: To infect anopheles cell lines with wolbachia and to examine and compare cellular interactions of wolbachia strains within host cells, using fluorescent microscopy and molecular methods to quantify the expression of immune genes.
2. to establish and characterise an. gambiae lines infected with wolbachia: To transfer wolbachia of various strains into an. gambiae, by microinjection of embryos and adults with purified bacteria from anopheles cell lines; to characterise the effects of wolbachia on plasmodium development, mosquito immunity, lifespan and host fitness, CI, maternal transmission and tissue distribution of introduced wolbachia
3. to develop a molecular age estimation assay for the an. gambiae complex: To identify an. gambiae genes showing age-specific patterns of expression using microarrays, allowing expression levels of all predicted genes in its genome to be compared at different ages; to develop a molecular method (based on quantitative reverse-transcription polymerase chain reaction) for an. gambiae age estimation and calibrate and assess this method using laboratory reared mosquitoes
4. to set up semi-field populations of an. arabiensis in Kenya: To set up a greenhouse population of an. arabiensis in Kenya that is viable over multiple generations and use release of marked mosquitoes to calibrate molecular age assays and compare age structure between greenhouse and wild populations
5. to assess field population age structure in Burkina Faso: To estimate population age structure in the M and S molecular forms of an. gambiae at three field sites in different seasons, using the molecular methodology and the CCM
6. to mathematically model wolbachia population dynamics including host age structure: To build a suite of models that will allow the analysis of different interventions that affect adult mosquito longevity; to incorporate the dynamics of wolbachia spread in an age-structured population, allowing an assessment of which strains will be most useful for creating an efficient malaria control strategy.

Project results:

Creation of wolbachia infected cell lines

Interspecific transfer of wolbachia strains to anopheles gambiae cell lines, which are not naturally infected with wolbachia, has been successfully achieved using a modified version of the shell vial technique using drosophila eggs or wolbachia infected insect cell lines. The procedure is based on the use of a syringe and a freeze-thaw cycle of the donor cells. The modified protocol has been applied to transinfect four anopheles cell lines using the wolbachia strains wAlbB from aedes albopictus, wAu from drosophila simulans Coffs and wMel and wMelPop both from drosophila melanogaster, wPip from Cx quinquefasciatus and wRi from D simulans DSR. Transinfected cell lines was sequentially transinfected with the same wolbachia strain twice in order to increase the bacterial load. The 4a3b haemocyte-like cell line, which was previously established from neonate A. gambiae larvae, was successfully transinfected with wAlbB, wAu, wMel and wMelPop and these infections remained at high density after 30 to 60 passages. Msq43 was also successfully transinfected with wMelPop, wAlbB and wMel. When 4a3b and Msq43 cell lines were transinfected either once or multiple times or multiple times using the antibiotic regimen every third passage, the proportion of wolbachia per cell increased from passage 30 to passage 60.

Comparative studies looking at the density of the four wolbachia and the effect on cell growth and division of the four different wolbachia infections in the three different hosts have been conducted using primers designed for the single copy genes from both wolbachia and anopheles, wolbachia densities were estimated using quantitative polymerase chain reaction for all cell cultures at thirty and sixty passages. These help determine how host genotype impacts on wolbachia ability to establish itself in a new host. wolbachia specific probes labeled with Rhodamine Green were used to detect wolbachia distribution in cell cytoplasm; wolbachia were distributed throughout the host cytoplasm in infected cells.

A wolbachia wMelPop strain transinfection in ae. aegypti mosquitoes was found using full genome microarrays, followed by quantitative reverse transcription polymerase chain reaction experiments, to induce a major constitutive upregulation of a large number of immune-related genes. We also showed that the development of Brugia filarial nematode parasites to the infectious L3 stage was strongly inhibited in the presence of wMelPop. The list of upregulated genes in wMelPop-infected ae. aegypti includes cecropins, LRIM1, TEP1 and several others whose anopheles homologs have known inhibitory effects on plasmodium parasites that cause malaria. We were therefore interested in determining to what extent different wolbachia strains induce immune upregulation in an. gambiae and how this immune upregulation translates into pathogen transmission blocking. For this purpose the levels of transcription of a number of innate immune genes in the transinfected an. gambiae cell lines have been characterised using quantitative reverse transcription polymerase chain reaction. Upregulation of an. gambiae cecropin and, defensin antimicrobial peptides, LRIM1, TEP1 and a number of other immune genes has been observed compared to controls in the wolbachia-transinfected cell lines and this effect is not restricted to the over-replicating wMelPop strain.

