Maternal provision of transformer-2 is required for female development and embryo viability in the wasp

In insect sex determination a primary signal starts the genetic sex determination cascade that, in most insect orders, is subsequently transduced down the cascade by a transformer (tra) ortholog. Only a female-specifically spliced tra mRNA yields a functional TRA-protein that forms a complex with TRA2, encoded by a transformer-2 (tra2) ortholog, to act as a sex specific splicing regulator of the downstream transcription factors doublesex (dsx) and fruitless (fru). Here, we identify the tra2 ortholog of the haplodiploid parasitoid wasp N. vitripennis (Nv-tra2) and confirm its function in N. vitripennis sex determination. Knock down of Nv-tra2 by parental RNA interference (pRNAi) results in complete sex reversal of diploid offspring from female to male, indicating the requirement of Nv-tra2 for female sex determination. As Nv-tra2 pRNAi leads to frequent lethality in early developmental stages, maternal provision of Nv-tra2 transcripts is apparently also required for another, non-sex determining function during embryogenesis. In addition, lethality following Nv-tra2 pRNAi appears more pronounced in diploid than in haploid offspring. This diploid lethal effect was also observed following Nv-tra pRNAi, which served as a positive control in our experiments. As diploid embryos from fertilized eggs have a paternal chromosome set in addition to the maternal one, this suggests that either the presence of this paternal chromosome set or the dosage effect resulting from the diploid state is incompatible with the induced male development in N. vitripennis caused by either Nv-tra2 or Nv-tra pRNAi. The role of Nv-tra2 in activating the female sex determination pathway yields more insight into the sex determination mechanism of Nasonia. © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Insect sex determination involves a cascade of regulatory genes that is evolving bottom-up (Wilkins, 1995). Doublesex (dsx), the interpreter for the sexual identity of the cell, is the most conserved gene at the bottom of the cascade (Shukla and Nagaraju, 2010;Verhulst and van de Zande, 2015). The primary signal at the top of the cascade directs sex determination into the female or male mode and is highly variable throughout the insect class (S anchez, 2008). The regulation of sex determination resembles an hourglass model in which the variable primary signal is considered as the instruction, and the multiple actions of dsx as the execution phase (Bopp et al., 2014). In most insects, the center of the hourglass is represented by transformer (tra), a fast-evolving gene that transduces the primary signal through sex-specific splicing (S anchez, 2008;Verhulst et al., 2010b). Only when the tra transcript is spliced in the female-variant, it can be translated into a functional TRA protein that will regulate female specific splicing of dsx transcripts. In addition, tra maintains its own female-specific splicing mode through an auto-regulatory loop that functions as a "memory" to ensure proper sexual differentiation (Bopp et al., 2014).
In Hymenoptera that have a haplodiploid reproductive system, males develop from haploid unfertilized eggs and females from diploid fertilized eggs. Two mechanisms of sex determination have thus far functionally been described: the Complementary Sex Determination (CSD) in Apis mellifera and the Maternal Effect Genomic Imprinting Sex Determination (MEGISD) in Nasonia vitripennis. CSD involves the complementation of alleles at the csd locus, in which heterozygosity at one or more csd loci leads to female development (Beye et al., 2003;Cook, 1993;van Wilgenburg et al., 2006). Its genetic basis has only been elucidated for the honeybee (Apis mellifera), where the csd locus was identified as a duplication of feminizer (fem), the honeybee ortholog of tra (Beye et al., 2003;Hasselmann et al., 2008;Gempe et al., 2009). The MEGISD model (Beukeboom et al., 2007;Beukeboom and van de Zande, 2010) involves maternal imprinting, maternal provision of tra mRNA and activation of zygotic tra transcription (Verhulst et al., 2010a(Verhulst et al., , 2013. In this model the non-silenced paternal allele of the hypothetical womanizer (wom) gene will initiate zygotic transcription of Nv-tra in diploid embryos from fertilized eggs only (van de Zande and Verhulst, 2014). Additionally, the Nasonia female sexdetermining mechanism is dependent on maternal provision of tra mRNA to the egg, like in many other insect species (e.g. the dipterans Ceratitis capitata and Musca domestica, and the coleopteran Tribolium castaneum (Hediger et al., 2010;Pane et al., 2002;Shukla and Palli, 2012). The roles of tra and dsx in Nasonia sex determination have been described in detail (Oliveira et al., 2009;Verhulst et al., 2010a), but a role of tra2 within this model has not yet been determined. Here, we describe the structure and splicing of Nv-tra2 and compare it to other known tra2 orthologs. We confirm a function of tra2 in Nasonia sex determination through parental RNA interference (pRNAi) experiments and also observed other roles of Nv-tra2 during development.

