Expression and Function of Toll Pathway Components in the Early Development of the Wasp Nasonia vitripennis

The Toll signaling pathway is the main source of embryonic DV polarity in the fly Drosophila melanogaster. This pathway appears to have been co-opted from an ancestral innate immunity system within the insects and has been deployed in different ways among insect taxa. Here we report the expression and function of homologs of the important components of the D. melanogaster Toll pathway in the wasp Nasonia vitripennis. We found homologs for all the components; many components had one or more additional paralogs in the wasp relative the fly. We also found significant deviations in expression patterns of N. vitripennis homologs. Finally, we provide some preliminary functional analyses of the N. vitripennis homologs, where we find a mixture of conservation and divergence of function.


Introduction
The parasitoid wasp, Nasonia vitripennis, is an emerging genetic model system that has experienced much growth in both its genetic toolkit and its applications over the last decade [1][2][3][4][5]. Embryonic patterning is a topic of particular interest in N. vitripennis, as it undergoes a rapid mode of long germ embryogenesis, that is very similar in many ways to that of the well-described model system D. melanogaster, but with some derived features reflecting the independent evolutionary derivation of this mode of embryogenesis [6,7].
While the fundamental similarity of the wasp and fly early embryonic patterning output was originally observed for anterior-posterior patterning [4,7], the similarity was later found to extend to the orthogonal dorsal-ventral (DV) axis [8]. For DV patterning, it was found that the expression of markers of cell fate was highly conserved between the fly and wasp just prior to the onset of gastrulation. However, the way the patterns arise earlier in development and how the tissues behave after gastrulation were quite distinct [8,9].
In N. vitripennis, the opposite pattern prevails. Loss of BMP signaling leads to loss of DV polarity, while loss of Toll signaling leads only to the loss of the ventral-most fates [36]. Previous examination of the BMP signaling components in N. vitripennis revealed some significant divergences from the D. melanogaster pathway [37], and the significance of

Discovery of Toll Pathway Orthologs
Protein sequences of Drosophila melanogaster Toll Pathway components were obtained from the NCBI "Protein" database. Potential N. vitripennis orthologs were obtained by submitting the corresponding D. melanogaster sequence into the query of the "blastp suite" of the NCBI Basic Local Alignment Search Tool (BLAST) [75]. Search settings were limited by "Database" (non-redundant protein sequences (nr)) and "Organism" (Nasonia vitripennis (taxid:7425)). "BLAST Results" were analyzed by reciprocal BLASTing the top hit under the "Sequences producing significant alignments" header. The goal of this work is to identify and initially characterize the components of the N. vitripennis Toll signaling pathway in detail in order to fill in our gaps of knowledge and, together with previous work, provide a more complete picture of the mechanism of this dynamic DV patterning system in the wasp. A candidate gene approach was used to find wasp orthologs to all known fly components and compare their expression through in situ hybridization. In addition, preliminary functional studies were conducted in N. vitripennis on major pathway components in order to elucidate the inner working of the pathway.

Discovery of Toll Pathway Orthologs
Protein sequences of Drosophila melanogaster Toll Pathway components were obtained from the NCBI "Protein" database. Potential N. vitripennis orthologs were obtained by submitting the corresponding D. melanogaster sequence into the query of the "blastp suite" of the NCBI Basic Local Alignment Search Tool (BLAST) [75]. Search settings were limited by "Database" (non-redundant protein sequences (nr)) and "Organism" (Nasonia vitripennis (taxid:7425)). "BLAST Results" were analyzed by reciprocal BLASTing the top hit under the "Sequences producing significant alignments" header.
The sequence of the top hit was entered into a new "blastp suite" query and BLASTed against the D. melanogaster database ("Organism": Drosophila melanogaster (taxid:7227)). If the original D. melanogaster sequence entered was the top hit of this second BLAST, then the N. vitripennis sequence found was considered an ortholog.
Since there is always the possibility of lineage-specific duplications, we repeated the reciprocal blast with the next best hit, until the N. vitripennis gene returned a D. melanogaster sequence other than the original query. All N. vitripennis sequences that returned the original query (or its known paralog, if any) were also saved and were considered potential orthologs. These potential candidates were then examined using PCR, RNAi, and in situ hybridization. Primers used to amplify fragments of each gene are given in Supplementary File S1.
Subsequently, we performed BLAST while including the beetle Tribolium castaneum and the bee Apis melifera and collected sequences from these four species that corresponded to the original reciprocal BLAST hits and, if present, the next most closely related D. melanogaster gene. These sets of sequences were aligned using Clustal Omega [76]. The resulting alignments were used for phylogenetic analysis with RAxML and support was estimated with 100 bootstrap replicates [77]. Command line used form: raxmlHPC-PTHREADS -T 4 -f a -x 127,745 -p 453,125 -# 100 -m PROTCATDAYHOFF -s ClustalAlign-mentInputGeneX -n OutputGeneX.
Trees were edited using FigTree (v1.4.4) (http://tree.bio.ed.ac.uk/software/figtree/ (accessed on 2 December 2021), and are presented in Supplementary File S2. Some of these analyses identified additional potential orthologs, as the phylogenetic clustering did not necessarily agree with reciprocal BLAST results. These additional candidates will be the subject of future analyses. Sequences that had only one BLAST result per species (Windbeutel, Myd88, Tube, and Bcl3) were not subjected to phylogenetic analysis.
Probe production and in situ hybridization were performed using standard protocols on ovarioles and embryos in order to characterize normal expression patterns of each transcript of interest during oogenesis and embryogenesis [8,9]. Samples were imaged at 20× magnification on a widefield, compound epi-fluorescent microscope (AXIO IMAGER M2, Zeiss, Jena, Germany).

