Trans‐generational plasticity in response to immune challenge is constrained by heat stress

Abstract Trans‐generational plasticity (TGP) is the adjustment of phenotypes to changing habitat conditions that persist longer than the individual lifetime. Fitness benefits (adaptive TGP) are expected upon matching parent–offspring environments. In a global change scenario, several performance‐related environmental factors are changing simultaneously. This lowers the predictability of offspring environmental conditions, potentially hampering the benefits of TGP. For the first time, we here explore how the combination of an abiotic and a biotic environmental factor in the parental generation plays out as trans‐generational effect in the offspring. We fully reciprocally exposed the parental generation of the pipefish Syngnathus typhle to an immune challenge and elevated temperatures simulating a naturally occurring heatwave. Upon mating and male pregnancy, offspring were kept in ambient or elevated temperature regimes combined with a heat‐killed bacterial epitope treatment. Differential gene expression (immune genes and DNA‐ and histone‐modification genes) suggests that the combined change of an abiotic and a biotic factor in the parental generation had interactive effects on offspring performance, the temperature effect dominated over the immune challenge impact. The benefits of certain parental environmental conditions on offspring performance did not sum up when abiotic and biotic factors were changed simultaneously supporting that available resources that can be allocated to phenotypic trans‐generational effects are limited. Temperature is the master regulator of trans‐generational phenotypic plasticity, which potentially implies a conflict in the allocation of resources towards several environmental factors. This asks for a reassessment of TGP as a short‐term option to buffer environmental variation in the light of climate change.

Under ongoing global change, shifts in abiotic factors (such as increased temperature variations or higher CO 2 levels) are accompanied by changes in biotic factors (e.g., parasites, predators) that may also favour acclimatization via the transfer of parental environmental experience (Vidal-Martinez, Pech, Sures, Purucker, & Poulin, 2010). Parasites are abundant and affect all living organisms (Price, 1990;Windsor, 1998), but their distribution, replication and virulence are susceptible to the conditions in their respective environments (Lafferty & Kuris, 1999;Marcogliese, 2008). The coevolutionary arms race between host and parasite (Anderson & May, 1982;Hamilton, 1980) selects for host immune systems (Altizer, Harvell, & Friedle, 2003;Boots & Bowers, 2004) that exhibit genetic specificity and phenotypic plasticity enabling them to combat abundant parasites and pathogens (Schmid-Hempel, 2011). If environmental change induces a shift in the parasite assemblage, this host plasticity promotes an efficient response that mediates acclimatization (Lazzaro & Little, 2009). The plasticity can be transferred from the parents to the offspring in invertebrates and vertebrates (trans-generational immune priming, TGIP; Beemelmanns & Roth, 2016a& Roth, , 2016b& Roth, , 2017Grindstaff, Brodie, & Ketterson, 2003;Hasselquist & Nilsson, 2009;Roth, Klein, Beemelmanns, Scharsack, & Reusch, 2012;Sadd, Kleinlogel, Schmid-Hempel, & Schmid-Hempel, 2005). TGIP boosts offspring immunity and is in vertebrates of particular importance during early life stages, when mortality selection is high due to the immature adaptive immune system (Hasselquist & Nilsson, 2009). As a particular case of trans-generational phenotypic plasticity, TGIP has the ability to compensate for the impact of alterations in pathogen and parasite assemblies as predicted under climate change.