Nasonia vitripennis strains and rearing
The N. vitripennis lab strain AsymCX (Werren et al., 2010) and the recessive red eye-colour mutant strain ST DR were used throughout the experiments. Homozygous ST DR females mated with AsymCX males produce diploid female offspring with wildtype eyes and red-eyed haploid male offspring). This allows the detection of wild-type diploid males resulting from knockdown of either Nv-tra2 or Nv-tra (Verhulst et al., 2010a). Wasps were reared on Calliphora sp. hosts and cultured at 25 C at a L16:D8 cycle.

RNA extraction and cDNA synthesis
Total RNA of individual wasps and embryo pools was extracted with TriZol (Invitrogen, Carlsbad, CA, USA) according to manufacturer's protocol. RNA was reverse-transcribed with oligo-dT and hexamer primers in a 1:6 ratio with the RevertAidTM H Minus First Strand cDNA Synthesis Kit (Fermentas, Hanover, MD, USA) and stored at À80 C. RNA samples for gene identification were individual male and female samples of the AsymCX strain. For 3 0 RACE (Rapid Amplification of cDNA Ends), RNA was reverse-transcribed with the RevertAidTM H Minus First Strand cDNA Synthesis Kit (Fermentas, Hanover, MD, USA) using 25 mM 3 0 RACE adapter (5 0 -GCG AGC ACA GAA TTA ATA CGA CTC ACT ATA GGT 12VN-3 0 ) from FirstChoice RLM-RACE kit (Ambion, Austin, TX, USA). For 5 0 RACE, RNA was processed according to manufacturer's instructions (FirstChoice RLM-RACE kit, Ambion, Austin, TX, USA) and reverse transcribed using the RevertAidTM H Minus First Strand cDNA Synthesis Kit (Fermentas, Hanover, MD, USA).

Identification of Nv-tra2 genomic structure
Primers were designed based on the predicted Nv-tra2 transcript (LOC100116671) from the N. vitripennis genome (Werren et al., 2010) (Table 1). 5 0 RACE-PCR was performed with outer primer Nvtra2_5Rout and inner primer Nvtra2_5Rin at 94 C for 3 min, 35 cycles of 94 C for 30 s, 58 C for 30 s and 72 C for 2 min, with a final extension of 10 min at 72 C. Outer primer Nvtra2_3Rout and inner primer Nvtra2_3Rin were used in 3 0 RACE-PCR with Phusion High-Fidelity DNA polymerase (Fermentas, Hanover, MD, USA). Cycling conditions were 98 C for 1 min, 35 cycles of 98 C for 10 s, 57(out)/55(in) C for 30 s and 72 C for 60 s, with a final extension of 10 min at 72 C. Resulting PCR fragments were run and visualized on ethidiumbromide-containing 1.5% agarose gel with 1 Â TAE buffer.
All RACE-PCR products were ligated into pGEM-T vector (Promega, Madison, WI, USA) after purification using GeneJET Gel Purification Kit (Fermentas, Hanover, MD, USA). Ligation reactions were used to transform competent JM-109 Escherichia coli (Promega, Madison, WI, USA). Colony-PCR was conducted by use of pGEM-T primers (5 0 -GTA AAA CGA CGG CCA GT-3 0 ) and 5 0 -GGA AAC AGC TAT GAC CAT G-3 0 ) at 94 C for 3 min, 40 cycles of 94 C for 30 s, 55 C for 30 s and 72 C for 2 min, with a final extension of 7 min at 72 C.
Additional Reverse Transcription-PCRs (RT-PCR) were used to identify splice variants. These PCRs were performed with primers Nvtra2_exon1F/Nvtra2_exon4R and Nvtra2_exon2F/Nvtra2_ex-on6R at 94 C for 3 min, 40 cycles of 94 C for 30 s, 55 C for 30 s and 72 C for 2 min, with a final extension of 7 min at 72 C. PCRfragments were sequenced on an ABI 3730XL capillary sequencer (Applied Biosystems) and reads were inspected in Chromas (Technelysium) and aligned in MEGA4 (Tamura et al., 2007). Exonintron structure of the genes was constructed by comparing the mRNA sequences to N. vitripennis genome and visualized with Exon-Intron Graphic Maker (http://wormweb.org/exonintron).

