Reprograming of epigenetic mechanisms controlling host insect immunity and development in response to egg-laying by a parasitoid wasp

Parasitoids are insects that use other insects as hosts. They sabotage host cellular and humoral defences to promote the survival of their offspring by injecting viruses and venoms along with their eggs. Many pathogens and parasites disrupt host epigenetic mechanisms to overcome immune system defences, and we hypothesized that parasitoids may use the same strategy. We used the ichneumon wasp Pimpla turionellae as a model idiobiont parasitoid to test this hypothesis, with pupae of the greater wax moth Galleria mellonella as the host. We found that parasitoid infestation involves the suppression of host immunity-related effector genes and the modulation of host genes involved in developmental hormone signalling. The transcriptional reprogramming of host genes following the injection of parasitoid eggs was associated with changes in host epigenetic mechanisms. The introduction of parasitoids resulted in a transient decrease in host global DNA methylation and the modulation of acetylation ratios for specific histones. Genes encoding regulators of histone acetylation and deacetylation were mostly downregulated in the parasitized pupae, suggesting that parasitoids can suppress host transcription. We also detected a strong parasitoid-specific effect on host microRNAs regulating gene expression at the post-transcriptional level. Our data therefore support the hypothesis that parasitoids may favour the survival of their offspring by interfering with host epigenetic mechanisms to suppress the immune system and disrupt development.


Introduction
Parasitic wasps are a polyphyletic group of insects within the order Hymenoptera that have evolved to use other insects as hosts for their offspring. Insects that adopt this so-called parasitoid lifestyle deposit their eggs either on (ectoparasitoid) or in (endoparasitoid) the host [1], and the latter often inject maternal venoms and viruses along with the eggs to suppress host immune responses such as multicellular encapsulation and melanization [2,3]. Idiobiont parasitoids not only suppress host defences but arrest host development when they deposit their eggs, whereas koinobiont parasitoids allow development to continue. Entomopathogenic bacteria and fungi influence the expression of immunity-related and developmental genes in host insects by interfering with epigenetic mechanisms [4,5]. It is plausible that idiobiont parasitoids also interfere with host epigenetic mechanisms to sabotage the immune system and arrest host development, and we therefore set out to test this hypothesis. As a model idiobiont, we selected the pupal endoparasitoid wasp Pimpla turionellae (Hymenoptera, Ichneumonidae). This species parasitizes many lepidopteran insects [6,7] including the greater wax moth Galleria mellonella (Lepidoptera, Pyralidae), an apicultural pest and favoured laboratory model. P. turionellae delivers virulence factors in the form of maternal venom (injected into the host insect along with the eggs) and the anal secretions of wasp larvae. The venom inhibits the encapsulation response in G. mellonella pupae, as well as reducing host cell viability, haemocyte numbers and the mitotic index [8][9][10]. We selected G. mellonella as a host for P. turionellae because it has recently emerged as a powerful model to study the impact of entomopathogenic bacteria and fungi on host insect epigenetics [11][12][13].
To determine the effect of parasitoids on innate immunity, we monitored the expression of 10 immunity-related effector genes by real-time polymerase chain reaction (PCR). We selected the genes encoding cecropin A, cecropin D, X-Tox, galiomycin and gloverin because they encode antimicrobial peptides that defend G. mellonella against bacteria [14][15][16][17], and the gene encoding c-type lysozyme, which is active against Gram-positive bacteria and fungi [16]. We also selected the gene encoding insect metalloproteinase inhibitor (IMPI), which inhibits virulence-associated microbial metalloproteinases [18,19]. Finally, we selected the genes PPO1 and PPO2 (encoding prophenoloxidases 1 and 2) and PO2 (encoding phenoloxidase 2), which mediate the production of melanin associated with multicellular defence reactions targeting parasitoid eggs [9,20,21]. To determine the effect of parasitoids on host development, we monitored the expression of 10 genes related to the juvenile hormone (JH) and edysteroid hormone signalling pathways [22,23].
To determine whether parasitoids influence epigenetic mechanisms in G. mellonella, we evaluated the levels of DNA methylation and histone acetylation/deacetylation, which regulate the initiation of transcription. The conversion of DNA cytidine residues to 5-methylcytidine influences the ability of DNA to interact with proteins, thus providing a mechanism for gene regulation [4,24]. The impact of parasitoids on host DNA methylation was determined by comparing the amount of 5methylcytidine monophosphate (5 m-dCMP) released from the genomic DNA of infested pupae and controls at different time points following the injection of wasp eggs. We also measured the expression of epigenetic marker genes encoding two components of the DNA methyltransferase complex, three histone acetyltransferases (HATs) and four histone deacetylases (HDACs), whose opposing activities are tightly regulated in G. mellonella [25], as well as measuring the levels of histone acetylation directly. Finally, we measured the expression of microRNAs (miRNAs), which regulate target genes at the posttranscriptional level [26]. These short, noncoding RNAs, 18-24 nucleotides in length, inhibit the translation of specific mRNAs by base-pairing with the untranslated regions or occasionally the coding region, and they play a role in insect antiviral responses [27]. We recorded the expression 603 miRNAs using a microarray spotted with more than 2000 insect miRNA probe sequences [28] which have been used to study entomopathogenic bacteria and fungi in G. mellonella [11][12][13].