Parental RNA Interference (pRNAi)
Yellow AsymCx pupae were injected with dsRNA (~1µg/mL in water) designed against each of the transcripts of interest (e.g., Nv-tollA, Nv-dl1, Nv-dl2, Nv-dl3, Nv-dl4, Nv-spz1) using standard protocols [78]. Injected pupae were allowed to eclose. Adult RNAi wasps were provided honey water for one day post eclosure and allowed to lay eggs. Wasps were reared individually, in isolated, solitary egg laying chambers [78] or socially in large communal egg laying chambers [37]. Fresh hosts were provided daily. Laid eggs were screened (see below) for quantity (clutch size) and viability (embryonic lethality or malformations).

Qualitative Polymerase Chain Reaction (qPCR)
RNA was isolated from 0-4 h (28 • C) knockdown embryos using standard TRIzolbased protocols (Ambion 15596018) and converted into cDNA using the Protoscript First Strand cDNA synthesis kit (NEB 63001), controlling for total RNA input. For each condition, multiple batches of wasps were injected. Embryos were pooled together from multiple egg lays within an injection batch, but not between different batches (biological replicates). Additionally, two cDNA technical replicates were synthesized per biological replicate. Embryos from mock, water-injected embryos were collected and prepared in a similar manner.
To assess knockdown, qPCR was performed on both knockdown and mock-treated embryos. Standard PCR reactions were assembled using the PowerUp SYBR Green Master Mix (Applied Biosystems: A25742). For each sample, reactions were performed using primers specific to the transcript of interest (Nv-dl1, Nv-dl2, Nv-dl3, Nv-dl4) and primers specific to a housekeeping gene (Nv-rp49). Reactions were performed in triplicate using standard parameters: (50 • C for 2 , 95 • C for 2 , 40 cycles of (95 • C for 15 s, 60 • C for 60 s, plate read, 72 • C for 60 s, plate read), 95 • C for 2 , gradient 60 • C→95 • C (0.2 • C for 1 s). Primer sequences are provided in Supplementary File S1.
Technical replicates (triplicates from both cDNA replicates) for each sample and primer set combination were combined to calculate an average C T value. Delta-Delta C T values were calculated for each knockdown condition and expressed as a relative expression (percentage of wildtype expression) after normalizing to rp49 levels.

Embryonic Lethality Screening
Overnight egg lays from RNAi wasps were collected and plated onto 1% PBS agar plates. Clutch sizes, the number of embryos laid by an individual wasp in a 24 h period, were recorded. Embryos were then screened for embryonic lethality and developmental malformations after being allowed to develop for 24 h at 28 • C (as described in [79]). Mock, water-injected embryos were also collected, plated, and screened as a control.

Lack of Expression of Eggshell-Modifying Component Homologs
Since the components and function of the EGF pathway in establishing polarity of the N. vitripennis DV axis has already been described, we started our analysis with the homologs of the genes downstream of EGF that transmit the signal from the follicle cells to the protease cascade through sulfation of vitelline membrane components [51]. Single orthologs were found for the crucial factors Wbl, Sll, and Pip using reciprocal BLAST and phylogenetic analysis for Sll and Pip (Supplementary Figures S1 and S2; only a single potential Wbl homolog was found). None of these had detectable expression patterns or levels in N. vitripennis ovaries ( Figure 2 [80]), in contrast to D. melanogaster, where pip is expressed strongly in ventral follicle cells and wbl and sll are expressed throughout the epithelium. Similar to the fly, all three factors lacked detectable expression in the wasp embryo (data not shown). The absence of detectable expression of all three of these specialized components is consistent with our observation that Nv-pipe knockdown does not affect DV patterning (not shown). This is also consistent with the lack of pipe homolog expression in the follicle cells of the honeybee [81]. Since Pipe has been shown to function in hemimetabolous insect DV patterning [80], it appears that a role for Pipe in DV patterning may have been lost early in hymenopteran evolution.
thelium. Similar to the fly, all three factors lacked detectable expression in the wasp embryo (data not shown). The absence of detectable expression of all three of these specialized components is consistent with our observation that Nv-pipe knockdown does not affect DV patterning (not shown). This is also consistent with the lack of pipe homolog expression in the follicle cells of the honeybee [81]. Since Pipe has been shown to function in hemimetabolous insect DV patterning [80], it appears that a role for Pipe in DV patterning may have been lost early in hymenopteran evolution.