Parental RNAi and sample collection
Parental RNAi knockdown was induced in ST DR females in the white pupal stage (Lynch and Desplan, 2006). Non-sex-specific regions were amplified with Nvtra2_RNAiF and Nvtra2_RNAiR primers for Nv-tra2 dsRNA (564bp) and NvTra_RNAiF1 and NvTra_RNAiR1 primers for Nv-tra dsRNA (452bp) ( Table 1). The 5 0 and 3 0 ends of those amplicons were provided with a T7 promotor sequence for dsRNA production using the Megascript RNAi kit (Ambion, Austin, Texas, USA) according to the manufacturer's protocol.
Approximately 600 female ST DR pupae were injected in the abdomen with 1.3 mg/ml of Nv-tra2 dsRNA, Nv-tra dsRNA or, as a negative control, sterile milliQ water. Nv-tra was used as a positive control (cf. Verhulst et al., 2010a). Injections were performed with Femtotips II (Eppendorf) needles connected to a Femtojet (Eppendorf, Hamburg, Germany) with continuous injection flow. Hundred and twenty adult females of each category were mated with AsymCX males and evenly divided over two sets for either embryo collections or scoring of offspring production. Virgin females (60 wasps) were set up to produce haploid male controls. Embryonic levels of Nv-tra2, Nv-tra and Nv-dsx mRNAs and sexspecific splicing of Nv-tra and Nv-dsx were assessed in early embryos of less than 3 h old. Injected and control females were provided with Calliphora sp. hosts in individual egg laying chambers assembled from filter tips (1 ml) capped with a cutoff 1.5 ml vial as a host carrier. The anterior side of the hosts was exposed to the parasitizing female to facilitate localization of the embryos during collection. After 2 h of oviposition in egg-laying chambers, hosts were gently opened to collect approximately 55 embryos in 100% ethanol per sample. Eight embryo samples per category were snapfrozen in liquid nitrogen and stored at À80 C until RNA extraction.
Levels and sex-specific splicing of Nv-tra2, Nv-tra and Nv-dsx mRNAs were determined in adult wasps that had been injected with sterile water, Nv-tra2 or Nv-tra dsRNA. Injected females were provided fresh hosts daily to oviposit for 24 h at 25 C. Five females of each category were individually used in RNA extractions. Parasitized hosts were incubated for 14 days at 25 C after which offspring emerged. Offspring of Nv-tra2 and Nv-tra injected females were phenotypically screened for sex and eye colour and 10 diploid and 10 haploid males were collected per gene knockdown. Expected sex ratio and eye colour of the offspring were compared to control-group females, and 10 diploid females and 10 haploid males were collected and stored at À80 C. Of all categories, another two adult offspring were collected and stored at À20 C for flow cytometry analysis to confirm ploidy (see 2.8).

mRNA levels in embryos and adults
Quantitative real-time PCR (qPCR) was performed with 5 ml of a 50-fold cDNA dilution and 300 nM PerfeCTaTM SYBR ® Green mix (Quanta Biosciences, Gaithersburg, MD, USA) on an Applied Biosystems 7300 Real Time PCR System (Foster City, CA, USA). Nv-tra2, Nv-tra and Nv-dsx were amplified with non-sex-specific, exonspanning primers at 250 nM (Table 1) (Verhulst et al., 2010a). The elongation factor 1 alpha (EF1a) was used as a reference gene using exon-spanning primers (250 nM) (Table 1) (Verhulst et al., 2010a). qPCR profiles were 95 C for 3 min, 45 amplification cycles of 15 s at 95 C, 58 C/58 C/57 C/55 C for Nv-tra2/Nv-tra/Nv-dsx/Nv-EF1a, 72 C for 30 s and followed up by a standard ABI7300 dissociation curve. Raw fluorescence data generated by 7300 System SDS Software (Applied Biosystems, Foster City, CA, USA) were base-line corrected and the N0 value calculated for PCR efficiencies per amplicon with LinRegPCR 11.0 (Ramakers et al., 2003). Relative levels were determined by dividing Nv-tra2, Nv-tra and Nv-dsx N0 values by Nv-EF1a N0. A one-way ANOVA was used for each life stage (embryo, adult offspring) to test relative differences between the Nv-tra2, Nv-tra dsRNA injected samples and the water-injected control samples.