Material and methods (a) Insect rearing
Stock cultures of G. mellonella and P. turionellae were maintained in the laboratory as previously described [29]. Last-instar G. mellonella larvae, each weighing 200-250 mg, were reared to the pupal stage. We presented the G. mellonella pupae (1-2 days after the onset of pupation) to P. turionellae adults for egg laying and further development into adults. Adult parasitoids were collected and held in glass jars without a host and fed on 40% (v/v) honey solution at 25 ± 2°C and 60% relative humidity, with a 12 h photoperiod.

(b) Parasitization technique
Male and female P. turionellae adults were reared in glass jars containing G. mellonella pupae. P. turionellae can lay eggs in 1-3 pupae per day, and we used the first parasitized pupa for further investigation. G. mellonella pupae were recovered from the jars after 4, 8, 24 and 96 h of parasitization for the isolation of nucleic acids and histone proteins. Non-parasitized pupae were used as negative controls, and pupae injected with Sephadex A25 beads 40-120 µm in diameter (Sigma-Aldrich, Taufkirchen, Germany) were used as positive controls because Sephadex induces encapsulation and melanization in G. mellonella larvae [30]. Three pupae representing each sample were pooled for further analysis.
(c) DNA isolation and global DNA methylation analysis DNA was isolated from parasitized G. mellonella pupae and controls as previously described [31]. DNA integrity was determined by 1% agarose gel electrophoresis and by measuring the absorbance ratios 260/280 nm and 260/320 nm by spectrophotometry. High-quality DNA (1250 ng per reaction) was hydrolysed to nucleotides with DNA Degredase (Zymo Research, Orange, CA, USA). All samples were passed through a 0.2 µm nylon Corning Costar Spin-X centrifuge tube filter (Sigma-Aldrich) before injection into the high-performance liquid chromatography (HPLC) column. Isocratic reversed-phase HPLC analysis was carried out using a PerkinElmer (Waltham, MA, USA) P200 instrument fitted with an ACE 5 C18 (125 × 4 mm) column [32]. The mobile phase, 50 mM ammonium orthophosphate (pH 4.1), was filtered (0.2 µm) and degassed thoroughly before use. For each reaction, we injected 1000 ng of DNA in a volume of 20 µl at a flow rate of 0.75 ml min −1 . To measure global DNA methylation, we used 5 m-dCMP as a standard (Cayman Chemicals, Ann Arbor, MI, USA; electronic supplementary material, figure S1). The global DNA methylation level was calculated as the amount (ng) of 5 m-dCMP in 1 µg of genomic DNA, detected at 280 nm.

