The role of a viral symbiont in the thermal mismatch of host–parasitoid interactions

ABSTRACT

High temperature events are becoming more severe with climate change, altering species interactions and ecological networks. Symbionts can influence the thermal tolerance of their hosts, yet the mechanisms underlying these effects are poorly understood. We tested the impact of a high temperature event on the molecular interactions among a caterpillar host, Manduca sexta, its parasitoid wasp, Cotesia congregata, and the wasp's symbiotic virus. As in many host–parasitoid systems, high temperatures are lethal to developing parasitoids, but not hosts. Typically, the parasitoid's viral symbiont immunosuppresses M. sexta. Here, we show that elevated temperatures led to an impairment of this immunosuppression, persisting for days after the event ended. Viral gene expression in the host was altered by heat, with distinct expression patterns tied to the virus's genomic architecture. Specifically, viral transcription varied according to the gene's position on viral circular genomic segments: genes located on circles known to integrate into host DNA exhibited increased or unchanged expression following high temperature exposure, while genes on non-integrating circles showed marked reductions in expression. These results demonstrate that high temperatures can disrupt parasitic immunosuppression, which could help explain the lower thermal tolerance of parasitoids relative to hosts. The genomic structure of the viral symbiont may be associated with these effects, but additional research is needed to evaluate this hypothesis. Our findings highlight the importance of complex interactions between environmental temperature, microbial symbionts and host immunity in the ecological responses of host–parasitoid systems to high temperature events.

Summary: Heat alters immune interactions and symbiont gene expression in a host–parasitoid–symbiont system, revealing potential mechanisms underlying thermal mismatch in ecological interactions during heat wave events.

INTRODUCTION

Climate change is increasing the frequency and intensity of extreme high temperature events such as heat waves (Colwell et al., 2012; Parmesan, 2006). Understanding how heat waves affect interacting species that form the basic units of communities and ecosystems is a high priority for ecologists (Buckley et al., 2023; Buckley and Kingsolver, 2019; Scranton and Amarasekare, 2017). When one or more species in a network has a lower thermal tolerance than others, thermal mismatches in species interactions can occur (Nowakowski et al., 2016). Thermal mismatches have been documented in a variety of host–symbiont and host–parasite systems; for example, coral/zooxanthellae (Hoegh-Guldberg, 1999), amphibian/fungi (Cohen et al., 2017) and insect/microbial symbionts (Dunbar et al., 2007; Wernegreen, 2012). However, the genetic and physiological mechanisms underlying thermal mismatches are poorly understood, especially in complex multi-species networks. Characterizing these mechanisms can improve our understanding for how molecular, organismal and ecological factors give rise to community- and ecosystem-level responses to climate change (Briscoe et al., 2023; Colwell et al., 2012; Riddell et al., 2017).

Parasitoids are incredibly ecologically and economically important insects as top-down controllers of insect pests (Meisner et al., 2014; Mugabo et al., 2019; Parker and Kingsolver, 2024), but are particularly vulnerable to climate change, making them ideal models for investigating climate impacts on trophic interactions (Hance et al., 2007; Malinski et al., 2023; Meisner et al., 2014; Moore et al., 2020, 2021, 2022, 2023; Tougeron et al., 2020; Tougeron and Iltis, 2022). Parasitoids are insects that obligately parasitize other insects for some portion of their life cycle (Polaszek and Vilhemsen, 2023). Parasitoid wasps are ubiquitous in nearly all terrestrial ecosystems and serve as ecosystem stabilizers and agricultural biocontrol agents of herbivorous insects, making them key players in community stability (Meisner et al., 2014; Mugabo et al., 2019). To create a favorable environment for parasitoid development, parasitoids must subdue, immunosuppress or otherwise manipulate the host insect through various specialized adaptations, including for some parasitoid wasp species the harboring of symbiotic polydnaviruses (Burke and Strand, 2014; Webb et al., 2006). These viral symbionts are often endogenized into the parasitoid's genome, where they produce dsDNA circular genomic segments packaged into viral particles called virions that are injected into hosts along with parasitoid eggs during oviposition (Gundersen-Rindal, 2012; Webb et al., 2006). Once inside the host insect, virions either integrate into host DNA or remain episomal (Wang et al., 2021), acting as gene delivery vectors to produce proteins that suppress host immunity, alter host endocrine and developmental pathways, and manipulate host behavior over the course of parasitism (Beckage et al., 1994; Burke and Strand, 2012, 2014; Webb et al., 2006; Wu et al., 2025).