Splice variant analysis of Nv-tra and Nv-dsx following pRNAi
Sex-specific fragments of Nv-tra and Nv-dsx in injected and control females, and their embryonic and adult offspring, were analyzed by RT-PCR. 5 ml of a 50-fold cDNA dilution was used in a PCR. For sex-specific Nv-tra amplification primers NvTra_F2 and NvTra_R3 were used, located at exon 2 and 3, yielding a single 228 bp fragment in females and three fragments of 514, 460 and 282 bp in males depending on their age. Primers NvDsxU_F3 and NvDsxM_R1 were used for sex-specific Nv-dsx amplification, yielding 543 bp in females and 651 bp in males (Verhulst et al., 2010a). For amplification and cDNA integrity control Nv-EF1a was amplified with NvEF1a_F1 and NvEF1a_R1 primers yielding a 174 bp fragment in both genders. The PCR profile was 45 cycles of 95 C for 15 s, 55 C for 30 s and 72 C for 30 s. Fragments were run and Table 1 Overview of primers used in this study. The binding site sequence for T7 RNA polymerase is noted between square brackets.

Flow cytometry of diploid males
Ploidy of Nv-tra2 and Nv-tra diploid males was confirmed by flow cytometry analysis. Adult male wasp heads were homogenized in Galbraith buffer (21 mM MgCl 2 , 30 mM tri-sodium citrate hydrate, 20 mM MOPS, 0,1% Triton X-100, 1 mg/l RNAse A) using Dounce homogenizers, filtered by 0.7 mm cell strainer caps (BD Falcon Cell strainer #352235, BD Biosciences, San Jose, California, USA) and stained in propidium iodide (Sigma, St. Louis, Missouri, USA). Samples were loaded on a BD FACS Aria II and analyzed by BD FACSDiva software (BD Biosciences, San Jose, California, USA). References for ploidy were set by haploid and diploid males from a polyploid strain (Beukeboom and Kamping, 2006).
2.9. Viability and fertility of offspring after parental Nv-tra2 and Nv-tra knockdown Brood sizes were analyzed to assess the survival of offspring from Nv-tra2 dsRNA, Nv-tra dsRNA, sterile water injected and noninjected mothers. The number of embryos was compared to the number of emerging adult offspring. ST DR female pupae were injected with dsRNA targeting Nv-tra2 (N ¼ 200), Nv-tra (N ¼ 200) and sterile water (N ¼ 200). Of each category 100 adult females were mated with AsymCX males, while the other 100 remained virgin. Mated and virgin females were individually cultured in egg laying chambers and divided over 2 batches which were alternately used for embryo or adult offspring counts. This prevented biases caused by knockdown efficiency or fecundity of the wasps. It additionally allowed exclusion of females that were not mated and females that were not injected from further analysis. Wasps were allowed to parasitize on the anterior side of the hosts for 2 h at 25 C, which was repeated on four consecutive days at a similar time of day. Females were allowed to oviposit for 2 h, after which they were removed and hosts were either opened for embryo counts or retained at 25 C for 14 days to count adult offspring. Numbers of embryos and adult offspring were compared in a general linear model and a post-hoc Tukey-test was used to examine sample differences in embryo and adult number. Differences in egg number laid by females in each category were compared with a Kruskal-Wallis test.
Adult male offspring from these experiments were used to assess fertility of haploid and diploid Nv-tra2 and Nv-tra pRNAi males compared to control males (haploid males from waterinjected mothers). Each male (N ¼ 20 per category) was mated to 7 virgin females very shortly after one another to induce sperm depletion (Ruther et al., 2009). These females were hosted on 4 Calliphora sp. pupae every other day over a period of 10 days (5 data points). Offspring were scored after emergence and the fraction of data points (of 7 females x 5 rehostings) producing daughters was calculated for each male. This provided a measurement of (diploid) male fertility. Differences between categories were compared with a one-way ANOVA.