Trans-generational plasticity has been shown to buffer single abiotic and biotic changes (Beemelmanns & Roth, 2016b;Donelson et al., 2012;Roth, Klein, et al., 2012;Salinas & Munch, 2012;Shama et al., 2014). However, whether TGP also has the potential to compensate multiple environmental modifications simultaneously remains unknown. Addressing this gap is of particular importance as anthropogenically induced global change alters a wealth of environmental factors concurrently, major threads in the ocean are higher temperature variations, enhanced CO 2 concentrations, a drop in salinity and the intermingled increased pathogen replication and virulence (Brook, Sodhi, & Bradshaw, 2008;Hoegh-Guldberg et al., 2007;Walther et al., 2002). Trans-generational acclimatization towards different environmental variables may involve the same physiological mechanisms. In such a scenario, TGP as a response to two changing environmental factors could be more than additive resulting in a synergistic beneficial offspring response. However, both TGP and TGIP are costly (Sadd & Schmid-Hempel, 2009;Sheldon & Verhulst, 1996), and they may only be adaptive when the offspring environment is predictable and matches the parental environment (Burgess & Marshall, 2014;Donald-Matasci, 2013;Fischer, Taborsky, & Kokko, 2011;Marshall & Uller, 2007;Mousseau & Fox, 1998). For example, if a temperature shift in the parental generation is combined with induced pathogen prevalence, the energetic costs for plastic acclimatization can rise. In addition, the probability for matching parental and offspring environments is diminished. This increases the selection for genetic adaptation, but also implies greater costs on the parental and offspring side due to reduced reproduction and survival (Carroll et al., 2014). This may lower the extent and benefits of TGP and induces costs. Two environmental factors that change in the parental generation could thus have antagonistic effects on offspring fitness, fading out the impact of TGP.
Investigating the synergistic and antagonistic impact of two changing environmental factors in the parental generation can help to unravel the limits of TGP.
To address the trans-generational impact of two interacting parental abiotic and biotic environmental changes, the parental generation of wild-caught pipefish of the species Syngnathus typhle were in a fully reciprocal mating design exposed to an ambient and an elevated temperature regime (18°C [cold] vs. 23°C [hot]) and immunological challenges with heat-killed Vibrio bacteria or no immunological activation. 18°C represents the ambient temperature during the breeding season along the coast of the western Baltic Sea, while increasing temperatures from 18°C to 23°C within 7 days represent a heatwave, which both in the temperature and the rate of change resemble natural conditions that have been observed in coastal environments during the last 20 years (Benston, Stephenson, Christensen, & Ferro, 2007;Schär & Jendritzky, 2004;Team, 2008;Vidale, 2007). The exposure to an immune challenge with heat-killed Vibrio bacteria mimicked an immunological activation upon a successful parasite infection. Pipefish pairs were fully reciprocally formed and mated (four parental environments: Vibrio cold: VC; Vibrio hot: VH; naïve cold: NC, naïve hot: NH).
Duration of male pregnancy and clutch size were evaluated. To simulate matching and nonmatching environmental conditions, offspring of each family were split directly after birth into cold or hot offspring environment. Half of the offspring from each temperature treatment were then exposed to an immune challenge with heat-killed Vibrio bacteria (four offspring environments: VC; VH; NC, NH), upon which gene expression and life-history responses were assessed.
We hypothesize that parental immune challenge and heatwave share the same physiological mechanisms and thus have an interactive effect on offspring performance. The direction of these effects can either be antagonistic such that the impact of parental Vibrio exposure on offspring immune defence (TGIP) is reduced when heat stress is experienced simultaneously. Or the experience of two environmental factors in the parental generation can also have synergistic beneficial impact on offspring performance, in particular in case of a matching parental and offspring environment (Burgess & Marshall, 2014;Mousseau & Fox, 1998;Salinas & Munch, 2012;Uller, 2008).
Alternatively, we hypothesize that parental immune challenge and temperature change do not interact but involve separate physiological mechanisms relying on segregated resource pools. Focusing on these two hypotheses, we assessed expression of a limited gene set (44) involved in immune defence or DNA modification and histone modification with a known role in TGIP (Beemelmanns & Roth, 2016a, 2016bRoth, Klein, et al., 2012). As life-history responses, we measured duration of male pregnancy, clutch size and offspring size.
Consistent with earlier studies in the pipefish S. typhle, parental Vibrio exposure induced offspring gene expression. However, if parents were exposed to elevated temperatures (23°C [hot]) in combination with an exposure to Vibrio, the impact of TGIP almost disappeared.
