High-sucrose diets increase mortality from infection with some bacterial pathogens

We first wanted to test whether high-sugar diets reduce immune defense in adult D. melanogaster independent of any developmental consequences. We therefore reared larvae on a yeast-cornmeal diet containing 4% sucrose, then split the population across the experimental diets during the non-feeding pupal stage such that eclosing adults emerged onto experimental diets that were 0%, 2%, 4%, 8%, 16%, or 24% (w/v) sucrose (Fig 1A). This covers the range of dietary sugar used in prior studies [3,12,32]. After 3–5 days on the experimental diets, adult female flies were given a systemic infection with the Gram-negative bacteria Providencia rettgeri and Serratia marcescens, which are known natural pathogens of D. melanogaster [35, 36]. Prior studies have also tested the effect of high-sugar diets on infection outcome using Gram-negative bacteria [2,12].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. High-sugar diet increases mortality after infection with some pathogens.

(A) Experimental setup for assaying effects of high-sugar diets on adult survival of infection. The six experimental diets used have the same yeast and cornmeal content, but they varied in sucrose (w/v). Diets with 0%, 2%, and 4% are represented in shades of blue while diets with 8%, 16%, 24% sucrose are represented in shades of orange. The rearing diet contains 4% sucrose. (B) Flies fed 8%, 16%, and 24% sucrose diets exhibit significantly higher mortality after Providencia rettgeri infection than flies given 0%, 2%, and 4% sucrose diets (p<0.0001; Cox Mixed Effects Model). There was no difference in survival among flies provided with the three high-sugar diets, while flies provided with the 4% sucrose diet exhibited significantly lower survivorship than flies provided with 0% added sucrose (p = 0.035; Cox Mixed Effects Model). (C) Flies fed 16% and 24% sucrose diets exhibit significantly higher mortality after Serratia marcescens than flies given diets with 8% sucrose or lower (p<0.0001; Cox Mixed Effects Model). Letters denote significant pairwise differences between diets (p<0.05). Fig 1A was generated using Biorender.com.

https://doi.org/10.1371/journal.ppat.1012447.g001

We found that flies provided with high-sucrose diets experienced higher mortality after infection with S. marcescens and P. rettgeri than flies fed low sugar (Fig 1B and 1C). This increased mortality occurred in a concentration-dependent manner depending on the pathogen. Flies provided with 0%, 2%, or 4% sucrose experienced low mortality and flies provided with 16% or 24% sucrose experienced high mortality after infection with both pathogens. However, on the 8% sucrose diet, flies infected with S. marcescens exhibited high survival (Fig 1B), but flies infected with P. rettgeri exhibited low survival (Fig 1C).

Gram-negative infections primarily stimulate the IMD pathway [19,21] whereas Gram-positive infections primarily stimulate Toll signaling [19,24]. Hypothesizing that the difference in immunological response might result in a different sensitivity to diet, we also tested whether high-sugar diets impact survival of infection with the Gram-positive bacteria Lactococcus lactis and Enterococcus faecalis, which are also natural pathogens to D. melanogaster [35,36]. We found there was no effect of diet on survivorship of infection with E. faecalis at the initial inoculation dose, although a small effect emerged at a higher infection dose (S1A and S1B Fig). We also found a weak relationship between diet and survival of infection with L. lactis (S1C Fig). Flies fed a 16% sucrose diet were significantly more likely to die than flies fed a 0% sucrose diet, although this increased mortality was not statistically significant compared to flies fed sucrose diets ranging from 2%-16%. At a higher infection dose, a significant difference in survivorship of L. lactis infection was observed between flies provided with 2% and 16% diets (S1D Fig). Flies provided with the 24% sucrose diet experienced the highest mortality compared to all other diets after L. lactis infection (S1C and S1D Fig). We attribute this to the diet being generally poor, as sham-infected control flies fed on the 24% sucrose diet also experienced elevated mortality (S2A Fig). Collectively, our data across all infections indicates that high-sugar diets increase susceptibility to bacterial infection, but that the quantitative increase in mortality depends on the pathogen a host is infected with.

Varying the diet of the host can alter the abundance and composition of microbiota associated with the fly [37–39] and the feeding substrate [40–42], which can yield effects on the host’s metabolic status [43–45]. Flies provided with 16% sucrose in our experiments showed a reduced abundance of microbiota compared to flies provided with 2% sucrose diet (S3A Fig). However, we found that high-sugar diets increased the probability of death from systemic P. rettgeri infection even in axenic flies with no microbiota (S3B Fig), suggesting that our observed effect of diet on immunity is not mediated by effects of associated microbiota.

Flies fed on high-sucrose diets become hyperglycemic despite reduced food intake

D. melanogaster will modulate feeding behavior depending on the content of their diet [46–48], and the nutritional composition of the diet influences metabolic state [3,34,49]. We therefore first assessed whether varying sucrose levels in the diet impacts feeding rate. We performed an excreta quantification assay (ExQ) [50] on uninfected females to measure feeding across the six experimental diets shown in Fig 1A. Flies were housed in ExQ chambers over a 2-day period with experimental diets containing the dye erioglaucine (1% w/v) and their excreta was collected every 24-hours. We found that flies ate significantly more of the three diets with the lowest sucrose (0%, 2%, and 4%) than the three diets with the highest sucrose (8%, 16%, 24%; p < 0.0001, Fig 2A). There was no significant difference in feeding rate among the three lowest-sucrose diets or among the three highest-sucrose diets. In subsequent experiments, we focus on the 2% and 16% sucrose diets as representative low- and high-sugar diets.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Flies fed high sugar sustain higher carbohydrate stores despite reduced feeding.

