SEARCH

SEARCH BY CITATION

Keywords:

  • host–parasite interaction;
  • immune defence;
  • insect;
  • parental effects;
  • trans-generational

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors’ contributions
  8. Acknowledgements
  9. References

1. Parasitized females in mammals, fish and birds can enhance the immune defence of their offspring by transferring specific antibodies for the embryo. Likewise, social insect mothers transfer immunity despite the fact that invertebrates lack antibodies.

2. Female trans-generational immune priming is consistent with parental investment theory, because mothers invest more into rearing their offspring than fathers. However, when immune priming is not directly linked to parental care, as is often the case in insects that abandon their eggs after oviposition, both sexes might benefit from protecting their offspring.

3. Using the red flour beetle, Tribolium castaneum, we show that after parental exposure to heat-killed bacteria, trans-generational immune priming occurs through fathers as well as mothers.

4. This novel finding challenges the traditional view that males provide only genes to their offspring in species without paternal care, and raises the possibility of a division of tasks with respect to immune protection between parents.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors’ contributions
  8. Acknowledgements
  9. References

Genomic imprinting (Haig 2000) and maternal effects (Bernardo 1996) can transfer information from the environment experienced by the parents and so adaptively change an offspring’s phenotype (Mousseau & Fox 1998; Bateson et al. 2004) without alteration of the genomic sequence. As selection is often most severe during early development and the juvenile phase (Rossiter 1996) such parental effects might have important evolutionary consequences (Grindstaff, Brodie & Ketterson 2003). In turn, parasites are probably the most ubiquitous selection factor (Hamilton, Axelrod & Tanese 1990). Therefore, if parasites are prevalent in the parental environment and offspring are likely to experience the same conditions, parents gain by protecting their offspring against the same or similar infections through immune priming, i.e. by stimulating the expression of immune responses. The biology of social insects defines such a situation and immune priming has indeed been found in such systems (Moret & Schmid-Hempel 2001; Sadd & Schmid-Hempel 2006, 2007). However, immune priming is also found in solitary but phylopatric species where offspring are likely to encounter a similar environment, such as in flour beetles inhabiting a storage of cereals (Roth et al. 2009) and in the woodlouse Porcellio scaber (Roth & Kurtz 2009).

Thus far, studies on trans-generational immune priming have focused on a transfer via the mother. Using the red flour beetle, Tribolium castaneum and two bacterial species, Bacillus thuringiensis and Escherichia coli, as a model system, we investigated the effect of trans-generational immune priming by mothers vs. fathers. Moreover, we explored the specificity of priming by comparing offspring survival upon a challenge in homologous (parents and offspring were exposed to the same bacterium) or heterologous (parents were exposed to a different bacterial species than their offspring) combinations. To avoid vertical transmission, we used here heat-killed bacteria for priming but tested for the effects by administering a potentially lethal dose of live bacteria to the offspring.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors’ contributions
  8. Acknowledgements
  9. References

The model system

Tribolium castaneum is a well-established model system for the investigation of ecology, behaviour, genetics and immunology of host-parasite interactions. Tribolium castaneum has further developed into a fully sequenced model system in which embryonic development and pesticide resistance can be genetically analysed (Lorenzen et al. 2005).

Tribolium castaneum is known to naturally harbour a range of protozoan and other parasites (Sokoloff 1974). However, for our experiments we used two bacteria species, namely E. coli (DSM No. 498) and B. thuringiensis (DSM No. 2046), both obtained from the German Collection of Micro-organisms and Cell Cultures (DSMZ). Escherichia coli was used as a widely distributed bacterium. The use of E. coli may be comparable with commonly used immune stimulation with impurities of peptidoglycane in lipopolysaccharide (LPS) (Moret & Schmid-Hempel 2000; Moret 2006). By contrast, B. thuringiensis is an insect-specific pathogen, also known to harm T. castaneum (Abdel-Razek et al. 1999). Both bacteria were obtained from a micro-organism supplier for reasons of standardization and to provide a novel challenge, i.e. to avoid that beetles might have previously encountered these bacterial strains. For the experiment, we used a randomly chosen outbred T. castaneum line from our laboratory stock-cultures. This line was produced by crossing ten lines provided by Michael J. Wade, collected in Indiana/USA and then kept in a large outbreeding population under standard laboratory conditions (30 °C, 70% humidity).

