Semaphorin-1a Is Required for Aedes aegypti Embryonic Nerve Cord Development

Although mosquito genome projects have uncovered orthologues of many known developmental regulatory genes, extremely little is known about mosquito development. In this study, the role of semaphorin-1a (sema1a) was investigated during vector mosquito embryonic ventral nerve cord development. Expression of sema1a and the plexin A (plexA) receptor are detected in the embryonic ventral nerve cords of Aedes aegypti (dengue vector) and Anopheles gambiae (malaria vector), suggesting that Sema1a signaling may regulate mosquito nervous system development. Analysis of sema1a function was investigated through siRNA-mediated knockdown in A. aegypti embryos. Knockdown of sema1a during A. aegypti development results in a number of nerve cord phenotypes, including thinning, breakage, and occasional fusion of the longitudinal connectives, thin or absent commissures, and general distortion of the nerve cord. Although analysis of Drosophila melanogaster sema1a loss-of-function mutants uncovered many similar phenotypes, aspects of the longitudinal phenotypes differed between D. melanogaster and A. aegypti. The results of this investigation suggest that Sema1a is required for development of the insect ventral nerve cord, but that the developmental roles of this guidance molecule have diverged in dipteran insects.


Introduction
Mosquito genome projects [1,2,3] have revealed orthologues of many genes that are known to regulate development of D. melanogaster, a well-characterized genetic model organism. Although characterization of the function of these mosquito genes could provide insight into the evolution of insect development or potentially reveal novel strategies for vector control, extremely little is known about mosquito development [4,5]. Studying the development of mosquitoes has proven to be technically challenging, and although excellent descriptive analyses of A. aegypti embryogenesis exist [6,7], expression of only a handful of mosquito embryonic genes has been described in A. aegypti or other vector mosquito species [8,9,10,11,12,13,14,15,16]. Given the importance of studying the biology of A. aegypti [4,17], we recently published a series of protocols for analysis of the development of this mosquito [18,19,20,21,22]. These methodologies, in addition to those published previously [8,10], are promoting analysis of mosquito developmental genetics.
A major research goal in our laboratory is to investigate the function of genes that regulate arthropod nervous system development. Analysis of genes that regulate mosquito nervous system development will promote a better understanding of the developmental basis of motor function, sensory processing, and behavior, key aspects of mosquito host location. The appearance of the mature ventral embryonic nerve cords of A. aegypti, A. gambiae, and D. melanogaster, as well as many other insects and crustaceans, are markedly similar [15,16,23]. Despite these similarities, differences in ventral nerve cord axonogenesis [15,23] have been described. For example, in the case of the brine shrimp Artemia franciscana, the mechanism for pioneering the ventral nerve cord is different than what has been observed in Drosophila and other insects, and temporal changes in the expression of the axon guidance molecule Netrin parallel these differences [23]. Furthermore, although early axonogenesis in A. aegypti embryos is similar to that of D. melanogaster, our recent studies in this vector mosquito suggest that the function of Netrin signaling during embryonic ventral nerve cord development has evolved in insects [16]. siRNA-mediated knockdown of the A. aegypti frazzled (Aae fra) gene, which encodes a Netrin receptor, suggests that the developmental mechanisms responsible for regulating axon guidance in the embryonic nerve cord may have diverged among insects. Here, we examine the function of a second axon guidance molecule, Semaphorin-1a (Sema1a), in vector mosquitoes.
Sema1a, a member of the Semaphorin family of axon guidance molecules, is expressed in the developing central nervous system (CNS) of Drosophila [24]. During nervous system development, Semas guide neuronal growth cones by acting as attractants or repellents [25,26]. In the fruit fly, Sema1a signals through the PlexA receptor and is critical for proper CNS formation and motor axon guidance, as evidenced by defasciculation abnormalities noted in mutant embryos [26,27,28]. Sema1a also has specific targeting functions in the developing Drosophila olfactory system [29,30] and regulates photoreceptor axon guidance in the fruit fly visual system [31]. Furthermore, recent studies indicate that Sema1a can promote growth during Drosophila development through induction of key cellular growth regulators [32]. Orthologues of Sema1a were identified in both the A. aegypti and A. gambiae genome projects [1,2], and it is hypothesized that Sema1a signaling will function in the development of the mosquito embryonic nervous system. In this investigation, we examine expression of sema1a in two mosquito species and investigate the impact of knocking down sema1a during A. aegypti embryonic nervous system development.