(d) RNA isolation and real-time polymerase chain reaction
Total RNA was isolated using TRI reagent (Sigma-Aldrich) and 1 µg of total RNA was reverse transcribed using the QuantiTect kit (Qiagen, Hilden, Germany). The cDNA was amplified by real-time PCR using a CFX96 instrument (Bio-Rad Laboratories, Hercules, CA, USA) and the Sensi Fast SYBRgreen kit (Bioline, Luckenwalde, Germany) according to the manufacturer's instructions. The target genes were amplified using the primers listed in electronic supplementary material, table S1. Each reaction was heated to 95°C for 15 min, followed by 39 cycles of denaturation at 94°C for 15 s, annealing at 56°C for 10 s and extension at 70°C for 30 s, then a final extension step at 65°C for 5 s. The 18S rRNA housekeeping gene was used for normalization [25], and the fold-change in gene expression was calculated using the 2 ΔΔCT method [33]. We estimated a primer efficiency of approximately 1 for all genes, including the 18S rRNA gene, given the similar curves in the logarithmic PCR amplification plots recorded by CFX Manager (Bio-Rad). This method to determine primer efficiency avoids the need to calculate separate efficiencies for a large number of genes in separate experiments and is estimated during analysis [34].
royalsocietypublishing.org/journal/rspb Proc. R. Soc. B 287: 20200704 (e) Histone protein isolation and acetylation analysis Histone proteins were isolated using the Total Histone Extraction kit (Epigentek, Farmingdale, NY, USA). The protein yield was measured using the Pierce BCA Protein Assay kit (Thermo Fisher Scientific, Waltham, MA, USA) with bovine serum albumin as the standard. The percentage acetylation of histones H3 and H4 (H3K9, H3K14, H4K5 and H4K12) was measured using the Epigentek colorimetric detection kit according to the manufacturer's instructions. In each experiment, 200 ng of extracted histone protein was used and the absorbance was measured at 450 nm using a Synergy H4 Hybrid microplate reader (Biotek, Winooski, VT, USA).

(f ) Microarray analysis of miRNAs
The expression profiles of conserved miRNAs in parasitized G. mellonella pupae and controls were monitored by microarray analysis. This is a high-throughput and relatively inexpensive technique for the analysis of miRNA expression although the low signal-to-noise ratio can generate false-positive and falsenegative results. Microarrays are often used to analyse miRNA expression in insects that lack annotated genome sequences [11,12]. An oligonucleotide microarray containing 2621 unique mature arthropod miRNA sequences was designed using miR-Base v. 21. The miRNA sequences were obtained from the fruit fly Drosophila melanogaster (1735 miRNAs), the pea aphid Acyrthosiphon pisum (103 miRNAs), the silkworm Bombyx mori (560 miRNAs), the red flour beetle Tribolium castaneum (422 miRNAs) and the western honeybee Apis mellifera (259 miRNAs), with some redundancy across species [28]. To determine changes in miRNA expression profiles during parasitization, total RNA was isolated from parasitized G. mellonella pupae and controls 4 h after egg deposition as described above. The RNA samples were pooled from three parasitized and three non-parasitized pupae. Microarray hybridization, detection and analysis were carried out by LC Sciences (Houston, TX, USA) as previously described [28].

(g) Prediction of miRNA targets
Target mRNAs were identification in silico by complementarity to miRNA probes using the sequence alignment editor BioEdit v. 7.2.5 as previously described [11]. Briefly, we identified open reading frames (ORFs) in all contigs in the sequenced G. mellonella transcriptome using the 'Find next ORF' option in BioEdit. Nucleotide sequences at the 3 0 end of individual contigs but outside confirmed ORFs were considered as potential 3 0 UTRs, and were screened for complementarity (seed sequence complementarity) with the expressed miRNA sequences identified by microarray analysis. Expressed miRNAs were defined as those for which the average microarray signal was above background in at least two different pools of the same treatment group. The Gene Ontology categories of the identified contigs were listed by consulting the UniProt database and a previous report [22]. The biological processes targeted by miRNAs in the parasitized and non-parasitized pupae were summarized using Cytoscape v. 3.2.1. The identified miRNA targets were validated using the RNA hybrid tool provided by the Bielefeld Bioinformatics Server v. 32 [35].

(h) Statistics
All experiments except the microarray analysis were carried out at least three times (biological replicates) with statistical analysis using SPSS v. 18.0 (SPSS, Chicago, IL, USA). Each parameter was analysed using an independent samples test with a significance threshold of p < 0.05. However, for the analysis of miRNAs, log2 values were assessed with significance threshold of p < 0.01.