The experiments

Two independent survival experiments were conducted (named 1 & 2 thereafter). Beetles experienced the following immune priming treatments: injection of E. coli (Ec), B. thuringiensis (Bt), insect saline solution (Ringer), or no challenge (naïve). The latter two groups served as controls. The following treatment groups with respect to sex were formed: (1) both sexes naïve, (2) female naïve, male Ringer, (3) female naïve, male Ec, (4) female naïve, male Bt, (5) male naïve, female Ringer, (6) male naïve, female Ec, (7) male naïve, female Bt. Offspring from beetles of these treatment groups were then again exposed to a challenge with either Ec, Bt, Ringer or they remained unchallenged (naïve) (Fig. 1). Offspring survival (in days post-challenge) was measured as a phenotypic outcome of the effect of immune priming. In experiment 1 further the activity of the key enzyme phenoloxidase was measured to directly determine immune system activity. In experiment 2, antimicrobial activity of the haemolymph of offspring as a token of immune system activity, offspring developmental time and their fecundity were measured, to explore whether parental priming bears any costs for offspring. Here, parental priming was full-factorial, such that also groups were included were both fathers and mothers received a priming, and where fathers received a different priming than mothers.

image

Figure 1.  Results of experiment 1. Survival of a bacterial challenge in Tribolium castaneum offspring after either maternal (a, c, e) or paternal (b, d, f) trans-generational immune priming with Bacillus thuringiensis or Escherichia coli. In panels a & b, data are pooled to show the comparison of offspring challenged with either the same (homologous) or a different bacterial species (heterologous) as their mothers or fathers were primed with. The other parts show data separated for the two bacteria challenges (panels c, d: B. thuringiensis; panels e, f: E. coli).

Download figure to PowerPoint

For the experiments, eggs of T. castaneum were individually distributed into 96-well plates, filled with flour. Animals were raised at 30 °C, 70% humidity in the dark. Individuals were checked regularly for their developmental stage. When the pupal stage was reached, animals were sexed and distributed individually into a fresh 96-well plate. Six weeks after the egg distribution all individuals had reached sexual maturity and the beetles were randomly assigned to one of the following immune priming treatments (n = 22 per sex and treatment): naïve, Ringer (which is an insect saline solution), Ec, Bt. The animals in the Ec – and the Bt – group were pricked between caput and thorax with a needle dipped in a sterile Ringer solution with 1010 bacteria mL−1 of heat-killed (30 min, 90 °C) bacteria obtained from an overnight culture (for details see Roth & Kurtz 2008). Animals in the sterile Ringer group were treated similarly, except that the Ringer solution contained no bacteria. 24 h after priming, the following breeding pairs were established (22 pairs per treatment; total n = 154 pairs): (1) both sexes naïve, (2) female naïve, male Ringer, (3) female naïve, male Ec, (4) female naïve, male Bt, (5) male naïve, female Ringer, (6) male naïve, female Ec, (7) male naïve, female Bt (Fig. 2).

image

Figure 2.  Graphical sketch of the experimental design. The parents were exposed to a priming with either heat-killed bacteria (Ec: Escherichia coli, Bt: Bacillus thuringiensis), to Ringer (R) or left naïve (N). Breeding pairs with seven different treatment combinations were formed and allowed to produce offspring. The offspring of every breeding pair was raised and after adult eclosion exposed to a possibly lethal dose of Ec or Bt, to control treatment Ringer or left naïve. After this, surviving animals were counted daily.

Download figure to PowerPoint

The breeding pairs were allowed to oviposit for 9 days, but offspring was sieved off every second day. Offspring were individually distributed to 96 well plates. 35 days later all offspring had grown to adults. For experiment 1, 10 families per priming treatment with at least eight offspring were randomly chosen and always two of the offspring were randomly assigned to one of four challenge treatments (naïve, Ringer, Ec, Bt). The animals were challenged as above. In total we challenged 10 families × 8 individuals × 7 priming treatments = 560 animals (for a total of 20 per treatment) in experiment 1. Thereafter, beetles were randomly and individually distributed to 96 well plates filled with flour. After the challenge survival of the animals was checked daily for 7 days, and again after 9 and 15 days. For experiment 2, 8 families per priming treatment with at least 16 offspring were randomly chosen and always four of the offspring were randomly assigned to one of the four challenge treatments (naïve, Ringer, Ec, Bt). In total we challenged 8 families × 16 individuals × 7 priming treatments = 896 animals (for a total of 32 per treatment) in experiment 2.