(A) Excreta quantification of uninfected female flies fed on diets ranging from 0%-24% sucrose was measured over two days. The flies were housed in groups of 8–10 per chamber with n = 48 biological replicates per diet. Flies feeding on the 8%, 16%, and 24% diets consumed significantly lower dye than those fed on the 0%, 2%, 4% sucrose diets (p < 0.001). (B) Uninfected flies fed on the 16% diet (n = 31) had higher glucose levels compared to flies fed on 2% sucrose (n = 31, p < 0.001; ANOVA with Tukey HSD post-hoc). After 24 hours of P. rettgeri infection, flies fed on high sugar (n = 20) continued to exhibit higher glucose levels compared to flies fed 2% sucrose (n = 25, p < 0.01; ANOVA with Tukey HSD post-hoc). However, there was no difference in glucose levels between flies provided with 2% (n = 32) and 16% diets (n = 31) 24-hours after S. marcescens infection (p = 0.658). Flies fed 16% sucrose exhibit significantly reduced glucose levels after S. marcescens infection (p<0.001), and there was a nearly significant decrease in glucose levels of flies fed 2% sucrose after S. marcescens infection (p = 0.0877) (C) Trehalose levels were significantly higher in flies fed 16% sucrose than in flies fed 2% sucrose when the flies were uninfected (p<0.001) or when they were infected with P. rettgeri (p<0.01). However, S. marcescens infection had no effect on trehalose levels between flies given the 2% and 16% diets (p = 0.38). P. rettgeri infection led to higher trehalose levels in flies on both 2% (p = 0.0554) and 16% (p = 0.0446) sucrose diets compared to uninfected flies, while S. marcescens had no effect on trehalose levels on either diet. (D) Glycogen levels were significantly higher in flies on 16% versus 2% diets when flies were uninfected (p = 0.0479), infected with P. rettgeri (p = 0.019), or infected with S. marcescens (p = 0.035). On both diets, infection with both pathogens led to reduced glycogen levels compared to uninfected (2% Providencia, p<0.01; 16% Providencia, p = 0.036; 2% Serratia, p<0.01; 16% Serratia, p < 0.01). Legend for panel figure: * = p<0.05, ** = p<0.01, *** = p <0.001.

https://doi.org/10.1371/journal.ppat.1012447.g002

To assess the effects of diet on the metabolic status of adults, we measured glucose, trehalose, glycogen, and soluble protein in the whole body of uninfected flies provided with either 2% or 16% sucrose diets. To account for any effect of diet on body size, we standardized our measurements to soluble protein levels [51]. Flies fed high sugar had ~10% less soluble protein than flies fed low-sugar diets (S4 Fig). We found that uninfected flies fed the high-sucrose diet exhibit higher levels of glucose (p < 0.001, Fig 2B), trehalose (p = 0.001, Fig 2C), and glycogen (Fig 2D, p = 0.0479) than flies fed the low-sugar diet (Fig 2B–2D). Given that infection can stimulate shifts in carbohydrate levels [26,52], we also tested whether flies fed high-sugar continue to exhibit elevated carbohydrate levels after infection. We focused on infection with S. marcescens and P. rettgeri provided with 2% and 16% diets. After P. rettgeri infection, we found that flies fed the high-sucrose diet continued to exhibit higher levels of glucose (p<0.01, Fig 2B), trehalose (p < 0.01, Fig 2C), and glycogen (p = 0.019, Fig 2C) than flies fed the low-sucrose diet at 24 hours. Interestingly, there was no difference in glucose content (Fig 2B) or trehalose content (Fig 2C) between flies fed low-sucrose or high-sucrose at 24-hours after infection with S. marcescens. However, flies given a high-sucrose diet exhibited higher glycogen stores after S. marcescens infection than flies given low sucrose (p = 0.035, Fig 2D). Despite reduced feeding on high sugar, flies on the high-sucrose diet still sustain higher levels of carbohydrate stores in both uninfected and infected states.

Feeding on high sucrose increases pathogen load in the early stages of infection

High pathogen load is associated with higher mortality from infection [53,54]. Since we observed that high-sugar diets lead to higher mortality during infection with the Gram-negative bacteria S. marcescens and P. rettgeri, we wanted to establish whether this high mortality arises from higher bacterial load. To test whether flies fed on high sugar have higher pathogen loads, we performed an in vivo bacterial growth assay where we sampled infected flies at two-hour intervals from 0–16 hours post-infection and at 24, 36, and 48 hours post-infection. At each time of sampling, individual flies were homogenized and plated on agar plates to estimate the number of live bacterial cells in the fly. Overall, the bacteria proliferated more quickly in flies fed on the high-sugar diet after both P. rettgeri (Fig 3A) and S. marcescens (Fig 3B) infection. However, the divergence in load between the two diets occurs at different times for the two pathogens. S. marcescens infections start to exhibit a higher pathogen burden as soon as 6 hours into the infection, while P. rettgeri burden becomes higher by 12 hours into the infection. Flies that died during the sampling period were not included in this assay, which may mean that pathogen burdens are underestimated if the surviving flies at any given timepoint tend to be the ones with lower pathogen burdens. At 10 hours and later post-infection with S. marcescens, we observed a lower proportion of surviving flies on the high-sugar diet than on the low-sugar diet (Fig 3B), which may explain the apparent absence of difference in pathogen burden 10–16 hours into the infection. We similarly observe a lower proportion of flies provided with high-sugar diet surviving P. rettgeri infection 10 hours and later, although the difference in pathogen burden between flies on the two diets remains significant (Fig 3A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Pathogens proliferate faster in flies fed high sugar.

