Introduction

Drosophila blood cells, called hemocytes, are functionally and developmentally analogous to vertebrate myeloid cells¹. Their conserved yet simple characteristics make hemocytes an attractive invertebrate model for studying hematopoiesis and innate immunity². Drosophila hemocytes were initially categorized based on their morphologies³ and subsequent studies refined these classifications using marker gene expressions and their biological functions⁴. Most recently, single-cell RNA sequencing (scRNA-seq) analysis allowed a detailed classification of Drosophila larval hemocytes, and several studies have provided in-depth insights into the heterogeneity of Drosophila hemocyte populations at a transcriptome level^5–9. Three classic types of mature hemocytes have been identified in Drosophila larvae: plasmatocytes, crystal cells, and lamellocytes². Plasmatocytes represent up to 95 % of larval hemocytes and function in phagocytosis, antimicrobial peptide expression, and tissue remodeling, similar to macrophages in vertebrates^10,11. Crystal cells facilitate wound healing and blood clotting and account for 2 ∼ 5 % of embryonic and larval hemocytes^12,13. Recently, crystal cells have been shown to regulate internal oxygen homeostasis, in collaboration with the trachea¹⁴. Lamellocytes are a rarely found in healthy animals but arise during immune or stress responses, where they play a role in encapsulation and melanization^15–17.

Similar to vertebrates, Drosophila develops hemocytes by two independent lineages: the embryonic lineage, which originates from the embryonic head mesoderm¹⁰, and the lymph gland lineage, which arises from the cardiogenic mesoderm^11,18,19. Embryonically derived hemocytes, known as embryonic hemocytes, proliferate and differentiate primarily into plasmatocytes and migrate throughout the embryo to participate in embryogenesis^10,20,21. During larval stages, these hemocytes either circulate in the hemolymph or colonize tissue microenvironments, where they continue to proliferate and differentiate^22,23. The second wave of Drosophila hematopoiesis occurs in the lymph gland, which contains hematopoietic progenitors and their microenvironment niche. Lymph gland hemocytes eventually differentiate and disintegrate at the onset of pupariation^{18,19,24–26}. Hemocytes from both lineages persist through the pupal and adult stages to continue to provide immune surveillance²⁷.

An effective immune response demands significant energy, which necessitates enhanced nutritional uptake, metabolic reprogramming of immune cells, and redistribution of body resources. Thus, coordination between metabolism and immunity is critical for effective immune responses and host survival, particularly during infection, across the animal kingdom. In Drosophila, the fat body serves as the primary organ for nutrient storage, metabolism, and humoral immunity. Upon immune activation, the fat body allocates stored energy to immune effectors, including hemocytes, to support the acute phase of immune responses^28–30. It also regulates the number, location, and activation of hemocytes in response to systemic metabolic states^31–33, thereby controlling overall metabolic homeostasis during immunity. In addition to the fat body, muscles play a key role by modulating whole-animal insulin and JAK/STAT pathways and by storing glycogen required for robust innate immune responses^34–36. Recent findings have revealed that hemocytes are not merely passive recipients of energy but also act as active regulators, establishing bi-directional communication with other tissues to adjust energy distribution through multiple signaling pathways^{30,32,37–39}. This is supported by recent single-cell transcriptome analyses^5–7,9, which showed that larval hemocyte subtypes express high levels of metabolism-related genes, implying the significance of hemocytes in bridging metabolism and immunity.

Despite the growing evidence for metabolic shifts in hemocytes under various developmental or stress conditions, and the significance of hemocytes in the crosstalk between immunity and metabolism, their metabolic profiles under steady-state or immune-activated conditions remain poorly characterized. In this study, we measured metabolic parameters of Drosophila larval hemocytes at various developmental timepoints, genotypes, both with and without immune activation. We identified that mitochondrial respiration drives metabolic activation in larval hemocytes. Hemocytes are metabolically dormant under homeostatic conditions but switch to an active state through mitochondrial activation. Furthermore, we demonstrated that lamellocytes exhibit unique metabolic activities, including enhanced trehalose catabolism and mitochondrial remodeling, required for their encapsulation response. Together, our study demonstrates the mitochondria-dependent metabolic activities of Drosophila larval hemocytes and their functional relevance in innate immunity.

Result

Hemocytes rely on mitochondrial respiration

To understand the metabolic characteristics of embryonically derived hemocytes in larvae, we measured metabolic parameters, including oxygen consumption rate (OCR) and extracellular acidification rate (ECAR), of larval hemocytes using the Seahorse XFe96 Analyzer⁴⁰. For the measurement of hemocytes, we adopted protocols from previous studies^41–43 and further optimized the process by exchanging the culture medium and filtering debris using a 40 μm cell strainer (see Methods). The validity of our modified protocol was verified by probing glycolytic signatures of larval brain as previously shown (Supplementary Figure 1a)⁴¹. To identify an optimal density of hemocytes in the measurement, we tested different numbers of live hemocytes – 0.43 x 10⁶ cells/mL, 1.13 x 10⁶ cells/mL, and 1.45 x 10⁶ cells/mL – that represent 0.5x, 1x, and 1.5x ratio, and observed proportional increments of OCR or ECAR values in accordance with the absolute number of hemocytes (Figure 1a-b). Based on the correlation, we set the number of hemocytes per one well to 2.7 x 10⁵ cells representing the optimal range of OCR or ECAR for each measurement without melanization. In the absence of drug application, hemocytes maintained constant OCR or ECAR up to 90-minute culture in the Seahorse analyzer, indicating that hemocytes are viable during the metabolic measurements within this time point (Supplementary Figure 1b-c).