Polydnaviruses play a key role in suppressing host immunity and facilitating the success of parasitoid eggs (Beckage et al., 1994; Burke and Strand, 2012). In many species, this immunosuppression targets host encapsulation and melanization pathways, main insect defenses against multicellular parasites (Dubovskiy et al., 2016; Gundersen-Rindal et al., 2013). Encapsulation forms a multilayered barrier of hemocytes (insect blood cells) around the parasite to sequester it (Zdybicka-Barabas et al., 2025). This capsule is then melanized through activation of the prophenoloxidase cascade, generating melanin and other cytotoxic molecules to kill the parasite (Dubovskiy et al., 2016; Gundersen-Rindal et al., 2013; Rantala et al., 2002; Rantala and Roff, 2007; Zdybicka-Barabas et al., 2025). Because these cytotoxic products can also damage host tissues, melanization is tightly regulated (Zdybicka-Barabas et al., 2025). Polydnaviruses often exploit this regulation to suppress melanization during parasitism (Beck and Strand, 2007; Chevignon et al., 2014, 2015; Wu et al., 2025). Encapsulation is disrupted by polydnaviruses, along with venom and other parasitoid products, by altering hemocyte abundance and motility (Teramoto and Tanaka, 2004; Wang et al., 2023; Wu et al., 2025). Although encapsulation and melanization are distinct processes, they function together in response to parasites: for example, melanization can occur independently in the hemolymph or during wound healing, but encapsulation of a parasite is followed by capsule melanization to kill the foreign invader (Zdybicka-Barabas et al., 2025). Hemocytes can initiate local melanization responses, while many immune components supporting these pathways are produced by the fat body; both tissues are key targets of immunosuppression by polydnaviruses (Li et al., 2018; Gunaratna and Jiang, 2013; Chevignon et al., 2014, 2015). Because host, parasitoid and viral symbiont responses to stressors can be independently measured in the lab (Adamo, 2005; Adamo et al., 2016; Gao et al., 2022; Gelman et al., 1999; Lavine and Beckage, 1996; Moore et al., 2021; Seehausen et al., 2017; Wang et al., 2021), these systems provide powerful tools for studying temperature effects in species relationships.

There is growing consensus that climate change is disrupting host–parasitoid interactions (Kankaanpää et al., 2020; Wenda et al., 2023), with parasitoids often facing particularly negative effects (Bahar et al., 2013; Furlong and Zalucki, 2017; Le Lann et al., 2021; Wu et al., 2016) (recently reviewed in Malinski et al., 2024). In insects, microbial symbionts can shape host responses to temperature stress (Dunbar et al., 2007; Wernegreen, 2012), but the role of polydnaviruses in parasitoid thermal tolerance is poorly understood. To address this gap, we use a model system comprising the larval tobacco hornworm, Manduca sexta (Lepidoptera: Sphingidae), its parasitoid wasp Cotesia congregata (Hymenoptera: Braconidae) and the parasitoid's symbiotic polydnavirus, the C. congregata bracovirus (CcBV). As in many other host–parasitoid systems, the thermal tolerance of C. congregata is lower than that of its M. sexta host (Moore et al., 2021, 2022). The developmental timing of this effect suggests that the CcBV symbiont may play a central role in shaping the thermal limits of the parasitoid: because viral manipulation of host physiology shifts over the course of parasitism (Lavine and Beckage, 1996), the timing of heat exposure is critical in determining the outcome for host, parasitoid and symbiont (Moore et al., 2022; Valls et al., 2020). High temperatures lead to the death of parasitoids during the egg stage but not later larval stages (Moore et al., 2021, 2022). This developmental window coincides with the period when the CcBV symbiont immunosuppresses the host, a function which is necessary for parasitoid survival (Amaya et al., 2005; Beckage et al., 1990). One possible explanation for parasitoid mortality at this developmental stage is that high temperatures impair CcBV function, allowing the host's immune response to eliminate parasitoid eggs.

We tested this hypothesis in the M. sexta–C. congregata–CcBV system by quantifying host immune responses and host and CcBV transcriptomic responses following high temperature exposure. Immune assays quantified the impacts of elevated temperature and parasitism on hemocyte-mediated defenses (encapsulation and capsule melanization). If host–parasitoid thermal mismatch is associated with reduced symbiont function, we predicted that elevated temperatures would weaken viral immunosuppression, resulting in stronger encapsulation and capsule melanization responses in heat-exposed parasitized hosts. In parallel, transcriptomics were used to evaluate host and symbiont molecular responses to high temperature and identify mechanisms in fat body tissue that may underlie thermal mismatch (parasitoid death). If thermal mismatch is driven by disruption of the viral symbiont (e.g. by degrading virions or preventing transcription of viral genes in host tissues), we expected viral gene expression to show strong patterns of dysregulation under high temperatures. Alternatively, if host response to heat drives thermal mismatch independent of viral transcription (e.g. if host immunity increases with heat), we expected viral gene expression to show minimal change at high temperatures, and host genes involved in supporting encapsulation and melanization to be upregulated. However, if host fat body tissue is not important to thermal mismatch, we expected to see no relationship between host immune gene expression and high temperatures. To evaluate the robustness of these effects across multiple populations, experiments were conducted in divergent host populations from a domesticated lab colony and from a wild field population. Finally, to test whether high temperatures impair parasitoid survival directly, we also developed a novel in vitro egg culturing assay to measure egg-to-hatching success under high temperature conditions.

MATERIALS AND METHODS

Study system and host populations

This study was conducted using Manduca sexta (Linnaeus 1763) hosts from both a wild field population and a domesticated laboratory population. Populations diverged >200 generations ago. Parasitoid (but not host) mortality at high temperatures was previously reported in both populations (Malinski et al., 2023; Moore et al., 2021, 2022); however, these populations differ markedly in their immune function (Alston et al., 2020; Diamond and Kingsolver, 2011) and thermal tolerance (Kingsolver and Nagle, 2007). These populations were compared in this study to identify and evaluate mechanisms driving thermal mismatch, as thermal mismatch with C. congregata occurs in both populations despite these biological differences.

Field population M. sexta eggs were collected from tobacco leaves at Upper Coastal Plains Research Station (Rocky Mount, NC, USA) and Central Crops Research Station (Clayton, NC, USA). Domesticated population M. sexta eggs were sourced from a lab colony at University of North Carolin (UNC, Chapel Hill, NC, USA), which has been maintained at 25°C constant with no parasitoid contact for >200 generations. All hosts were kept at a constant 25°C under a 14 h:10 h light:dark photoperiod in environmental chambers (Percival Scientific 36VL) and provided with an artificial wheat germ-based diet supplemented with tobacco powder ad libitum (adapted from Kingsolver and Woods, 1998). Host larvae entered the experiment upon molt to 3rd instar. Ethical approval was not required for this work, but all organisms were treated as humanely as possible.

