Macronutrients modulate resistance to infection and immunity in Drosophila

Immunity and nutrition are two essential modulators of individual fitness However, while the implications of immune function and nutrition on an individual’s lifespan and reproduction are known, the interplay between feeding behaviour, infection, and immune function, remains poorly understood. In this study, we used the fruit fly, Drosophila melanogaster, to investigate how infection through septic injury modulates nutritional intake, and how macronutrient balance affects survival to infection by the pathogenic Gram-positive bacterium Micrococcus luteus. Our results show that infected flies maintain carbohydrate intake, but reduce protein intake, thereby shifting from a protein-to-carbohydrate (P:C) ratio of ~1:4 to ~1:10 relative to non-infected and sham-infected flies. Strikingly, we found that the proportion of flies dying after M. luteus infection was significantly lower when flies were fed a low-P high-C diet, revealing that flies shift their macronutrient intake as means of nutritional self-medication against bacterial infection. This is likely due to the effects of macronutrient balance on the regulation of the constitutive expression of innate immune genes, as a low-P high-C diet was linked to an up-regulation in the expression of key antimicrobial peptides. Together, our results reveal the intricate relationship between macronutrient intake and resistance to infection, and integrate the molecular cross-talk between metabolic and immune pathways into the framework of nutritional immunology.


43
Infection and nutrition are intricately and intimately linked (Kelley and Bendich, 1996;Sheldon 44 and Verhulst, 1996;Samartin and Chandra, 2000;Rolff and Siva-Jothy, 2003;Cunningham-45 Rundles et al., 2005;Bauer et al., 2006;Calder, 2006;Falagas and Kompoti, 2006;Amar et al., 46 2007;Klasing, 2007;Wu et al., 2007;Ayres and Schneider, 2009;Falagas et al., 2009;Lazzaro 47 and Little, 2009;Sorci and Faivre, 2009;Hawley and Altizer, 2010;Ponton et al., 2011a;Schmid-48 Hempel, 2011;Huttunen and Syrjanen, 2013;Ponton et al., 2013;Genoni et al., 2014;Martinez et 49 al., 2014, Vogelweith et al. 2015. Recent studies have allowed a detailed molecular understanding 50 of the cross-regulation between nutrition and immunity, with nutrient sensing pathways being 51 identified as important regulators of innate immunity (Becker et al., 2010;Martin et al., 2012;52 Varma et al., 2014). Immunity can be activated independently to an infection and this regulation 53 can act under conditions of fluctuating nutrient availability. While the underlying mechanisms are 54 far from being fully understood, the relationship between diet, diet-induced metabolic diseases and 55 infections is clearly multi-factorial, with impairments of immune function playing a key role 56 (Martí et al., 2001;Nave et al., 2011). Better understanding the nutritional components that 57 influence immunity and resistance to infection is an important challenge with implications for 58 animal and human health. 59 There is an ongoing debate on the effects of diet on immune responses to infections. Food 60 deprivation, and/or protein shortage has been reported to negatively affect immunity responses and 61 survival after infection (Siva-Jothy and Thompson, 2002;Pletcher, Macdonald, Marguerie et al., 62 2002, Brunner et al., 2014 with infected hosts selecting a protein-biased diet that provided them 63 with a better survival after infection (Lee et al., 2006;Povey et al., 2009;Povey et al., 2014). In 64 Drosophila, while diet restriction has been shown to decrease the capacity of the host to clear the 65 4 infection (i.e., "resistance"), it provided the host with the ability to reduce the damage of the 66 infection on its health, also called "tolerance" (Ayres andSchneider, 2009, 2012). More recently, 67 it has been shown that yeast restriction affects tolerance specifically to one strain of bacterium in 68 a time-dependent manner; however, no effect on resistance was detected (Kutzer and Armitage, 69 2016, see also Miller andCotter, 2017 andHowick andLazzaro, 2014). 70 Finally, a negative effect of protein and/or a positive effect of carbohydrate on resistance 71 have been revealed (Graham et al., 2014;Kay et al., 2014;Mason et al., 2014) with, for instance, nutrient ratio simultaneously, which hinders the ability to specifically measure the effects of food 77 components and/or caloric content on immunity [but see (Cotter et al., 2011)]. There is now 78 evidence that considering the interactive effects of nutrients is essential and offers a more 79 ecologically relevant understanding (Cotter et al., 2011;Simpson and Raubenheimer, 2012;80 Simpson et al., 2015). 81 Here, we explored the nutritional responses of Drosophila melanogaster after bacterial   91 We first hypothesized that infection through septic injury with the pathogen Micrococcus luteus 92 would modulate the nutritional selection of Drosophila melanogaster. Adult flies were offered a 93 choice between two capillaries filled with either a sucrose or a yeast solution, and food intake was 94 measured every two days for six days (Ja et al., 2007) Table 1). Protein consumption was the lowest for flies infected with M. luteus and was the greatest 100 for sham-infected and non-infected flies (Fig. 1). This reduction in protein intake by infected flies 101 resulted in a marked change in the ingested dietary P:C ratio, such that flies infected with M. luteus 102 balanced their diet to a P:C ratio close to 1:9.6 (i.e., 9% protein, Fig. 1) and non-and sham-infected 103 to a P:C ratio of 1:3.8 (i.e., 20% protein) and 1:3.2 (i.e., 25% protein), respectively ( Fig. 1).

