Consequences of chronic bacterial infection in Drosophila melanogaster


Authors: Moria Cairns Chambers aff001;  Eliana Jacobson aff001;  Sarah Khalil aff001;  Brian P. Lazzaro aff001
Authors place of work: Department of Entomology, Cornell University, Ithaca, New York, United States of America aff001;  Department of Biology, Bucknell University, Lewisburg, PA, United States of America aff002;  Cornell Institute of Host-Microbe Interactions and Disease, Cornell University, Ithaca, New York, United States of America aff003
Published in the journal: PLoS ONE 14(10)
Category: Research Article
doi: 10.1371/journal.pone.0224440

Summary

Even when successfully surviving an infection, a host often fails to eliminate a pathogen completely and may sustain substantial pathogen burden for the remainder of its life. Using systemic bacterial infection in Drosophila melanogaster, we characterize chronic infection by three bacterial species from different genera - Providencia rettgeri, Serratia marcescens, and Enterococcus faecalis–following inoculation with a range of doses. To assess the consequences of these chronic infections, we determined the expression of antimicrobial peptide genes, survival of secondary infection, and starvation resistance after one week of infection. While higher infectious doses unsurprisingly lead to higher risk of death, they also result in higher chronic bacterial loads among the survivors for all three infections. All three chronic infections caused significantly elevated expression of antimicrobial peptide genes at one week post-infection and provided generalized protection again secondary bacterial infection. Only P. rettgeri infection significantly influenced resistance to starvation, with persistently infected flies dying more quickly under starvation conditions relative to controls. These results suggest that there is potentially a generalized mechanism of protection against secondary infection, but that other impacts on host physiology may depend on the specific pathogen. We propose that chronic infections in D. melanogaster could be a valuable tool for studying tolerance of infection, including impacts on host physiology and behavior.

Keywords:

Bacterial diseases – Drosophila melanogaster – Enterococcus faecalis – Enterococcus infections – Gene expression – Pathogens – Serratia marcescens – Serratia infections

Introduction

Animal physiology is profoundly impacted by constant association with microbes. Commensal microbes that live on epithelial surfaces, in the gut, and in other body compartments have recently come to prominent attention. Another source of microbial association comes from pathogenic infections that remain in the host even after symptoms resolve. We and others have observed that many bacterial infections in Drosophila melanogaster are never eradicated; they are only controlled into chronic persistence [15]. Previous research has focused on the processes leading to chronic infection and the consequent immune stimulation [13,5]. However, aside from a small set of studies evaluating the reproductive output of females [1,4], there has been less effort focused on the longer term consequences of chronic infection.

Our work expands our understanding of chronic infection using three bacterial pathogens—Providencia rettgeri, Serratia marcescens, and Enterococcus faecalis–which are frequently employed in experimental infection of D. melanogaster. Most mortality from these infections occurs within three days post-infection [1,2,59]. Individuals that survive sustain stable bacterial loads for the remainder of their life with minimal impact on life-span in the laboratory [2]. Chronic pathogen burdens for these bacterial species can be as high as ten thousand colony forming units (CFU) per fly [1,2,5,9,10].

Sustaining an infection for the remainder of life may have immunological and physiological consequences for the host. The presence of bacteria and bacterially-derived peptidoglycan stimulates the D. melanogaster immune system [1118]. Induced expression of antimicrobial peptide genes is positively correlated with bacterial load during lethal infections in D. melanogaster, the red flour beetle (Tribolium castaneum) and the buff-tailed bumblebee (Bombus terrestris) [1921]. In the mealworm beetle (Tenebrio molitor), sustained bacterial infection is associated with continued expression of antimicrobial peptide genes [22,23]. These observations led us to hypothesize that chronically infected D. melanogaster may be in a state of chronic immune stimulation, which may be protective against future infection. For example, previous studies have shown that immune stimulation after clearing a live bacterial infection, or after injection of heat-killed bacteria, provides some protection against future infections in D. melanogaster [24,25]. Similar effects have been seen in other insect species [26,27]. However, on the other hand, compounding combinatorial infection could overwhelm the fly and cause increased probability of death. The strength of protection in D. melanogaster varies depending on the bacterial species of both the initial inoculation and secondary challenge [24,25], and lingering live bacteria during chronic infection might continue to damage the host.

In this work, we determine the consequences of chronic infection with three extracellular opportunistic pathogens: Providencia rettgeri, Serratia marcescens, and Enterococcus faecalis. These data provide insight into the physiological consequences of chronically sustained extracellular bacterial infection in D. melanogaster, establishing the foundation for future studies on tolerance of infection in Drosophila and other insects.

Materials and methods

D. melanogaster strains and husbandry

Unless specified otherwise, experiments were done using wild-type Canton-S (CS, Bloomington stock # 1). Immune reporter flies that express GFP under promoters from the antibacterial peptide genes attacin A, diptericin A, and drosomycin [28] were used for visual readout of the immune response. The flies were reared as described previously in Chambers et al. [2]. For rearing each line, approximately fifteen males and fifteen females were placed in plastic bottles with 50mL of a glucose-cornmeal-yeast medium (50g/L yeast, 60g/L cornmeal, 40g/L glucose, 7g/L Drosophila Agar, 12mL of Acid Mix (41.5mL Phosphoric Acid + 418mL Propionic Acid + 540.5mL distilled water) and 26.5mL/L methylparaben (Moldex)) and allowed to lay eggs for two days. Adults were cleared and offspring were given ten days to develop. Eclosed adults were transferred to a fresh bottle, kept overnight to allow mating, and sorted on the subsequent day into vials of twenty males. Flies were aged for four additional days before injection, such that all experiments were performed with flies aged 5–7 day post-eclosion. Experiments were performed with male flies, which were flipped to new vials every seven days to keep flies on relatively fresh food. All flies were kept at 25°C in an incubator with a 12:12 hour light-dark cycle.

Bacterial strains and infection conditions

Three bacterial species were used: Providencia rettgeri (strain Dmel, isolated as an infection of field-caught D. melanogaster [29]), Serratia marcescens (strain BPL, an attenuated strain derived from the Type strain ATCC 13880 [30]) and Enterococcus faecalis (isolated as an infection of field-caught D. melanogaster [29]).