Experimental design and heat shock treatments

The experiment consisted of two treatment types (parasitism, temperature) each with two levels (P, parasitized; NP, non-parasitized control; 40°C, heat shock temperature; 25°C, control temperature) for both lab and field M. sexta populations (four treatments per population: 25NP, 25P, 40NP, 40P) (Fig. 1). Hosts in the parasitized treatment were parasitized in the morning (between 09:00 and 11:00 h) by exposing them to adult female Cotesia congregata (Say 1836) from the UNC colony until an oviposition event lasting >2 s was observed (Moore et al., 2020). To compare between parasitized and non-parasitized treatment groups directly, an immune reaction to a multicellular organism was stimulated in non-parasitized hosts by sham injection with Sephadex beads directly into the hemocoel (Diamond and Kingsolver, 2011). Sham injections were prepared as 0.25 mg sterile Sephadex A-25 beads suspended in 5 µl sterile phosphate-buffered saline and administered using a 25-gauge 100 µl syringe (Hamilton 700 series) (Lavine and Beckage, 1996). Hosts were held at 25°C until the afternoon of the same day (between 14:00 and 16:00 h), then moved to temperature treatments (40°C heat shock or 25°C control) in environmental chambers for 24 h. Following the 24 h temperature exposure, hosts were returned to constant 25°C for 24 h to recover from acute effects of heat stress. Following this recovery period, hosts entered the immune assays or were flash-frozen intact in liquid nitrogen and stored at −80°C for later tissue dissection for transcriptomics.

Fig. 1.

Experimental time line and methodology for immune assays and transcriptomics. Treatments included temperature and parasitism, each with two levels (P, parasitized; NP, non-parasitized sham injection; 25°C, control temperature; 40°C, heat shock temperature) for a total of four treatment groups. The experiment was run with two populations of Manduca sexta (lab, field). The time line is shown on the x-axis, and temperature is shown on the y-axis. Line color and type indicate the temperature treatment groups (solid purple line, 25°C control; dashed orange line, 40°C heat shock).

Host immune assays

CcBV primarily suppresses host immunity by disrupting encapsulation and melanization, primary insect defenses against multicellular parasites (Dubovskiy et al., 2016; Gundersen-Rindal et al., 2013; Wu et al., 2025). Host encapsulation and capsule melanization responses were quantified using an immune assay protocol modified from Rantala and Roff (2007); nylon monofilaments (4±0.1 mm length, 0.1 mm diameter) were inserted beneath the cuticle at the base of the last larval proleg. Larvae were allowed to encapsulate and melanize the filament for 24 h before it was gently removed and frozen at −20°C until imaging (Diamond and Kingsolver, 2011; Rantala and Roff, 2007). This assay quantifies the strength of encapsulation and capsule melanization responses at the physiological level in the host organism. Filaments were imaged under 10× magnification (Nikon AlphaPhot-2 YS2-T microscope) and standardized light conditions via a microscope-mounted camera (OMAX A35180U3) and Touplite software (v.2.1.17309.20200616). Images were analyzed in ImageJ software (v.1.53i) for encapsulation area (area within the outline of the implant+adhered hemocytes) and capsule melanization (mean gray value within the outline of the implant+adhered hemocytes) using a custom macro script. Detailed imaging and analysis protocols can be found in the Supplementary Materials and Methods.

To determine whether and how temperature, parasitism or their interaction impacted encapsulation or capsule melanization, ANOVAs were conducted using the ‘aov’ function in R. For encapsulation, the response variable was mean encapsulated area of adhering hemocytes; for capsule melanization, the response variable was mean gray value of adhering hemocytes. To explicitly test whether the two host populations differed in their responses, three-way ANOVAs were conducted with temperature, parasitism, population and their interactions as predictor variables (Table S1). To evaluate population-specific effects, two-way ANOVAs were conducted in each population separately with temperature, parasitism and their interaction as predictor variables (Table 1).

Table 1.

Results of two-way ANOVA for encapsulation and capsule melanization responses in immune assays for lab and field population Manduca sexta hosts

Population	Immune response	Treatment	F-value	Pr(>F)
Lab	Encapsulation	Temperature	0.188	0.666
		Parasitism	1.474	0.227
		Temperature×Parasitism	0.474	0.492
	Capsule melanization	Temperature	0.183	0.669
		Parasitism	0.225	0.636
		Temperature×Parasitism	14.538	0.0002*
Field	Encapsulation	Temperature	0.514	0.475
		Parasitism	20.537	1.7e−05*
		Temperature×Parasitism	13.37	0.0004*
	Capsule melanization	Temperature	2.151	0.145
		Parasitism	103.7	<2e−16*
		Temperature×Parasitism	14.54	0.0002*

Significant Pr(>F) is indicated with an asterisk.