Novel Paralogs and Expression Patterns of Protease Cascade Components
The components within the perivitelline space are important for activating the Toll signaling pathway on the ventral half of the embryo. Signals from the follicle cells activate a cascade of sequential protein cleavages, leading to the cleavage and activation of the Toll ligand Spätzle. Homologous proteins for all six components (nudel, gastrulation defective, snake, easter, serpin 27A, spätzle) were identified in N. vitripennis. In situ hybridization compared the expression of these homologs in the ovarioles and embryos of D. melanogaster and N. vitripennis.

Nudel
Nudel (Ndl) is exclusively expressed in D. melanogaster follicle cells in late oogenesis in the ovary. Initially, we found a single ortholog of Ndl in the N. vitripennis genome, based on reciprocal BLAST results. However, a more recent search found an additional sequence (XP_031786376) that also BLASTs back to fly Nudel and clusters with it in phylogenetic analysis ( Figure S3). This additional potential paralog will be investigated in future analyses. We termed the originally discovered ortholog (XP_003424379) Nv-ndl. Like its D. melanogaster ortholog [53,54], Nv-ndl is expressed in the follicle cells surrounding the oocyte in late oogenesis, but not in the nurse cells or in the embryo ( Figure 3A), consistent with a previous report and the conserved expression of Nudel throughout insects [80]. sequence (XP_031786376) that also BLASTs back to fly Nudel and clusters with it in phylogenetic analysis ( Figure S3). This additional potential paralog will be investigated in future analyses. We termed the originally discovered ortholog (XP_003424379) Nv-ndl. Like its D. melanogaster ortholog [53,54], Nv-ndl is expressed in the follicle cells surrounding the oocyte in late oogenesis, but not in the nurse cells or in the embryo ( Figure 3A), consistent with a previous report and the conserved expression of Nudel throughout insects [80].

Gastrulation Defective
Gastrulation defective (Gd) is the most upstream component of the cytoplasmic protease cascade that leads to cleavage of Spz. Dm-gd is expressed in the oocyte, the surrounding follicle cells, and later on in the embryo [55]. Two paralogs of Dm-gd were found in N. vitripennis based on reciprocal BLAST results and phylogenetic analysis ( Supplementary Figure S4, which also suggests that CG9649 is paralogous to gd in the fly). Nv-gd1 is expressed throughout the oocyte and follicle cells early on ( Figure 3B). In mid and late egg chambers, mRNA expression is restricted to a single side of the follicle epithelium ( Figure 3C). Nv-gd1 is not expressed in the embryo. Nv-gd2 is neither expressed in the oocyte nor the follicle cells (not shown), but is expressed ubiquitously at low levels in the early embryo and then at higher levels in the yolk of the syncytial blastoderm ( Figure 4A,B). Thus, together the two N. vitripennis paralogs recapitulate the full expression of D. melanogaster gd.

Snake
D. melanogaster snk is expressed in the oocyte, and its protein is secreted into the perivitelline space and acts upstream of Easter [21]. Reciprocal BLAST results supported the protein XP_008207491.1 as the most likely N. vitripennis ortholog of Snake (hereafter referred to as Nv-snk). However, Nv-Snk is more similar to the N. vitripennis protein XP_031786740.1 than it is to any D. melanogaster sequence, suggesting that the two genes in N. vitripennis split from an ancestral Nv-snk. Contrary to this, XP_031786740.1 is a reciprocal best BLAST hit with the relatively unknown D. melanogaster proteins encoded by the genes CG11841 and CG11842, indicating a set of two ancestral snk-like lineages, and we provisionally refer to the gene encoding XP_031786740.1 as Nv-snk2. Nv-snk and Nv-snk2 are barely detectable in the ovaries (not shown) and only Nv-snk shows weak expression in the early embryo ( Figure 4C). Phylogenetic analysis of genes with similarity to Snake suggests several other candidates that might be part of the Snake lineage and which could play a role in the N. vitripennis system (Supplementary Figure S5).
ing follicle cells, and later on in the embryo [55]. Two paralogs of Dm-gd were found in N. vitripennis based on reciprocal BLAST results and phylogenetic analysis (Supplementary Figure S4, which also suggests that CG9649 is paralogous to gd in the fly). Nv-gd1 is expressed throughout the oocyte and follicle cells early on ( Figure 3B). In mid and late egg chambers, mRNA expression is restricted to a single side of the follicle epithelium ( Figure  3C). Nv-gd1 is not expressed in the embryo. Nv-gd2 is neither expressed in the oocyte nor the follicle cells (not shown), but is expressed ubiquitously at low levels in the early embryo and then at higher levels in the yolk of the syncytial blastoderm ( Figure 4A,B). Thus, together the two N. vitripennis paralogs recapitulate the full expression of D. melanogaster gd. Nv-gd2 is initially expressed ubiquitously at low levels in the pre-blastoderm stages (A) before being restricted from the cortex and having elevated expression in the cytoplasm/yolk of the syncytial blastoderm (B). (C) Nv-snk is expressed ubiquitously at low levels in the early (pre-to early blastoderm (shown)) embryo. (D-H) Nv-spn27A is initially expressed in a narrow, anterior stripe along the dorsal midline (D) before elongating to the posterior pole during the late syncytial blastoderm stage (E). The stripe later retracts from the posterior pole (F), expands laterally (G), and finally becomes localized to the presumptive extraembryonic material after gastrulation (H). (I,J) Nv-spz2 is ubiquitously expressed throughout the early blastoderm embryo, with elevated expression along the dorsal midline (I). Elevated expression then expands laterally, creating a broad ring around the entire circumference of the embryo, from the thoracic to the anterior abdominal segments at the late blastoderm stages (J). Anterior is to the left and dorsal is at the top (except in D, F, and H; dorsal views).