We have also shown that recombinant wolbachia surface protein (WSP) is a strong immune elicitor in an. gambiae cell lines, stimulating highly increased transcription of a number of immune genes. At least two of the upregulated genes, TEP1 and APL1, are known to be important in plasmodium killing in this species.

Creation of wolbachia transinfections in adult an. gambiae

Effects of wolbachia somatic infections on immunity and plasmodium

A reproducible and effective protocol for the purification of wolbachia from cell lines has been established. This technique allows us to grow wolbachia to high quantities in a timely manner reproducibly, in order to test in vivo, via somatic infection of adult mosquitoes, the influence that different wolbachia strains may have in establishing an infection in adults.

We created transient somatic infections of wolbachia by intrathoracic inoculation and these individuals also showed increased expression of the malaria-related immune genes compared to uninfected, E. coli-injected or buffer-injected controls. Highly significant reductions in plasmodium infection intensity were then observed in the wMelPop-infected cohort compared to all the controls. This inhibition effect was lost in the presence of TEP1 gene knockdown, evidence that the upregulation of this gene in the presence of wMelPop plays a major role in the phenotype.

An experiment to compare the levels of upregulation in somatic and stably inherited wMelPop infections in ae. aegypti revealed that levels of upregulation were much lower in the somatic infections than in the stably transinfected line; inhibition of development of Brugia filarial nematodes was nevertheless observed in the somatic wMelPop infected females. The effects observed in an. gambiae are also likely to be larger if stably inherited wolbachia transinfections can be created and that somatic infections of wolbachia provide a useful model for examining effects on pathogen development or dissemination.

Our experiments have also shown that different wolbachia have different effects on the survival of mosquitoes in somatic injections, some increasing and others decreasing lifespan. Similar comparative studies have also looked at wolbachia growth kinetics, involving immune knock-downs of the host and immune activation assays have helped determine the effect of this host response to wolbachia density and growth.

Experiments to create a stable inherited transinfection in an. gambiae

The first attempts to transfect anopheles gambiae with wolbachia concentrated on the transfection of the wMelPop strain using wolbachia purified from mosquito cells in culture. After injection of 18 000 pre-blastoderm embryos using this method, only a small number of samples positive for wolbachia were obtained in the G0. It was concluded that the wMelPop strain is too virulent to allow for stable transfection, paralleling work in another lab in which this strain was introduced into ae. albopictus and found to dramatically reduced egg hatching. Therefore injection attempts shifted onto injection of lower density / less virulent strains such as wMel, wAlb and wPip. Embryos were injected using wolbachia purified from cells, but cytoplasm transfer was the main method used it appeared from initial experiments to be to be more effective at generating G0 transfections (although it is a technically more deminading procedure and throughput is low). We have been most successful at generating G0 and G1 transinfections in the injection generation using the strain wAu, which does not cause CI and is not virulent. However no transinfections have been stably inherited over multiple generations and many females found to be positive for wolbachia using PCR for detection of the wsp gene, were found to be negative in the ovarian tissues.

In order to determine if An gambiae ovaries can sustain wolbachia infection we have established a direct in vivo ovary injection technique. Mock dye ovary injections showed an average of 65 % fully stained ovaries and only 7% of all injected ovaries showed no staining. Injected females had an average 70 % survival rate after injections and a egg hatched rate of roughly 40 % when both ovaries were injured. Single ovary analysis on injected ovaries through time, using a quantitative polymerase chain reaction assay, have shown that wolbachia is capable of replication in the ovaries. However, it has not so far been possible to observe maternal transmission and thus transinfect subsequent generations, using this methodology.