The immune assays

Phenoloxidase activity

In the offspring generation of experiment 1, 10 beetles were sampled at random from each treatment, 24 h after the immune challenge. The haemolymph was collected by puncturing the pleural membrane between pronotum and occiput with a sterile hypodermic needle. The outflowing droplet of hemolymph was collected in a sterile, pre-chilled glass capillary. For each insect, 0·05 μL of hemolymph was collected and flushed into a 96 well plate containing 20 μL Bis–Tris buffer (0·1 M, pH 7·5) and stored at −80 °C. To determine phenoloxidase (PO) activity, 50 μL of Aqua dest and 50 μL Bis–Tris were given in a 96-well plate (flat bottom) with 20 μL of the hemolymph in Bis–Tris. After adding 50 μL of L-Dopa (4 mg mL−1 L-Dopa dissolved in Bis–Tris), absorbance was measured on a Tecan Infinite M200 plate reader at 490 nm at 37 °C for 90 min, once every minute (kinetics). Phenoloxidase activity was determined as the fastest change in absorbance over 15 minutes (V max).

Antimicrobial activity

Escherichia coli and B. thuringiensis were grown in medium (per 1000 mL of water: 10 g Bacto-tryptone, 5 g yeast etract, 10 g NaCl, pH 7) over night (33 °C). The next morning the bacteria concentration was determined. For the preparation of the plates, medium (as described above) with 1% agar was autoclaved and cooled down in a water bath to 45 °C. Either E. coli (concentration: 1 × 105 cells mL−1) or B. thuringiensis (concentration: 4 × 105 cells mL−1) were added. 6 mL of the medium was put into a Petri dish, swirled immediately to provide an even distribution. Full body homogenates of the beetles (frozen 24 h after challenge in 30 μL Ringer supplemented with Phenylthiourea acid (PTU) to inhibit melanization) were prepared. After centrifugation 2 μL of the supernatant was filled into pre-punctured holes (diameter: 2·00 mm) on the agar plates, every sample was used on E. coli and on B. thuringiensis agar plates to determine antimicrobial activity against the respective bacterium. As positive control on every plate Streptomycin was added, as a negative control Ringer was used. The plates were incubated at 30 °C for 12 h (E. coli) or for 18 h (B. thuringiensis). Inhibition zone were measured horizontally and vertically. The mean value of both measurements was calculated, to provide a more reliable result. For measuring, the software Optimas® was used.

Statistics

All statistical analyses were performed in JMP 6 (SAS Institute Inc., US).

In order to test for maternal and paternal immune priming two separate analyses of experiment 1 were performed, using either a three-way Proportional Hazard survival analysis or a generalized linear model (which addresses survival to the end of the experiment as the response variable). In both cases, priming (B. thuringiensis, E. coli, control), challenge (B. thuringiensis, E. coli, control) and the sex of the parent (maternal or paternal) were taken as fixed factors. The response variable was survival in days post-challenge for Proportional Hazard Analysis, with animals still alive after 15 days taken as censored values, or the state of the animals (dead or alive) at the last day of the experiment (day 15) with the generalized linear model using a binomial error distribution.

Experiment 2 was designed to support the paternal immune priming result of the first experiment. In experiment 2, the survival after challenge was much higher; the dose used was not as lethal as in the first experiment perhaps due to adaptation by the beetle populations in the meantime. Thus, we did not have enough power to find all interactions as in the previous experiment. To simplify the analysis for the central question of parental immune priming in the survival analysis, we therefore excluded animals that were not exposed to a challenge by bacteria from the analysis, and the two kinds of bacteria were combined into one general group of bacterial challenge. Besides, the same response variables and fixed factors as in experiment 1 were used.