Snake
D. melanogaster snk is expressed in the oocyte, and its protein is secreted into the perivitelline space and acts upstream of Easter [21]. Reciprocal BLAST results supported the protein XP_008207491.1 as the most likely N. vitripennis ortholog of Snake (hereafter referred to as Nv-snk). However, Nv-Snk is more similar to the N. vitripennis protein Nv-gd2 is initially expressed ubiquitously at low levels in the pre-blastoderm stages (A) before being restricted from the cortex and having elevated expression in the cytoplasm/yolk of the syncytial blastoderm (B). (C) Nv-snk is expressed ubiquitously at low levels in the early (pre-to early blastoderm (shown)) embryo. (D-H) Nv-spn27A is initially expressed in a narrow, anterior stripe along the dorsal midline (D) before elongating to the posterior pole during the late syncytial blastoderm stage (E). The stripe later retracts from the posterior pole (F), expands laterally (G), and finally becomes localized to the presumptive extraembryonic material after gastrulation (H). (I,J) Nv-spz2 is ubiquitously expressed throughout the early blastoderm embryo, with elevated expression along the dorsal midline (I). Elevated expression then expands laterally, creating a broad ring around the entire circumference of the embryo, from the thoracic to the anterior abdominal segments at the late blastoderm stages (J). Anterior is to the left and dorsal is at the top (except in D,F,H; dorsal views).

Easter
Easter (Ea) is the protease that cleaves Spätzle, producing the active Toll ligand [82]. Melanization protease 1 (MP1) and the Spätzle processing enzyme (Spe) are the top BLAST hits of Ea, and are likely paralogs in D. melanogaster (Supplementary Figure S6), but only Ea has been shown to act in the DV patterning pathway [21]. We found two proteins that are reciprocal best hits to Ea, MP1, and Spe in the N. vitripennis genome. We named them Nv-ea1 and Nv-ea2, and they are examined here. Phylogenetic analysis revealed a very complex set of relationships, and additional potential paralogs of easter may be present (Supplementary Figure S6). Nv-ea1 is expressed strongly throughout the oocyte and the posterior region of the nurse cells ( Figure 3D) and is only weakly detected in the embryo (not shown). Nv-ea2 was at very low or undetectable levels in both the ovary and embryo (not shown).

Serpin27A
Serpin27A (Spn27A) is an important regulator of the protease cascade and controls its activity over time [21]. A single ortholog was found in N. vitripennis (Nv-spn27A) via BLAST and phylogenetic analysis (Supplementary Figure S7). Nv-spn27A is not expressed in the ovarioles (not shown), but is expressed in a dynamic pattern in the embryo ( Figure 4D-H). mRNA is first localized in a short, narrow stripe at the dorsal midline, near the anterior pole of the embryo ( Figure 4D). This expression domain then expands to the posterior pole, retracts back from the posterior pole, and then widens laterally ( Figure 4E-G). This lateral expansion increases during gastrulation until it is expressed in the entire presumptive serosa ( Figure 4H). This expression pattern is quite different from what is seen in the fly. Dm-spn27A is expressed in the oocyte and the early embryo. However, expression remains constant and ubiquitous through the cellular blastoderm stage.

Spätzle
Spätzle is the target of the protease cascade and sets up the gradient of Toll pathway activation and is secreted from the oocyte [83]. We identified two potential paralogs of Spz in the N. vitripennis genome with reciprocal BLAST and support from phylogenetic analysis (Supplementary Figure S8), indicating a duplication in the wasp lineage. Nv-spz1 is expressed in the ovarioles in a similar pattern to that of Nv-ea1, throughout the oocyte and localized to the posterior of the nurse cells ( Figure 3E). Expression is also similarly lacking in the follicle cells and the embryo (not shown). Nv-spz2 is not expressed in any portion of the ovarioles (not shown), but is expressed in the embryo. Expression is ubiquitous throughout the embryo, but elevated along the dorsal midline ( Figure 4I). Elevated expression then expands to a broad lateral domain encircling the entire trunk of the embryo in later blastoderm stages ( Figure 4J). This late and dynamic expression seen in Nv-spz2 has not been described in D. melanogaster.