(A) Pathogen load was assayed at 2-hour intervals from 0–16 hours post-infection and at 24, 36, and 48 hr post-infection in flies fed 2% and 16% sucrose diets. After P. rettgeri infection, flies provided with high-sugar diets have higher pathogen loads at 12, 14, 16, 36, and 48 hr than flies given low-sugar diet (12 hr, p = 0.031; 14 hr, p <0.001, 16 hr, p< 0.001; 36 hr, p < 0.01; 48 hr, p = 0.022; Generalized Least Square Means (GLS) with Tukey HSD post-hoc for pairwise comparisons). Flies given high-sugar diets have consistently lower survival to time of sampling at hour 10 and onward. (B) After S. marcescens infection, flies given high-sugar diets have higher pathogen loads than flies given low-sugar diet at 4 hr, 6 hr, 24 hr, and 36 hr post-infection (4 hr, p = 0.011; 6 hr, p = 0.044, 24 hr, p< 0.001; 36 hr, p < 0.01; Generalized Least Square Means (GLS) with Tukey HSD post-hoc for pairwise comparisons). Flies given high-sugar diets have consistently lower survival to time of sampling at hour 8 and beyond. At 2 hours post-infection, flies given low-sugar diets have higher pathogen load than flies given high-sugar diets (p < 0.01). Each dot represents the pathogen burden of an individual fly. Sample sizes for each timepoint and diet ranged from 8–26 individual flies, depending on how many were still alive at the sampling point. Legend for panel figure: * = p<0.05, ** = p<0.01, *** = p <0.001. The font colors in the table correspond to the proportion of flies that survived at each time point: black = 100%-80%, purple = 79%-50%, red = 50%-0%.

https://doi.org/10.1371/journal.ppat.1012447.g003

Nutrient availability promotes S. marcescens proliferation within hosts on high-sugar diets

Having determined that D. melanogaster provided with high-sucrose diets are both more susceptible to infection and sustain higher internal carbohydrate stores, we next sought to understand whether the increased susceptibility to infection on the high-sugar diet is due to impaired immune system activity or to an altered nutritional environment experienced by the infecting pathogen. To test this, we measured pathogen proliferation during in vivo infection of immune compromised D. melanogaster on the low- and high-sugar diets. We predicted that the effect of diet on sensitivity to infection would be eliminated in immunocompromised flies if the dietary effect is due to impaired immunity, but it would remain if the dietary effect were due to altered nutritional environment for the pathogen. We used the D. melanogaster strain ΔAMP10 [55], which has intact Toll and IMD signaling but is missing 10 genes encoding major infection-inducible antimicrobial peptides. Antimicrobial peptides are small, cationic peptides that are secreted from the insect fat body in response to a systemic infection, and they are critical for killing pathogens by disrupting function of specific microbial processes [56–60]. We can use these flies to monitor in vivo pathogen growth in response to diet without any interference from the humoral immune response. We measured pathogen burden in ΔAMP10 flies at 2-hour intervals post-infection until the last time point prior to the complete death of the infected population, which is 12-hrs post-infection with P. rettgeri and 8-hours post-infection with S. marcescens. When we compare pathogen loads from the wildtype CS flies used in Fig 3A and 3B to the pathogen burdens in the ΔAMP10 mutants in Fig 4A and 4B, we see that the mutants exhibit significantly higher bacterial loads than the CS across all infection and diet conditions (S5A–S5D Fig). At the last time point with surviving flies, the pathogen loads carried in the ΔAMP10 mutants were ~log₂ [20], equal to ~1x10⁶, cells of S. marcescens, and ~log₂ [22], equal to ~4x10⁶, cells of P. rettgeri. These loads are close to the previously reported lethal burden [61].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Nutrient availability promotes Serratia marcescens proliferation within hosts on high-sugar diets.

(A) Diet did not affect the proliferation of P. rettgeri in flies lacking 10 major inducible antimicrobial peptides (p> 0.05 time*diet; GLS model). (B) S. marcescens proliferate significantly more rapidly in ΔAMP10 flies fed high sugar (p<0.001 time*diet; GLS model). Two hours into the infection, flies fed high sugar exhibit a lower Serratia burden (p < 0.01; GLS model followed by Tukey HSD post-hoc), but flies fed high sugar exhibit significantly higher loads by hours by hours 4 (p<0.001) and 6 (p<0.0001; GLS model followed by Tukey HSD-post-hoc). For each time point, approximately 10–28 flies were sampled per diet. (C) There was no effect of sugar supplementation on P. rettgeri growth over 1080 mins in M9 minimal media supplemented with 1% casamino acids (CAS; p> 0.05; ANOVA followed by Tukey HSD post-hoc). CAS supplementation significantly increased OD at 1080 minutes compared to media without casamino acids (p<0.001; ANOVA). (D) S. marcescens grew significantly more in minimal media supplemented with the combination of casamino acids and glucose (p<0.001; ANOVA with Tukey HSD post-hoc) or sucrose (p < 0.01, ANOVA with Tukey HSD post-hoc) than in media supplemented with casamino acids alone. S. marcescens OD at 1080 mins was significantly lower in media supplemented only with glucose (p < 0.001; ANOVA with Tukey HSD). Each curve represents 6–8 replicates per treatment, pooled. Letters in figures denote significant pairwise differences with p <0.05. Legend: G = glucose, S = sucrose, C = casamino acids. Colors for in vitro curves: 1% glucose (light orange), 1% sucrose (light blue), 1% casamino acids (black), or 1% glucose & casamino acids (dark orange), and 1% sucrose & casamino acids (dark blue). (E) Glucose levels declined significantly more after infection with S. marcescens but increased in flies infected with P. rettgeri on both 2% (p < 0.01; ANOVA followed by Tukey HSD post-hoc) and 16% (p< 0.0001; ANOVA followed by Tukey HSD post-hoc) sucrose diets. There was no significant difference in change in glucose levels after P. rettgeri infection between diets (p = 0.24; ANOVA followed by Tukey HSD post-hoc). There was a trend for increased relative loss of glucose after S. marcescens infections in flies fed the 16% sucrose diet compared to 2%, however this was not statistically significant (p = 0.0601, ANOVA followed by Tukey HSD post-hoc). (F) Glycogen levels were reduced after infection with both P. rettgeri and S. marcescens. Flies fed 2% sucrose had significantly higher loss of glycogen stores compared to flies fed 16% sucrose after infection with both P. rettgeri (p = 0.0178, ANOVA with Tukey HSD post-hoc) and S. marcescens (p = 0.0291, ANOVA with Tukey HSD post-hoc). (G) There was no significant difference in change upon infection in trehalose levels between P. rettgeri and S. marcescens on either the 2% diet (P.r., n = 25; S.m., n = 32; p = 0.197) or the 16% sucrose diet (P.r., n = 32; S.m., n = 31; p = 0.215). There was a significant reduction in trehalose levels of flies infected with S. marcescens on 16% sucrose diets compared to flies fed 2% sucrose (p = 0.0372, ANOVA with Tukey HSD post-hoc). There was a trend toward reduced trehalose in flies provided with 16% sucrose diet after P. rettgeri infection, but it was not statistically significant (p = 0.0816, ANOVA with Tukey HSD post-hoc). Legend for panel figure: * = p<0.05, ** = p<0.01, *** = p <0.001.

