- Email: [email protected]
Cold temperature preference in bacterially infected Drosophila melanogaster improves survival but is remarkably suboptimal
Journal of Insect Physiology 93–94 (2016) 36–41
Contents lists available at ScienceDirect
Journal of Insect Physiology journal homepage: www.elsevier.com/locate/jinsphys
Cold temperature preference in bacterially infected Drosophila melanogaster improves survival but is remarkably suboptimal Kenneth M. Fedorka ⇑, Ian C. Kutch, Louisa Collins, Edward Musto University of Central Florida, Department of Biology, 4000 Central Florida Blvd., Orlando, FL 32816, United States
a r t i c l e
i n f o
Article history: Received 7 June 2016 Received in revised form 11 August 2016 Accepted 12 August 2016 Available online 13 August 2016 Keywords: Behavioral fever Behavioral chill Temperature preference Drosophila melanogaster Pseudomonas aeruginosa Parasite manipulation
a b s t r a c t Altering one’s temperature preference (e.g. behavioral fever or behavioral chill) is a common immune defense among ectotherms that is likely to be evolutionarily conserved. However, the temperature chosen by an infected host may not be optimal for pathogen defense, causing preference to be inefficient. Here we examined the efficiency of temperature preference in Drosophila melanogaster infected with an LD50 of the gram negative bacteria Pseudomonas aeruginosa. To this end, we estimated the host’s uninfected and infected temperature preferences as well as their optimal survival temperature. We found that flies decreased their preference from 26.3 °C to 25.2 °C when infected, and this preference was stable over 48 h. Furthermore, the decrease in temperature preference was associated with an increased chance of surviving the infection. Nevertheless, the infected temperature preference did not coincide with the optimum temperature for infection survival, which lies at or below 21.4 °C. These data suggest that the behavioral response to P. aeruginosa infection is considerably inefficient, and the mechanisms that may account for this pattern are discussed. Future studies of infected temperature preferences should document its efficiency, as this understudied aspect of behavioral immunity can provide important insight into preference evolution. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Fever in endotherms is a highly conserved, metabolicallygenerated increase in host body temperature that improves immune defense (Kluger 1978; Kluger et al. 1998). Ectotherms can also alter their body temperature adaptively in response to infection. However, this is accomplished by seeking out warmer (behavioral fever) or cooler (behavioral chill) temperatures, not by altering metabolic activity. For instance, invertebrate house flies (Musca domestica) and grasshoppers (Melanoplus sanguinipes) improve their chance of surviving fungal infections by relocating to warmer environments (Watson et al. 1993; Inglis et al. 1996). Infected vertebrate zebrafish (Danio rerio) likewise improve immune defense by seeking out warmer waters (Boltana et al. 2013). Such behavioral alterations to infection have been reported across a wide variety of ectothermic taxa and the phenomenon is likely to be evolutionarily conserved in animals (Kluger et al. 1998). Still, numerous studies have failed to detect changes in ⇑ Corresponding author at: University of Central Florida, Department of Biology, Bldg. 20, 4000 Central Florida Blvd., Orlando, FL 32816, United States. E-mail addresses: [email protected] (K.M. Fedorka), [email protected] (I.C. Kutch), [email protected] (L. Collins), [email protected] (E. Musto). http://dx.doi.org/10.1016/j.jinsphys.2016.08.005 0022-1910/Ó 2016 Elsevier Ltd. All rights reserved.
the thermal preferences of infected ectothermic hosts, even when fitness would be increased by such a response (Stahlschmidt and Adamo 2013), suggesting that the expression and/or evolution of the behavior is complex. Assuming that the host is capable of the behavior, there are several reasons why altered temperature preferences might not be elicited upon infection with a virulent pathogen. First, a change in temperature preference does not improve host fitness. This may occur if the host’s temperature-fitness function is flat for the pathogen in question, or if the host is already at its thermal optimum. Second, the cost of the behavioral response outweighs the benefit for a given host. This may occur when an individual host has little remaining residual reproductive value (e.g. older hosts or hosts infected with sterilizing pathogens), causing it to forgo an immune investment and instead invest in reproduction (e.g. terminal investment; Fedorka et al. 2013; Fedorka 2014). Third, there is a lack of coevolutionary history between the host and pathogen, which could result in the host not recognizing an appropriate Pathogen Associated Molecular Pattern (PAMP) to initiate the behavior. Fourth, the behavior is inhibited by the invading pathogen. Behavioral inhibition is plausible when one considers that pathogen manipulation of host behavior is a common occurrence (Moore 2002). Unfortunately, the causes underlying the lack of behavior are difficult to assess and are generally not explored.
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
Even if an infected host exhibits an adaptive shift in thermal preference, it does not mean the response is efficient. That is to say, the temperature chosen by a host when infected may not be the optimal temperature to maximize fitness. Inefficient temperature preference is an understudied aspect of ectothermic immune function that can provide important insights into the behavior’s evolution. One potential reason for an inefficient preference is the aforementioned lack of host-pathogen coevolutionary history. This can be caused by (i) weak selection for the optimal infected temperature preference due to infrequent host-pathogen interactions, or (ii) a lack of evolutionary time to allow selection to shape the preference, even if selection is relatively strong (Fig. 1A). A second potential reason for an inefficient preference is pathogen inhibition. In order to maximize its own fitness, a pathogen might manipulate the infected host’s temperature preference (ITP) that results in a perpetual tug-of-war that keeps the host away from its temperature optimum (Fig. 1B). It is important to note that a ‘‘weaker” response by a certain class of host (e.g. older hosts) due
37
to a higher behavioral cost does not represent an inefficient behavior, considering that the ‘‘weaker” response is still optimal for that class of host (Fig. 1C). It could be mistaken for inefficiency, however, if an experimental design suffered from sampling bias (e.g. not controlling for age, sex or reproductive history among treatments). Regrettably, most studies that document a host’s altered temperature preferences when infected do not estimate the optimal temperature when infected (e.g. Boorstein and Ewald 1987; Adamo 1998; Karban 1998; Moore and Freehling 2002), making the prevalence of inefficient responses unknown among animals. Here we examine the efficiency of temperature preference in Drosophila melanogaster infected with the gram negative bacterium Pseudomonas aeruginosa. We chose D. melanogaster as the host due to its prevalence as an invertebrate immunological model (Hoffmann 2003; Lemaitre and Hoffmann 2007) and P. aeruginosa as the pathogen because it is commonly used in invertebrate immunological studies (D’Argenio et al. 2001; Linder et al. 2008; Wittman and Fedorka 2015). Pseudomonas species exist naturally in wild populations of D. melanogaster, where oral-fecal transmission is the most likely route of infection, although infection via wounding may also occur (Corby-Harris et al. 2007; Chandler et al. 2011; Martins et al. 2013). Once infected, D. melanogaster rely on a variety of immunological responses against gram-negative bacteria, including ROS production, lysozymes, phenoloxidase cascade, and IMD-related antimicrobial peptides (Vallet-Gely et al. 2008). No previous D. melanogaster study has found evidence of behavioral fever/chill (Ballabeni et al. 1995; Arnold et al. 2015). Still, the elicitation of a behavioral response is likely to be pathogen specific (Adamo 1998), and a response against bacterial pathogens has yet to be examined (previous work focused on nematodes and viruses). To examine the existence and efficiency of such a response in D. melanogaster, we estimated the host’s uninfected (UTP) and infected (ITP) temperature preferences as well as their optimal body temperature when infected (ITO). If a host’s infected temperature preference (ITP) coincides with its infected temperature optimum (ITO), then the response would be deemed efficient. However, if ITP fails to reach the optimum, then the behavior would be considered inefficient, and one of the above mechanisms would be involved in shaping this important immunological response. 2. Methods 2.1. Stock maintenance
Fig. 1. The evolution and expression of behavioral fever. (A) When a lack of coevolutionary history exists between the host and pathogen, the infected host may not reach their infected temperature optimum (ITO), resulting in a suboptimal/ inefficient infected temperature preference (ITP). UTP represents the temperature preference of the uninfected host. (B) Pathogen manipulation may also result in a suboptimal ITP due to selective pressure on the pathogen to reach its optimal host body temperature (PO). (C) Multiple optima may exist for different groups of infected hosts within a species or population. Here, the fitness curves for two different hosts groups (A and B) are shown. For host B, a fully realized response may be too costly, and these hosts may exhibit a ‘‘weaker” infected temperature preference (ITP-B) due to a decreased infected temperature optimum (ITO-B) compared with host A’s preference and optimum (ITP-A and ITO-A). Although this response would not be considered inefficient, it could appear as such if the experimental design did not account for host sampling bias.