104
The level of expression of several immune genes was measured for the different treatments  sham-infected flies compared to non-infected insects, levels of expression of these genes remained 114 more elevated in infected individuals compared to sham-infected and non-infected individuals 115 (Fig. 2).

116
These first results show that when flies are infected with M. luteus, they shift their 117 nutritional choice to a carbohydrate-biased (lower P:C) diet, which is above and beyond the stress 118 of physical injury (i.e. compare sham-infected vs. infected). 121 We then hypothesized that the shift to a low P:C diet observed for infected flies had survival 122 significance. In this second experiment, non-, sham-and M. luteus-infected flies were fed one of 123 three diets (high, medium and low P:C in a no-choice experiment) and survival was followed. As 124 expected, the interaction between the dietary P:C and the treatment significantly influenced 125 survival rate of flies (Cox regression, Treatment X Diet: χ 2 =26.97, df=4, p<0.001, Treatment: 126 χ 2 =66.28, df=2, p<0.001, Diet: χ 2 =606.57, df=2, p<0.001). Survival was reduced on higher P:C 127 diets for the three groups of flies compared to the two other diets (Fig. 3). However, while naïve 128 flies survived in similar proportions on medium and low P:C diets (i.e., 24% and 4% protein) (Log  The P:C ratio influences the constitutive expression of antimicrobials 143 We next investigated the underlying mechanisms mediating the effect of carbohydrate-biased diet 144 on immune state. Given our findings, we hypothesized that a low-protein, high-carbohydrate diet  N=158; 75% mortality, χ 2 =51.345, df=6, p≤0.001, N=153; Fig. 4). Expression level of the genes 156 coding for AMPs was overall negatively associated with dietary P:C and this was observed at the 157 three sampling times, though there is some suggestion of non-linear trends in the earlier sampling  Table 6).

162
When we looked in more details at the effect of dietary P:C on the expression level of the 163 specific genes, we found a significant negative non-linear relationship between the level of 164 expression and the percentage of dietary protein for six out of nine genes coding for antimicrobial 165 peptides (Fig. 5), which reveals that antimicrobial peptide expression is tightly linked with the 166 macronutrient balance in the diet. This diet-dependent effect on antimicrobial peptide expression 167 was consistent throughout the flies' lifespan (see Supplementary Fig. 3).

168
Interestingly, we found that the effect of dietary P:C can vary depending on the sampling 169 point and the specific gene. For example, for the pattern recognition proteins, gene expression was 170 positively associated with P:C for PGRPSC2, GNBP1 (at 25% mortality only for both genes) and 171 PGRPLC (at 50% mortality only), whereas there was a negative association for PGRPSA (at 25% 172 and 75% mortality) and PGRPSB1 (at all sampling points) (see Supplementary Fig. 3). Expression 173 of genes coding for proteins involved in the immune-signal transduction (i.e., Dif, Imd, Relish,

327
Target gene expression levels were normalized using four reference genes (i.e., Ef1α100E, 328 αTub84B, RpL32 and 18SrNA, see Supplementary Table 3). All samples were run on an ABI 329 model 7900HT sequence detection system according to the protocol supplied by the manufacturer.

330
Results were summarized using the 2 -∆∆Ct method. We log transformed the response variable 331 before making statistical inferences, although all plots are of the raw data.

332
The effect of the percentage of dietary P:C was then tested for each gene and time point 333 individually using generalized additive models (GAMs) that allowed for no a priori decision for