Downloaded from http://rspb.royalsocietypublishing.org/ on April 12, 2017

Transcriptomics of an extended rspb.royalsocietypublishing.org phenotype: parasite manipulation of social behaviour shifts expression of caste-related genes Research Amy C. Geffre1, Ruolin Liu2, Fabio Manfredini4, Laura Beani5, Cite this article: Geffre AC, Liu R, Manfredini Jeyaraney Kathirithamby6, Christina M. Grozinger7 and Amy L. Toth1,3 F, Beani L, Kathirithamby J, Grozinger CM, Toth AL. 2017 Transcriptomics of an extended 1Department of Ecology, Evolution, and Organismal Biology, 2Department of Electrical and Computer phenotype: parasite manipulation of wasp Engineering, and 3Department of Entomology, Iowa State University, Ames, IA, USA 4 social behaviour shifts expression of School of Biological Sciences and Centre for Systems and Synthetic Biology, Royal Holloway, University of London, London, UK caste-related genes. Proc. R. Soc. B 284: 5Department of Biology, University of Florence, Florence, Italy 20170029. 6Department of Zoology, University of Oxford, Oxford, UK 7 http://dx.doi.org/10.1098/rspb.2017.0029 Center for Pollinator Research and Department of Entomology, Pennsylvania State University, State College, PA, USA FM, 0000-0002-9134-3994; CMG, 0000-0001-5738-1201; ALT, 0000-0003-0824-0933

Received: 5 January 2017 Parasites can manipulate host behaviour to increase their own transmission Accepted: 13 March 2017 and fitness, but the genomic mechanisms by which parasites manipulate hosts are not well understood. We investigated the relationship between the social paper wasp, dominula, and its parasite, vesparum (Insecta: ), to understand the effects of an obligate endoparasitoid on its host’s brain transcriptome. Previous research suggests that X. vesparum shifts Subject Category: aspects of host social caste-related behaviour and physiology in ways that Behaviour benefit the parasitoid. We hypothesized that X. vesparum-infested (stylopized) females would show a shift in caste-related brain gene expression. Specifically, Subject Areas: we predicted that stylopized females, who would normally be workers, would behaviour, evolution, genomics show gene expression patterns resembling pre-overwintering queens (gynes), reflecting gyne-like changes in behaviour. We used RNA-sequencing data to characterize patterns of brain gene expression in stylopized females and Keywords: compared these with those of unstylopized workers and gynes. In support host–parasite interactions, social wasp, of our hypothesis, we found that stylopized females, despite sharing numer- parasite manipulation, social caste, gene ous physiological and life-history characteristics with members of the expression, eusociality worker caste, show gyne-shifted brain expression patterns. These data suggest that the parasitoid affects its host by exploiting phenotypic plasticity related to social caste, thus shifting naturally occurring social behaviour in a way that is beneficial to the parasitoid. Authors for correspondence: Amy C. Geffre e-mail: [email protected] Amy L. Toth 1. Background e-mail: [email protected] is a widespread and highly successful life-history strategy. Parasites can reduce the fitness of their hosts, often causing profound changes in host physiology and behaviour [1–3]. In some cases, referred to as ‘adaptive host manipulation’, phenotypic changes in the host may benefit the parasite [4]. There are several striking cases in which parasites induce unusual and novel behaviours that are not typically part of the host’s normal behavioural reper- toire [5]. Alternatively, parasites may cause more subtle alterations to host behaviour, such as changes in frequency or timing of otherwise normal behav- iour [6,7]. There are many open questions about the mechanisms by which Electronic supplementary material is available parasites accomplish host behavioural manipulation, and how novel parasite- induced host behaviours evolve. Advances in genomics offer an exciting online at https://dx.doi.org/10.6084/m9. opportunity to examine the mechanistic underpinnings and evolution of figshare.c.3726622. adaptive host manipulation [8].

& 2017 The Author(s) Published by the Royal Society. All rights reserved. Downloaded from http://rspb.royalsocietypublishing.org/ on April 12, 2017

Phenotypic plasticity is an important driver of evolutionary 2 change and phenotypic novelty [9]. One strategy a parasite 7 r s might take to shift its host’s phenotype would be to exploit p 6 b . r existing host phenotypic plasticity (and associated genetic 1 o y a l pathways) in ways that benefit the parasite. This has the poten- s o c i

tial to be an efficient strategy for parasite manipulation of host e

A t y phenotypes. Previous studies have documented a wide array of 2 p u b

host behavioural, physiological and gene expression changes l i B s h associated with parasitization [2,10–15]. However, few studies i n g . have examined changes in host gene expression through the E/5 o

3 r lens of adaptive parasite manipulation, also viewed as an g

‘extended phenotype’ of the parasite’s genome [16]. Because P r o

C c phenotypic plasticity is the result of shifts in gene expression, D . R one powerful way to examine how parasites affect their hosts . S o is using transcriptomics, which is now easily applicable to c . ecological model systems [17]. Although previous studies 4 B 2 have examined the effects of parasites on their hosts’ transcrip- 8 4 :

