Contributed by Anthony A. James, May 7, 2012 (sent for review January 28, 2012)
Anopheles stephensi mosquitoes expressing m1C3, m4B7, or m2A10 single-chain antibodies (scFvs) have significantly lower levels of infection compared to controls when challenged with Plasmodium falciparum, a human malaria pathogen. These scFvs are derived from antibodies specific to a parasite chitinase, the 25 kDa protein and the circumsporozoite protein, respectively. Transgenes comprising m2A10 in combination with either m1C3 or m4B7 were inserted into previously-characterized mosquito chromosomal “docking” sites using site-specific recombination. Transgene expression was evaluated at four different genomic locations and a docking site that permitted tissue- and sex-specific expression was researched further. Fitness studies of docking site and dual scFv transgene strains detected only one significant fitness cost: adult docking-site males displayed a late-onset reduction in survival. The m4B7/m2A10 mosquitoes challenged with P. falciparum had few or no sporozoites, the parasite stage infective to humans, in three of four experiments. No sporozoites were detected in m1C3/m2A10 mosquitoes in challenge experiments when both genes were induced at developmentally relevant times. These studies support the conclusion that expression of a single copy of a dual scFv transgene can completely inhibit parasite development without imposing a fitness cost on the mosquito.
Anopheles mosquitoes transmit malaria parasites among humans. Approximately 91% of the 216 million cases estimated to occur in 2010 were due to Plasmodium falciparum (1). Insecticide and parasite drug resistance continue to hinder malaria eradication efforts and innovative approaches to disease control are needed to complement traditional methods (2). One novel approach proposes the use of transgenic mosquitoes to introgress a parasite resistance gene into wild, malaria-susceptible mosquito populations, thereby interrupting transmission (3⇓–5).
Plasmodium falciparum must progress through several developmental stages within the mosquito before becoming infective to humans. Parasites enter the midgut as gametocytes during bloodfeeding. Gametocytes produce sexual forms, the male microgametes and female macrogametes, within the blood bolus, which then fuse to form zygotes. Zygotes mature into motile ookinetes that penetrate the peritrophic matrix surrounding the bloodmeal to reach the midgut epithelium. After traversing this tissue, the parasites rest beneath the basal lamina and form oocysts, within which thousands of sporozoites develop. Once matured, sporozoites exit oocysts and travel through the mosquito open circulatory system to reach the salivary glands from which they can be released during a subsequent blood meal.
Parasite resistance genes should be designed to encode products that inhibit parasite development without having major fitness effects on the mosquito host (3). Single-chain antibodies (scFvs) are promising candidates due to their specificity, efficacy, and small size. Transgenic An. stephensi expressing the scFvs m1C3, m4B7 or m2A10, produce significantly fewer parasites than controls when challenged with P. falciparum (6). The m1C3 and m4B7 scFvs were derived from monoclonal antibodies that bind the P. falciparum ookinete proteins Chitinase 1 and Pfs25, respectively. The m2A10 scFv binds the circumsporozoite protein (CSP), the predominant surface protein of sporozoites. An additional feature of m4B7 and m2A10 is the joining of the An. gambiae cecropin A peptide to the scFvs by a polypeptide linker. Cecropin has microbiocidal activity against both bacteria and Plasmodium species (7), and the resulting scFv-peptide proteins could exert both parasite-binding and antimicrobial activity.
We posited that a mosquito expressing two scFvs that target different P. falciparum life stages would completely inhibit parasite development. We tested this hypothesis using site-specific recombination to produce An. stephensi strains expressing dual transgenes comprising either m1C3 or m4B7 linked to m2A10. Anopheles gambiae carboxypeptidase A (AgCPA) gene regulatory elements were joined to m1C3 and m4B7 to direct bloodmeal-induced midgut expression of the scFvs. The An. stephensi Vitellogenin 1 (AsVg1) gene regulatory elements were used to direct bloodmeal-induced fat body expression of m2A10.
The φC31 integration system was selected for its ability to integrate large transgenes in a site-specific and irreversible manner (8⇓⇓–11). Transgene integration occurs when an attB site in the transgene-bearing plasmid recombines with an attP site (docking site) in the mosquito genome (12). Studies of the φC31 system in Drosophila melanogaster demonstrated that the strength of transgene expression in different tissues varies among docking sites (13). Anopheles stephensi, a vector of malaria in the Indian subcontinent, is transformed efficiently, facilitating the analyses of transgene expression in several genomic locations. The scFv transgenes were recombined into five An. stephensi recipient lines carrying one to three copies of the attP docking site, four of which were analyzed previously and shown to have no significant fitness load (14). A mutated φC31 integrase based an enzyme copy that showed increased catalytic activity and specificity in cultured human cells (10, 15) was evaluated alongside the wild-type enzyme. We hypothesized that the mutations in the amino acid sequence and the codon-optimization of the integrase gene would produce greater integration efficiency.
A docking site strain permissive of expression of two scFvs was identified and found to have only a minimal fitness cost. When compared with this recipient line, mosquitoes expressing scFvs displayed no fitness cost, and few or no sporozoites in the majority of P. falciparum challenge experiments. No sporozoites could be detected in experiments with mosquitoes expressing m1C3 and m2A10 at relevant developmental stages. These studies support the use of dual scFv transgenes as effector molecules in population replacement strategies to control malaria parasite transmission.
