Targeted Mutagenesis in the Malaria Mosquito Using TALE Nucleases

Anopheles gambiae, the main mosquito vector of human malaria, is a challenging organism to manipulate genetically. As a consequence, reverse genetics studies in this disease vector have been largely limited to RNA interference experiments. Here, we report the targeted disruption of the immunity gene TEP1 using transgenic expression of Transcription-Activator Like Effector Nucleases (TALENs), and the isolation of several TEP1 mutant A. gambiae lines. These mutations inhibited protein production and rendered TEP1 mutants hypersusceptible to Plasmodium berghei. The TALEN technology opens up new avenues for genetic analysis in this disease vector and may offer novel biotechnology-based approaches for malaria control.


Introduction
Malaria is caused by Plasmodium parasites transmitted to their human hosts by the bite of anopheline mosquitoes. Malaria has been charged with causing more human deaths than any other disease in human history and continues to kill about 660,000 annually [1]. A reduction in the malaria death toll has been achieved thanks to vector control using insecticides and insecticide-impregnated bednets, better health care and progress in medical treatment, but this success is currently mitigated by the spread of resistance both of mosquitoes to insecticides and of Plasmodium to antimalarial drugs. Sequencing the genome of the main malaria vector, Anopheles gambiae [2], enabled the identification of hundreds of genes involved in the vector's capacity to transmit Plasmodium. Altering the mosquito genome in a way that abates Plasmodium transmission, through transgenesis or other sophisticated genetic engineering tools, can offer new perspectives in the fight against malaria. On one hand, experimentally altering mosquito genes of interest will advance our fundamental understanding of the biological interactions between mosquito and parasite, and may help target vulnerable points in the parasite cycle. On the other hand, the prospect of releasing engineered male mosquitoes to propagate malaria resistance genes through wild susceptible vector populations has been receiving increasing attention [3,4]. Novel methods to disrupt or alter target genes of interest in the malaria mosquito would promote rapid progress towards these goals. Furthermore, targeted genetic modifications that do not require the permanent introduction of transposons in the genome are particularly desirable, as they eliminate the potential risk of subsequent unplanned transposon mobilization by natural sources of transposition factors.
Recently, a novel class of DNA-binding protein domain derived from the Xanthomonas Transcription Activator Like Effector (TALE) proteins [5,6] has been successfully harnessed to custom-design sequence-specific endonucleases [7,8]. These TALE nucleases (TALENs), in which the TAL DNAbinding domain is fused to the FokI endonuclease domain, are easy to engineer (e.g., [9,10]), highly predictable in their sequence specificity, and highly mutagenic [11] making them an attractive alternative to Zinc Finger Nucleases that have less predictable binding specificities and require in vitro optimization [12]. Mutations arise by imprecise repair of the TALEN-generated double-stranded breaks by the nonhomologous end joining (NHEJ) repair pathway. In a number of animal species, injection of mRNA encoding TALENs have readily allowed researchers to generate mutants in their genes of interest [13][14][15][16][17][18][19]. Very recently, mutagenesis of an eye pigmentation gene was achieved in Aedes mosquitoes using this method [20]. In rice, disease resistant mutants have been produced by transgenic expression of the TALENs, whose respective transgenes were eliminated by subsequent genetic crosses once the desired mutation had been fixed [21]. We hypothesized that a similar approach would be applicable to insect vectors and set out to use TALENs to target the TEP1 gene, a key component of the mosquito immune system.