Development of a transcriptional profiling method for age determination in an. gambiae

The age distribution of female mosquitoes in the field is a critical component of vectorial capacity, due to the extrinsic incubation period of mosquito-borne pathogens. However this parameter has not been well characterised in malaria vectors due to methodological difficulties; transcriptional profiling provides a potential new approach for age determination. In anopheles gambiae, microarrays were used to examine global gene expression over adult life. Nine genes were selected from the 2714 gene transcripts that displayed age-related transcription patterns and quantitative RT-PCR used to select the four best performing genes. The resulting age estimation assay was able to predict quite accurately the mean female age from lab-reared samples in the insecaries at the University of Oxford.

Therefore the assay was taken forward to testing using field-collected samples. When the assay was used to estimate the age of wild-caught specimens from Burkina Faso for calibration, the accuracy of age prediction was lower than for lab-reared specimens. This is probably due to the logistical difficulties such as post-capture dehydration stress which is likely to influence gene expression patterns, transporting specimens from the field to the lab in ribonucleic acid (RNA)later rather than being able to process them fresh and the need to also dissect wild-caught specimens prior to placing in order to identify them to species. It is also possible that sequence variation exists in the primers used for assay in wild specimens compared to laboratory colonies. Thus the transcriptional profiling method remains a potentially very useful tool for malaria epidemiological studies, but has proven logistically difficult to apply successfully in the field.

Continuous semi-field populations of an. arabiensis established in Kenya

Anopheles arabiensis populations were colonised in Kenya by collecting indoor resting blood-fed or gravid an. gambiae s.l from the villages within and near Mbita area. The females were held individually in glass vials and those laying eggs are killed and their carcass subjected to species identification by polymerase chain reaction. An. arabiensis offspring were pooled and the resulting adults kept in the insectary and females periodically allowed to obtain blood meals. A semi field system is a recently re-screened double wall screened house measuring 11m x 7.1m divided into two enclosures. Each enclosure has a mud-walled and grass thatched hut measuring 2m x 1.8m. Mosquito larval habitats were constructed from soil from natural larval habitats. Artificial resting sites were provided for by clay pots embedded on the mud walls. The system holds a variety of indigenous plant species growing and these provide a source of sugar and also offer refuge sites for the captive male and female mosquito population. The establishment of these populations will facilitate a variety of behavioural, ecological and lifespan studies of the an. gambiae complex.

Field population age structure characterisation in an. gambiae from Burkina Faso

In a first set of activities, gravid female mosquitoes of the anopheles gambiae complex were collected in human dwellings in the village of Bama (11°23'N; 4°24'W), 30 km north of Bobo Dioulasso, during the dry season (February and April 2009). They were stored individually in plastic cups and allowed to lay their eggs. Soon after egg-laying, the females were killed, their deoxyribonucleic acid (DNA) extracted and they were identified to species and molecular forms using a specific polymerase chain reaction assay. All females were anopheles gambiae M form, as expected from earlier collections conducted in the area at this period of the year.

The progenies were reared in the insectaries at Institut de Recherche en Sciences de la Santé (IRSS) Bobo Dioulasso under standard rearing conditions (T = 28 ± 1 °C, 80 ±10 % relative humidity and 12:12 light:dark cycle) and fed ad libidum (0.15 mg TetraMin baby fish food per larva per day). Emerging adult mosquitoes were maintained in small netted cages (20 x 20 x 20 cm) in the insectaries where they were fed glucose ad libidum. Depending on the treatment, female mosquitoes were offered, or not, a bloodmeal. Adults were sampled at regular time intervals at the day of emergence (D0) and every five days afterwards. A sample of 20 to 50 females was obtained from the pooled progenies at D0, D5, D10, D15, D20 and D25 post-emergence. Their gonotrophic status (Unfed, Half-Gravid, Gravid) was recorded, their ovaries were dissected and their parity status was checked (Parous/Nulliparous), their insemination status was checked after dissection of the spermatheca, two legs were dissected out to serve as DNA template for molecular identification (as above) and the remaining of the carcass was stored in RNAlater at -20 °C. Samples of males mosquitoes (N=10 per sample) were also collected at D0 and D5 and stored in RNAlater. Every specimen was PCR-identified at the IRSS and the corresponding biological samples stored in RNAlater were sent to UOXF to serve as calibration standards for the age grading assay based on quantitative reverse transcription (qRT) PCR.