Host and CcBV transcriptomics

Fat body isolation and RNA extraction

Fat body tissue was chosen for transcriptomic analysis because of its importance in insect immune responses and as one of the main tissue targets of host immunosuppression by CcBV (Chevignon et al., 2015; Gunaratna and Jiang, 2013). Fat body tissue supports humoral immunity and melanization in the hemolymph; note that transcriptomics in this study do not measure the response of hemocytes, which are captured by the immune assays. RNA was extracted from fat body tissue of 8–10 M. sexta larvae per treatment group per population using the Direct-zol RNA Miniprep Kit (Zymo Research) according to the manufacturer's instructions, including a DNase I (Zymo Research) treatment step to remove genomic DNA from samples. RNA quality was assessed via Qubit 4 Fluorometer High Sensitivity RNA assay and NanoDrop Lite Plus Spectrophotometer (both Thermo Fisher Scientific).

Library preparation and sequencing

RNA libraries were constructed via QuantSeq 3′ mRNA-Seq FWD Library Prep Kit for Illumina sequencing and purified via Purification Module with Magnetic Beads PCR Add-on Kit (both Lexogen) according to the manufacturer's instructions. This method generates sequences from the 3′ end of mRNA transcripts, and has been demonstrated as an accurate and cost-effective alternative to whole mRNAseq approaches (Lohman et al., 2016). Library quality was assessed after the final PCR step via Qubit (information as above); libraries were then sequenced via Illumina NovaSeq6000 SP with single-end 100 bp reads.

Read processing and differential gene expression analysis

Sequences were trimmed using cutadapt (v.4.1) (Martin, 2011) to remove adapters and reads shorter than 20 bp, as per the library preparation kit manufacturer’s guidelines. Trimmed sequences were assessed for quality control using FastQC (v.0.11.5), and aligned to both the M. sexta genome (GCF_014839805.1_JHU_Msex_v1.0) (Gershman et al., 2021) and the circular CcBV genome (GCA_000844405.1_ViralMultiSegProj14556) (Espagne et al., 2004) via STAR alignment software (v.2.7.7a) (Dobin et al., 2013). Uniquely mapped reads were counted via featureCounts read summarization software (v.2.0.4) (Liao et al., 2014). Detailed program specifications for alignment are reported in the Supplementary Materials and Methods. Only reads which aligned to either the M. sexta or CcBV reference genomes were used in transcriptomic analyses.

All subsequent analyses were conducted in R (v.4.3.1) unless otherwise specified. Pre-filtering was performed by removing genes with <20 counts across at least three libraries to increase analytical power (Love et al., 2014, 2015). Normalization of genes and differential gene expression analysis was performed for each alignment using the ‘DESeqDataSetFromMatrix’ and ‘DESeq’ functions in the DESeq2 package (v.1.40.2) (Love et al., 2014, 2015) for each population separately, using temperature, parasitism and their interaction as coefficients. log₂Fold-change values were shrunk via the ‘ashr’ method to preserve signal and reduce noise (Stephens, 2017). Differentially expressed genes (DEGs) were considered statistically significant according to the Wald test and Benjamini–Hochberg correction (Haynes, 2013) with a false discovery rate (FDR) cutoff of <0.05 and log₂fold-change >|1|.

Gene annotation

CcBV

Because of incomplete functional characterization and Gene Ontology (GO) annotation, CcBV genes were categorized based on gene family (PTP, protein tyrosine phosphatase; ANK, viral ankyrin; EP, early protein; cystatin, cystatin proteins; CrV1, CrV1 proteins). CcBV genes were further classified based on their known ability to integrate into host M. sexta DNA: the genes of the CcBV genome are apportioned into 36 dsDNA circularized segments excised from their proviral form in the wasp's linear genome, and encapsidated into virion particles in the wasp's ovaries (Bézier et al., 2013; Espagne et al., 2004; Louis et al., 2013). Some viral circles integrate into the DNA of the caterpillar host to express viral genes, while others transcribe genes episomally (non-integrated) (Chevignon et al., 2018; Muller et al., 2021; Wang et al., 2021). Current understanding for integration status is partial for CcBV (Chevignon et al., 2018), but more robust for a closely related species whose genome is organized into the same 36-circular segment structure (Cotesia typhae bracovirus, CtBV) (Muller et al., 2021). We investigated whether CcBV differential gene expression was correlated with viral circle integration status using existing evidence from CcBV and this close relative.

Manduca sexta

To assess transcription of host immune genes typically suppressed by CcBV, we evaluated expression of M. sexta genes involved in encapsulation, melanization and melanization inhibition (negative regulation of the melanization pathway). Genes with a predicted function in these pathways were included based on GO terms. Genes with an established function in these pathways in M. sexta were compiled from lists provided by several recent transcriptomic and proteomic studies (Cao et al., 2020; Chevignon et al., 2015; Gunaratna and Jiang, 2013; He et al., 2016). The identity of M. sexta-specific gene names was validated by searching protein IDs from the M. sexta translated CDS (GCF 014839805.1_JHU_Msex_v1.0) against a non-redundant, high-quality database of complete M. sexta protein sequences reported by He et al. (2016). Significantly differentially expressed genes in these pathways were explicitly evaluated for effects across temperature and parasitism treatment groups. The list of genes for M. sexta and CcBV, their species-specific gene synonyms, and their assigned categories are reported in Dataset 1. GO enrichment was also conducted for each population, but is not particularly informative because of limited availability of M. sexta-specific annotations (Figs S1 and S2).

