By adding 30% coconut oil into our conventional fly diet (CD), a high-fat diet was used to feed flies for 5 days before the expression of IMD/NFκB activity readout, DptB, was measured. DptB expression in the midgut did not show any difference between flies raised on CD and HFD ( Figure 1A ), suggesting that the local gut immunity was not altered by a HFD. In addition, the abdominal fat body and the Malpighian tubules did not show upregulation of DptB upon HFD either. Interestingly, we found that the fly head, which includes the brain, eyes, head cuticles and head fat bodies that interiorly attach to the cuticle, consistently upregulated DptB expression ( Figure 1A ). DptB is expressed in the head fat body cells as previously shown using a reporter gene (Barajas-Azpeleta et al., 2018), suggesting remote activation of IMD signaling in the fat tissue of the head by HFD. mRNA levels of several additional immune effectors including DptA, AttD, CecA1, Dro, BaraA, Mtk, Drs and IM3 (BomS3) were also measured from the head of flies raided on CD and HFD, and none of them differed significantly between the two diet conditions ( Figure 1B ). Thus, HFD appears to specifically induce DptB but not other effectors only in the head fat body.

Using a Relish null mutant, we found HFD-induced activation of DptB was completely Relish-dependent ( Figure 1C ). To further study if HFD-induced activation of DptB was mediated by the canonical IMD/NFκB pathway, a number of other flies mutant for key components of the IMD pathway were analyzed. In mutants with blocked IMD signal transduction (TAK1, Key and Dredd mutants), HFD-induced DptB upregulation was totally abolished as that in Relish mutant flies ( Figure 1D ). Interestingly, the induction of DptB by HFD required the pattern recognition receptor PGRP-LC that locates to the plasma membrane but not the cytosolic PGN receptor PGRP-LE. Of note, the higher level of DptB expression in the head of PGRP-LE mutant flies raised on CD is consistent with a recent observation that gut-specific depletion of PGRP-LE caused systemic immune activation during gut infection due to the inability to control gut bacteria (Joshi et al., 2023). Taken together, HFD upregulated fat body immunity dependent on the canonical IMD pathway.

To rule out the possibility that HFD causes damages to the gut epithelium and consequently makes a leaky gut, we first measured the expression of three proposed fly cytokines (Unpaired 2 (Upd2), Upd3 and Spaetzle (Spz)) (Woodcock et al., 2015; Yu et al., 2022) in the midgut, the head, the abdominal fat body and the Malpighian tubules of flies raised under CD and HFD, but again failed to detect any difference ( Figures 1E–G ). Furthermore, Smurf assay measuring gut permeability (Rera et al., 2012) also argued against a general defect in the integrity of the intestinal epithelium as we did not detect any positive individuals with pathological gut permeability in both CD and HFD conditions (data not shown). These results reinforced that instead of causing an immune activation or epithelial damages locally in the gut, HFD led to inflammation remotely.

The IMD pathway is activated by sensing bacterial PGN. We therefore hypothesized that changes in the gut microbiota may underline the inflammatory response to the HFD in fly head. To test this idea, we removed the gut bacteria either by feeding flies a cocktail of antibiotics in the adult stage or by generating axenic flies that are not confronted with any microbes from birth onwards (Guo et al., 2014). We first confirmed both methods efficiently eliminated bacteria from flies (data not shown). Then, such microbe-less flies were raised either on CD or on HFD conditions and measured for DptB expression in the head. We found that DptB induction by the HFD was totally repressed using both methods, pointing to a role of gut microflora in causing inflammation in remote tissues ( Figures 2A, B ).

Then we sought to detect changes in the gut microbiota due to HFD. We quantified overall microbial load by measuring colony-forming units (CFUs) in flies raised on CD and HFD, using selective plates to identify Lactobacillae (MRS), Acetobacteriaceae (MFM), and bacteria grown on nutrient-rich medium (LB). The bacterial load increased in flies raised on HFD compared to CD in all the three culturing media ( Figures 2C ). This implies a general increase in bacterial load. To further dissect the microbiota structure, 16S metagenomic analyses were performed using six biological replicates for flies raised on each culturing conditions concerning the potential variability of the microbiome. This analysis uncovered that our HFD significantly altered the composition of the microbial community in comparison with that of flies raised on a CD. Linear discriminant analysis Effect Size (LEfSe) method revealed that Acetobacter was the most enriched genus upon HFD feeding. On the other hand, flies fed a CD exhibited enrichment of species belonging to Faecalibacterium, Escherichia, Bacteroides as well as some uncultured bacteria. To our surprise, while most of the bacterial genera showed similar relative abundance in CD and HFD conditions, Acetobacter species consistently increased their composition in the gut microflora of flies fed with the HFD ( Figures 3A, B ). This made Acetobacter the top bacterium in HFD condition. We further validated the increase of Acetobacter species using quantitative PCR measuring 16S rRNA regions specific for Acetobacter and identified a 15-fold increase in the amount of Acetobacter ( Figures 3C, D ). Thus, Acetobacter strongly increased their abundance in the intestine of flies raised on our HFD.