In addition, wild mosquito populations were sampled in three field sites in Burkina Faso, including Soumousso (11°00'N; 04°02'W), a typical savanna village with seasonal malaria transmission, Bama (above), a rice fields area with year-long presence of an. gambiae M from and Dandé (11°34'N;04°33'W), a typical savannah village with high ITNs coverage. Wild female mosquitoes were collected using centre for disease control (CDC) light-traps during the night and by aspiration when resting in human dwellings in the morning. Mosquitoes from the anopheles gambiae complex were visually sorted from other culicidae. Samples of 250 to 300 females an. gambiae s.l. were collected in each collection round in each study site. As above, each specimen was dissected in the field, their gonotrophic status, nulliparous/parous status and insemination status were determined and recorded, two legs were dissected out and kept in labelled tubes for PCR genotyping and the remaining of the carcass was stored in RNAlater and kept at -20 °C pending age grading. A set of N=50 mosquito per site and collection date was identified to species and molecular form.

Samples collected in Bama consisted only of the M molecular form of an. gambiae at all collection dates whereas samples from the other two field sites presented a mixture of all three taxa. In Soumousso, the S form was predominant in field collections during the rainy season, although an.arabiensis predominated in the early rainy season (May). In Dande, the relative proportion of the three taxa in the samples was more variable: the M form of an. gambiae predominated at the onset of the rainy season and the S form was predominant during the rainy season.

Residual demography in an. gambiae from Burkina Faso

Building on recent advances in empirical and analytical methods developed by Carey and others on medfly biodemography, we used a fine-grained sampling in order to estimate the mean biological age of wild mosquito populations and to explore changes in the means over time, rather than attempting to estimate the entire age distributions.

Wild mosquitoes were sampled daily for a period of one month (e.g. 30 consecutive days) in the village of VK7, our study site in southwestern Burkina Faso where the M form of an. gambiae can be found all year long. Samples of 20 females an. gambiae were collected each day, through aspiration while resting inside human dwellings. Wild-caught females were placed each in an individual cup covered with a net and was monitored in the insectary until death. Mosquitoes were maintained on water and 10 % sugar solution available ad libitum and reared under standard insectary conditions (T = 28 ± 1 °C, 80 ± 10 % relative humidity and 12:12 light:dark). Cups were checked every day at roughly the same time to record mortality data. Reference cohorts were established from eggs laid by the gravid females collected in the field. They were reared in the same room as the adults, in standard insectary pans at a density of 200 larvae/pan and fed Tetramin baby fish food. Upon emergence, a set of 200 randomly selected females and 100 randomly selected males were placed in individual cups covered with a net and their longevity was monitored as described above to establish reference life tables.

In order to document changes in mean biological age of the wild mosquito population, mosquito collections were conducted from February to March in 2011 and 2012 (dry season), from July to August 2011 (middle of the rainy season) and from November to December, 2011 (late rainy/early dry season) in Bama. A subsample of test specimens (i.e. 50 mosquitos per session, including mothers of the offspring mosquitoes used to build the reference cohort) were checked by PCR and all belonged to the M molecular form of an. gambiae. Enzyme-linked immunosorbent assay (ELISA) tests for plasmodium infection detection in wild-caught females were performed at IRSS, to account for a possible effect of infection on longevity. Post-capture longevity in field-collected females ranged from 2 to 38 days. Interestingly, no female in the reference cohort survived more than 26 days and males did not survive beyond 23 days.