Results and Discussion
Expression of sema1a and plexA in vector mosquito embryos Orthologues of sema1a and plexA were identified in both the A. aegypti [2] and A. gambiae [1] genome projects. Orthology assignments were confirmed through phylogenetic analyses which are presented in Fig. S1. Expression of A. aegypti sema1a (Aae sema1a) and A. gambiae sema1a (Aga sema1a) were analyzed through whole-mount in situ hybridization at the onset of nerve cord development in both species. Aae sema1a expression initiates in most developing neurons of the CNS just prior to establishment of the axonal scaffold and is maintained during ventral nerve cord formation (Fig. 1A,C). Comparable sema1a expression patterns are detected in the developing nervous system of A. gambiae (Fig. 1B). A similar sema1a pattern of expression was previously reported in the developing CNS of D. melanogaster [24]. In Drosophila, Sema1a signals through the PlexA receptor [26,27]. The expression patterns of plexA in both A. aegypti (Fig. S2A,C) and A. gambiae (Fig. S2B) are comparable to those of sema1a in both mosquito species (Fig. 1) and to the published patterns of Drosophila plexA expression [27]. These data are consistent with the hypothesis that Sema1a signaling regulates embryonic nervous system development in mosquitoes.
siRNA-mediated knockdown of sema1a during A. aegypti development We next functionally tested the hypothesis that sema1a expression is required for proper development of the ventral nerve cord in mosquitoes. This was accomplished through knockdown of the A. aegypti sema1a gene through use of RNAi, which was recently shown to be an effective method of inhibiting gene function during embryonic development of A. aegypti [16,20]. Two siRNAs corresponding to different regions of the Aae sema1a gene, siRNA 890 and siRNA 1198 , were used to target sema1a. Control experiments were completed using a control siRNA that has no sequence homology in the A. aegypti genome.
siRNAs were injected pre-cellular blastoderm, and Aae sema1a knockdown was confirmed through both quantitative real-time PCR (qRT-PCR) and whole-mount in situ hybridization. Multiple qRT-PCR replicates at three different time points, including 24, 48, and 72 hrs., confirmed knockdown throughout embryogenesis following injection of either siRNA 890 or siRNA 1198 . Results are reported for 48 hrs. (Fig. 2A), a time point that corresponds to early neurogenesis. Injection of siRNA 890 resulted in an average of 59% reduction of sema1a transcripts (N = 9, p,0.001, where N is the number of biological replicates), and injection of siRNA 1198 resulted in an average of 63% reduction (N = 7, p,0.001) of Aae sema1a transcripts at 48 hrs. of development. siRNA mix , a combination of siRNA 890 and siRNA 1198 , yielded the highest levels of knockdown. At 48 hrs. post injection of siRNA mix , sema transcript levels were on average 72% less than that of the controlinjected group (N = 7, p,0.01, Fig. 2A). A maximum average of 83% knockdown was achieved in one siRNA mix replicate. Knockdown in the developing CNS was also verified through in situ hybridization at 54 hrs. of development, immediately preceding the time point at which axon phenotypes were assessed (see below). In situs confirmed reduced levels of sema1a transcripts in siRNA mix -injected embryos at levels comparable to those detected by qRT-PCR, and which revealed nearly complete knockdown in a number of siRNA mix -injected embryos (Fig. 2B).