Results
Since the embryo microinjection procedure is technically challenging in Anopheles gambiae (as judged by poor survival and relatively low success rate of transgenesis in this species), we expected that mutant recovery after direct injection of TALEN-encoding plasmids or mRNA might be difficult. For this reason and to enable controlled mutagenesis experiments, we preferred transgenic expression of the TALENs in the mosquito germ cells. To obtain the proof-of-principle for gene targeting in A. gambiae via transgenic TALENs, we selected the wellcharacterized immune gene TEP1 as a target. TEP1, a protein similar to vertebrate complement factor C3, binds Plasmodium parasites as they invade the mosquito intestine and kills them in a manner probably dependent on its thioester site located in the C-terminus [22][23][24][25]. TEP1 mutants will be instrumental in further dissecting the antiparasitic complement-like system in mosquitoes.
TALENs function in pairs, each member of which binds a chosen 12 to 24-nucleotide sequence. The two selected target sequences are separated by 14-16 nucleotides, the optimal distance for the two FokI domains of the TALENs to properly dimerize and create a double-stranded break. Cleavage of the bound DNA molecule occurs near the center of the sequence separating the two target sites. We designed a single pair of TALENs to target a site within the TEP1 gene centered on an NcoI restriction site (5'-CCATGG-3'), offering the possibility to easily screen individual mosquitoes for mutations that destroyed the NcoI site ( Figure 1A). Mutations in this region are expected to strongly affect TEP1 protein function: frame-shifts resulting in premature stop codons would remove the Cterminal third of the protein including its thioester domain, while amino-acid deletions or insertions would alter the length of the alpha-helix connecting the CUB domain to the thioester domain [26], presumably resulting in destabilization of the protein's structure.
A distinct transgenic mosquito line was generated for each TALEN of the pair, the expression of which was driven by the A. gambiae Vasa promoter, active in the mosquito germline [27]. Therefore, F1 mosquitoes arising from a cross between the two lines will express both TALENs simultaneously and are expected to produce F2 gametes carrying mutations in the TEP1 gene. To screen individual larvae within the F2 progeny, we PCR-amplified a TEP1 fragment spanning the target site and subjected the amplification product to a restriction digest with NcoI ( Figure 1B). Of 310 screened F2 larvae, 16 (5.16%) carried a heterozygous mutation at the target locus, as evidenced by the appearance of a PCR product that NcoI was unable to cleave. Thus, at least 2.58% of TEP1 copies were mutated after exposure to one TALEN dose (i.e., one generation). This figure is a conservative estimate of mutation frequency, as we subsequently observed some mutations that left the NcoI recognition sequence intact. In order to obtain a mosquito population containing a higher frequency of mutations, we self-crossed successive generations of mosquitoes expressing both TALENs of the pair. This was facilitated by automated COPAS selection of larvae [28] that had inherited one copy each of the left and right TALEN genes, which are respectively associated with a red and yellow fluorescent marker expressed in the nervous system [29]. At least 1000 double-TALEN larvae were COPAS-selected and cultured for each generation. In the 7th generation (i.e., exposure to 6 TALEN mutagenic doses), 51% (49 out of 96) of the examined individual mosquitoes carried a heterozygous mutation in TEP1. This indicated that the frequency of mutations increased faster than predicted if mutations accumulated linearly from one generation to the next. This observation is consistent with a model where mosquitoes already carrying a heterozygous mutation employ homologous recombination to repair new TALEN-induced breaks in the wildtype chromosome, thereby effectively copying the existing mutation onto the newly damaged chromosome. This suggests that in addition to NHEJ, TALEN-caused breaks can also be repaired by homologous recombination. Therefore, the observed number of NHEJ mutations is an underestimation of the true rate of TALEN activity.
We wondered if a single TALEN of the pair is capable of causing mutations in TEP1. To investigate this, we sampled mosquito larvae from lines carrying a single TALEN maintained at high population levels for 8 (right TALEN) or 10 (left TALEN) generations, and again sampled larvae after about 16 generations. We expected that such a high number of generations would have allowed rare "monotalenic" mutational events to accumulate in the population. Out of 96 individual larvae tested by PCR for each sample, none carried a mutation in the TEP1 NcoI site. The observed absence of mutations among 576 haploid genomes exposed to single TALEN activity for 7, 9 or 15 generations suggests that single TALENs never or rarely induce mutations at their target site. To strengthen this point, we purified DNA from 2600 pooled larvae whose genomes had been exposed to one TALEN of the pair for 7 or 9 generations, and PCR-amplified the target region. PCR products appeared to be fully cleaved by NcoI, pointing to the absence of TALEN-induced mutations. To increase the chance of detection of a minor fraction of mutated products, we purified the region of the gel in which uncleaved PCR products may exist, cloned them into a plasmid, and examined E. coli transformants containing single copies of the PCR products. Again, all cloned fragments were cleaved by NcoI. Although we note that deep sequencing of amplicons would provide a more sensitive assay to detect rare mutations, this result further suggests that single TALENs rarely or never generate mutants. The apparent absence of single TALEN background activity was likely facilitated by our design scheme in which we used obligate heterodimeric FokI domains in TALEN construction [30,31].