Phenoloxidase measurements were Box-Cox-transformed (with PO = (PO0·2−1)/2·354) to reach normal distribution. Differences in phenoloxidase activity were thereafter analysed for every parental sex in a two-way anova taking priming and challenge as fixed factors. Effects of parental priming on antimicrobial activity were analysed in a two-way anova (fixed factors: maternal and paternal).

Developmental time until adult eclosion was analysed in a two-way anova with maternal and paternal priming as fixed factors. Reproduction of animals whose parents either had a maternal, a paternal priming or both, was analysed in a two-way anova (maternal and paternal priming as fixed factors).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors’ contributions
  8. Acknowledgements
  9. References

Analysing the results with pair-wise comparisons using a proportional hazard test, we found no difference in offspring survival when comparing the two control treatments (Ringer vs. naïve) with each other, neither for parental priming controls, nor for offspring challenge controls in both experiments (χ2 tests, all P ≥ 0·5).

For this reason, we pooled Ringer and naïve treatments as controls for further analyses.

Experiment 1 – survival

Both models indicate that priming (parental exposure to bacteria vs. control) and challenge (offspring exposure to bacteria vs. control) had a significant effect on the survival of our animals (Table 1). Whereas a bacterial challenge decreased the survival, parental priming by bacteria was advantageous for the survival of the offspring (Fig. 1). The significant ‘priming × challenge’ interaction further suggests that the effect of trans-generational immune priming was dependent on the combination of bacteria used for priming (B. thurinigiensis or E. coli) and challenge. Offspring that were exposed to the same bacteria as their parents survived longer, indicating that parental immune priming is specific for the bacterial species. The absence of a significant sex main effect points out that trans-generational immune priming was consistent for maternal and paternal priming.

Table 1.   Statistics of experiment 1. Effects of (fixed) factors sex (mother, father), parental priming (Bacillus thuringiensis, Escherichia coli, control) and challenge (B. thuringiensis, E coli, control) on post-challenge offspring survival in days (a, proportional hazard test) and the chances of being alive at the end of the experiment (b, generalized linear model). DF = degrees of freedom, *means P < 0.05
a) Proportional hazard test
FactorDFChi-squareP
Sex1 1·2090·271
Parental priming2 8·3230·016*
Sex × parental priming2 0·3410·843
Offspring challenge225·31<0·0001*
Sex × offspring challenge2 1·9040·386
Parental priming × offspring challenge411·900·018*
Par. priming × offspr. challenge × sex4 6·3960·172
b) Generalized linear modelDistribution: binomial, link: logit
FactorDFChi-squareP
Sex10·8300·363
Parental priming26·3060·043*
Sex × parental priming20·7050·703
Offspring challenge224·79<0·0001*
Sex × offspring challenge22·3490·309
Parental priming × offspring challenge412·150·016*
Par. priming × offspr. challenge × sex47·9160·095

Experiment 2 – survival

The second survival experiment supported the conclusion of paternal trans-generational immune priming when the environment poses a risk of bacterial infections. The results of this second experiment confirmed the finding that priming of parents with heat-killed bacteria significantly improved survival of the offspring after a challenge with live bacteria. This effect was again consistent for maternal and paternal priming suggesting that priming occurs through both sexes (Table 2) (Fig. 3).

Table 2.   Statistics of experiment 2. Effects of (fixed) factors sex (mother, father) and parental priming (bacteria, control) on post-challenge offspring survival in days (a, proportional hazard test), and the chances of being alive at the end of the experiment (b, generalized linear model). *Means P < 0.05
a) Proportional hazard test
FactorDFChi-squareP
Parental sex10·8480·357
Parental priming15·4840·019*
Sex × parental priming10·0520·819
b) Generalized linear modelDistribution: binomial, link: logit
FactorDFChi-squareP
Parental sex10·8110·368
Parental priming16·9100·009*
Sex × parental priming10·1570·692
image

Figure 3.  Results of experiment 2. Survival of a bacterial challenge in Tribolium castaneum offspring after either maternal (a, c, e) or paternal (b, d, f) trans-generational immune priming with Bacillus thuringiensis or Escherichia coli. The data are shown here in the same way as in Fig. 1 (see legend).