Toll Receptors
In D. melanogaster, the Toll receptor receives positional information from the follicle cells and perivitelline space through its ligand Spätzle. Toll is then responsible for activating the signaling cascade that will fully polarize the embryo.
There is a single ortholog of the Toll receptor responsible for activating Dorsal in embryonic patterning and immunity. In contrast, we found four Toll paralogs in the N. vitripennis genome that reciprocally BLAST to fly Toll and which are direct orthologs of Dm-Toll, based on phylogenetic analysis (Supplementary Figure S9), distinct from the larger family of Toll-like receptors found among insects [84][85][86]). Three of these (Nv-TollA, C, D) are expressed in both the oocyte and embryo; however, the details of their expression vary. Nv-TollA is expressed in the posterior nurse cells and throughout the oocyte, with strong posterior enrichment and localization ( Figure 5A). Nv-TollC is expressed more strongly in the nurse cells and highly enriched in the most posterior ones, and the mRNA is highly enriched at both the anterior and posterior poles of the oocyte ( Figure 5B). Nv-TollD is more ubiquitously expressed throughout the nurse cells and accumulates throughout the oocyte (as opposed to the localization at the poles in Nv-TollA and C) ( Figure 5C).
In the early blastoderm embryo, Nv-TollA is ubiquitously expressed at low levels, with higher levels of expression in both the anterior and posterior polar regions ( Figure 5D). Later, expression becomes tightly restricted to the two poles and then starts to form a very thin stripe along the ventral midline ( Figure 5E). Nv-TollC is initially expressed at the anterior and posterior poles of the early N. vitripennis blastoderm and also shows a narrow ventral midline stripe early on ( Figure 5F), similar to Nv-TollA. This stripe quickly evolves to a domain that outlines the borders of the presumptive mesoderm ( Figure 5G). Once gastrulation begins, expression is strong throughout the entire presumptive mesoderm ( Figure 5H). Nv-TollD is expressed ubiquitously at low levels in the embryo ( Figure 5I). Nv-TollB is neither expressed during oogenesis nor embryogenesis.
This asymmetric localization of Nv-Toll mRNAs in both the ovary and embryo is unexpected. Dm-Toll is also maternally expressed in the oocyte and embryo, but it is ubiquitously expressed through the blastoderm stages similar to the expression of NvTollD [87][88][89][90]. It is not until gastrulation that zygotic Toll starts to be differentially expressed [87][88][89][90].

Membrane-Interacting Mediators of Toll Signaling
Several components interact with the membrane-bound activated Toll receptor and transmit the signal into the cell. D. melanogaster myeloid differentiation primary response gene 88 (Myd88) is required to localize Tube to the membrane upon Toll activation. In turn, Tube binds to Pelle, which leads to Pelle phosphorylation. Phosphorylation of Pelle leads to phosphorylation of Cactus, and the translocation of Dorsal to the nucleus [11,71]. Single orthologs were detected in N. vitripennis for Tube and Myd88 with reciprocal BLAST, and two paralogs were confirmed for N. vitripennis Pelle (Nv-Pll) (Supplementary Figure S10). N. vitripennis tube (Nv-tub) is expressed in both ovaries and the early embryo ( Figure  6A,B). The two Nv-pll paralogs, as well as Nv-Myd88, are not detected by in situ hybridization in the ovary (not shown), but are expressed at low levels in the early embryo (Figure 6C-E).

Membrane-Interacting Mediators of Toll Signaling
Several components interact with the membrane-bound activated Toll receptor and transmit the signal into the cell. D. melanogaster myeloid differentiation primary response gene 88 (Myd88) is required to localize Tube to the membrane upon Toll activation. In turn, Tube binds to Pelle, which leads to Pelle phosphorylation. Phosphorylation of Pelle leads to phosphorylation of Cactus, and the translocation of Dorsal to the nucleus [11,71]. Single orthologs were detected in N. vitripennis for Tube and Myd88 with reciprocal BLAST, and two paralogs were confirmed for N. vitripennis Pelle (Nv-Pll) (Supplementary Figure S10). N. vitripennis tube (Nv-tub) is expressed in both ovaries and the early embryo ( Figure 6A,B). The two Nv-pll paralogs, as well as Nv-Myd88, are not detected by in situ hybridization in the ovary (not shown), but are expressed at low levels in the early embryo ( Figure 6C-E).