The pBacDsRed-m4B7/m2A10-attB plasmid was constructed using the AgCPA-m4B7 and AsVg1-m2A10 cassettes assembled previously (Fig. 1) (6). The pBacDsRed vector expresses DsRed, a fluorescent marker distinguished easily from the cyan fluorescent protein (CFP) expressed by recipient-line mosquitoes (16). An attB sequence inserted into the left-hand terminal repeat DNA of the piggyBac transposon (pBac LH) allows transgene recombination and insertion at the attP-containing docking site.
Schematic representation of integration of a dual single-chain antibody (scFv) transformation construct into an attP docking site transgene. (Top) the m4B7 scFv An. gambiae Cecropin A epitope-tag gene (E tag-m4B7-CecA) is flanked by An. gambiae carboxypeptidase A epitope-tag gene regulatory sequences (CP 5′, CP 3′). The m2A10 scFv An. gambiae Cecropin A gene (CecA-m2A10-E tag) is flanked by An. stephensi vitellogenin 1 gene regulatory sequences (VG 5′, VG 3′). Arrows denote the direction of transcription. The scFv transgenes are joined to the piggyBac DsRed plasmid containing the pUC18 vector, piggyBac Left- and Right-hands (LH, RH), and the DsRed marker of transformation (DsRed) flanked by the Pax3 promoter (3XP3) and the sv40 polyadenylation sequence. A bacterial attachment site (attB) is present within the LH. (Middle) The docking site transgene consists of piggyBac LH and RH sequences joined to a CFP transformation marker and a phage attachment site (attP) [adapted from Nimmo et al. (8)]. (Bottom) The φC31 integrase-catalyzed recombination of the attB and attP sites forms right and left attachment sites (attR and attL) and results in integration of the scFv plasmid.
The pBacDsRed-m4B7/m2A10-attB plasmid was microinjected with mutated φC31 integrase mRNA into embryos of docking-site lines attP 30, attP 43, and attP 44 (Table 1) (10). This plasmid also was microinjected into attP 20 and attP 19A embryos with wild-type integrase mRNA. All docking-site lines except attP 30 yielded recombinant offspring with attP 19A, 43, and 44 producing seven, two, and five lines, respectively. Individual DsRed-positive mosquitoes from lines containing multiple docking site transgenes (attP 19A, 43, 44) were outcrossed to wild-type (non-transgenic) mosquitoes to establish independent lines. DsRed-positive individuals from line attP 20, containing one docking site, were intercrossed to establish a single transgenic line.
Summary of results of pBacDsRed-scFv-attB plasmid microinjections into attP transgenic lines
Fluorescent hybridization in situ and gene amplification (inverse PCR) were used to characterize the docking-site insertions in DsRed-positive attP 44 individuals (Fig. S1). Hybridization of a CFP-specific probe to polytene chromosomes revealed three attP 44 docking sites, designated attP 44-A, attP 44-B, and attP 44-C, on the X chromosome. The genomic DNA sequences flanking the three docking sites were identified using inverse PCR.
Fluorescent hybridization in situ analyses based on colocalization of probes specific to m2A10 and a particular docking site detected integration exclusively in one of the available sites (attP 19A-B, attP 43-B, attP 44-C) of each line (Fig. 2 and Fig. S2). Colocalization of m2A10 and attP 20 probes confirmed a single, site-specific integration in this line. Consequently, one transgenic line of each genotype, attP 19A m4B7/m2A10, attP 20 m4B7/m2A10, attP 43 m4B7/m2A10, and attP 44 m4B7/m2A10, was established.
Fluorescent hybridization in situ of polytene chromosomes of attP 44 m4B7/m2A10 transgenic Anopheles stephensi. The four images presented display (A) the chromosome stain, (B) the Cy3-labeled m2A10 DNA probe, (C) the Cy5-labeled attP 44-C docking site DNA probe, and (D) a merged image. The attP docking site DNA probe has sequence complementary to the genomic DNA insertion site of the attP 44-C docking site transgene. Arrows point to probe signals.
Gene amplification (RT-PCR) was used to evaluate scFv transcript expression in the four attP m4B7/m2A10 lines at discrete times after a bloodmeal coinciding with the maximum expression of the endogenous gene from which the corresponding control DNA sequences were derived (17, 18) (Fig. 3). Previously-characterized scFv-expressing mosquitoes (6) and wild-type females were used as positive and negative controls, respectively. In addition, scFv expression was compared to that of the endogenous An. stephensi AsVg1 and AsCPA genes. The m4B7 transcripts were detected in all transgenic lines exclusively in midgut samples of RNA collected from females 4 h post bloodmeal (hPBM). The m2A10 expression evaluated in females at 24 hPBM detected transcripts only in lines attP 43 m4B7/m2A10 and attP 44 m4B7/m2A10. Neither m2A10 nor m4B7 transcripts were detected in transgenic males (Fig. S3).
