PGRP-SC–mediated innate immune memory against pathogenic bacteria in Drosophila

Abstract

Trained immunity is essential for innate immune cells to retain a memory of previously encountered pathogens, strengthening the hosts’ response against these pathogens. However, the mechanisms governing trained immunity have not been well elucidated. In this study, flies of different genotypes were trained with heat-killed gram-negative (G⁻) bacteria and subsequently reinfected with live pathogens. The innate immune responses against reinfection were evaluated through assessments of survival rates, antimicrobial peptide expression levels, and bacterial loads, complemented by transcriptomic and chromatin immunoprecipitation (ChIP) analyses. We found that flies trained with heat-killed gram-negative bacteria exhibited a higher survival rate upon secondary infection compared to unprimed flies, which was associated with increased expression of antimicrobial peptides. Priming with G⁻ bacteria increased the sensitivity of the immune deficiency pathway to a second bacterial infection owing to lower levels of peptidoglycan recognition protein SC (PGRP-SC) after the first infection. The gut was the major tissue involved in the downregulation of PGRP-SC expression. The histone H3 lysine 9 trimethylation (H3K9me3) levels were higher at the PGRP-SC loci in immune-trained flies compared to untrained flies, contributing to the suppression of PGRP-SC expression. PGRP-SC overexpression in the fly gut abolished the effect of trained immunity. Taken together, our studies identify an innate immune memory in Drosophila that is regulated by gut-derived PGRP-SC through H3K9me3-mediated epigenetic repression of the PGRP-SC.

Graphical abstract

Highlights

• •Training with heat-killed gram-negative (G⁻) bacteria has a protective effect against reinfection in Drosophila.
• •Peptidoglycan recognition protein SC (PGRP-SC) expression is decreased in the gut of immune-trained flies.
• •PGRP-SC is downregulated because of increased histone H3 lysine 9 trimethylation (H3K9me3) at the promoter.
• •Overexpression of PGRP-SCs in the gut during training abolishes trained immunity.

INTRODUCTION

Innate immunity is an evolutionarily conserved mechanism that primarily defends organisms against pathogen invasion. Immune memory is considered a characteristic function of the adaptive immune system [1]. However, the absence of adaptive immunity in many organisms raises the question of whether and how memory mechanisms operate to cope with reinfections. In fact, over the past decade, the prevailing dogma that innate immune cells lack memory has been challenged by the discovery of various types of innate immune cells and tissue-resident stem cells that exhibit adaptive-like memory [2].

Mammalian innate immune cells can be trained, and epigenetic modifications may play important roles in this process [2]. In lipopolysaccharide (LPS)-exposed murine macrophages, the induction of antimicrobial effectors can occur more quickly and reach a much greater level upon real pathogen infection. Changes in gene expression are associated with modifications of histone H3 lysine 4 trimethylation (H3K4me3) in the promoter regions of Fpr1 and Ptges [3]. The exposure of monocytes to Candida albicans or β-glucans leads to H3K4me3-based cellular functional reprogramming, resulting in increased cytokine production both in vivo and in vitro in response to subsequent immune stimulation [4]. It has also been reported that Set7-mediated histone H3 lysine 4 monomethylation (H3K4me1) is associated with the trained immunity of macrophages and monocytes [5]. Dendritic cells isolated from immunized mice exhibit an enhanced immune response to pathogen-specific cytokine production, with specific histone modifications detected during this process [6].

Insects possess a robust innate immune system, rendering them among the most abundant animals on earth [7]. In Drosophila, there is no adaptive immunity but rather a simple yet evolutionarily conserved innate immune system, making it an excellent model for studying innate immunity [8,9]. The innate immune response in Drosophila consists of humoral and cellular components. The systemic and local immune responses are regulated primarily by the Toll and immune deficiency (IMD) pathways, which are expressed mainly in the fat body and gut of insects and produce a variety of antimicrobial peptides. The cellular defense mechanisms primarily involve phagocytosis, regulated by scavenger receptors [10,11]. The insect immune system can also be trained or primed [12]. An earlier study in Drosophila reported that trained immunity is mediated by hemocytes, in which Toll signaling is involved [13]. However, a recent study revealed that the IMD pathway is required for gram-negative (G⁻) bacterium-induced trained immunity in Drosophila [14]. Another study reported that the chromatin remodeler Ada2b plays a role in the immune priming process in Drosophila [15]. However, the precise mechanisms of trained immunity in insects remain largely unknown.

Here, we employed a heat-killed G⁻ bacteria training system to investigate the memory basis for trained immunity in Drosophila. Our results revealed that, after pretreatment with the heat-killed G⁻ bacterium Serratia marcescens (S.m.), the flies employed the secretory peptidoglycan recognition protein SC (PGRP-SC), which is a key molecule that sensitize them to future infection. This process occurred mainly in the gut. Training the flies with heat-killed G⁻ bacteria reduced the expression of PGRP-SC family proteins, increasing the competence of the flies for future infection. At the molecular level, the decreased levels of PGRP-SC proteins were associated with an increase in histone H3 lysine 9 trimethylation (H3K9me3) in the promoter region of PGRP-SC genes. Therefore, we have identified a molecular mechanism underlying trained immunity, wherein downregulation of PGRP-SC expression resulted in the establishment of an immune-competent state in Drosophila.

RESULTS

Training with Heat-killed G⁻ Bacteria Triggers a Protective Effect Against Reinfection in Drosophila

To induce possible memory of innate immunity in flies, we treated the adult flies with heat-killed bacteria before a subsequent infection with live bacteria 7 days after the first treatment (Figure 1A). Compared with control flies pretreated with phosphate-buffered saline (PBS), those pretreated with heat-killed S. marcescens (hkSm) or heat-killed Pseudomonas aeruginosa PAO1 (hkPAO1) presented an increased survival rate upon reinfection with the same live pathogens (Figure 1B–1C). To exclude the possibility of persistent overactivation of immune responses after pretreatment, we assessed the activity of the IMD pathway, which is triggered mainly by G⁻ bacteria. We found that the expression levels of the antimicrobial peptides (AMPs), Dpt and Cec, in immune-trained flies returned to basal levels at 3 or 5 days after the injection of hkSm or hkPAO1 (Figure S1A–S1D). To address the question of whether the protective effect is limited to situations where priming and reinfection involve the same pathogen, flies were trained with one of three heat-killed strains, S.m., PAO1, or Erwinia carotovora carotovora 15 (Ecc15), before infection with another strain of the three. Compared with the PBS training control, all the combinations provided significant protection after training (Figure 1D–1F), suggesting that training with one G⁻ bacterial strain may establish a memory against another G⁻ strain.

