FORUM

Wolbachia-Induced Mortality as a Mechanism to Modulate Pathogen Transmission by Vector Arthropods JASON L. RASGON,1 LINDA M. STYER,

AND

THOMAS W. SCOTT

Department of Entomology, University of California, Davis, CA 95616

J. Med. Entomol. 40(2): 125Ð132 (2003)

ABSTRACT Insecticide resistance and absence of clinical cures or vaccines for many vector-borne diseases has stimulated interest in using genetically modiÞed arthropod vectors for disease control. Current transgenic strategies focus on vector susceptibility to pathogen infection, which is an inefÞcient target for pathogen transmission interference. Manipulation of vector survival is theoretically more effective, resulting in larger reductions in the expected number of human infections. A hypothetical method to manipulate vector survival is to drive mortality-inducing Wolbachia into populations. For varying patterns and degrees of induced mortality, we outline the conditions under which virulent Wolbachia introductions into vector populations are expected to succeed and quantify the resultant reduction in pathogen transmission. The most critical component to the success of this strategy is the pattern of induced mortality. For operationally feasible introductions, induced mortality must be delayed until after vector reproduction begins. If this condition is not met, introduction thresholds become exceedingly high, ranging from ⬇40% to 90% of the total adult population. Delayed induced mortality patterns can reduce introduction thresholds to ⬇15Ð 45% of the total adult population. Reduction in cytoplasmic incompatibility with male age has negligible effects on introduction success regardless of the induced mortality pattern. Under proper circumstances, symbiont-induced manipulation of vector survival can theoretically result in up to 100% reduction in pathogen transmission, depending on Wolbachia parameters, magnitude and pattern of induced mortality, and duration of pathogen incubation in the vector. Our results indicate that a broadening of the current paradigm for genetic manipulation of vectors to parameters other than arthropod vector competence is justiÞed and will reveal new research possibilities for vector-borne disease control. KEY WORDS Wolbachia, popcorn, vector-borne disease, Culex pipiens, age-structure

EACH YEAR, BILLIONS of people worldwide are at risk of contracting vector-borne diseases (Beaty 2000). Concerns for insecticide resistance (Hemingway and Ranson 2000) and the lack of suitable vaccines and clinical cures for important diseases such as malaria and dengue (Beaty 2000) are stimulating the pursuit of novel disease control strategies based on genetic manipulation of arthropod vectors (Pettigrew and OÕNeill 1997). For the most part, these approaches focus on attempting to reduce arthropod vector competence, i.e., the proportion of vectors feeding on an infected host that imbibes and becomes capable of biologically transmitting a pathogen (Powers et al. 1996, Higgs et al. 1998, Ito et al. 2002). Vector susceptibility to pathogen infection was chosen based on the notion that because vector-borne pathogens are dependent on arthropods for transmission from one vertebrate host to the next, human infection and resultant disease can be curtailed by rendering arthropods genetically incapable of transmission (Ito et al. 2002). Although this argument is intuitively attractive, it is not well sup1

E-mail: [email protected].

ported theoretically or empirically. Theory indicates that vector susceptibility to pathogen infection is among the least important components of an arthropodÕs role in pathogen transmission (Spielman 1994). Unless the genetic block on vector pathogen susceptibility is close to 100% efÞcient under Þeld conditions, disease will not be eliminated and may not even be signiÞcantly reduced (Boete and Koella 2002). There are numerous examples under natural conditions when difÞcult-to-infect arthropods were efÞcient vectors because other aspects of their biology compensated for low levels of pathogen susceptibility (Miller et al. 1989, Walker et al. 1998, Mellor et al. 2000). Consequently, there are theoretical and empirical reasons to question the feasibility of a disease control strategy based on manipulation of arthropod susceptibility to pathogen infection. The most sensitive component of a vectorÕs role in pathogen transmission is its daily probability of survival (Garrett-Jones 1964). Disease control strategies that increase vector mortality are expected to be more efÞcient in reducing pathogen transmission than altering vector competence, because relatively small