Figure 1.

To monitor the mitochondrial activity of hemocytes in the late third instar larvae, we introduced a series of oxidative phosphorylation (OXPHOS) complex inhibitors: oligomycin, rotenone, and antimycin A to hemocytes. Oligomycin, a potent ATP synthase (complex V) inhibitor, impairs the electron flow of the electron transport chain, reducing mitochondrial respiration or OCR⁴⁴. Rotenone or antimycin A is a complex I or complex III inhibitor, respectively, and a mixture of the two compounds shuts down mitochondrial respiration and allows us to measure the non-mitochondrial OCR^45,46. For measuring glycolysis, 2-deoxy-glucose (2-DG), a glucose analog, was used to competitively block hexokinase and its downstream metabolism⁴⁷ (Figure 1c). First, we carried out the glycolytic rate assay that accounts for compensatory glycolysis following mitochondrial inhibition⁴⁸. Proton efflux in live cells is generated by two main pathways: glycolysis- and mitochondrial-derived acidification. Thus, if tissue is glycolytic, ECAR is significantly elevated after Rot/AA treatment due to compensatory activation of glycolytic proton production (Figure 1d). However, hemocytes considerably reduced ECAR by Rot/AA injection and did not respond to 2-DG (Figure 1e). Also, Rot/AA decreased OCR similar to the trend shown in ECAR (Supplementary Figure 1d), suggesting that OCR and ECAR in larval hemocytes are largely independent of glycolysis. Second, we performed the glycolysis stress test that provides key parameters of glycolytic function, including glycolytic capacity, glycolytic reserve, and non-glycolytic acidification. Most cells are able to switch between glycolysis and oxidative phosphorylation and glucose in cells is converted into pyruvate to be utilized in lactate production in the cytoplasm or TCA cycle. Therefore, glucose injection in a saturated concentration enhances ECAR through lactose dehydrogenase (Ldh)-mediated acidification of cells, which is further elevated by oligomycin treatment due to the lack of OXPHOS (Figure 1f). Hemocytes marginally induced ECAR upon glucose administration, which was suppressed by oligomycin but not by a following 2-DG injection (Figure 1g). Third, we substantiated these results by real-time ATP rate assay designed to measure total ATP production rates and quantitative bioenergetic balance in cells. Consistent with the glycolytic rate assay and glycolysis test, real-time ATP rate assay demonstrated that larval hemocytes generate ATP mainly by mitochondrial activities (mitoATP, 80.1 %) with a minor contribution of glycolysis (glycoATP, 19.9 %) (Figure 1h; Supplementary Figure 1e-f). In addition to hemocytes from w¹¹¹⁸ wild-type animals used for the above experiments, we ran real-time ATP rate assay with hemocyte-derived Drosophila S2 cells and hemocytes from different genetic backgrounds, including Oregon R and Hml^Δ-gal4 UAS-EGFP, and obtained similar mitochondria-dependent ATP production rates (Figure 1h; Supplementary Figure 1g-k). Finally, mito stress test was used to further investigate a key parameter of mitochondrial functions. Oligomycin decreases electron flow through the electron transport chain (ETC) and FCCP lifts limits of the ECT maximizing oxygen consumption by complex IV (Figure 1i). Mito stress test using hemocytes revealed that FCCP injection did not lead to a rebound in OCR higher than basal levels (Figure 1j), suggesting that hemocytes do not hold the spare mitochondrial capacity, a functional parameter for mitochondrial reserve. Taken together, we conclude that embryonically derived hemocytes in Drosophila larvae predominantly utilize mitochondrial respiration to generate energy and exhibit minimal glycolysis rates under unchallenged conditions.

Metabolic activities of hemocytes change over development

Previous studies have indicated that hemocytes of the embryo are released into the hemolymph upon hatching and undergo an extensive proliferation during the second to third instars^11,22,49. Hemocytes at the late third instar right before pupariation could be less proliferative than those in earlier stages, and therefore, may differ in their metabolism. Based on the proliferation profile of hemocytes, we reasoned whether hemocytes experience metabolic fluctuations during larval development. To assess developmental changes in larval hemocyte metabolism during the second and third instars, we measured OCR or ECAR of larval hemocytes at 72 h, 96 h, or 120 h after egg laying (AEL), representing mid-second, early third, and late third instars, respectively. With the real-time ATP rate assay, we observed that hemocytes at 96 h AEL generate larger ATP production rate than those in 72 h or 120 h AEL (Figure 2a, Supplementary Figure 2a-b). At all stages, ATP production relied heavily on mitochondrial respiration, and the proportion of glycolysis-mediated ATP generation remained low, which rather diminished over development (Figure 2a). Consistently, ATP-linked respiration corresponded well with the ATP production rate (Figure 2b). This phenotype was recapitulated by biochemical ATP measurement assay (Supplementary Figure 2b). Proton leak, an indicator for mitochondrial fitness, decreased over time (Figure 2c). Furthermore, hemocytes did not respond to 2-DG at all developmental time points (Supplementary Figure 2c). These results indicate that mitochondrial respiration is the main process for hemocyte metabolism and ATP production peaks at 96 h AEL during hemocyte development.