Structure, splicing and conservation of tra2 in N. vitripennis
To identify a tra2 homolog in N. vitripennis, the reference genome of AsymCX (Werren et al., 2010) was screened with an A. mellifera TRA2 query (Genbank accession AFJ15561) using tblastn. A putative homolog of tra2 (Genbank accession XP_001601106, predicted isoform X3) could be detected that had previously been described in an amino acid comparison in Nissen et al. (2012). This predicted sequence was used for RACE-PCR and RT-PCR primer design to identify the gene structure. Four splice variants of Nv-tra2 were detected in N. vitripennis based on a combination of RACE-PCRs and verified by RT-PCRs (Fig. 1). The most commonly detected variant was Nv-tra2 A . The less abundant variants Nv-tra2 B and Nv-tra2 C were identified by their coding region corresponding to exon 5 and 6a (Fig. 1). Variant Nv-tra2 D was detected in low quantities, includes exon 3 and leads to a truncated ORF. An additional 24 bp (5 0 -CATCATTTGCTACCTTACACACAG-3 0 ) was present in very low frequencies in one or more splice variants, including the 5 0 end of exon 2. Splice variant Nv-tra2 A matches exactly with the predicted isoform X3 (Genbank accession XP_001601106). Only splice variants Nv-tra2 C and Nv-tra2 D were not predicted by the automated computational analysis of the tra2 locus (NCBI Nasonia vitripennis Annotation Release 101). No sexspecific splice variants were detected. All splice variants are deposited in Genbank with accession number: KY938035 (Nv-tra2 A ); KY938036 (Nv-tra2 B ); KY938037 (Nv-tra2 C ); KY938038 (Nvtra2 D ).
The structure of TRA2, consisting of a single RNA-binding domain (RBD) flanked by RS-rich regions, is conserved in N. vitripennis. Unlike in A. mellifera (Nissen et al., 2012) no variation was found in the length of the first RS-rich domain. The 24 bp stretch is just upstream of this region. The second RS-rich region is entirely absent in the low abundant splice variants C and D. Variant Nv-Tra2 D lacks a large part of the RBD region, which is fully present in the other splice variants (Fig. 1).
The amino acid sequence of NV-TRA2 was aligned to known TRA2 homologs, revealing strong conservation, in particular within the Hymenoptera (Fig. 2). A notable feature, compared to other insect TRA2 peptides, is the presence of a glycine-rich region at the 3' end of the Nv-tra2 A , Nv-tra2 B and Nv-tra2 C splice variants (presence in Nv-tra2 A depicted in Fig. 2 between positions 266 to 288). This glycine-rich region is present in D. melanogaster RBP1 which is an important co-factor in the TRA-TRA2 regulation of dsx splicing (Heinrichs and Baker, 1997). The Gly-rich domain in RBP1 is directly involved in protein-protein interactions between RBP1 and TRA2. It is also present in the chalcid Ceratosolen solmsi (XP_011500657 (Xiao et al., 2013)) and B. mori (Niu et al., 2005) (Fig. 2), which could signify a gain of TRA2 function in these systems.

Sex reversal effects of Nv-tra2 parental RNAi knockdown
A pilot RT-PCR experiment revealed Nv-tra2 presence in early embryos (<3 h old) indicating maternal provision of Nv-tra2 to the eggs (original data not shown, but the presence of maternally provided Nv-tra2 is depicted here in Fig. 3a control and Nv-tra pRNAi samples). The bulk of zygotic expression in embryos laid by oviparous animals is initiated at the midblastula transition, a precise developmental point prior to gastrulation (Langley et al., 2014). In N. vitripennis blastoderm formation occurs 4e9 h after egg laying after which gastrulation starts (Bull, 1982), indicating that the midblastula transition in N. vitripennis starts no earlier than 4 h after egg laying. We assume this 4 h time point as the start of largescale zygotic transcription in N. vitripennis and have shown that zygotic transcription of Nv-tra starts at 5 h of development in embryos from fertilized eggs (Verhulst et al., 2010a). As all embryonic samples in this study were less than 3 h of age, the Nv-tra transcripts have a maternal origin (Verhulst et al., 2010a) and the Nv-tra2 transcripts are also considered to be of maternal origin. By using pRNAi it is then possible to knockdown the maternal provision of Nv-tra2 mRNA to the eggs. In order to target this maternal provision of all Nv-tra2 splice variants, a 565 bp sequence containing the RBD and spanning all splice variants was used to synthesize dsRNA (primer positions shown in Fig. 1). This Nv-tra2 dsRNA was injected into females in the pupal stage. RNAi off-target effects were predicted with WaspAtlas (Davies and Tauber, 2015) and no targets other than Nv-tra2 were detected. Parental knockdown (pRNAi) with Nv-tra2 dsRNA led to a 10-fold reduction of Nv-tra2 transcripts in the offspring compared to that of water injected females (F (2,15) ¼ 27.1, P < 0.001) (Fig. 3a), but no effect on the level of Nv-tra mRNA (Tukey-test: P ¼ 0.22) (Fig. 3b). Parental knockdown of Nv-tra led to a significant decrease in Nv-tra mRNA levels in the offspring compared to that of water-injected females (F (2,15) ¼ 13.3, P < 0.001) (Fig. 3b), but no significant effect on Nv-tra2 mRNA levels (Tukey-test: P ¼ 0.18) (Fig. 3a). Hence, pRNAi of either gene does not interfere with the maternal transcript provision of the other.
Adult offspring of Nv-tra2 knockdown mated females were scored after emergence; they contained no daughters but consisted solely of a modest number of diploid males and a larger number of haploid males (Table 2). Mated females injected with water produced diploid female and haploid male offspring as expected under haplodiploidy, indicating that the parental knockdown of Nv-tra2 transcripts in females caused a sex reversal of their diploid offspring, turning them into males whereas haploid offspring were unaffected. Nv-tra pRNAi also resulted in diploid and haploid male offspring, in agreement with earlier observations (Verhulst et al., 2010a). The ploidy of males with wild-type and red eye colour was confirmed with flow cytometry to be diploid and haploid, respectively. The water-injected mated females produced progeny with a low sex ratio (calculated as the number of males divided by the total number of offspring). In the adult offspring of injected mothers, relative levels of Nv-tra2 and Nv-tra mRNA had recovered, after the initial reduction in early embryonic stages. The sexspecific splicing of sex determination genes Nv-tra and Nv-dsx was assessed in the adult offspring of Nv-tra2 dsRNA-, Nv-tra dsRNA-and water-injected females. Diploid females and haploid males from water-injected mothers have normal female-or malespecific splicing of Nv-tra and Nv-dsx as expected. Both Nv-tra2 and Nv-tra dsRNA-injected mated females produced an all-male progeny, consisting of haploid and diploid males, that all showed splicing of Nv-tra and Nv-dsx in the male-specific mode (Fig. 4). This suggests that the role of Nv-tra2 in the sex determination cascade is conserved at the level of interaction with Nv-tra. Furthermore, both Nv-tra and Nv-tra2 transcripts are detected prior to the presumed start of zygotic transcription around 4 h after egg laying. This early presence is presumed to result from maternal provision of the Nvtra and Nv-tra2 transcripts to the zygote and required for female development of fertilized eggs.