0022-2585/03/0125Ð0132$04.00/0 䉷 2003 Entomological Society of America

126

JOURNAL OF MEDICAL ENTOMOLOGY

changes in daily survival can result in relatively large changes in the number of new vertebrate host infections. Despite this known relationship, transgenic disease control approaches focusing on vector survival have not been rigorously examined. For disease transmission to occur, a vector must imbibe a pathogen during bloodfeeding and survive until the pathogen can be biologically transmitted to a vertebrate host. This time period, known as the extrinsic incubation period (EIP), can vary from days to weeks depending on ambient temperature and details of the vector-parasite system. Effective vectors are those that survive long enough to transmit the pathogen. Introducing a trait into a vector population that shortens life-span would decrease both vector survival through the EIP and the expectation of infective life, i.e., the number of days a vector is expected to live after becoming infectious. These two linked effects would reduce the number of vectors that are physiologically capable of transmitting pathogens. To counteract the Þtness disadvantages conferred by a trait that increases mortality and to promote spread of the trait into the population, it will be necessary to link the trait to a factor that can actively drive it through the population to high frequency (Spielman 1994). Bacterial symbionts in the genus Wolbachia potentially fulÞll that requirement (Sinkins and OÕNeill 2000). Transovarially transmitted Wolbachia symbionts are widespread among invertebrate taxa and are associated with reproductive host effects including cytoplasmic incompatibility (CI) (Stouthamer et al. 1999). CI partially or completely sterilizes matings between infected males and uninfected females. Matings of infected females are fertile regardless of the infection status of the male. Because of the reproductive advantage conferred to infected females, CI can drive Wolbachia infection through host populations to high frequency despite Þtness costs (Turelli 1994, Turelli and Hoffmann 1999). In some systems, Wolbachia does not seem to confer major Þtness costs to its arthropod host (Hoffmann et al. 1998, Zchori-Fein et al. 2000). However, a pathogenic Wolbachia strain (denoted popcorn or wMelPop) has been shown to kill adult Drosophila melanogaster by over-replicating in the central nervous system. Average adult life span of infected ßies is approximately one-half that of uninfected ßies (Min and Benzer 1997). If popcorn or a popcorn-like Wolbachia strain were transferred into disease vectors such as mosquitoes, it might be possible to use CI to counteract the Þtness disadvantages conferred by infection and spread pathogenic symbionts through the population, thus reducing pathogen transmission by increasing vector mortality (Sinkins and OÕNeill 2000). Although this strategy has been discussed in a qualitative sense (Zimmer 2001), its feasibility has not been explored in a quantitative fashion. In this report we describe and explore, using a quantitative modeling framework, the conditions under which mortalityinducing Wolbachia introductions would be expected to succeed and quantify the reduction in pathogen

Vol. 40, no. 2

transmission expected to result from such an introduction. Materials and Methods Dynamics of mortality-inducing Wolbachia. An existing analytical model describing the dynamics of mortality-inducing Wolbachia (Fine 1978) assumes discrete generations and does not account for the effect of population age structure. This approach does not allow one to examine the effects of different induced patterns of mortality on Wolbachia dynamics and can underestimate the infection frequency that must be exceeded for Wolbachia to become established in the population; i.e., the introduction threshold. To address these issues, we developed an agestructured matrix model with overlapping generations to explore how different patterns and degrees of induced mortality affect Wolbachia introductions and spread. We chose for this analysis to examine pathogenic Wolbachia spread through populations of the mosquito Culex pipiens L. Cx. pipiens complex mosquitoes naturally harbor Wolbachia (Yen and Barr 1973) and are important vectors of Þlarial parasites and arboviruses (Savage et al. 1993, Krida et al. 1998, Campbell et al. 2000, Lindsay and Thomas 2000). Although Wolbachia infection dynamics are expected to vary according to the host species, our results will constitute a basis for comparison and have broad application to other vector-borne pathogen systems. We constructed a life table with 40 age classes: 10 preadult (egg, larvae, pupae) and 30 adult age classes based on Cx. pipiens development and maintenance at 25Ð27⬚C in our laboratory and from published reports (Hays and Hsi 1975). Age-speciÞc fecundity was estimated from published reports (Go´ mez et al. 1977) and scaled for 30 adult age classes (Fig. 1A). We assumed random mating in the population, a 1:1 sex ratio, and monogamous female mating (Kitzmiller and Laven 1958) of 3-d-old females with egg rafts Þrst appearing on day 6 of adult life (Weidhaas et al. 1971). Altering these life-table parameters may numerically alter our results but will not qualitatively change our conclusions. To simplify calculations, we assumed that males and females exhibit similar mortality patterns and that population size is constant and not regulated in a density-dependent manner; natural populations may not conform to these assumptions. For Wolbachia model parameters, we used previously published terminology (Turelli and Hoffmann 1999) where ␮ ⫽ percentage of uninfected offspring produced from an infected female (␮ ⫽ 0 if Wolbachia transmission is perfect), F ⫽ relative fecundity of infected versus uninfected females (F ⫽ 1 if Wolbachia has no effect on fecundity), H ⫽ relative hatch rate of an incompatible versus compatible cross (H ⫽ 0 if CI results in complete sterility), and Sh ⫽ (1 ⫺ H). We initially assumed that CI does not change with male age and altered this assumption in a later analysis (see below). Variables involved in construction of birth (Bx) and survival (px) parameters in the matrices (mx ⫽ age-speciÞc fecundity and px ⫽ age-speciÞc sur-

