Immune Stimulation via Wounding Alters Chemical Profiles of Adult Tribolium castaneum

Group-living individuals experience immense risk of disease transmission and parasite infection. In social and in some non-social insects, disease control with immunomodulation arises not only via individual immune defenses, but also via infochemicals such as contact cues and (defensive) volatiles to mount a group-level immunity. However, little is known about whether activation of the immune system elicits changes in chemical phenotypes, which may mediate these responses. We here asked whether individual immune experience resulting from wounding or injection of heat-killed Bacillus thuringiensis (priming) leads to changes in the chemical profiles of female and male adult red flour beetles, Tribolium castaneum, which are non-social but gregarious. We analyzed insect extracts using GC-FID to study the chemical composition of (1) cuticular hydrocarbons (CHCs) as candidates for the transfer of immunity-related information between individuals via contact, and (2) stink gland secretions, with analysis of benzoquinones as main active compounds regulating ‘external immunity’. Despite a pronounced sexual dimorphism in CHC profiles, wounding stimulation led to similar profile changes in males and females with increases in the proportion of methyl-branched alkanes compared to naïve beetles. While changes in the overall secretion profiles were less pronounced, absolute amounts of benzoquinones were transiently elevated in wounded compared to naïve females. Responses to priming were insignificant in CHCs and secretions. We suggest that changes in different infochemicals after wounding may mediate immune status signaling in the context of both internal and external immune responses in groups of this non-social insect, thus showing parallels to social immunity. Supplementary Information The online version contains supplementary material available at 10.1007/s10886-022-01395-x.


Bacteria Cultivation and Immunostimulant Preparation
To prepare the bacterial culture, 100 µL of a Bacillus thuringiensis bv. tenebrionis freezer stock (stored at -80 ºC) was inoculated into a 500 mL baffled flask containing 50 mL Luria Broth (Carl Roth GmbH + Co. KG) and the culture was incubated at 30 ºC, 200 rpm for 15 h in dark.
Subsequently, the culture was transferred into a 50 mL Falcon conical tube and centrifuged at 3,645 × g at 4 ºC for 10 min. The bacterial pellet was washed once with phosphate buffered saline (PBS, Calbiochem®) and the concentration adjusted to 1 × 10 9 cells mL -1 . PBS (as injection/wounding control) and bacteria solution (for heat-killed B. thuringiensis bv. tenebrionis injecting treatment, 'priming') were heated at 95 ºC for 30 min to heat-kill the bacteria. All procedures were carried out under a horizontal laminar air-flow hood to keep the conditions sterile.

Details on Data Processing
Data pre-processing was done using the GCsolution Postrun software (Version 2.30.00, Shimadzu), exporting peak areas with at least 100 (CHCs) and 5 counts (stink gland secretion), respectively.
Retention time alignment of peaks was done in R (version 4.0.3, (R Core Team 2020) using the package GCalignR (Ottensmann et al. 2018), allowing a maximum shift in retention time by 0.05 min. Features that could not be distinguished from those occurring also in blank samples were excluded. The resulting datasets were both normalised to the internal standard and only features detected in at least half of the replicates per treatment and sex (and time point in case of stink gland secretions) were considered for further analysis. When relative values were used for analysis, these were calculated by expressing the amount of a certain feature as the proportion of the total amount of features considered for a sample.

Details on Statistical Analyses
All statistical analyses were done using R (version 4.0.3). To visualise dissimilarities between groups of beetles, non-metric multidimensional scaling (NMDS) was applied to all datasets using the metaMDS function in the package vegan (version 2.5-7, (Oksanen et al. 2020). For this, datasets For the secretions, we further used the datasets acquired 24 h after treatment and identified features that differed between treatments in both females and males using a conditional random forest classification (seed 245, n trees = 500, n variables per split = 5) (Strobl et al. 2009a, b;De Moraes et al. 2014). The analysis was done on relative feature amounts to control for different total amounts of secretions due to individual variations and random gland emptying before and during experimental procedures, thereby capturing changes in the proportion of features to one another.
Random forest analysis identified one feature as important for each sex, whose relative concentrations were further analysed in post hoc testing. We tested for data normality, heteroscedasticity, presence of multicollinearity and influential cases and obtained diagnostics plots (Q-Q plots and histograms of model residuals distribution and a plot to estimate non-linear patterns of residuals). Not meeting the model assumptions, the two features were analysed with nonparametric Kruskal-Wallis tests and equivalent pairwise comparisons between treatment groups.
We further analysed the effect of treatments on the absolute concentration of secretion compounds with known biological activity, EBQ, MBQ and 1-pentadecene (1-C15-ene) separately for both sexes at both time points using linear models (LMs). We tested the model assumptions as described above and employed the package bestNormalize (Peterson 2021)  Cox. When significant treatment effects were founds, Tukey HSD post hoc comparisons were performed between treatment groups with P-value corrections using the BH procedure (a = 0.05).      Table S1 Cuticular hydrocarbons (CHCs) and their relative composition in Tribolium castaneum beetles of different immune treatment groups (N = 6 per sex and treatment). Beetles were untreated ('Naïve'), or treated by injection of phosphate buffered saline ('Wounded') or of phosphate buffered saline with heat-killed Bacillus thuringiensis bv. tenebrionis ('Primed'), 18 h before extraction. Features were putatively identified by comparing their retention index (RI) value to Lockey (1978), Alnajim et al. (2019) and Awater-Salendo et al. (2020). Forward slash between positions of methyl groups in CHC names indicates that positions are unclear or mixtures could not be separated. If literature suggested different identities, names were given based on better matching RI value (No. 12), better peak separation (No. 6, 7), or most recent publication (No. 3,8,11,12,15 Li et al. (2013). Average proportions of all features are shown as percentage of the total amount of all features considered for a sample. The average proportions were calculated across females and males of different treatments and time points. * Identity confirmed by reference standard. ** Li et al. (2013) state that positions of double bonds are assumed based on matching peaks in Suzuki et al. (1975) and Görgen et al. (1990). Suzuki et al. (1975)