https://doi.org/10.1371/journal.ppat.1012447.g004

When we compare pathogen growth within the immunocompromised flies across the two diets, there was no difference in the rate of P. rettgeri proliferation in ΔAMP10 hosts provided with either low or high sucrose (Fig 4A), indicating that the typically observed effect of diet on sensitivity to this infection may be a consequence of diet-dependent immune impairment. In contrast, Serratia marcescens continued to proliferate faster in ΔAMP10 flies fed high sucrose than in those fed low sucrose (Fig 4B), indicating an effect of diet beyond direct mediation of the immune system.

As an independent, albeit indirect, test of whether S. marcescens might utilize excess available carbon to grow faster in hosts provided with high-sugar diets, we measured in vitro growth curves of P. rettgeri and S. marcescens in minimal medium supplemented with either a carbon source or a nitrogen source. We performed an 18-hour growth curve assay in M9 minimal media supplemented with different combinations of casamino acids, sucrose, and glucose. We found that supplementation with 1% casamino acids enabled P. rettgeri growth, but there was no further change in the growth trajectory or final OD with supplementation with 1% glucose, 1% sucrose, or no additional sugar (p >0.05, Fig 4C). These data indicate that P. rettgeri is nitrogen-limited but not carbon-limited in minimal media. The addition of either glucose or sucrose dramatically increased the growth of S. marcescens in minimal media and resulted in a significantly higher OD at the end of the assay (p<0.01, Fig 4D), indicating a greater dependence on environmental carbon under these growth conditions. While we do not know precise bioavailability of carbon to bacteria infecting D. melanogaster, the data in Fig 2 show that flies on high-sugar diets have higher levels of circulating and stored sugars, and in combination these data offer conceptual support to the hypothesis that the increased proliferation of S. marcescens within hosts provided with a high-sugar diet may arise in part from increased availability of carbon to the infecting bacteria.

We had observed that flies provided with high-sucrose diets show elevated glucose and trehalose levels after infection with S. marcescens (Fig 2B and 2C) despite reduction in glycogen stores in both diets (Fig 2D), and we hypothesized that S. marcescens might consume excess circulating sugar within the host. We measured the change in carbohydrate levels after infection with either P. rettgeri or S. marcescens relative to uninfected flies on both diets using the data illustrated in Fig 2B–2D. We found that flies infected with S. marcescens have a significantly higher loss in glucose levels compared to flies infected with P. rettgeri on both the low-sugar (p < 0.01) and high-sugar (p<0.001, Fig 4E) diets, although they exhibit similar relative loss in glycogen (p > 0.05, Fig 4F). Interestingly, we observed significantly higher loss of glycogen in flies provided with low-sugar compared to high-sugar diets after both P. rettgeri (p = 0.0178) and S. marcescens infection (p = 0.029, Fig 4F). There was no significant difference between the infections in proportional change in trehalose levels on either diet (Fig 4G). Our observation that both infections stimulate loss in glycogen stores but only S. marcescens infection results in a significant reduction in glucose levels is consistent with S. marcescens consumption of free sugars during infection.

Drosophila on high sugar diets produce less AMP peptide after infection

We inferred that high-sugar diets increase sensitivity to P. rettgeri infection by impairing the immune system because the effect of diet was eliminated in immunocompromised flies. However, the inference that S. marcescens may consume excess carbohydrates present in flies on high-sugar diets does not preclude the possibility that those flies might also have reduced immune responses. To directly test whether high-sugar diets impair the immune response, we first measured mRNA transcripts of a panel of genes encoding six D. melanogaster antimicrobial peptides (AMPs). Gene expression was measured in flies fed low or high-sugar diets at 8-hours after infection with either S. marcescens or P. rettgeri. We found that diet had no effect on the infection-induced expression levels of Diptericin A, Attacin A, Cecropin A1, Drosocin, Drosomycin, or Defensin after either P. rettgeri and S. marcescens infection (Fig 5A and 5B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. High sugar diets reduce production of AMP peptides in response to infection without affecting AMP gene expression.