Parasitoid egg survival assays

To separate the effects of temperature on host immunity versus the ability of parasitoid eggs to survive high temperatures, an in vitro parasitoid egg survival assay was developed and conducted. This assay involved parasitoid egg isolation and subjection to a heat shock that causes complete mortality inside hosts (Malinski et al., 2023; Moore et al., 2021, 2022). Cotesia congregata eggs were dissected from 20 adult females, cultured in Grace's medium in microcentrifuge tubes, and exposed in vitro to either a 40°C heat shock or a control temperature of 25°C for 24 h (>50 eggs per treatment). Cultures were then returned to 25°C and hatching success was assessed 10 days later (eggs typically hatch within 5–8 days post-oviposition at this temperature inside hosts; K.H.M., personal observation). Survival was analyzed via generalized linear mixed models using the glmer function in the ‘lme4’ package (v.1.1.35.1) in R (v.4.3.1). Number of eggs initially cultured and temperature treatment were used as predictor variables, number of hatched larvae observed as the response variable, and female wasp ID as a random effect. To evaluate whether temperature treatment significantly impacted survival to hatching, a simpler model dropping the temperature effect was compared with the full model via Chi-square test. A detailed description of the culturing methods can be found in the Supplementary Materials and Methods.

RESULTS

Host immune assays

Host immune responses to parasitism and temperature differed between populations in several ways (Fig. 2). Parasitism, along with the parasitism×temperature interaction term, was a significant predictor of both immune responses (ANOVA P<0.05; Table S1). Population was a significant predictor of capsule melanization strength (ANOVA P<0.05) but not encapsulation (ANOVA P=0.285). However, the three-way interaction term was a significant predictor of encapsulation (ANOVA P<0.05; Table S1), indicating that the interaction of temperature and parasitism gave rise to different responses in lab and field populations (Fig. 2).

Fig. 2.

Immune responses in M. sexta host populations 24 h after removal from treatment temperatures. (A) Encapsulation, represented by mean encapsulated surface area of hemocytes (mm²); (B) capsule melanization, represented by mean gray value of hemocytes adhering to the filament as a proportion of total black (1=black, most melanized; 0=white, least melanized). Parasitism status is depicted on the x-axis (NP, non-parasitized sham control; P, parasitized), and temperature treatment by line type and color (solid purple line, 25°C control; dashed orange line, 40°C heat shock). Means±s.e.m. are shown as solid points, with raw data shown as semi-translucent points. Sample sizes are indicated by numbers adjacent to points. Statistically significant terms according to two-way ANOVA tests (Table 1) are indicated with asterisks in the top right of each panel (T, temperature; T×P, temperature×parasitism interaction; n.s., not significant).

When lab and field population immune responses were analyzed separately, several differences between their responses were highlighted. For field population M. sexta, parasitism significantly reduced host encapsulation and capsule melanization strength at 25°C (Fig. 2) (ANOVA P<0.05, Table 1), consistent with prior work on the impacts of parasitism alone (Amaya et al., 2005; Lavine and Beckage, 1996). At 40°C, parasitized hosts exhibited a partial recovery of these immune responses (Fig. 2), statistically supported by the temperature×parasitism interaction term in the two-way ANOVA analyses (ANOVA P<0.05; Table 1). Note that the sampling time point for immune assays began 24 h after the heat shock ended (Fig. 1), so immune responses reflect sustained effects of prior heat exposure rather than acute effects.

In the lab population, parasitism reduced capsule melanization at 25°C, but the magnitude of the reduction was weaker than in the field population, indicating that lab population hosts were immunosuppressed by parasitism less effectively than field population hosts (Fig. 2). Parasitized lab hosts also showed a partial recovery of the capsule melanization response in the heat shock relative to control temperature treatment, statistically supported by the significant temperature×parasitism interaction (ANOVA P<0.05; Table 1). Encapsulation, in contrast, was not statistically affected by any treatment in the lab population (Fig. 2, Table 1). Encapsulation strength in unparasitized hosts at 25°C was nearly 2× lower in the lab than in the field population (0.0069 mm² lab, 0.012 mm² field), reflecting differences in baseline encapsulation capacity between these two populations.

Host and CcBV transcriptomics

Read processing, differential gene expression and annotation

Libraries contained 4.9 million reads on average before and 4.5 million reads after trimming. An average of 3.9 million reads per sample aligned to the M. sexta reference genome. An average of 36,317 reads per sample aligned to the CcBV reference genome for parasitized hosts, and 92 for non-parasitized hosts (presumably erroneously, as these hosts were not parasitized and thus should not have CcBV genes expressed) – two libraries from parasitized treatment hosts with <300 reads aligning to the CcBV genome were assumed to be not successfully parasitized and were removed from the analysis. Final sample sizes included libraries from 7–10 samples per treatment group per population. Note that the sampling time point for gene expression was 24 h after the heat shock ended (Fig. 1), so transcriptomic measures reflect sustained effects of prior heat exposure rather than acute effects.

CcBV

Between lab and field host populations, CcBV gene expression was consistent for the number of genes detected after prefiltering (97 for each population), the proportion of genes upregulated and downregulated by the prior heat shock (over a third of detected genes were significantly differentially expressed; Fig. 3) and the identity of the top DEGs (Fig. 4). Gene families most strongly downregulated by prior heat exposure in both host populations were the cystatin, CrV1 and early protein families (with the exception of EP1-like 1) (Fig. 4; Fig. S7), which are likely important for host immunosuppression and manipulation of host physiology immediately following parasitism (Chevignon et al., 2014, 2015; Espagne et al., 2005). Gene families most strongly upregulated by prior heat exposure were the PTP and ankyrin families (Fig. 4; Fig. S7). PTP genes are typically expressed at low abundance early in parasitism, and PTPs likely play a role in manipulating host signal transduction and cellular phosphorylation in the later stages of parasitism (Chevignon et al., 2014). Ankyrins are thought to be important to suppressing early host immune responses in this system (Magkrioti et al., 2011), including melanization in a related species (Bae and Kim, 2009). Significant CcBV DEGs which did not fall into any established gene family are reported in Fig. S3.