Flies in this study originated from 40 gravid females collected in 2010 from a single location in Orlando Florida and maintained as a large outbred stock (600 individuals per generation) to minimize the loss of genetic variation due to drift. Individuals were feed a cornmeal, yeast, sugar and agar food medium and kept in vertical incubators (Percival, Perry, IA, USA) at 25 °C with a 12 h:12 h light:dark photoperiod. Prior to all experiments, adult flies were collected upon eclosion and maintained in sex-specific vials at medium density (20 individuals per vial) to ensure virginity. All experimental flies were 5 ± 1 days old virgin females. 2.2. Experimental design To determine if D. melanogaster exhibits a shift in thermal preference when infected with P. aeruginosa, and whether this preference is efficient, we estimated three key parameters: UTP, ITP, ITO. These parameters were estimated using three separate treatments including an uninfected temperature-choice treatment (estimates UTP), an infected temperature-choice treatment (estimates ITP), and an infected no-choice temperature treatment
38
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
(estimates ITO). For the infected treatments, we inoculated flies with an LD50 of P. aeruginosa, while flies in the uninfected treatment were inoculated with the bacterial vehicle only (Luria broth – Fisher Scientific BP1426). We did not estimate the optimal host body temperature for P. aeruginosa transmission, but assumed that it resembled its optimal in vitro growth temperature of 37 °C (Linder et al. 2008). 2.3. Bacterial infections To create the LD50 concentration, bacteria were incubated in sterilized Luria broth at 37 °C until log phase. This solution was diluted with sterile broth to an absorbance of 0.3 at 490 nm using a microplate reader (Bio-Rad Model 680, Hercules, CA, USA). The new solution was further diluted to a concentration of 5 10 3, which corresponds to an LD50 for flies maintained at 25 °C and a 12:12 L:D photoperiod. The final bacterial solution was divided into 0.5 ml aliquots and stored at 4 °C. Prior to use, each aliquot was removed from refrigeration, vortexed and pipetted into a shallow microcentrifuge tube cap. Bacterial inoculation was achieved by dipping a thin needle (Fine Science Tools item # 26002-20) into the microcentrifuge cap and immediately piercing the thorax of the CO2 anesthetized fly just below and to the front of the wing attachment. This is a common method for infection (Lazzaro et al. 2004; Linder et al. 2008; Kutch et al. 2014) that provides reproducible results (Apidianakis and Rahme 2009), although variation in inoculation dose between flies does exist. Bacterial solution aliquots were used for no longer than 30 min before being replaced with a fresh solution. Between each infection, needles were dipped in 95% ethonol and allowed to air dry. Flies in the uninfected-choice treatment were inoculated with Luria broth only. 2.4. Temperature gradient To create the temperature gradient, we used an aluminum bar (91.4 30 2.5 cm; TECA, Chicago, IL, model #TGB-5030) atop of two cold/hot plates (TECA model #AHP-1200CPV). Each cold/hot plate has an operating range of 0 °C–50 °C, allowing a stable gradient of any length between these ranges to be produced. On top of the gradient were two types of assay apparatus, including a choice and a no-choice temperature apparatus (L W H = 68 8.5 4 cm each). Both types were constructed by imbedding 11 inverted, lidless petri dishes (D H = 60 15 mm; hereafter referred to as chambers) within spray foam insulation (Great StuffTM, Dow Chemicals), all contained within a wooden frame (Fig. 2A). The only difference between the two types of apparatus was that the choice apparatus had the portion of the walls removed where the chambers abutted, creating a long corridor that allowed flies to traverse the length of the gradient. On the top of each chamber, an 8 mm hole was created in which was placed a 1.5 ml microcentrifuge tube containing food media (the bottom of the tube was cutoff to allow access to the food). Along the top of each apparatus, we placed an LED strip connected to a dimmer switch to create a consistent light level in each chamber (RibbonFlex ProÒ, Armacost Lighting). This was done to minimize the confounding effects of phototaxis on parameter estimates. On the bottom of each apparatus was a removable layer of plastic wrap that minimized fly escapes between the apparatus and gradient bar. Each apparatus was then topped with a layer of polyurethane foam insulation. The layers of insulation were paramount in maintaining consistent air temperatures within each chamber, which was verified using button-sized temperature loggers (Embedded Data Systems model DS1922L; temperature range 40 °C–85 °C ± 0.5 °C; D H = 14 5 mm) adhered to the top of each chamber with a thin layer of modeling clay (Fig. 2B).
Fig. 2. Bottom-up view of the behavior and fitness assay apparatus. (A) Assay apparatuses were constructed out of petri dishes (i.e. chambers) surrounded by foam insulation and an LED light source. The temperature-choice apparatus had the chamber walls removed to create a movement corridor along the gradient. All chambers had access to food. (B) The air temperature of each chamber was accurate and consistent. Dashed line represents the programed temperature while the filled circles represent the actual air temperature means (standard errors are too small to see).
2.5. Estimation of parameters To estimate ITO, 110 flies were lightly CO2 anesthetized, inoculated with bacteria, and placed into a no-choice apparatus (10 flies per chamber). After the flies recovered from the anesthesia, the apparatus was placed on the gradient bar for 48 h, which ranged from 20 °C to 40 °C (Fig. 2B). At the end of hour 48, the apparatus was removed and the proportion of dead flies in each chamber counted. Previous work in our laboratory suggests that 25 °C flies that succumb to infection do so between hours 24 and 36. Previous work also suggests that flies alive at hour 48 show no increased mortality rates over the following seven days compared with controls (Kutch et al. 2014). Thus, flies that were alive at hour 48 were assumed to have fully recovered from infection. Estimation of ITO was replicated four times for a total of 440 flies across 11 different temperature regimes. In addition, we estimated the temperature tolerance of uninfected flies as a control. To this end, 330 flies where pricked with a needle dipped in sterile broth only and placed in the no-choice apparatus (10 flies per chamber) for 48 h, after which their survival was assayed. To estimate ITP and UTP, flies were lightly CO2 anesthetized and randomly assigned to either an infected-choice or an uninfectedchoice assay, respectively. Flies in the infected-choice assay were inoculated with bacteria, while flies in the uninfected-choice assay were pricked with a pin dipped in Luria broth only. Flies were
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
then placed into the central chamber of their respective temperature-choice apparatus (a total of 20 flies were released into each apparatus). Upon recovery from anesthesia, each apparatus was placed on the thermal gradient for 48 h. Using the information we gained from estimating ITO above, we modified the gradient range for this experiment to between 18 °C and 32 °C. At 12 h intervals, each apparatus was gently raised 1 m above the gradient for approximately 15 s while a second research photographed the underneath of the apparatus. This picture recorded the distribution of flies across the apparatus, which allowed ITP and UTP to be estimated at each time interval (we have no reason to believe this physically gentle and temporally brief disturbance had a longterm effect on fly distributions). ITP and UTP was calculated by assigning each fly to the temperature of the chamber in which it was found and averaging ‘‘fly temperatures” within treatments. Considering that the width of the gradient bar could accommodate three apparatuses, each replicate was composed of either one uninfected and 2 infected assays, or the converse. In each replicate, treatments were randomly assigned to one of the three choice apparatuses and to one of three positions on the gradient (top, middle or bottom). In total, 480 flies were assayed among 10 infected-choice replicates and 11 uninfected-choice replicates. The choice chambers were cleaned with 95% ethanol between each replicate run and allowed to dry.
39
3. Results There was a significant relationship between survival and the thermal environment for infected flies (Fig. 3A; chamber temperature: X210 = 113.7, P < 0.0001; replicate: X23 = 5.4, P = 0.1431). Specifically, flies had a greater probability of surviving the infection as temperatures decreased, with flies exposed to 21.4 °C exhibiting the greatest survival (57%). Survival quickly approached zero once temperature exceeded 28 °C. Pairwise significant tests based on 95% C.I.s suggest the infected temperature optimum (ITO) for our experimental design was 21.4 °C. The relationship between average chamber temperature and average survival was best described by a second order polynomial (chamber: F1, 9 = 79.5, P < 0.0001; chamber2: F1, 9 = 7.76, P = 0.0237) compared with a simple linear model and a third order polynomial (Fig. 3A; AIC scores: 17.4, 15.1, and 10.1 respectively). The lack of support for a third order polynomial model suggests that the true infected temperature optimum was not identified within our experimental temperature range. Uninfected flies survived at 100% until they reached 35.2 °C, at which point survival dropped to 16.0 ± 6.7% (Fig. 3A).