Figure 1.

Training Drosophila with heat-killed G⁻ bacteria confers a shared protective response against subsequent encounters with G⁻ bacteria

(A) Graphical outline of immune training with heat-killed bacteria and subsequent reinfection with live bacteria in Drosophila. (B) Survival of wild-type flies w¹¹¹⁸ (with or without hkSm training) after infection with S.m. (C) Survival of w¹¹¹⁸ flies (with or without hkPAO1 training) after infection with PAO1. (D) Survival of S.m.-infected w¹¹¹⁸ flies trained with hkSm or hkPAO1. (E) Survival of PAO1-infected w¹¹¹⁸ flies trained with hkSm or hkPAO1. (F) Survival of S.m.-infected w¹¹¹⁸ flies trained with hkSm or hkEcc15. Survival experiments were conducted independently at least three times, and the data were combined and analyzed using the log-rank test to determine statistical significance (^∗P < 0.05; ^∗∗P < 0.01; ^∗∗∗P < 0.001; ^∗∗∗∗P < 0.0001; ns, not significant). Abbreviations: G⁻, gram-negative bacterium; PBS, phosphate-buffered saline; S.m., Serratia marcescens; hkSm, heat-killed S. marcescens; PAO1, Pseudomonas aeruginosa PAO1; hkPAO1, heat-killed P. aeruginosa PAO1; hkEcc15, heat-killed Erwinia carotovora carotovora 15.

Training with Heat-killed G⁻ Bacteria Prepares Flies with a More Potent IMD Response

The expression of AMPs plays a crucial role in regulating the immune response in Drosophila [16]. To examine the IMD pathway activity in heat-killed bacteria-exposed flies after they were reinfected with live bacteria, we assessed the expression of AMPs via reverse transcription-quantitative polymerase chain reaction (RT-qPCR). The immune-trained flies were reinfected with S.m., or PAO1, or with PBS as a control, and subsequently harvested 12 h post-infection (hpi), 24 hpi, and 36 hpi. The results revealed that the expression levels of multiple AMPs, such as Att, Cec, Dpt, and Mtk, but not of Dros, in hkSm-trained flies were significantly greater after S.m. reinfection than they were in unprimed flies, especially in the early infectious stage of reinfection (Figure 2A–2E). Similarly, PAO1 reinfection resulted in higher expression levels of Att, Cec, Dpt, Dros, and Mtk in trained flies than in untrained flies (Figure 2F–2J).

Figure 2.

Heat-killed G⁻ bacteria training enhances IMD pathway activity and improves the antibacterial capacityor bacterial tolerance of Drosophila

(A–E) RT-qPCR was performed to detect the expression levels of Dpt, Att, Cec, Dros, and Mtk in PBS- or hkSm-trained flies after reinfection with S.m. The sampling time points are indicated. The PBS-treated flies following the primary training served as controls. The sample size for each group ranged from 6 to 15. Data are presented as mean ± SEM in panels A–E (^∗P < 0.05; ^∗∗P < 0.01; ^∗∗∗P < 0.001; ns, not significant; Student's t-test). (F–J) RT-qPCR was performed to assess the expression of Dpt, Att, Cec, Dros, and Mtk in PBS- or hkPAO1-trained flies after reinfection with PAO1. Post-reinfection sampling time points are indicated. The control groups received PBS rather than live PAO1 during the second infection. The sample size for each group was 6. Data are presented as mean ± SEM in panels F–J (^∗P < 0.05; ^∗∗P < 0.01; ^∗∗∗P < 0.001; ns, not significant; Mann–Whitney test). (K–N) Bacterial load (K, M) and BLUD (L, N) tests were conducted to evaluate the number of bacteria at the indicated time points after reinfection or upon the death of the infected flies. All the flies were pretreated with PBS, hkSm, or hkPAO1 and subsequently reinfected with the indicated bacteria. In panels K and L, data are presented as mean ± SD (^∗P < 0.05; ns, not significant; Student's t-test). The sample size for each group ranged from 46 to 48 in panel K and 39–69 in panel L. In panels M and N, data are presented as mean ± SD in (^∗P < 0.05; ns, not significant; Mann–Whitney test). The sample size for each group ranged from 62 to 96 in panel M and 18–127 in panel N. Experiments were conducted independently at least three times, and the data were combined and analyzed to determine statistical significance. Abbreviations: G⁻, gram-negative bacterium; IMD, immune deficiency; RT-qPCR, reverse transcription-quantitative polymerase chain reaction; PBS, phosphate-buffered saline; S.m., Serratia marcescens; hkSm, heat-killed S. marcescens; PAO1, Pseudomonas aeruginosa PAO1; hkPAO1, heat-killed P. aeruginosa PAO1; BLUD, bacterial load upon death; SEM, standard error of the mean; SD, standard deviation.

To address why the survival rate was increased in hkSm- or hkPAO1-trained flies, bacterial load and bacterial load upon death (BLUD) analyses were performed in the reinfected flies after pretreatment. The BLUD test measures the maximal bacterial load that a fly can tolerate [17]. Flies trained with hkSm did not exhibit any changes in bacterial loads but presented an increased BLUD value (Figure 2K–2L), indicating that hkSm training conferred tolerance on flies to cope with reinfection. Interestingly, hkPAO1-trained flies presented a decreased bacterial load compared with that of PBS-trained flies upon reinfection. However, no difference was detected in the BLUD test (Figure 2M–2N), suggesting that hkPAO1 training enhanced bacterial clearance efficiency in flies during the second infection. These results suggest that different mechanisms underlie the trained immunity triggered by hkSm and hkPAO1. Taken together, heat-killed G⁻ bacterial training may improve the IMD pathway response of flies upon subsequent reinfection, as indicated by lower bacterial loads or higher BLUDs and higher expression of AMPs in the reinfected flies.