Figure 2.

Upon hatching, larval hemocytes either freely circulate the larval hemolymph or colonize resident tissues including hematopoietic pockets of the larval body wall to control their proliferation and differentiation⁵⁰. Considering that hemocyte metabolism is modified during larval development, we asked whether circulating or sessile hemocytes display differential cellular metabolism due to their distinct microenvironment. Glycolysis-mediated ATP production was lower in sessile hemocytes (10 %) than in circulating hemocytes (18.6 %) (Figure 2d). However, most metabolic profiles, such as ATP production, ATP-linked respiration, and proton leak, were comparable regardless of the hemocyte location (Figure 2e-f; Supplementary Figure 2d-e). Overall, we conclude that hemocytes developmentally elevate ATP production rates at 96 h AEL and that hemocytes at two locations – sessile and circulating – are metabolically indistinguishable.

Hemocyte proliferation enhances mitochondrial respiration

Our findings indicate that larval hemocytes rely on mitochondrial respiration without reserving spare mitochondrial capacities. However, metabolic activities of hemocytes fluctuate during larval development, implying that these cells are able to transform cellular metabolism according to bioenergetic requirements. Based on this, we sought to understand the metabolic signatures of hemocytes in various genetic backgrounds that promote the proliferation or differentiation of hemocytes or upon immune challenges. First, we investigated whether induced proliferation alters hemocyte metabolism by using an active form of Ras, UAS-Ras^v12 (Hml^Δ-Gal4 UAS-Ras^v12). We confirmed that late-third instar larvae carrying both transgenes show over a 10-fold increase in the number of Hml⁺ plasmatocytes with a relatively minor proportion of lamellocytes differentiation (Supplementary Figure 3a-b). A fixed number of hemocytes (2.7 x 10⁵ cells) expressing Ras^v12 exhibited 2-fold higher basal OCR or ECAR levels than those in controls, which was significantly decreased to non-mitochondrial OCR levels following oligomycin and Rot/AA injections (Supplementary Figure 3c-d). Additionally, the expression of Ras^v12 considerably enhanced the ATP production rate without changing the preference for mitochondrial respiration over glycolysis (Figure 3a). Both ATP-linked respiration (Figure 3b) and proton leak (Figure 3c) were increased in this genetic background. Different from wild-type hemocytes, hemocytes expressing Ras^v12 showed higher OCR rebound than basal levels following the FCCP treatment (Figure 3d), indicating that Ras^v12-expressing hemocytes elevate the spare respiratory capacity. Thus, we conclude that Ras^v12-expressing hemocytes rewire mitochondrial capacities as well as energetic activities.

Figure 3.

Recent studies including single-cell transcriptome analyses of larval hemocytes indicated that larval plasmatocytes are composed of transcriptionally diverse subtypes, in addition to the Hml⁺ population^5,6. To test whether every plasmatocyte, regardless of its subtype represented by a marker gene expression, displays similar metabolic profiles during proliferation, we expanded Hml^- plasmatocytes by genetically ablating Hml⁺plasmatocytes. Expression of the pro-apoptotic genes hid and rpr in Hml⁺ plasmatocytes (Hml^Δ-Gal4 UAS-hid,rpr) reduces the number of Hml⁺ plasmatocytes while expanding Hml^- but Pxn⁺ plasmatocytes ⁵¹. Similar to Ras^v12-expressing Hml⁺ plasmatocytes, Hml^-Pxn⁺ plasmatocytes exhibited higher basal OCR or ECAR rates (Supplementary Figure 3e-f). These rates sharply declined following oligomycin and Rot/AA treatment in a similar trend seen by those expressing Ras^v12 (Supplementary Figure 3e-f). Consistently, ATP production rates (Figure 3e), ATP-linked respiration (Figure 3f), proton leak (Figure 3g), and the spare respiratory capacity (Figure 3h) were all increased, demonstrating that proliferation of plasmatocytes, regardless of their subtypes, augments hemocyte metabolism.

Lastly, we assessed the metabolic activities of crystal cells, the second largest population under normal growing conditions. Since crystal cells comprise only ∼ 5 % of total hemocytes, we increased their proportion by accelerating crystal cell differentiation through Notch^ICD expression (Hml^Δ-Gal4 UAS-N^ICD). Overexpression of Notch^ICD successfully expanded the crystal cell pool (Supplementary Figure 3g). However, number-matched Notch^ICD-expressing hemocytes displayed marginally decreased or similar metabolic levels comparable to control hemocytes, including ATP production, ATP-linked respiration, and proton leak (Supplementary Figure 3h-j). Moreover, the mito stress test was unchanged in this genetic background (Supplementary Figure 3k), suggesting N^ICD-expressing hemocytes with expanded crystal cell population exhibit lower metabolic activities than wild-type hemocytes.

Overall, these results establish that an increase in the number of plasmatocytes, but not crystal cells, is accompanied by enhanced respiratory activity and increased mitochondrial reserve.