Dorsal Inhibitors
In D. melanogaster, Cactus binding prevents Dorsal from translocating to the nucleus. Toll signaling leads to Cactus phosphorylation and degradation, freeing Dorsal. While a single cactus gene performs this function in the fly, three genes orthologous to Cactus (Nv-cact1, 2, and 3) were found in the N. vitripennis genome by reciprocal BLAST and phylogenetic analysis (Supplementary Figure S11).
All three Nv-cact paralogs are expressed in the nurse cells and oocyte, with slightly different patterns in the nurse cells for each ( Figure 7A-C). Nv-cact1 is also expressed in the follicle cells ( Figure 7A). In addition to the ubiquitous maternal contribution, Nv-cact1 and Nv-cact2 have patterned zygotic expression in the embryo. Nv-cact1 is strongly expressed along the ventral midline ( Figure 7D). This narrow stripe persists from the early blastoderm until the onset of gastrulation ( Figure 7E). Nv-cact2 is ubiquitously expressed in the early embryo ( Figure 7F). Later blastoderm stages have slightly elevated expression along the ventral midline, similar to, but slightly weaker than, what is seen in Nv-cact1 ( Figure 7G). These patterns may indicate feedback control of Nv-cact1 and 2 by the Toll signaling pathway, as seen in other insects [91,92]. Nv-cact3 is not expressed in the embryo.
We also found an additional potential Dorsal inhibiting factor in the N. vitripennis genome. It appears to be most closely related to the B-cell lymphoma-3 (Bcl3) genes in vertebrates, which are distantly related to the IκB factors, like Cactus, and perform similar roles in controlling NfκB nuclear translocation [93]. We also found homologs of Bcl3 in other insects, including the bee Apis meliferra and the beetle Tribolium castaneum. Nv-bcl3 is expressed in the nurse cells, strongly but ubiquitously in the oocyte, but not in the follicle cells ( Figure 8A). In the embryo, Nv-bcl3 is expressed dynamically in a weakly periodic fashion along the AP axis, with broad, alternating regions of high and low expression that become more stripe-like over time ( Figure 8B,C). After gastrulation, Nv-bcl3 is expressed in scattered cells across the ectoderm ( Figure 8D).

Dorsal Inhibitors
In D. melanogaster, Cactus binding prevents Dorsal from translocating to the nucleus. Toll signaling leads to Cactus phosphorylation and degradation, freeing Dorsal. While a single cactus gene performs this function in the fly, three genes orthologous to Cactus (Nv-cact1, 2, and 3) were found in the N. vitripennis genome by reciprocal BLAST and phylogenetic analysis (Supplementary Figure S11).
All three Nv-cact paralogs are expressed in the nurse cells and oocyte, with slightly different patterns in the nurse cells for each ( Figure 7A-C). Nv-cact1 is also expressed in the follicle cells ( Figure 7A). In addition to the ubiquitous maternal contribution, Nv-cact1 and Nv-cact2 have patterned zygotic expression in the embryo. Nv-cact1 is strongly expressed along the ventral midline ( Figure 7D). This narrow stripe persists from the early blastoderm until the onset of gastrulation ( Figure 7E). Nv-cact2 is ubiquitously expressed in the early embryo ( Figure 7F). Later blastoderm stages have slightly elevated expression along the ventral midline, similar to, but slightly weaker than, what is seen in Nv-cact1 ( Figure 7G). These patterns may indicate feedback control of Nv-cact1 and 2 by the Toll signaling pathway, as seen in other insects [91,92]. Nv-cact3 is not expressed in the embryo.
We also found an additional potential Dorsal inhibiting factor in the N. vitripennis genome. It appears to be most closely related to the B-cell lymphoma-3 (Bcl3) genes in vertebrates, which are distantly related to the IκB factors, like Cactus, and perform similar roles in controlling NfκB nuclear translocation [93]. We also found homologs of Bcl3 in other insects, including the bee Apis meliferra and the beetle Tribolium castaneum. Nv-bcl3 is expressed in the nurse cells, strongly but ubiquitously in the oocyte, but not in the follicle cells ( Figure 8A). In the embryo, Nv-bcl3 is expressed dynamically in a weakly periodic fashion along the AP axis, with broad, alternating regions of high and low expression that become more stripe-like over time ( Figure 8B,C). After gastrulation, Nv-bcl3 is expressed in scattered cells across the ectoderm ( Figure 8D).

Duplications in Dorsal
Dorsal nuclear translocation is the functional output of Toll signaling. In the fly, a single dorsal gene, expressed maternally, performs the patterning function. In N. vitripennis, we found four paralogous genes related to dorsal (Nv-dl1-4) (Supplementary Figure S12).
All four N. vitripennis dorsal transcripts are expressed maternally in the germline cells ( Figure 9A-D). Nv-dl1 shows uniform expression in the nurse cells and oocyte ( Figure 9A), while Nv-dl2 appears to be present in a posterior-to-anterior gradient in the nurse cells ( Figure 9B). Nv-dl3 and Nv-dl4 show strong concentration in the posterior nurse cells adjacent to the oocyte ( Figure 9C,D). These differences indicate distinct modes of regulation of these paralogs in the germline, the functional significance of which is at present unclear. All four paralogs are expressed uniformly at the blastoderm stage (with very low levels seen in Nv-dl3) and show no patterned expression ( Figure 9E-H).