Functionality of Nv-tra2: viability and additional knockdown effects
All diploid individuals observed after Nv-tra2 knockdown were males, but their number was very low compared to the number of diploid males after Nv-tra knockdown and of diploid females after injections with water (Table 2). This reduction in offspring after pRNAi could result from a reduced oviposition rate or from lethality during development. To determine which was the case, the number of eggs and adult offspring produced by mated and virgin females were compared. A large discrepancy between the number of oviposited eggs and emerging adults, in absence of larval or pupal remains, would indicate embryonic inviability, whereas equally lowered numbers of eggs and adults would indicate reduced oviposition.
In Nv-tra2 pRNAi virgin females, a significant difference was detected between the number of oviposited eggs and the number of emerged offspring (F (4,1014) ¼ 9.8, P < 0.001, Tukey-test: P < 0.001). This result indicates an effect of Nv-tra2 on offspring viability. As no dead larvae or pupae were observed in opened hosts, this lethal effect appears to occur during embryonic development. The number of eggs deposited by Nv-tra pRNAi virgin females did not significantly differ from the number of emerged offspring (Tukeytest: P ¼ 0.74) (Fig. 5a). Mated females injected with either Nv-tra2 (Tukey-test: P < 0.001) or Nv-tra dsRNA (Tukey-test: P < 0.001) showed a discrepancy between number of oviposited eggs and adult emergence (F (4,971) ¼ 34.9, P < 0.001) (Fig. 5b), indicating embryonic lethality in both dsRNA treatments. Overall, fewer embryos were found for both Nv-tra and Nv-tra2 dsRNA-injected females (H (3,507) ¼ 84.7, P < 0.001) (Fig. 5a) when compared to noninjected females, but this reduction was also present in the waterinjected control category, suggesting an effect of handling and injection trauma more so than potential dsRNA side effects.
Since Nv-tra2 pRNAi caused mortality in the offspring of the treated wasps, the fitness of the surviving male offspring was assessed by testing their fertility. This showed that haploid and diploid Nv-tra pRNAi male offspring as well as haploid and diploid Nv-tra2 pRNAi male offspring had normal capabilities to fertilize females compared to the control haploid male offspring of waterinjected mothers (Fig. 6). A significant difference was only found between haploid and diploid Nv-tra pRNAi male offspring (F (4,94) ¼ 3.5916, P ¼ 0.009, Tukey: P ¼ 0.003). As neither haploid nor diploid Nv-tra pRNAi male offspring differed in fertilization abilities from the haploid control male offspring, we suggest that this significant difference is an effect of the small sample size used in this experiment (N ¼ 20 per category) or an effect of experimental procedure caused by slight timing differences in collection of males and set-up of matings.