tomes [4,12–15], none have specifically examined how and if 2 0 1

parasites influence host behaviour via hijacking inherent host 7 0 0

phenotypic plasticity. 2

Figure 1. Polistes dominula host and Xenos vesparum parasite life cycle. 9 To begin to explore the hypothesis that parasites can Stylopized indicated with red outline. Letters indicate P. dominula life manipulate hosts by shifting plastic phenotypes, we focus on cycle: (A) founding phase, (B) worker phase, (C) reproductive phase, (D) decline transcriptome-wide gene expression in the well-studied case and (E) overwintering aggregation. Numbers indicate X. vesparum life cycle: of paper wasp hosts, Polistes dominula, and their parasites, (1) first-instar X. vesparum larvae infect host P. dominula larvae in the nest, Xenos vesparum (Insecta: Strepsiptera). This system affords an (2) endoparasitic larvae develop inside host pupa, (3) male cephalotheca/ exciting opportunity to study mechanisms and evolution of female cephalothorax extrudes from between-host tergites, (4) aberrant aggre- host manipulation for several reasons [18,19]. First, P. dominula gations: male X. vesparum emerge as free-living adults and mate, then die; colonies are often heavily parasitized by X. vesparum, which are neotenic females remain endoparasitic in hosts, (5) stylopized wasps joined obligate endoparasitoids, meaning that they must develop by unstylopized gynes to overwinter, (6) stylopized wasps leave aggregation within the host and also lead to reproductive death via cas- after unstylopized gynes have begun founding, (7) stylopized wasps forage tration [20,21]. Second, P. dominula are a model with and visit conspecific nests, but do not found nests themselves. Female substantial behavioural and developmental plasticity in the X. vesparum drops first-instar larvae directly on nests or on flowers (drawing form of flexible social castes [22,23]. Third, X. vesparum have not to scale, by A.C. Geffre) (Online version in colour.) well-documented effects on their hosts, including behavioural and physiological changes that suggest they create ‘wrong caste’ phenotypes in hosts (described below) [18,19]. Fourth, endoparasitic, with an extruded cephalothorax with no eyes P. dominula have new genomic data available, including a or appendages [20,21,39]. genome and caste-related transcriptome [18], and identified A parasitized female wasp (hereafter ‘stylopized female’, genes and pathways related to caste differences [24–29]. after suborder Stylopidia) undergoes a dramatic physio- The typical colony cycle of temperate Polistes wasps involves behavioural deviation [40] (figure 1; summarized in electronic several phases, and in each, there are distinct groups of females supplementary material, table S1), including loss of ovaries, showing extensive phenotypic plasticity in physiology and be- asocial behaviour and prolonged lifespan, when infested. haviour (figure 1). Colonies are founded in the spring by one Early emerging stylopized females desert their natal colonies or a few cooperating founding females, who build the nest and in mid-summer, during peak colony activity, and form extrani- rear a first set of offspring, and all females, who become workers dal aggregations with stylopized females from other colonies upon adult emergence [30]. Upon worker emergence, the found- [19,41]. This aberrant host behaviour occurs close to the time ing female becomes the primary reproductive (queen) of the of male parasitoid emergence and appears to help facilitate colony, while the workers perform tasks related to brood care mate location by the parasitoids [38]. Later, stylopized females and colony growth [31]. Later in the colony cycle, larvae are are joined by unstylopized aggregating gynes, forming reared by workers and emerge as males or female ‘gynes’— overwintering clusters. Stylopized females remain with the non-working pre-overwintering queens with large fat stores unstylopized gynes until the next spring when the latter [32]. Gynes will leave the colony in fall to form extranidal aggre- become nest-founding queens [19,40,41]. The stylopized gations with other gynes, where they overwinter until they females do not found nests, and instead, transmit parasitoid disperse to found new colonies the following spring [30]. larvae by visiting other newly founded P. dominula nests or Xenos vesparum infest P. dominula larvae in various host via shared floral forage sites [33,36,37]. larval instars in multiple colony phases [33], including early This study focuses on the unusual nest desertion and aggre- in the season when the founding queen is laying worker- gation behaviour of stylopized females. Because the stylopized destined eggs [33–37] (figure 1). Xenos vesparum remains females described here typically emerge in early summer, they endoparasitic until after the host emerges as an adult [36]. should be workers, engaging in colony work such as nest Then, X. vesparum extrudes its anterior region between the building, foraging and brood care. However, they exhibit be- host’s abdominal segments. The male parasitoid emerges as a haviour that is typical of unstylopized gynes—lack of nest fully motile, winged adult that will disperse to mate [21,33,38], work, nest desertion and formation of pre-overwintering whereas the female remains neotenic, living permanently aggregations. Despite our detailed understanding of the Downloaded from http://rspb.royalsocietypublishing.org/ on April 12, 2017

behavioural effects of stylopization, the molecular mechanisms library preparation with barcoding. We performed two rounds of 3 underlying this behavioural deviation are unknown. oligo(dT) selection of the poly(A) RNA with the MicroPoly (A) r

TM s Here, we provide transcriptomic data underlying the Purist Kit (Ambion LifeTech). cDNA libraries were prepared p b .

starting from 100 to 500 ng poly(A) RNA and using the SOLiD r extended phenotype of X. vesparum parasitoids in the context o y

Total RNA-Seq Kit according to the manufacturer’s instructions a l of manipulation of their P. dominula hosts’ behaviour. We used s o

(Applied Biosystems /Life Technologies). Sequencing was per- c i

high-throughput RNA-sequencing to measure global brain e

formed on the Applied Biosystems (Life Technologies) SOLiDv4 t y gene expression patterns in unstylopized workers, unstylopized p according to the manufacturer’s instructions. All samples were u b

gynes in pre-winter aggregations and early emerging stylopized l i

loaded into a single flow cell of a SOLiD 5500xl Sequencer, produ- s h females. Importantly, we measured gene expression in stylo- i cing 76 base single-end reads (total number of reads per sample n g . pized females before they deserted their nests; thus, we were listed in electronic supplementary material, table S2). o r able to capture gene expression patterns that preceded the full To minimize the noise from sequencing mistakes, raw reads were g expression of the parasitoid-induced behaviour and are more processed using the Perl script (SOLiD_preprocess_filter_v2.pl). P r o

Reads with quality scores below 10 were discarded. A comparison c likely to be causal to, rather than a consequence of, the shifted . R behaviour. First, we hypothesized that the overall pattern of between two popular colour space alignment tools, BFAST and . S

SHRiMP, led us to use BFAST for improved mapping rates (elec- o brain gene expression in stylopized females would be shifted c . from towards gyne-like patterns. Second, we also hypothesized tronic supplementary material, table S2). Using BFAST, we first B

built two reference genomes both in colour space and in nucleotide 2 that the parasitoid-induced shift in gene expression would 8 4

space. Then, we created indexes for the reference genome in colour :