Expression of m4B7 and m2A10 transcripts in attP m4B7/m2A10 females. (A) RT-PCR was used to detect m4B7 transcripts in RNA isolated from homogenates of dissected midguts (gut) and the remaining carcasses (car.) of four females 4 h post-bloodmeal (hPBM). Previously-characterized transgenic mosquitoes expressing m4B7 transcript (m4B7) (6) and wild-type (WT) mosquitoes were used as positive and negative controls, respectively. The m4B7 transcript accumulation was compared to that of the endogenous An. stephensi Carboxypeptidase-A (AsCPA). (B) RT-PCR was used to detect m2A10 transcripts in RNA isolated from homogenates of two to three females 24 hPBM. Previously-characterized transgenic mosquitoes expressing m2A10 transcript (m2A10) were used as a positive control (6). The m2A10 transcript accumulation was compared to that of endogenous An. stephensi vitellogenin (AsVg1). The An. stephensi S26 ribosomal protein transcript (rpS26) was amplified as a loading control in both m4B7 and m2A10 experiments. Parallel reactions in which the reverse transcription reaction step was omitted, assayed for genomic DNA contamination (rpS26-RT).
Immunoblot analyses with an antibody specific to the peptide epitope tag, E tag, were used to compare expression of m2A10 and m4B7 scFvs in different docking-site lines and a previously-characterized m2A10-expressing line (6) (Fig. 4). Although both scFvs contain the E tag, m4B7 expression is induced at an earlier post-bloodmeal time point, making it possible to distinguish the similarly-sized proteins. Consistent with RT-PCR findings, m2A10 expression was detected in attP 43 m4B7/m2A10 and attP 44 m4B7/m2A10 females at 12–72 hPBM. An immunoblot of attP 44 m4B7/m2A10 female samples confirmed that the scFv was secreted into the hemolymph. In contrast with RT-PCR findings, no signals corresponding to the predicted size of the m4B7 scFv were detected in any 4 hPBM samples.
Bloodmeal-induced scFv expression. (A) Immunoblot analyses of transgenic attP m4B7/m2A10 An. stephensi. An anti-E tag antibody was used to detect scFv protein in whole-body homogenates of transgenic females. A homogenate from a previously-characterized m2A10-expressing transgenic line was used as a positive control [m2A10 (6)] in each experiment. Non-bloodfed (NBF) females and bloodfed females 4, 12, 24, 48, and 72 h post-bloodmeal (hPBM) were examined. Wild-type females (WT) served as a negative control. Each protein sample was prepared from two whole females. The identity of each attP m4B7/m2A10 transgenic line is listed to the Left of each blot. (B) An immunoblot of a hemolymph preparation of ten 24 hPBM females from transgenic line attP 44 m4B7/m2A10. A whole-body homogenate from a previously-characterized m2A10-expressing transgenic line serves as a positive control [m2A10 (6)]. The identity of each sample is listed above the blot. Coomassie stain was used to visualize proteins present in the polyacrylamide gel following transfer (Right).
The pBacDsRed-m1C3/m2A10-attB transgene was produced by replacing the m4B7 gene of pBacDsRed-m4B7/m2A10-attB with m1C3 (Fig. S4). The attP 44 recipient line was chosen because it expressed consistently both m4B7 and m2A10 in the previous experiments. Three pools of G0 adults (P3, P7, P9) produced transgenic DsRed-positive larvae (Table 1). As the attP 44 recipient line contains three docking sites, Southern blotting was used to analyze transgene copy number and insertion site (Fig. S4). Genomic DNA isolated from two P3 and two P9 individual females was examined. Samples from four individual P7 females were analyzed because two distinct DsRed fluorescent larval phenotypes suggested the presence of multiple genotypes. A probe specific for m2A10 was used in combination with an NheI restriction enzyme digest to produce signals that would vary in size according to genomic location and number of independent insertions. Two fragments of approximately 8 and 10 kb in length were observed. The approximately 8 kb fragment also was present in attP 44 m4B7/m2A10, indicating transgene insertion in attP 44-C. PCR genotyping analysis indicated that the approximately 10 kb fragment represents transgene integration into attP 44-B. Among the three pools, transgene integration appeared in site B, site C, or both sites. Subsequent PCR genotyping analysis of 53 additional G1 P7 individuals identified three females containing the transgene integrated in docking site A, two of which had integrations in all three docking sites.
RT-PCR analyses to evaluate scFv expression resulting from integrations at site B, site C, or sites B and C, revealed both m1C3 and m2A10 transcripts in all lines (Fig. S5). Transgene expression from docking site A was not evaluated because lines were not established from these mosquitoes. An attP 44 m1C3/m2A10 line containing a single copy of the transgene integrated in attP 44-C was used for all further analyses to facilitate comparison with the attP 44 m4B7/m2A10 transgenic line. Immunoblot analyses of this line detected m2A10 expression in females 24–72 hPBM in a pattern similar to that of attP 44 m4B7/m2A10 mosquitoes. No protein corresponding to the predicted size of the m1C3 scFv was detected in females 4 hPBM.
Meiotic recombination was used to separate the attP 44-A and B docking sites from the attP 44, attP 44 m4B7/m2A10, and attP 44 m1C3/m2A10 transgenic lines prior to performing fitness analyses. This removed the possibility that these additional docking sites could influence these analyses. Hemizygous transgenic females were outcrossed to wild-type males and their progeny analyzed by gene amplification to select individuals containing only the attP 44-C docking site (Fig. S6).