March 2003

RASGON ET AL.: MORTALITY-INDUCING Wolbachia

冢 冣冢 U

N 0,t N 1,t U N 2,t ⫽ ⯗ U N ␻,t U

⫻

127 U

N 0,t⫹1 ⫹ U U N 1,t⫹1 U N 2,t⫹1 ⯗ U N ␻,t⫹1

冣

[3]

where U

Bx ⫽ 共mx 兲共1 ⫺ Sh Wx 兲

[4]

Wx ⫽ Wolbachia infection frequency among all mating adults at the time the female in age class x mated (infection in premating adults is not relevant). This calculation only incorporates the reduction in uninfected eggs oviposited at time t ⫹ 1 due to CI. The number of uninfected eggs produced by infected mothers caused by imperfect transmission (U) must also be calculated for each age class x and summed over all age classes. Assuming that uninfected offspring produced by imperfect maternal transmission are equivalent to offspring produced by uninfected mothers, U takes the value U ⫽ 兺共mx 兲共I Nx,t 兲共␮F关1 ⫺ Wx 兴 ⫹ ␮FH关Wx 兴兲 Fig. 1. (A) Culex pipiens age-speciÞc fecundity (mx). E, egg; L, larvae; P, pupae; Pr, premating adult; MA, mating adult; Po, postmating adult; and OA, ovipositional adult. The model assumes a 1:1 sex ratio of offspring. (B) Age-dependent daily survival patterns for pathogenic Wolbachia-infected mosquitoes. All age-dependent patterns have the same initial survival (0.9) that declines to 0 after varying latent periods: A, 100% mortality at day 14 of adult life; B, 100% mortality at day 18 of adult life; C, 100% mortality at day 22 of adult life; D, 100% mortality at day 26 of adult life.

vival) came directly from Fig. 1 or as stated in the text (see Results). I and U refer to infected and uninfected, respectively (␻ ⫽ oldest age class). Nx,t ⫽ number of mosquitoes in age class x at time t.

冢

I

B0 p0 0 ⯗ 0

I

I

... ... ... ⯗ 0

B1 0 I p1 ⯗ 0

l

B␻ ⫺1 0 0 ⯗ I p␻ ⫺1

I

冣冢 冣冢 冣

B␻ 0 0 ⫻ ⯗ 0

I

N0,t N1,t I N2,t ⫽ ⯗ I N␻ ,t I

I

N0,t⫹1 N1,t⫹1 I N2,t⫹1 ⯗ I N␻ ,t⫹1 I

[1]

where I

Bx ⫽ 共mx 兲共1 ⫺ ␮兲 F

[2]

It is possible to input the age-speciÞc fecundity of Wolbachia-infected individuals directly into the lifetable and eliminate the F term. However, for purposes of comparison, it is more informative to include the term to estimate the relative effects of different fecundity levels. This approach assumes that Wolbachia affects fecundity of all age classes equally.

冢

U

B0 p0 0 ⯗ 0

U

U

B1 0 U p1 ⯗ 0

... ... ... ⯗ 0

l

B␻ ⫺1 0 0 ⯗ U p␻ ⫺1

U

B␻ 0 0 ⯗ 0

冣

[5]

The modeling framework outlined can be easily adapted to vary parameters in an age-dependent manner. One parameter where this may be of interest is the age-speciÞc expression of CI by infected males. There is laboratory evidence that CI expression can be modulated with male age in a number of host species (Singh et al. 1976, Turelli and Hoffmann 1995, Jamnongluk et al. 2000, Kittayapong et al. 2002). Assuming random mating, one can calculate the average CI level in the population (H*) as the sum of all age-speciÞc CI values (H) weighted by the age-speciÞc proportions of infected males in each age class (⌿x) at the time that females in age class x mated: H* ⫽ and

冘

Hx ⌿x

S*h ⫽ 1 ⫺ H*

[6] [7]

These parameters can be substituted for H and Sh in Equations 4 and 5. When we examined the effect of age-independent (constant) and age-dependent (dynamic) mortality patterns (Fig. 1B) on Wolbachia dynamics and pathogen transmission, we assumed that Wolbachia-infected individuals were initially released as 6-d-old males and gravid females into a population at a stable age distribution. In all cases, the daily probability of survival for uninfected mosquitoes was set at 0.9 (Macdonald et al. 1968). The Wolbachia introduction threshold was deÞned as the frequency of infected adults that must be exceeded for Wolbachia to successfully be established in the population (number released/total adult population including those released). The stable equilibrium level was deÞned as the infection frequency Wolbachia will reach in the adult population if it successfully becomes established. Introduction thresholds and equilibrium levels were calculated numerically.