Lamellocyte differentiation or active immunity triggers mitochondrial respiration

While plasmatocytes are the major hemocyte population under unchallenged conventional conditions, lamellocytes are the predominant cell type that give rise during active immunity or stress responses^15,16. To gain further insights into the lamellocyte differentiation and mitochondrial function, we expressed a constitutively active form of the Janus kinase Hopscotch (hop^Tum-l) in hemocytes (Hml^Δ-gal4 UAS-hop^Tum-l) to genetically induce a large number of lamellocytes (Supplementary Figure 4a-c)^17,52,53. Hemocyte-specific expression of hop^Tum-l considerably elevated basal OCR or ECAR and raises ATP production rates, ATP-linked respiration, and proton leak (Figure 3i-k; Supplementary Figure 4d-e). Also, we detected spare respiratory capacities in hop^Tum-lexpressing hemocytes similar to those found in Ras^v12-or hid, rpr-expressing hemocytes (Figure 3l).

Drosophila larvae respond to wasp-egg deposition within a few hours and trigger a series of innate immune responses, including a massive differentiation of lamellocytes⁵⁴. Immature lamellocytes appear by 24 h post-infestation (PI) and develop into mature lamellocytes by 48 h PI to encapsulate wasp eggs^5,55. Given that proliferation or differentiation of hemocytes stimulated by gene expression alters hemocyte metabolism, we asked whether innate immunity caused by a natural infection influences hemocyte respiration. Hemocytes at 24 h PI when mature lamellocytes have not yet given rise showed OCR or ECAR levels similar to unchallenged controls at 96 h AEL and did not change other metabolic indicators (Supplementary Figure 4f-g). On the other hand, hemocytes at 48 h PI expressed significantly higher OCR or ECAR by two folds than wild-type hemocytes at 120 h AEL (Supplementary Figure 4h-i). Moreover, these hemocytes increased their ATP production rate, ATP-linked mitochondrial respiration, and proton leak (Figure 3m-o). The mito stress test recognized that hemocytes at 48 h PI result in elevated spare respiratory capacities (Figure 3p), suggesting that immune-activated hemocytes and mature lamellocytes amplify mitochondrial respiratory rates and capacity.

Overall, these data suggest that while larval hemocytes under unchallenged conditions rely largely on mitochondrial respiration and exhibit low metabolic activity, hemocyte proliferation, lamellocyte differentiation, or active immune responses significantly elevate mitochondrial as well as glycolytic activities to promote ATP production.

Rearrangement of mitochondrial structures in plasmatocytes and lamellocytes

Since metabolic alterations often accompany structural modifications of mitochondria⁵⁶, we hypothesized that mitochondria in hemocytes undergo morphological or quantitative changes during hemocyte development or upon active immunity. To address this, we used UAS-mitoGFP to visualize mitochondrial morphology in embryonically derived hemocytes (Srp-Gal4 UAS-mitoGFP). Over larval development under unchallenged conditions, plasmatocytes experienced an increase in mitochondrial size during 72 to 96 h AEL and maintained the size until the third instar 120 h AEL (Figure 4a-c). In contrast, the number of mitochondria per one plasmatocyte decreased, particularly between 72 and 96 h AEL (Figure 4a-c); therefore, the total mitochondrial size per one plasmatocyte remained relatively constant throughout the developmental period. A stark morphological change during 72 to 96 h AEL corresponded well with fluctuations of metabolic parameters of hemocytes at this stage (Figure 2a-c), supporting that metabolic plasticity is associated with mitochondrial rearrangement. Under wasp parasitism, immature lamellocytes given rise at 24 h PI contained a markedly higher number of mitochondria with a smaller size than age-matched plasmatocytes at 96 h AEL (Figure 4a-c). At 48 h PI, mature lamellocytes marginally reduced the number of mitochondria but increased the average mitochondrial size compared to immature lamellocytes at 24 h PI (Figure 4a-c). These results suggest two different regulatory modes of mitochondrial plasticity in plasmatocytes and lamellocytes: developing plasmatocytes maintain relatively constant numbers of mitochondria with an increase in size during the second to third instar. However, lamellocytes undergo extensive mitochondrial fragmentation during the initial differentiation process, which is later attenuated and shifted towards the size expansion during maturation.

Figure 4.