Mosquitoes were examined for their survival in larval development, longevity as adults, ability to produce offspring, and wing length. Results and statistical analyses are presented in Table 2 and Table S1, respectively, and Fig. S7. The fitness impact of the attP 44-C transgene was evaluated by comparing attP 44-C and wild-type mosquitoes. The attP 44-C m4B7/m2A10 and attP 44-C m1C3/m2A10 mosquitoes were compared to attP 44-C to measure the additional impact of integration and expression of the scFv transgenes.
Fitness evaluation of wild-type, attP 44-C, attP 44-C m4B7/m2A10, and attP 44-C m1C3/m2A10 mosquitoes
Statistically significant differences between experimental and control samples were detected in only four experiments: larval to pupal development time (attP 44-C m4B7/m2A10 vs. attP 44-C), the percentage of females that laid no eggs (attP 44-C vs. wild-type), female longevity (attP 44-C m1C3/m2A10 vs. attP 44-C), and male longevity (attP 44-C vs. wild-type). A load was associated with transgene integration in only one instance: attP 44-C males had reduced daily survival after approximately 20 days and a lower median longevity of approximately 3 days than wild-type controls. All other significant differences reflected a benefit of transgene presence: comparisons of attP 44-C and wild-type mosquitoes indicated that fewer attP 44-C females laid no eggs, while comparisons of attP 44-C with scFv-expressing mosquitoes revealed that attP 44-C m1C3/m2A10 females had a higher median longevity and attP 44-C m4B7/m2A10 larvae had a reduced time to pupation.
Wild-type, attP 44-C, attP 44-C m4B7/m2A10, and attP 44-C m1C3/m2A10 mosquitoes were challenged with P. falciparum to assess parasite prevalence and numbers (Table 3). Mosquitoes in Experiments 1 and 2 were provided with uninfected bloodmeals on the 3rd, 8th, and 13th days post-infection (PI) to maintain expression of the bloodmeal-induced scFvs, and those in Experiments 3–5 received an additional bloodmeal on the 10th day PI. Accumulation of scFv transcripts was verified by RT-PCR in Experiment 3 (Fig. S8). The m4B7 and m1C3 transcripts had weak amplification signals in non-blood fed (NBF) females and stronger signals in 4 h PI females in identical reaction conditions. The m2A10 transcripts were not detected in NBF females, but had strong signals in females 24 h PI and day 14 PI (sampled 24 h PBM).
Oocyst and sporozoite prevalence and numbers in P. falciparum-challenged mosquitoes
The attP 44-C and attP 44-C m4B7/m2A10 females were homozygous for their respective transgene in all experiments. At the start of the challenge experiments, the elimination of non-essential attP docking sites had not yet been completed for the attP 44 m1C3/m2A10 transgenic line. As a consequence, the m1C3/m2A10 females in Experiments 1–3 were either hemizygous or homozygous for the scFv-expressing transgene and also carried the additional empty attP 44-A and B docking-site transgenes. Females in Experiments 4 and 5 did not carry the additional docking sites. Furthermore, the female population was ½ homozygous and ½ hemizygous in Experiment 4, while the female population in Experiment 5 was completely homozygous.
Oocyst infection of the midgut was evaluated 7–9 days PI, while sporozoite infection of the salivary glands (contained within head/thorax tissues) was evaluated 16–19 d PI. Oocyst prevalence (percentage of midguts infected) and mean intensities of infection (MII, average number of oocysts in infected midguts), and sporozoites per infected female (total number of sporozoites in head-thorax tissue pool/number of females in pool/oocyst prevalence) were assessed (Table 3). No consistent trends in oocyst prevalence were evident among controls and transgenic lines. The number of oocysts in attP 44-C mosquitoes did not differ significantly from wild-type controls (Mann–Whitney U test, two-tailed P > 0.05) except in Experiment 2 (Mann–Whitney U test, two-tailed P = 0.05). The number of oocysts in scFv-expressing mosquitoes also did not differ significantly from attP 44-C controls (Mann–Whitney U test, one-tailed P > 0.05).
Sporozoite infection results revealed striking differences between control and scFv-expressing mosquitoes. Averages of between 1,112 and 17,573 sporozoites per infected female were detected in attP 44-C and wild-type samples. In contrast, three of four attP 44-C m4B7/m2A10 experiments had samples that contained few (< 200) or no sporozoites (Experiments 3–5; Table 3). The exception (Experiment 1; Table 3), had an attP 44-C m4B7/m2A10 sample with a greater number of sporozoites than the attP 44-C controls. Notably, no sporozoites were detected in attP 44-C m1C3/m2A10 samples in any of the four experiments in which they were challenged (Experiments 2–5; Table 3).
The impact of m1C3 and m2A10 on parasite development was examined further in challenge experiments in which the number of uninfected bloodmeals following infection was varied (Table 4). Homozygous attP 44-C and attP 44-C m1C3/m2A10 females were provided 0–4 uninfected bloodmeals using the same bloodfeeding regimen as the previous challenge experiments. A low oocyst MII (< 3 oocysts/midgut) was observed in all mosquitoes in Experiment 6. Both oocyst prevalence and MII were lower in attP 44-C m1C3/m2A10 than control attP 44-C mosquitoes. Furthermore, every attP 44-C control sample contained on average 3,420–10,215 sporozoites/infected female, while no attP 44-C m1C3/m2A10 sample contained any sporozoites.