Download figure to PowerPoint

Immune assays

Our data indicate that, on the one hand, immune priming via the father may lead to up-regulated phenoloxidase activity (Fig. 4, Table 3). On the other hand, paternal immune priming with E. coli and Ringer decreased the antibacterial activity of the offspring compared to paternal priming with B. thuringiensis or no priming. Maternal priming produced a similar difference, but seems to be a minor effect (Fig. 5, Table 4a and b).

image

Figure 4.  Phenoloxidase activity measurements of hemolymph samples taken 24 h after the offspring’s immune challenge in experiment 1. The four panels show PO activity for animals after a Bacillus thuringiensis challenge (a), after an Escherichia coli challenge (b), without a challenge (c) or after a Ringer challenge (i.e. wounding) (d).

Download figure to PowerPoint

Table 3.   Two-way anova testing for differences in the phenoloxidase (PO) – activity of offspring from parents that had received a immune priming (Bt, Ec, naïve or Ringer; c.f.Fig. 2). Two separate analyses were performed for maternal (a) and paternal (b) immune priming, since PO activity in animals with maternal immune priming was lower compared to paternal immune priming. Before analysis data were Box-Cox-transformed (with PO = (PO0·2-1)/2·354) to reach normal distribution. PO seems involved in rather unspecific paternal immune priming, whilst we expect a different, more specific but as yet unknown mechanism to be responsible for maternal immune priming
(a) Maternal immune priming
anovaDF
Model15
Error165
Total180
Effect testsDFSum of SqF ratioP
Priming30·02073·82250·0111
Challenge30·0254·62580·0039
Priming × challenge90·0432·6490·0068
Post hoc tests (Tukey HSD)Least SqDifferent
Priming
 Ec−0·168A
 Naïve−0·1834AB
 Ringer−0·1886AB
 Bt−0·1982B
Challenge
 Bt−0·1734A
 Ec−0·1756A
 Ringer−0·1832AB
 Naïve−0·2056B
Priming × challenge
 Ec, Bt−0·1327A
 Naïve, Bt−0·158AB
 Ringer, Ringer−0·161AB
 Ec, Ec−0·1663AB
 Naive, R−0·1722AB
 Ringer, Ec−0·1737AB
 Naive, Ec−0·1799AB
 Bt, Ec−0·1826AB
 Ec, naïve−0·1851AB
 Ec, Ringer−0·1868AB
 Bt, naïve−0·1957AB
 Bt, Bt−0·1996B
 Ringer, Bt−0·2019B
 Bt, Ringer−0·2129B
 Ringer, naïve−0·2179B
 Naïve, naïve−0·2218B
(b) Paternal immune priming
anovaDF
Model15
Error245
Total260
Effect testsDFSum of SqF ratioP
Priming30·029675·01270·0022
Challenge30·080513·605< 0·0001
Priming × challenge90·01550·8750·5482
Post hoc tests (Tukey HSD)Least SqDifferent
Priming
 Ec−0·1493A
 Naïve−0·1602A
 Ringer−0·1627AB
 Bt−0·1834B
Challenge
 Bt−0·1457A
 Ec−0·1521A
 Ringer−0·164A
 Naïve−0·1938B
image

Figure 5.  Antimicrobial activity of offspring in relation to parental immune priming. Measurements were obtained from experiment 2 using zone of inhibition assays on agar plates with the respective bacteria (concentration: Ec 4 × 105 mL−1, Bt 1 × 105 mL−1; grown over night for 18 h at 30 °C). The two bars show antimicrobial activity against Bt or Ec. Either the father (paternal), the mother (maternal) or both were primed fully reciprocally with Bacillus thuringiensis (Bt), Escherichia coli (Ec) Ringer (R) or left naïve (N). Bars show mean + standard errors.