involve caste-related genes, i.e. those that show differential 2

space and found candidate alignment locations (CALs). Next, 0 1

expression between unstylopized workers and gynes. 7

local alignments were performed in CALs using the P. dominula 0 0

reference genome [24] in nucleotide space. Finally, low-quality 2 9 2. Material and methods alignments were filtered by a post-processing step. (a) Sample collection (d) Analysis of gene expression Wasps were collected at three sites in Tuscany, Italy, including We used HTSeq (http://www-huber.embl.de /users /anders/ Siena, Impruneta and Sesto, in the spring and summer of 2009. HTSeq/) for counting the number of reads that mapped to 11 506 Upon collection, wasp heads were immediately removed and transcripts based on the P. dominula genome annotation r1.0 [24]. placed in RNAlater q (Life Technologies, AM7023), stored at 48C These counts were used by edgeR for differential expression analy- for 24 h (as per the manufacturers’ instructions), then transferred ses. We filtered low expressed transcripts by the following criterion. to 2208C and kept frozen until brain dissections. Bodies were We calculated transcript CPM (counts per million mapped reads) for frozen for later dissection. We collected three groups of females: each sample and preserved transcripts which satisfied CPM . 3 in 21 unstylopized workers from nine nests in June (herein referred at least three samples. By using 3 as a cut-off, 8484 (approx. 74%) to as ‘Workers, W’); 20 unstylopized gynes from pre-overwintering transcripts remained for the differential expression analyses. aggregations in August (Gynes, G) and 15 stylopized females from Raw read counts were analysed with R using the edgeR pack- six nests in June (Stylopized, S). Stylopized females were collected age from Bioconductor [44], using the standard protocol outlined from a subset of the same nests as workers. in the edgeR user guide, except we used a more stringent CPM cut-off. The edgeR user guide recommends keeping a gene when at least two samples with CPM . 1. However, considering our (b) Dissections and RNA extractions relatively high sample size and read counts, we set our threshold Abdomens were removed from the wasps’ bodies and dissected. for keeping a gene as at least three samples with CPM . 3. Nor- To identify caste status and parasitoid presence, the following malization was performed using the TMM (Trimmed Mean of information was recorded for each wasp: ovary development M-values) method. First, we wished to examine which of the score [42], fat body size score [42] and presence of X. vesparum, three wasp phenotypes had distinct expression profiles versus all including number, developmental stage and sex of the parasitoids of the others, and also which phenotype explained most of the [43]. We selected 15 wasps that were stylopized by at least one difference in global gene expression. For this purpose, we adopted neotenic female X. vesparum (although some were stylopized by a generalized linear model (GLM) fit to the count data and ident- more than one female or a female and a male). ified differentially expressed genes using planned linear contrasts, Brains from each individual were dissected in RNAlater q at room as described in Mikheyev & Linksvayer [45]. Second, we analysed temperature and then stored at 2808C until RNA extraction. Total focal pairwise comparisons to identify groups of genes that were RNA from each brain was extracted using the RNeasy q Mini Kit significantly differentially expressed between wasp phenotypes. (Qiagen, 74 104), including a 15 min DNase I (Qiagen, 79 254) step We compared the following wasps: workers versus gynes, stylo- before ethanol washes. RNA was quantified using a NanoDrop pized versus workers and stylopized versus gynes. For this 1000 q (Thermo Fisher Scientific) spectrophotometer and Qubit approach, read counts were log transformed and corrected for Fluorometer (Life Technologies). Samples with 260 : 230/260 : 280 , 2 skew. We detected differential levels of gene expression using a 1.9 were not included in the transcriptomic analysis. A subset of modified Fisher’s exact test, which took into account the dispersion samples were quantified by Bioanalyzer (Agilent), which also and the multiple samples. Finally, raw p-values for each gene were served to verify quality based on RNA integrity numbers. Following corrected for multiple comparisons with an false discover rate quantification, RNA was stored at 2808C until later use. One set was (FDR) cut-off of 0.05. The HTSeq raw read counts for the 11 506 allocated exclusively for transcriptome sequencing and another for transcripts, the R scripts for processing the data and creating the qRT-PCR validation. RNA from 14 pooled samples were used for Venn diagram, data files for the condition-averaged raw reads sequencing (five gynes, five workers and four stylopized). The com- counts for the 8484 genes and FDR values for each pairwise com- position of each pooled sample varied slightly, from two to four parison are available on the Github repository (https://github. individual brains, to provide enough RNA for sequencing. com/ruolin/ WaspsProject). For global analyses of gene expression, we used hierarchical (c) RNA-sequencing clustering (Ward method) in JMP PRO v. 10.0 (SAS, Cary, NC) At the Pennsylvania State University Genomics Core Facility, each and principal component analysis in R. To perform Gene Ontology of the pooled RNA samples was used for a standard SOLiD mRNA (GO) analyses, we obtained Drosophila melanogaster orthologues Downloaded from http://rspb.royalsocietypublishing.org/ on April 12, 2017

with BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) for all of stylopized females were also apparent from a principal com- 4 P. dominula transcripts, including those that were significantly ponent analysis on the 367 DETs (see electronic supplementary r s differentially expressed between treatments (see electronic sup- material, figure S1). p b . plementary material, table S4 for a complete list). We used GO r Overall, gynes were the most divergent group in terms o y functional annotation clustering in DAVID v. 6 [46,47] with a l

of brain gene expression. Gynes had a total of 282 DETs (142 s o

medium stringency; significance was assessed at p , 0.05 and c i

were upregulated and 140 were downregulated, p , 0.05, e with a Benjamini –Hochberg correction for multiple testing. To t y

FDR-adjusted) relative to the other two groups. Stylopized p identify overrepresented biological functions (enrichment analy- u b

wasps were the least divergent in brain gene expression relative l i sis), we compared the annotation composition in our list of s h

to the other two groups (59 DETs: 53 upregulated and six i differentially expressed genes with that of a population back- n g . ground composed by all the P. dominula transcripts (represented downregulated), which agree with the idea that this phenotype o r in the RNA-sequencing data) with Drosophila orthologues. has intermediate features between worker and gyne pheno- g

types. Finally, workers had 94 DETs relative to the other two P r o c

wasp phenotypes (39 upregulated and 55 downregulated). . R

(e) Real-time qRT-PCR validation Analysis via focal pairwise comparisons (figure 2b) ident- . S

We selected two genes to validate RNA-sequencing data from o ified 305 caste-related DETs between workers and gynes (146 c . transcripts found differentially expressed in each phenotype: upregulated in workers and 159 downregulated), 51 DETs B P. dominula defensin precursor (Defensin; GU327374.1) and 2 between stylopized and workers (six upregulated in stylopized 8 4

P. dominula Immune-response protein 30 (IRP30; JN181874.1). These :

and 45 downregulated) and 90 DETs between stylopized and 2 genes were chosen for validation because they had large and 0 1

gynes (30 upregulated in stylopized and 60 downregulated). 7 robust expression differences between all three experimental 0 0 groups. RNA from additional P. dominula samples not already There were only three DETs that differed between all three 2 9 used for RNA-sequencing was used for validation. We only had contrasts (figure 2b, centre), including two immune-related a small number of samples available for this validation exercise genes, Defensin and IRP30. Because of these strong expression