Hence, just as was observed with medflies, despite being maintained under exactly the same conditions in the laboratory as adults, the captive lifespan of 'once-wild' mosquitoes of unknown age exceeded the lifespans of never-wild mosquitoes reared from eggs obtained in the laboratory and thus were of known age. We explored the hypothesis that reduced lifespan in females from the reference cohort resulted from the fact that females were unmated and/or had no chance to experience blood feeding prior to their capture, as many of the field-collected females would have. For this purpose, we implemented a series of experimental treatments, using a laboratory mosquito colony of an. gambiae M form originated from Bama and maintained at IRSS. The longevity of mosquitoes was monitored as above, from emergence till death, in inseminated versus non inseminated females fed with glucose and/or blood. Access to a bloodmeal did not significantly increase longevity in non-inseminated females, provided access to glucose is granted. Bloodfeeding, however, seemed to have a much more important effect when the females are inseminated, probably reflecting physiological trade-offs in resources allocation for longevity and reproduction. Moreover, even when fed glucose and blood, the longevity of inseminated as well as non inseminated females did not exceed the upper limit of 26 days, as found in our previous experiment for the reference cohort.

Altogether, these results suggested that the potential longevity of wild-caught (female) mosquitoes can be substantially greater than that of mosquitoes that have emerged in the laboratory. One hypothesis currently under investigation is that once-wild mosquitoes might have access, at some point during their development or through their diet in the field, to certain amounts or kinds of important microorganisms with an impact on longevity in never-wild females which do not have access to these micro-fauna.

Further insights into the age-structure and age-structure variation in wild an. gambiae populations can however be gained when monitoring raw and mean postcapture lifespans. Our results obtained for the first field session conducted during the dry season suggest that mean postcapture lifespan steadily decreased from February to March 2011, as the dry season progresses in the area.

Estimation of male population age structure in an. gambiae from Burkina Faso

Recent studies have shown that the M and S molecular forms of an. gambiae have species-specific swarming behaviour and dispersal which suggests that the environmental and endogenous cues that determine the nature of swarming/mating sites must influence male age structure in the field.

To test whether we were able to age-grade male an. gambiae following the methodology described in Huho et al. (2006), we first conducted a preliminary experiment on the correlation between male age and morphology of the testes in the laboratory. Males an. gambiae s.s. M form from three age groups (1, 5 and 10 days-old virgin males, 30 individuals per group) belonging to an F1 generation of wild mosquitoes collected from VK7 were dissected. The number of spermatocysts and the volume of the sperm reservoir were assessed using a 20x magnification microscope and compared between age groups. We found that male age was strongly correlated with the morphology of the testes. First, the number of spermatocysts decreased significantly with male age with highest number of spermatocysts in one day-old males and lowest in 10 days-old males. Second, the size of sperm reservoir was positively associated with male age, with less than 40 % of reservoir filled with sperm in one day-old males and more than 80 % in more than four days-old males.

Collectively, these results indicate that the methodology previously developed by Huho et al. is a reliable morphological technique to gauge the age of male an. gambiae. We then used this technique to gauge the age of wild male mosquitoes in natural populations of an. gambiae M and S forms collected in VK7 (M-form) and Soumousso (S-form) and to investigate seasonal changes in males age-structure.

The global male age structure did not differ between villages with a higher percentage of old males (more than four days) in collections from swarms than from indoor resting sites inside the villages (86 % versus 66 % for Soumousso) or from indoor-resting sites close to a breeding site (86 % versus 47 %, for Soumousso). There was, however, a difference between villages in the effect of collection site on male age structure.

In Soumousso, males collected from indoor resting sites in the village were not significantly different in age than those collected near breeding sites (respectively 69 % of more than4 days-old males versus 67 %). In contrast, in VK7 the proportion of males that were more than4 days-old was significantly higher in the indoor resting sites than those collected near breeding sites (66 % versus 47 %). Older males (more than4 days-old) were mostly found toward the second half of the rainy season, particularly between August and October, with the highest proportions recorded in September, irrespective of the sampling sites. Conversely, younger males were mainly collected from June to August irrespective of sampling sites, peaking in July.