This supports the hypothesized interactive effects of parental immune challenge and exposure to elevated temperature on offspring gene expression.

| Experiment
The parental pipefish generation was caught end of April 2014 by snorkelling with handnets in a seagrass meadow in northern Germany near Gelting in 1-3 m water depth. Fish were transported within 2 hr after capture to our aquaria facilities at GEOMAR in Kiel and slowly acclimated to laboratory conditions (18°C, salinity according to natural conditions in Kiel harbour [14][15][16][17][18]). During this time, animals were kept sex-separated in six 200-L barrels. All animals were healthy and did not show any symptoms of ongoing infections. On May 16, fish were moved into the glass aquaria system, two fish of the same sex per 80-L aquarium, a total of 64 fish in 32 aquaria. The next day, these aquaria were randomly assigned to either the "cold" or the "hot" circulation system (16 aquaria each).
In the aquaria belonging to the "hot" parental group, temperature was slowly raised by 1°C per day until 23°C was reached. This represents a recent heatwave scenario that can occur in summer in coastal waters of the Baltic Sea (Benston et al., 2007;Schär & Jendritzky, 2004). The other aquaria remained at 18°C. Fresh Baltic sea water was added daily (300 L/day, about 10% of the water in the aquaria system). In addition, water was exchanged between the two circulation systems to avoid confounding effects of the distinct aquaria systems.
On May 22, animals from the cold and the hot group were either exposed to an immune challenge with heat-killed Vibrio bacteria (peritoneal injection of 50 μl of 10 9 bacteria/ml) of an Italian Vibrio isolate I9K1 (Beemelmanns & Roth, 2016a, 2016bRoth, Keller, Landis, Salzburger, & Reusch, 2012;Roth, Klein, et al., 2012), or they were left naïve. In each aquarium, one female and one male were placed that were either both immune-challenged but kept at cold temperatures (VC), both immune-challenged but exposed to elevated temperatures (VH), both not immune-challenged and kept at cold temperatures (NC) or both not immune-challenged and exposed to elevated temperatures (NH). These pairs were designated as parental generation according to the four different treatments: NC, NH, VC, VH, eight replicates each, in total 32 pairs. A total of 29 pairs mated within 48 hr (7 NC, 6 NH, 8 VC, 8 VH), thereafter named families. The temperature was kept at 18°C or 23°C, respectively, throughout pregnancy. During pregnancy, seven families were either lost due to death (two families), as they jumped out of the aquaria (three families), or as they lost their brood (two families). Males of 22 families successfully gave birth to offspring (seven NC, five NH, six VC, four VH) between June 9 and 18. Pregnancy lasted on average 17.4 ± 0.167 days (mean ± SE) for animals kept at hot temperatures, and 23.6 ± 0.55 days for animals kept at cold temperatures. Males were checked for signs of ongoing birth four times a day. Immediately after birth, the clutch was split and transferred into small aquaria (2 L), one kept at 18°C, the other one at 23°C. The small aquaria were swimming in the large tanks of the aquaria system due to their polystyrene surrounding, and water exchange was permitted over two circular sections that were cut and covered with fine-mesh nets.
Eight days after birth, juveniles of each half-clutch were again split and either exposed to an immune challenge with heat-killed Vibrio bacteria over a pricking with a needle dipped in a solution of 10 10 heat-killed Vibrio bacteria/ml, or stayed without an immunological treatment (naïve) as described in Roth (2016a, 2016b). For this treatment, four families per parental treatment (16 families in total) with an equal distribution of five replicates per each of the four offspring treatment groups (NC, NH, VC, VH) were used (20 offspring per family resulting in 16 × 20 = 320 animals). 24 hr after the offspring immune challenge, the total length of the animals was measured in mm (from the tip of the snout to the tip of the caudal fin), and animals were killed by decapitation and for later usage stored in RNA later at −20°C.