(A) There was no effect of diet on expression of genes encoding the AMPs Diptericin, Attacin A, Cecropin A1, Drosocin, Drosomycin, and Defensin at 8-hours post-infection with P. rettgeri (2%, n = 6; 16%, n = 6; p>0.05) or (B) S. marcescens (2%, n = 5; 16%, n = 6; p > 0.05). (C) P. rettgeri infection resulted in significantly higher levels of tagged Drosocin than in uninfected flies on both 2% (uninfected, n = 21; p < 0.0001) and 16% (uninfected, n = 18; p = 0.0002) diets. Flies on 16% sucrose diet (n = 21) had significantly lower amounts of tagged Drosocin six hours after P. rettgeri infection than flies on 2% sucrose diet (n = 21; p = 0.008). (D) Tagged Drosocin levels were higher at 6 hours after S. marcescens infection than in uninfected flies on both 2% (uninfected, n = 14; p < 0.001) and 16% (uninfected, n = 25; p < 0.001) diets, but there was no difference between the diets after S. marcescens infection (2%, n = 27; 16%, n = 28; p = 0.71). (E) The probability of detectable Cecropin A1 peptide was similar after P. rettgeri infection on both diets (2%, n = 29; 16%, n = 32; p = 0.45), while flies fed the 2% diet (n = 30) were significantly more likely to have detectable peptide present after S. marcescens infection than those fed the 16% diet (n = 27; p = 0.003). (F) Of the flies with detectable Cecropin A1 levels, there were significantly higher peptide levels in flies fed the 2% diet compared to 16% diet after P. rettgeri (2%, n = 13; 16%, n = 11; p = 0.0229), but it was not significantly different after S. marcescens infection (2%, n = 21;, 16%, n = 8; p = 0.211). Each point represents the amount of either the tagged Drosocin or Cecropin A1 peptide measured in an individual fly.

https://doi.org/10.1371/journal.ppat.1012447.g005

Measuring the amount of mRNA transcript does not directly demonstrate the abundance of translated peptide [62], and translation can be impaired by cellular stress during the D. melanogaster response to infection [63]. To test whether high-sugar diets impact AMP protein levels in response to infection, we used sandwich ELISA to quantify Drosocin and Cecropin A1 from flies that have hemagglutinin (HA) and FLAG epitopes tagged to the endogenous AMPs. Infection-induced production of tagged AMP peptides is detectable as early as 3 hours post-infection (S6A and S6B Fig). We quantified tagged AMPs from individual flies at 6-hours after infection with either P. rettgeri or S. marcescens. Flies fed on the high-sugar diet produced significantly lower levels of tagged Drosocin than flies fed on the low-sugar diet after P. rettgeri infection (p = 0.008, Fig 5C). However, we observed no effect of diet on Drosocin levels in flies infected with S. marcescens (Fig 5D). An unexpectedly high number of flies did not have detectable levels of tagged Cecropin peptide after infection with either P. rettgeri or S. marcescens (S7A and S7B Fig). We performed a hurdle model to statistically test for differences in Cecropin abundance. The hurdle model first tests the probability that detectable levels of tagged Cecropin are present using a binomial logistic regression, and subsequently tests whether there are quantitative differences in peptide levels conditional on having observed a detectable level of peptide. While the probability of detecting Cecropin was equivalent in flies infected with P. rettgeri on both diets (Fig 5E), flies infected with S. marcescens had lower probability of expressing detectable Cecropin when fed the high-sugar diet (p < 0.01, Fig 5E). In the flies for which Cecropin was detectable, peptide abundance was significantly lower in the flies on the high-sugar sucrose diet than in the flies on the low-sugar diet after infection with P. rettgeri (p = 0.023, Fig 5F). Diet had no effect on the abundance of detectable Cecropin after S. marcescens infection (p = 0.21, Fig 5F). Overall, flies provided with 16% sucrose diets produce less Drosocin and Cecropin after infection with P. rettgeri and have a lower probability of detectable Cecropin 6 hours post-infection with S. marcescens. Thus, the translational production of AMPs seems to be impaired in flies on high-sucrose diets even when there is no decrease in transcription level.

Fly stocks and husbandry

All flies were reared and maintained at 25°C on a 12-hour light/dark cycle. The wild-type strain Canton S (CS) were used for all experiments unless otherwise indicated. Mated females were used in all assays because this is the physiologically more common condition as Drosophila can rapidly mate post-eclosion [82], and mating occurs naturally in our experimental setup where males and females are cohoused after eclosion on the experimental diets (see Experimental Diet section). Additionally, mated females are anticipated to be more sensitive to immunological effects of dietary nutrition due to the energetic demands of reproduction [83]. Flies were aged to 3–5 days post-eclosion prior to infection for all experiments. Flies were anesthetized on CO₂ pads for collection and during infection.

Generation of Tagged AMP Flies

Flies carrying tagged Cecropin A1 or Drosocin were generated using CRISPR-facilitated homology-directed repair. In brief, guide sequences for CecA1 or Dro were cloned into pCFD5 (CecA1 guides: guide 1: CCATTGGACAATCGGAAGCT; guide 2: ATAATTATAAATAATCATCG; Dro guides: guide 1: CAAAAACGCAAGCAAGCAGC; guide 2: CAATCAATTGTGACACAATG) [84]. Homology-directed repair cassettes were cloned into pHD-Scarless-dsRed, encoding the replacement of endogenous CecA1 and Dro with peptides that included FLAG and HA tags (predicted sequences of mature peptides: CECA1^HF, GWLDYKDDDDKKIGKKIERVGQHTRDATIQGLGIAQQAANVAATYPYDVPDYAR; DRO^HF, GDYKDDDDKPRPYSPRPTSHPRPIYPYDVPDYARV) [85]. These two plasmids were then injected together into embryos of the genotype y¹ P(nos-cas9, w-) M(RFP-.attP)ZH-2A w*. Male flies emerging from this injection were crossed with females of the genotype w¹¹¹⁸; If / SM6a or w¹¹¹⁸; TM2 / TM6c, Sb¹, as appropriate. Males expressing dsRed in the eye were then selected and crossed to flies carrying CyO, Tub-PBac to enable excision of the 3xP3-dsRed cassette. The final structure of each locus was verified by PCR.