(nW ¼ 4, nG ¼ 3 and nS ¼ 5). cDNA was transcribed from individ- differences, these two genes were used for qRT-PCR validation ual RNA samples, spiked with an external control sequence of the RNA-seq results (described below). (mCherry [48], with SuperScript III First-Strand Synthesis Kit Of the 51 DETs that differed between stylopized females w (Invitrogen)). qRT-PCR was performed, using Power SYBR and workers, 14 were also related to caste differences. This Green PCR Master mix (Life Technologies), on the CFX384 gene set represents top candidates for parasite manipulation TM Touch Real-Time PCR Detection System. of host caste phenotypic plasticity and is labelled as ‘Parasite Control genes were identified from the P. dominula genome manipulation candidate genes’ in figure 2b. These genes [24], using best BLAST hits to control genes used in other studies include five uncharacterized proteins (electronic supplemen- and related taxa: these genes were homologous to other hyme- nopteran Actin and Elongation-factor 1a (EF1a) sequences. We tary material, table S7), as well as genes putatively related to also used an external control mCherry (Discosoma sp., [48]). hormone binding, transcriptional regulation, oxidoreductase Primer sequences for each of the control and focal genes were activity, cell adhesion and proteolysis (table 1). Several of designed using Primer Quest (Integrated DNA Technologies). these functions have been previously associated with caste Full primer sequences and additional protocol details are differential gene expression in Polistes [24,29]. Strikingly, a found in electronic supplementary material, table S4. comparison of read count ratios for the 14 overlapping DETs between workers and gynes (W : G ratio), and workers and sty- lopized females (W : S ratio) were tightly correlated (Pearson 3. Results correlation coefficient R2 ¼ 0.9831, p , 0.0001; figure 2c). (a) Sequencing data (c) Gene functions Fourteen libraries of SOLiD RNA-seq generated 904 464 442 We performed GO analysis separately on each of the pairwise single-end reads of 76 bases in length. The majority (72%) of contrasts in order to identify functional categories of genes reads (657 626 532) passed quality control steps. Most (11) of associated with caste differences (worker versus gyne) and the samples preserved nearly 90% reads after filtering. There parasitism (stylopized versus gyne and stylopized versus were four samples retaining approximately 30% reads after fil- worker); these DET lists returned 43 clusters, nine clusters tering (electronic supplementary material, table S3), and these and one cluster of GO terms, respectively (electronic sup- were kept in the analysis because they were evenly distributed plementary material, table S6). For the caste comparison across the sample groups. Using BFAST [49] to align SOLiD (worker versus gyne), the top clusters related to largely biosyn- RNA-seq reads, 502 657 739 (76.4%) reads were uniquely thetic and metabolic functions (including oxidoreductase and mapped to the reference genome of P. dominula [24]. cytochrome P450), stimulus response and post-translational gene silencing. For the stylopization comparisons, stylopized (b) Gene expression patterns versus gyne clusters primarily consisted of lifespan and metal-binding functions, and the stylopized versus worker We based estimates of gene expression only on uniquely mapped cluster was related to immune function (electronic supplemen- reads aligned to the P. dominula gene annotation r1.0 [24]. Over- tary material, table S6); however, no terms were significantly all, we identified 367 differentially expressed transcripts (DETs) enriched after correction for multiple testing. (see electronic supplementary material, table S5). Hierarchical clustering analysis of all DETs (figure 2a) revealed that the overall gene expression patterns of stylopized females were more similar (d) qRT-PCR validation of select RNA-seq results to those of gynes than workers, despite the fact that stylopized We selected two genes to validate our RNA-seq results using females were collected on worker phase nests alongside qRT-PCR. We chose Defensin and IRP30 because both of these other normal workers. Gyne-shifted gene expression patterns genes showed strong expression differences between the T ( T e a a e F t h o x x c h n a i g r u e p p P P P P P P P P P t e d b o r t u r r d d d d d d d d d s m l s e e a t g e s r o o o o o o o o o i o s s n d e o r m m m m m m m m m s s p t o e s i e s 1 h o u 2 c t M M M M M M M M M d c . e n o r p . a p R R R R R R R R R i f s b T 1 n p p r ( N N N N N N N N N o o a a 6 t d a t ( h m p A A A A A A A A A ) s t c i d e s t e I o 0 0 0 0 0 0 0 0 0 H e i h D d a c s l 1 1 5 6 1 0 0 6 1 r u e g t a a i n 4 2 2 4 1 4 2 5 2 n e a m o n r e s 6 9 2 5 2 8 0 7 9 e g s t n n d m d 1 5 1 4 4 1 1 8 4 n f u i l ‘ o ...... d s i p s a 1 1 1 1 1 1 1 1 1 s s D ) r ( t –2.0 a t a p . y a E t S P r l e a R a T )( s o o l a h s s e p s l g i n i o d o okrsyoie gyne stylopized worker t s i ( z e d e w t f e a n i r r G X X X X X X X X X a n m i s o G e t e n M M M M M M M M M c i e c p m o s c g w l a a n r u n ______e n e r f h B s d e w s o 0 0 0 0 0 0 0 0 0 s i u ( p e e e s a r 1 1 1 1 1 1 1 1 1 w m m o n u d n i n 5 5 5 5 5 5 5 5 5 o r X l o t o 3 3 3 3 3 3 3 3 3 m k a k n s fi a . r e s 3 3 3 2 2 2 1 1 3 t v k v t a i r v d 5 5 3 2 5 5 6 8 5 o e a e Downloaded from r e s 2 3 7 3 0 3 0 9 4 o r n v i y i l s w 2 8 8 1 9 0 9 2 2 e a m a p v o c r d 3 0 4 7 8 6 5 4 6 b a e n s a f i 1:1 ...... d l l u r r r 1 1 1 1 1 1 1 1 1 n e a m s u i e s ) d t r u g m . i e i o s g i u d ( a s n y b l a n s p a a n ) t g t t a l e y e e i i V v r o l x M C P H B M P C D ( b g o a g e e % a n p a a o o o u e e a s p n e n o e n r m m l l i n m f r s i n ( ( ( ( ( i i g n t e n , a d o z p s s 7 7 7 8 8 e e t s s a p p e o b t t n e u e a s d 2 7 8 1 2 i s e e c s s o o m d u m d B r g l i m % % % % % i s s ’ h t o o e n n w a . s n y L o i a n n c g i o o g a l c c ( ) ) ) ) ) l a A i http://rspb.royalsocietypublishing.org/ r l e t o o g c a a t t t r n r a i t e p n h ) a u u S u e p n h n n i r r r a p m o e s s T m w i t r o P o a a +2.0 i z t n r e v n u t s t e d d a t e fl fl i s a y o h u l e t o a s e e d t p a o o h e r s ) n r t r i f r n n t e h k a n r r ) a t i s d i v s s s i i t p n o o a l s a n d d h a i i e t a s r n s s r o n g a ( a a r e t f e n p a p 7 w o s a t l n n s o d i c o u e 3 u o n r e a u u n o n r s ( % c f t n u e e s s r a 7 i a i r s a m t h w e P l 5 t ) e s ( manipulation s c i o . s % b v l c o b h o a m u e ) d ) e r t ) g p k