Laboratory-reared Drosophila melanogaster harbors a simple gut microbiome with only several dominating bacterial species that belong to the Lactobacillus and Acetobacter genera (Broderick et al., 2014). However, we noticed that even in our CD condition, the abundance of Lactobacillus and Acetobacter species in our laboratory were generally very low, potentially suggesting a strong impact of diets and environmental factors on the structure of gut microbiome (Martino et al., 2017).

We next sought to identify the Acetobacter species nourished by the HFD and see if it underlies HFD-induced inflammation in remote tissues. By plating on mannitol agar serial dilutions of homogenized gut of flies raised on HFD, uniform colonies with similar morphological features were seen. Ten representative isolates were cultured and characterized by PCR amplification followed by sequencing of the 16S rRNA gene. We found that they all represented the same bacterial species, with the closest relationship to a previously isolated Acetobacter malorum strain JCM 17274 using BLAST search ( Figure 4A ). The JCM 17274 strain can only be traced back to the following study (Cleenwerck et al., 2002) without detailed information whether it belongs to a gut commensal of flies. We named our strain as Acetobacter malorum Z2311. To test if the Acetobacter malorum Z2311 strain mimicked HFD-induced effects, we mono-associated axenic flies with our Z2311 strain and found this strain produced a strong upregulation of DptB in the head, to a comparable level of Ecc15 oral infection ( Figures 4B, C ). Ecc15 belongs to the phytopathogenic bacterial genus Erwinia (now Pectobacterium) and is a well characterized strain used to induce fly immune response (Buchon et al., 2009b; Zhai et al., 2018a). In addition to local gut immune response, oral Ecc15 infection also activates a systemic immune response that is dependent on the IMD pathway (Basset et al., 2000). To see if oral ingestion of the Z2311 strain resulted in an infection-like condition, flies were fed with bacteria of different concentrations ranging from OD₆₀₀ 0 to 20, and checked for DptB expression in both the gut and the head. While Acetobacter malorum Z2311 had only very mild induction of DptB in the gut across all the concentrations tested, it induced DptB expression for around 70 folds in the head at OD₆₀₀ 20 ( Figure 4D ). This implies that the Acetobacter malorum Z2311 bacteria trigger only very limited local gut immune response thus remain largely unnoticed by the gut epithelium, but instead are competent to induced remote DptB expression in the head.

Gut-bacteria-derived PGN can translocate into the hemolymph that bathes most adult organs and tissues and activate the IMD/NF-κB pathway remotely (Charroux et al., 2018). We wondered if HFD-induced elevation of Acetobacter malorum Z2311 activated immune response in remote tissues by increasing PGN release into the circulation. Silkworm Larvae Plasma (SLP) assay was used to quantify circulating bacterial PGN in the hemolymph (Troha et al., 2019). We found that the amount of circulating bacterial PGN was significantly higher in flies raised on HFD than on CD ( Figure 4E ). Furthermore, mono-association of flies with Acetobacter malorum Z2311 significantly increased circulating bacterial PGN ( Figure 4F ). Further supporting the role of PGN in promoting remote immune activation, flies carrying deletions each removing key amidase PGRPs (PGRP-LB and PGRP-SCs) showed higher DptB expression already under CD, but did not further increase their immune activation under HFD ( Figure 4G ). This is consistent with the notion that the inability to remove PGN locally from the gut and systemically form the circulation renders flies to uncontrolled immune activation. Finally, to better dissect the host-microbe interactions in a nutritional environment, we have measured the growth of Acetobacter malorum Z2311 by directly culturing them in both control and HFD fly media in the absence of Drosophila. Interestingly, the Z2311 strain grew much faster on HFD than on control diet ( Figure 4H ), supporting that Z2311 itself is better adapted to the high-fat environment. It remains interesting to test in future whether this Acetobacter malorum strain also impacted the host adaptation to the HFD condition. In sum, our study supports that HFD-induced changes in the gut microbiome increased the systemic release of PGN that remotely activates immune response.

Fruit flies were cultured in 21°C and under 60% humidity with a 12:12 light dark cycle. Conventional fly diet (CD) contains per liter 32.3g yeast, 69.2g corn flour, 9.2g soybean meal, 61.5mL syrup, 1.7g Nipagin methyl ester and 7.7g Agar. The high-fat diet (HFD) was made by adding into CD 30% coconut oil (w/v). Unless otherwise noted, flies were cultured on CD or HFD for 5 days before they were analyzed. Fly strains used in study are isogenic w¹¹¹⁸ as wild type, Rel^E20 , PGRP-LB^KO , PGRP-SC^KO , PGRP-LE¹¹² , PGRP-LC^E12 , Dredd^B118 , TAK1^D10 , key^c02831 and Dpt^sk1 .