JOURNAL OF MEDICAL ENTOMOLOGY

128

Mosquito survival and pathogen transmission. We developed a model to evaluate the vectorial capacity of a population (C) (expected number of new inoculations/infectious host/d, which is a measure of the pathogenÕs basic reproductive rate when considering only entomological components of transmission) (Garrett-Jones 1964) in an age-structured population where mosquitoes can exhibit age-independent or age-dependent mortality patterns. Total vectorial capacity (C) for a pathogen with EIP of n days (nCT) was calculated (for each age class) as C of Wolbachiainfected fraction of the population plus C of the uninfected fraction of the population (nCT ⫽ nCI ⫹ nCU) and summed over all age classes x such that

冋 冉写 冊 x⫹n

CI ⫽ 兺nCIx ⫽ 兺 共da2␤兲

n

册

pj 共Iex⫹n兲共⍀x兲共wx兲

I

j⫽x

[8]

and

冋 冉写 冊 x⫹n

CU ⫽ 兺nCUx ⫽ 兺 共da2␤兲

n

册

pj 共Uex⫹n兲共⍀x兲共I ⫺ wx兲

U

j⫽x

[9]

where I ⫽ Wolbachia-infected, U ⫽ Wolbachia-uninfected, n ⫽ pathogen EIP (days), x ⫽ age class when mosquito bites an infectious host, d ⫽ mosquito density (female mosquitoes/vertebrate host), a ⫽ mosquito biting rate (bites/host/d), ␤ ⫽ mosquito vector competence (the value [da2␤] is held constant for both Wolbachia-infected and -uninfected mosquitoes in this analysis and thus does not affect results), ⍀x ⫽ proportion of adult mosquitoes in age class x among all biting adults, wx ⫽ Wolbachia infection frequency x⫹n I,U in age class x, 冲j⫽x p j ⫽ product of the daily survival rates for Wolbachia-infected or -uninfected mosquitoes from the age that they Þrst bite an infectious host through the pathogen EIP (the probability of surviving the pathogen EIP), and I,Uex⫹n ⫽ infected or uninfected life expectancy at age x ⫹ n (a measure of how long the mosquito is expected to live once becoming infectious) as calculated by standard demographic techniques (Carey 1993). To estimate the reduction in pathogen transmission, we calculated the relative vectorial capacity after mortality-inducing Wolbachia have successfully invaded the population and reached a stable equilibrium frequency. The relative change in vectorial capacity before and after an intervention attempt gives an estimate of the relative treatment efÞcacy (Dye 1990) and is calculated as nCR ⫽ (nCT/nCO), where nCO ⫽ vectorial capacity before Wolbachia introduction, as calculated by our model when the population is at the stable age distribution determined by matrix projection (Carey 1993) and nCT is calculated as described above. For each invasion scenario, we determined the reduction in pathogen transmission for pathogens with EIP values ranging from 3 to 25 d.

Vol. 40, no. 2 Results

In general, mortality-inducing Wolbachia introductions become more difÞcult as the vertical transmission rate and CI decrease and Þtness (fecundity and mortality) effects increase. Introduction thresholds and equilibrium levels depend on degree and pattern of induced mortality, Wolbachia infection parameters (␮, H, F), life-stage of introduced infected insects, and population age structure at the time of introduction. Although there is no analytical expression that describes the range of Wolbachia parameter values over which invasions will be possible, limits can be approximated by numerical analysis. Simulations show that, with the assumptions we made in our analysis, transmission must be high (␮ ⱕ 0.1), CI strong (H ⱕ 0.2), and fecundity effects negligible (F ⱖ 0.9) for mortality-inducing Wolbachia to successfully invade a population, although the exact limits are ßexible and depend on the conditions outlined above, i.e., introductions with less ideal Wolbachia infection parameters are possible with lesser mortality effects, higher introduction levels, and/or alterations in the initial population age structure. For examples in this report, we used constant Wolbachia parameters at the midpoints of these general ranges (95% transmission [␮ ⫽ 0.05], 90% CI [H ⫽ 0.1], and 5% fecundity effect [F ⫽ 0.95]) and varied age-independent and agedependent Wolbachia-induced mortality. Induced mortality patterns were speculative except where noted. Age-independent induced mortality. We modeled age-independent induced reductions in daily survival from 0.9 to 0.86, in 0.01 increments. Reductions in daily survival ⱕ0.85 increased introduction thresholds to ⱖ94.4% of the total adult population (an infected to uninfected release ratio of ⱖ17:1). Releases at this level are too high to be operationally feasible for C. pipiens, and thus, were not considered further. Wolbachia that induce age-independent mortality are able to invade the host population, but introduction thresholds are exceedingly high. Introductions ranged from 41.2 to 88.7% of the total adult population (release ratio of 0.7:1Ð7.85:1) depending on the magnitude of the survival reduction. Equilibrium levels did not signiÞcantly change across induced mortality values (Fig. 2A). The relative vectorial capacity after invasion ranged from 0.90 to 0.32 of the initial level before introduction (10 Ð 68% control), depending on the magnitude of the survival reduction and pathogen EIP (Fig. 3A). Age-dependent induced mortality. We examined the dynamics of pathogenic Wolbachia that caused four different patterns of delayed elevated mortality. All had initial daily survival of 0.9 (similar to uninfected) that was constant for varying periods of time before rapidly killing the mosquito at adult age 14 (A), 18 (B), 22 (C), or 26 (D) days (Fig. 1B). The shape of the decline in survival was estimated in all cases from the survival pattern exhibited by popcorn-infected Drosophila (Min and Benzer 1997).