Bacterial stocks were maintained and infection procedures were performed as described previously [2,31]. Bacterial stocks were stored at -80°C in Luria Bertani (LB) broth containing 15% glycerol. Bacteria were first streaked on an LB agar plate from the stock, grown overnight at 37°C, and subsequently stored at 4°C for up to a month. For infection, liquid bacterial cultures were inoculated from a single bacterial colony from the plate and grown overnight for 16–20 hours. P. rettgeri and S. marcescens were grown in 2ml LB broth at 37°C with shaking in an orbital shaker at 120 rotations per minute. E. faecalis, which is microaerophilic, was grown in 2mL BHI broth at 37°C standing. The overnight cultures were then diluted in Phosphate Buffer Saline (PBS, pH 7.4) to a desired optical density (Absorbance600). A volume of 23nL of bacterial suspension was injected into each fly using a Nanoject II (Drummond, www.drummondsci.com) as described in Khalil et al. [31]. To initiate chronic infection, injections were done in the abdomen to minimize mortality [2] allowing us to efficiently focus on later stages of infection. For experiments where we wanted to initiate lethal secondary infection, we injected bacteria into the thorax because this generates a higher bacterial burden and more lethal infection [2]. Bacterial suspensions were generally injected at an optical density OD (A600) of 0.1, which corresponds to a dose of about 2,000–4,500 Colony Forming Units (CFU; viable bacteria) per fly for P. rettgeri, 2,000–6,000 CFU for S. marcescens and 2,000–4,000 CFU for E. faecalis. In the experiment explicitly testing infectious dose, other densities were used as needed. Sterile PBS was injected as a control. Flies were anesthetized with light CO2 for less than five minutes during the infection procedure. All controls were exposed to CO2 for the same amount of time and no CO2-induced mortality was observed. Infected flies were kept in groups with a maximum of 20 per vial (typically 16–20 flies) at 25°C in an incubator with a 12:12 hour light-dark cycle.

Infectious dose survival experiments

For each experimental treatment, flies were injected in the abdomen with bacterial suspensions ranging from OD of 0.001 to 5.0 or with PBS as a control. Flies were housed in groups, with a maximum of 20 flies per vial (typically 18–20). Experimental blocks contained 2–3 vials of flies per experimental treatment, and there were three experimental blocks (independently performed experiments with a new bacterial culture and set of flies) completed for each bacterial species. After inoculation, death was recorded daily for two weeks, with dead flies left in the vials. After one week, surviving flies were transferred to fresh vials, so the number of flies per vial was subsequently more variable as it was dependent on mortality that occurred prior to transfer. Survival curves were plotted as Kaplan-Meier plots using Graphpad Prism 7.0. Statistical significance was determined using Cox Proportional Hazards Mixed Effects Model (coxme) with the package “Coxme” in R (http://www.r-project.org/). Fixed effects included were experimental block (B) and the infectious dose (D), and the interaction between B and D. Fixed effects are incorporated in sequence and factors are listed in order of inclusion in result tables. To account for the experimental structure, vial (V) nested inside of both B and D was included as a random effect. Infectious dose was estimated by using the optical density of the injected bacterial suspensions. Our models fulfill the assumption of Cox Proportional Hazards in that the proportional hazards do not cross over-time.

Model A: coxme(status,time) = B + D + (B x D) +(1|B/D/V)

Bacterial load

For each experimental block, approximately 120 flies were injected in the abdomen with bacterial suspensions at each optical density (0.001, 0.01, 0.1, and 0.5) or injected with PBS as control. Flies were housed in groups, with a maximum of 20 flies per vial (typically 18–20) at 25°C. Pathogen burden at various time points post-infection was determined as the number of colony forming units (CFU) recovered from homogenized flies as described previously by our research group [2,31]. P. rettgeri and S. marcescens bacterial loads were determined using a WASP II autoplate spiral plater (Microbiology International). In these experiments, six or more flies from the same vial were individually homogenized in 250uL of PBS at the relevant time-point. For zero hour time-points, flies were placed on ice within two minutes of injection and homogenized within thirty minutes. Infection conditions expected to yield bacterial loads greater than 104 CFU/fly were diluted ten-fold. The spiral plater lays 50uL of homogenate in a continuously decreasing concentric spiral on an LB plate. Plates were grown overnight at 37°C and then counted using the EZ-Count Automated Colony Counter (Microbiology International), which calculates the number of CFU per fly based on the number of colonies and their position along the spiral, allowing for effective resolution of three orders of magnitude on a single plate. E. faecalis load was determined using spot-plating due to high variability in load among individuals, making prediction of the appropriate dilution for spiral plating difficult. In order to conserve plates, we moved to spot plating, which can resolve bacterial load over six orders of magnitude on a single plate but is more technically difficult. In these experiments, 8–16 individual flies from the same vial were collected at each time point. Each individual fly was homogenized in 100uL, then the homogenate was diluted serially 1:10, 10uL of each dilution spotted onto LB agar and grown overnight at 37°C. Spots containing 30–300 colonies were counted and used to calculate CFU per fly. For all experiments, flies injected with sterile PBS were homogenized and plated to test for contamination and experiments where growth was detected in controls were discarded. Three experimental blocks were completed for both P. rettgeri and E. faecalis, while two experimental blocks were completed for S. marcescens because one block was discarded due to contamination found in sterile PBS injected controls. Analysis of variance (ANOVA) was used to determine the impact of infectious dose on bacterial loads one week and two weeks post-infection using packages “lme4” and “nlme” in R (http://www.r-project.org/). Factors included were log10 average infectious dose (D) and block (B). Infectious dose was estimated by using the average bacterial load at zero hours post-infection within each experimental block. Each factor is incorporated in sequence and factors are listed in order of inclusion in result tables. All models reported fulfilled assumptions of normality as determined by the Shapiro Test.

Model B: log10(persistent bacterial load) = B + D + (B x D)

Quantitative RT-PCR

For each experimental block, groups of sixty flies were inoculated in the abdomen with either P. rettgeri, E. faecalis or S. marcescens and sixty flies were injected with sterile PBS in the abdomen. Following injection, the flies were housed in vials in groups of twenty and left at 25°C for seven days, during which time mortality remained under twenty percent. Surviving flies were sorted into groups of fifteen, homogenized in TRIzol (Life Technologies) and stored at -80°C until further processing. Three experimental blocks were completed, each with three homogenates for each condition. Samples were processed as described previously by our research group [2]. RNA was isolated using a TRIzol extraction, RNA samples were treated with DNase (Promega), and cDNA was generated using MLV-RT (Promega) and then diluted 1:20 in water. Quantitative PCR was performed using the SSO Advanced SYBR Green Kit (Bio-Rad) following the manufacturer’s directions, scaled down for 15μL reactions. For consistency across studies, we used the same qPCR primers as previous studies [2,31,32]. To allow calculation of primer efficiency, a standard curve generated by ten-fold serial dilutions of pooled cDNA was run on each plate. Primer sequences are reported in Table 1. For each cDNA sample, three technical replicates were performed for each primer set and the average threshold cycle (Ct) was used for further analysis. Relative expression values for each sample were calculating using the Pfaffl method, which is similar to the commonly used 2-ΔΔCt method except that it uses experimentally determined amplification efficiencies instead of assuming perfect efficiency for every set of primers [33]. In these analyses, expression levels of the gene of interest are normalized to a reference gene (rp49). The threshold cycles for the reference gene were all within one Ct, indicating that there was consistent isolation and synthesis of cDNA across all samples. The output of the Pfaffl method for each sample was then used as the response variable in an ANOVA performed using the “aov” command in R. Factors included were experimental block (B), infection treatment (T: Sterile PBS, P. rettgeri, S. marcescens, or E. faecalis) and the interaction between them.