Analysis of the A. aegypti sema1a knockdown CNS phenotype
In both A. aegypti and D. melanogaster, a scaffold of developing axon pathways give rise to the embryonic ventral nerve cord, which has a ladder-like appearance (Fig. 3A, Fig. 4A). A pair of bilaterally symmetrical longitudinal axon tracts are pioneered separately on either side of the midline in each segment. Although a fraction of early growth cones project only on their own side, most CNS interneurons project their axons across the midline in either the anterior or posterior commissural axon tracts before extending rostrally or caudally in the developing longitudinals [33,34]. At 56 hrs. of development, the mature ventral embryonic nerve cord of A. aegypti (Fig. 3A) resembles that of the mature ventral nerve cord found in a St. 16 D. melanogaster embryo (Fig. 4A, [16]).
The impact of sema1a knockdown on A. aegypti embryonic nerve cord development was assessed through anti-acetylated tubulin staining at 56 hrs. of development. sema1a siRNA-injected embryos display a variety of nerve cord defects, including thinning, breakage, or fusion of the longitudinals (Fig. 3B-F), as well as thin ( Fig. 3B-D) or missing ( Fig. 3D-F) commissures. In some cases, the nerve cord was very severely distorted, almost appearing to be twisted upon itself (Fig. 3F). Injection of either siRNA independently ( Fig. 3B and C) or a mixture of the two (siRNA mix , Fig. 3D-F) generated comparable axon phenotypes. Generation of the sema1a knockdown nerve cord phenotype with two separate siRNAs corresponding to different regions of the sema1a gene suggests that these phenotypes result from knockdown of sema1a and do not result from off-site targeting.
Although anti-Fasciclin II (Fas II) staining previously revealed motor neuron defects in D. melanogaster [26], attempts to score such phenotypes in A. aegypti were abandoned for a number of reasons. First, markers for particular motor neurons, i.e. anti-Fas II, are not yet available in A. aegypti. While the Drosophila anti-Fas II antibody has been useful in the characterization of fruit fly sema1a mutants [26], the Drosophila antibody does not crossreact with A. aegypti Fas II (not shown). Secondly, due to the severe nature of the midline phenotypes observed in A. aegypti (Fig. 3), it would be difficult to rule out that any motor neuron phenotypes observed were not simply secondary to defects at the midline.

Comparative analysis of sema1a loss in D. melanogaster
We wished to assess if Drosophila sema1a loss of function mutant embryos exhibit ventral nerve cord pathfinding defects comparable to those observed in A. aegypti. Yu et al. [26] previously used Fas II staining to uncover defects in the third longitudinal connective of sema1a[P1] null mutants, which were reported to have a thin, discontinuous, and wavy appearance. We further assessed the sema1a[P1] loss of function phenotype through BP102 staining (which labels all axons of the CNS [35]) in stage 16 germ-band retracted embryos, a time point that corresponds to the developmental time at which A. aegypti sema1a knockdown phenotypes were assessed ([4], Fig. 3). This staining revealed that Sema1a[P1] null loss of function mutant embryos exhibit many of the nerve cord defects observed in the A. aegypti sema1a knockdown experiments. These include breakage and occasional fusion of the longitudinal connectives, thin or absent commissures, and general distortion of the nerve cord ( Fig. 4B-D). Although sema1a[P1] is a null allele [26], the range of severity of the phenotypes observed varied just as it did in the mosquito knockdown experiments. Such variation is often observed in conjunction with null alleles for other axon guidance genes, such as Netrin or frazzled, and is typically attributed to the notion that the multiple guidance molecules regulating axonogenesis can compensate for each other [36,37,38].
In order to be certain that the defects associated with the sema1a[P1] allele were not specific to this fly stock, the sema1a[P1] allele was outcrossed to two additional Sema1a Minos-insertion alleles, Sema1a[MB05185] and Sema1a[MB07938] (both of which  A number of studies have characterized the molecular functions of Sema1a in the Drosophila motor and sensory systems. These studies may provide insight into the molecular roles of Sema1a in the fruit fly and mosquito embryonic nerve cord. Yu et al. [39] demonstrated through manipulation of Sema1a, Fas II, and Connectin levels that Sema1a mediates defasciculation in the motor system by functioning as a repellent. They concluded that proper axon guidance resulted from a balance between attractive and repulsive guidance cues. More recently, Zlatic et al. [40], who studied embryonic sensory neuron axons, demonstrated that Semas pattern the dorso-ventral axis in fruit fly embryos. They concluded that sensory axons are delivered to particular regions of the neuropile by their responses to systems of positional cues, including different levels and combinations of Semas in various dorso-ventral layers of the neuropile. Yu et al. [31] also demonstrated that reverse signaling of Sema1a, which interacts genetically with Rho1, regulates axon guidance in the D. melanogaster photoreceptor system. Finally, Sema1a is known to promote growth and induction of key growth regulators such as Myc and Cyclin D in flies, suggesting that it may promote axon growth in addition to guidance [32]. These studies suggest that Sema1a, through regulation of cell adhesion molecules such as Fas II, signaling molecules such as Rho1, and growth regulators such as Myc and Cyclin D, may control axon repulsion, attraction, defasciculation, growth, and/or dorso-ventral patterning in the fly and vector mosquito embryonic ventral nerve cord.