We quantified the mRNA level of 44 preselected target genes already used in previous studies (Beemelmanns & Roth, 2016a;Birrer, Reusch, & Roth, 2012;Roth, Keller, et al., 2012;Roth, Klein, et al., 2012) with quantitative real-time polymerase chain reaction (qPCR). The genes were originally identified and selected from transcriptomes of several S. typhle individuals that were previously exposed to natural Vibrio isolates (Haase et al. 2013).
Total RNA was extracted of 320 whole-body samples with an and 3.15 μl pre-amplified PCR products. An assay mix was prepared by combining 0.7 μl of 50 μM primer pair mix, 3,5 μl Assay Loading Reagent (Fluidigm) and 3.15 μl low EDTA-TE buffer. At the end, 5 μl of each sample and assay mix were filled into the GE chips and measured in the BioMark system, applying the GE-fast 96.96 PCR protocol according to the manufacturer's instructions (Fluidigm). The samples were distributed randomly across chips, and each of these included no template controls, controls for gDNA contamination (− RT) and standards and two technical replicates per sample and gene (protocol and reference genes according to Roth (2016a, 2016b)).
For each of the two technical replicates per sample, the mean cycle time (Ct), the standard deviation (SD) and the coefficient of variance (CV) were calculated. If CV was larger than 4%, samples were removed due to potential measurement errors (Bookout and Mangelsdorf 2003). The housekeeping genes ubiquitin (Ubi) and ribosome protein (Ribop) showed the highest stability (geNorm M > 0.85) (Hellemans et al. 2007). Their geometric mean was thus used to quantify relative expression of each target gene by calculating ∆Ct values. A total of 27 animals had to be excluded from further statistical analyses due to measurement errors in gene expression profiles.

| Data analysis and statistics
This study aimed to evaluate how the combination of parental immune challenge and temperature change affected duration of male pregnancy and clutch size. Upon offspring exposure to both parental temperature treatment and Vibrio immune challenge (split design within each family) expression of immune genes, genes mediating epigenetic signalling (DNA modification and histone modification) and impact on offspring body size, an important life-history trait, were evaluated. This permitted the study of usage of similar or distinct physiological mechanism when two environmental stressors were applied both during the parental and the offspring generation. Doing so, we could assess the potentially interactive effect of TGP according to two environmental alterations, and its adaptive characteristics in case of matching or nonmatching parental and offspring environmental conditions.
A permutational multivariate analysis of variance (PERMANOVA) was applied for immune gene expression (29 target genes) as well as epigenetic regulation genes (15 target genes) (320 samples).
The PERMANOVA model ("vegan" package-"adonis" function in R) was based on a Bray-Curtis matrix of nontransformed ∆Ct values with 1,000 permutations per model. Our data fulfilled the requirements for multivariate homogeneity of group dispersion, which was tested with betadisper ("vegan" package). We applied tparent (temperature parents), Vparent (Vibrio exposure parents), toffspring (temperature offspring), Voffspring (Vibrio exposure offspring) as fixed factors and included "family" as random effect (strata = family). The PERMANOVA that included all genes as response variables was followed by PERMANOVAs for the following functional gene groups: (i) adaptive immune response, (ii) innate immune response, (iii) complement system, (iv) methylation/demethylation and (v) acetylation/ deacetylation) (all genes in Table 1).
Statistical univariate approaches were applied for body size and as post hoc tests for factors showing a significant effect in the PERMANOVAs to evaluate the contribution of single genes. Only factors with a significant effect in the PERMANOVAs were considered. A linear mixed effect model ("nmle" package-"lmer" function in R) was fitted using tparent (temperature parents), Vparent (Vibrio exposure parents), toffspring (temperature offspring), Voffspring (Vibrio exposure offspring) as fixed factors and "family" as random effect. Prior to the analysis, response variables (gene expression, size) were box-coxtransformed, and data and residuals were tested for normal distribution and variance homogeneity (Shapiro-Wilk test, Levene's test). Post hoc Student's t tests to examine interactions in more detail followed significant ANOVAs.