Oocyst and sporozoite prevalence and numbers in P. falciparum-infected mosquitoes provided 0–4 uninfected blood meals
Experiment 7 had a higher oocyst MII (> 8 oocysts/midgut) than Experiment 6 in attP 44-C controls, resulting in 10,035–20,422 sporozoites/infected female. Here again, reductions in oocyst prevalence and MIIs were seen in attP 44-C m1C3/m2A10 when compared with attP 44-C. Notably, attP 44-C m1C3/m2A10 females receiving zero or one uninfected bloodmeals following the initial infectious meal averaged as many as 6,163 sporozoites/infected female, while those receiving at least two uninfected bloodmeals contained no sporozoites. We interpret these results to indicate that the early induction of m1C3 and m2A10 is sufficient to prevent the development of sporozoites at low oocyst MII levels, but that subsequent production of m2A10 is essential at later times to block sporozoites resulting from high oocyst MII.
These studies support the conclusion that single copies of dual effector genes can abolish sporozoites in transgenic mosquitoes when expressed at relevant parasite developmental stages. The target of zero sporozoites in salivary glands was set as the most epidemiologically-relevant phenotype following experiments in an avian malaria model in which few parasites (< 10) were needed to infect the vertebrate host (19). The absence of sporozoites in mosquitoes carrying the transgenes tested here, in particular those comprising m1C3 and m2A10, meets this target. We propose that the m1C3/m2A10 dual transgene can be used in the further development of a population replacement strategy to control the transmission of the human pathogen, P. falciparum, by An. stephensi.
The φC31 system successfully recombined additional DNA into the majority of docking site lines. While used previously in Aedes aegypti and An. gambiae (10, 11), this is the first demonstration of this system in An. stephensi. The finding that some docking sites did not have insertions may reflect a low recombination efficiency for these sites. Alternatively, since the attP recipient lines were not necessarily homozygous for every docking site, and multiple sites in lines may not be linked, it is possible that some docking sites are present only in a portion of the embryos injected. The pBacDsRed-m1C3/m2A10-attB microinjection, which had twice as many surviving adults as the pBacDsRed-m4B7/m2A10-attB experiment, produced both transgene integrations at docking sites that had not been seen previously to recombine and integrations into multiple attP sites within the same individual mosquito.
Both wild-type and mutated φC31 integrases were capable of catalyzing the recombination-mediated insertion of a transgene. Since the two integrase genes were injected into different docking site lines, a precise comparison of their activity cannot be made. However, experiments with Ae. aegypti showed no significant differences in the efficiency of the mutated and wild-type integrase for inserting a transgene into the same site and our experiments are consistent with these findings (10). In addition, since no integration into pseudo-attP sites was observed, we conclude that the mutated enzyme preserves the site-specificity of the wild-type integrase (15).
Fitness analyses of attP 44-C vs. wild-type revealed no significant differences between experimental and control mosquitoes for the majority of tests. A significant cost was detected in the median lifespan of attP 44-C males. However, the difference in the daily survival of attP 44-C and wild-type males manifest late in their lifespan and it is unlikely that this would affect reproduction, as a study of An. stephensi males found that mating activity is maximal at 3–5 d post emergence (20). The finding that the attP 44-C line permitted expression of genes from three promoters (3XP3, AgCPA, AsVg1) and displayed only a minimal fitness cost supports the conclusion that it is a useful tool for integration and expression of different anti-plasmodial transgenes.
All mosquitoes in the attP 44-C vs. attP 44-C m1C3/m2A10 experiments displayed reduced survivorship when compared with those of the wild-type vs. attP 44-C and attP 44-C vs. attP 44-C m4B7/m2A10 experiments. The environmental and nutritional impact of providing sucrose instead of raisins to the attP 44-C and attP 44-C m1C3/m2A10 mosquitoes may have contributed to this result. Furthermore, the wild-type vs. attP 44-C and attP 44-C vs. attP 44-C m4B7/m2A10 survivorship experiments were performed a year prior to the attP 44-C vs. attP 44-C m1C3/m2A10 experiments. Fluctuations in the insectary environment may have impacted mosquito survival.
Fitness studies comparing the scFv transformed lines to the attP 44-C recipient line support the conclusion that expression of the transgenes does not confer a fitness load. The m1C3/m2A10 scFvs may even confer a fitness advantage, as the median lifespan of attP 44-C m1C3/m2A10 females was 4 days greater than that of attP 44-C controls. By comparison, analyses of transgenic mosquito lines with and without effector genes found the majority to be less fit than wild-type mosquitoes (21). For example, Akt-expressing transgenic An. stephensi, the only other Anopheles engineered to be completely resistant to P. falciparum, display a reduced average lifespan (22). This fitness cost may result from the role Akt has in mosquito signal transduction pathways. No fitness load was associated with a transgene-induced elevation of Rel2-responsive components of the endogenous mosquito immune system, but the resulting phenotype did not achieve complete elimination of the parasites (23). The m1C3, m4B7, and m2A10 all target parasite ligands, so they may be less likely to interfere with mosquito biology and confer a load. Furthermore, limiting transgene expression to the female midgut and fat body tissues also may minimize the fitness impact of the scFv transgenes. Experiments that place wild-type and transgenic mosquitoes in direct competition are needed to evaluate fitness parameters involved in mating competitiveness.