March 2003

RASGON ET AL.: MORTALITY-INDUCING Wolbachia

Fig. 2. Pathogenic Wolbachia introduction threshold levels and equilibrium frequencies. (A) Introduction thresholds and equilibrium frequencies for Wolbachia that induce ageindependent (constant) reductions in survival. [Introduction thresholds] and (equilibrium frequencies) for different daily probabilities of survival are 0.86 [0.887] (0.984); 0.87 [0.791] (0.988); 0.88 [0.637] (0.991); and 0.89 [0.412] (0.992). (B) [Introduction thresholds] and (equilibrium frequencies) for Wolbachia that induce age-dependent (dynamic) reductions in survival according to patterns shown in Fig. 1B: A, [0.45] (0.993); B, [0.239] (0.994); C, [0.171] (0.994); and D, [0.151] (0.994).

By allowing mosquitoes to reproduce before the onset of induced mortality, Wolbachia introduction thresholds are reduced to more manageable levels compared with when mortality is age-independent (Fig. 2B). Introduction thresholds ranged from 15.1 to 45% of the total adult population (release ratio of 0.18:1Ð 0.82:1), depending on the pattern of age-dependent induced mortality. Equilibrium frequencies were not signiÞcantly affected by induced mortality (Fig. 2B). The relative vectorial capacity after invasion ranged from 0.88 to 0.006 (12Ð⬎99% control), depending on the length of the induced mortality latent period and pathogen EIP (Fig. 3B). Attempts to control disease using pathogenic Wolbachia that induce age-dependent delayed mortality can result in signiÞcant control over a wide range of pathogen EIP values. Because induced mortality does not occur in eggs, larvae, or pupae, and is delayed until after early adult life, preovipositional mortality is minimized and results in comparatively lower, potentially feasible introduction thresholds (Fig. 2B). Male age and CI. Although CI reduction with male age in Cx. pipiens has been measured (Singh et al. 1976), the data are presented in such a way as to make parameter estimation difÞcult. We therefore used data estimated from Aedes albopictus infected with a single Wolbachia strain (Kittayapong et al. 2002). The data

129

Fig. 3. Reduction in vectorial capacity (C) across a range of pathogen extrinsic incubation periods (EIP) after invasion of pathogenic Wolbachia, expressed as percent of initial vectorial capacity before invasion. (A) Reduction in C after invasion of pathogenic Wolbachia that decrease daily probability of survival in an age-independent manner from 0.9 to 0.86 in 0.01 increments. (B) Reduction in C after invasion of pathogenic Wolbachia that reduce daily probability of survival according to age-dependent mortality patterns shown in Fig. 1B.

were estimated from a 60-d time series. We scaled the data for 30 d and obtained a linear regression plot (Fig. 4A). The predicted values of the regression were used to estimate changes in CI expression due to male age in our model. Reduced CI expression with male age has only negligible effects on the dynamics of mortality-inducing Wolbachia. For age-independentÐinduced mortality, the male age effect increases introduction thresholds 2Ð3% (Fig. 4B). Introduction thresholds are increased only 1Ð2% for age-dependent induced mortality patterns (Fig. 4C). Stable equilibrium levels are affected ⬍0.5% in all simulations. Discussion Although it is technically possible for Wolbachia that induce age-independent reductions in survival to invade populations, introduction thresholds are too high for practical purposes because of the cumulative effect of increased mortality before infected females can lay their Þrst raft of eggs. We expect that it will not be technically or economically feasible to exceed these threshold levels because of the unrealistically large numbers of infected adults that must be released, although multiple introduction attempts may serve to lower the magnitude of individual releases.

130

JOURNAL OF MEDICAL ENTOMOLOGY

Fig. 4. CI attenuation with male age. (A) Estimated increase in hatch rate in incompatible crosses as males age. Data were estimated from published reports and scaled for 30 adult age classes. (B and C) Effect of male age CI reduction on introduction thresholds for pathogenic Wolbachia that induce (B) age-independent and (C) age-dependent reductions in daily survival. Black circles represent introduction thresholds without male age CI attenuation (from Fig. 2); white circles represent introduction thresholds where CI follows the pattern in A.