The TALEN-generated TEP1 mutations were very similar in nature to mutations obtained in other organisms ( Figure 1C) and consisted mainly of small deletions and insertions (indels) that are the hallmark of imprecise NHEJ-mediated DNA repair. Some mutations deleted a number of nucleotides in multiples of three, resulting in the deletion of one to a few amino acids from the TEP1 protein. Other mutations introduced a frameshift in the TEP1 coding sequence, resulting in the loss of the entire C-terminal half of the protein, which contains features crucial for TEP1 function including the thioester domain. Using a PCR selection procedure from single legs taken from live mosquitoes, we recovered five homozygous mutant mosquito lines. These mutant lines were representative of the main classes of mutations: line M ∆T and M ∆MV are deletions of 1 (Threonine) and 2 (Methionine and Valine) amino-acids, respectively; line M ∆3+6 lacks 3 endogenous amino-acids but gained an insertion of 6 exogenous ones, line M ∆ct1 and M ∆ct2 are frame-shift mutations causing the loss of the entire protein C-terminus ( Figure 2A). Immunoblotting analysis failed to detect TEP1 in the Mut ∆ct1 and M ∆ct2 lines, while it revealed strongly reduced protein levels in M ∆T , M ∆MV and M ∆3+6 lines ( Figure 2B, top). Although small amounts of TEP1 protein could be detected in whole mosquito extracts of the M ∆T, M ∆MV and M ∆3+6 lines, these proteins did not undergo cleavage and were impaired in their secretion to the hemolymph, as no TEP1 signal was observed in immunoblotting of hemolymph samples ( Figure 2B, bottom). Therefore, the alpha helix connecting the CUB and thioester domains of TEP1 [26,32] seems to be very sensitive to insertion or deletion of single amino acids that destabilize the structure and prevent proper protein synthesis and secretion.
We next assessed the phenotype of these mutant lines in comparison to the parental lines by infection assays using Plasmodium berghei-infected mice (Figure 3). Similar to the RNA interference knockdown phenotype of TEP1 [22], all mutations in the homozygous state resulted in a dramatic increase in the number of developing parasites within the mosquito gut. These results confirm the pivotal role of TEP1 in antiparasitic responses. While it cannot be fully excluded that RNAi-mediated TEP1 knockdown may also affect genes sharing some nucleotide identity with the TEP1 sequence, such as other closely-related genes in the TEP family [33], mutations in TEP1 should leave the expression of other genes unaffected.
Interestingly, while midguts from TEP1 mutant mosquitoes often displayed impressive oocyst numbers that could exceed 1500, each experiment also yielded mutant midguts bearing no or only a few oocysts. Upon blood feeding on a single infected mouse, only well-gorged mosquito females were selected for subsequent dissection. It is therefore unlikely that some females ingested dramatically reduced numbers of P. berghei gametocytes. Thus, it is apparent that TEP1-independent mechanisms are at work to limit Plasmodium infection in a subset of mosquitoes.
To examine the effect of TEP1 dosage in the process of parasite killing, we compared oocyst infection levels between TEP1 heterozygous mutant, homozygous mutant and control mosquitoes (Figure 3d). The heterozygous mutant had an intermediate susceptibility phenotype, which was closer to, but significantly different from, the control. This suggests that the efficiency of parasite killing depends on TEP1 protein levels.
The antiparasitic role of TEP1 was discovered and characterized using RNA interference assays, in which synthetically produced double-stranded RNA homologous to a fragment of native TEP1 is injected at a high concentration in the body of adult mosquitoes. This technique was generalized for the functional characterization of hundreds of mosquito genes [34]. However, injection per se was reported to impinge on Plasmodium development by the potential induction of the wounding response [35]. Therefore the mutant TEP1 lines developed here will be useful in studies that must exclude the confounding effect of the injection procedure itself. To examine how the classical TEP1 RNAi phenotype compares with the mutant phenotype, we compared parasite loads in M ∆T to dsTEP1-injected mosquitoes of the parental line (Figure 3e). The mutant and RNAi phenotypes were very similar, validating a posteriori the high efficiency of RNAi knockdown.