2.6. Follow-up experiment Once the temperature parameters were estimated (see results section), we conducted a follow-up experiment to verify the results. To this end, 40 flies were inoculated with an LD50 of P. aeruginosa as before and randomly assigned to either a 26.5 °C or a 25 °C treatment. Treatment temperatures were created using the no-choice apparatus sitting atop a temperature gradient. Twenty flies from each treatment were then placed into a single no-choice chamber maintained at the appropriate treatment temperature. Twenty uninfected control flies were simultaneously placed in a central no-choice chamber maintained at approximately 25.75 °C. This experiment was replicated four times and a total of 160 experimental flies and 80 control flies were used.
2.7. Statistical analysis To estimate ITO, we employed a general linear model using the binomial distribution, with chamber temperature and replicate as discrete predictor variables and status (alive or dead) as the binomial response. Replicate was included to account for variation in inoculation dose and hence mortality rates among the replicate groups. To determine pairwise differences between the chambers, we employed 95% confidence intervals; i.e. if the 95% C.I.s did not overlap chamber means, they were deemed to be significantly different from one another. To estimate UTP and ITP, we used a two-factor ANOVA with treatment and replicate as discrete predictor variables and chamber temperature as the response. Again, replicate was included to account for variation in inoculation dose, which could obscure differences between the treatments. Considering that there were four time points assessed (12, 24, 36 and 48 h), we analyzed each time interval separately. No interactions were statistically significant and therefore interactions were not included in the final model. For the follow-up experiment, we again employed a general linear model using the binomial distribution, with chamber temperature and replicate as discrete predictor variables and status (alive or dead) as the binomial response. All analyses were conducted using JMPÒ Pro version 11.0.0.
Fig. 3. (A) Estimation of the infected host’s optimal temperature, ITO. As temperature decreased, survival of infected flies (filled circles) increased, with the maximum survival recorded at the gradient minimum temperature of 21.4 °C. Chambers connected by the same letter were not significantly different based on 95% C.I.s. The relationship between mean chamber temperature and mean chamber survival was best described by a second order polynomial. Uninfected flies (open circles) exhibited 100% survival until 35.2 °C, at which point survival dropped to 16.0 ± 6.7% (B) When given a choice, uninfected hosts (open circles) preferred an average temperature of 26.3 °C over the 48 h period, while infected hosts (closed circles) preferred 25.2 °C (mean ± se). These preferences appeared consistent across time intervals. Dashed lines and corresponding numbers represent the probability of survival at that average temperature, based on the second order polynomial fit of the data. Asterisks represent statistical significance at a = 0.05.
40
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
When infected flies were presented with a choice in temperatures, they exhibited a behavioral chill. Uninfected flies exhibited an average temperature preference (UTP) of 26.3 + 0.1 °C over the 48 h period, while infected flies exhibited a preference (ITP) of 25.2 ± 0.1 °C. The difference in preference between uninfected and infected flies was apparent at each time interval, but was maximal at hour 36 (26.5 °C versus 25.2 °C, respectively; Fig. 3b and Table 1). These preferences also appeared to be stable over time for both treatments (infected time: F3,834 = 0.38, P = 0.7640; replicate: F7,830 = 1.9, P = 0.0544, versus uninfected time: F3,729 = 1.09, P = 0.3515; replicate: F7,725 = 4.17, P = 0.0002). The second order polynomial based on the no-choice data (Fig. 3A) indicated that a decrease in temperature from 26.3 °C to 25.2 °C, would result in a 23% increase in survival. Furthermore, the 25 °C no-choice chamber exhibited a significant 54% increase in survival over the 26.5 °C no-choice chamber based on 95% C.I.s. Thus, the observed behavioral chill appears to improve survival. Give these results, we conducted a follow-up experiment to verify that the preferred infection temperature provides a survival benefit compared with the preferred uninfected temperature. We found that infected flies at 25 °C exhibited a 48% increase in survival compared with those at 26.5 °C (mean survival: 63.9% versus 43.3%, respectively; chamber temperature: X21 = 6.7, P < 0.0097; replicate: X23 = 2.16, P = 0.5396). No uninfected control flies died during the follow-up experiment. In short, infected flies preferred cooler temperatures compared to uninfected flies, which coincided with improved survival. Moreover, the temperature preference of the infected flies did not approach their thermal optimum for combating the pathogen, causing the behavior to be inefficient. 4. Discussion The potential for a behavioral fever or chill to be inefficient has been largely overlooked in the literature. However, the occurrence of an inefficient response may be quite common given that (1) a complete lack of a behavioral response in infected hosts has been reported in a wide variety of systems (Stahlschmidt and Adamo 2013) and (2) an inefficient response can be caused by the same mechanisms driving a complete lack of response (see introduction). Here we show that when infected with a bacterial pathogen D. melanogaster exhibited a behavioral chill that improved survival. Specifically, infected hosts exhibited a 1.1 °C decrease in their normal preference, which remained consistent over the 48 h examination period. However, the infected temperature preference did not coincide with the optimum temperature for survival, remaining firmly between host and pathogen temperature optima. These data
Table 1 The effect of infection on temperature preference. Source
df
F
P
12 h Model Replicate Treatment
8, 407 7, 408 1, 414
1.81 1.68 3.03
0.0736 0.1108 0.0825
24 h Model Replicate Treatment
8, 394 7, 395 1, 401
4.97 3.56 10.7
<0.0001 0.0010 0.0012
36 h Model Replicate Treatment
8, 367 7, 368 1, 374
1.97 0.96 11.3
0.0491 0.4587 0.0009
48 h Model Replicate Treatment
8, 367 7, 368 1, 374
2.23 1.77 6.63
0.0245 0.0924 0.0104
suggest that the behavioral response to P. aeruginosa infection is considerably inefficient. One mechanism that may be responsible for the observed pattern is weak selection on the behavioral response to P. aeruginosa. Several known and unknown Pseudomonas species have been associated with D. melanogaster in the wild (Corby-Harris et al. 2007; Chandler et al. 2011), although none of these species have been definitively identified as P. aeruginosa. If contact with P. aeruginosa is infrequent, then selection may not have had the opportunity to optimize the behavioral response, causing it to be inefficient. However, elicitation of invertebrate immune systems is generally based on simple, highly conserved pathogen motifs (such as lipopolysaccharides, peptidoglycans and beta glucans), which act as pyrogens across a wide variety of animal systems (Stahlschmidt and Adamo 2013). Furthermore, P. aeruginosa does elicit an immune response in D. melanogaster by increasing antimicrobial peptide production and cellular response activity (Linder et al. 2008; Ye et al. 2009). Thus, weak selection may not be the most parsimonious explanation driving our results. A second mechanism that may underlie our results is host manipulation by P. aeruginosa. Previous work in D. melanogaster has shown that survival against P. aeruginosa is much improved at 17 °C compared with 25 °C. Furthermore, the in vitro growth rate of P. aeruginosa at 25 °C far exceeds its growth rate at 17 °C (Linder et al. 2008). Thus, P. aeruginosa would greatly increase its chances at transmission by keeping its host away from cooler temperatures. Moreover, P. aeruginosa has been shown to manipulate D. melanogaster antimicrobial peptide gene expression during the initial stages of infection (Apidianakis et al. 2005). Work in other systems has also indicated that parasite manipulation of host temperature preference does occur. For instance, Plasmodium mexicanum appears to increase the thermal preference of its intermediate sand