Mathematical modelling of wolbachia population dynamics

Wolbachia spread between insects by maternal transmission from mother to offspring. The bacteria employ a mechanism known as CI to increase their frequency across successive generations of the host population. The trade-off between the fitness advantages and costs incurred by wolbachia infection means that wolbachia must obtain a threshold frequency in the host population before it can spread. A mathematical modelling approach was developed to investigate the dynamics and spread of wolbachia in anopheles gambiae mosquitoes. The models aim to guide the development of strategies for releasing wolbachia into wild mosquito populations in order to prevent the transmission of mosquito-borne diseases. Model development focused on incorporating major aspects of the demography and population dynamics of an. gambiae to investigate how these processes influence the spread of wolbachia. Previous models of wolbachia dynamics have not considered the hosts' population dynamics in detail and have assumed temporally constant host population size with a limited representation of host age-structure. The modelling approach that we have developed includes an age-structured representation of the mosquito host population that considers juvenile and adult life stages and juvenile, or larval, density-dependent survival.

These demographic aspects are known to be central to the population dynamics of an. gambiae and have been shown by this analysis to have important implications for the spread of wolbachia in this host species. Our analysis explores seasonal fluctuation in mosquito abundance and how this can affect the design of strategies for wolbachia release. The model has also been extended to explicitly consider the different sexes of adult mosquito which is significant because only adult females are responsible for the transmission of human diseases. Model analysis has explored the relative benefits of male-biased releases of wolbachia-infected mosquitoes in terms of the short and long term reduction in disease transmission achieved by the releases. Finally, the model was further developed to explore spatial mosquito population dynamics including spatial subdivision of juvenile habitat and adult dispersal. These spatial processes are clearly central to the population dynamics of an. gambiae. This investigation has shown that spatial heterogeneity in the mosquito habitat strongly affects the spatial spread of wolbachia once the infection has become established locally.

Modelling the effect of mosquito demography and population dynamics on wolbachia spread

We developed models to explore interactions between host population dynamics and wolbachia infection frequency for an age-structured insect population regulated by larval density dependence. The model was formulated mathematically using age-structured integral equations. Using this model we derived a new expression for the unstable equilibrium wolbachia infection frequency. The unstable equilibrium infection frequency is an important result of mathematical analysis of wolbachia dynamics because it predicts the frequency of wolbachia infection that must be achieved in the host population before the bacteria will spread. The expression that we have derived is analogous to formula obtained by previous modelling studies. Our result generalises previous results to allow for juvenile density-dependent survival, age-dependent mortality in adult mosquitoes and imperfect transmission of wolbachia.

The dynamic properties of this age-structured model have important qualitative differences to those resulting from models that do not consider the host population age-structure. For example, in an age-structured insect population, a transient elevation of the infection frequency of wolbachia above the unstable equilibrium value does not necessarily cause wolbachia to spread. Repeated small releases of wolbachia-infected mosquitoes may be more effective in triggering spread than a single large release. This is because repeated small releases avoid causing a large increase in the density-dependent mortality experienced by the larval population. The unstable equilibrium frequency more accurately predicts the spread of wolbachia when the mosquitoes are released in a large number of small releases over an extended time period. However, this modelling analysis has emphasised that predicting wolbachia dynamics in wild mosquitoes can be a complex process. For example, we explored the effect of a range of assumptions about the form of density-dependent mortality experienced in by mosquito larvae on the model results.

The assumed form of density-dependence considerably affected the conditions necessary for wolbachia spread and the release strategy that is most likely to achieve spread. In general the analysis demonstrates that detailed field data describing the demography and ecology of the target mosquito species will be necessary to accurately predict the dynamics and spread of wolbachia.