Discussion
Here, we obtained proof of principle that TALENs can be used for targeted mutagenesis in the genome of the malaria mosquito, which is notably difficult to manipulate genetically. Recently, Aryan et al. [20] reported disruption of the eye pigmentation kmo gene in the dengue vector mosquito Aedes aegypti by injection of TALEN-encoding plasmid DNA. In the same species, successful mutagenesis of GFP and of the odorant receptor co-receptor (orco) was achieved by injecting mRNA encoding Zinc Finger Nucleases [36]. In both cases and in other animal systems for which TALEN mutagenesis has been reported, mutants were obtained directly upon injection of DNA or RNA into embryos. In contrast, we employed transgenically-expressed TALENs for this purpose. Besides the assurance of expressing the pair of TALENs in all germ cells, this offered the possibility to increase the frequency of mutant alleles in successive TALEN-expressing generations of mosquitoes. We disrupted the antiparasitic gene TEP1 as a first target; future work will make use of the obtained hypersusceptible mutant lines to further dissect the role of this important anti-malarial factor in parasite killing. Beyond the research field of mosquito immunity, this study paves the way for numerous other applications such as obtaining mutations in Anopheles genes considered to be essential for Plasmodium parasite development [37][38][39]. Disrupting these genes, or altering specific domains on the encoded proteins, may render homozygous mutant mosquitoes unable to support parasite development. Such parasite-refractory mutant mosquitoes could be used in anti-malaria intervention schemes, including vector population replacement in endemic regions. Unless a gene-drive strategy is specifically designed to spread desired mutations, this could be achieved by the repeated release of mass-produced male mosquitoes. Where knockout mutants in a given target gene might compromise the fitness of the mosquitoes, TALEN mutagenesis offers the possibility of a gene therapy to cure mosquitoes of malaria by selecting mutations that prevent a protein's interactions with parasite factors while preserving its other vital functions. TALEN mutagenesis could also be employed to knock out male fertility genes for use in the Sterile Insect Technique to reduce vector populations [40]. Of note, the TALEN transgenes used to obtain a desired mutation can subsequently be discarded by selection, thereby rendering the obtained homozygous mutant mosquitoes transgene-free. This could facilitate mosquito  release interventions that adhere to local regulations regarding genetically modified organisms.