fly host by 2 °C in order to facilitate its development and transmission to its definitive host (Fialho and Schall 1995). In short, parasite manipulation appears to be a viable mechanism for the patters observed here. It is interesting to note that a previous study showed that D. melanogaster did not modify their temperature preference when infected with a sterilizing nematode, even though an increased host body temperature would improve host defense (Ballabeni et al. 1995). It may be that behavioral response was inhibited by the nematode parasite. It is important to note that the failure of flies to reach their infected thermal optimum was not due to some high cost associated with such a response. Our experimental design avoided the types of sampling bias among our treatments (e.g. not accounting for age, sex, reproductive status or genetic background) that can make preferences appear inefficient (Fig. 1C). In addition, the optimal infected temperature (ITO) in our design (21.4 °C) increased the probability of survival over the infected temperature preference (25.2 °C) by 75% according to the second order polynomial fit of the data (Fig. 3A), indicating strong benefits for cold seeking flies. Moreover, flies surviving an LD50 of P. aeruginosa remain reproductively competent post infection (personal observation), indicating a relatively small reproductive cost for flies exhibiting a behavioral response. Our results show that D. melanogaster prefers colder temperatures when infected with P. aeruginosa. This behavioral chill makes sense, considering that warmer temperatures would improve the growth rate of the bacteria. Similar behavioral chills have been documented in other insect systems. Bumblebees infected with a parasitoid conopid fly spend their nights outside of the nest where they were exposed to colder temperatures, which improved their immune defense (Muller and Schmidhempel 1993). Likewise, cockroaches infected with temperature-sensitive acanthocephala worms retarded worm development by spending time in colder environments (Moore and Freehling 2002). Although this cold
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
seeking behavior is not a ‘‘fever”, it represents the same immune strategy; that is, modification of the host’s temperature ‘‘setpoint” to improve immune defense capabilities. In summary, we show that D. melanogaster’s behavioral immune response to P. aeruginosa is significantly inefficient. This inefficiency could be due to either weak selection on the behavior or parasite manipulation. Future work in this system should attempt to disentangle these mechanisms. This study is also the first to document a behavioral response to infection in D. melanogaster and reinforces the notion that such responses may be pathogen specific, most likely for the reasons listed above. Our work suggests that future studies examining fever or chill in other systems should document the efficiency of the response, as this may provide important insights into the selective pressures that shape this immune trait. Acknowledgments We thank Roxane Sperling for help with data collection. We also thank the University of Central Florida’s Office of Undergraduate Research and UCF’s Biology Department for funding this research. References Adamo, S.A., 1998. The specificity of behavioral fever in the cricket Acheta domesticus. J. Parasitol. 84, 529–533. Apidianakis, Y., Mindrinos, M.N., Xiao, W.Z., Lau, G.W., Baldini, R.L., Davis, R.W., Rahme, L.G., 2005. Profiling early infection responses: pseudomonas aeruginosa eludes host defenses by suppressing antimicrobial peptide gene expression. Proc. Natl. Acad. Sci. U.S.A. 102, 2573–2578. Apidianakis, Y., Rahme, L.G., 2009. Drosophila melanogaster as a model host for studying Pseudomonas aeruginosa infection. Nat. Protoc. 4, 1285–1294. Arnold, P.A., White, C.R., Johnson, K.N., 2015. Drosophila melanogaster does not exhibit a behavioural fever response when infected with Drosophila C virus. J. Gen. Virol. 96, 3667–3671. Ballabeni, P., Benway, H., Jaenike, J., 1995. Lack of behavioral fever in nematodeparasitized Drosohila. J. Parasitol. 81, 670–674. Boltana, S., Rey, S., Roher, N., Vargas, R., Huerta, M., Huntingford, F.A., Goetz, F.W., Moore, J., Garcia-Valtanen, P., Estepa, A., MacKenzie, S., 2013. Behavioural fever is a synergic signal amplifying the innate immune response. Proc. R. Soc. B-Biol. Sci. 280. Boorstein, S.M., Ewald, P.W., 1987. Costs and benefits of behavioral fever in Melanoplus sanguinipes infected by Nosema acridophagus. Physiol. Zool. 60, 586–595. Chandler, J.A., Lang, J.M., Bhatnagar, S., Eisen, J.A., Kopp, A., 2011. Bacterial communities of diverse drosophila species: ecological context of a hostmicrobe model system. PLoS Genet.
41
Corby-Harris, V., Pontaroli, A.C., Shimkets, L.J., Bennetzen, J.L., Habel, K.E., Promislow, D.E.L., 2007. Geographical distribution and diversity of bacteria associated with natural populations of Drosophila melanogaster. Appl. Environ. Microbiol. 73, 3470–3479. D’Argenio, D.A., Gallagher, L.A., Berg, C.A., Manoil, C., 2001. Drosophila as a model host for Pseudomonas aeruginosa infection. J. Bacteriol. 183, 1466–1471. Fedorka, K.M., 2014. Reproductive and immune system interactions in the context of life history and sexual selection theory. In: Ottaviani, E., Malagoli, D. (Eds.), Eco-immunology: Evolutive Aspects and Future Perspectives. Springer. Fedorka, K.M., Copeland, E.K., Winterhalter, W.E., 2013. Seasonality influences cuticle melanization and immune defense in a cricket: support for a temperature-dependent immune investment hypothesis in insects. J. Exp. Biol. 216, 4005–4010. Fialho, R.F., Schall, J.J., 1995. Thermal ecology of a malarial parasite and its insect vector - consequences for the parasites transmission success. J. Anim. Ecol. 64, 553–562. Hoffmann, J.A., 2003. The immune response of Drosophila. Nature 426, 33–38. Inglis, G.D., Johnson, D.L., Goettel, M.S., 1996. Effects of temperature and thermoregulation on mycosis by Beauveria bassiana in grasshoppers. Biol. Control 7, 131–139. Karban, R., 1998. Caterpillar basking behavior and nonlethal parasitism by tachinid flies. J. Insect. Behav. 11, 713–723. Kluger, M.J., 1978. Evolution and adaptive value of fever. Am. Sci. 66, 38–43. Kluger, M.J., Kozak, W., Conn, C.A., Leon, L.R., Soszynski, D., 1998. Role of fever in disease. In: Kluger, M.J., Bartfai, T., Dinarello, C.A. (Eds.), Molecular Mechanisms of Fever. New York Acad Sciences, New York, pp. 224–233. Kutch, I.C., Sevgili, H., Wittman, T., Fedorka, K.M., 2014. Thermoregulatory strategy may shape immune investment in Drosophila melanogaster. J. Exp. Biol. 217, 3664–3669. Lazzaro, B.P., Sceurman, B.K., Clark, A.G., 2004. Genetic basis of natural variation in D-melanogaster antibacterial immunity. Science 303, 1873–1876. Lemaitre, B., Hoffmann, J., 2007. The host defense of Drosophila melanogaster. In: Annu. Rev. Immunol.. Annual Reviews, Palo Alto, pp. 697–743. Linder, J.E., Owers, K.A., Promislow, D.E.L., 2008. The effects of temperature on hostpathogen interactions in D-melanogaster: who benefits? J. Insect Physiol. 54, 297–308. Martins, N.E., Faria, V.G., Teixeira, L., Magalhaes, S., Sucena, E., 2013. Host adaptation is contingent upon the infection route taken by pathogens. PLoS Pathog. 9, 8. Moore, J., 2002. Parasites and the Behavior of Animals. Oxford University Press. Moore, J., Freehling, M., 2002. Cockroach hosts in thermal gradients suppress parasite development. Oecologia 133, 261–266. Muller, C.B., Schmidhempel, P., 1993. Exploitation of cold temperature as defense agasint parasitoids in bumblebees. Nature 363, 65–67. Stahlschmidt, Z.R., Adamo, S.A., 2013. Context dependency and generality of fever in insects. Naturwissenschaften 100, 691–696. Vallet-Gely, I., Lemaitre, B., Boccard, F., 2008. Bacterial strategies to overcome insect defences. Nat. Rev. Microbiol. 6, 302–313. Watson, D.W., Mullens, B.A., Petersen, J.J., 1993. Behavioral fever reponse of Musca domestica (Diptera, Muscidae) to infection by Entomophithora muscae (Zygomycetes, Entomophithorales). J. Invertebr. Pathol. 61, 10–16. Wittman, T., Fedorka, K.M., 2015. Male mate choice for unparasitized females in Drosophila melanogaster. J. Insect. Behav. 28, 37–43. Ye, Y.X.H., Chenoweth, S.F., McGraw, E.A., 2009. Effective but costly, evolved mechanisms of defense against a virulent opportunistic pathogen in Drosophila melanogaster. PLoS Pathog. 5.