Results
(a) Parasitoid-dependent transcriptional reprogramming of immunity-related genes and hormonal pathway genes in Galleria mellonella pupae We compared the expression levels of 10 immunity-related genes in parasitized G. mellonella pupae and controls by realtime PCR (figure 1). We found that all 10 genes were strongly downregulated at most time points after egg deposition (4, 8, 24 and 96 h) but that the genes naturally fell into two groups based on their expression profiles. The first group comprised those genes showing transient upregulation at 24 h (galiomycin, cecropin A and IMPI, with IMPI showing a particularly striking sixfold induction) and those following the same trend in which the suppression was transiently lifted, but not quite enough to reach basal levels (X-Tox and lysozyme). The second group comprised those genes that were downregulated throughout the period of parasitization (gloverin, cecropin D, PPO1, PPO2 and PO2).
We also compared the expression levels of 10 hormonerelated genes involved in development. Eight of the genes encoded components of the JH pathway, namely JH-inducible (JH-Ind), JH-binding proteins 1-4 (JH-BP1, JH-BP2, JH-BP3 and JH-BP4), JH epoxide hydrolases 1 and 2 (JH-EHyd1 and JH-EHyd2) and JH esterase (JH-Est). The last two genes encoded ecdysteroid hormone 22-kinase (ECD-22Ki) and ecdysteroid-regulated protein (ECD-RP). As above, we compared parasitized G. mellonella pupae and controls at four time points (4,8,24 and 96 h) after egg deposition (figure 2). The expression profiles of the hormone-related genes were more complex than the immunity-related genes, with more diverse responses. JH-Ind, JH-EHyd2 and ECD-RP formed a group that showed general downregulation or minimal change but a spike of induction after 8 h, whereas JH-BP1, JH-EHyd1 and JH-Est followed the profile of the first set of immunity-related genes, with general downregulation but a spike of induction (or partial recovery from repression) after 24 h. JH-BP2 was strongly induced after egg deposition followed by a return to basal expression after 96 h. JH-BP3 was similar to JH-BP2, but the return to basal levels occurred after 24 h and was followed by strong repression. ECD-22Ki also fitted this profile, but the switchover between induction and repression occurred somewhere between the 8 h and

(d) Parasitoid-dependent effects on miRNA expression in Galleria mellonella pupae
We investigated the expression profiles of 603 candidate miRNAs and selected 82 for further analysis based on their moderate or strong modulation. We found that 58 of these miRNAs were upregulated and 24 were downregulated in parasitized G. mellonella pupae compared to controls ( p < 0.01), with many of the upregulated miRNAs showing moderate (<twofold) induction but several of the downregulated sequences showing much stronger effects. The most extreme was bmo-miR-3365, which was downregulated 14-fold in the parasitized pupae (electronic supplementary material, figure S3 and S4). The mRNA targets of selected miRNAs were tested against the G. mellonella transcriptome [22] as previously described [28]. We found that these miRNAs targeted genes with important roles in host-parasitoid interactions, such as defence responses and host development (electronic supplementary material, figure S5).