Transgenic lines that differ only in the effector gene can be compared to determine the relative transmission-blocking efficacy of each gene. Our data support the conclusion that m1C3 is more efficient than m4B7 in contributing to the inhibition of parasite development. The ability of attP 44-C m4B7/m2A10 females to inhibit parasite development varied among experiments, indicating that m4B7 and m2A10 partially blocked parasite development. However, the combination of m1C3 and m2A10 was effective in completely inhibiting sporozoite development in all experiments where m2A10 was induced one or more times after day 8 PI. The fact that the only genetic difference between attP 44-C m4B7/m2A10 and attP 44-C m1C3/m2A10 mosquitoes is the expression of different midgut scFvs supports the interpretation that m1C3 contributes to the “no sporozoite” phenotype. Importantly, only one copy of the transgene appears to be necessary for blocking, as the attP 44-C m1C3/m2A10 mosquitoes challenged in Experiment 4 were an equal mix of hemizygous and homozygous females.
The results of challenge Experiments 6 and 7 support the conclusion that scFvs target both early and late parasite stages. The scFvs induced by the infective bloodmeal were sufficient for complete inhibition of sporozoite development in the experiment with a low oocyst MII. In contrast, at higher oocyst MIIs, only females in which scFv expression was induced by bloodmeals on days 3 and 8 completely blocked infection. It is likely that these scFvs would effectively impair P. falciparum transmission in field conditions, as infected wild-caught An. gambiae often carry three or fewer oocysts (24).
Interestingly, neither scFv-expressing mosquito line reproduced the reduction in prevalence and mean intensity of oocyst infection observed in piggyBac-transformed mosquitoes expressing scFvs (6). This may be due to differences in the copy number and pattern of transgene expression. The m4B7 and m1C3 were present in 4 and 8 copies, respectively, per haploid genome in the previous work. It is possible that the single transgene copy number in the experiments reported here results in lower expression levels of m1C3 and m4B7. Expression analyses in which the transcripts encoding the AgCPA-regulated scFvs are seen, but no m1C3 or m4B7 proteins detected, are consistent with this interpretation. The scFvs may not be detected due to a lack of sensitivity of the immunoblot assay. The finding that RT-PCR detected different levels of m2A10 expression in the attP 43 m4B7/m2A10 and attP 44 m4B7/m2A10 lines, while the immunoblot displayed similar results for both lines, supports this interpretation.
If the m4B7 and m1C3 scFvs are indeed expressed in low quantities, their binding may lead to a delay or perturbation of oocyst development rather than complete inhibition of formation. We expect that oocysts are exposed to both midgut and hemolymph-expressed scFvs because diffusion of antibodies through mosquito tissues has been demonstrated previously (ingested anti-25 kD and anti-sporozoite antibodies were detected within oocysts) (25, 26). Should midgut-expressed scFvs impair oocyst development, such a phenotype may not be detected in the oocyst counts as they are performed here. Additional studies of attP 44-C m4B7/m2A10 and attP 44-C m1C3/m2A10 mosquitoes will determine whether the observed oocysts are viable. Furthermore, the absence of sporozoites in head and thoracic tissues of m1C3/m2A10 mosquitoes could result from m2A10 interactions with these parasites before or after their budding from the oocyst.
Future studies will reveal how scFvs inhibit parasite development. The m1C3 scFv was derived from the 1C3 monoclonal antibody, which was shown to bind the ookinete surface (27). Its transmission-blocking activity could be due either to its interaction with the ookinete surface or with its binding to the secreted chitinase enzyme. As CecA-scFv fusions, m4B7 and m2A10 could display both parasite binding and antimicrobial activity. Characterization of the mechanism by which these scFvs inhibit P. falciparum development could help guide future effector molecule strategies.
Testing additional scFv-promoter combinations could expand the repertoire of P. falciparum resistance transgenes. For example, the promoter of the An. gambiae G12 gene may promote higher levels of expression of midgut-stage scFvs than AgCPA (28). The PfNPNA-1 scFv, which is based on a human recombinant monoclonal antibody that recognizes CSP, could be tested as an alternative to m2A10 (29). The scFv strategy also could be adapted for use in other species of mosquitoes, such as An. gambiae, or with other human malaria parasites, such as P. vivax. If coupled with a mechanism for gene spread, scFv-expressing, malaria-resistance transgenes could become a self-sustaining disease control tool.
A colony of Anopheles stephensi (gift of M. Jacobs-Lorena, John Hopkins University) bred in our insectary for > 5 years was used in the experiments. The mosquitoes were maintained at 27 °C with 77% humidity and 12 h day/night, 30 min dusk/dawn lighting cycle. Larvae were fed a diet of powdered fish food (Tetramin, Melle, Germany) mixed with yeast. Adults were provided water and either raisins or a 10% sucrose solution ad libitum. Transgenic and wild-type (non-transgenic) mosquitoes used in parasite challenge experiments were reared in parallel using standardized insectary procedures. Mosquitoes were maintained at 26 °C ± 1 °C with 70–75% humidity and 12 hr day/night.
Table S2 lists oligonucleotide primer names and sequences used for gene amplification.