Experimental diets

All flies were reared from egg to pupation in rearing chambers with interchangeable food plates on a standard base sucrose-yeast-cornmeal diet (per liter: 60 g yeast, 60 g cornmeal, 40 g sucrose, 7 g Drosophila Agar, 0.5 mL phosphoric acid, 5 mL propionic acid, and 6 g methylparaben dissolved in 26.5 mL 100% ethanol). D. melanogaster pupated on the chamber walls and, prior to eclosion, the food plates containing the standard rearing media were swapped with one of the experimental diets. Emerging adult flies were aged on the experimental media for 3–5 days post-eclosion prior to infection for all experiments. Experimental diets were identical to the rearing diet aside from the concentration of sucrose. Experimental diets contained either 0%, 2%, 4%, 8%, 16%, and 24% (w/v) of sucrose. Our method of transferring adult flies to the variable-sugar diets at eclosion was chosen for experimental convenience. However, there is carryover of larval fat body cells for up to two days into a newly eclosed adult [86]. In principle, this could affect our experiments because the fat body is metabolically responsive to dietary nutrition and is the primary immune organ. To test whether this methodology has any consequences for our main results, we performed an additional infection survival experiment with adult flies that eclosed on standard rearing diet and were aged for 3 days before being transferred to experimental 2% or 16% sucrose diets, where they remained for an additional 3 days before infection. Flies fed 16% sugar still showed significantly higher mortality after infection with both S. marcescens and P. rettgeri than flies fed low sugar (S8A and S8B Fig).

Infection survival assays

Flies were anesthetized on CO₂ and injected in the thorax with either a bacterial suspension or sterile PBS (as injury control) using a pulled glass capillary needle and nano-injector [87]. The following bacteria were injected in at least four experimental blocks: Providencia rettgeri strain Dmel, Serratia marcescens strain 2698B, Enterococcus faecalis strain Dmel, Lactococcus lactis strain Dmel. All bacteria were originally isolated as natural infections of D. melanogaster [36,88] except S. marcescens 2698B, which is a clinical isolate [89]. Cultures for P. rettgeri and S. marcescens were started from a frozen glycerol stock then cultured overnight in liquid lysogeny broth (LB) at 37°C on a shaker, then subcultured the next morning in fresh LB to achieve growth phase. L. lactis was cultured overnight from a frozen glycerol stock in liquid Brain Heart Infusion (BHI) at 37°C with shaking, then subcultured the next morning in fresh BHI to achieve log-phase. E. faecalis was cultured from a frozen glycerol stock at room temperature (~24°C) in BHI. All bacteria were resuspended in 1x phosphate-buffered saline (PBS), diluted to A₆₀₀ = 0.1, and injected at a volume of 23 nL per fly. There were ~4000 P. rettgeri cells, ~3000 S. marcescens cells, ~1000 cells E. faecalis cells injected, or ~ 1000 L. lactis cells injected per fly in the primary experiment. An additional experiment was performed with injections of ~3000 E. faecalis and L. lactis cells. Flies were housed in groups of ten and each experimental block contained 3–4 vials of flies per experimental diet. Survival curves were assessed using a Cox proportional hazards mixed effect model (coxme) with the “CoxMe” package in R.

Model A: coxme(Surv(status, time) ~ diet + (1|block)

Diet was treated as a main effect and experimental block was treated as a random factor in our model. This model was used to perform all pairwise comparisons using the “emmeans” package in R with a Tukey p-value correction (p<0.05).

Bacterial load trajectory assay

To assay bacterial proliferation during in vivo infection, flies were infected with P. rettgeri or S. marcescens as described above, then collected for measurement of bacterial load at 0, 2, 4, 6, 8, 10, 12, 14, 16, 24, 36, and 48 hours post-infection. Single, live flies were homogenized in 500 μL sterile 1x-PBS with a metal bead and 50 μL of the homogenate was plated onto LB agar using a spiral plater (Whitley Automated Spiral Plater-2). To ensure there was no overgrowth of bacteria on the plates, fly homogenates were diluted with PBS at time points where we anticipated higher bacterial growth. For wild-type CS flies, fly homogenates from timepoints 0, 2, and 4 hours were directly plated, homogenates from timepoints 6–12 were diluted 1:10, and homogenates from timepoints 14–48 were diluted 1:100. For ΔAMP10 mutants, fly homogenates were directly plated for time points 0, 2, and 4, diluted 1:10 at hour 6, and 1:100 at time points 8, 10 and 12. Agar plates were incubated at 37°C overnight. The resulting colonies were counted using PROTOCOL3 software to estimate the number of colony forming units (CFU) per mL of fly homogenate. We then used the CFU/mL value to approximate the number of bacteria in the individual fly at the time of sampling by using the following formula:

CFU/fly = CFU/mL * (dilution factor) * (0.5 mL/fly of original homogenate)

Two experimental blocks were completed for P. rettgeri and S. marcescens in CS flies, and three experimental blocks were completed for ΔAMP10 mutants with each pathogen. A generalized least squares (GLS) model from the “nlme” package in R was used to determine the effect of diet on pathogen loads over time. Fixed effects in the model include time point sampled (T) and diet (D). The function weights = varIdent(form = ~1|time point) from the “nlme” package was incorporated into the model to account for unequal variances across time points, which arises as the pathogen grows from the initial injection time point. The emmeans function was used for pairwise comparisons between the experimental diets within each time point using the Tukey method to determine p-values (p<0.05)