The Trained Immunity Effect in Drosophila is IMD Pathway-dependent

We have shown that S.m. or PAO1 training led to better survival and a better IMD response to G⁻ bacterial infections. Thus, trained immunity likely operates in Drosophila to improve the response of flies to repeated bacterial infections. To address how trained immunity is related to the IMD pathway, we assessed how trained flies react in an IMD-compromised context. The Kenny (Key) protein is an essential component for IMD pathway activation [18]. We set out to examine the effect of key loss-of-function mutations on trained immunity. Using the GeneSwitch (GS) system, we were able to temporally manipulate key RNAi by adding RU486 (mifepristone) to fly food [19]. da-GS is a ubiquitously expressed GS driver [20]. To slow the death rate of flies and facilitate survival analysis, the flies were maintained at 18 °C after reinfection. Strikingly, a trained immunity effect was still detected in da-GS>key^RNAi flies reared without RU486, as an increased survival rate was detected in hkSm-trained flies (Figure 3A). In contrast, da-GS>key^RNAi flies fed RU486 did not display immune training effects, as no significant difference in survival rate was detected (Figure 3B). RT-qPCR revealed significant knockdown of key expression in da-GS>key^RNAi flies fed RU486 (Figure 3C). Therefore, our results suggest that the IMD pathway is required for the trained immunity effect.

Figure 3.

Trained immunity in Drosophila depends on the IMD pathway

(A–B) Survival rate after S.m. infection in PBS-treated or hkSm-trained da-GS>key^RNAi flies with or without RU486 feeding. RU486 was fed to induce the overexpression of target genes. EtOH was control. (C) RT-qPCR was used to detect the knockdown efficiency of key three days after the RU486 feeding (data are presented as the mean ± SEM, ^∗P < 0.05, Student's t-test, n = 3 for each group). All of the infected flies were reared at 18 °C. Survival experiments were conducted independently at least three times, and the data were combined and analyzed using the log-rank test to determine statistical significance (^∗∗P < 0.01; ns, not significant). Abbreviations: IMD, immune deficiency; PBS, phosphate-buffered saline; S.m., Serratia marcescens; hkSm, heat-killed S. marcescens; EtOH, ethanol; RT-qPCR, reverse transcription-quantitative polymerase chain reaction; SEM, standard error of the mean.

The Fat Body and the Gut are Indispensable for Trained Immunity in Drosophila

The fat body, hemocytes, and midgut are major immune organs of Drosophila. The fat body plays a crucial role in the systemic immune response of Drosophila. Hemocytes are responsible for cellular immune responses, and the midgut serves as a local immune organ [9]. To investigate which tissue(s) are essential for trained immunity in Drosophila, several strategies have been employed to dampen immune function in these tissues. Apoptosis can be induced by the overexpression of rpr or hid [21,22]. GS.hmlD-Gal4 (Hml-GS) is a hemocyte-specific RU486-induced GeneSwitch line [23]. We used Hml-GS flies to specifically overexpress rpr and hid in hemocytes for targeted elimination. Trained immunity effects were detected in Hml-GS>rpr&hid flies reared with or without RU486 after hkSm training (Figure 4A–4B). Hml is specifically expressed in hemocytes [24], which allows for the examination of hemocyte ablation efficiency through RT-qPCR. As shown in Figure 4C, the ablation efficiency was greater than 95% in the RU486 group. These results suggest that hemocytes are likely dispensable for establishing trained immunity in our experimental settings.

Figure 4.

Both the fat body and the gut are required for trained immunity

(A‒B) Survival rate after S.m. infection for hkSm-trained Hml-GS>rpr&hid flies fed with or without RU486. RU486 was fed before the training for 3 days to induce the overexpression of target genes to eliminate the hemocytes in adults. EtOH was control. (C) Hml expression levels were measured to evaluate hemocyte ablation efficiency in the indicated flies 3 days after feeding of RU486 (data are presented as mean ± SEM, ^∗∗P < 0.01, Student's t-test, the sample size (n) was 6 for the +EtOH group and 4 for the +RU486 group). (D–E) Survival of hkSm-trained flies without (r4-Gal4/+) or with (r4>key^RNAi) key knockdown in the fat body after infection with S.m. The infected flies were reared at 18 °C. (F–G) Survival tests for hkSm-trained GSG4073>key^RNAi flies that were fed with or without RU486 after reinfection with S.m. Survival experiments were conducted independently at least three times, and the data were combined and analyzed using the log-rank test to determine statistical significance (^∗P < 0.05; ^∗∗P < 0.01; ^∗∗∗∗P < 0.0001; ns, not significant). Abbreviations: PBS, phosphate-buffered saline; S.m., Serratia marcescens; hkSm, heat-killed S. marcescens; EtOH, ethanol; SEM, standard error of the mean.

As mentioned above (Figure 3A–3C), the key gene plays an important role in trained immunity. Thus, we used key RNAi under the control of gut- or fat body-specific Gal4 to deplete tissue-specific immune functions. When we specifically knocked down key expression in the fat body of r4>key^RNAi flies, which were maintained at 18 °C after reinfection, the immune training effect of hkSm was significantly dampened, whereas the control groups of r4-Gal4/+ exhibited normal trained immunity (Figure 4D–4E). Additionally, when the key gene was specifically knocked down in the gut, the trained immunity effect was also inhibited (Figure 4F–4G). Together, our data suggest that the fat body and the gut play crucial roles in establishing trained immunity in Drosophila.