The impact of the parental immunological treatment (Vparent) and parental temperature treatment (tparent) on duration of pregnancy and clutch size was analysed in an ANOVA with Vparent and tparent as fixed factors.
Heatmaps were depicted for graphical visualization of gene expression (Figures 1-7). For normalization (-∆∆C t ), the ∆Ct value of T A B L E 1 All genes assessed and discussed in this manuscript, grouped according to their functional categories. Gene names and functions are given for each gene. In the references, the according primers and accession numbers on NCBI can be found.

| Multivariate analysis-all genes
Both parental temperature and bacterial experience, but also offspring bacterial exposure (tparent, Vparent and Voffspring), significantly changed gene expression in offspring (PERMANOVA over all genes).
In addition, the interaction of the two parental environmental factors,  which was also found for genes involved in the complement system.

| Multivariate analysis-functional gene groups
Expression of genes of the adaptive immune system was affected by a combination of both the parental temperature experience and the bacterial experience, and this effect in turn depended on offspring environmental temperature (tparent × Vparent × toffspring). This three-way interaction effect was also identified for genes mediating acetylation/deacetylation. The temperature offspring were exposed to (toffspring), only influenced expression of adaptive immune genes (Table 2).

| Univariate analyses-single genes
For the significant main effects and interactions identified in the PERMANOVA per functional gene group, statistical univariate models (mixed linear effect models) were calculated, followed by post hoc analyses (Student's t test) (Tables S1-S5). For ten genes, the parental temperature experience played a significant role (tparent). Of those ten genes, three are involved in innate immune defence (LectpI, Cf, Kin), three in adaptive immune defence (HIVEP2, HIVEP3, Bcell.rap31) and one in the complement system (C1), ASH mediates histone methylation, and HDAC3 and BROMO are involved in histone deacetylation and acetylation. For eight of these ten genes, expression was lower if parents were exposed to an elevated temperature. Only Cf and Bcell. rap31 showed the opposite pattern ( Figure 1).
For six genes, parental Vibrio injection resulted in a higher expression: Kin, Tyroprot (innate immune system), Lymphag75 und Bcell.rap31    showed lowest expression of No66 if they were also kept at cold temperature (NCC; Figure 7).
In summary, there was only one case for adaptive TGP supported  None of the factors assesses had an impact on offspring size that was measured 8 days after birth (linear mixed effect model). This suggests that during the shorter pregnancy at hot temperatures, development must have been accelerated and that the temperature regime juveniles were exposed to after birth did not influence size (Table S6).
Because the resources allocated to plasticity are limited, the adaptive capacity for response to multiple environmental factors may also be constrained (Bubliy, Kristensen, Kellermann, & Loeschcke, 2011).
Analogous to negative trait correlations (Etterson & Shaw, 2001;Stearns & Magwene, 2003), a full adjustment to two parental exposures may thus be impossible in the offspring. In this study, heat stress dominated TGP and limited the ability of pipefish offspring to respond to parental immunological challenge.

Methylation (8)
Acetylation (  (lysine-specific-histone demethylase, gene silencing) and TPR (lysinespecific demethylase, gene activation) and HDAC6 (histone deacetylase, gene silencing) (all four associated with DNA modification and histone modification). The expression of eleven of these twelve genes was highest if parents were exposed to an immunological activation with Vibrio bacteria at ambient temperatures (parental VC treatment), implying a boost of offspring immune defence upon parental Vibrio challenge (Beemelmanns & Roth, 2016a, 2016bRoth, Klein, et al., 2012 Immune genes LectpI (lectin protein type I, pathogen recognition receptor), Kin (kinesin, intracellular transport) from the innate immune system, HIVEP 2 and HIVEP3 from the adaptive immune system and C1 (recognition subcomponent, formation of the antigen-antibody complex) from the complement system were all downregulated in offspring if parents were exposed to elevated temperatures ( Figure 1).