Tab. 1. Primers for qRT-PCR.
Primers for qRT-PCR.

Model C: Log2(relative expression) = B + T+ (B x T)

Each factor is incorporated in sequence and factors are listed in order of inclusion in result tables. Pairwise comparisons between individual conditions were determined by Tukey’s multiple comparison tests using the “TukeyHSD” command in R. Expression levels for all five AMP genes were assessed using the same fly homogenates, so to control for multiple testing we used a Bonferroni correction, which divides the threshold p-value by the number of tests, lowering our significance threshold to p-values < 0.01 instead of 0.05. Since this correction assumes that all tests are independent, it is conservative in the current context where the tested genes are at least partially coordinately regulated.

Microscopy of GFP reporter flies

Ten flies of each reporter line were injected with 23nL of sterile media or bacterial suspension in the abdomen, with a goal of recovering at least six surviving flies for imaging (anticipating about 10–20% mortality). One week after injection, surviving GFP-reporter flies were anaesthetized on ice for 30 min. Flies were then imaged in groups of six using both brightfield and epifluorescent illumination with a Leica M165 FC microscope fitted with a Leica DFC 450 camera. Images were captured with Leica Application Suite V4.0. Images for all settings were captured in a single sitting using identical settings on the microscope.

Secondary infection experiments

For each experiment, flies were first injected in the abdomen with bacterial suspensions (OD 0.1) to initiate chronic infection or injected with sterile PBS as control. The injections were done in the abdomen to minimize mortality [2] allowing us to efficiently focus on later stages of infection. Flies were housed in groups with a maximum of 20 flies per vial (typically 18–20). Infected flies were kept at 25°C in an incubator with a 12:12 hour light-dark cycle for one week to allow chronic infection to develop, during which time mortality remained under twenty percent. To initiate lethal secondary infection, surviving flies were then injected with either bacterial suspensions (OD 0.1) or sterile PBS as a control. Secondary injections were done in the thorax, which generates a more lethal infection, although delivery of sterile PBS through this route does not cause significant mortality [2]. Experimental blocks contained 2–3 vials of 20 flies per experimental treatment. All chronic infection-secondary infection pairings were repeated in three independent blocks except for the P. rettgeri-P. rettgeri, P. rettgeri-S. marcescens and E. faecalis-E. faecalis experiments, which were repeated in four independent blocks. After secondary infection, death was recorded daily for one week. Survival curves are plotted as Kaplan-Meier plots using the “survival” package in R. As there was minimal mortality of control flies injected with sterile PBS during the secondary injection, statistical analysis was performed on only flies that received live bacterial infection during the secondary bacterial injection. The impact of chronic infection on survival after secondary infection was determined using Cox Proportional Hazards Mixed Effects Model (coxme) with the package “coxme” in R (http://www.r-project.org/). Each factor is incorporated in sequence and factors are listed in order of inclusion in result tables. Fixed effects included were experimental block (B), infection treatment (T: media control or bacterial infection) and the interaction between B and T. To account for experimental structure, Vial (V) nested inside both B and T was included as a random effect. Our models fulfill the assumption of Cox Proportional Hazards in that the proportional hazards do not cross over-time.

Model D: coxme(status,time) = B + T + (B x T) + (1|B/T/V)

Starvation assay

For each experiment, flies were first injected in the abdomen with bacterial suspensions (OD 0.1) to initiate chronic infection or injected with sterile PBS as control. The flies were housed in groups of 20 flies per vial, during which time mortality stayed under twenty percent. One week after injection, surviving flies (16–20 per vial) were flipped to vials containing 1% agarose and monitored for survival approximately every hour for 35 hours. These vials were kept in storage containers surrounded by damp paper towels to keep the environment humid and ensure that flies did not die from desiccation. Three experimental blocks were completed for both S. marcescens and E. faecalis, while six experimental blocks were completed for P. rettgeri. Survival curves were plotted as Kaplan-Meier plots using Graphpad Prism 7.0 and statistical significance was determined using Cox Proportional Hazards Mixed Effects models (coxme–Model D, described above) with the package “coxme” in R (http://www.r-project.org/). Our models fulfill the assumption of Cox Proportional Hazards in that the proportional hazards do not cross over time.

Results

Infectious dose impacts both survival and the persistent bacterial load

It is a common observation that larger inoculation doses lead to a greater risk of host death [5,21,24,3437]. Perhaps less intuitively, increasing the initial inoculum of P. rettgeri also increases the chronic bacterial load sustained by survivors [5]. To test whether the relationship between infectious dose and chronic burden is general, we determined the bacterial load in flies that survive infection with P. rettgeri, S. marcescens and E. faecalis across a 5000-fold range of initial infectious doses over two weeks post-infection.

As expected, mortality was dependent on infectious dose for all three species (p<0.01, Fig 1, Table 2), with the risk of death dramatically increasing with increasing dose. At the highest infectious dose of approximately 105−106 CFU per fly (A600nm = 5), P. rettgeri infection caused 100% mortality within 2 days, S. marcescens infection caused 100% mortality within 10 days, and E. faecalis infection caused 93% mortality over two weeks (Fig 1). The lowest infectious dose, fewer than 100 bacteria per fly, caused less than 20% mortality after infection with each of the three bacterial species. P. rettgeri infection showed a strong dose-dependency for the risk of death even at the lower infectious doses whereas the other two bacteria elicited very little mortality except at the highest infection dose (Fig 1, Table 2).

Survival of infection decreases as initial dose increases.
Fig. 1. Survival of infection decreases as initial dose increases.
Flies were injected with 23nL of bacterial suspension at the reported absorbance at 600nm and monitored for survival for 2 weeks. Results from experimental blocks were combined for graphical depiction.
Tab. 2. Impact of infectious dose on survival.
Impact of infectious dose on survival.