Conservation of tra2 in the N. vitripennis sex determination cascade
We showed that an ortholog of tra2 is present in the genome of N. vitripennis, and that it is not sex-specifically spliced. This is corroborated by observations of tra2 splicing in other insects (Martín et al., 2011;Nissen et al., 2012;Niu et al., 2005;Shukla and Palli, 2013). The tra2 gene is an important component of the Nasonia sex-determining cascade, as prevention of Nv-tra2 maternal transcript provision by pRNAi leads to male-specific splicing of both Nv-tra and Nv-dsx pre-mRNA in diploid fertilized eggs and subsequent differentiation into functional males.
The sex reversal from female to male development of fertilized eggs upon parental knockdown of Nv-tra2 expression indicates functional and positional conservation of tra2 in the Nasonia sex determination cascade. Absence of Nv-tra2 mRNA in early embryonic stages fails to direct the sex determining pathway of diploid embryos towards female development, as only male-specific splice variants of Nv-tra and Nv-dsx are detected. Although it is possible that pRNAi also reduces zygotic transcripts in later embryonic stages, our results indicate that maternally provided transcripts are crucial for a timely activation of female sex determination in N. vitripennis. This is supported by the fact that in both the CSD mechanism of A. mellifera (Gempe et al., 2009) and the sex determination mechanism in the hymenopteran A. tabida (Geuverink, unpublished data) maternal provision of fem or female-specific tra mRNA is absent and female sex determination ensues differently. Our results also indicate that NV-TRA2 is required for initiation of the female-specific Nv-tra auto-regulatory loop. After all, if NV-TRA2 would only interact with Nv-dsx, maintenance of femalespecific splicing of Nv-tra would be unaffected and only malespecific splicing of Nv-dsx would occur. Similar to A. mellifera (Nissen et al., 2012), in Nasonia TRA2 acts on two levels in the sex determination cascade, regulating the splicing of both tra and dsx.

Role of Nv-tra2 in embryo viability
Following pRNAi of Nv-tra2, a high lethality of both haploid and diploid offspring was observed, while Nv-tra pRNAi led to only high lethality in the diploid offspring. This suggests that (1) Nv-tra2 has Relative levels of Nv-tra2 mRNA (a) and Nv-tra mRNA (b) in control embryos (water), embryos of Nv-tra2 pRNAi females and embryos of Nv-tra pRNAi females. Different letters above the bars indicate significant differences between treatment groups (Tukey test: P < 0.05).