Fig. 3.

Number of significantly differentially expressed genes (DEGs) for each treatment and population. Data are for transcripts which aligned to the M. sexta genome (A) or to the CcBV circular genome (B). Differential expression (red, upregulated; blue, downregulated) is shown relative to the reference level for each treatment [P<0.05, log₂fold-change(FC)>|1|]: the reference level was the control temperature for the heat shock treatment and the non-parasitized sham control for the parasitism treatment. DESeq model coefficients are shown at the top: heat shock (T), parasitism (P), interaction between heat shock and parasitism (T×P). For CcBV transcripts, CcBV gene expression is not applicable in non-parasitized hosts, so the heat shock treatment is shown for parasitized treatment groups only.

Fig. 4.

Heatmap of CcBV genes differentially expressed in host fat body tissue in the heat shock relative to the control treatment. Data are for field (left) and lab (right) population hosts. log₂FC in response to the heat shock treatment is shown as a color gradient: dark blue, strongly downregulated; white, no change; dark red, strongly upregulated. The top DEGs with P<0.05 and log₂FC>|1| in at least one population for each gene family are shown. CcBV gene expression is not applicable in non-parasitized hosts, so only parasitized treatment groups are shown. The viral genomic circle is indicated beside each gene name, with integrating circles in bold and non-integrating circles underlined.

CcBV expression patterns were correlated with viral circle integration status. Genes on non-integrating circles were generally downregulated after heat exposure, while genes on integrating circles tended to be upregulated (Fig. 5). Genes on a given circle were always differentially expressed in the same direction, with the exception of two circles in lab population hosts. Downregulation of non-integrating circles was a more robust effect in both populations, whereas upregulation of integrating circles was consistent only for circles 1 and 18 (and partially 32) between the two populations (Fig. 5).

Fig. 5.

Proportion of CcBV genes significantly differentially expressed under the heat shock relative to the control treatment out of all expressed genes detected from each viral circle. Data are for field (A) and lab (B) population hosts. log₂FC in response to the heat shock treatment (P<0.05, log2FC>|1) is indicated (red, upregulated; blue, downregulated). CcBV circles with a demonstrated or predicted ability to integrate into host DNA are shown on the left; those with no evidence for integration are on the right. CcBV circles which have been previously confirmed to integrate into host DNA after a parasitism event are shown in dashed fill; circles with integration status only inferred from a related species (C. typhae bracovirus) are shown in solid fill.

Manduca sexta

Host populations expressed similar numbers of M. sexta genes: 9814 detected in field and 9799 in lab hosts. Lab population hosts had substantially more DEGs than field hosts in response to heat shock, parasitism and their interaction: 0.45–0.87% of genes were differentially expressed in the field population compared with 1.1–4.1% in the lab population (Fig. 3). The response of both populations to the heat shock×parasitism interaction was stronger (more DEGs) than the direct response to parasitism alone, especially in the lab population.

Gene expression patterns for melanization, melanization inhibition and encapsulation pathways did not align with expectations from immune assay results, suggesting that transcriptomic changes in fat body tissue (which supplies many immune components to the hemolymph; Zdybicka-Barabas et al., 2025) do not explain the physiological recovery of encapsulation and capsule melanization in parasitized, heat-shocked hosts. In the field population, parasitism upregulated genes involved in melanization, especially at 25°C (Fig. 6; Fig. S4). In contrast, parasitism reduced expression of melanization-related genes in the lab population at both temperatures. Genes involved in encapsulation were generally downregulated by parasitism in the field population, with little influence of heat shock, while expression in the lab population was largely unaffected by any treatment. Identities of the top host DEGs that did not fall into the established immune or stress response pathways are reported in Fig. S5.

Fig. 6.

Heatmap for M. sexta DEGs involved in immune and heat stress response pathways. Pathways are indicated on the right of the panels for the field (left) and lab (right) population of M. sexta. Heatmap columns correspond to differential expression for each term: heat shock (T), parasitism (P), interaction between heat shock and parasitism (T×P). Reference levels for each treatment were control temperature (25°C) for the heat shock treatment and non-parasitized sham control for the parasitism treatment. Genes with confirmed function in M. sexta pathways are in bold; otherwise, genes have predicted function based on homology (according to GO terms). The top 20 significantly DEGs with P<0.05 and log₂FC>|1| in at least one term for each category are shown (note, some categories had fewer than 20 genes with significant differential expression).

Parasitoid egg survival assays

To test whether parasitoid eggs are capable of surviving high temperatures, the effect of a 24 h heat shock on the survival of C. congregata eggs cultured in vitro was assessed. Survival to hatching was observed in 5 out of 20 female clutches (25%), with overall hatch rates of 3.1% at 40°C and 5.9% at 25°C (Fig. S6). The generalized linear mixed effects models indicated no significant effect of temperature on hatching (Chi-square, d.f.=1, P=0.873). Survival was poor at both temperatures, likely due to the disturbance of dissection and the imperfect in vitro simulation of the internal host environment. Given low hatch rates across both treatments, quantitative estimates of temperature effects on egg survival using this in vitro culturing method are limited, and the possibility that eggs are negatively impacted by heat cannot be ruled out. However, the presence of at least some parasitoid hatchlings in multiple replicates of the high temperature treatment suggests that parasitoid eggs are capable of surviving exposure to high temperature conditions previously demonstrated to cause complete mortality inside hosts (40°C for 24 h; Moore et al., 2022).