Modelling male-biased wolbachia releases and the transmission of mosquito-borne diseases

An important consideration when introducing wolbachia-carrying mosquitoes into natural populations is the minimisation of any transient increase in disease risk or biting nuisance. This may be achieved by predominantly releasing male mosquitoes. To explore this, we use a sex-structured model of wolbachia-mosquito interactions. We first show that wolbachia spread can be initiated with very few infected females provided the infection frequency in males exceeds a threshold. We then consider realistic introduction scenarios involving the release of batches of infected mosquitoes, incorporating seasonal fluctuations in population size. For a range of assumptions about mosquito population dynamics we find that male-biased releases allow the infection to spread after the introduction of low numbers of females, many fewer than with equal sex-ratio releases. We extend the model to estimate the transmission rate of a mosquito-borne pathogen over the course of wolbachia establishment. For a range of release strategies we demonstrate that male-biased release of wolbachia infected mosquitoes can cause substantial transmission reductions without transiently increasing disease risk. The results show the importance of including mosquito population dynamics in studying wolbachia spread and that male-biased releases can be an effective and safe way of rapidly establishing the symbiont in mosquito populations.

Modelling the spatial dynamics of wolbachia infections

Previous modelling studies of the spatial dynamics of wolbachia have used a reaction-diffusion approach which provides a simple analytic description of key properties of spatial spread such as the equilibrium speed of a travelling wave of wolbachia infection. However the reaction-diffusion approach does not account for significant details of host population dynamics. We developed a metapopulation model describing the spatial dynamics of wolbachia in an age-structured host insect population regulated by juvenile density-dependent competition. The model produces similar dynamics to the reaction-diffusion model in the limiting case where the host's habitat quality is spatially homogeneous and wolbachia has a small effect on host fitness. When habitat quality varies spatially, wolbachia spread is usually much slower and the conditions necessary for local invasion are strongly affected by immigration of insects from surrounding regions. Spread is most difficult when variation in habitat quality is spatially correlated. The results show that spatial variation in the density-dependent competition experienced by juvenile host insects can strongly affect the spread of wolbachia infections, which is important to the use of wolbachia to control insect vectors of human disease and other pests.

Potential impact:

The major components of the project have provided impacts and advances over the current state of the art for the mosquito / malaria / wolbachia research communities and ultimately for control programmes in disease endemic countries (DECs), as detailed below.

1. The project has demonstrated that a number of different strains can successfully establish themselves in anopheles gambiae cell lines and as somatic infections in adult anopheles mosquitoes. Somatic infections of wolbachia have been shown to result in strong inhibition of the development of plasmodium parasites and the mechanism has been shown to involve the upregulation of important immune effector genes. These data provide a stimulus for further wolbachia-Anopheles research in this area and provide motivation for a significant shift in focus from life-shortening to plasmodium inhibition, as a means of directly reducing the susceptibility of the mosquito to the development of plasmodium parasites. They also provide a stimulus for investigating the use of other bacteria which may inhibit plasmodium development as biocontrol tools, but may be more easily introduced into anopheles than wolbachia.
2. The mathematical models of wolbachia spread dynamics allow a realistic assessment of the likely impact of wolbachia-based strategies on pathogen transmission and the extent to which released wolbachia are likely to spread in local populations and which strains will be most effective for example the wolbachia-based control of the dengue vector mosquito aedes aegypti and aedes albopictus.
3. The project has delivered two new methodologies for the estimation of mean anopheles gambiae age in field populations. These tools, particularly the use of the CCM, provide a significant advance for the study of malaria epidemiology in Africa and facilitate the examination of differences in transmission associated with mean female lifespan in different seasons and under different ecological conditions. They will assist in the assessment of the impact of different interventions and assessment of effectiveness and for operational planning, of any novel vector control intervention that can be expected to impact vector population age structure. This particularly applies to any novel insecticides that act via residual dosage effect rather than immediately and to the use of entomopathogenic fungi.
4. The establishment of stable greenhouse malaria vector populations in Kenya viable over multiple generations and the associated protocols and guidelines, provide a tool for monitoring and evaluating the impact of innovative interventions - trials of modified / sterile mosquitoes, entomopathogenic fungi, or novel insecticides under strictly contained semi-field conditions.

Project website: http://www.zoo.ox.ac.uk/group/anopopage/

E-mail: steven.sinkins@zoo.ox.ac.uk