Duplications in Dorsal
Dorsal nuclear translocation is the functional output of Toll signaling. In the fly, a single dorsal gene, expressed maternally, performs the patterning function. In N. vitripennis, we found four paralogous genes related to dorsal (Nv-dl1-4) (Supplementary Figure  S12).
All four N. vitripennis dorsal transcripts are expressed maternally in the germline cells ( Figure 9A-D). Nv-dl1 shows uniform expression in the nurse cells and oocyte ( Figure 9A), while Nv-dl2 appears to be present in a posterior-to-anterior gradient in the nurse cells ( Figure 9B). Nv-dl3 and Nv-dl4 show strong concentration in the posterior nurse cells adjacent to the oocyte ( Figure 9C,D). These differences indicate distinct modes of regulation of these paralogs in the germline, the functional significance of which is at present unclear. All four paralogs are expressed uniformly at the blastoderm stage (with very low levels seen in Nv-dl3) and show no patterned expression ( Figure 9E-H).

Functional Analysis of Pathway Components Functional Analysis of Nv-Dorsal Paralogs
While four Nv-Toll paralogs were found, it has already been shown that the knockdown of Nv-TollA by pRNAi results in the dorsalization of the embryo and high levels of embryonic lethality [36]. However, knocking down Nv-TollB, C, or D in a similar way did not result in these or any other observable developmental phenotype.
In order to similarly determine which of the four Nv-Dorsal paralogs is functionally equivalent to Dm-Dl, each paralog was also knocked down by pRNAi and screened for phenotypes. In D. melanogaster, Toll and Dorsal mutants both cause a similar dorsalizing phenotype, therefore it is predicted that the functional N. vitripennis paralog would cause similar phenotypes, as seen in the Nv-Toll knockdowns [21]. Following pRNAi, none of the predicted phenotypes were observed for any of the four paralogs. The average embryonic lethality observed for each paralog was under 15% and in situ hybridization revealed no clear disruption of DV patterning (not shown). Quantitative PCR confirmed that the lack of observed phenotypes was not due to a failure in knockdown, as transcript levels were reduced to below 30% of wildtype levels in multiple clutches and for each paralog ( Figure 10). While both degree of knockdown and frequency of lethality were variable, there was no clear correlation between the two (i.e., strong knockdowns did not show higher embryonic lethality, and weaker knockdowns still showed lethality well above control levels). These results cannot differentiate between incomplete knockdown and redundant function among the paralogs as the explanation for the lack of significant phenotypes when knocking down Nv-dl paralogs. Stronger loss of function disruptions (e.g., CRISPR-induced frameshifts) and multi-target approaches will be necessary to determine which Dorsal paralogs function in N. vitripennis embryonic patterning.

Functional Analysis of Nv-Dorsal Paralogs
While four Nv-Toll paralogs were found, it has already been shown that the knockdown of Nv-TollA by pRNAi results in the dorsalization of the embryo and high levels of embryonic lethality [36]. However, knocking down Nv-TollB, C, or D in a similar way did not result in these or any other observable developmental phenotype.
In order to similarly determine which of the four Nv-Dorsal paralogs is functionally equivalent to Dm-Dl, each paralog was also knocked down by pRNAi and screened for phenotypes. In D. melanogaster, Toll and Dorsal mutants both cause a similar dorsalizing phenotype, therefore it is predicted that the functional N. vitripennis paralog would cause similar phenotypes, as seen in the Nv-Toll knockdowns [21]. Following pRNAi, none of the predicted phenotypes were observed for any of the four paralogs. The average embryonic lethality observed for each paralog was under 15% and in situ hybridization revealed no clear disruption of DV patterning (not shown). Quantitative PCR confirmed that the lack of observed phenotypes was not due to a failure in knockdown, as transcript levels were reduced to below 30% of wildtype levels in multiple clutches and for each paralog ( Figure 10). While both degree of knockdown and frequency of lethality were variable, there was no clear correlation between the two (i.e., strong knockdowns did not show higher embryonic lethality, and weaker knockdowns still showed lethality well above control levels). These results cannot differentiate between incomplete knockdown and redundant function among the paralogs as the explanation for the lack of significant phenotypes when knocking down Nv-dl paralogs. Stronger loss of function disruptions (e.g., CRISPR-induced frameshifts) and multi-target approaches will be necessary to determine which Dorsal paralogs function in N. vitripennis embryonic patterning.