Contents lists available at ScienceDirect
Journal of Insect Physiology journal homepage: www.elsevier.com/locate/jinsphys
Cold temperature preference in bacterially infected Drosophila melanogaster improves survival but is remarkably suboptimal Kenneth M. Fedorka ⇑, Ian C. Kutch, Louisa Collins, Edward Musto University of Central Florida, Department of Biology, 4000 Central Florida Blvd., Orlando, FL 32816, United States
a r t i c l e
i n f o
Article history: Received 7 June 2016 Received in revised form 11 August 2016 Accepted 12 August 2016 Available online 13 August 2016 Keywords: Behavioral fever Behavioral chill Temperature preference Drosophila melanogaster Pseudomonas aeruginosa Parasite manipulation
a b s t r a c t Altering one’s temperature preference (e.g. behavioral fever or behavioral chill) is a common immune defense among ectotherms that is likely to be evolutionarily conserved. However, the temperature chosen by an infected host may not be optimal for pathogen defense, causing preference to be inefficient. Here we examined the efficiency of temperature preference in Drosophila melanogaster infected with an LD50 of the gram negative bacteria Pseudomonas aeruginosa. To this end, we estimated the host’s uninfected and infected temperature preferences as well as their optimal survival temperature. We found that flies decreased their preference from 26.3 °C to 25.2 °C when infected, and this preference was stable over 48 h. Furthermore, the decrease in temperature preference was associated with an increased chance of surviving the infection. Nevertheless, the infected temperature preference did not coincide with the optimum temperature for infection survival, which lies at or below 21.4 °C. These data suggest that the behavioral response to P. aeruginosa infection is considerably inefficient, and the mechanisms that may account for this pattern are discussed. Future studies of infected temperature preferences should document its efficiency, as this understudied aspect of behavioral immunity can provide important insight into preference evolution. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Fever in endotherms is a highly conserved, metabolicallygenerated increase in host body temperature that improves immune defense (Kluger 1978; Kluger et al. 1998). Ectotherms can also alter their body temperature adaptively in response to infection. However, this is accomplished by seeking out warmer (behavioral fever) or cooler (behavioral chill) temperatures, not by altering metabolic activity. For instance, invertebrate house flies (Musca domestica) and grasshoppers (Melanoplus sanguinipes) improve their chance of surviving fungal infections by relocating to warmer environments (Watson et al. 1993; Inglis et al. 1996). Infected vertebrate zebrafish (Danio rerio) likewise improve immune defense by seeking out warmer waters (Boltana et al. 2013). Such behavioral alterations to infection have been reported across a wide variety of ectothermic taxa and the phenomenon is likely to be evolutionarily conserved in animals (Kluger et al. 1998). Still, numerous studies have failed to detect changes in ⇑ Corresponding author at: University of Central Florida, Department of Biology, Bldg. 20, 4000 Central Florida Blvd., Orlando, FL 32816, United States. E-mail addresses: [email protected] (K.M. Fedorka), [email protected] (I.C. Kutch), [email protected] (L. Collins), [email protected] (E. Musto). http://dx.doi.org/10.1016/j.jinsphys.2016.08.005 0022-1910/Ó 2016 Elsevier Ltd. All rights reserved.
the thermal preferences of infected ectothermic hosts, even when fitness would be increased by such a response (Stahlschmidt and Adamo 2013), suggesting that the expression and/or evolution of the behavior is complex. Assuming that the host is capable of the behavior, there are several reasons why altered temperature preferences might not be elicited upon infection with a virulent pathogen. First, a change in temperature preference does not improve host fitness. This may occur if the host’s temperature-fitness function is flat for the pathogen in question, or if the host is already at its thermal optimum. Second, the cost of the behavioral response outweighs the benefit for a given host. This may occur when an individual host has little remaining residual reproductive value (e.g. older hosts or hosts infected with sterilizing pathogens), causing it to forgo an immune investment and instead invest in reproduction (e.g. terminal investment; Fedorka et al. 2013; Fedorka 2014). Third, there is a lack of coevolutionary history between the host and pathogen, which could result in the host not recognizing an appropriate Pathogen Associated Molecular Pattern (PAMP) to initiate the behavior. Fourth, the behavior is inhibited by the invading pathogen. Behavioral inhibition is plausible when one considers that pathogen manipulation of host behavior is a common occurrence (Moore 2002). Unfortunately, the causes underlying the lack of behavior are difficult to assess and are generally not explored.
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
Even if an infected host exhibits an adaptive shift in thermal preference, it does not mean the response is efficient. That is to say, the temperature chosen by a host when infected may not be the optimal temperature to maximize fitness. Inefficient temperature preference is an understudied aspect of ectothermic immune function that can provide important insights into the behavior’s evolution. One potential reason for an inefficient preference is the aforementioned lack of host-pathogen coevolutionary history. This can be caused by (i) weak selection for the optimal infected temperature preference due to infrequent host-pathogen interactions, or (ii) a lack of evolutionary time to allow selection to shape the preference, even if selection is relatively strong (Fig. 1A). A second potential reason for an inefficient preference is pathogen inhibition. In order to maximize its own fitness, a pathogen might manipulate the infected host’s temperature preference (ITP) that results in a perpetual tug-of-war that keeps the host away from its temperature optimum (Fig. 1B). It is important to note that a ‘‘weaker” response by a certain class of host (e.g. older hosts) due
37
to a higher behavioral cost does not represent an inefficient behavior, considering that the ‘‘weaker” response is still optimal for that class of host (Fig. 1C). It could be mistaken for inefficiency, however, if an experimental design suffered from sampling bias (e.g. not controlling for age, sex or reproductive history among treatments). Regrettably, most studies that document a host’s altered temperature preferences when infected do not estimate the optimal temperature when infected (e.g. Boorstein and Ewald 1987; Adamo 1998; Karban 1998; Moore and Freehling 2002), making the prevalence of inefficient responses unknown among animals. Here we examine the efficiency of temperature preference in Drosophila melanogaster infected with the gram negative bacterium Pseudomonas aeruginosa. We chose D. melanogaster as the host due to its prevalence as an invertebrate immunological model (Hoffmann 2003; Lemaitre and Hoffmann 2007) and P. aeruginosa as the pathogen because it is commonly used in invertebrate immunological studies (D’Argenio et al. 2001; Linder et al. 2008; Wittman and Fedorka 2015). Pseudomonas species exist naturally in wild populations of D. melanogaster, where oral-fecal transmission is the most likely route of infection, although infection via wounding may also occur (Corby-Harris et al. 2007; Chandler et al. 2011; Martins et al. 2013). Once infected, D. melanogaster rely on a variety of immunological responses against gram-negative bacteria, including ROS production, lysozymes, phenoloxidase cascade, and IMD-related antimicrobial peptides (Vallet-Gely et al. 2008). No previous D. melanogaster study has found evidence of behavioral fever/chill (Ballabeni et al. 1995; Arnold et al. 2015). Still, the elicitation of a behavioral response is likely to be pathogen specific (Adamo 1998), and a response against bacterial pathogens has yet to be examined (previous work focused on nematodes and viruses). To examine the existence and efficiency of such a response in D. melanogaster, we estimated the host’s uninfected (UTP) and infected (ITP) temperature preferences as well as their optimal body temperature when infected (ITO). If a host’s infected temperature preference (ITP) coincides with its infected temperature optimum (ITO), then the response would be deemed efficient. However, if ITP fails to reach the optimum, then the behavior would be considered inefficient, and one of the above mechanisms would be involved in shaping this important immunological response. 2. Methods 2.1. Stock maintenance
Fig. 1. The evolution and expression of behavioral fever. (A) When a lack of coevolutionary history exists between the host and pathogen, the infected host may not reach their infected temperature optimum (ITO), resulting in a suboptimal/ inefficient infected temperature preference (ITP). UTP represents the temperature preference of the uninfected host. (B) Pathogen manipulation may also result in a suboptimal ITP due to selective pressure on the pathogen to reach its optimal host body temperature (PO). (C) Multiple optima may exist for different groups of infected hosts within a species or population. Here, the fitness curves for two different hosts groups (A and B) are shown. For host B, a fully realized response may be too costly, and these hosts may exhibit a ‘‘weaker” infected temperature preference (ITP-B) due to a decreased infected temperature optimum (ITO-B) compared with host A’s preference and optimum (ITP-A and ITO-A). Although this response would not be considered inefficient, it could appear as such if the experimental design did not account for host sampling bias.