d log (gyne RC/worker RC) i r o o o f e candidate r r okrvru ye(0)stylopizedversusgyne(90) worker versusgyne(305) i m u o l –2 o e f n r o f n f g u parasite e i g c n o c u p D r o y e t u f l genes E m i a ( n l o l ( T w a t o t n n p h h S p z s b s i t t i i n g o a e o e o a y g a o y u h e i n g t k t ( r g r l t n r d A m t l e o s r b i r H r p k k e i e y a o n a s e e r b g t l s e n ( o a p r d o M i t e o t t e t d d o r s r d r s i i e u i i a x o e B n v t i r c c u u e t X f - worker versusstylopized(51) x n ( lo i e p s t y l p b o e e s r r 4 L d W c c a - p 3 o a d p w ) g s r i p o t t l t - A n b , r c - c o - g i f y a a i a x ) l e e 245 k o B i i t l a v h S –3 –2 –1 i t R , r s s s a (st i p i p s k e - n e t e 3 k - u o T e e e e p l s s P e 2 t 0 1 2 3 4 5 t l e a 0 n v r t e y i p i r S c t e t k y b s d ¼ A a o h d o 4 2 0 h r o y - e e , l o e a t e 14* o t o i l m e e lo e i h c f t l b t p t k 5 i y onApril12,2017 i r 0 e s h a w e e p n m i e 2 l t p z i v . e - t n a 9 e p - o e w i C ized RC/workerRC) o l r o d 8 e d r i o m i d k u e k s 3 a 18 A i o e e e ( e r t 1 43 e n 3 S . l r n n , a ) s o e T n r p e c a h l r b e a a g n f e e c s d o , 16 a d t t t r n e w r l c g o i i s 0 s e o s e y n t . m a u o D w n p s p p c f 28 f e ( 0 n i e c t c u e w n r n x u r r 0 N u e r , h a e e o o l i n t u c 1 l s o d x A r a t t d p s g t t ( e l e u o c . c a r g e e s G e o p r i . r t g k ( n t p s a e g o o i t b a ( e r u ) r R n e i e s u n e O s g i p ) o . u l r n o i r t v r C y m e , n d l d l a j n a p e D n a b s e e ) e n h a d b u l v r l ) s i b t m c r i i , o n t v s . e i e a e c n i n o s g t n l f o ( a r d t h t m r y i b d s e e u T c m l g o a h n s t y o a t r n h i i u i c n s n , o e , e n r o v c o v p o t e l e a s n r a o c u g n e i a r h d n l n o o t e t g o p r s r , , i g a r e l e e s u s i l r g y t i e r g c a o n p c y g d i e c c g o r u o s t r r e n i i , d n m r i a u i t y i r e d c n l e m i v p l c e e n a r n n b l l c a d a i i a t t i t ( i t t e e i e e b n s s l z t i a d m c n V y e o l p b e s h a i o a d i e o p s n r t o t a t a o n n c m a h n h i h o s i r n v t a o o n o t y l e o h n o r a o i l r o e l n g , n d o s , w t o a y f c r r c i b , a t u h s u e t p o t u h s w t [ t r y s a r y c r n s h r 5 t ) a t t a e b . l a i y t . h s e e ) o t 0 n a l l s h n l i l ) r r r a g ] i o e u B s r a e m s r s ) t c p e m n e e h c , r S n s s r i d o s o i h z 7 p i t p c e v p i m e i ) r f t o g e s B f . t d i i p i e o n r o h i s L o a r t l s n f l A e o n l i e e g l a ( n g S m r l h g T t u o i t a e w e a i

h l n n s l e

s l ) i g e y s t ) .

9 2 0 0 7 1 0 2 : 4 8 2 B . c o S . . c o P g o . g n i h s i l b u p y t e i c o s l a y o . b p s R r r r r 5 Downloaded from http://rspb.royalsocietypublishing.org/ on April 12, 2017

gyne of genes differentially expressed because of stylopization 6 worker closely mirrored patterns found in gynes versus workers

100 r

Defensin s 90 stylopized (figure 2c). These data provide new insights into how parasites p b

80 . r 70 * can manipulate host brains and behaviour, and suggest that o y

60 p = 0.2 a l

hijacking of plasticity in gene expression could be an effective s

50 o c

40 i

mechanism by which parasites can shift multiple, coordinated e

30 t y

20 host phenotypes to their advantage. p u

10 b

The most pronounced differences in gene expression (as far l i

0 s % normalized expression 3 4 5 5 5 4 h as the number of genes) that we uncovered were related to caste i n

qRT-PCR RNA-seq g .

differences. There were numerous gene expression differences o r between workers and gynes, including genes with putative g IRP30 20 functions in oxidation–reduction processes, carbohydrate P r o

18 c 16 and lipid metabolism, and visual stimulus response. Several .

* R

14 of these gene functions have been identified in previous studies . 12 S * o 10 of caste differential gene expression in Polistes metricus [29,42] c .