PGRP-SC Expression is Decreased in the Guts of Immune-trained Flies in Association with Increased H3K9me3

To understand how trained immunity is formed after the first treatment with heat-killed G⁻ bacteria, RNA sequencing (RNA-seq) and weighted correlation network analysis (WGCNA) were performed at the whole-fly level. The results revealed a gene module (pink) that was negatively correlated with hkSm training at 168 h at the gene expression level (Figure S2A, Table S1). Gene Ontology (GO) analysis revealed that peptidoglycan metabolic/catabolic genes were enriched in the pink module (Figure S2B, Table S2). In addition, the transcripts per million (TPM) values indicated a downregulation of the expression of peptidoglycan recognition proteins PGRP-SC1a, PGRP-SC1b, and PGRP-SC2 (Figure S2C–S2E, Table S3). RT-qPCR confirmed that hkSm training resulted in significant downregulation of PGRP-SC1a/b and PGRP-SC2 expression. In the trained flies, the levels of PGRP-SC1a/b and PGRP-SC2 significantly decreased by approximately 60% and 55%, respectively, in whole flies compared with the control flies at 168 h (Figure 5A–5B). PGRP-SC1a/b and PGRP-SC2 are expressed mainly in the gut [25]. Therefore, the guts of hkSm-trained flies were dissected and subjected to RT-qPCR. Consistently, we observed significantly downregulated PGRP-SC expression in the gut (Figure 5C–5D).

Figure 5.

hkSm training leads to the downregulation of PGRP-SC family genes and increased H3K9me3 levels in the promoters of PGRP-SC

(A–B) Transcriptional levels of PGPP-SC1a/b and PGRP-SC2 in hkSm-trained flies 168 h after the injection of hkSm. The samples are from whole flies, and the flies injected with PBS were used as controls. In panel A, the sample size (n) was 10 for the PBS group and 11 for the hkSm group. In panel B, each group had a sample size (n) of 13. (C–D) Transcriptional levels of PGPP-SC1a/b and PGRP-SC2 in the guts of hkSm-trained flies 168 h after the injection of hkSm. The flies injected with PBS were used as controls. The sample size (n) was 3 for each group. (E–G) ChIP assays revealed that H3K9me3, but not H3K27m3, levels in the promoters of PGRP-SC were significantly increased upon training with hkSm. The samples used for the ChIP assay were obtained from the guts of hkSm-treated flies at 168 h post-treatment (n = 3, for each group). Data are presented as the mean ± SEM; ^∗P < 0.05; ^∗∗P < 0.01; ns, not significant. Experiments were conducted independently at least three times, and the data were combined and analyzed to determine statistical significance. Abbreviations: hkSm, heat-killed Serratia marcescens; PGRP-SC, peptidoglycan recognition protein SC; H3K9m3, histone H3 lysine 9 trimethylation; PBS, phosphate-buffered saline; H3K27m3, histone H3 lysine 27 trimethylation; ChIP, chromatin immunoprecipitation; SEM, standard error of the mean.

It is well known that epigenetic regulation plays a crucial role in gene expression, and an increasing number of studies suggest that this type of regulation plays an important role in the formation of innate immune memory [2]. Since PGRP-SC proteins are known as negative regulators of the systemic IMD signaling [[25], [26], [27]], we hypothesized that the immune training induces the epigenetic reprogramming of PGRP-SC genes, leading to the downregulation of PGRP-SC, which consequently upregulates the IMD pathway. Given that PGRP-SCs are mainly expressed in the gut [25,26], we dissected the guts of the trained flies and performed chromatin immunoprecipitation (ChIP) experiments. The results revealed that the amount of H3K9me3 was approximately 3–6 times greater in hkSm-trained flies than in PBS-treated flies at 168 h, whereas the histone H3 lysine 27 trimethylation (H3K27me3) level did not notably change (Figure 5E–5G). As H3K27me3 and H3K9me3 are repressive markers of gene expression [28], our data suggest that hkSm training significantly increases H3K9me3 levels in the promoter regions of PGRP-SC family genes and concomitantly decreases PGRP-SC gene expression, which may consequently promote the IMD signaling once a second infection occurs.

Gut-expressed PGRP-SC Regulates Trained Immunity Effects in Drosophila

The results described thus far led us to propose that, in hkSm-treated flies, the downregulation of PGRP-SC expression likely allows the flies to acquire immune competence. If true, we would expect that the overexpression of PGRP-SC proteins during training would abolish the competence of the immune response to a second infection. To test the hypothesis, we overexpressed PGRP-SC1a/b or PGRP-SC2 proteins either ubiquitously or specifically in the guts and then assessed the training effects. The results revealed that the overexpression of PGRP-SC1a/b or PGRP-SC2 ubiquitously in flies with da-GS-Gal4 during hkSm training terminated the trained immunity effects in the RU486 groups, as evidenced by the lack of any significant difference in survival rates between hkSm-trained flies and PBS-trained flies, whereas trained immunity effects were clearly observed in the ethanol control groups (Figure 6A–6D). Moreover, gut-specific overexpression of PGRP-SC1a/b or PGRP-SC2 driven by GSG4073-Gal4 and NP1^ts-Gal4 (NP1-Gal4, Tub-Gal80^ts), during hkSm training, both abolished trained immunity (Figures 6E‒6H and S3A‒S3C). Consistent with the above results, knocking down PGRP-SC in the gut improved the survival of S.m.-infected flies (Figure S4A–S4C). These results indicate that gut PGRP-SC mediates immune memory in flies.

Figure 6.