Mitochondrial fission is required for lamellocyte immune function

To understand how hemocytes acquire differential mitochondrial activity during development and immune activation, we examined transcriptional profiles of metabolic genes in wild-type and wasp-infested larval hemocytes extracted from the single-cell RNA sequencing^5,7. We integrated single-cell RNA transcriptomes of hemocytes into pseudo-bulk levels and clustered cell types into six groups: developing plasmatocytes at 96 h AEL or 120h AEL, plasmatocytes upon wasp infestation at 24 h PI or 48 h PI, and lamellocytes at 24 h PI or 48 h PI, corresponding 96 h or 120 h AEL, respectively. Metabolic gene expressions in plasmatocytes at 96 h or 120 h AEL were nearly identical, except for a minor decrease in hexokinase, Hex-A (Figure 4d). However, plasmatocytes changed their metabolic gene expressions during wasp parasitism by downregulating Hex-A, UQCR-6.4 (mitochondrial electron transport, complex III), ND-AGGG (mitochondrial electron transport, complex I), and up-regulating Gapdh2, Cyt-c1, or ATPsynbeta (Figure 4d). Lamellocytes at 24 h or 48 h PI, on the other hand, displayed considerably distinct metabolic gene expression profiles than plasmatocytes. Four representative categories of genes were highly induced in lamellocytes: glucose homeostasis (Gapdh1, Gapdh2, Tpi, Pglym78, and Eno), mitochondrial fission (Drp1 and Fis1), SLC2 family sugar transporter (CG1208, Tret1-1, and CG4607), and starch and sucrose metabolism (Treh) (Figure 4d). These genes in lamellocytes were already elevated by 24 h PI and their expression further increased as lamellocytes mature by 48 h PI (Figure 4d). The elevation of mitochondrial fission genes such as Drp1 and Fis1 in the lamellocyte corresponded with the changes in mitochondrial morphologies (Figure 4d). However, we found no apparent changes in the mitochondrial fusion gene, Marf, in lamellocytes at these time points (Supplementary Figure 5a). To test whether mitochondrial fission is associated with lamellocyte differentiation or function, we analyzed lamellocyte numbers and encapsulation success rates after knocking-down Drp1 in hemocytes (Hml^Δ-Gal4 UAS-GFP UAS-Drp1 RNAi). Despite the marked upregulation in Drp1 mRNA, Drp1 RNAi did not alter lamellocyte differentiation (Supplementary Figure 5a-b). However, hemocyte-specific expression of Drp1 RNAi significantly reduced encapsulation of wasp eggs (Figure 4e-f), suggesting that mitochondrial remodeling is required for melanization response mediated by lamellocytes. Together, these findings indicate that relative to both unchallenged and challenged plasmatocytes, lamellocytes emerging upon wasp infestation exhibit differential expression of genes associated with sugar metabolism and mitochondrial structure and activity. Additionally, morphological rearrangement of mitochondria is critical for immune effector function of lamellocytes (Figure 4g).

Functional lamellocytes utilize sugars to fuel mitochondrial respiration

Given that hemocytes at 48 h PI stimulate energy production and lamellocytes express increased levels of sugar metabolism and mitochondria-related genes, we investigated whether hemocytes at 48 h PI respond differently to sugars. Trehalose, a primary sugar type used by Drosophila, is either stored in the fat body or mobilized to glucose as an energy course⁵⁷. To test if hemocytes under unchallenged conditions and immune-activated hemocytes differentially respond to sugars, we injected trehalose or glucose into control hemocytes at 120 h AEL or lamellocytes at 48 h PI. Hemocytes at 120 h AEL marginally elevated ECAR levels after glucose inoculation, which subsequently declined following oligomycin administration (Figure 4h; Supplementary Figure 5c). However, trehalose injection did not raise ECAR levels in control hemocytes, implying a specific response of developing hemocytes to glucose (Figure 4h; Supplementary Figure 5c). In contrast, hemocytes at 48 h PI, mainly comprised of lamellocytes, reacted to both glucose and trehalose injection by instantly increasing their ECAR levels, which were suppressed by oligomycin (Figure 4h; Supplementary Figure 5d). The significance of glycolysis in the ECAR elevation was probed by applying 2-DG to these hemocytes following the oligomycin treatment. 2-DG did not alter ECAR similarly to unchallenged controls (Supplementary Figure 5c-d). These data suggest that lamellocytes utilize glucose and trehalose mainly through mitochondrial respiration rather than activating glycolysis.

Innate immunity counteracts nutrient storage^58,59 and consequently changes the hemolymph sugar levels^39,60,61. To understand how systemic sugar levels correlations with hemocyte metabolism, we measured hemolymph sugar levels before and after wasp infestation and across development. Between 96 to 120 h AEL, larvae showed a dramatic increase in glycemic levels that represent both trehalose and glucose in the hemolymph (Figure 4i). Similar to developing hemocytes, wasp infestation led to a robust increase in glycemic levels at 48 h PI, reaching levels higher than those observed in 120h AEL (Figure 4i). To test whether elevated glycemia and sugar responsiveness during wasp parasitization contribute to hemocyte differentiation or proliferation, we quantified total hemocytes, Pxn⁺ plasmatocytes, and lamellocytes after hemocyte-specific RNAi against HexA or Treh RNAi (Srp-Gal4 UAS-Treh RNAi or Srp-Gal4 UAS-HexA RNAi). Knockdown of these sugar metabolism or transporter genes did not alter absolute hemocyte counts, including lamellocytes (Supplementary Figure 6a-c). However, it significantly reduced the efficiency of wasp egg encapsulation (Figure 4j-k), indicating that sugar metabolism is essential for the functionality of lamellocyte-mediated encapsulation, similar to the inhibition of mitochondrial fission (Figure 4e-f). Together, these results indicate that developing hemocytes are selectively responsive to glucose, whereas hemocytes stimulated during wasp parasitism (48h PI), which largely comprise lamellocytes, utilize both glucose and trehalose to support elevated metabolic activities. Furthermore, metabolic reprogramming, associated with mitochondrial remodeling and sugar responsiveness, is critical for the activation of innate immune function of lamellocytes.