We measured bacterial load at 12 hours, 24 hours, 48 hours, 72 hours, 7 days, and 14 days post-infection over a 500-fold range of inoculation doses, excluding the highest dose that caused complete and rapid mortality during the survival experiments. Infection with P. rettgeri was characterized by an initial increase in bacterial load after inoculation, but by three days post-infection, after most mortality had occurred, the load in survivors stabilized (Fig 2A). In contrast, there was little proliferation of E. faecalis or S. marcescens after the initial inoculation, which is consistent with the low mortality caused by these infections (Fig 2B and 2C). E. faecalis and S. marcescens chronic loads stabilized just above the level of the initial inoculum (Fig 2B and 2C). Infectious dose was a significant predictor of chronic load at both one and two weeks post-infection for all three bacterial species (p < 0.0001, Table 3), with increased inocula resulting in higher chronic loads sustained. This is consistent with previous work in the lab using these three species [2,5,9]. In two cases, we observed significant interaction terms between block and infectious dose (S. marcescens one-week post-infection, p = 0.03, and E. faecalis two-weeks post-infection, p = 0.0002; Table 3). In both cases, all blocks showed dose-dependence that was consistent in direction but the strength of the dependence was variable across blocks.

Bacterial load over time depends on infectious dose.
Fig. 2. Bacterial load over time depends on infectious dose.
Flies were injected with 23nL of bacterial suspension at the reported absorbance at 600nm and bacterial load assessed at the specified time-points post-injection. Median infectious dose for each absorbance is reported in the legend in parentheses. Results from experimental blocks were combined for graphical depiction.
Tab. 3. Impact of infectious dose on persistent bacterial load.
Impact of infectious dose on persistent bacterial load.

Chronic infection results in sustained immune activation and resistance against secondary infection

The Drosophila immune system is stimulated by bacterial peptidoglycan, with DAP-type peptidoglycan (characteristic of Gram-negative bacteria and Bacillus) more strongly stimulating the Imd pathway and Lys-type peptidoglycan (characteristic of most Gram-positive bacteria) more strongly stimulating the Toll pathway [12,38]. Activation of each pathway leads to the expression of antimicrobial peptide genes, some of which are more elicited dominantly through either Imd or Toll and some of which are co-stimulated by both pathways [12,28]. Measuring the expression level of various antimicrobial peptide genes therefore provides an indication of the degree to which the different arms of the immune system are activated.

We assessed five antimicrobial peptide genes that have diverse immunological activity. First, we assessed diptericin A, which is strongly regulated by the Imd pathway [12] and is the primary determinant of susceptibility to P. rettgeri infection [39,40]. We additionally assayed expression of two other antimicrobial peptide genes (attacin A and metchnikowin) that can be induced by both Toll and Imd signaling and two antimicrobial peptide genes (defensin and drosomycin) that are most strongly regulated by the Toll pathway [12]. These peptides have antimicrobial effects against different types of pathogens [39,4144].

We hypothesized that chronic infection would result in sustained expression of antimicrobial peptides. To test this, we infected flies with P. rettgeri, S. marcescens, and E. faecalis at a dose that is expected to yield a chronic burden of approximately 104 bacteria per fly and measured antimicrobial peptide gene expression by qPCR and GFP fluorescence from reporter constructs. Consistent with our hypothesis, we found that chronic infection with all three infections resulted in significantly upregulated expression of diptericin A, attacin A, metchnikowin, and drosomycin at one-week post-infection, although infection with the Gram-positive pathogen E. faecalis stimulated lower levels of antimicrobial peptide gene expression than infections with P. rettgeri and S. marcescens (Fig 3A). In contrast, defensin was only significantly upregulated by the Gram-negative pathogens, S. marcescens and P. rettgeri (Fig 3A). This can also be shown visually with reporter lines that express green fluorescent protein (GFP) under the control of antimicrobial peptide promoters. These reporter lines generally exhibited comparable mortality to Canton S flies of 10–20%, except for E. faecalis infection of the drosomycin reporter line, which unexpectedly resulted in 80% mortality. GFP fluorescence driven by diptericin A, attacin A, and drosomycin promoters was strong at one week post-infection with S. marcescens (Fig 3B). GFP expression from the Imd-regulated diptericin A and attacin A promoters was moderate at one week post-infection with P. rettgeri but was much weaker from the Toll-regulated drosomycin promoter (Fig 3B), despite the continued significant upregulation of the native drosomycin gene as measured by qPCR (Fig 3A). In contrast, reporter lines infected with E. faecalis showed negligible GFP expression above controls (Fig 3B). These results demonstrate sustained immune activation from chronic infection with all three chronic infections, although a much weaker elicitation from chronic E. faecalis infection.

Antimicrobial peptide gene expression one week post-infection.
Fig. 3. Antimicrobial peptide gene expression one week post-infection.
Flies were injected in the abdomen with 23nL of bacterial suspension (A600nm = 0.1; approximately 2000–6000 CFU) and antimicrobial peptide transcription was assessed at one week post-injection by qPCR (A) and through use of AMP-promoter>GFP constructs (B). (A) Relative expression was calculated using the Pfaffl method and was scaled to media-injected controls for graphical representation. Significantly different expression values across treatment within each antimicrobial peptide transcript were determined by Tukey’s multiple comparison test and are indicated by the capital letters, where treatments that share a letter are not significant with p > 0.01. More detailed results of the statistical analysis are in S1 and S2 Tables. (B) All panels contain 6 flies except the E. faecalis-drosomycin panel, which contains only 2 flies due to high mortality, and images for each reporter construct (diptericin A, attacin A and drosomycin) were taken with the same exposure for each infection condition. The scale bar indicates 2mm, and corresponding brightfield images are shown in S1 Fig.

We next hypothesized that sustained immune activation elicited by chronic infection might result in protection against secondary bacterial infection. To test this, we chronically infected flies with each of the three pathogens and subsequently subjected them to a more intense challenge with either the same pathogen or one of the other two. Expected lethality from the secondary infection was increased by delivering the inoculum into the thorax, which results in increased bacterial proliferation and higher levels of mortality [2]. Overall, chronic infection was protective against mortality from secondary infection, with eight out of nine combinations of chronic and secondary infection showing significant protection (Fig 4; Table 4). More specifically, chronic infection with P. rettgeri provided significant protection against all three secondary infections (P. rettgeri: p<0.0001; S. marcescens: p = 0.001, E. faecalis: p = 0.02). Chronic infection with S. marcescens gave significant protection against all three secondary infections (P. rettgeri: p = 0.01; S. marcescens: p = 0.0006, E. faecalis: p = 0.01). Chronic infection with E. faecalis also gave significant protection against P. rettgeri (p = 0.03) and S. marcescens (p = 0.003). Surprisingly, E. faecalis chronic infection did not give significant protection against secondary infection with E. faecalis (p = 0.38) although the data trend in the direction of protection. For the two combinations that showed a significant block by infection interaction (P. rettgeri-P. rettgeri: p = 0.03, S. marcescens-P. rettgeri: p = 0.003), every block individually showed that flies carrying chronic infection and secondary infection survived better than flies only carrying the secondary infection, although the magnitude of this difference varied among individual blocks.