The overexpression of PGRP-SC suppresses the hkSm training-induced protective effect

(A–D) Survival of PBS-trained or hkSm-trained flies (with or without overexpression of PGRP-SC1a/b or PGRP-SC2 in the whole fly during immune training) after infection with S.m. (E–H) Survival of PBS-trained or hkSm-trained flies (with or without overexpression of PGRP-SC1a/b or PGRP-SC2 in the gut during immune training) after infection with S.m. For panels A–H, feeding with RU486 induced the overexpression of the indicated genes, whereas pretreatment with EtOH was the control. (^∗∗P < 0.01 ^∗∗∗∗P < 0.0001; ns, not significant; log-rank test). (I–L) Expression levels of Att and Dpt in the fat body of GSG>PGRP-SC1a/b or GSG>PGRP-SC2 flies 24 h after the injection with PBS or live S.m. The flies were reared with or without RU486 (data are presented as the mean ± SEM; ^∗P < 0.05, ^∗∗P < 0.01; Student's t-test). In panel I and J, the sample size (n) was 7 for the PBS (EtOH) and PBS (+RU486) groups and 18 for the S.m. (EtOH) and S.m. (+RU486) groups. In panel K and L, the sample size (n) was 3 for the PBS (EtOH) and PBS (+RU486) groups and 13 for the S.m. (EtOH) and S.m. (+RU486) groups. Experiments were conducted independently at least three times, and the data were combined and analyzed to determine statistical significance. Abbreviations: PGRP-SC, peptidoglycan recognition protein SC; S.m., Serratia marcescens; hkSm, heat-killed S. marcescens; PBS, phosphate-buffered saline; EtOH, ethanol; SEM, standard error of the mean.

As antibacterial effectors, AMPs are expressed mainly in fat body cells during systemic infection [29,30], whereas the PGRP-SC secretory proteins are expressed primarily in the gut [25]. We hypothesized that gut-expressed PGRP-SC may regulate IMD pathway activity in the fat body following the establishment of trained immunity. To investigate potential crosstalk between PGRP-SC and the immune response of the fat body, RT-qPCR was performed on the fat body of S.m.-infected GSG4073>PGRP-SC flies. As expected, the overexpression of either PGRP-SC1a/b or PGRP-SC2 in the adult gut resulted in systemic downregulation of Att and Dpt expression in the fat body after infection with S.m., as assessed at 24 hpi (Figure 6I–6L). Together, these results suggest that gut-expressed PGRP-SC suppresses trained immunity and AMP expression in the fat body.

DISCUSSION

The innate immune response serves as the primary defense mechanism against pathogen invasion. It was originally believed that innate immunity lacks immunological memory. However, 97% of organisms in this world lack an adaptive immune system; thus, immunological memory is unlikely to be unique to the adaptive immune system [2]. An increasing number of studies in the last decade have reported that multiple types of innate immune cells have a form of memory of previously encountered pathogens. For example, macrophages can be immune trained upon exposure to LPS or β-glucan, leading to increased expression of genes encoding antimicrobial effector molecules, chemokines, and pro-inflammatory cytokines, which provide additional immune protection to the host [3,31,32]. Monocytes can be immune trained by C. albicans and increase the levels of the inflammation-related cytokines tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) [4]. Similarly, dendritic cells can be immune trained by Cryptococcus neoformans through promoting interferon signaling pathways [6]. However, the mechanisms underlying innate immune memory formation remain to be further investigated. In this study, we found that the treatment of flies with heat-killed S. marcescens resulted in the establishment of an innate immune memory that was dependent on gut cells. Treatment with hkSm downregulated PGRP-SC expression in the fly gut through a mechanism involving increased H3K9me3 epigenetic modifications. The PGRP-SC family encodes amidases that cleave bacterial-derived meso-diaminopimelic acid (DAP)-type peptidoglycan, the ligand responsible for the activation of the IMD pathway [33]. Therefore, downregulation of PGRP-SC expression in the Drosophila gut enhanced the flies’ immune priming, enabling a more effective response upon secondary infection with either the same or a different gram-negative bacterium (Figure 7). Wild flies inhabit a bacterium-rich environment, where they may encounter nonpathogenic or nonlethal-dose pathogenic bacteria through septic wounding or oral intake. In this scenario, their immune system could be trained to provide stronger immune protection against the subsequently encountered pathogens with similar pathogen-associated molecular patterns, which was elucidated in this study. Thus, immune memory would enable wild flies to survive better in a pathogen-filled environment.

Figure 7.

Model for the PGRP-SC mediated innate immune memory in Drosophila

In untrained flies infected with G⁻ bacteria, gut-expressed amidase PGRP-SCs are secreted into the hemolymph, degrading the DAP-PGN, which functions as an immune-stimulatory factor of the IMD signaling pathway. Low-dose PGN induces limited activation of the IMD pathway, leading to low levels of AMPs. In trained flies that encounter a secondary G⁻ bacterial infection, pretreatment-induced high levels of H3K9me3 in the promoter suppress the expression of PGRP-SC. As a result, more DAP-PGN persists in the hemolymph to induce a more potent IMD response, indicated by high levels of AMPs and improved survival. Abbreviations: PGRP-SC, peptidoglycan recognition protein SC; G⁻, gram-negative bacterium; IMD, immune deficiency; DAP, meso-diaminopimelic acid; PGN, peptidoglycan; AMP, antimicrobial peptide; H3K9me3, histone H3 lysine 9 trimethylation.

Both hkSm and hkPAO1 training promoted AMP production. The colony-forming unit (CFU) number of hkPAO1-trained flies was lower than that of control flies, indicating greater bacterial clearance efficiency in hkPAO1-trained flies. However, hkSm-trained flies did not exhibit increased bacterial killing capacity; instead, they presented increased tolerance to secondary S.m. infection (Figure 2K–2N). A similar observation was reported in the Enterococcus faecalis immune training system [34]. Emerging evidence indicates that AMPs possess dual antimicrobial functions: direct bactericidal activity and suppression of bacterial virulence factors [35]. This dual functionality may explain our contrasting observations: hkSm-trained flies demonstrated improved survival despite elevated bacterial loads in the BLUD assays, whereas hkPAO1-trained specimens presented increased survival coupled with more efficient bacterial clearance. Our results highlight the indispensable role of the IMD pathway in mediating the effects of G⁻ bacterium-induced trained immunity in Drosophila.