To reduce Þtness costs associated with pathogenic Wolbachia infection, it would be more effective to delay the onset of elevated mortality until infected individuals have had an opportunity to reproduce. Interestingly, this is exactly the phenomenon observed in natural popcorn-infected DrosophilaÑ elevated mortality is delayed until adult ßies are ⬇6 Ð7 d old, which allows them to mate and oviposit before dying (Min and Benzer 1997). A similar mortality pattern in infected mosquitoes would allow for the transmission of infection to offspring and for the expression of CI, yet would kill many adults before they complete the EIP, decrease their expectation of infective life, and consequently reduce the number of vectors transmitting pathogens.

Vol. 40, no. 2

We demonstrated that reduction in CI expression with male age does not have a signiÞcant effect on the success of mortality-inducing Wolbachia introductions. Although attenuation of CI expression in older males can be severe, there are relatively few old males in an age-stratiÞed population. Because the majority of males are young, the average CI level is skewed toward the value for young males. To be conservative, we scaled the data to a 30-d time frame, giving us an estimate for CI modulation that was twice as severe as the original study indicated. Reduction of CI expression under Þeld conditions may be greater or less than we estimated; appropriate values will have to be empirically determined. It might be expected that selection pressures will result in the attenuation of induced mortality after virulent Wolbachia have been introduced into populations. Currently, there is no Þeld data available to address this issue. In the laboratory, the only known pathogenic Wolbachia strain (popcorn) retained the ability to induce host mortality through years of culture, presumably despite selection. In addition, popcorn continued to induce mortality 29 generations after transfection into D. simulans, despite the fact that the fecundity costs attributed to infection were attenuated in response to selection pressures (McGraw et al. 2002). Although these data are encouraging, it remains to be seen how mortality-inducing Wolbachia strains will behave under Þeld conditions. Surveys for natural Wolbachia infections that affect host life history traits and studies of popcorn/host interactions in both natural and transfected systems under Þeld conditions are necessary. The disease control strategy we outlined relies on a Wolbachia strain that induces both CI and causes reductions in arthropodÐ host survival. Popcorn does not cause CI in its natural D. melanogaster host (Min and Benzer 1997). However, it exhibits strong CI and induces mortality when transfected into D. simulans (McGraw et al. 2001, 2002) and is thus a potential candidate strain for future disease control efforts in mosquitoes. Alternatively, if the genetic basis for induced mortality can be understood, it may be possible to engineer a Wolbachia strain with the desirable mortality/CI phenotype. The success of artiÞcial Wolbachia transfers in a number of insect systems (Chang and Wade 1996, Van Meer and Stouthamer 1999, Sasaki and Ishikawa 2000) indicates that it may be feasible to transfer popcorn, or other mortality-inducing strains, to disease vectors where they would cause both mortality and CI. If the transfected strain does not cause CI in a novel mosquito background, it will be necessary to infect mosquitoes with both a mortality-inducing strain and a second CI-inducing Wolbachia strain. The modeling framework we developed can be adapted to examine the dynamics of a double Wolbachia infection, something that was beyond the scope of the current study. To prevent disassociation from the CI-causing Wolbachia and its elimination from the population, the mortality-inducing strain would have to be transmitted with 100% Þdelity (Turelli and Hoffmann 1999). The CI-inducing driver

March 2003

RASGON ET AL.: MORTALITY-INDUCING Wolbachia

strain can be transmitted with less than perfect Þdelity, because offspring that do not receive the driver strain will still be infected with the mortality-inducer and will be eliminated in subsequent generations because of CI expression. If a Wolbachia strain that causes both CI and elevated mortality can be engineered, the chance of disassociation would be reduced. Our analyses predict that mortality-inducing Wolbachia can be driven into a vector population and dramatically affect disease transmission dynamics. This idea is a signiÞcant departure from current transgenic strategies that aim to control disease by reducing arthropod vector competence and is theoretically more effective because of the nature of the relationship between vector survival and pathogen transmission (Garrett-Jones 1964, Spielman 1994). Understanding the pattern of induced mortality is essential for determining whether or not pathogenic Wolbachia can successfully invade vector populations, because it has a large inßuence on the magnitude of critical introduction thresholds. For introductions to be operationally feasible, induced mortality must be delayed until after the onset of mosquito reproduction. We expect that this caveat will hold for any attempt to reduce vector-borne disease by manipulating patterns of vector mortality by heritable symbionts or other methods. Broadening of the current paradigm for genetic manipulation of vectors is justiÞed by our results and would be beneÞcial. Expanding the focus for research on genetically modiÞed vectors to transmission parameters other than vector competence is likely to reveal new research possibilities for vector-borne disease control. Within the caveats we outlined, altering vector survival dynamics would be a powerful new tool in the Þght against vector-borne diseases. Acknowledgments We thank Michael Turelli and James R. Carey for helpful modeling advice and for reviewing earlier versions of this manuscript. This research was supported by the National Institutes of Health (grant GM-20092 to J.L.R.) and the University of California Mosquito Research Program.