Chronically infected flies better survive secondary bacterial infections.
Fig. 4. Chronically infected flies better survive secondary bacterial infections.
Flies were injected with 23nL of bacterial suspension (A600nm = 0.1) or sterile media and maintained at 25°C for one week to allow chronic infection to develop. The flies were then given a secondary injection by injecting 23nL of bacterial suspension (A600nm = 0.1) or sterile media into the thorax. Host survival was monitored for one week. Chronic infection for each panel is as follows: P. rettgeri (A, D, G), S. marcescens (B, E, H), and E. faecalis (C, F, I). Secondary infections for each panel is as follows: P. rettgeri (A, B, C), S. marcesens (D, E, F), and E. faecalis (G, H, I). Results from experimental blocks were combined for graphical depiction. The statistical significance of the protective effect is overlaid on each panel with the complete statistics provided in Table 4.
Tab. 4. Effect of chronic infection on survival to a secondary acute infection.
Effect of chronic infection on survival to a secondary acute infection.

Chronic infection with Providencia rettgeri causes decreased starvation resistance

It seems plausible that bearing a substantial chronic infection could be energetically costly, in part due to the costs of sustained immune activation. Time to starvation in the absence of nutrients provides a reliable indication of overall energy state [45]. To assess whether chronic infection puts an energetic strain on the host, we subjected chronically infected flies to intense starvation stress by flipping them to non-nutritious agarose for 35 hours. We monitored the flies’ survival at one hour intervals. Flies infected with P. rettgeri died significantly faster than controls (p<0.0001, Fig 5A, Table 5). While flies with chronic S. marcescens infections appeared to starve marginally more quickly than controls, the effect was not significant (p = 0.41, Fig 5B, Table 5). Due to the significant interaction between block and S. marcescens infection (p = 0.02, Table 5), we examined each block individually and found that increased sensitivity to starvation occurred in only a single block and that the other two blocks showed no effect of chronic infection. Chronic infection with E. faecalis also did not cause any difference in starvation sensitivity (p = 0.65, Fig 5C, Table 5).

Flies chronically infected with <i>P</i>. <i>rettgeri</i> are more susceptible to starvation.
Fig. 5. Flies chronically infected with P. rettgeri are more susceptible to starvation.
Survival of chronically infected flies during starvation stress was compared to uninfected controls by flipping the flies to 1% non-nutritive agarose and monitoring survival for 35 hours. Results from experimental blocks were combined for graphical depiction. The statistical significance of chronic infection’s impact on starvation resistance is overlaid on each panel with the complete statistics provided in Table 5.
Tab. 5. Effect of chronic infection of starvation resistance.
Effect of chronic infection of starvation resistance.

Discussion

In this study, we demonstrate that diverse bacteria are able to establish chronic infections in the D. melanogaster host (Figs 1 and 2), and that the number of chronically persistent bacteria is predicted by the initial inoculation dose (Table 3). Chronic infection results in chronically upregulated immune reactions (Fig 3), and in turn is partially protective against secondary infection (Fig 4, Table 4). This protection is nonspecific and could potentially be explained by sustained production of antimicrobial peptides. We found that chronic infection with P. rettgeri caused host flies to become sensitive to starvation (Fig 5, Table 5), suggesting an energetic cost of bearing that infection. This work lays an initial foundation for studies of the causes and consequences of chronic infection in the experimentally tractable D. melanogaster host.

Diverse bacterial species are able to establish chronic infection in D. melanogaster. These include Providencia rettgeri [1,2,5, present study], Serratia marcescens [9, present study], Escherichia coli [5,9] Pectobacterium (formerly Erwinia) carotovora [5,9], Enterococcus faecalis [2,5,9, present study], Lactococcus lactis [4], and several other species of Providencia (unpublished observations). All of these are opportunistic infectors of Drosophila and other organisms, and they are in some cases very distantly related to each other. It seems unlikely, then, that they share a specialized adaptation for infection of an insect host. We suspect it is much more likely that the chronic infection emerges through a common property of bacterial physiology, possibly adaptive for withstanding harsh abiotic environmental conditions. It is then incidentally triggered in a subset of infected Drosophila leading to a relatively asymptomatic infection that the host is able to tolerate.

We observed here that the chronic bacterial load was correlated with the initial infection dose for P. rettgeri, S. marcescens and E. faecalis (Fig 2), as was previously observed for P. rettgeri [5]. Duneau et al. [5] proposed a model to describe bacterial infection in D. melanogaster. In this model, the bacteria grow logarithmically over the first few hours of infection, during a lag time in which the host immune response is becoming activated (termed tc in that paper). Duneau et al. hypothesized a critical threshold for bacterial density, proposing that if the bacteria reach this threshold prior to sufficient activation of the host immune response, the bacteria would continue to proliferate and would kill the host. However, if the bacteria fail to reach this threshold density, it was hypothesized that the infection would freeze at whatever density had been reached at tc and become chronic. Our data are consistent with this model. Assuming a constant rate of net bacterial proliferation, higher inoculation doses would be expected to yield a higher probability of death because the infection begins closer to the critical threshold, and surviving flies would be expected to bear higher pathogen burdens during the chronic phase. We observe both of these phenomena in our results, lending support for this theoretical model.

We found that chronic infection with all three bacteria stimulates chronic upregulation of D. melanogaster genes encoding antimicrobial peptides (Fig 3), although this immune activation is clearly not sufficient to eliminate the infections (Fig 2). Notably, both of the Gram-negative bacteria–P. rettgeri and S. marcescens—stimulate sustained induction of antimicrobial peptide genes that are regulated by both the Toll and Imd pathways, despite the canonical expectation that Gram-negative bacteria should predominantly stimulate Imd signaling [12,17,46]. Although chronic infection with the Gram-positive E. faecalis stimulates increased expression of four of the five antimicrobial peptide genes tested (Fig 3A), it does not show significant upregulation of defensin, a gene that encodes a peptide with potent antibacterial properties against Gram-positive bacteria that can be elicited by stimulation of the Toll pathway [12,17,46]. This adds to an increasing body of evidence that the Toll and Imd pathways do not dichotomously and simplistically discriminate between Gram-positive and Gram-negative infections [9,4749]. The differences among the three bacteria in the degree of chronic immune activation is not due to differences in the chronic bacterial load, as all three bacteria persist at similar levels (Fig 2), but S. marcescens stimulates a much stronger sustained immune reaction and E. faecalis stimulates a much weaker one (Fig 3). Earlier work looking at the correlation between bacterial load and antimicrobial peptide gene expression in D. melanogaster, Tribolium castaneum, and Bombus terrestris focused on a single bacterial species within each host: Listeria monocytogenes, Bacillus thuringiensis, and Escherichia coli, respectively [1921]. However, our findings show that strength of immune stimulation is highly dependent on bacterial species and suggest that the relationship between bacterial load and immune stimulation may vary across host-bacteria pairings.