Evolution of Sema1a function in insects
The results of this investigation suggest that Sema1a plays a critical role during development of both the A. aegypti and D. melanogaster ventral nerve cords. Many similar defects, including thinning or loss of the commissural axons and breaks or fusion of the longitudinals, were observed when Sema1a function was  (Figs. 3,4). However, slight differences in the phenotypes, including thinning of the longitudinals in the mature nerve cord of A. aegypti (Fig. 3C-F) but not D. melanogaster (Fig. 4), were also observed. Likewise, defects specific to the third fascicle (Fig. 3A) previously observed in D. melanogaster [26] were not observed at a comparable time point in A. aegypti (Fig. 3B-F). These observations suggest that while the overall function of Sema1a in nerve cord development is largely conserved between A. aegypti and D. melanogaster, the developmental functions of this gene have diverged slightly between the two insects.
These results suggest that further analysis of embryonic nerve cord development in mosquitoes may uncover underlying differences between D. melanogaster and mosquito nervous system development. In support of this idea, our recent analysis of the fra knockdown phenotype in A. aegypti suggests that while this gene is required for commissural axon guidance in both mosquitoes and fruit flies, the penetrance and severity of the A. aegypti knockdown phenotype is greater than that of the Drosophila fra null mutant [16]. The combined results of these investigations indicate that characterizing the functions of additional axon guidance genes in A. aegypti and other insect species may uncover other differences in gene function. Although sema1a expression in A. gambiae is comparable to that of A. aegypti (Fig. 1), functional analysis of Aga sema1a would be necessary to verify if its developmental roles are conserved in different species of mosquitoes. This would require the application of siRNA-mediated embryonic knockdown strategies recently developed for A. aegypti [20] to A. gambiae, which is a future goal of this laboratory. In general, although temporal changes in axon guidance gene expression that reflect underlying differences in nerve cord development have been noted in crustaceans [23], the apparent divergence of axon guidance gene functions within dipterans is somewhat surprising given the many similarities in insect embryonic CNS development that had previously been reported (for example, see [41,42]). However, as discussed previously [15,23], there are numerous examples in which homologous nerve cords have been produced despite the fact that earlier developmental processes have diverged. Functional genetic analyses in mosquitoes and other emerging arthropod models may therefore continue to reveal functional differences of phylogenetically orthologous nervous system development genes. These functional differences might be minor, such as those reported in this investigation, or may perhaps be quite significant, especially when comparing the function of less closely related arthropod species.
In the future, we also hope to apply the Aae sema1a knockdown strategies employed in this investigation to later stages of development. In particular, it would be interesting to determine if Sema1a, which regulates axon targeting during D. melanogaster olfactory development [29,30], functions in the developing olfactory systems of vector mosquitoes. It is important to extend comparative analysis of nervous system developmental genetics to the olfactory system, which has not been particularly well-analyzed in an evolutionary developmental genetic context in arthropods. Furthermore, the mosquito olfactory system is a tissue of particular interest to the vector community, as location of human hosts is an olfactory-driven behavior.

Ethics statement
This investigation was performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The animal use protocol (Study # 11-036) was approved by the University of Notre Dame Institutional Animal Care and Use Committee.

Mosquito Rearing, Egg Collection, and Fixation
A. aegypti Liverpool-IB12 (LVP-IB12) and A. gambiae M Form mosquitoes were used in these investigations. Procedures for mosquito rearing and egg collection [21,43] have been described previously. A. aegypti were fixed as described [19], and A. gambiae fixation was completed through use of a comparable procedure, except that eggs were fixed at room temperature.