In our study, we found that the overexpression of PGRP-SC in the gut not only inhibited trained immunity in hkSm-trained flies but also attenuated AMP expression in S.m.-infected flies. Given the predominant intestinal expression of the PGRP-SC1a/b and PGRP-SC2 isoforms [25], our findings position the gut as a crucial immunoregulatory center governing trained immunity in Drosophila. To our surprise, the ablation of hemocytes did not alter the effects of trained immunity in our experimental settings. These findings differ from those of a previous study highlighting the critical involvement of hemocytes in training responses against the G⁺ bacterium Streptococcus pneumoniae [13]. A recent study revealed that the fat body, rather than hemocytes, is responsible for G⁻ bacterium-induced trained immunity [14]. Flies may have evolved distinct strategies to orchestrate trained immunity against different pathogens. In fact, Drosophila have evolved multiple defense mechanisms to combat pathogenic infections, including phagocytosis, AMP production, and melanization cascades. A previous study also indicated that S. pneumoniae fails to elicit robust humoral immune responses in the flies [13]. This observation potentially explains why hemocytes, rather than fat bodies, mediate S. pneumoniae-induced trained immunity in Drosophila.

Deciphering epigenetic modifications is fundamental to understanding the molecular basis of trained immunity. Previous studies have demonstrated that various histone modifications, such as H3K4me3 and H3K27me3, are associated with trained immunity in mammalian innate immune cells [3,4,6]. Our study revealed that hkSm training induces an increase in H3K9me3 modifications, but not H3K27me3 modifications, within the promoter regions of PGRP-SC family genes, leading to a closed chromatin structure and transcriptional repression of these genes. Studies have demonstrated that H3K9me3 is generally regarded as a stable repressive epigenetic mark, playing a critical role in the establishment of constitutive heterochromatin, whereas H3K27me3 represents a more dynamically regulated repressive modification [36,37]. Our findings suggest that Drosophila utilize H3K9me3-mediated epigenetic modifications to establish and maintain stable immune memory in trained flies.

The gut in mammals plays a crucial role not only in local innate immune regulation but also in systemic innate immune regulation. For example, gut-resident immune cells secrete cytokines, such as type I interferon (IFN-I) and colony stimulating factor 3 (CSF3), to increase the survival rate when mice are systemically infected by vesicular stomatitis virus (VSV) or Escherichia coli [38,39]. PGRP-SC, which is expressed mainly in the gut, is a negative regulator of IMD in flies [25,27]. In this study, the overexpression of PGRP-SC1a/b or PGRP-SC2 in the gut eliminated hkSm-induced trained immunity effects, likely by suppressing AMP expression in the fat body.

Our study revealed systemic regulation from the gut to the fat body in trained immunity, mediated by the downregulation of PGRP-SC expression. However, which signal(s) guides the increased trimethylation of H3K9 in the promoter regions of PGRP-SC genes requires further investigation.

MATERIALS AND METHODS

Fly Stocks and Husbandry

Flies were raised on standard cornmeal media (1 L of food containing 32.19 g of yeast, 9 g of agar, 0.726 g of CaCl₂, 31.62 g of sucrose, 63.2 g of glucose, 77.7 g of cornmeal, 2 g of potassium sorbate, and 15 mL of 10% nipagin) with 12 h light:12 h dark cycles at 25 °C and 65% humidity unless otherwise indicated. The following fly stocks were utilized: the wild-type w¹¹¹⁸ (Cat#3605), GSG4073 (Cat#62084), r4-Gal4 (Cat#33832), and UAS-GFP.nls (Cat#4776) flies were obtained from the Bloomington Drosophila Stock Center (BDSC; Indiana, USA). The UAS-key-RNAi stock (Cat#TH03812.N) was obtained from TsingHua Fly Center (THFC; Beijing, China). The UAS-PGRP-SC-RNAi stock (Cat#51238) was obtained from the Vienna Drosophila Resource Center (VDRC; Vienna, Austria). da-GS is kept in our lab. UAS-rpr&hid (P{UAS-rpr. C}, P{UAS-hid}) was a kind gift from Akira Goto at the Institute of Molecular and Cellular Biology (IBMC; Strasbourg, France). NP1^ts-Gal4 (NP1-Gal4, Tub-Gal80^ts) was a kind gift from Dominique Ferrandon at the IBMC. GS.hmlD-Gal4 (Hml-GS) was obtained from Lei Pan's laboratory at the Shanghai Institute of Immunity and Infection (Shanghai, China). To induce the GeneSwitch system, RU486 (mifepristone; Cat#M830038; Macklin, Shanghai, China) was dissolved in ethanol and subsequently added to the food at a final concentration of 150 μg/mL.

Transgenic flies with PGRP-SC1a/b and PGRP-SC2 were constructed as previously described [40]. In brief, the cDNAs of PGRP-SC1a/b and PGRP-SC2 were amplified from the wild-type genome of w¹¹¹⁸ via the primers listed in Table S4, and then cloned and inserted into pUAST-attB vectors. Microinjection was performed by the Guangzhou Drosophila Resource Center (GDRC; Guangdong, China) to integrate the transgenes into the 3rd chromosome of flies (Cat#24749; BDSC, Indiana, USA).

For other specific fly lines utilized in this study: virgin females of r4-Gal4 or GSG4073 were individually crossed with w¹¹¹⁸ males to obtain r4-Gal4/+ and GSG4073/+ offspring; GSG4073 virgin females were mated with UAS-GFP.nls males to generate GSG4073>GFP.nls flies; virgin females of da-GS, r4-Gal4, or GSG4073 were crossed individually with male UAS-key-RNAi flies, yielding da-GS>key^RNAi, r4>key^RNAi and GSG4073>key^RNAi flies; GSG4073 virgins were crossed with UAS-PGRP-SC-RNAi males to get GSG4073>PGRP-SC^RNAi flies; virgin female Hml-GS flies were crossed with male UAS-rpr&hid flies to generate Hml-GS>rpr&hid flies; virgin females of da-GS, GSG4073 or NP1^ts-Gal4 flies were crossed with UAS-PGRP-SC1a/b or UAS-PGRP-SC2 males to generate da-GS>UAS-PGRP-SC1a/b, GSG4073>UAS-PGRP-SC1a/b, NP1^ts>UAS-PGRP-SC1a/b, da-GS>UAS-PGRP-SC2, GSG4073>UAS-PGRP-SC2 and NP1^ts>UAS-PGRP-SC2 strains.