Ethics statement
Our experimental protocols were approved by Comité de Qualification Institutionnel (CQI), the ethics evaluation committee of INSERM (IRB00003888, FWA00005831). Mosquitoes were reared and blood-fed on anesthetized mice in compliance with French and European laws on animal house procedures (agreement E67-482-2 of the Direction of Veterinary services of the French Ministry of Agriculture).

Assembling the TALENs
TALENs targeting the sites shown in Figure 1 were constructed by Golden Gate Cloning according to [9]. For TALEN assembly, we prepared two transgenesis-compatible destination vectors (annotated sequences provided in Text S1) encoding a Venus yellow fluorescent and a DsRed fluorescent transgenesis reporter gene, respectively. These vectors also contain a phage ϕC31 attB site for genomic integration into transgenic lines harboring attP sites. The first module used in Golden Gate assembly provided the Vasa promoter characterized in [27]. The last module closed the TALEN assembly with either a FokI-DD or a FokI-RR domain [30], designed for obligate heterodimerization of the two TALENs and codon-optimized for A. gambiae. The annotated sequence of the three plasmids is provided in Text S1. The TALEN C and N-terminal domains flanking the repeat region were identical to those used in [7].

A. gambiae lines and mosquito transgenesis
The TALEN-encoding vectors were inserted by ϕC31 integrase-mediated transgenesis [41] into the genome of A. gambiae lines X1 and X13, which are derived from laboratory strain G3. These two lines carry a PiggyBac transgene on chromosome II, containing an attP docking site. Individual transgenic larvae carrying the inserted left or right TALEN genes at the X1 or X13 attP site were identified by their red or yellow fluorescence, respectively. Resulting adults were crossed to their non-fluorescent parental line. In the F2 generation, fluorescent homozygous larvae were COPASselected [28] to establish stable TALEN-expressing lines.

Identification of TEP1 Mutants
In putative mutants (mosquitoes arising from parents expressing both TALENs), we PCR amplified a region of TEP1 spanning the TALEN target site using Phusion or Phire Polymerases (Thermo, Fisher) and primer 5'-TCAACTTGGACATCAACAAGAAGGCCGA-3' in combination with either 5'-GCATATCTTTGTGCCACACTTT-3' or 5'-GCCACCGTAACGAATTTCCA-3'. The PCR product was digested with NcoI (Fermentas), the recognition site of which is centrally located in the sequence cut by the TALENs. PCR products corresponding to mutant TEP1 alleles were not digested by NcoI. These PCR products were sequenced directly in the case of homozygous individuals, or cloned (CloneJET PCR Cloning Kit, Fermentas) and subsequently sequenced.

Mutagenesis and mutant line recovery
For mutagenesis experiments, the two TALEN lines were crossed. From the F2 generation onwards, 1000 mosquito larvae that inherited a single copy of each TALEN were COPAS-purified to initiate the next generation. To isolate TEP1 mutants, mosquitoes from this population were out-crossed to the parental line. The progeny carried a single TALEN and was therefore no longer subjected to TALEN mutagenesis. Among this progeny, we screened single adult mosquitoes by PCR on one leg using the Phire direct animal tissue PCR kit (Thermo, Fisher). The identified heterozygous mutants were individually crossed to the parental line. In the F2 progeny, we identified homozygous mutants by leg PCR and pooled them to start distinct homozygous mutant families.

RNAi and infection assays
RNAi silencing of TEP1 by double-stranded RNA injection into the thorax of adult mosquitoes and mosquito infections on mice carrying Plasmodium berghei GFP-con 259cl2 were performed as described [22]. To semi-automatically quantify the number of oocysts in photographs of dissected mosquito midguts, we used the "watershed segmentation" and "analyze particles" plugins of the ImageJ software after digital subtraction of the image background and smoothing of the signal. Oocyst counts obtained by this method are consistent with those obtained by manual counting of oocysts.

Immunoblotting
Whole-body mosquito extracts were obtained by grinding one adult female mosquito in 80 µl of protein sample buffer. The sample was denatured for 3 min at 95°C and centrifuged. 8 µl were loaded on 8% SDS-polyacrylamide gels. Hemolymph samples were prepared as described [42], 10 µl of the samples were loaded. Immunoblotting was performed using standard procedures [43] with rabbit polyclonal antibodies raised against the prophenoloxidase PPO 2 [44] or against the C-terminal half of TEP1 [45].

Supporting Information
Text S1. The nucleotide sequences of plasmids and building blocks used to construct the TALEN-expressing transgenesis vectors are provided. Building blocks not listed here are published in references 5,7. (DOCX)