Social insects are prone to pathogen infection because of high exposure rates from social interactions. However, it remains unclear whether queens have enhanced pathogen resistance, since reproduction is largely confined to queens. Here, we used a natural host-pathogen system, the subterranean termite Reticulitermes chinensis and the entomopathogenic fungus Metarhizium anisopliae, to investigate the differences in allo-grooming, locomotion, and immune gene expression between queens and workers against pathogen infection. We found that fungal infection significantly reduced survival in both queens and workers. Infected queens received significantly more grooming time from sanitary nestmates than infected workers, but they returned much less grooming time to sanitary nestmates than infected workers. Infection resulted in a reduction in the average locomotion speed and distance of queens but had no effect on worker locomotion. Infection resulted in upregulated expression of two immune genes (termicin and transferrin), two antioxidant genes (CAT and SOD), and phosphate genes CYP450 in queens but not in workers. Our results indicated that eusocial termites evolved strategies that prioritize the reproductive castes welfare in defending against the pathogen infection to ensure continued reproduction and colony persistence.

Study termites: Two colonies of R. chinensis were collected from Lion Mountain, Wuhan, China in 2022. Adult termites were split out from the cracks of dead wood and each 50 individuals (48 workers and 2 soldiers) were housed in a 6-cm Petri dish containing moist filter paper as a food and water source. Each colony was divided into 60 sub-colonies (120 sub-colonies in total). All sub-colonies were kept in a climatic chamber (25^oC, 70% relative humidity, 12 h day/night cycle). Once a week, we checked the consumption of a piece of moist filter paper and provided corresponding supplements. Previous studies have shown that workers of R. chinensis are able to develop into neotenics (supplementary reproductive) when the primary queen dies or is lost (Su et al., 2015). Here, in our observations, after about three weeks, the workers will differentiated into neotenics. We classed the neotenics as queens for all experiments. These all sub-colonies with a queen were used for the following experiments.

Fungal pathogens: We used the entomopathogenic fungus M. anisopliae (strain NRRL13969) to infect the focal workers and queens. The fungus was cultivated on potato dextrose agar (Solarbio, China) for 3 weeks, and used 0.1% Tween 80 to collect the conidial suspensions, then stored at 4°C. All conidial suspensions had a germination rate of >95% as determined directly before starting each experiment. We exposed individual termite workers by applying 0.35μl droplet of a suspension of 10⁴ conidia/ml in 0.1% Tween 80 (infected treatment). and the control treatment consisted of an application of a 0.35μl droplet of a 0.1% Tween 80 only. Subsequently, all termites were refrigerated at 4°C for one hour to slow their movements and to precipitate the conidia onto their cuticle before starting the experiments (Traniello et al. 2002).

Survival: To determine the effect of infection with M. anisopliae on the different termite castes, 100 workers and 100 queens were randomly picked out from 100 sub-colonies (50 sub-colonies per colony), and each termite individual was assigned to one of four treatments: control worker (n=50), control queen (n=50), infected worker (n=50), and infected queen (n=50). All of these treatments were reared individually in 3.5-cm Petri dish with moist filter paper for providing food and water, and kept in a constant climatic chamber (25°C, 70% relative humidity, 12 h day/night cycle). The number of dead termites was recorded daily over 12 days, after which the dead termites were removed. 

Locomotion and allogrooming: To observe the locomotion and allogrooming behavior of workers and queens, we randomly picked 8 individuals (7 workers and 1 queen) from each sub-colony to be placed into a new 3.5-cm Petri dishes, totally set 12 groups (6 control groups and 6 infected groups). Of these 8 individuals, 6 workers were provided to act as caregivers, a single queen and another worker were infected with Tween 80 or M. anisopliae. Each individual was marked by a unique paint color dot on its thorax (Street Graffiti, purchased by MINISO, Guangzhou China). All replicates were kept in a constant climatic chamber (25^oC, 70% relative humidity, 12 h day/night cycle) for three days to allow them to adapt to their new environment. Every 24h, we checked the color code of each individual, and if the paint of an individual was missing, we re-marked it immediately.