Drosophila Genetics
Embryos of the following genotypes were scored in this investigation:

Immunohistochemistry
Immunohistochemistry was performed in mosquitoes as previously described [18]. Drosophila embryos were prepared and stained according to the Patel [35] protocol. Anti-acetylated tubulin (Zymed, San Francisco, CA) was used at a final concentration of 1:100, and BP102 (supplied by N.H. Patel) was used at 1:10. HRP-conjugated secondary antibodies (Jackson Immunoresearch, West Grove, PA) were used at a concentration of 1:200.

RNA interference
Knockdown was performed through embryonic microinjection of siRNAs targeting Aae sema1a (for transcript information, see AAEL002653 at http://www.vectorbase.org, [46]). siRNA design and microinjection were performed as previously described [20]. The following siRNAs were synthesized by Dharmacon RNAi Technologies (Lafayette, CO): siRNA 890 sense: 59 AUCGUUA-GAACCAUGCAAUUUUU 39 and antisense: 59 UUUAGCA-AUCUUGGUACGUUAAA 39 (corresponds to base pairs 890-1,011 of Aae sema1a), siRNA 1198 sense: 59 GCAAGGUUACA-GAGGUAUGUU 39 and antisense: 59 UUCGUUCCAAUGU-CUCCAUAC 39 (corresponds to base pairs 1,198-1,219 of Aae sema1a), and control siRNA sense: 59 UUCAGACUCGCUGAA-CACGUUUU 39 and antisense: 59 UUAAGUCUGAGCGA-CUUGUGCAA 3.9 The control siRNA is a scrambled version of an siRNA targeting Aae sema1a; Blast searches confirmed that this scrambled sequence does not target other genes in the A. aegypti genome. siRNAs were injected at a concentration of 6-8 ug/uL. Measurement of knockdown effectiveness was determined both through in situ hybridization (see above) and through qRT-PCR. qRT-PCR experiments were performed as previously described [47]. In summary, for each replicate total RNA was extracted from ,30 pooled siRNA-microinjected mosquito embryos using Trizol (Invitrogen, Carlsbad, CA). cDNA was prepared with the High Capacity RNA to cDNA Kit (Applied Biosystems, Foster City, CA) per the manufacturer's instructions. Real-time quantification was performed through use of the SYBR Green I PCR kit (Applied Biosystems, Foster City, CA) in conjunction with an Applied BioSystems Step One Plus Real-Time PCR System. Primer sets for Aae sema1a were: For 59 CGCTGATGGATGAAAATGTG 39 and Rev 59 CCCACCGGGAGTTTTAATTT 39. Primer sets for the housekeeping gene RPS17, which was included as the internal standard for data normalization [47], are: For 59 AGA CAA CTA CGT GCC GGA AG 39 and Rev 59 TTG GTG ACC TGG ACA ACG ATG 39. At least seven independent biological replicates were conducted, and all PCR reactions were performed in triplicate. Quantification of results was made by standardizing reactions to RPS17 levels and then using the DDCt method as described [48]. Results were expressed as fold-difference as compared to controlinjected embryos. The Student's t-test was used to analyze qRT-PCR data from replicate experiments.

Phylogenetic analyses
The orthologues of Drosophila sema and plex genes in the mosquito genomes were identified from OrthoDB ( [49], http://cegg. unige.ch/orthodb3/) and Biomart ( [50], http://www.biomart. org/). The orthologous prediction of these genes was consistent between both databases. The amino acid sequences of Sema and Plex from mosquitoes and the twelve Drosophila species were obtained from Vectorbase ( [46], www.vectorbase.org) and Flybase ( [44], www.flyabse.org). Multiple sequence alignment was performed using ClustalW [51]. A phylogenetic tree was drawn using the Neighbor-Joining method [52]. The bootstrap consensus tree was inferred from 1000 replicates [53]. Tree linearization was performed using methodology in reference [54] and assuming equal evolutionary rates in all lineages. The evolutionary distances were computed using the Poisson correction method [55] and correspond to the number of amino acid substitutions per site. Phylogenetic analyses were conducted in MEGA4 [56].