Discussion
Parasitoids have evolved a successful reproductive strategy in which they use the larvae or pupae of host insects to provide a source of nutrition for their offspring [1]. Koinobiont parasitoids feed on the host insect as it develops, but idiobiont parasitoids arrest host insect development and the underlying mechanisms are not yet understood. However, the host is not defenceless against parasitoid eggs and can attack them with various defence responses, including melanization and multicellular encapsulation [36][37][38]. The coevolution of parasitoids and their hosts has led to counterstrategies in parasitoid wasps to circumvent such host defence reactions. For example, P. turionellae not only arrests the development of G. mellonella pupae but can also suppress the host immune system [8][9][10]. We hypothesized that P. turionellae achieves this by subverting the epigenetic mechanisms of the host in order to effect the transcriptional reprogramming of immunity-related and developmental genes, thus sabotaging both host defence responses and normal development. Accordingly, we monitored the expression of 10 immunity-related genes and 10 genes involved in the hormonal control of development in G. mellonella pupae parasitized by P. turionellae. The expression profiles of the 10 immunity-related genes confirmed their profound repression 4 and 8 h after egg deposition, but five of the genes mounted a partial or complete transient recovery at the 24 h time point before succumbing to repression thereafter. Three of these genes (encoding cecropin A, X-Tox and lysozyme) recovered to the point of restoring near normal basal expression (equivalent to the expression level in control larvae), whereas the genes encoding galliomycin and IMPI showed a strong spike of induction at 24 h, with a mean sixfold upregulation in the case of IMPI. Interestingly, IMPI is part of an immunity-related signalling cascade that does not require microbial pathogen-associated molecular patterns (PAMPs) to elicit the synthesis of antimicrobial peptides, but senses the presence of danger-associated molecular patterns (DAMPs) [39,40]. Microbial metalloproteinases can digest targets such as collagen type IV in insects [41], generating peptide fragments that operate as DAMPs [42], which in turn activate immune responses including the synthesis of IMPI, galiomycin and cecropin A [43]. IMPI is the only known defence peptide that can inhibit virulenceassociated metalloproteinases [44]. Similar to parasitic fungi that produce metalloproteinases as virulence factors, leading to the activation of innate immune responses in G. mellonella [43], we postulate that P. turionellae injects metalloproteinases  royalsocietypublishing.org/journal/rspb Proc. R. Soc. B 287: 20200704 along with its eggs into the infected host insect, which could elicit a strong immune response if the parasitoid does not block the corresponding transcriptional reprogramming. Indeed, the venom glands of P. turionellae and another endoparasitoid (Toxoneuron nigriceps) are known to synthesize a metalloproteinase [45,46]. The transient nature of the recovery we observed for IMPI, galliomycin, cecropin A, X-Tox and lysozyme, followed by strong suppression at the 96 h time point, suggests that this defence response is actively targeted and overcome by the parasitoid eggs. Furthermore, the silencing of genes encoding PPOs and PO in the parasitized pupae indicates that the parasitoids have also evolved a counterstrategy to protect the eggs from melanization. We also found that the deposition of eggs by P. turionellae into G. mellonella pupae caused the profound transcriptional reprogramming of host genes representing the JH and ecdysteroid hormone pathways, which play important roles in lepidopteran growth and development. In the corpora allata, secreted juvenile hormones are carried to their target tissues by JH-BPs 1-4, which protect the hormones from nonspecific esterases and JH epoxide hydrolase activity [47][48][49][50]. The upregulation of JH-BP2 at all time points (up to 20-fold) and JH-BP3 until the 24 h time point (up to 420-fold) indicated that JH signalling increases after parasitization, as already reported in other host-parasitoid systems [51,52]. The transformation of lepidopteran pupae into adults requires high levels of ecdysteroid hormones in the absence of JH [53,54]. Our data show that parasitization triggers the profound downregulation of enzymes that degrade JH (JH-Ehyd1 and JH-Est at all time points, and JH-Ehyd1 at later time points) as well as ECD-RP at all time points. Furthermore, parasitization also triggers the transient upregulation of ECD-22Ki (until the 24 h time point), which phosphorylates and thereby inactivates ecdysteroid hormones. These combined effects would help to explain the developmental arrest of parasitized G. mellonella pupae and agree with earlier reports showing that wasp venom can inhibit the degradation of JH, while suppressing the accumulation of ecdysteroid hormones [52]. JH can suppress the immune response in D. melanogaster by blocking the ability of 20hydroxy-ecdysone to stimulate the production of antimicrobial peptides [55,56]. Together, these data indicate that parasitoids interfere with hormonal regulation to arrest host development and to compromise innate immunity.