The pBacDsRed-m4B7/m2A10-attB plasmid was produced in three cloning steps. First, the AgCP5′-m4B7-AgCP3′ sequence from the previously described pSLFA-AgCP5′-m4B7-AgCP3′ plasmid was inserted into pBacDsRed (6, 16). The AgCP5′-m4B7-AgCP3′ insert was joined at the 5′-end by a Kpn I/Asc I blunt ligation and at the 3′-end by an Fse I ligation. An additional component of this cloning step was to digest the pSLFA-AgCP5′-m4B7-AgCP3′ plasmid with Xmn I to cut and remove the unnecessary pSLFA vector. The AsVg5′-m2A10-AsVg3′ was excised from the previously-described pSLfa-AsVg5′-m2A10-AsVg3′ plasmid and joined to the pBacDsRed-AgCP5′-m4B7-AgCP3′ plasmid at an Fse I site (6). Amplification of the attB sequence from the plasmid pTA-attB (30) using primers containing the Sph I recognition sequence, pCRattBFOR and pCRattBREV, allowed for digestion of the purified PCR product with Sph I and ligation into the piggyBac LH region of the pBacDsRed-AgCP5′-m4B7-AgCP3′-AsVg3′-m2A10-AsVg5′ plasmid. The name of this final plasmid is abbreviated as pBacDsRed-m4B7/m2A10-attB. The pBacDsRed-m1C3/m2A10-attB plasmid was produced by digesting both the pBacDsRed-m4B7/m2A10-attB and the previously-described pBacEGFP-AgCP5′-m1C3-AgCP3′ plasmid with Asi SI and Nhe I (6). The subsequent ligation allowed for replacement of m4B7 sequence with that of m1C3.
Microinjection was performed using in vitro-transcribed integrase as described previously (6, 8). Embryos were injected with a solution containing 300 ng/μl plasmid DNA and 400 ng/μl integrase mRNA. G0 males and females were outcrossed to wild-type mosquitoes in pools of approximately 5 G0 males or 15–30 G0 females. G1 progeny were screened as larvae for DsRed fluorescence under UV-fluorescence microscopy.
Southern blotting and hybridization techniques, performed as described previously (31), were used to detect transgene integration. The m2A10 probe was generated by restriction enzyme digestion of the pBSK-m2A10 plasmid (6) with BamHI and BstBI and purification from an agarose gel. Probes were labeled with 32P using the Megaprime DNA labeling system (Amersham).
The sequence of genomic DNA flanking the three docking sites in line attP 44 was isolated through inverse PCR as described previously (8, 14). The piggyBac3REV primer used in these experiments contained one base pair difference from the published sequence (Table S2). Genomic DNA from transgenic individuals was digested either with Taq I, for amplification of the 5′-end flanking sequence, or Hae III for amplification of the 3′ flanking sequence. Digestion with Hha I was used for isolation of the 3′-end flanking sequence of attP 44-C.
A DNA probe complementary to the m2A10 gene, amplified using m2A10 FOR and REV primers, was labeled with Cy3-AP3-deoxyuridine triphosphate (dUTP) (GE Healthcare) using the Random Primers DNA Labeling System (Invitrogen, Carlsbad, CA, USA). Probes specific to each copy of the attP docking site were labeled with Cy5-AP3-dUTP (GE Healthcare). The primer pairs 3endREV5′ and L19ARendBnew, L43endBFOR2 and L43endBREV1, and attP44-C-5′ and attP44-C-3′ were used to produce probes specific for the attP 19A-B, attP 43-B, and attP 44-C docking sites, respectively. Amplification of all three probes was completed using An. stephensi wild-type genomic DNA as template. The probe specific for attP 20 was produced by digesting a plasmid containing the attP 20 insertion site sequence in a pCR4-TOPO vector with Afl II (14). The cy5-labeled CFP probe was synthesized from a 320 bp DNA fragment, amplified previously using ECFPFOR and ECFPREV primers and RNA extracted from attP transgenic mosquitoes. Fluorescent hybridization in situ was performed as described (14).
Primers based on the attP 44 A, B, and C insertion site sequences were used to genotype individual females for copy number. The attP44-A-5′, attP44-B-5′, and attP44-C-5′ primers were mixed with piggyBac5REV to identify individuals containing copies of the attP docking site transgene in sites A, B, or C. Integration of the m1C3/m2A10 transgene into attP 44 sites A, B, or C was detected using the primer pCRattBREV in combination with attP44-A-3′, attP44-B-3′, or attP44-C-3′.
Total RNA was isolated using Trizol (Invitrogen) and treated with DNAseI (Promega). Gene-specific primers and a OneStep RT-PCR Kit (Qiagen) were used for amplification of diagnostic products from m1C3, m4B7, m2A10, AsCPA, AsVg1, or rpS26 transcripts (Table S2) (17). The annealing temperatures for these reactions were 59, 56, 67, 59, 67, and 67 °C, respectively. Four parallel RT-PCR reactions containing equal quantities of RNA were prepared for each analysis of scFv expression: scFv, AsCPA or AsVg1, rpS26, and rpS26-RT. The rpS26-RT reaction tested for the presence of genomic DNA contamination using rpS26 gene primers but omitting the reverse transcription step. Multiple biological replicates (≥2) were performed for selected time points for each of the RT-PCR series of experiments. Plasmodium falciparum-infected mosquitoes were analyzed using the MMLV Reverse Transcription reagents (Invitrogen) and goTaqDNA polymerase (Promega). The annealing temperatures for the m4B7 and m1C3 reactions were 54 and 59.5 °C, respectively.