Download figure to PowerPoint

Table 4.   A two-way anova (fixed factors: maternal and paternal) testing for differences in antimicrobial activity in animals with either paternal or maternal priming. Tables represent either antimicrobial activity against Bt (a) or against Ec (b). Post hoc tests were performed with a Students t-test. *Means P < 0.05
a) Antimicrobial activity against Bt
anovaDF
Model15
Error461
Total476
Effect testsDFSum of SqF ratioP
Maternal3131·63·0540·028*
Paternal3252·55·8580·0006*
Maternal × paternal910097·807< 0·0001*
Post hoc tests (Students t-test)Least SqDifferent
Maternal
 Naïve17·656A
 Bt16·942AB
 Ec16·580B
 Ringer16·160B
Paternal
 Naïve17·594A
 Bt17·550A
 Ringer16·109B
 Ec16·088B
Maternal × paternal
 Ec, Bt19·411A
 Bt, Bt18·854A
 Ringer, naive18·734A
 Naïve, Ec18·693AB
 Naïve, naïve17·752ABC
 Naïve, Bt17·744ABC
 Ec, naïve17·520ABCD
 Bt, Ec16·725BCDE
 Ec, Ringer16·718BCDE
 Naïve, Ringer16·437CDE
 Bt, naïve16·369CDE
 Ringer, Ec16·264CDE
 Bt, Ringer15·822DEF
 Ringer, Ringer15·458EF
 Ringer, Bt14·184FG
 Ec, Ec12·671G
b) Antimicrobial activity against Ec
anovaDF
Model15
Error446
Total461
Effect testsDFSum of SqF ratioP
Maternal37·6723·2350·022*
Paternal3114·3394·8220·0026*
Maternal × paternal918·9972·6700·0050*
Post hoc tests (students t-test)Least SqDifferent
Maternal
 Naïve5·970A
 Bt5·956A
 Ec5·950A
 Ringer5·661B
Paternal
 Naïve6·115A
 Bt5·947AB
 Ringer5·765BC
 Ec5·711C
Maternal × paternal
 Ringer; naive6·306A
 Naïve, naïve6·254AB
 Bt, Bt6·247AB
 Naïve, Ec6·184ABC
 Bt, naïve6·042ABCD
 Bt, Ringer5·998ABCD
 Ec, Bt5·928ABCDE
 Naïve, Bt5·924ABCDE
 Ringer, Ec5·904ABCDE
 Ringer, Ringer5·899ABCDE
 Ec, naïve5·856BCDE
 Ringer, Bt5·689CDE
 Ec, Ringer5·642DEF
 Bt, Ec5·537EF
 Naïve, Ringer5·519EF
 Ec, Ec5·218F

Costs

We here further investigated the costs of parental immune priming for the offspring by measuring developmental time and fecundity of beetles whose mothers and/or fathers were primed (with all combinations of B. thuringiensis, E. coli, Ringer, naïve). Maternal priming with E. coli led to a slightly longer developmental time (Table 5). Paternal priming significantly reduced offspring fecundity (Table 5, Fig. 6).

Table 5.   Developmental time until adult eclosion was analysed in a two-way anova with maternal and paternal priming as fixed factors. Maternal and paternal priming was performed in a fully reciprocal design with either Bacillus thuringiensis (Bt), Escherichia coli (Ec), Ringer (R) or left naïve (N). Data were Box-Cox-transformed (log (developmental time) × 30·309) to reach normal distribution. Post hoc tests for significant main effects were performed with Tukey HSD. *Means P < 0.05
Developmental time
anovaDF
Model15
Error369
Total384
Effect testsDFSum of SqF ratioP
Maternal344·1753·7760·0181*
Paternal329·5992·5300·0570
Maternal × paternal9108·1943·0830·0014*
Post hoc tests (Tukey HSD)Least SqDifferent
Maternal
 Ec103·845A
 Bt103·564AB
 R103·119AB
 Naïve102·991B
Maternal × paternal
 Ec, Ec104·749A
 Ec, Ringer104·715A
 Bt, Ec104·074AB
 Ringer, Bt103·686AB
 Bt, Bt103·672AB
 Naïve, naïve103·538AB
 Ec, naïve103·590AB
 Naïve, Ec103·324AB
 Bt, naïve103·281AB
 Bt, Ringer103·227AB
 Ringer, Ec103·179AB
 Ringer, Ringer103·016AB
 Ringer, naïve102·593B
 Naïve, Bt102·578B
 Naïve, Ringer102·423B
 Ec, Bt102·329B
image