Bacteria, Infection, and Heat-killed Bacteria Immune Training

The E. carotovora carotovora 15 (Ecc15) strain was described previously [41]. P. aeruginosa PAO1 was obtained from Dominique Ferrandon's laboratory at IBMC. S. marcescens (S.m.) (Cat#1.2818) was obtained from the China General Microbiological Culture Collection Centre (CGMCC; Beijing, China). Ecc15 and S.m. were cultured overnight in standard LB media at 30 °C. PAO1 was cultured overnight in brain heart infusion (BHI) medium at 37 °C. Prior to infection, the bacteria were washed three times with 1× PBS and then diluted to the specified optical density at 600 nm (OD₆₀₀) values of 2 (S.m.) and 0.01 (PAO1) unless otherwise specified in the figure legends.

For infection experiments, flies (2–5 vials, each containing 15–25 flies in an individual test) were anesthetized on a CO₂ pad before bacteria were quantitatively injected into the fly thorax using a microinjector (Nanoject III, Drummond Scientific Company, Pennsylvania, USA). The injection volumes were 4.6 nL per fly for S.m. and 9.2 nL per fly for PAO1. These volumes delivered a bacterial dose that elicited a moderate mortality rate, which allowed for easy observation of the trained immunity phenotype. The control group was injected with the corresponding volume of 1× PBS. Unless specifically indicated in the figure legends, the S.m.-infected flies were cultured at 29 °C, and PAO1-infected flies were cultured at 18 °C.

For heat-killed bacterial immune training, the bacteria were washed three times with 1× PBS, subsequently diluted to the specified OD₆₀₀ of 5, and then incubated at 100 °C for 30 min. One- to three-day-old flies were picked at random and injected with 4.6 nL of heat-killed bacteria. The immune-trained flies were cultured at 29 °C for 7 days.

To analyze the survival of S.m.-infected flies with key knockdown (Figures 3A–3B and 4F–4G), adult flies were reared on food with RU486 during both heat-killed bacterial training and subsequent bacterial reinfection. Post-challenge survival was recorded daily for 13 days in Figure 3A–3B and for 10 days in Figure 4F–4G. For the tests in Figure 4A–4C, the adult flies were first reared on RU486-containing food for 3 days to induce hemocyte apoptosis and then transferred to normal food for subsequent assays. The survival analysis of S.m.-infected flies with hemocyte elimination (Figure 4A–4B) was monitored for 7 days. For survival tests of S.m.-infected flies with PGRP-SC overexpression driven by da-GS or GSG4073-Gal4 in Figure 6A–6H, the flies were reared on food with RU486 during training and subsequently transferred to normal food without RU486 after they were reinfected with S.m. Survival was then monitored for 9–12 days. For AMP level assays in S.m.-infected flies overexpressing PGRP-SC (Figure 6I–6L) or survival analyses of S.m.-infected flies with PGRP-SC knockdown (Figure S4A), flies were fed with RU486-containing food for 7 or 9 days, respectively, before S.m. infection and continued on the same diet thereafter. Survival was recorded daily for 10 days post-infection in Figure S4A.

For survival analyses of S.m.-infected flies with PGRP-SC overexpression driven by NP1^ts in Figure S3A–S3C, adult flies were heat-shocked overnight at 29 °C to induce target gene overexpression, followed by immune training with hkSm. The trained flies were maintained at 29 °C for 7 days before the secondary infection with live S.m., then transferred to 25 °C and monitored for survival over 11 days.

Bacterial Load and BLUD Assay

To determine the CFU of S.m.- or PAO1-infected flies, individual flies were harvested at the indicated time points after reinfection and placed into tubes containing 200 μL of precooled PBS. Then, 3 mm sterile zirconia beads were added to the tubes, and the samples were homogenized via vibration at a frequency of 30 Hz for 30 s using a Retsch MM400 machine (Retsch, North Rhine-Westphalia, Germany). Next, 10 μL of homogenate from each sample was subjected to serial dilution at a ratio of 1:10. Finally, 10 μL of the diluted homogenate was inoculated onto standard LB agar plates and incubated overnight at 37 °C for colony growth. Each diluted sample was analyzed in duplicate, and the number of colonies for each sample was counted. For the BLUD assay, flies were promptly collected within 15–30 min of death, and procedures similar to those used in the CFU assay were performed to determine the bacterial load upon death of the flies.

RT-qPCR

To quantify the mRNA levels of the indicated Drosophila genes, approximately seven flies were collected from each sample for homogenization. For homogenization, a sterile zirconia bead (3 mm) was added to a tube with 200 μL of precooled RNAiso solution. The sample was homogenized by vibrating the zirconia bead at a frequency of 30 Hz for 3 min using the Retsch MM400 machine. The total RNA for each treatment was extracted using the RNAiso Plus kit (Takara, Shiga, Japan) according to the manufacturer's instructions. Subsequently, 1 μg of RNA was reverse-transcribed using the PrimeScript RT reagent kit (Takara, Shiga, Japan) according to the manufacturer's instruction. Quantitative PCR analyses were performed using the TB Green premix Ex Taq II kit (Takara, Shiga, Japan) on a Bio-Rad C1000 Touch Thermal Cycle (Bio-Rad, California, USA) according to the manufacturer's instructions. Finally, the expression levels of the target genes were normalized to rp49 (data are presented as ΔCT, 2^{(Ct of rp49 – Ct of target genes)}). The primers used are listed in Table S4.