Flies in this study originated from 40 gravid females collected in 2010 from a single location in Orlando Florida and maintained as a large outbred stock (600 individuals per generation) to minimize the loss of genetic variation due to drift. Individuals were feed a cornmeal, yeast, sugar and agar food medium and kept in vertical incubators (Percival, Perry, IA, USA) at 25 °C with a 12 h:12 h light:dark photoperiod. Prior to all experiments, adult flies were collected upon eclosion and maintained in sex-specific vials at medium density (20 individuals per vial) to ensure virginity. All experimental flies were 5 ± 1 days old virgin females. 2.2. Experimental design To determine if D. melanogaster exhibits a shift in thermal preference when infected with P. aeruginosa, and whether this preference is efficient, we estimated three key parameters: UTP, ITP, ITO. These parameters were estimated using three separate treatments including an uninfected temperature-choice treatment (estimates UTP), an infected temperature-choice treatment (estimates ITP), and an infected no-choice temperature treatment
38
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
(estimates ITO). For the infected treatments, we inoculated flies with an LD50 of P. aeruginosa, while flies in the uninfected treatment were inoculated with the bacterial vehicle only (Luria broth – Fisher Scientific BP1426). We did not estimate the optimal host body temperature for P. aeruginosa transmission, but assumed that it resembled its optimal in vitro growth temperature of 37 °C (Linder et al. 2008). 2.3. Bacterial infections To create the LD50 concentration, bacteria were incubated in sterilized Luria broth at 37 °C until log phase. This solution was diluted with sterile broth to an absorbance of 0.3 at 490 nm using a microplate reader (Bio-Rad Model 680, Hercules, CA, USA). The new solution was further diluted to a concentration of 5 10 3, which corresponds to an LD50 for flies maintained at 25 °C and a 12:12 L:D photoperiod. The final bacterial solution was divided into 0.5 ml aliquots and stored at 4 °C. Prior to use, each aliquot was removed from refrigeration, vortexed and pipetted into a shallow microcentrifuge tube cap. Bacterial inoculation was achieved by dipping a thin needle (Fine Science Tools item # 26002-20) into the microcentrifuge cap and immediately piercing the thorax of the CO2 anesthetized fly just below and to the front of the wing attachment. This is a common method for infection (Lazzaro et al. 2004; Linder et al. 2008; Kutch et al. 2014) that provides reproducible results (Apidianakis and Rahme 2009), although variation in inoculation dose between flies does exist. Bacterial solution aliquots were used for no longer than 30 min before being replaced with a fresh solution. Between each infection, needles were dipped in 95% ethonol and allowed to air dry. Flies in the uninfected-choice treatment were inoculated with Luria broth only. 2.4. Temperature gradient To create the temperature gradient, we used an aluminum bar (91.4 30 2.5 cm; TECA, Chicago, IL, model #TGB-5030) atop of two cold/hot plates (TECA model #AHP-1200CPV). Each cold/hot plate has an operating range of 0 °C–50 °C, allowing a stable gradient of any length between these ranges to be produced. On top of the gradient were two types of assay apparatus, including a choice and a no-choice temperature apparatus (L W H = 68 8.5 4 cm each). Both types were constructed by imbedding 11 inverted, lidless petri dishes (D H = 60 15 mm; hereafter referred to as chambers) within spray foam insulation (Great StuffTM, Dow Chemicals), all contained within a wooden frame (Fig. 2A). The only difference between the two types of apparatus was that the choice apparatus had the portion of the walls removed where the chambers abutted, creating a long corridor that allowed flies to traverse the length of the gradient. On the top of each chamber, an 8 mm hole was created in which was placed a 1.5 ml microcentrifuge tube containing food media (the bottom of the tube was cutoff to allow access to the food). Along the top of each apparatus, we placed an LED strip connected to a dimmer switch to create a consistent light level in each chamber (RibbonFlex ProÒ, Armacost Lighting). This was done to minimize the confounding effects of phototaxis on parameter estimates. On the bottom of each apparatus was a removable layer of plastic wrap that minimized fly escapes between the apparatus and gradient bar. Each apparatus was then topped with a layer of polyurethane foam insulation. The layers of insulation were paramount in maintaining consistent air temperatures within each chamber, which was verified using button-sized temperature loggers (Embedded Data Systems model DS1922L; temperature range 40 °C–85 °C ± 0.5 °C; D H = 14 5 mm) adhered to the top of each chamber with a thin layer of modeling clay (Fig. 2B).
Fig. 2. Bottom-up view of the behavior and fitness assay apparatus. (A) Assay apparatuses were constructed out of petri dishes (i.e. chambers) surrounded by foam insulation and an LED light source. The temperature-choice apparatus had the chamber walls removed to create a movement corridor along the gradient. All chambers had access to food. (B) The air temperature of each chamber was accurate and consistent. Dashed line represents the programed temperature while the filled circles represent the actual air temperature means (standard errors are too small to see).
2.5. Estimation of parameters To estimate ITO, 110 flies were lightly CO2 anesthetized, inoculated with bacteria, and placed into a no-choice apparatus (10 flies per chamber). After the flies recovered from the anesthesia, the apparatus was placed on the gradient bar for 48 h, which ranged from 20 °C to 40 °C (Fig. 2B). At the end of hour 48, the apparatus was removed and the proportion of dead flies in each chamber counted. Previous work in our laboratory suggests that 25 °C flies that succumb to infection do so between hours 24 and 36. Previous work also suggests that flies alive at hour 48 show no increased mortality rates over the following seven days compared with controls (Kutch et al. 2014). Thus, flies that were alive at hour 48 were assumed to have fully recovered from infection. Estimation of ITO was replicated four times for a total of 440 flies across 11 different temperature regimes. In addition, we estimated the temperature tolerance of uninfected flies as a control. To this end, 330 flies where pricked with a needle dipped in sterile broth only and placed in the no-choice apparatus (10 flies per chamber) for 48 h, after which their survival was assayed. To estimate ITP and UTP, flies were lightly CO2 anesthetized and randomly assigned to either an infected-choice or an uninfectedchoice assay, respectively. Flies in the infected-choice assay were inoculated with bacteria, while flies in the uninfected-choice assay were pricked with a pin dipped in Luria broth only. Flies were
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
then placed into the central chamber of their respective temperature-choice apparatus (a total of 20 flies were released into each apparatus). Upon recovery from anesthesia, each apparatus was placed on the thermal gradient for 48 h. Using the information we gained from estimating ITO above, we modified the gradient range for this experiment to between 18 °C and 32 °C. At 12 h intervals, each apparatus was gently raised 1 m above the gradient for approximately 15 s while a second research photographed the underneath of the apparatus. This picture recorded the distribution of flies across the apparatus, which allowed ITP and UTP to be estimated at each time interval (we have no reason to believe this physically gentle and temporally brief disturbance had a longterm effect on fly distributions). ITP and UTP was calculated by assigning each fly to the temperature of the chamber in which it was found and averaging ‘‘fly temperatures” within treatments. Considering that the width of the gradient bar could accommodate three apparatuses, each replicate was composed of either one uninfected and 2 infected assays, or the converse. In each replicate, treatments were randomly assigned to one of the three choice apparatuses and to one of three positions on the gradient (top, middle or bottom). In total, 480 flies were assayed among 10 infected-choice replicates and 11 uninfected-choice replicates. The choice chambers were cleaned with 95% ethanol between each replicate run and allowed to dry.
39
3. Results There was a significant relationship between survival and the thermal environment for infected flies (Fig. 3A; chamber temperature: X210 = 113.7, P < 0.0001; replicate: X23 = 5.4, P = 0.1431). Specifically, flies had a greater probability of surviving the infection as temperatures decreased, with flies exposed to 21.4 °C exhibiting the greatest survival (57%). Survival quickly approached zero once temperature exceeded 28 °C. Pairwise significant tests based on 95% C.I.s suggest the infected temperature optimum (ITO) for our experimental design was 21.4 °C. The relationship between average chamber temperature and average survival was best described by a second order polynomial (chamber: F1, 9 = 79.5, P < 0.0001; chamber2: F1, 9 = 7.76, P = 0.0237) compared with a simple linear model and a third order polynomial (Fig. 3A; AIC scores: 17.4, 15.1, and 10.1 respectively). The lack of support for a third order polynomial model suggests that the true infected temperature optimum was not identified within our experimental temperature range. Uninfected flies survived at 100% until they reached 35.2 °C, at which point survival dropped to 16.0 ± 6.7% (Fig. 3A).