The overall goal of our study was to determine whether the parasitoid-dependent transcriptional reprogramming of immunity-related and developmental genes in the host involves epigenetic mechanisms. We therefore compared the levels of DNA methylation and histone acetylation in the parasitized pupae and controls, as well as measuring the level of selected miRNAs. The global DNA methylation    royalsocietypublishing.org/journal/rspb Proc. R. Soc. B 287: 20200704 level in control pupae was 30 ng 5 m-dCMP per µg genomic DNA (3% total methylation) but the deposition of eggs by P. turionellae caused a transient decline, with a minimum value of approximately 35% below normal after 4 h, followed by a recovery to approximately 33% below normal after 8 h and approximately 5% below normal after 24 h. This matches the transient recovery of the five immunity-related genes discussed above as well as two genes directly related to DNA methylation (encoding DNA cytosine 5-methyltransferase and DNA methyltransferase-1 associated protein, respectively). The methylation-related genes also showed a moderate recovery or at least no significant increase in downregulation after 24 h, before the recovery was quashed and stronger suppression became evident after 96 h. The loss of DNA methylation was also observed in diamondback moth (Plutella xylostella) larvae following the deposition of eggs by the parasitoid wasp Cotesia plutellae, commensurate with the suppression of genes encoding DNA methyltransferases 1 and 2 [57]. Therefore, it appears that parasitoids may trigger global DNA demethylation in the host in order to suppress immunity-related and developmental genes that require the presence of methylated sites for full expression.
The impact of P. turionellae on the epigenetic mechanisms of G. mellonella was also evident at the level of histone acetylation. Six of seven genes encoding HATs and HDACs were silenced in pupae following the deposition of parasitoid eggs (figure 4). The exception was the gene encoding HDAC 8 isoform 2, which was initially strongly induced before falling back towards basal levels after 24 h and succumbing to strong repression (compared to control pupae) after 96 h. The transcriptional activation of HDACs leads to histone deacetylation and the suppression of gene expression. The transient transcriptional activation of HDAC 8 isoform 2 may therefore trigger the downregulation of innate immunityrelated genes via the deacetylation of H3K14, H3K9 and H4K12 [5]. The sabotage of histone acetylation by the parasitoid offers another plausible mechanism for the suppression of immunity-related and developmental genes in the host insect because this epigenetic mechanism plays a key role in the regulation of transcriptional reprogramming during metamorphosis and infection in G. mellonella [27].
Finally, we identified 24 miRNAs that were downregulated in parasitized G. mellonella pupae, some strongly, and 58 that were moderately upregulated (electronic supplementary material, figure S4). The expression of miRNA genes in insects is known to change during development [58,59] and is also modulated by immune challenges [28] or parasitization [60]. We identified seven G. mellonella miRNAs that were strongly suppressed by parasitization, ranging from threefold to 13fold downregulation (bmo-miR-989a, tca-miR-989-3p, ame-miR-989, dme-miR-989-3p, bmo-miR-989b, dps-miR-989, and mse-miR-989). Interestingly, all of them appear to play a role in the regulation of cell migration, which is a key developmental process as well as an important part of the innate immune response [61]. For example, miR-989 is normally expressed in somatic cells of the D. melanogaster ovary and its absence delays their migration and arrests oogenesis [62]. Additionally, border cell migration in D. melanogaster involves several conserved signalling pathways such as the JAK/STAT pathway [62], which coordinates cytokine-dependent immune responses and regulates homeostasis via multiple mechanisms [63]. The parasitization of D. melanogaster has been shown to suppress the secretion of JAK/STAT pathway ligands by haemocytes [63]. These data suggest that P. turionellae venom may suppress the expression of miRNAs to inhibit cell migration as a component of development and immunity. We identified a larger number of G. mellonella miRNAs that were slightly upregulated by parasitization, including mja-miR-6492, dme-miR-263a-5p, ame-miR-3796, tca-miR-2c-5p and ame-miR-3756, which are known to target immunityrelated and developmental genes (electronic supplementary material, figure S4). Interestingly, the parasitoid wasp Cotesia vestalis is known to produce miRNAs in its venom and teratocytes, which are introduced in the host during oviposition and in order to modulate host genes [64]. Accordingly, it is unclear whether the upregulated miRNAs we detected in the parasitized G. mellonella pupae are derived from the host or the parasitoid. Among the miRNA induced most strongly in the parasitized pupae was api-miR-124, which controls neural plasticity and transient memory in honeybees [65]. Similarly, miR-124 controls gene expression in the sensory nervous system of the nematode worm Caenorhabditis elegans [66]. Therefore, P. turionellae may use miR-124 to modulate neural gene expression in the host.
In summary, we found that the parasitoid wasp P. turionellae interferes with the epigenetic mechanisms of its host, which may help to sabotage the host immune system and arrest its development. All three major epigenetic mechanisms are affected, resulting in the global depletion of DNA methylation, changes in the levels of specific acetylated histones and the modulation of miRNA expression, in line with the observed transcriptional reprogramming of immunity-related and developmental genes. Further work should focus on the components of the venom injected into the host during oviposition, which may reveal the identity of venom compounds that directly interfere with the epigenetic mechanisms in the host.