RNA-seq and Bioinformatics Analyses

The total RNA of PBS- and hkSm-trained w¹¹¹⁸ flies (about 60 male flies for each biological replicate) was extracted by RNAiso Plus at 24, 72, and 168 h post injection, followed by storage at −80 °C until use. Eukaryotic mRNA was enriched with oligo(dT) beads, and rRNA was removed with a Ribo-Zero Magnetic Kit (Epicenter Madison, Wisconsin, USA). Next, the RNA was quantified, reverse-transcribed, and sequenced via the BGISEQ/MGISEQ sequencing platform (MGIseq2000, MGI, Guangdong, China). The procedures for mRNA enrichment, rRNA elimination, cDNA library preparation, and sequencing were conducted by BGI Tech (Shenzhen) Co., Limited (Guangdong, China). The raw sequencing data were processed via Trim Galore software (v0.6.4) to eliminate any remaining adapter sequences in the reads and low-quality reads (q value ≤ 20). The resulting clean data were aligned to the fly genome dm6 and its corresponding gtf file, which was obtained from the ensemble database (http://ensemblgenomes.org/) using STAR software (v2.7.2b) with the STAR function [42]. The following bioinformatics analyses were performed via R (v3.5.3). The gene expression matrix, raw count matrix, or TPM matrix was generated by RSEM software (v1.3.1) [43]. The TPM values were log₂-transformed and subsequently subjected to WGCNA. WGCNA was conducted via the WGCNA package (v1.69) [44]. The parameter for RsquaredCut in the pickSoftThreshold function in the WGCNA package was set to 0.84. GO biological process (BP) analysis was conducted via the clusterProfiler package (v3.10.1) [45]. The genes whose expression was downregulated at 168 h (log(fold change) < 1 and P value < 0.05) were subjected to GO analysis. Differentially expressed genes were identified via the DESeq2 package (v1.22.2) using the raw count matrix generated by RSEM software [46]. Unless otherwise specified, all analyses were performed using default parameters.

ChIP Assay

PBS- or hkSm-trained flies were collected at 168 h for dissection of the guts (approximately 60 male flies for each biological replicate). The dissected guts were subsequently fixed with 1% formaldehyde at 37 °C for 10 min before the addition of glycine buffer (125 nM) for 2 min to stop crosslinking. Later, the fixed guts were washed with ice-cold PBS (with 2× Protease Inhibitor Cocktail; Roche, Basel-City, Switzerland) 2 times and then lysed with lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris, pH 8.1). The genomic DNA was ultrasonicated into 200–500 bp fragments on ice via a CX130 machine (80% power, 10 s on, 5 s off, 100 times; Sonics, Connecticut, USA). Thereafter, the resulting histone‒DNA complexes were diluted 10 times with dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8.1, and 167 mM NaCl). The diluted samples were then immunoprecipitated with anti-H3K27me3 (1:200; Cat#39155; Active Motif, California, USA), anti-H3K9me3 (1:200; Cat#ab8898; Abcam, Cambridgeshire, UK) and control IgG (1:200; Cat#B900610, Proteintech, Hubei, China) for 16 h at 4 °C with agitation. The corresponding 10% volume of the diluted sample, which was used for immunoprecipitation in the previous step, was retained and stored at −80 °C. The resulting antibody-histone-DNA complexes were separated with protein A/G agarose for 2 h at 4 °C. The beads were then sequentially washed once with low-salt immune complex wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8.1, 150 mM NaCl), once with high-salt buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8.1, 500 mM NaCl), once with LiCl (0.25 M LiCl, 1% NP40, 1% deoxycholic acid [sodium salt], 1 mM EDTA, 10 mM Tris, pH 8.1), and twice with TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0). Finally, the histone-DNA complexes were eluted from the beads for 30 min at room temperature with 500 μL of elution buffer (1% SDS, 0.1 M NaHCO₃). Subsequently, the resulting histone-DNA complexes and their corresponding 10% input samples were reversed with 20 μL and 2 μL of 5 M NaCl, respectively, at 65 °C for 4 h, followed by digestion with proteinase K (40 μL/mL, Cat#V900887, Sigma–Aldrich, Missouri, USA) at 55 °C for 1 h. The resulting DNA was then purified and stored at −80 °C for subsequent qPCR. The primers used are listed in Table S4.

Statistical Analysis

Statistical analysis was conducted using GraphPad Prism v8.0 software. In brief, the log-rank test and Gehan-Breslow-Wilcoxon test were used for comparisons of survival curves, and the unpaired t-test and Mann–Whitney U test were used for comparisons between two groups. Multiple comparisons were performed via Brown–Forsythe and Welch ANOVA tests coupled with Benjamini–Yekutieli multiple comparison post hoc analysis for Figure S1. The details of the statistical analysis parameters’ settings and results are provided in Table S5.

AUTHOR CONTRIBUTIONS

Zihao Deng: conceptualization, funding acquisition, methodology, investigation, validation, visualization, writing – original draft. Jiazhen Luo: methodology, investigation, data curation, formal analysis, visualization. Yufu Zhang: methodology, investigation, validation, data curation, formal analysis, visualization. Chen Li: resources, investigation, data curation. Shixin Deng: resources, investigation, data curation. Jiazhi He: resources, investigation, data curation. Zihao He: investigation, data curation. Wenqi Wu: writing – review & editing. Renjie Jiao: conceptualization, project administration, funding acquisition, supervision, writing – original draft, writing – review & editing. Jiyong Liu: conceptualization, project administration, funding acquisition, supervision, writing – original draft, writing – review & editing.

DECLARATION OF GENERATION AI AND AI-ASSISTED TECHNOLOGIES IN THE WRITING PROCESS

During the preparation of this work, the authors only used Doubao and DeepSeek to polish the language to increase readability. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

ETHICS APPROVAL

Ethics approval is not needed for the Drosophila study.

DECLARATION OF COMPETING INTERESTS

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.

Appendix A. Supplementary data

The following are the Supplementary data to this article.

DATA AVAILABILITY

The raw RNA sequencing data used in this study can be accessed in the Genome Sequence Archive of the National Genomics Data Center (https://ngdc.cncb.ac.cn/) with the accession number CRA020678. The datasets of the raw data in this study are provided in Table S6.