Discussion

In this study, we characterized the metabolic characteristics of Drosophila larval hemocytes and identified that these cells primarily rely on oxidative phosphorylation (OXPHOS) for energy metabolism. We showed that OXPHOS activity scales linearly with hemocyte proliferation during development and in response to genetic perturbations, except crystal cells. Furthermore, lamellocyte differentiation induced either by the hemocyte-specific expression of hop^TumL or by parasitic wasp infestation significantly enhances both OXPHOS capacity and activity. This metabolic shift was accompanied by mitochondrial remodeling and the upregulation of trehalose metabolism-related genes. While trehalose metabolism was required for an effective melanization response, it did not significantly modify the differentiation of lamellocytes or the underlying OXPHOS-dependent respiration even during immune challenge. Collectively, our findings indicate OXPHOS-dependent respiration and its associated metabolic pathways are closely associated with hemocyte proliferation and innate immune activation, implying a distinctive metabolic strategy of myeloid-like immune cells in Drosophila.

Macrophages serve as sentinels of cellular stress and immune challenge, exhibiting remarkable plasticity in response to various environmental stimuli⁶². In vertebrates, activated macrophages are traditionally categorized into pro-inflammatory M1 and anti-inflammatory M2 phenotypes, which display profound metabolic divergence. M1 macrophages preferentially utilize aerobic glycolysis and the pentose phosphate pathway (PPP) to support rapid energy production and the biosynthesis of inflammatory intermediates^63,64, whereas M2 macrophages rely on OXPHOS⁶⁵. In this context, our results demonstrate that Drosophila hemocytes, mainly consisting of macrophage-like plasmatocytes, show a metabolic signature similar to that of M2 macrophages, characterized by a high dependence on OXPHOS-driven metabolism. Interestingly, and in contrast to the vertebrate M1 paradigm, Drosophila hemocytes do not undergo a significant glycolytic shift even under immune challenge. Instead, they elevate total ATP production by the upregulation of OXPHOS capacity. By maintaining OXPHOS, hemocytes may be able to achieve the high metabolic efficiency required to support cellular remodeling of lamellocytes and support the synthesis of immune effectors for the melanization cascade. Alternatively, hemocytes may engage glycolysis transiently to mount a rapid response to acute infection and then revert to OXPHOS during the post-acute phase. This program may further support immune functions by directing carbon flux into the PPP pathway to maintain redox homeostasis⁶⁶. Our data suggest the possibility that insect immune cells favor metabolic economy over the metabolic wastefulness of the Warburg effect. This strategy may be advantageous in the open hemolymph system where circulating sugar supports both systemic growth and immunity as a shared limiting source. Thus, immune activation may not universally require a glycolytic shift but may instead depend on the organism’s life traits and systemic metabolic preference.

Previous studies have shown that larval hemocytes are metabolically quiescent at steady-state, unchallenged conditions. However, sensory neurons triggered by wasp odors enhance GABA release to mount an immune response via lamellocyte differentiation⁶⁷. The same pathway also modulates the ROS levels through succinate conversion⁶⁸, which all contribute to metabolic activities. These studies indicate that TCA cycle may not be fully active in hemocytes under unchallenged conditions, which likely attenuates TCA cycle during the surveillance phase and constrains baseline metabolic activity of hemocytes. This strategy may be important for minimizing the systemic energy cost of maintaining a large immune cell population. Interestingly, this hypothesis is also consistent with studies demonstrating that a high sugar diet is detrimental to the development and immunity of larval hemocytes^51,69. These collective findings, together with our analyses, support a hypothesis that larval hemocytes maintain a tightly regulated metabolic baseline that prioritize energy conservation during homeostasis while preserving sufficient bioenergetic plasticity for rapid metabolic reprogramming upon pathogen encounter.

In the late third instar, plasmatocytes comprise ∼ 95 % of hemocytes. Accordingly, our measurements primarily reflect the metabolic dynamics of plasmatocyte population. Despite efforts to profile each hemocyte type in isolation, it was not feasible to completely separate the three lineages. Instead, we leveraged genetic backgrounds that bias expansion of specific hemocyte types, including Hop^TumL, Ras^V12, and Notch^ICD, to enrich for the population of interest. Expression of Ras^V12 robustly expanded plasmatocytes by 10 to 50 folds with minimal lamellocyte expansion, whereas Hop^TumLor wasp infestation enabled metabolic quantification of lamellocytes alongside activated plasmatocytes^16,24. Notch^ICD expression increased crystal cell differentiation but likely included a residual plasmatocyte fraction¹³. These enrichment strategies, while informative, introduce activation-state and admixture caveats that can influence metabolic readouts. Future cell-type resolved approaches, including advanced cell-sorting or single-cell metabolomic profiling, will be essential to define these programs and the underlying regulatory mechanisms across fully resolved hemocyte types and hematopoietic origins. Moreover, incorporating single-cell resolution analysis and space- and time-resolved measurements after immune or genetic perturbation will further clarify whether observed shifts reflect intrinsic metabolic rewiring or context-dependent activation.