2.6. Follow-up experiment Once the temperature parameters were estimated (see results section), we conducted a follow-up experiment to verify the results. To this end, 40 flies were inoculated with an LD50 of P. aeruginosa as before and randomly assigned to either a 26.5 °C or a 25 °C treatment. Treatment temperatures were created using the no-choice apparatus sitting atop a temperature gradient. Twenty flies from each treatment were then placed into a single no-choice chamber maintained at the appropriate treatment temperature. Twenty uninfected control flies were simultaneously placed in a central no-choice chamber maintained at approximately 25.75 °C. This experiment was replicated four times and a total of 160 experimental flies and 80 control flies were used.
2.7. Statistical analysis To estimate ITO, we employed a general linear model using the binomial distribution, with chamber temperature and replicate as discrete predictor variables and status (alive or dead) as the binomial response. Replicate was included to account for variation in inoculation dose and hence mortality rates among the replicate groups. To determine pairwise differences between the chambers, we employed 95% confidence intervals; i.e. if the 95% C.I.s did not overlap chamber means, they were deemed to be significantly different from one another. To estimate UTP and ITP, we used a two-factor ANOVA with treatment and replicate as discrete predictor variables and chamber temperature as the response. Again, replicate was included to account for variation in inoculation dose, which could obscure differences between the treatments. Considering that there were four time points assessed (12, 24, 36 and 48 h), we analyzed each time interval separately. No interactions were statistically significant and therefore interactions were not included in the final model. For the follow-up experiment, we again employed a general linear model using the binomial distribution, with chamber temperature and replicate as discrete predictor variables and status (alive or dead) as the binomial response. All analyses were conducted using JMPÒ Pro version 11.0.0.
Fig. 3. (A) Estimation of the infected host’s optimal temperature, ITO. As temperature decreased, survival of infected flies (filled circles) increased, with the maximum survival recorded at the gradient minimum temperature of 21.4 °C. Chambers connected by the same letter were not significantly different based on 95% C.I.s. The relationship between mean chamber temperature and mean chamber survival was best described by a second order polynomial. Uninfected flies (open circles) exhibited 100% survival until 35.2 °C, at which point survival dropped to 16.0 ± 6.7% (B) When given a choice, uninfected hosts (open circles) preferred an average temperature of 26.3 °C over the 48 h period, while infected hosts (closed circles) preferred 25.2 °C (mean ± se). These preferences appeared consistent across time intervals. Dashed lines and corresponding numbers represent the probability of survival at that average temperature, based on the second order polynomial fit of the data. Asterisks represent statistical significance at a = 0.05.
40
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
When infected flies were presented with a choice in temperatures, they exhibited a behavioral chill. Uninfected flies exhibited an average temperature preference (UTP) of 26.3 + 0.1 °C over the 48 h period, while infected flies exhibited a preference (ITP) of 25.2 ± 0.1 °C. The difference in preference between uninfected and infected flies was apparent at each time interval, but was maximal at hour 36 (26.5 °C versus 25.2 °C, respectively; Fig. 3b and Table 1). These preferences also appeared to be stable over time for both treatments (infected time: F3,834 = 0.38, P = 0.7640; replicate: F7,830 = 1.9, P = 0.0544, versus uninfected time: F3,729 = 1.09, P = 0.3515; replicate: F7,725 = 4.17, P = 0.0002). The second order polynomial based on the no-choice data (Fig. 3A) indicated that a decrease in temperature from 26.3 °C to 25.2 °C, would result in a 23% increase in survival. Furthermore, the 25 °C no-choice chamber exhibited a significant 54% increase in survival over the 26.5 °C no-choice chamber based on 95% C.I.s. Thus, the observed behavioral chill appears to improve survival. Give these results, we conducted a follow-up experiment to verify that the preferred infection temperature provides a survival benefit compared with the preferred uninfected temperature. We found that infected flies at 25 °C exhibited a 48% increase in survival compared with those at 26.5 °C (mean survival: 63.9% versus 43.3%, respectively; chamber temperature: X21 = 6.7, P < 0.0097; replicate: X23 = 2.16, P = 0.5396). No uninfected control flies died during the follow-up experiment. In short, infected flies preferred cooler temperatures compared to uninfected flies, which coincided with improved survival. Moreover, the temperature preference of the infected flies did not approach their thermal optimum for combating the pathogen, causing the behavior to be inefficient. 4. Discussion The potential for a behavioral fever or chill to be inefficient has been largely overlooked in the literature. However, the occurrence of an inefficient response may be quite common given that (1) a complete lack of a behavioral response in infected hosts has been reported in a wide variety of systems (Stahlschmidt and Adamo 2013) and (2) an inefficient response can be caused by the same mechanisms driving a complete lack of response (see introduction). Here we show that when infected with a bacterial pathogen D. melanogaster exhibited a behavioral chill that improved survival. Specifically, infected hosts exhibited a 1.1 °C decrease in their normal preference, which remained consistent over the 48 h examination period. However, the infected temperature preference did not coincide with the optimum temperature for survival, remaining firmly between host and pathogen temperature optima. These data
Table 1 The effect of infection on temperature preference. Source
df
F
P
12 h Model Replicate Treatment
8, 407 7, 408 1, 414
1.81 1.68 3.03
0.0736 0.1108 0.0825
24 h Model Replicate Treatment
8, 394 7, 395 1, 401
4.97 3.56 10.7
<0.0001 0.0010 0.0012
36 h Model Replicate Treatment
8, 367 7, 368 1, 374
1.97 0.96 11.3
0.0491 0.4587 0.0009
48 h Model Replicate Treatment
8, 367 7, 368 1, 374
2.23 1.77 6.63
0.0245 0.0924 0.0104
suggest that the behavioral response to P. aeruginosa infection is considerably inefficient. One mechanism that may be responsible for the observed pattern is weak selection on the behavioral response to P. aeruginosa. Several known and unknown Pseudomonas species have been associated with D. melanogaster in the wild (Corby-Harris et al. 2007; Chandler et al. 2011), although none of these species have been definitively identified as P. aeruginosa. If contact with P. aeruginosa is infrequent, then selection may not have had the opportunity to optimize the behavioral response, causing it to be inefficient. However, elicitation of invertebrate immune systems is generally based on simple, highly conserved pathogen motifs (such as lipopolysaccharides, peptidoglycans and beta glucans), which act as pyrogens across a wide variety of animal systems (Stahlschmidt and Adamo 2013). Furthermore, P. aeruginosa does elicit an immune response in D. melanogaster by increasing antimicrobial peptide production and cellular response activity (Linder et al. 2008; Ye et al. 2009). Thus, weak selection may not be the most parsimonious explanation driving our results. A second mechanism that may underlie our results is host manipulation by P. aeruginosa. Previous work in D. melanogaster has shown that survival against P. aeruginosa is much improved at 17 °C compared with 25 °C. Furthermore, the in vitro growth rate of P. aeruginosa at 25 °C far exceeds its growth rate at 17 °C (Linder et al. 2008). Thus, P. aeruginosa would greatly increase its chances at transmission by keeping its host away from cooler temperatures. Moreover, P. aeruginosa has been shown to manipulate D. melanogaster antimicrobial peptide gene expression during the initial stages of infection (Apidianakis et al. 2005). Work in other systems has also indicated that parasite manipulation of host temperature preference does occur. For instance, Plasmodium mexicanum appears to increase the thermal preference of its intermediate sand fly