DISCUSSION

High temperature events associated with climate change are disrupting host–parasitoid interactions and reshaping their surrounding communities (Le Lann et al., 2021; Malinski et al., 2024). Parasitoids often display lower thermal tolerance than hosts, but the role of symbionts in mediating this pattern is not well understood. The goal of this study was to evaluate the physiological and molecular mechanisms involved in the thermal mismatch between a model host–parasitoid–endosymbiont system. Our results showed that at least some parasitoid eggs are capable of surviving heat shocks of the same temperature and duration previously demonstrated to cause complete mortality inside hosts (Malinski et al., 2023; Moore et al., 2022). While low hatch rates across the experiment limited the possibility of quantifying the effects of temperature on survival, these results suggest that the thermal mismatch in this system was likely not driven by increased thermal sensitivity in the egg stage. We then tested whether high temperatures disrupted the typical immunosuppression of the host by the virus, and characterized host and symbiont transcriptomic profiles underlying these responses.

High temperatures impair host immunosuppression

Heat shock resulted in a partial recovery of immune responses in hosts typically immunosuppressed by parasitism. We found significant interactive effects of temperature and parasitism treatments on encapsulation and capsule melanization immune responses in both populations, with the exception of encapsulation in the lab population (discussed below) (Fig. 2, Table 1). These results provide strong evidence that heat disrupts host immunosuppression, which may help explain the thermal mismatch in this system during the early parasitoid developmental window (Malinski et al., 2023; Moore et al., 2022). This effect was sustained for at least 24 h after the heat exposure ended, suggesting that high temperature events can have lasting impacts on physiological and transcriptomic responses in host–parasitoid–symbiont interactions. We also found that in both the immune assays and transcriptomic analyses, lab population hosts displayed weaker baseline encapsulation capacity in the double control treatment (i.e. in the absence of parasitism or heat stress) (Figs 2 and 6). This effect was likely due to immunological differences between host lineages, potentially arising from relaxed selective pressure for response to multicellular parasites in domesticated lab environments (Alston et al., 2020; Diamond and Kingsolver, 2011).

Transcriptomic analyses revealed that a large proportion (over a third) of detected CcBV genes were significantly differentially expressed following heat shock in parasitized hosts from both populations (Fig. 3), including genes with key roles in host immunosuppression (discussed below). CcBV genes showed markedly similar patterns of expression between host populations (Figs 3 and 4), despite previously established population differences in heat tolerance and immune responses (Alston et al., 2020; Diamond and Kingsolver, 2011). The consistent CcBV gene expression in lab and field hosts suggests that the influence of temperature on the virus was a robust response to heat rather than a response to heat-induced changes in host physiology.

The physiological immune assays demonstrated clear interactive effects of parasitism and high temperature (with the exception of the encapsulation response in the lab population, discussed above) (Fig. 2), but these effects did not appear to be driven by changes in host gene expression in fat body tissue. Parasitism reduced expression of encapsulation-related genes (Fig. 6; Fig. S4) consistent with other research (Chevignon et al., 2015), but heat had no significant effect on the expression of these genes in fat body. Similar results were observed for melanization genes in the lab population and melanization inhibition genes in both populations, e.g. expression of several serpins (Fig. 6; Fig. S4). There was a subset of unexpectedly upregulated genes which fell into the melanization category (collagenase, trypsin and chymotrypsin families; Fig. 6) capable of producing melanin and other cytotoxic compounds but are not part of the known melanization pathway in M. sexta (Beck et al., 2001), suggesting they could be involved in some putative novel role in immunity or stress. These transcriptomic results suggest that the reduced suppression of host immune responses at high temperatures (Fig. 2) is not driven by transcriptomic changes in host fat body tissue. Additional work is needed to determine causal molecular mechanisms, e.g. evaluating transcriptomic changes in hemocytes after parasitism and heat shock, and to confirm whether the observed circle-dependent changes in CcBV gene expression (Figs 3 and 4) are involved (discussed below).

Transcriptomic response to heat varies according to viral genomic architecture

The results of this study indicate that high temperatures impaired typical CcBV gene expression in parasitized hosts, and these changes were associated with the virus's genomic organization. Genes located on the same viral circular genome circle were almost always differentially expressed in the same direction (Fig. 5). This directionality depended on whether viral circles exhibited the ability to integrate into host DNA or not (Chevignon et al., 2018; Muller et al., 2021). Gene expression from circles known to integrate into host DNA showed upregulation or little change following the heat shock, while gene expression from circles which are not known to integrate into host DNA were generally downregulated (Fig. 5). In some cases, this circle-dependent differential expression was observed for 75–100% of genes present on the circle (e.g. integrating circle 18 and non-integrating circle 9; Fig. 5). Note, however, that prior research determining circle integration status in this species evaluated only hemocytes, not fat body tissue (Chevignon et al., 2018). It is possible that viral circle integration may be consistent across fat body and hemocyte tissues because integration mechanisms operate through host integration motifs, which are genomic and should not differ between cell types (Chevignon et al., 2018).