Figure 6.  Reproduction of animals whose fathers, mothers or both were immune primed with either Bacillus thuringiensis (Bt), Escherichia coli (Ec) Ringer (R) or were left naïve (N) (experiment 2). Reproduction was measured by forming breeding pairs within each treatment and measuring reproduction over 2 weeks.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors’ contributions
  8. Acknowledgements
  9. References

Our study shows that when challenged with heat-killed bacteria both, mothers and fathers transfer a corresponding resistance to their offspring. Note that selection on parental populations cannot explain this result, since survival of parents is not affected by injection of heat-killed bacteria as used here (Roth & Kurtz 2008; Roth et al. 2009), and no difference in parental fecundity was observed (data not shown). Our study provides the first evidence for paternal trans-generational immune priming. Previous studies in both vertebrates and invertebrates have either focused on maternal trans-generational effects or have not tried to disentangle maternal from paternal effects (e.g. Moret 2006, Linder & Promislow 2009). A recent study (Freitak, Heckel & Vogel 2009) found strong effects of maternal diet on offspring immunity, but no significant paternal effects. Kurtz & Sauer 1999 found stronger paternal than maternal heritability of phagocytic capacity but did not address potential effects of priming. Our finding of paternal immune priming has important evolutionary and ecological consequences. If trans-generational immune priming takes place via both parents as observed here, information about pathogens in the environment of both mothers and fathers can be transferred to the offspring. Moreover, when different kinds of immunity are transferred, the protection offspring receives from mothers and fathers may even be more than additive. This could either result in a general better protection against parasites due to up-regulation of immune responses or in improved phenotypic plasticity in the immune response of the offspring, leading to a more specific and stronger reaction upon secondary exposure.

With respect to the underlying mechanism of this phenomenon, we can only speculate. The observed trans-generational paternal immune priming cannot be based on the transfer of antibodies as is known from vertebrates. Instead, protection may be realized through a programmed up-regulation of specific immune defences in the offspring, or alternatively, through enhanced tolerance against the respective bacterial species (Kurtz & Franz 2003; Raberg, Sim & Read 2007; Roth et al., 2009; Sadd & Schmid-Hempel 2006). Sadd & Schmid-Hempel (2006) could demonstrate that female bumblebees transfer immune protection via a factor residing inside the eggs. Here observed paternal immune priming might either relate to priming of the genetic information (e.g. imprinting through CpG methylation) or transfer of epigenetic factors such as regulatory RNA (RNAi, miRNAs) that might affect the expression of immune genes (Rassoulzadegan et al. 2006; Slotkin & Martienssen 2007).

Both immune traits measured seem to be affected by parental immune priming. Whereas maternal immune priming had no effect on the rather unspecific phenoloxidase activity (Söderhall & Cerenius 1998), it was up-regulated by paternal immune priming with bacteria, but also by Ringer pricking (Fig. 4, Table 3). This indicates that the transfer of induced phenoloxidase activity serves as a general signal of wounding and infection. Parental immune priming with E. coli and Ringer decreased the antibacterial activity of the offspring compared to parental priming with the gram-positive bacterium B. thuringiensis or no priming. However, these effects were stronger in paternal immune priming and seem to be of minor effect in maternal priming (Fig. 5, Table 4a and b). This suggests the Toll-pathway that is activated via the recognition of gram-positive bacteria to play a different role than the Imd-pathway, activated by gram-negative bacteria. Further, the fact that the non-pathogenic bacterium E. coli led to the same level of antimicrobial activity as the wounding control agent Ringer could account for a stronger selection for paternal immune priming from natural insect pathogens like B. thuringiensis.