While plasmatocytes in the unchallenged state express basal levels of carbohydrate transporters, lamellocytes significantly induced trehalose transporters and trehalase upon differentiation^5–7. In addition, an elegant ¹³C-isotope labeling approach showed that activated hemocytes channel carbon sources primarily into the cyclic pentose phosphate pathway (cyclic PPP) to generate NADPH, together with ATP, pyruvate, or pentose⁶⁶. Integrating the prior work with our analyses, we propose that activated plasmatocytes and lamellocytes maximize cyclic PPP flux to bolster antioxidant buffering for host protection, while sustaining viability via OXPHOS-mediated energy metabolism. By contrast, increasing the number of crystal cells did not significantly alter overall metabolic parameters. Although crystal cells were viable at the time of measurement, they may have adjusted their metabolic program in response to environmental changes in vitro. Moreover, N^ICD-induced crystal cells may not fully recapitulate developmentally programmed crystal cell differentiation and maturation. Given that crystal cells are non-mitotic and expand by transdifferentiation⁷⁰ and that crystal cells express a significant number of oxygen-carrying proteins¹⁴, these cells may adopt a distinct metabolic strategy relative to lamellocytes or plasmatocytes.

Materials and methods

Drosophila melanogaster stocks and genetics

The following Drosophila stocks were used in this study: Hml^Δ-Gal4 (S.Sinenko), UAS-Ras^V12 (BL4847), UAS-hid, rpr (Nambu J. R.), UAS-hop^Tum-l (K.Brueckner), Srp-Gal4 (L.Waltzer), UAS-mito-HA-GFP (BL8442), Drp1 RNAi (BL27682), Hex-A RNAi (BL35155), Treh RNAi (BL51810), UAS-Notch^ICD (U. Banerjee).

Fly stocks used in this study were maintained at 18 °C. Unless otherwise indicated, Gal4-UAS fly crosses were incubated at 25 °C and 70 % humidity. To synchronize larval development, approximately one hundred adult flies were allowed to lay eggs on a grape-juice agar plate for two-hour period. At 23 hours after egg laying (AEL), any hatched larvae were eliminated from the plate. Larvae were then transferred to and maintained on standard cornmeal-based food at 24 h AEL.

Hemocyte isolation and immunohistochemistry

Before hemocyte isolation, larvae were vortexed with glass beads (Sigma-Aldrich, G9268) for three minutes. This step was performed to extract total hemocytes, including sessile and circulating populations. Vortexed larvae were bled onto a slide glass (Immuno-Cell Int, 61.100.17) which was kept at 4 °C (or on ice) for 45 minutes to allow hemocytes to settle and adhere. Samples were fixed with 3.7 % formaldehyde in 1x PBS for 30 minutes. Following fixation, samples were washed three times in 0.4 % Triton X-100 in 1× PBS for 10 minutes and blocked in 10 % normal goat serum (NGS) in 0.4 % TritonX, 1× PBS for 30 minutes. The primary antibody (α-Pxn, 1:2000) was added and samples were incubated in a humidified chamber overnight at 4 °C. The following day, samples were washed three times in 0.4 % Triton X in 1× PBS and incubated in the secondary antibody or Phalloidin (ThermoFisher, A12380, 1:400) for 3 hours at room temperature⁷¹. After washing 3 times with 0.4% Triton X in 1× PBS, samples were mounted using Vectashield (Vector Laboratory) with DAPI and imaged using the Zeiss Axio Imager M2, Nikon C2 Si-plus, or Zeiss LSM900 with Airyscan confocal microscope. Images were analyzed by Image J (Version 1.54) or Imaris (Bitplane).

Analysis of D. melanogaster samples using the Agilent Seahorse XFe96 analyzer

The Agilent Seahorse XFe96 analyzer was preset and maintained at 28 °C (heater on). The day prior to the analysis, the Agilent Seahorse XFe96 cartridge (Agilent, 102416-100) was hydrated using a utility plate (Agilent, 102416-100) filled with 200 μl of cell culture grade sterile water and incubate in a 28 °C non-CO₂ chamber. Simultaneously, the Seahorse XF calibrant buffer (Agilent, 102416-100) was pre-warmed in 28 °C non-CO₂ chamber. The following day, water was removed from both the utility plate and the cartridge. The utility plate was refilled with 200 μl of pre-warmed calibrant buffer in each well. The cartridge was combined with utility plate and incubated in the 28 °C non-CO₂ chamber over 30 minutes before running the analyzer. Larval brain was dissected in phosphate-buffered saline (PBS) and loaded onto the Agilent Seahorse XF96 cell culture microplates (101085-004) containing assay media. The brains were carefully settled to the bottom center of each well in the microplate. To ensure optimal signal, we recommend to gently put the brain to the center of the well between the three spherical supports using a pipet tip.

Hemocytes and S2 cells were prepared in Schneider’s Drosophila Medium (Thermo Fisher, 21720024) and filtered using a 40 μm cell strainer (Falcon, 352340). Filtered cells were centrifuged at 6000 rpm for 5 minutes and suspended into the assay media. After adjusting the cell concentration to 1.5x10⁶ cells/ml, 180 μl of the assay media was added to each well.

The sample-loaded microplate was centrifuged at 3000 rpm for 3 minutes (Eppendorf, EP022628146) to ensure settling and adherence. The cartridge combined with utility plate was loaded with assay solution on injection ports and placed onto the Seahorse XFe96 analyzer and the run was initiated. The utility plate was replaced with the sample microplate after the equilibration phase.

Assay media and solutions

Assay media were prepared immediately prior to each experiment based on the specific metabolic test. The base media for all assays were Seahorse XF DMEM medium, pH 7.4 (Agilent, 103575-100). For real-time ATP rate assay, glycolytic rate assay, and mito stress test, the base medium was supplemented with 10 mM of XF glucose (103577-100), 1mM of XF pyruvate (103578-100), and 2mM of XF glutamine (103579-100). For glycolysis stress test, the assay was proceeded in Seahorse XF DMEM medium, pH 7.4 mixed with 2mM of XF glutamine (103579-100).