host by 2 °C in order to facilitate its development and transmission to its definitive host (Fialho and Schall 1995). In short, parasite manipulation appears to be a viable mechanism for the patters observed here. It is interesting to note that a previous study showed that D. melanogaster did not modify their temperature preference when infected with a sterilizing nematode, even though an increased host body temperature would improve host defense (Ballabeni et al. 1995). It may be that behavioral response was inhibited by the nematode parasite. It is important to note that the failure of flies to reach their infected thermal optimum was not due to some high cost associated with such a response. Our experimental design avoided the types of sampling bias among our treatments (e.g. not accounting for age, sex, reproductive status or genetic background) that can make preferences appear inefficient (Fig. 1C). In addition, the optimal infected temperature (ITO) in our design (21.4 °C) increased the probability of survival over the infected temperature preference (25.2 °C) by 75% according to the second order polynomial fit of the data (Fig. 3A), indicating strong benefits for cold seeking flies. Moreover, flies surviving an LD50 of P. aeruginosa remain reproductively competent post infection (personal observation), indicating a relatively small reproductive cost for flies exhibiting a behavioral response. Our results show that D. melanogaster prefers colder temperatures when infected with P. aeruginosa. This behavioral chill makes sense, considering that warmer temperatures would improve the growth rate of the bacteria. Similar behavioral chills have been documented in other insect systems. Bumblebees infected with a parasitoid conopid fly spend their nights outside of the nest where they were exposed to colder temperatures, which improved their immune defense (Muller and Schmidhempel 1993). Likewise, cockroaches infected with temperature-sensitive acanthocephala worms retarded worm development by spending time in colder environments (Moore and Freehling 2002). Although this cold
K.M. Fedorka et al. / Journal of Insect Physiology 93–94 (2016) 36–41
seeking behavior is not a ‘‘fever”, it represents the same immune strategy; that is, modification of the host’s temperature ‘‘setpoint” to improve immune defense capabilities. In summary, we show that D. melanogaster’s behavioral immune response to P. aeruginosa is significantly inefficient. This inefficiency could be due to either weak selection on the behavior or parasite manipulation. Future work in this system should attempt to disentangle these mechanisms. This study is also the first to document a behavioral response to infection in D. melanogaster and reinforces the notion that such responses may be pathogen specific, most likely for the reasons listed above. Our work suggests that future studies examining fever or chill in other systems should document the efficiency of the response, as this may provide important insights into the selective pressures that shape this immune trait. Acknowledgments We thank Roxane Sperling for help with data collection. We also thank the University of Central Florida’s Office of Undergraduate Research and UCF’s Biology Department for funding this research. References Adamo, S.A., 1998. The specificity of behavioral fever in the cricket Acheta domesticus. J. Parasitol. 84, 529–533. Apidianakis, Y., Mindrinos, M.N., Xiao, W.Z., Lau, G.W., Baldini, R.L., Davis, R.W., Rahme, L.G., 2005. Profiling early infection responses: pseudomonas aeruginosa eludes host defenses by suppressing antimicrobial peptide gene expression. Proc. Natl. Acad. Sci. U.S.A. 102, 2573–2578. Apidianakis, Y., Rahme, L.G., 2009. Drosophila melanogaster as a model host for studying Pseudomonas aeruginosa infection. Nat. Protoc. 4, 1285–1294. Arnold, P.A., White, C.R., Johnson, K.N., 2015. Drosophila melanogaster does not exhibit a behavioural fever response when infected with Drosophila C virus. J. Gen. Virol. 96, 3667–3671. Ballabeni, P., Benway, H., Jaenike, J., 1995. Lack of behavioral fever in nematodeparasitized Drosohila. J. Parasitol. 81, 670–674. Boltana, S., Rey, S., Roher, N., Vargas, R., Huerta, M., Huntingford, F.A., Goetz, F.W., Moore, J., Garcia-Valtanen, P., Estepa, A., MacKenzie, S., 2013. Behavioural fever is a synergic signal amplifying the innate immune response. Proc. R. Soc. B-Biol. Sci. 280. Boorstein, S.M., Ewald, P.W., 1987. Costs and benefits of behavioral fever in Melanoplus sanguinipes infected by Nosema acridophagus. Physiol. Zool. 60, 586–595. Chandler, J.A., Lang, J.M., Bhatnagar, S., Eisen, J.A., Kopp, A., 2011. Bacterial communities of diverse drosophila species: ecological context of a hostmicrobe model system. PLoS Genet.
41
Corby-Harris, V., Pontaroli, A.C., Shimkets, L.J., Bennetzen, J.L., Habel, K.E., Promislow, D.E.L., 2007. Geographical distribution and diversity of bacteria associated with natural populations of Drosophila melanogaster. Appl. Environ. Microbiol. 73, 3470–3479. D’Argenio, D.A., Gallagher, L.A., Berg, C.A., Manoil, C., 2001. Drosophila as a model host for Pseudomonas aeruginosa infection. J. Bacteriol. 183, 1466–1471. Fedorka, K.M., 2014. Reproductive and immune system interactions in the context of life history and sexual selection theory. In: Ottaviani, E., Malagoli, D. (Eds.), Eco-immunology: Evolutive Aspects and Future Perspectives. Springer. Fedorka, K.M., Copeland, E.K., Winterhalter, W.E., 2013. Seasonality influences cuticle melanization and immune defense in a cricket: support for a temperature-dependent immune investment hypothesis in insects. J. Exp. Biol. 216, 4005–4010. Fialho, R.F., Schall, J.J., 1995. Thermal ecology of a malarial parasite and its insect vector - consequences for the parasites transmission success. J. Anim. Ecol. 64, 553–562. Hoffmann, J.A., 2003. The immune response of Drosophila. Nature 426, 33–38. Inglis, G.D., Johnson, D.L., Goettel, M.S., 1996. Effects of temperature and thermoregulation on mycosis by Beauveria bassiana in grasshoppers. Biol. Control 7, 131–139. Karban, R., 1998. Caterpillar basking behavior and nonlethal parasitism by tachinid flies. J. Insect. Behav. 11, 713–723. Kluger, M.J., 1978. Evolution and adaptive value of fever. Am. Sci. 66, 38–43. Kluger, M.J., Kozak, W., Conn, C.A., Leon, L.R., Soszynski, D., 1998. Role of fever in disease. In: Kluger, M.J., Bartfai, T., Dinarello, C.A. (Eds.), Molecular Mechanisms of Fever. New York Acad Sciences, New York, pp. 224–233. Kutch, I.C., Sevgili, H., Wittman, T., Fedorka, K.M., 2014. Thermoregulatory strategy may shape immune investment in Drosophila melanogaster. J. Exp. Biol. 217, 3664–3669. Lazzaro, B.P., Sceurman, B.K., Clark, A.G., 2004. Genetic basis of natural variation in D-melanogaster antibacterial immunity. Science 303, 1873–1876. Lemaitre, B., Hoffmann, J., 2007. The host defense of Drosophila melanogaster. In: Annu. Rev. Immunol.. Annual Reviews, Palo Alto, pp. 697–743. Linder, J.E., Owers, K.A., Promislow, D.E.L., 2008. The effects of temperature on hostpathogen interactions in D-melanogaster: who benefits? J. Insect Physiol. 54, 297–308. Martins, N.E., Faria, V.G., Teixeira, L., Magalhaes, S., Sucena, E., 2013. Host adaptation is contingent upon the infection route taken by pathogens. PLoS Pathog. 9, 8. Moore, J., 2002. Parasites and the Behavior of Animals. Oxford University Press. Moore, J., Freehling, M., 2002. Cockroach hosts in thermal gradients suppress parasite development. Oecologia 133, 261–266. Muller, C.B., Schmidhempel, P., 1993. Exploitation of cold temperature as defense agasint parasitoids in bumblebees. Nature 363, 65–67. Stahlschmidt, Z.R., Adamo, S.A., 2013. Context dependency and generality of fever in insects. Naturwissenschaften 100, 691–696. Vallet-Gely, I., Lemaitre, B., Boccard, F., 2008. Bacterial strategies to overcome insect defences. Nat. Rev. Microbiol. 6, 302–313. Watson, D.W., Mullens, B.A., Petersen, J.J., 1993. Behavioral fever reponse of Musca domestica (Diptera, Muscidae) to infection by Entomophithora muscae (Zygomycetes, Entomophithorales). J. Invertebr. Pathol. 61, 10–16. Wittman, T., Fedorka, K.M., 2015. Male mate choice for unparasitized females in Drosophila melanogaster. J. Insect. Behav. 28, 37–43. Ye, Y.X.H., Chenoweth, S.F., McGraw, E.A., 2009. Effective but costly, evolved mechanisms of defense against a virulent opportunistic pathogen in Drosophila melanogaster. PLoS Pathog. 5.