Functional Analysis of Other DV Patterning Components
Previous work has demonstrated that the reduction of EGF signaling by pRNAi leads to a ventralized, partially duplicated axis phenotype in N. vitrpennis embryos [43]. In D. melanogaster a similar, but stronger, effect is mediated by expansion of the pipe expression domain [45], allowing unrestrained cleavage of spz [94]. Given the lack of detectable, patterned expression of Nv-pipe, it seemed unlikely that Nv-pipe would play a role in mediating the EGF signal. Indeed, neither Nv-pipe nor Nv-wbl pRNAi led to patterning defects or reduced viability in embryos, further indicating that this part of the pathway has been deleted in N. vitrpennis.
Preliminary analysis of the protease cascade has not resulted in clear results for the functional conservation of this part of the pathway. No phenotypes were recovered in preliminary pRNAis against Nv-gd1, Nv-gd2, Nv-snk1, Nv-ea1, or Nv-ea2 (not shown). We believe that this is the result of the combination of incomplete knockdown and possible redundant functionality of paralogs for all of these proteases. Nv-ndl was the exception in giving an embryonic lethal phenotype, but the knockdown resulted in defective eggshells, and no embryonic material could be obtained to determine whether patterning was disrupted.
Downstream of the protease cascade, at least one of the Nv-spz paralogs, Nv-spz1, is required for DV patterning, and phenotypes similar to weak Nv-TollA knockdowns were obtained with Nv-spz1 pRNAi ( Figure 11). Only the previously published Nv-TollA gave dorsalizing phenotypes, indicating that it is the primary receptor mediating Nv-Spz1 signaling to the embryo. The potential role of the presumed membrane-associated mediators of Nv-Toll signaling, Nv-myd88, -pll1, -tub, were also investigated. Their role in DV patterning could not be assessed, as the injected wasps were all sickly, with short life-spans, and did not lay eggs. This is likely due to the crucial roles of these genes in innate immunity. The knockdown of Nv-pll2 did not result in sterile wasps. The embryos laid by these wasps did not hatch and had larval deformities commonly seen in other dorsalizing knockdown phenotypes (not shown). Misexpression of downstream DV genes following RNAi still needs to be investigated to confirm this defect in polarity. Finally, pRNAi against the Nv-cact paralogs and Nv-bcl3 resulted in normal embryos, and no DV patterning defects were detected (not shown). Again, we believe this is likely due to incomplete knockdowns and redundancy among paralogs. A summary of these experiments is presented in Table 1.

Discussion
Here we have shown that orthologs of all known components of the Toll signaling pathway are present and well conserved in the wasp N. vitrpennis. For most of the fly Toll pathway genes, at least one N. vitripennis ortholog is expressed in a pattern consistent with having a conserved function. Unfortunately, we could not confirm the presence of conserved function for many of the identified components in this preliminary analysis. More focused and intensive functional testing will be required to confidently ascribe or eliminate potential functions for many of these factors.
An exception to the general rule of conserved expression is the cassette of proteins needed for sulfation of the vitelline membrane: Sll, Wbl, and Pipe. While present in the genome, none of these are expressed in the ovary of N. vitrpennis, and do not give phenotypes. Based on the recent discovery of a Drosophila-like function of Pipe in the cricket [80], it appears that this pathway has been deleted from the DV patterning hierarchy in the N. vitrpennis lineage. The absence of pipe expression in the follicle cells in the honeybee indicates that the loss likely happened early in hymenopteran evolution [81]. Interestingly, a recent investigation of vitelline membrane proteins in N. vitrpennis [95] found that their knockdown leads to phenotypes at the cuticular level quite similar to those where the DV axis is disrupted ( [43,96], J.A.L., personal observations). This indicates that vitelline membrane components still play a role in establishing DV polarity, but the mechanism and potentially the effect may be different.
We also observed large scale gene duplication resulting in multiple paralogs of several components of the system, most notably four Toll paralogs, four dorsal paralogs, and three cactus paralogs. The functional relevance of so many paralogs is so far not clear. It is not a general feature of N. vitrpennis signaling pathways, as the wasp BMP pathway exhibited very few lineage-specific paralogs [37]. Are these duplication events adding robustness and protection to vital portions of the pathway, potentially through subfunctionalization or specialization of the various paralogs? Or are they paired with modified/diversified expression pattern domains because they are evolving to meet some species-specific function or need? In depth analysis and more robust gene disruption techniques will be needed to answer these questions.
Perhaps the biggest unanswered question is: how is the EGF signal from the oocyte to the follicle cells transmitted to the embryo? We have shown that EGF-mediated asymmetry signals are conserved and required to polarize the embryo. Here we have also shown that Spz1 is likely the ligand primarily responsible for embryonic Toll activation. So far, however, how the EGF polarity information is converted into localized production of active Spz is unknown. We do know that Nv-Pip is not involved. Have novel proteins been coopted into the pathway to fill Nv-Pip's void? Or is a completely new mechanism in place to relay the EGF signal from the follicle cells to the oocyte? The expression of Nv-gd1 in a subset of follicle cells is potentially significant, since fly Gd protein becomes enriched at the ventral side of the eggshell in a Pipe-dependent manner, and thus plays a crucial role in polarizing the protease cascade and producing ventral cleavage of Spz [97]. If Nv-Gd1 protein localization reflects its mRNA, this could provide a method to polarize the embryo in the absence of Nv-Pip function. While preliminarily Nv-gd1 knockdown gave no phenotype, future studies could employ more complete knock-out approaches, such as CRISPR, to test this idea more conclusively.
In general, it is clear that new approaches are needed to rigorously test the function of Toll signaling components in N. vitripennis, which will then allow more meaningful comparisons with other insect species with the goal of understanding the evolutionary significance of this ancient pathway in the diversity of insects.

Conflicts of Interest:
The authors declare no conflict of interest.