When the risk of parasitism fluctuates over generations, trans-generational immune priming comes at an additional cost (von Schantz et al. 1999) because selection acts on both, parents and offspring (Kirkpatrick & Lande 1989; Mousseau & Fox 1998). These selection episodes may frequently oppose each other (Kirkpatrick & Lande 1989; Wolf et al. 1998) and result in the ‘wrong’ priming, which entails a cost that can either be paid by the parents, the offspring, or by both (Gallizi, Guenon & Richner 2008). Furthermore, parents face a resource allocation trade-off as the transfer of substances to the eggs or to the sperm is energetically costly (Grindstaff et al. 2003). Offspring, in turn, mount a costly immune response as the parental products stimulate its expression (Carlier & Truyens 1995). Sadd & Schmid-Hempel (2009) recently detected costs of maternal immune priming, if the parasite exposure of the offspring does not fit the priming of the mothers. Also Linder & Promislow (2009) found costs of maternal immune priming for offspring in Drosophila, but no benefits in terms of increased protection of offspring. We here considered costs of parental priming for the offspring by measuring developmental time and fecundity of beetles whose mothers and/or fathers were primed. Maternal priming with E. coli was associated with a slightly longer developmental time (Table 5). Paternal priming significantly reduced offspring fecundity (Table 6, Fig. 6). This suggests that costs of maternal immune priming were comparably relatively lower than the costs of paternal priming. This may be explained by the possible mechanisms behind maternal and paternal immune priming. If maternal immune priming based on transfer of substances via the eggs as has been shown in bumblebees (Sadd & Schmid-Hempel 2007), we could assume this investment to be costly for the female rather than the offspring. By contrast, paternal immune priming that is likely to rely on epigenetic changes would hence likely not induce high costs on the paternal side but mainly for the offspring which consequently have to invest into an up-regulated immune defence.

Table 6.   Reproduction of offspring after parental priming (mothers, fathers or both parents were primed), analysed with a two-way anova (maternal and paternal priming as fixed factors). Maternal and paternal priming was performed in a fully reciprocal design with either Bacillus thuringiensis (Bt), Escherichia coli (Ec), Ringer (R) or left naïve (N). Post hoc tests for significant main effects were performed with Tukey HSD. *Means P < 0.05
Reproduction
anovaDF
Model15
Error127
Total142
Effect testsDFSum of SqF ratioP
Maternal3609·010·5230·6670
Paternal38124·276·9820·0002*
Maternal × paternal95389·131·5440·1395
Post hoc tests (Tukey HSD)Least SqDifferent
Paternal
 Ec64·545A
 Naïve58·925A
 R52·458AB
 Bt44·306B

Our results challenge the notion that in species without parental care only mothers can invest into offspring, since they produce the egg, whereas fathers are typically limited to passing on their genes only, via sperm (Arnqvist & Rowe 2005). Whilst these limitations still hold, fathers also seem to help their offspring by somehow priming their sperm or seminal fluid. This kind of sperm-based help, perhaps via genetic imprinting or transfer of modifying epigenetic factors (Ashe & Whitelaw 2007), would also be more likely to evolve than, for example, helping young in a nest, since passing successful sperm conveys absolute certainty in paternity. Moreover, this raises the possibility that fathers and mothers share tasks in the kind of protection they supply to their offspring, or that they might transfer information that is specific to the parasitic environment they have encountered. Paternal immune priming is likely to have an impact on sexual selection, as female choice may depend on the parasitic experience fathers made as they provide immunity to the offspring. Paternal immune priming may even represent another explanation for the two-fold cost of sex (West, Lively & Read 1999). The parasitic experience of both parents transferred can possibly result in offspring better adapted to the local conditions; this would, especially in environmental conditions with high parasitic pressure, improve the fitness of sexually reproducing individuals significantly.

Authors’ contributions

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors’ contributions
  8. Acknowledgements
  9. References

O.R. and J.K. designed the experiments. Animals rearing and all experimental and laboratory work was performed by O.R., H.E., J.H. and J.D. O.R., G.J. and P.S.H. analysed the data. O.R., G.J., P.S.H., and J.K. wrote the paper.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors’ contributions
  8. Acknowledgements
  9. References

We would like to thank people from the Experimental Ecology Group at the ETH Zürich and people from the Institute for Evolution and Biodiversity at the University of Münster. We are grateful for support in the laboratory by Annette Bernhardt, Barbara Hasert and Joe Lange. We thank Sophie Armitage and Thorsten Reusch for comments on this manuscript and Jukka Jokela for statistical advice. This study was supported by a grant from the Swiss National Science Foundation (3100A0-112992 to J.K.) and ETH (nr. TH-09 60-1 to P.S.H). G.J. was supported by the German Academic Exchange Service (DAAD) and the Roche Research Foundation. O.R. was supported by the Volkswagen Stiftung.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors’ contributions
  8. Acknowledgements
  9. References