Model B: gls(log₂(pathogen load) ~ T * D, weights = varIdent(form = ~ 1|T))

Excreta Quantification (ExQ) to measure feeding

Methods for the ExQ assay are further detailed in Wu (2020) [50]. Briefly, food was prepared by melting down experimental diets and adding 1% (w/v) of dry erioglaucine powder (Sigma # 861146). The dyed food mixture was then dispensed into microcentrifuge tubes caps filled up to 50 μL volume. ExQ chambers were prepared from 50 mL conical vials (Cell Treat # 229421). Air holes were poked along the sides of tubes and the lid using a pushpin (~0.5 mm diameter). A precision knife was used to cut a hole in the lid of the conical vial to fit the diameter of the food caps. Naïve flies were anesthetized with CO₂ and sorted into ExQ chambers in groups of 8–10. Every 24 hours, the flies were transferred to new chambers with fresh food. After 48 hours, flies were emptied from chambers and 3 mL of 1x-PBS was used to collect the excreta. Total dye concentration was determined by using a spectrophotometer to measure absorbance of the dye excreted from flies. A standard curve was prepared from a stock solution of erioglaucine by dissolving 10 mg of powder dye in 10 mL of PBS. A serial 2-fold dilution was then created from this initial stock, and the final range of standards used to create the standard curve was 312.5 μg/mL to 5 μg/mL. Absorbance was measured at 630 nm in a 96-well plate on a spectrometer (Molecular Biosciences Spectra Max Series Plus 384). Standards and samples were run in duplicates with 100 μL per well and then averaged for data analysis. Dye consumed per fly within a single ExQ chamber was determined using the following equation:

Dye (μg) per fly = [(Absorbance–Slope)]/(Intercept * # flies)

A two-way Analysis of Variance (ANOVA) in R was used to test for differences in dye consumed by flies on each diet on Day 1 and Day 2 of feeding. Then the emmeans function in R was used for pairwise comparisons between diets within each day using the Tukey test to determine p-values (p<0.05).

Model C: aov(dye ~ diet + day)

Quantitative real-time PCR (RT-qPCR) for Antimicrobial Peptide Gene Expression

Female flies provided with 2% and 16% sucrose diets were injected with ~3000 CFU of log-growth S. marcescens or P. rettgeri then sampled at 8 hours after injection for quantification of gene expression. Flies were pooled in groups of 5 and frozen at -80°C for subsequent RNA extraction. In 1.5 mL microcentrifuge tubes, RNA was isolated and purified using a TRIzol extraction procedure according to the manufacturer’s instructions (Zymo Kit direct-Zol RNA kit #R2050). Briefly, flies were homogenized in TRIZoL (Invitrogen #15596018) using a motorized pestle in a microcentrifuge tube and RNA was then extracted using the Zymo Spin Column (Zymo Research #C1078). Samples were then treated with the kit’s DNAse and RNA was dissolved 50 μL of molecular grade water. RNA concentration was determined using a Qubit4 Broad Sense RNA kit (Thermo #Q10210). cDNA was generated using 500 ng of total RNA with the iSynthesis cDNA kit (Bio-Rad #1708891). RT-qPCR was performed on a CFX Connect Real-Time Detection System (Bio-Rad) using PerfeCTa SYBER Green fast mix (VWR #101414–270). Expression of the AMPs Drosocin, Defensin, CecropinA, Dipterin, Attacin A, Drosomycin was normalized to the expression of the housekeeping gene Actin 5C. Uninfected controls were not subject to injury. Primer sequences used are listed in S1 Table. A linear model was used to test the contributions of diet and infection status to expression of each AMP gene, standardized to the housekeeping gene [90].

Model D: lm(AMP_CT ~ Actin5C_CT + Diet + Infection Status * Diet)

Diet and Infection Status were fixed effects in the model. Estimated marginal mean CT values for the Diet x Infection interaction terms were extracted using the emmeans package in R. Then we subtracted the estimated marginal mean CT value for uninfected flies from the estimated marginal means from the infected flies for each diet. This difference in estimated marginal mean CT between uninfected and infected values indicates infection-induced change in AMP gene expression on a log₂ scale. Using the emmeans function, we compared infection-induced expression values between experimental diets with Tukey’s corrections test for p-values (p<0.05).

Sandwich ELISA to quantify tagged AMPs

HA/FLAG tagged AMPs were quantified 6 hr post-infection. The fly line w¹¹¹⁸; Dro^HF/Dro^Hf was used to quantify tagged Drosocin and w¹¹¹⁸; Cec^HF/+ was used to quantify tagged Cecropin. Maxisorp Nunc 96-well ELISA plates (Thermo Scientific 232698) were coated with 2.5 μg/mL of anti-FLAG antibody (Sigma-Aldrich F1804) in coating buffer (0.2 M sodium carbonate/bicarbonate buffer) overnight at 4°C, then the plates were blocked for 2 hours at room temperature with 2% bovine serum albumin (Sigma-Aldrich A2153) prior to use. At the time of sampling, a single fly was homogenized in 200 μL of 0.2% Triton-X buffer and centrifuged at 10,000 RPM for 5 mins to pellet fly debris. Drosocin samples were diluted 1:20 and Cecropin samples were diluted 1:10 so that they fell within linear range of a HA/FLAG peptide standard curve (sequence NH2-DYKDD DDKGG GGGSY PYDVP DYA-NH2 made from AAPTec). The standard curve was prepared by performing a 2-fold serial dilution starting from a stock solution of 2 ng/mL of HA/FLAG peptide to yield 14 standard concentrations. For sample incubation, 50 μL of each standard and sample were added to the ELISA plate and then incubated overnight. Any unused wells were filled with 50 μL of 0.2% Triton-X buffer to serve as blanks. The next day, anti-HA diluted 1:5000 in 0.2% Triton-X was added to each well for another overnight incubation at 4°C. The following morning, 100 μL of room temperature 1-Step Ultra TMB-ELISA Substrate Solution (ThermoFisher #34028) was added to each well and the plate was incubated at 25°C for 30 min. After incubation, 100 μL of 2M sulfuric acid was added to each well to stop the reaction. Plates were then read on a spectrometer at 450 nm to measure absorbance. To determine the amount of peptide per fly, the detected mass of the HA/FLAG peptide was determined from the HA/FLAG standard curve with the following equation:

Detected mass of HA/FLAG (pg) = [(Absorbance-Slope) * Dilution Factor]/ (Intercept * Fly Homogenate)

The synthetic HA-FLAG peptide used to make the standard curve has a different relative molecular mass (Mr) than the Drosocin-tagged and CecropinA1-tagged AMPs produced in the fly (see S2 Table). The detected mass of the HA/FLAG peptide per fly was converted to the mass of the tagged-AMP using the following equation:

Mass of tagged-AMP = (Detected mass of HA/FLAG/ Mr of standard peptide) * Mr of tagged-AMP

To determine whether diet affects Drosocin peptide levels, a linear mixed effects model was applied, with diet and infection status as main factors and experimental block as a random factor. The emmeans function in R was used to compare peptide yield from flies on each diet within infection status.

Model E: lmer(Dro_levels ~ diet * infection + diet + (1|block))

To determine effects of diet on Cecropin peptide levels, a hurdle model was applied. In the first step, a binomial generalized linear model (GLM) was used to test whether diet and infection status (Serratia or Providencia infection versus uninjured) affected the probability of detecting Cecropin peptide in the samples.

Model F: glm(Cec_present ~ diet * infection + diet, family = binomial)

In the second step, a GLM was performed on samples with detectable peptide to test whether diet and infection resulted in quantitative differences in peptide abundance.

Model G: glm(Cec_level ~ diet *infection + diet)

The significance of the contrast between diets within each infection was determined with a post-hoc Tukey test (p<0.05) using the emmeans package in R.

Measurement of nutritional indices

Nutritional indices were measured in naïve flies and in flies 24 hours post-infection with either P. rettgeri or S. marcescens. Flies were homogenized in pools of three in 150 μL of cold lysis buffer (10 mM Tris, 1 mM EDTA, pH 8.0 with 0.1% (v/v) Triton-X-100) with a motorized pestle in 1.5 mL microcentrifuge tubes. Homogenates were centrifuged for 1 minute at 13,000 rpm at 4°C to pellet fly debris. The supernatants were transferred to fresh 1.5ml microfuge tubes, and 10 μL of homogenized sample was removed for protein quantification. The remaining samples were placed in a 72°C water bath for 20 minutes to denature endogenous enzymes. The tubes were stored at -80°C for later use.

Bacteria growth curves in vitro

Serratia marcescens and Providencia rettgeri were measured for in vitro growth using the kinetic cycle in a spectrometer (Molecular Biosciences Spectra Max Series Plus 384) at room temperature. Bacteria were first cultured overnight with shaking in liquid LB at 37°C. The next morning, bacterial cells were resuspended in 1X M9 minimal media (M9) to achieve A₆₀₀ = 1.0, then further diluted to A₆₀₀ = 0.20 in one of six different experimental M9 media conditions. The combinations of M9 media used were 1% glucose, 1% sucrose, 1% casamino acids (CAS), 1% casamino acids and 1% glucose, 1% casamino acids and 1% sucrose, and a M9 blank. CAS was added to the media to provide a nitrogen source while glucose and sucrose provide a carbon source. In a sterile 96-well plate with a clear flat bottom, 100 μL of bacteria resuspended in the experimental media were pipetted into individual wells in 3-fold replicate. Two wells of uninoculated experimental media were included to monitor for contamination. The plate was covered with a lid and inserted into the spectrometer to measure A₆₀₀ at 15-minute intervals over 18 hours at 37°C. We used a linear mixed effect model to test for the main effects of CAS and sugar with block as a random factor to test whether the addition of CAS and/or sugar led to increased bacterial growth, measured as a higher final A₆₀₀. A post-hoc Tukey HSD was performed using the emmeans package in R to test pairwise comparisons (p<0.05) between media containing sugars and CAS.

Model H: lmer(OD600 at 18 hours ~ sugar *CAS + sugar + (1|block))

Preparation of axenic flies

Axenic flies were prepared by embryo bleaching. Conventionally reared (microorganism associated) flies were fed on lab standard yeast cornmeal diet (4% sucrose w/v), and approximately 100–150 flies were placed into an egg laying chamber and allowed to lay eggs on grape juice agar plates. Embryos were collected after 20 hours and removed from agar plates with a clean paint brush into a cell strainer. In a six-well tissue culture plate, embryos were washed within the cell strainer first in 10% bleach twice for 1.5 minutes each, followed by two 30-second washes in 70% ethanol. Lastly, embryos were washed twice for 10 seconds with autoclaved 1X PBS. Embryos were continuously mixed throughout the process to prevent them from sticking to each other or the cell strainer. Once sterilized, embryos were transferred with a sterile brush onto autoclaved yeast-cornmeal rearing media in groups of 50. During pupation, the rearing media was replaced with autoclaved 2% and 16% sucrose diets for adult flies to eclose onto.