After the three-day adaptation period in the new Petri dish, one queen and one worker from each of 6 groups were treated with conidia-free 0.1% Tween 80, and one queen and one worker from each of the other 6 groups were infected with M. anisopliae as described above. The behaviors of these 12 groups were recorded with a camera (AX100E, SONY) for 30 min. The mortality rates of control workers, control queens, infected workers, and infected queens of these 12 groups were recorded after the behavioral measurements were taken. Ultimately, the video recordings were used to analyse the behaviors of allogrooming and locomotion related data with the help of a computerized EthoVision XT 15 tracking system (Noldus Information Technology, Beijing China): 1) grooming-received: the total duration time of control and infected individual received grooming from other 7 nestmates; 2) grooming-given: the total grooming time given to the other 7 nestmates by the control and infected individual; 3) velocity: the average moving speed of control and infected individuals in the petri dish during 30-min observation; 4) distance: the total distance that the focal individual covered during 30-min observation. EthoVision was also used to generate motion trajectories and heatmaps for the distribution of the termites around the test arenas during each of the 30-min recordings (for details see Appendix S1).

Quantitative real-time PCR (RT-qPCR): The expression of six genes termicin, transferrin, cytochromeP450 (CYP450), glutathione S-transferase (GST), superoxide dismutase (SOD), and catalase (CAT) were analyzed in control workers, control queens, infected workers, and infected queens by real-time PCR. We randomly picked out 40 workers and 8 queens from 8 sub-colonies, of which 20 workers and 4 queens were treated with conidia-free 0.1% Tween 80, and another 20 workers and 4 queens were infected with M. anisopliae. Four replicates (each worker replicate containing a pool of five workers and queen replicate containing one queen) were set up for each treatment (2 replicates per colony). The gene expressions were measured 48h after initial fungal infection.

Each sample was crushed in a 1.5-mL centrifuge tube with liquid nitrogen, using sterilized disposable tissue grinding pestles. Total RNA was extracted from the whole body of workers and queens using RNeasy mini kit (QIAGEN), respectively. After RNA extraction, the quality and quantity (purity of protein and salt) were analysed using a NanoDrop Lite spectrophotometer (Model: ND Lite-4470). Then the cDNA was synthesized from the RNA by GoScriptTM Reverse Transcription Mix, Oligo(dT) (Promega), and was preserved at -20˚C. The quantitative real-time PCR (qRT-PCR) was performed on the Applied Biosystems® StepOne™ system (Thermo) with QuantiNovaTM SYBR® Green PCR Kit (Qiagen). All the procedures were implemented in accordance to the manufacturer’s protocol. Since β-actin is considered the most reliable reference gene for Reticulitermes termites (Tong et al. 2009; Ishitani and Maekawa, 2010), this was included in our experiment. Relative gene expression levels were calculated using a typical 2^-ΔΔCt method. We performed four biological replicates and four experimental replicates for all RT-qPCR experiments. The primers used for real-time PCR are listed in Table 1.

Statistical analysis: All statistical analyses were performed with IBM SPSS Statistics version 22.0 and the program R v3.6.3. All graphics were produced with GraphPad Prism 8.0. Cox proportional regression was used to evaluate the effect of variables (separately reared and group-reared) on the survival of termites. Since the survival curves of workers and queens in different treatments (control vs. infected) were crossing, a two-stage method (Qiu & Sheng, 2008) was used using R’s TSHRC package. For the allogrooming behavior, locomotion behavior, and immune-gene expression, we applied the generalized linear mixed model (GLMM) with a normal distribution to analyze the effects of castes (workers vs. queens) and treatment types (control vs. infected), as well as their interaction, respectively. In these three models, the castes and treatment types as the fixed factors, and the colony as the random factor. In addition, post hoc pairwise comparisons with LSD tests were performed to examine the differences of allogrooming, locomotion, and immune gene expression in workers and queens under different treatments (control and infected).