8 and other social taxa [51], suggesting that these data B

6 2

4 reflect typical caste-related expression patterns. We saw fewer 8 4

2 :

0 differences in gene expression associated with stylopiza- 2 % normalized expression 0

4 4 5 5 5 4 1

tion than related to social caste, suggesting that X. vesparum 7

qRT-PCR RNA-seq 0 0

targets a fairly specific set of transcriptomic and behaviou- 2 Figure 3. Validation of RNA-seq results using qRT-PCR. Normalized expression ral responses. Some stylopization-related genes also had 9 is considered as a percentage increase or decrease of expression compared with putative functions associated with immunity, with high that of the worker group (i.e. W ¼ 1 ¼ 100% worker expression). Sample sizes expression of both Defensin and IRP30 in stylopized females shown below the bars, statistical significance indicated with *. (a) Defensin (figure 3). Upregulation of Defensin in stylopized females expression was similar in both qRT-PCR and RNA-seq experiments. Expression agrees with a previous study on X. vesparum effects on larval levels significantly different with RNA-seq (indicated by *), but not qRT-PCR. wasp immune response [52], suggesting that wasps are (b) IRP30 expression was also similar in both RNA-seq and qRT-PCR. Expression immunostimulated during the stylopization process. levels were significantly different with RNA-seq (indicated by *) and also qRT- Although we saw clear evidence of the predicted shift in 2 PCR when normalized by actin (Wilcoxon: x ¼ 6.1091, p ¼ 0.0471) or stylopized female brain gene expression towards being more 2 normalized by Ef-1 (Wilcoxon: x ¼ 6.303, p ¼ 0.0428). gyne-like (figure 2), our data do not allow us to conclude whether these gene expression responses are the cause or the consequence of the observed behavioural shifts. Importantly, stylopized wasps were collected before they exhibited gyne- three groups (note that they are not among our top candidate like nest desertion and aggregation behaviour. This suggests genes for host manipulation). We found overall similar gene that some of the identified genes could have causal roles in expression patterns between our RNA-seq results and qRT- some of the subsequent parasitoid-induced shifts in caste- PCR validation (using different biological replicates) for both related behaviour. Most promising candidates include genes Defensin and IRP30 (figure 3). For Defensin, expression was related to transcriptional regulation (knirps and HMX3-B), extremely high in stylopized females, intermediate in gynes oxidation–reduction (hydroxypyruvate reductase) and behaviou- and lowest in workers in both RNA-seq and qRT-PCR ral response to starvation (Takeout, figure 2c and table 1). (figure 3a). For IRP30, expression was lowest in workers and However, more studies are needed to understand whether higher in stylopized females (figure 3b), but the pattern in these gene expression changes are a direct cause of parasitoid gynes was not consistent across the two types of analysis. The manipulation, or are downstream from other parasitoid-induced Wilcoxon tests showed significant differences for IRP30 in the changes in host physiology or hormones. qRT-PCR (IRP30: x2¼ 6.1091, p ¼ 0.0471) and RNA-seq. For This study also highlights the potential of using host– Defensin, the qRT-PCR was not significant ( p ¼ 0.20), but a parasite systems to elucidate the molecular basis of complex trend in the same direction as the RNA-seq data was apparent. phenotypes. Although hundreds of caste-related gene expres- sion differences have been previously identified in Polistes [53], it has been difficult to pinpoint which genes relate to dif- 4. Discussion ferent aspects of the suite of caste-related behavioural and In this study, we provide novel evidence supporting the physiological traits. Because stylopization only appears to hypothesis that parasites can manipulate host behaviour by target certain modules of gyne-like behaviour (such as nest inducing changes in gene expression related to innate host desertion), rather than the entire phenotype, it can serve as a phenotypic plasticity. The main finding of this study is that ‘natural experiment’ [18] to help hone in on key molecular X. vesparum appears to be manipulating its P. dominula wasp components involved in specific caste endophenotypes. host by partially shifting its social caste, and this is associated In summary, we provide novel data on the transcriptomic with changes in the expression of caste-related genes. Three basis of parasite manipulation of host social behaviour. Our main lines of evidence support this idea. (i) Overall gene data suggest that X. vesparum shifts the brain expression of expression patterns in stylopized females were shifted towards genes related to plastic social castes in their host, eliciting an being gyne-like (figure 2a). (ii) A substantial fraction of the ordinarily quiescent gyne phenotype in the ‘wrong caste’. genes that changed in expression because of stylopization This is an excellent example of subtle behaviour manipulation, were caste-related genes (i.e. also differed in expression and our results suggest that parasitism can solicit a complex between normal gynes and workers, figure 2b). (iii) A subset behavioural response in hosts via small but specific shifts in Downloaded from http://rspb.royalsocietypublishing.org/ on April 12, 2017

host brain gene expression. We suggest that plastic host behav- be found in the following repository (https://github.com/ruolin/ 7 ioural and physiological responses, such as alternate morphs WaspsProject). r s or even personality types [7], may be particularly susceptible Authors’ contributions. A.C.G.: brain dissection, RNA extraction and p b .

submission of samples for sequencing, qRT-PCR validation, data r to parasite manipulation. By tapping into underlying host o y

analysis and writing; R.L.: processing of raw sequencing data, a l phenotypic plasticity, parasites could more efficiently alter s

preparation of DET dataset and writing; F.M.: sample collection, prep- o c i multiple, correlated host behavioural and physiological traits aration of samples for RNA-sequencing and writing; L.B.: study e t y via fairly targeted shifts in host regulatory mechanisms. conception, sample collection, experimental design and writing; p u

J.K.: study conception, sample collection and writing; C.M.G.: RNA- b

Further research is needed to investigate the generality of l i s

sequencing/analysis infrastructure and resources; A.L.T.: study h such a mechanism across other host–parasite systems, but i n

conception and experimental design, sample collection and processing, g . the plasticity perspective has the potential to provide new o

and writing. r g insights into our understanding of host–parasite interactions Competing interests. We have no competing interests. and their coevolution. P Funding. No funding has been received for this article. r o c Acknowledgements. We thank Jimena Carrillo-Tripp and Adam Dolezal . R

Ethics. Wasps in this study were collected and euthanized humanely, for mCherry primers and qRT-PCR standards and Apis mellifera . S according to standard collection protocols. o

sequences used to design control genes in P. dominula, Dave Galbraith c .