Flies were allowed to lay eggs on apple juice agar plates (containing 2% agar and 50% apple juice) for 4 hours at 25˚C. The eggs were then collected with deionized water using a brush and net baskets. We washed the eggs three times with 70% ethanol, and disinfected them with bleach diluted with deionized water. The eggs were then washed thoroughly with deionized water before being transferred to sterilized food supplemented with an antibiotic cocktail (500μg/mL Ampicillin; 50μg/mL Tetracycline; 200μg/mL Rifamycin) used previously (Jia et al., 2021). To make microbe-less flies starting only from the adulthood, flies were conventionally raised to 3-5 days post eclosion, and shifted to food supplemented with the antibiotic cocktail mentioned above. The axenic state of flies was confirmed by culturing method.

Flies were cultured on CD with blue dye (Sigma, UAS, Cat. #861146) for 1 day after they were treated on CD or HFD for 5 days. Smurf flies that have a leaky gut were recorded as those exhibiting a visible blue color throughout their hemocoel within the body cavity (Rera et al., 2012).

CO2 anesthetized flies were surface cleaned with 70% ethanol and homogenized in group of 15 flies with the Bertin Precellys Evolution Homogenizer. Then, their homogenates were serially diluted and plated on plates made with either Luria-Bertani medium (LB), Man-Rogosa-Sharpe medium (MRS) or Mannitol Ferment Medium (MFM). After the plates were cultured in 29°C for 48 hours, colonies were manually counted and calculated as CFU/fly.

To check bacterial growth rate in the CD and HFD fly media in the absence of Drosophila, 40μL (OD₆₀₀ 0.001) Acetobacter malorum Z2311 was added to sterilized CD and HFD food tubes and cultured at 21°C for 2 days. The tubes were washed with 1mL sterile water that was serially diluted and plated on LB plates. The plates were cultured at 29°C for 48 hours, and then colonies were manually counted and calculated as CFU/tube.

Homogenate of fruit flies fed with HFD was cultured on selective MFM medium for Acetobacter. Multiple single colonies were picked, cultured and sequenced using universal primers (27F: 5’-AGAGTTTGATCCTGGCTCAG-3’ and 1429R: 5’-GGTTACCTTGTTACGACTT-3’) for the bacterial 16S rRNA to confirm their identity. The sequences of our isolated Acetobacter malorum strain (accession number: PRJNA1063874) were Blast searched, and the phylogenetic tree was drawn together with the most closely related Acetobacter species using MEGA11. For re-association experiment, Acetobacter malorum Z2311 was orally fed to flies for 12 hours at concentrations of OD₆₀₀ 0.1, OD₆₀₀ 0.5, OD₆₀₀ 1, OD₆₀₀ 5, OD₆₀₀ 10 and OD₆₀₀ 20. In another experiment, Acetobacter malorum Z2311 and Ecc15 were orally fed to flies for 12 hours at the concentration of OD₆₀₀ 20. The feeding of bacteria was done essentially via soaking into a filter paper fully covering fly food as previously described (Zhai et al., 2018a).

Hemolymph collection and PGN quantification were done essentially according to a recent paper (Chen et al., 2022). Briefly, to collect hemolymph, 100 decapitated female adults were centrifuged at 1500g for 15min at 4°C, followed by a 5 min heating at 70°C. The supernatant was collected and further centrifugated at 12000g for 10 min at 4°C. The extracted hemolymph was 1:10 diluted before PGN quantification. SLP-HS Single Reagent Set II (Fujifilm Wako Pure Chemical Corporation, Japan, Cat. #296-81001) was used to detect PGN in the hemolymph according to the manufacturer instructions.

Whole flies were used for 16S rRNA sequencing. Flies were surface cleaned before being processed further. Bacterial DNA was co-extracted with fly genomic DNA and amplified with primers (343F: 5’-TACGGRAGGCAGCAG-3’; 798R: 5’-AGGGTATCTAATCCT-3’) specific for bacterial 16S rRNA gene (V3/V4 region). Sequencing library preparation and MiSeq (Illumina) high-throughput sequencing were done in oebiotech (Shanghai, China). Sequencing data were processed using standard procedure and raw sequencing data are available under BioProject ID PRJNA1047347.

Total RNA was extracted from the head, the fat body, the gut and the Malpighian tubules of 15-20 flies using RNAiso Plus (TaKaRa, Dalian, China, Cat. #9109). cDNA was synthesized using the PrimeScript RT reagent Kit (TaKaRa, Dalian, China, Cat. #RR037A). 0.2-0.5μg total RNA was used for reverse transcription with oligo (dT), and the 1^st strand cDNA was diluted 10-20 times with water and further used in real time PCR. Real time PCR was performed in at lease duplicate for each sample using LightCycler 480 SYBR Green (Roche, Switzerland, Cat. #04887352001) on a q225 qPCR System from Quantagene (Kubo Technology, Beijing, China). Bacterial DNA contained in the fly gut was extracted using a Bacteria DNA Kit (TIANGEN, China, Cat. #DP302). The expression value was calculated by ΔΔ Ct method, and the relative expression was normalized to RpL32 or GAPDH. Primers used are shown in Table 1 .

Data are shown as means ± SEM from at least three replicates of each experiment. Statistical significance between two groups was calculated with Unpaired Student’s t test to assess differences using GraphPad Prism 9. Statistical significance is presented as *p < 0.05, **p < 0.01, ***p < 0.001.