We additionally found that chronic infection is generically protective against secondary infection (Fig 4, Table 4) with no evidence of specificity in that protection. The two bacteria that stimulated the most elevated expression of antimicrobial peptide genes relative to control during the chronic phase of infection, P. rettgeri and S. marcescens (Fig 3), provided the most robust protection against secondary infection. Our finding of elevated antimicrobial peptides gene expression suggests that this protection is likely due to increased killing of the secondary infection, but other factors that could influence protection during secondary infection include competition among the chronically and secondarily infecting bacteria for resources and/or increased tolerance of infection by the host. It is also possible that the ten to twenty percent mortality during the first week of the initial infection (Fig 1) selects for the subset of flies that are more robust to secondary infection. However, this is unlikely to account for the entire protective effect we observe. Our results show that chronic infection can increase survival of secondary infection by thirty percent or greater compared to flies without a chronic infection (Fig 4A, 4B & 4D). This suggests that if selection plays a role, it works in combination with other factors.

Our findings support the idea that immune stimulation by one systemic infection may be generally protective against subsequent systemic infections in insects. Work in the caterpillar Manduca sexta demonstrated that infection with non-pathogenic Escherichia coli provided protection against later infection with the pathogenic Photorhabdus luminescens [50]. This is in contrast to other studies in both D. melanogaster and the red flour beetle (Tribolium casteum), which have shown specificity in the protective effect of prior exposure to bacteria [24,51]. This may result from differences in experimental methodology as both of those experiments used heat-killed bacteria to test the specificity of their responses. It is possible that live bacterial infection is be required to elicit generalized protection against future challenges. In the buff-tailed bumblebee (Bombus terrestris) there is also evidence for cross-protection, although protection was stronger for prior sub-lethal infection with the same pathogen than with another bacterial species [52]. In that experiment, the initial infections were cleared and no longer present at the time of secondary infection, which could have reduced the intensity of any sustained immune stimulation relative to what we saw in our experiment, where the initial infection was sustained throughout the secondary infection as well.

Sustaining a long-term infection seems potentially costly because of the energetic demands of sustaining the bacteria, the demands of chronically sustained immune reactions, or both. We used resistance to starvation as an indirect measure of the energetic state of the host [45]. Surprisingly, only chronic infections with P. rettgeri put enough strain on the host to make them significantly more susceptible to starvation (Fig 5). Chronic S. marcescens and E. faecalis infections did not reduce resistance to starvation. Interestingly, chronic infection with P. rettgeri did not significantly impact female fecundity [1] or lifespan [2] in previous studies. Thus, chronic infection appears to have only limited energetic consequence, despite appreciable bacterial loads (Fig 2) and chronic upregulation of the immune response (Fig 3). This finding is somewhat surprising because sustained infections by other bacteria have been shown to deplete energy stores [5355] and there is evidence for a trade-off between immune response and energy storage [1,56]. However, the bacteria causing the most dramatic impact on energy storage in previous studies established intracellular infections that may be more physiologically disruptive or energetically demanding than our current extracellular chronic infections. Additionally, perhaps the level of antimicrobial peptide gene expression we observed does not place a large demand on host energetic resources. Nevertheless, there is a modest cost to sustaining chronic infection with P. rettgeri even under laboratory conditions, and costs in the field may be even more severe.

Overall, this work demonstrates that chronic infection of Drosophila melanogaster is a general phenomenon that can be established by multiple distantly related bacteria, and with varied physiological impact on the host. The chronic infection seems to offer nonspecific protection against secondary bacterial infection, although any benefit from the primary infection needs to be weighed against the risk of death from that initial infection. Chronic infection upregulates immune responses, with the intensity of that upregulation depending more on the identity of the infecting bacterium than it does on the bacterial titer. Chronic infection can also carry energetic cost to the host, although this too varies among pathogens and is not strictly predicted by bacterial titer or magnitude of immune induction. Determining the mechanistic origins of these effects on the host, as well as how the bacteria are able to establish non-lethal chronic infection, should be fruitful areas for future study.

Supporting information

S1 Fig [tif]
Brightfield images for chronically infected GFP reporter flies.

S1 File [zip]
Data for Publication.

S1 Table [docx]
Results from ANOVA analysis for antimicrobial peptide gene expression.

S2 Table [docx]
Results from Tukey HSD post-hoc analysis on antimicrobial peptide gene expression.


Zdroje

1. Howick VM, Lazzaro BP. Genotype and diet shape resistance and tolerance across distinct phases of bacterial infection. BMC Evol Biol. 2014;14: 56. doi: 10.1186/1471-2148-14-56 24655914

2. Chambers MC, Jacobson E, Khalil S, Lazzaro BP. Thorax injury lowers resistance to infection in Drosophila melanogaster. Infect Immun. 2014;82: 4380–4389. doi: 10.1128/IAI.02415-14 25092914

3. Clemmons AW, Lindsay SA, Wasserman SA. An effector peptide family required for Drosophila Toll-mediated immunity. PLOS Pathog. 2015;11: e1004876. doi: 10.1371/journal.ppat.1004876 25915418

4. Kutzer MAM, Armitage SAO. The effect of diet and time after bacterial infection on fecundity, resistance, and tolerance in Drosophila melanogaster. Ecol Evol. 2016;6: 4229–4242. doi: 10.1002/ece3.2185 27386071

5. Duneau D, Ferdy J-B, Revah J, Kondolf H, Ortiz GA, Lazzaro BP, et al. Stochastic variation in the initial phase of bacterial infection predicts the probability of survival in D. melanogaster. eLife. 2017;6: e28298. doi: 10.7554/eLife.28298 29022878

6. Galac MR, Lazzaro BP. Comparative pathology of bacteria in the genus Providencia to a natural host, Drosophila melanogaster. Microbes Infect. 2011;13: 673–683. doi: 10.1016/j.micinf.2011.02.005 21354324

7. Nehme NT, Quintin J, Cho JH, Lee J, Lafarge M-C, Kocks C, et al. Relative roles of the cellular and humoral responses in the Drosophila host defense against three Gram-positive bacterial infections. PLOS ONE. 2011;6: e14743. doi: 10.1371/journal.pone.0014743 21390224