The roles of integrating and episomal viral genomic circles are not fully understood and likely involve additional complexity. Firstly, exceptions to the circle-dependent expression patterns, such as circle 17 (Fig. 5), could reflect differences in the timing of circle integration, for example if circle 17 does not integrate into host DNA until later in parasitism (as seen in other bracovirus species; Wang et al., 2021). Depending on the parasitoid species, some bracoviruses use only integrating circles (e.g. MdBV: Beck et al., 2011), while some appear to use a mix of integrating and episomal circles (e.g. GiPDV: Gundersen-Rindal and Dougherty, 2000; and CtBV: Muller et al., 2021) and others show no evidence of host integration machinery (e.g. CiV: Annaheim and Lanzrein, 2007), though the vast majority of species have not been evaluated.

Integration status and the functions of genes encoded on integrating and episomal circles may be intertwined with the temporal dynamics of host manipulation as parasitism progresses from early to late stages (Kim et al., 2023; Wang et al., 2021). For example, the integration of viral genes into host DNA can enable the persistence of viral gene expression in hosts by preventing viral dilution as host cells replicate (Beck et al., 2011; Muller et al., 2021). These genes are likely important to the later stages of parasitism, e.g. for disrupting host development and behavior during parasitoid pupation (Adamo, 2019; Adamo et al., 2016; Beck et al., 2011). The circle-dependent transcription in response to high temperatures (Fig. 5) suggests that integration could also play a protective role against abiotic factors such as heat stress. Along the same line, the episomal transcription of genes on non-integrating circles are hypothesized to enable the rapid production of viral gene products immediately after oviposition (Harwood et al., 1998). The potential role of viral circle integration in the temporal dynamics of viral gene expression could explain why high temperature events can disrupt host–parasitoid–endosymbiont interactions differently depending on the timing of exposure. Future research should evaluate whether episomal circles impose tradeoffs with thermal stability in other parasitoid bracovirus species.

Although most polydnavirus genes remain poorly characterized, we found that several classes of viral genes with established or hypothesized roles in immunosuppression were downregulated following heat exposure (cystatin, CrV1 and early protein gene families) (Fig. 4). Cystatin proteins are typically expressed at high levels soon after parasitism and inhibit host cathepsins involved in antiviral immune defense (Espagne et al., 2005; Saikhedkar et al., 2015). CrV1 encodes a glycoprotein that disrupts hemocyte encapsulation by interacting with actin and hemolin, a host recognition protein (Labropoulou et al., 2008). Early proteins, some of the earliest expressed viral products, are hypothesized to drive immediate changes to host physiology following oviposition (Harwood et al., 1998; Harwood and Beckage, 1994; Le et al., 2003).

We also observed enhanced expression of PTP and ankyrin genes in both populations after heat shock, more strongly in the field population (Fig. 4). PTP genes are the largest family of CcBV genes, implicated in disrupting signal transduction and cellular phosphorylation (Gao et al., 2022; Provost et al., 2004). There is evidence that ankyrins, including the two upregulated in this study, are involved in early immunosuppression of the M. sexta host by CcBV via suppression of the antimicrobial peptide pathway (Magkrioti et al., 2011), and in a congener species, via suppression of induced melanization (Bae and Kim, 2009). The upregulation of ankyrin 5 and ankyrin 6 under high temperatures in the present study is not consistent with the results of the immune assays (Fig. 2); however, CcBV expression can vary across host tissues (Chevignon et al., 2015), so this result could differ between fat body and hemocytes, which was not tested here. Similarly, host melanization genes also did not display expression patterns consistent with the immune assays (Fig. 6; Fig. S4). Sampling additional immune-relevant tissues will be required to establish whether these gene products are linked to the loss of immunosuppression under heat demonstrated in the immune assays.

While our study provides potential explanations for the disruption in host–parasitoid ecology under high temperatures, these data are not definitive, and evaluation of alternative explanations is needed. A logical next step for future inquiry will be hemocyte motility, abundance and gene expression in parasitized, heat shocked hosts. Endocrine changes and intermediate metabolism may also be important to host melanization (Thompson and Kavaliers, 1994), which should be evaluated for its role in host–parasitoid thermal mismatch. Additionally, while venom is not required for immunosuppression or parasitism success in this species, recent work suggests that venom may be involved in early host immunosuppression (Moreau et al., 2025), and venom proteins can enhance CcBV's effectiveness via cross-reaction with virus-produced proteins (Beckage et al., 1994). High temperature could impact venom proteins, as demonstrated in other parasitoid species (Cavigliasso et al., 2021), but was not evaluated here. In addition, this study did not consider teratocytes (specialized parasitoid cells involved in host manipulation; Strand, 2014). Further study is needed to explore the consequences of heat stress during parasitism on these biological components of host–parasitoid–symbiont interactions.

Conclusions

This study demonstrates how extreme high temperature events, which are increasing with climate change (Buckley et al., 2023; Parmesan, 2006), can impact interactions between species at the molecular and physiological levels to shape downstream organismal and ecological outcomes. The studies described here provide evidence that the interactions between a host, parasitoid and symbiont are altered immunologically by heat, which may be associated with the thermal mismatch in host–parasitoid interactions between M. sexta and C. congregata. This work provides some of the first evidence that the dynamics of polydnavirus genomic circle integration into host DNA may play an important role in parasitoid thermal tolerance. Further research is needed to characterize the biological roles of polydnavirus circle integration versus non-integration in interactions with the host insect, to determine the extent to which these effects are involved in the lower parasitoid thermal tolerance relative to hosts, and to characterize cascading ecological consequences for the broader community (e.g. hostplants; Parker and Kingsolver, 2024). Determining how these effects interact and scale is crucial for improving our mechanistic understanding of how molecular and immunological processes give rise to community- and ecosystem-level responses to a changing climate.