Table 2
Offspring number, ploidy and sex of Nv-tra2 dsRNA, Nv-tra dsRNA and water-injected females. Numbers of tested females (P: females (RNAi)) and counts of their offspring (F1: haploid males, F1: diploid females and F1: diploid males) are shown. Mean numbers of haploid and diploid offspring per female are based on these counts.  other functions in early development, as lowered maternal Nv-tra2 provision leads to inviable haploid offspring and (2) the lack of Nv-tra2, and possibly also Nv-tra, maternal mRNA impairs the development of the diploid zygote (Fig. 7). Unfortunately, we cannot distinguish between the causes of the observed inviable offspring and diploid lethality after Nv-tra2 knockdown. Our first conclusion is corroborated by Nissen et al. (2012) who reported effects of tra2 knockdown on embryogenesis in A. mellifera that were not sexspecific and independent of tra regulation. This suggests a similar role of Nv-tra2 in embryogenesis as in A. mellifera, but our experimental approach differs from Nissen et al. (2012). Our pRNAi blocks maternal provision, whereas in A. mellifera dsRNA is injected directly in embryos to silence zygotic transcription. In dipterans and lepidopterans, RNAi studies have not shown any function for tra2 in embryogenesis (Burghardt et al., 2005;Salvemini et al., 2009;Suzuki et al., 2012). However, in T. castaneum (Shukla and Palli, 2013) RNAi with tra2 dsRNA at the larval stage led to developmental arrest, suggesting that tra2 could have acquired additional developmental functions in Hymenoptera and Coleoptera or has lost such functionality in Diptera and Lepidoptera. Our second conclusion may be caused by an additional mechanism. A decreased survival of diploid offspring following Nv-tra pRNAi can also be seen in the data of Verhulst et al. (2010a) where mated mothers produced 44% haploid offspring. This deviates from a normal progeny sex ratio of a single foundress that typically contains about 15% haploid (male) offspring, as observed in the progeny sex ratio of the water injected females of this study. Nv-tra pRNAi does not seem to impact haploid offspring that, under normal developmental conditions, would not produce a functional TRA protein.
Viable diploid males are observed in a polyploid mutant N. vitripennis strain (Whiting, 1960), however, these individuals emerge from unfertilized diploid eggs, and thus do not carry a paternal genome set (Beukeboom et al., 2007). In this study, diploid offspring in both pRNAi classes differ from their haploid brothers in obtaining a paternal chromosome set upon fertilization, hence they are biparental. One possible explanation for the higher mortality of these diploid embryos resulting from Nv-tra or Nv-tra2 pRNAi may be that the absence of the TRA-TRA2 complex is incompatible with the activated genes from the paternal genome, which would then lead to an early developmental arrest, but further research is necessary to understand the cause of the mortality. Offspring counts of (a) virgin and (b) mated females. Virgin females produce only haploid offspring and mated females produce a majority of diploid offspring when hosted individually. The number of eggs laid in a host is plotted in dark grey and the number of adult offspring which emerged from a host is plotted in light grey. The stars indicate the level of significance between the number of embryos and adults of the same treatment: ***P < 0.001. Fig. 6. Fertility of knockdown male offspring. Mean fraction of daughters/total offspring per vial produced by haploid and diploid males of Nv-tra2 pRNAi, Nv-tra pRNAi and waterinjected females. Different letters above the bars indicate significant differences between categories (p < 0.05).

Surviving diploid males do not differ from haploids in splicing and fertility
Despite the reduced viability of diploid Nv-tra pRNAi male offspring and an almost complete inviability of diploid Nv-tra2 pRNAi male offspring, the surviving males do not appear to suffer subsequent effects. Their sex determination is firmly fixed in the male mode with male-specific splicing of both Nv-tra and Nv-dsx. Furthermore, their ability to mate with females and fertilize their eggs does not appear to be impacted. This suggests that only very early processes, possibly at the start of zygotic transcription, result in the observed inviability. Once the male developmental pathway is firmly established, these males appear not to encounter any further costs of their diploid status.

Activation of tra under CSD and MEGISD
The requirement of maternally provided Nv-tra and Nv-tra2 for survival of diploid embryos highlights the importance of maternal effects in the N. vitripennis MEGISD sex determination (Verhulst et al., 2010a). Maternal provision of Nv-tra alone is not sufficient to start the female-specific cascade; Nv-tra2 mRNA is required in the early embryo as well to ensure both female development and proper embryonic development. In the honeybee, RNAi of fem does not lead to noticeable mortality (Beye et al., 2003;Gempe et al., 2009;Hasselmann et al., 2008) and mortality after tra2 embryonic RNAi is independent of ploidy (Nissen et al., 2012). Diploid males resulting from homozygous csd alleles can develop in A. mellifera, but are killed by workers before reaching adulthood (Woyke, 1963). Apparently, silencing this transducing stage of the sex determination cascade has different consequences for the CSD mechanism of A. mellifera than for the MEGISD mechanism of N. vitripennis. Fem F in A. mellifera is not maternally provided contrary to the tra F mRNA provision in N. vitripennis (Gempe et al., 2009;Verhulst et al., 2010a). The female-specific cascade in A. mellifera is activated in the presence of two different csd alleles. N. vitripennis sex determination relies on a silencing mechanism in the mother, or an activating signal in the father. These differently imprinted chromosome sets may result in additional detrimental effects when present in an embryo developing into the opposite sex. Fig. 7. Overview of maternal provision and resulting offspring development in non-manipulated and pRNAi N. vitripennis wasps. Depicted at the top is the normal embryonic development of N. vitripennis; at the bottom is the embryonic development following pRNAi with Nv-tra or Nv-tra2 dsRNA. The ploidy of the embryo is represented by one or two chromosomes, with the maternal chromosome in light grey and the paternal chromosome in dark grey. Maternal provision is indicated by horizontal bars representing mRNA of Nvtra (tra F ) and Nv-tra2 (tra2). Knockdown of maternal provision by pRNAi is portrayed with a dotted outline, while the unaffected maternal provision is indicated with a solid outline. Underneath, the resulting sexual development is stated with the observed viability.