Data accessibility. Raw data files from the SOLiD RNA-sequencing project and Brendan Hunt for assistance with the RNA-seq data analysis, B described above have been deposited in the NCBI short read archive, Zoltan Acs for wasp samples used to generate preliminary data (not 2 8 4

BioProject (BioProject: PRJNA312511, Experiments SRX1629425 included in the final study), Daniel Standage for data storage and the : 2

through SRX1629436). All scripts used to process RNA-Seq data can Toth laboratory for feedback on the experiment and manuscript. 0 1 7 0 0 2 9 References

1. Lefevre T, Adamo S, Biron D, Misse D, Hughes D, behaviour in vertebrates. Physiol. Behav 79, outside if parasitized: parasite manipulation? Behav. Thomas F. 2009 Invasion of the body snatchers: the 441–449. (doi:10.1016/S0031-9384(03)00163-X) Ecol. 15, 1037–1043. (doi:10.1093/beheco/arh111) diversity and evolution of manipulative strategies in 11. Adamo S. 2013 Parasites: evolution’s 20. Kathirithamby J. 1989 Review of the order host–parasite interactions. In Advances in neurobiologists. J. Exp. Biol. 216, 3–10. (doi:10. Strepsiptera. Syst. Entomol. 14, 41–92. (doi:10. parasitology: natural history of host–parasite 1242/jeb.073601) 1111/j.1365-3113.1989.tb00265.x) interactions, vol. 68 (ed. JP Webster), pp. 45–83. 12. Hoek RM, Van Kesteren RE, Smit AB, de Jong-Brink 21. Kathirithamby J. 2009 Host–parasitoid associations Burlington, MA: Academic Press. M, Geraerts WP. 1997 Altered gene expression in in Strepsiptera. Ann. Rev. Entomol. 54, 227–249. 2. Libersat F, Delago A, Gal R. 2009 Manipulation of the host brain caused by a trematode parasite: (doi:10.1146/annurev.ento.54.110807.090525) host behaviour by parasitic and insect neuropeptide genes are preferentially affected 22. Starks P, Turillazzi S. 2006 Polistes paper wasps: parasites. Annu. Rev. Entomol. 54, 189–207. during parasitosis. Proc. Natl Acad. Sci. USA 94, emergence of a model . Ann. Zool. Fenn. 43, (doi:10.1146/annurev.ento.54.110807.090556) 14 072–14 076. (doi:10.1073/pnas.94.25.14072) 385–624. 3. Perrot-Minot M, Cezilly F. 2010 Parasite-induced 13. Biron DG, Marche´ L, Ponton F, Loxdale HD, Gale´otti 23. Jandt J, Tibbetts E, Toth A. 2015 Polistes paper behavioural change: mechanisms. In Encyclopedia N, Renault L, Joly C, Thomas F. 2005 Behavioural wasps: a model genus for the study of social of behaviour (eds M Breed, J Moore), manipulation in a grasshopper harbouring dominance hierarchies. Insect. Soc. 61, 11–27. pp. 618–627, 3rd edn. Oxford, UK: Academic Press. hairworm: a proteomics approach. Proc. R. Soc. (doi:10.1007/s00040-013-0328-0) 4. Hoover K, Grove M, Gardner M, Hughes D, Lond. B. 272, 2117–2126. (doi:10.1098/rspb. 24. Standage D, Berens A, Glastad K, Severin A, Brendel McNeil J, Slavicek J. 2011 A gene for an 2005.3213) V, Toth A. 2016 Genome, transcriptome, and extended phenotype. Science 333, 1401. 14. Feldmeyer B, Mazur J, Beros S, Lerp H, Binder H, methylome sequencing of a primitively eusocial (doi:10.1126/science.1209199) Foitzik S. 2016 Gene expression patterns underlying wasp reveal a greatly reduced DNA methylation 5. Perrot-Minot M, Kaldonski N, Cezilly F. 2007 parasite-induced alterations in host behaviour and system in a social insect. Mol. Ecol. 25, 1769–1784. Increased susceptibility to predation and altered life history. Molec. Ecol. 25, 648–660. (doi:10.1111/ (doi:10.1111/mec.13578) anti-predator behaviour in an acanthocephalan- mec.13498) 25. Sumner S, Pereboom J, Jordan W. 2006 Differential infected amphipod. Int. J. Parasitol. 37, 645–651. 15. de Bekker C, Ohm RA, Loreto RG, Sebastian A, gene expression and phenotypic plasticity in (doi:10.1016/j.ijpara.2006.12.005) Albert I, Merrow M, Brachmann A, Hughes DP. 2015 behavioural castes of the primitively eusocial wasp, 6. Knell R, Webberly K. 2004 Sexually transmitted Gene expression during zombie ant biting behavior Polistes canadensis. Proc. R. Soc. B 273, 19–26. disease of insects: distribution, evolution, ecology reflects the complexity underlying fungal parasitic (doi:10.1098/rspb.2005.3291) and host behaviour. Biol. Rev. 79, 557–581. behavioral manipulation. BMC Genomics 16, 620. 26. Toth A et al. 2007 Wasp gene expression supports (doi:10.1017/S1464793103006365) (doi:10.1186/s12864-015-1812-x) an evolutionary link between maternal behaviour 7. Poulin R. 2013 Parasite manipulation of host 16. Hughes D. 2013 Pathways to understanding and eusociality. Science 318, 441–444. (doi:10. personality and behavioural syndromes. J. Exp. Biol. the extended phenotype of parasites in their 1126/science.1146647) 216, 18–26. (doi:10.1242/jeb.073353) hosts. J. Exp. Biol. 216, 142–147. (doi:10.1242/ 27. Toth A, Varala K, Henshaw M, Rodriguez-Zaz S, 8. Houte S, Ros VI, Oers MM. 2013 Walking with jeb.077461) Hudson M, Robinson G. 2010 Brain transcriptomic insects: molecular mechanisms behind parasitic 17. Rokas A, Abbot P. 2009 Harnessing genomics analysis in paper wasps identifies gene associated manipulation of host behaviour. Molec. Ecol. 22, for evolutionary insights. Trends Ecol. Evol. 24, with behaviour across social insect lineages. 3458–3475. (doi:10.1111/mec.12307) 192–200. (doi:10.1016/j.tree.2008.11.004) Proc. R. Soc. B 277, 2139–2148. (doi:10.1098/rspb. 9. West-Eberhard M. 2003 Developmental plasticity 18. Beani L. 2006 Crazy wasps: when parasites 2010.0090) and evolution. New York, NY: Oxford University manipulate the Polistes phenotype. Ann. Zool. 28. Ferreira P, Patalano S, Chauhan R, Ffrench-Constant Press. Fennici. 43, 564–574. R, Gabaldon T, Guigo R, Sumner S. 2013 10. Klein S. 2003 Parasite manipulation of the 19. Hughes D, Kathirithamby J, Turilazzi S, Beani L. Transcriptome analyses of primitively eusocial wasps proximate mechanisms that mediate social 2004 Social wasps desert the colony and aggregate reveal novel insights into the evolution of sociality Downloaded from http://rspb.royalsocietypublishing.org/ on April 12, 2017