8. Teixeira N, Varahan S, Gorman MJ, Palmer KL, Zaidman-Remy A, Yokohata R, et al. Drosophila host model reveals new Enterococcus faecalis quorum-sensing associated virulence factors. PLoS ONE. 2013;8. doi: 10.1371/journal.pone.0064740 23734216

9. Troha K, Im JH, Revah J, Lazzaro BP, Buchon N. Comparative transcriptomics reveals CrebA as a novel regulator of infection tolerance in D. melanogaster. PLOS Pathog. 2018;14: e1006847. doi: 10.1371/journal.ppat.1006847 29394281

10. Duneau DF, Kondolf HC, Im JH, Ortiz GA, Chow C, Fox MA, et al. The Toll pathway underlies host sexual dimorphism in resistance to both Gram-negative and Gram-positive bacteria in mated Drosophila. BMC Biol. 2017;15. doi: 10.1186/s12915-017-0466-3 29268741

11. Lemaitre B, Reichhart J-M, Hoffmann JA. Drosophila host defense: Differential induction of antimicrobial peptide genes after infection by various classes of microorganisms. Proc Natl Acad Sci. 1997;94: 14614–14619. doi: 10.1073/pnas.94.26.14614 9405661

12. De Gregorio E, Spellman PT, Tzou P, Rubin GM, Lemaitre B. The Toll and Imd pathways are the major regulators of the immune response in Drosophila. EMBO J. 2002;21: 2568–2579. doi: 10.1093/emboj/21.11.2568 12032070

13. Takehana A, Katsuyama T, Yano T, Oshima Y, Takada H, Aigaki T, et al. Overexpression of a pattern-recognition receptor, peptidoglycan-recognition protein-LE, activates imd/relish-mediated antibacterial defense and the prophenoloxidase cascade in Drosophila larvae. Proc Natl Acad Sci. 2002;99: 13705–13710. doi: 10.1073/pnas.212301199 12359879

14. Leulier F, Parquet C, Pili-Floury S, Ryu J-H, Caroff M, Lee W-J, et al. The Drosophila immune system detects bacteria through specific peptidoglycan recognition. Nat Immunol. 2003;4: 478. doi: 10.1038/ni922 12692550

15. Kaneko T, Goldman WE, Mellroth P, Steiner H, Fukase K, Kusumoto S, et al. Monomeric and polymeric Gram-negative peptidoglycan but not purified LPS stimulate the Drosophila IMD pathway. Immunity. 2004;20: 637–649. doi: 10.1016/s1074-7613(04)00104-9 15142531

16. Royet J, Dziarski R. Peptidoglycan recognition proteins: pleiotropic sensors and effectors of antimicrobial defences. Nat Rev Microbiol. 2007;5: 264–277. doi: 10.1038/nrmicro1620 17363965

17. Buchon N, Silverman N, Cherry S. Immunity in Drosophila melanogaster—from microbial recognition to whole-organism physiology. Nat Rev Immunol. 2014;14: 796–810. doi: 10.1038/nri3763 25421701

18. Vaz F, Kounatidis I, Covas G, Parton RM, Harkiolaki M, Davis I, et al. Accessibility to peptidoglycan is important for the recognition of Gram-positive bacteria in Drosophila. Cell Rep. 2019;27: 2480–2492.e6. doi: 10.1016/j.celrep.2019.04.103 31116990

19. Tate AT, Graham AL. Dissecting the contributions of time and microbe density to variation in immune gene expression. Proc R Soc B Biol Sci. 2017;284. doi: 10.1098/rspb.2017.0727 28747473

20. Erler S, Popp M, Lattorff HMG. Dynamics of immune system gene expression upon bacterial challenge and wounding in a social insect (Bombus terrestris). PLOS ONE. 2011;6: e18126. doi: 10.1371/journal.pone.0018126 21479237

21. Louie A, Song KH, Hotson A, Thomas Tate A, Schneider DS. How many parameters does it take to describe disease tolerance? PLoS Biol. 2016;14. doi: 10.1371/journal.pbio.1002435 27088212

22. Haine ER, Moret Y, Siva-Jothy MT, Rolff J. Antimicrobial defense and persistent infection in insects. Science. 2008;322: 1257–1259. doi: 10.1126/science.1165265 19023083

23. Makarova O, Rodriguez-Rojas A, Eravci M, Weise C, Dobson A, Johnston P, et al. Antimicrobial defence and persistent infection in insects revisited. Philos Trans R Soc B Biol Sci. 2016;371. doi: 10.1098/rstb.2015.0296 27160598

24. Pham LN, Dionne MS, Shirasu-Hiza M, Schneider DS. A specific primed immune response in Drosophila is dependent on phagocytes. PLOS Pathog. 2007;3: e26. doi: 10.1371/journal.ppat.0030026 17352533

25. Christofi T, Apidianakis Y. Drosophila immune priming against Pseudomonas aeruginosa is short-lasting and depends on cellular and humoral immunity. F1000Research. 2013;2. doi: 10.12688/f1000research.2-76.v1 24358857

26. Milutinović B, Kurtz J. Immune memory in invertebrates. Semin Immunol. 2016;28: 328–342. doi: 10.1016/j.smim.2016.05.004 27402055

27. Cooper D, Eleftherianos I. Memory and specificity in the insect immune system: current perspectives and future challenges. Front Immunol. 2017;8. doi: 10.3389/fimmu.2017.00539 28536580

28. Tzou P, Ohresser S, Ferrandon D, Capovilla M, Reichhart J-M, Lemaitre B, et al. Tissue-specific inducible expression of antimicrobial peptide genes in Drosophila surface epithelia. Immunity. 2000;13: 737–748. doi: 10.1016/s1074-7613(00)00072-8 11114385

29. Juneja P, Lazzaro BP. Providencia sneebia sp. nov. and Providencia burhodogranariea sp. nov., isolated from wild Drosophila melanogaster. Int J Syst Evol Microbiol. 2009;59: 1108–1111. doi: 10.1099/ijs.0.000117-0 19406801

30. Lazzaro BP, Sceurman BK, Clark AG. Genetic basis of natural variation in D. melanogaster antibacterial immunity. Science. 2004;303: 1873–1876. doi: 10.1126/science.1092447 15031506

31. Khalil S, Jacobson E, Chambers MC, Lazzaro BP. Systemic bacterial infection and immune defense phenotypes in Drosophila melanogaster. J Vis Exp. 2015 [cited 28 Feb 2016]. doi: 10.3791/52613 25992475

32. Schwenke RA, Lazzaro BP. Juvenile hormone suppresses resistance to infection in mated female Drosophila melanogaster. Curr Biol. 2017;27: 596–601. doi: 10.1016/j.cub.2017.01.004 28190728