Immunoblot analyses were performed as described previously (6).
Plasmodium falciparum gametocytes from the NF54 isolate were produced by automated culture (32). Culture conditions and preparation for membrane feeding were as described previously (33). Female mosquitoes were 1–11 days post-emergence at the time of infection. Females were allowed to feed in darkness for 15 minutes. Un-engorged females were removed and discarded. The remaining mosquitoes were provided a 10% sucrose solution containing 0.05% p-aminobenzoic acid. An uninfected blood meal was provided on the 3rd, 8th, and 13th days PI. In experiments 3–5, an additional uninfected bloodmeal was provided on the 10th day PI. Un-engorged females were removed after each of the uninfected bloodmeals. Midguts were dissected 7–9 days PI and stained with bromo-fluorescein for detection of oocysts. Sporozoites were isolated from dissected head/thorax tissues 16–17 days PI using a modified Ozaki method (34). Briefly, pooled dissected head/thorax tissues in RPMI 1640 medium were centrifuged on a glass wool column to separate sporozoites for detection in a chamber slide with a counting grid. GraphPad Prism software was used to perform a Mann-Whitney U test of oocyst infection data. One-tailed and two-tailed P values are listed for scFv-expressing vs. attP 44-C and attP 44-C vs. WT comparisons, respectively.
The mosquitoes evaluated in these experiments were the product of crossing approximately 125 virgin hemizygous transgenic females and 25 wild-type males. In this way, 50% of the offspring of these pairings inherit the experimental genotype and 50% inherit the control genotype. Evaluation of all experimental and control replicates for each fitness parameter was performed simultaneously. However, the attP 44-C vs. wild-type, attP 44-C m4B7/m2A10 vs. attP 44-C, and attP 44-C m1C3/m2A10 vs. attP 44-C experiments were performed at different times. Insectary procedures were standardized in an effort to control for the impact of manipulations and/or environmental conditions upon fitness evaluations. All adults evaluated in the attP 44-C vs. attP 44-C m1C3/m2A10 experiments were provided a 0.3 M sucrose solution instead of damp cotton balls and raisins.
For each experiment, 1,200 larvae were distributed, 50 larvae/pan, into pans (29 × 20 × 8.5 cm) containing 400 ml of ddH2O and 0.6 ml food slurry. At the time of hatching, an additional sample of approximately 100 larvae were examined microscopically to assess fluorescence phenotype, then discarded. Larvae were examined microscopically at day 5 post-hatching, counted, sorted according to fluorescence phenotype, and replaced in the pans. Pupae were removed and counted each day until there were no surviving individuals.
Fifty experimental or control mosquitoes aged 0–4 days post-emergence were placed in separate cages. Dead mosquitoes were removed and recorded daily until all died.
One experimental or control male was mated to five virgin wild-type females. Males were 0–2 days post emergence and females were 3–4 days post-emergence when placed together in a cage. Females were provided a bloodmeal three days after being placed with the males. Egg cups were provided two days after blood feeding, examined daily for the presence of larvae and categorized as either producing or not producing larvae.
For each treatment, approximately 100 virgin females, aged 0–3 days post-emergence, and 20 wild-type males, aged 0–4 days post-emergence, were combined in a cage. Females were provided a bloodmeal 4–5 days after being placed with the males. Two days after the bloodmeal, females were put individually into plastic vials lined with damp filter paper. Two days later, eggs were counted and placed into pans of water containing food slurry. Larvae were removed and counted daily for the following three days. The number of eggs laid by each female was recorded as an index of fecundity. Females that laid no eggs were not included in this calculation. The number of larvae hatching in each pan was recorded as a measure of fertility.
Wings were measured from the axial incision to the intersection of the R 4 + 5 vein and the margin, not including the fringe of scales.
GraphPad Prism software was used to perform statistical analyses of fitness experiments. A Mann–Whitney U test was used to calculate a two-tailed P value for both fecundity and egg hatchability data. A chi-square test was used to evaluate the number of females that laid no eggs and the number of larvae of each phenotype at hatching and 5 days post-hatching. A two-tailed P value produced by an unpaired t-test with Welch′s correction is reported for analyses of larval development time. A two-tailed P value produced by an unpaired t-test is reported for analyses of wing length. Adult survivorship curves were generated using the Kaplan-Meier method and compared using a log-rank test.
The authors are grateful to Aniko Fazekas, Rebeca Juarez, and Trung Dinh for mosquito husbandry. Research was supported by grants from the NIH NIAID (AI29746) to A.A.J. and from the Institut Pasteur to C.B. In part, A.I. was supported by a Chateaubriand pre-doctoral fellowship from the Embassy of France in the United States.
Author contributions: A.T.I., N.J., I.T., A.Z., C.B., and A.A.J. designed research; A.T.I., N.J., M.T., I.T., A.Z., and C.B. performed research; A.T.I. contributed new reagents/analytic tools; A.T.I. and A.A.J. analyzed data; and A.T.I., C.B., and A.A.J. wrote the paper.
The authors declare no conflict of interest.
See Author Summary on page 11070 (volume 109, number 28).
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1207738109/-/DCSupplemental.
Freely available online through the PNAS open access option.