and the origin of alternative phenotypes. Genome 37. Kinzelbach R. 1978 Fa¨cherflu¨gler (Strepsiptera). Die 45. Mikheyev A, Linksvayer T. 2015 Genes associated 8 Biol. 14, R20. (doi:10.1186/gb-2013-14-2-r20) Tiewelt eutschlands, vol. 65. Jena, Germany: VEB with ant social behaviour show distinct r s 29. Berens A, Hunt J, Toth A. 2014 Comparative Gustav Fischer Verlac. transcriptional and evolutionary patterns. eLife 4, p b . r transcriptomics of convergent evolution: different 38. Beani L, Giusti F, Mercati D, Lupetti P, Paccagnini E, e04775. (doi:10.7554/eLife.04775) o y a l

genes but conserved pathways underlie caste Turilazzi S, Dallai R. 2005 Mating of Xenos vesparum 46. Huang D, Sherman B, Lempicki R. 2009 Systematic s o c i

phenotypes across lineages of eusocial insects. Mol. (Rossi) (Strepsiptera, Insecta) revisited. J. Morphol. and integrative analysis of large gene lists using e t y

Biol. Evol. 32, 690–703. (doi:10.1093/molbev/ 265, 291–303. (doi:10.1002/jmor.10359) DAVID Bioinformatics Resources. Nat. Protoc. 4, p u b

msu330) 39. Erezylimaz D, Hayward A, Huang Y, Paps J, Acs Z, 44–57. (doi:10.1038/nprot.2008.211) l i s h

30. Reeve H. 1991. Polistes. In The social biology of Delgado J, Collantes F, Kathirithamby J. 2014 47. Huang D, Sherman B, Lempicki R. 2009 i n g .

wasps (eds K Ross, R Matthews). Ithaca, NY: Cornell Expression of the pupal determination broad during Bioinformatics enrichment tools: paths toward the o r University Press. metamorphic and neotenic development the comprehensive functional analysis of large gene g

31. Pardi L. 1948 Dominance order in Polistes wasps. strepsipteran Xenos vesparum Rossi. PLoS ONE 9, 1–8. lists. Nucleic Acids Res. 37, 1–13. (doi:10.1093/ P r o c

Physiol. Zool. 21, 1–13. (doi:10.1086/physzool.21.1. 40. Beani L, Dallai R, Mercati D, Cappa F, Giusti F. 2011 nar/gkn923) . R

30151976) When a parasite breaks all the rules of a colony: 48. Carrillo-Tripp J, Krueger E, Harrison R, Toth A, Miller . S o

32. Hunt J. 2007 The evolution of social wasps. morphology and fate of wasps infected by a A, Bonning B. 2014 Lymantria dispar virus 1 c .

New York, NY: Oxford University Press. strepsipteran endoparasite. Anim. Behav. 82, (LdIV1), a new model to study iflaviral persistence B 2

33. Hughes D, Beani L, Turilazzi S, Kathirithamby J. 1305–1312. (doi:10.1016/j.anbehav.2011.09.012) in lepidopterans. J. Gen. Virol. 95, 2285–2296. 8 4 :

2003 Prevalence of the parasite Strepsiptera in 41. Beani L, Massolo A. 2007 Polistes dominulus wasps (doi:10.1099/vir.0.067710-0) 2 0 1

Polistes as detected by dissection of immatures. (: Vespidae), if parasitized by Xenos 49. Homer N, Merriman B, Nelson S. 2009 Local alignment 7 0 0

Insect. Soc. 50, 62–68. (doi:10.1007/s0004 vesparum (Strepsiptera: Stylopidae), wander among of two-base encoded DNA sequence. BMC Bioinformatics 2 00300010) nests during the pre-emerging phase. Redia 90, 10, 175. (doi:10.1186/1471-2105-10-175) 9 34. Manfredini F, Massolo A, Beani L. 2010 161–164. 50. Oliveros, J. 2007 VENNY: an interactive tool for A difficult choice for tiny pests: host-seeking 42. Toth A, Bilof K, Henshaw M, Hunt J, Robinson G. comparing lists with Venn diagrams. See http:// behaviour in Xenos vesparum triungulins. Ethol. 2009 Lipid stores, ovary development and brain bioinfogp.cnb.csic.es/tools/venny/index.html. Ecol. Evol. 22, 247–256. (doi:10.1080/03949370. gene expression in Polistes metricus females. Insect. 51. Chen X et al. 2012 Transcriptome comparison 2010.502319) Soc. 59, 77–84. (doi:10.1007/s00040-008-1041-2) between honey queen- and worker-destined 35. Hughes D, Kathirithamby J. 2005 Cost of 43. Manfredini F, Giusti F, Beani L, Dallai R. 2007 larvae. Insect. Biochem. Molec. 42, 665–673. strepsipteran macroparasitism for immature Developmental strategy of the endoparasite Xenos (doi:10.1016/j.ibmb.2012.05.004) wasps: does sociality modulate virulence? vesparum (Strepsiptera, Insecta): host invasion and 52. Manfredini F, Beani L, Taormina M, Vannini L. 2010 Oikos 220, 428–434. (doi:10.1111/j.0030-1299. elusion of its defense reactions. J. Morphol. 268, Parasitic infection protects wasp larvae against a 2005.13791.x) 588–601. (doi:10.1002/jmor.10540) bacterial challenge. Microb. Infect. 12, 727–735. 36. Hughes D, Kathirithamby J, Beani L. 2004 44. Robinson M, McCarthy D, Smyth G. 2010 edgeR: a (doi:10.1016/j.micinf.2010.05.001) Prevalence of the parasite Strepsiptera in adult Bioconductor package for differential expression 53. Jandt J, Toth A. 2015 Physiological and genomic Polistes wasps: field collections and literature analysis of digital gene expression data. mechanisms of social organization in wasps overview. Ethol. Ecol. Evol. 16, 363–375. (doi:10. Bioinformatics 26, 139–140. (doi:10.1093/ (Hymenoptera: Vespidae). Adv. Insect. Physiol. 48, 1080/08927014.2004.9522627) bioinformatics/btp616) 95–130. (doi:10.1016/bs.aiip.2015.01.003)