Metabolic inhibitors or substrates were prepared as 10X stock solutions and diluted into the respective basal assay media: Oligomycin A (Sigma-Aldrich, O4876), 2-DG (Sigma-Aldrich, D8375), Rotenone (Sigma-Aldrich, R8875), FCCP (Sigma-Aldrich, C2920), glucose (103577-100), or trehalose (PHR1344). These media were loaded into the Seahorse XFe96 analyzer injection ports immediately before the assay run.

Wasp infestation

Drosophila melanogaster larvae were collected at 72h AEL and exposed to Leptopilina boulardi (G486) wasps for 6 hours. Following the exposure, adult wasps were removed. The infestation process was maintained under controlled conditions of 25 °C and 70 % humidity. Infestation was confirmed by observation of wasp eggs or larvae inside the host, which was verified during bleeding.

Preprocess of scRNA-seq datasets

Drop-seq UMI count matrices of circulating hemocytes from 96 and 120 h AEL under steady state and wasp infection were retrieved from our previous study GSE141273 ⁷. Total of 12 libraries, three for each condition, were downloaded and merged. Because the gene annotation version was different between steady state (BDGP 6.22) and wasp infested (BDGP 6.02) data, an FBgn to Annotation ID conversion table from FlyBase (https://flybase.org; fbgn_annotation_ID_fb_2019_03.tsv) was used to convert gene IDs to the BDGP 6.22 version. Because these UMI matrices were already preprocessed (i.e., filtration of low-quality cells based on UMI, gene counts, and mitochondrial contents), no further filtration was performed. Briefly, a few outlier cells were first filtered using UMI thresholds; 96 and 120 h AEL wild type, UMI count > 50,000; 96 h AEL infested, > 70,000; 120 h AEL infested, > 55000. Then cells having UMI counts higher than two standard deviations from the mean UMI count were also removed to exclude possible multiplets. Low quality cells having only limited number of genes were also removed; 96 and 120 h AEL wild type, gene count < 200; 96 h AEL infested, < 300; 120 h AEL infested, < 400. Because mitochondrial contents in circulating hemocytes were a bit higher than those in lymph gland counterparts (a threshold of < 10% was used), we applied 20% as a threshold. Remaining cells were 1006 (96 h AEL wild type), 1671 (120 h AEL wild type), 5986 (96 h AEL infested), and 6693 cells (120 h AEL infested).

Integration of scRNA-seq data from different conditions

For each dataset (96 h AEL steady state, 96 h AEL wasp infested, etc.), libraries were first predicted for cell types using a lymph gland from 72 to 120 h AEL as a reference as we performed in the previous study ⁷. In details, each dataset was first transformed to a Seurat object and three independent sequencing libraries were integrated using FindIntegrationAnchors() and IntegrateData() with default parameters ⁷². Then, cell type annotation was transferred from the lymph gland dataset using FindTransferAnchors() and TransferData() with default parameters. Non-hematopoietic cell types (dorsal vessel, neurons, ring gland, and posterior signaling center cells) and a few minor subclusters (< 0.1% of total population) were removed. In the end, 995 (96 h AEL steady state), 1356 (120 h AEL steady state), 5674 (96 h AEL wasp infested), 6376 cells (120 h AEL wasp infested) were remained. After the prediction step, UMI counts were normalized, log-transformed, and scaled, and PCA analysis was performed which identified 40 significant principal components (PCs). Four datasets were integrated using Harmony ⁷³ with default parameters. t-SNE and UMAP plots were generated using the selected numbers of PCs.

Supplementary figrues

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.

Supplementary Figure 4.

Supplementary Figure 5.

Supplementary Figure 6.

Data availability

All data generated during and/or analyzed during the current study are included in this published article and its supplementary files. In-house Python and R source codes used in this study are publicly available on GitHub repository (https://github.com/sangho1130/dmel_metabolism). Analyses using public tools were performed using default parameters, otherwise, described in the Method section. All analyses were performed using Python (version 2.7.5), R (version 3.5.3), and R Studio (version 1.1.383). Detailed software versions are described in the Methods.

Acknowledgements

We thank all members of the Shim laboratory for their helpful comments. We thank Bloomington Drosophila Stock Center (NIH P40OD018537) and the Drosophila Genomics Resource Center (NIH 2P40OD010949), FlyBase (release 2025_04, 2019_03), and the Korean Drosophila Resource Center for supporting the fly community. This work was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (RS-2024-00349703 and RS-2025-02232977) and the New Faculty Startup Fund from Seoul National University to J.S.

Additional information

Author contributions

DL performed Seahorse analyses and wasp infestation analyses. FK and NC performed wasp infestation analyses and associated hemocyte counting. SHY, KTL, and JWN performed in silico analysis using hemocyte single cell RNA sequencing. DL, YVK, and JS designed this study and wrote the manuscript. YVK and JS supervised the project and conceived the idea.

Funding

National Research Foundation of Korea (NRF) (RS-2024-00349703)

• Jiwon Shim

National Research Foundation of Korea (NRF) (RS-2025-02232977)

• Jiwon Shim