References Cited Beaty, B. J. 2000. Genetic manipulation of vectors: a potential novel approach for control of vector-borne diseases. Proc. Natl. Acad. Sci. USA 97: 10295Ð10297. Boete, C., and J. C. Koella. 2002. A theoretical approach to predicting the success of genetic manipulation of malaria mosquitoes in malaria control. Malaria J. 1: 1Ð7. Campbell, G. L., C. S. Ceianu, C. S. Ceianu, and Savage, H. M. 2000. Epidemic West Nile encephalitis in Romania: waiting for history to repeat itself. Ann. NY Acad. Sci. 51: 94 Ð101. Carey, J. R. 1993. Applied demography for biologists with special emphasis on insects. Oxford, Oxford, UK. Chang, N. W., and M. J. Wade. 1996. An improved microinjection protocol for the transfer of Wolbachia pipientis between infected and uninfected strains of the ßour beetle Tribolium confusum. Can. J. Microbiol. 42: 711Ð714.

131

Dye, C. 1990. Epidemiological signiÞcance of vector-parasite interactions. Parasitology 101: 409 Ð 415. Fine, P.E.M. 1978. On the dynamics of symbiote-dependent cytoplasmic incompatibility in Culicine mosquitoes. J. Invertebr. Pathol. 30: 10 Ð18. Garrett-Jones, C. 1964. The human blood index of malaria vectors in relations to epidemiological assessment. Bull. WHO 30: 241Ð261. Go´ mez, C., J. E. Rabinovich, and C. E. Machado-Allison. 1977. Population analysis of Culex pipiens fatigans Wied. (Diptera: Culicidae) under laboratory conditions. J. Med. Entomol. 13: 453Ð 463. Hays, J., and B. P. Hsi. 1975. Interrelationships between selected meterologic phenomena and immature stages of Culex pipiens quinquefasciatus Say: study of an isolated population. J. Med. Entomol. 12: 299 Ð308. Hemingway, J., and H. Ranson. 2000. Insecticide resistance in insect vectors of human disease. Annu. Rev. Entomol. 45: 371Ð391. Higgs, S., J. O. Rayner, K. E. Olson, B. S. Davis, B. J. Beaty, and C.D.S. Blair. 1998. Engineered resistance in insect vectors of human disease. Am. J. Trop. Med. Hyg. 58: 663Ð 670. Hoffmann, A. A., M. Hercus, and H. Dagher. 1998. Population dynamics of the Wolbachia infection causing cytoplasmic icompatibility in Drosophila melanogaster. Genetics 148: 221Ð231. Ito, J., A. Ghosh, L. A. Moreira, E. A. Wimmer, and M. Jacobs-Lorena. 2002. Transgenic anopheline mosquitoes impaired in transmission of a malaria parasite. Nature (Lond.) 417: 452Ð 455. Jamnongluk, W., P. Kittayapong, K. J. Baisley, and S. L. O’Neill. 2000. Wolbachia infection and expression of cytoplasmic incompatibility in Armigeres subalbatus (Diptera: Culicidae). J. Med. Entomol. 37: 53Ð57. Kittayapong, P., P. Mongkalangoon, V. Baimai, and S. L. O’Neill. 2002. Host age effects and expression of cytoplasmic incompatibility in Þeld populations of Wolbachiasuperinfected Aedes albopictus. Heredity 88: 270 Ð274. Kitzmiller, J. B., and H. Laven. 1958. Tests for multiple fertilization in Culex mosquitoes by the use of genetic markers. Am. J. Hyg. 67: 207Ð213. Krida, G., A., B., F. Rodhain, and A. B. Failloux. 1998. Variability among Tunisian populations of Culex pipiens: genetic structure and susceptibility to a Þlarial parasite, Brugia pahangi. Parasitol. Res. 84: 139 Ð142. Lindsay, S. W., and C. J. Thomas. 2000. Mapping and estimating the population at risk from lymphatic Þlariasis in Africa. Trans. R. Soc. Trop. Med. Hyg. 94: 37Ð 45. Macdonald, W. W., A. Sebastian, and M. M. Tun. 1968. A mark-release-recapture experiment with Cules pipiens fatigans in the village of Okpo, Burma. Ann. Trop. Med. Parasitol. 62: 200 Ð209. McGraw, E. A., D. J. Merritt, J. N. Droller and S. L. O’Neill. 2001. Wolbachia-mediated sperm modiÞcation is dependent on host genotype in Drosophila. Proc. R. Soc. Lond. Ser. B Biol. Sci. 268: 2565Ð2570. McGraw, E. A., D. J. Merritt, J. N. Droller and S. L. O’Neill. 2002. Wolbachia density and virulence attenuation after transfer into a novel host. Proc. Natl. Acad. Sci. USA 99: 2918 Ð2923. Mellor, P. S., J. Boorman, and M. Baylis. 2000. Culicoides biting midges: their role as arbovirus vectors. Annu. Rev. Entomol. 45: 307Ð340. Miller, B. R., T. P. Monath, W. J. Tabachnick, and V. I. Ezike. 1989. Epidemic yellow fever caused by an incompetent mosquito vector. Trop. Med. Parasitol. 40: 396 Ð399.