33. Pfaffl MW. A new mathematical model for relative quantification in real-time RT–PCR. Nucleic Acids Res. 2001;29: e45. doi: 10.1093/nar/29.9.e45 11328886

34. Dionne MS, Ghori N, Schneider DS. Drosophila melanogaster is a genetically tractable model host for Mycobacterium marinum. Infect Immun. 2003;71: 3540–3550. doi: 10.1128/IAI.71.6.3540-3550.2003 12761139

35. Mansfield BE, Dionne MS, Schneider DS, Freitag NE. Exploration of host-pathogen interactions using Listeria monocytogenes and Drosophila melanogaster. Cell Microbiol. 2003;5: 901–911. 14641175

36. Hotson AG, Schneider DS. Drosophila melanogaster natural variation affects growth dynamics of infecting Listeria monocytogenes. G3 GenesGenomesGenetics. 2015;5: 2593–2600. doi: 10.1534/g3.115.022558 26438294

37. Gupta V, Vale PF. Nonlinear disease tolerance curves reveal distinct components of host responses to viral infection. R Soc Open Sci. 2017;4. doi: 10.1098/rsos.170342 28791163

38. Myllymäki H, Valanne S, Rämet M. The Drosophila Imd Signaling Pathway. J Immunol. 2014;192: 3455–3462. doi: 10.4049/jimmunol.1303309 24706930

39. Hanson MA, Dostálová A, Ceroni C, Poidevin M, Kondo S, Lemaitre B. Synergy and remarkable specificity of antimicrobial peptides in vivo using a systematic knockout approach. MacPherson AJ, Garrett WS, Hornef M, Hooper LV, editors. eLife. 2019;8: e44341. doi: 10.7554/eLife.44341 30803481

40. Unckless RL, Howick VM, Lazzaro BP. Convergent balancing selection on an antimicrobial peptide in Drosophila. Curr Biol. 2016;26: 257–262. doi: 10.1016/j.cub.2015.11.063 26776733

41. Dimarcq J-L, Hoffmann D, Meister M, Bulet P, Lanot R, Reichhart J-M, et al. Characterization and transcriptional profiles of a Drosophila gene encoding an insect defensin. Eur J Biochem. 1994;221: 201–209. doi: 10.1111/j.1432-1033.1994.tb18730.x 8168509

42. Fehlbaum P, Bulet P, Michaut L, Lagueux M, Broekaert WF, Hetru C, et al. Insect immunity. Septic injury of Drosophila induces the synthesis of a potent antifungal peptide with sequence homology to plant antifungal peptides. J Biol Chem. 1994;269: 33159–33163. 7806546

43. Levashina EA, Ohresser S, Bulet P, Reichhart JM, Hetru C, Hoffmann JA. Metchnikowin, a novel immune-inducible proline-rich peptide from Drosophila with antibacterial and antifungal properties. Eur J Biochem. 1995;233: 694–700. doi: 10.1111/j.1432-1033.1995.694_2.x 7588819

44. Wang L-N, Yu B, Han G-Q, Chen D-W. Molecular cloning, expression in Escherichia coli of Attacin A gene from Drosophila and detection of biological activity. Mol Biol Rep. 2010;37: 2463–2469. doi: 10.1007/s11033-009-9758-1 19711194

45. Chippindale AK, Chu TJF, Rose MR. Complex trade-offs and the evolution of starvation resistance in Drosophila melanogaster. Evolution. 1996;50: 753–766. doi: 10.1111/j.1558-5646.1996.tb03885.x 28568920

46. Dionne MS, Schneider DS. Models of infectious diseases in the fruit fly Drosophila melanogaster. Dis Model Mech. 2008;1: 43–49. doi: 10.1242/dmm.000307 19048052

47. Engström Y. Induction and regulation of antimicrobial peptides in Drosophila. Dev Comp Immunol. 1999;23: 345–358. doi: 10.1016/s0145-305x(99)00016-6 10426427

48. Stenbak CR, Ryu J-H, Leulier F, Pili-Floury S, Parquet C, Hervé M, et al. Peptidoglycan molecular requirements allowing detection by the Drosophila Immune Deficiency Pathway. J Immunol. 2004;173: 7339–7348. doi: 10.4049/jimmunol.173.12.7339 15585858

49. Hedengren-Olcott M, Olcott MC, Mooney DT, Ekengren S, Geller BL, Taylor BJ. Differential activation of the NF-κB-like factors Relish and Dif in Drosophila melanogaster by fungi and Gram-positive bacteria. J Biol Chem. 2004;279: 21121–21127. doi: 10.1074/jbc.M313856200 14985331

50. Eleftherianos I, Marokhazi J, Millichap PJ, Hodgkinson AJ, Sriboonlert A, ffrench-Constant RH, et al. Prior infection of Manduca sexta with non-pathogenic Escherichia coli elicits immunity to pathogenic Photorhabdus luminescens: Roles of immune-related proteins shown by RNA interference. Insect Biochem Mol Biol. 2006;36: 517–525. doi: 10.1016/j.ibmb.2006.04.001 16731347

51. Roth O, Sadd BM, Schmid-Hempel P, Kurtz J. Strain-specific priming of resistance in the red flour beetle, Tribolium castaneum. Proc R Soc B Biol Sci. 2009;276: 145–151. doi: 10.1098/rspb.2008.1157 18796392

52. Sadd BM, Schmid-Hempel P. Insect immunity shows specificity in protection upon secondary pathogen exposure. Curr Biol. 2006;16: 1206–1210. doi: 10.1016/j.cub.2006.04.047 16782011

53. Chambers MC, Song KH, Schneider DS. Listeria monocytogenes infection causes metabolic shifts in Drosophila melanogaster. PLoS ONE. 2012;7: e50679. doi: 10.1371/journal.pone.0050679 23272066

54. Dionne MS, Pham LN, Shirasu-Hiza M, Schneider DS. Akt and foxo Dysregulation Contribute to Infection-Induced Wasting in Drosophila. Curr Biol. 2006;16: 1977–1985. doi: 10.1016/j.cub.2006.08.052 17055976

55. Paredes JC, Herren JK, Schüpfer F, Lemaitre B. The role of lipid competition for endosymbiont-mediated protection against parasitoid wasps in Drosophila. mBio. 2016;7: e01006–16. doi: 10.1128/mBio.01006-16 27406568

56. Buchanan JL, Meiklejohn CD, Montooth KL. Energetic stress and infection generate immunity-fecundity tradeoffs in Drosophila. bioRxiv. 2018; 318568. doi: 10.1101/318568


Článek vyšel v časopise

PLOS One


2019 Číslo 10