This suggests a less efficient immune defence when parents experienced a rise of temperature, which goes in line with our finding of decreased pipefish immune response at warmer temperatures (Landis, Kalbe, et al., 2012). In addition, also ASH, HDAC3 and BROMO involved in histone modification were downregulated under these conditions, suggesting a rearrangement (remodelling) of gene expression via histone modification. No effects on DNA methylation were identified. Only Cf (coagulation factor II, blood clotting and inflammation) and Bell.rap31 (B-cell activation) showed the opposite pattern. This main effect of parental temperature is, however, also influenced by the above-discussed impact of immune challenge that was only effective at ambient temperatures as indicated by the significant interaction of the two parental effects (Vparent × tparent). Within the limited number of genes addressed in this study, we cannot find support for the hypothesized segregated resource pools. Nevertheless, it is very likely that parental exposure to elevated temperature rather induced a set of genes related to heat stress. With one exception (Hsp60), we did not include heat stress genes in our study. Differential gene expression analyses upon full transcriptome comparison should be much more indicative to answering this hypothesis in a future study.
Juveniles kept at hot temperatures induced the expression of genes of the adaptive immune system. If this constitutive upregulation of gene expression at increased temperatures independent of the parental treatment could also be found under natural conditions, it may imply an adaptive reaction towards more virulent pathogens and spreading infectious disease in warmer waters (Harvell et al. 2002). Alternatively, the significant interaction among parental temperature exposure, parental immune challenge and the temperature offspring were kept at (tparent × Vparent × toffspring) which may rather imply that TGP reaches its limits once it needs to be attributed to two changing parental environmental factors (Bubliy et al., 2011). This effect was consistently identified for immune genes and DNA-and histone-  Recent whole transcriptome gene expression (RNAseq) approaches identified several immune genes involved in thermal TGP (Shama et al., 2016;Veilleux et al., 2015). Immune genes thus seem to reflect the processes of both thermal trans-generational acclimatization and TGIP. The immune genes investigated in our study are known to correlate with cellular immune defence (Beemelmanns & Roth, 2016b;Birrer et al., 2012), implying their role in host physiology, which adds to the aim to understand the molecular basis of TGP. Epigenetic marks were recently claimed to mediate TGP (Munday, 2014). While studies already confirmed that DNA-and histone-modification genes are influenced by TGIP in pipefish (Beemelmanns & Roth, 2016a, 2016b, 2017, epigenetic regulation also mediates thermal trans-generational adjustments. ASH, HDAC3 and BROMO (acetylation) were downregulated in offspring upon parental heat stress. The downregulation of the histone methyltransferase ASH could imply that parental temperature challenge negatively influences embryonic development, yet development was faster at elevated temperature with no negative effect on offspring size.
The lower expression of the histone deacetylase HDAC3 will enhance deacetylation of lysine residues and induce transcription, while downregulation of the histone acetyltransferase BROMO will result in negative regulation of transcription over chromatin structure rearrangement.
Benefits of TGP are always context dependent (Marshall, 2008), and TGP can be maladaptive (Schade, Clemmesen, & Wegner, 2014) even if parental and offspring conditions match. While the parental immune challenge induced offspring gene expression, elevated temperature in the parental generation had a smaller impact on the offspring gene expression profiles. The combination of the two parental effects revealed the same pattern as the sole application of a temperature change in the parental generation. We thus identified a dominant parental temperature effect, as the offspring gene expression upon an elevated parental temperature exposure remained. Independent of the applied parental immune challenge, temperature was the master regulator of phenotypic plasticity. Our data suggest that the potential of trans-generational effects to compensate stressful environmental conditions during offspring maturation is hampered when multiple environmental stressors are applied simultaneously in the parental and offspring generation, potentially because the capacity for TGP is limited. This sheds new light on how animals can cope with changing environmental conditions in nature that usually impact several abiotic and biotic factors simultaneously, and may raise the question whether phenotypic plasticity remains an effective short-term response that permits acclimation to global change.