132

JOURNAL OF MEDICAL ENTOMOLOGY

Min, K. T., and S. Benzer. 1997. Wolbachia, normally a symbiont of Drosophila, can be virulent, causing degeneration and early death. Proc. Natl. Acad. Sci. USA 94: 10792Ð 10796. Pettigrew, M. M. and S. L. O’Neill. 1997. Control of vectorborne disease by genetic manipulation of insect vectors: technological requirements and research priorities. Aust. J. Entomol. 36: 309 Ð317. Powers, A. M., K. I. Kamrud, K. E. Olson, S. Higgs, J. O. Carlson, and B. J. Beaty. 1996. Molecularly engineered resistance to California serogroup virus replication in mosquito cells and mosquitoes. Proc. Natl. Acad. Sci. USA 93: 4187Ð 4191. Sasaki, T., and I. Ishikawa. 2000. Transinfection of Wolbachia in the mediterranean ßour moth, Ephestia kuehniella, by embryonic microinjection. Heredity 85: 130 Ð135. Savage, H. M., G. C. Smith, C. G. Moore, C. J. Mitchell, M. Townsend, and A. A. Marfin. 1993. Entomologic investigations of an epidemic of St. Louis encephalitis in Pine Bluff, Arkansas, 1991. Am. J. Trop. Med. Hyg. 49: 38 Ð 45. Singh, K. R., C. F. Curtis, and B. S. Krishnamurthy. 1976. Partial loss of cytoplasmic incompatibility with age in males of Culex fatigans. Ann. Trop. Med. Parasitol. 70: 463Ð 466. Sinkins, S. P. and S. L. O’Neill. 2000. Wolbachia as a vehicle to modify insect populations, pp. 271Ð287. In A. M. Handler and A. A. James (eds.), Insect transgenesis: methods and applications. CRC, New York. Spielman, A. 1994. Why entomological antimalarial research should not focus on transgenic mosquitoes. Parasitol. Today 10: 374 Ð376. Stouthamer, R., J. A. Breeuwer, and G. D. Hurst. 1999. Wolbachia pipientis: microbial manipulator of arthropod reproduction. Annu. Rev. Microbiol. 53: 71Ð102.

Vol. 40, no. 2

Turelli, M. 1994. Evolution of incompatibility-inducing microbes and their hosts. Evolution 48: 1500 Ð1513. Turelli, M., and A. A. Hoffmann. 1995. Cytoplasmic incompatibility in Drosophila simulans: dynamics and parameter estimates from natural populations. Genetics 140: 1319 Ð 1338. Turelli, M., and A. A. Hoffmann. 1999. Microbe-induced cytoplasmic incompatibility as a mechanism for introducing transgenes into arthropod populations. Insect Mol. Biol. 8: 243Ð255. Van Meer, M. M., and R. Stouthamer. 1999. Cross-order transfer of Wolbachia from Muscidifurax uniraptor (Hymenoptera: Pteromalidae) to Drosophila simulans (Diptera: Drosophilidae). Heredity 82: 163Ð169. Walker, E. D., E. P. Torres, and R. T. Villanueva. 1998. Components of the vectorial capacity of Aedes poicilius for Wuchereria bancrofti in Sorsogon province, Philippines. Ann. Trop. Med. Parasitol. 92: 603Ð 614. Weidhaas, D. E., R. S. Patterson, C. S. Lafgren, and H. R. Ford. 1971. Bionomics of a population of Culex pipiens quinquefasciatus Say. Mosq. News 31: 177Ð182. Yen, J. H., and A. R. Barr. 1973. The etiological agent of cytoplasmic incompatibility in Culex pipiens. J. Invertebr. Pathol. 22: 242Ð250. Zchori-Fein, E., Y. Gottlieb, and M. Coll. 2000. Wolbachia density and host Þtness components in Muscidifurax uniraptor (Hymenoptera: pteromalidae). J. Invertebr. Pathol. 75: 267Ð272. Zimmer, C. 2001. Wolbachia: a tale of sex and survival. Science 292: 1093Ð1095. Received for publication 9 September 2002; accepted 13 November 2002.