Lipid droplet profiling during neutrophil differentiation by stimulated Raman scattering microscopy

Abstract

Lipid droplets (LDs) are dynamic organelles that serve as metabolic hubs and emerging regulators of immune cell fate. Although LDs have been implicated in immune regulation, how LD metabolism is remodeled during neutrophil differentiation and how stage-specific LD dynamics shape mature neutrophil function remain poorly defined. Here, we profiled LD dynamics during neutrophil development and evaluated their role in innate immune function. Using stimulated Raman scattering (SRS) microscopy, we performed label-free, quantitative mapping of LD accumulation in two complementary differentiation systems: murine Hoxb8 myeloid progenitors and human pluripotent stem cell (hPSC)-derived neutrophils. To define metabolic requirements, we pharmacologically modulated LD biosynthesis and catabolism throughout neutrophil differentiation. Perturbing LD metabolism did not impair differentiation efficiency or lineage commitment, but significantly altered the functional output of mature neutrophils. In the murine system, inhibition of adipose triglyceride lipase (ATGL)-mediated LD breakdown enhanced reactive oxygen species (ROS) production and increased anti-tumor cytotoxicity against GL261 glioma cells. In the hPSC model, ATGL inhibition during the myeloid progenitor-to-neutrophil transition selectively increased intracellular LD accumulation without compromising neutrophil yield or purity. This metabolic rewiring also elevated ROS production in hPSC-derived neutrophils, although cytotoxic enhancement against U87MG glioblastoma cells was less pronounced than in the murine system. Collectively, these findings define a stage-specific LD metabolic landscape during neutrophil development and highlight targeted LD modulation as a potential strategy to enhance the functional potency of therapeutic neutrophils.

INTRODUCTION

Neutrophils are the most abundant leukocytes in circulation, comprising ~50–70% of white blood cells in humans and ~10–25% in mice. However, their short lifespan and inability to proliferate after maturation present significant challenges for neutrophil biology studies. Despite these limitations, neutrophils have emerged as key players in a wide range of physiological and pathological processes, including acute injury, tissue repair, autoimmunity, chronic inflammation, and cancer.^1–6

Immune cell differentiation and activation are accompanied by extensive metabolic reprogramming, which in turn shapes cellular identity and function. Although mature neutrophils primarily rely on glycolysis for ATP production, emerging evidence suggests that their function is closely linked to metabolic pathways beyond glucose metabolism.^7–10 Among lipid-associated metabolites, lipid droplets (LDs) have garnered increasing attention for their roles in regulating immune cell behavior.

LDs are intracellular organelles composed of a neutral lipid core, primarily cholesteryl esters and triacylglycerols (TAGs), surrounded by a phospholipid monolayer embedded with specific proteins. LD in leukocytes commonly contain neutral lipid stores enriched in TAG. Traditionally viewed as inert lipid reservoirs or buffers against lipotoxicity, LDs are now recognized as metabolically active organelles that interact with other cellular compartments to regulate immune responses^9,10, influence tumor progression¹¹, and govern cell fate decisions^12–14. Recent studies have further established LDs as dynamic metabolic and signaling platforms that regulate immune cell function across diverse cell types, including T cells^15,16, macrophages^17–19, and dendritic cells^20,21. Adipose triglyceride lipase (ATGL), which catalyzes the rate-limiting step of LD breakdown, plays a central role in controlling the availability of lipid mediators. For instance, ATGL deficiency in macrophages leads to excessive LD accumulation and reduced release of free fatty acids (FAs), resulting in decreased ATP production and impaired phagocytic capacity.²² In addition, the tumor microenvironment and tumor-derived signals, such as IL-33 and arachidonic acid, can induce lipid-driven neutrophil phenotypes that promote metastasis and therapeutic resistance.^23,24 Both extracellular lipid droplets, which can carry alarmins to promote neutrophil infiltration during type 2 inflammation, and intracellular lipid metabolic remodeling have been implicated in driving pro-metastatic and immunosuppressive neutrophil phenotypes in cancer.²⁵ Notably, lipids released from LDs directly participate in metabolic and signaling pathways that regulate gene transcription, energy homeostasis, and cellular signaling networks.

Beyond immune effector functions, lipid metabolic remodeling has emerged as a critical determinant of cell differentiation and lineage commitment.^26–28 Notably, autophagy-related gene 7 (Atg7) plays a key role in neutrophil maturation in mice. Atg7 deficiency disrupts LD-associated autophagy, leading to LD accumulation, impaired neutrophil maturation, and increased retention of immature neutrophils in the bone marrow.²⁹ These findings suggest that precise regulation of LD dynamics is essential not only for neutrophil function but also for proper neutrophil development.

However, despite increasing recognition of LDs in immune regulation, the dynamic control of LD metabolism during neutrophil differentiation and how stage-specific LD remodeling influences mature neutrophil function remain poorly understood. In this study, we quantitatively profile LD dynamics across distinct stages of neutrophil differentiation using both Hoxb8-derived neutrophils and human pluripotent stem cell (hPSC)-derived neutrophils, leveraging Stimulated Raman Scattering microscopy (SRS). The estrogen-regulated Hoxb8 myeloid progenitor system was selected as a tractable murine model because it enables scalable expansion of progenitors and synchronized differentiation into neutrophil-like cells upon estrogen withdrawal, thereby providing a robust and reproducible platform for longitudinal imaging and pharmacological perturbation studies. In parallel, hPSC-derived neutrophils enabled assessment of whether LD remodeling is conserved across the human system. Unlike conventional fluorescence imaging, SRS is a label-free technique that detects lipid molecules based on their intrinsic vibrational signatures,³⁰ enabling high-resolution, real-time quantification of lipid droplets without the need for exogenous dyes that may perturb cellular metabolism. By integrating pharmacological perturbations with SRS imaging, we delineate stage-specific roles of LD metabolism during neutrophil differentiation and assess the functional consequences in mature neutrophils. Collectively, our findings reveal a critical role for LD-mediated metabolic regulation in governing both neutrophil development and functional output.

MATERIALS & METHODS

hPSC maintenance and neutrophil differentiation

H9 hPSCs (WiCell, WA09) were maintained on Matrigel-coated surfaces in mTeSR plus medium (STEMCELL Technologies, Cat#100–0276). For lineage induction, cells were dissociated and re-seeded onto iMatrix 511-coated 24 -well plates at a density of 1–8×10⁴ cells/cm² in mTeSR plus supplemented with 5 μM Y27632 (Cayman Chemical, Cat#10005583). Differentiation was initiated on day 0 by treating cells with 6 μM CHIR99021 (Cayman Chemical, Cat#13122) in DMEM (Gibco, Cat# 11320033) supplemented with 100 μg/mL ascorbic acid (DMEM/Vc). From days 1 to 4, endothelial specification was induced using LaSR basal medium (Gibco, Cat# 12491015) with 50 ng/mL VEGF added beginning on day 2. On day 4, cultures were transitioned to Stemline II medium supplemented with 10 μM SB431542, 25 ng/mL SCF, and 25 ng/mL FLT3L to promote endothelial-to-hematopoietic transition (EHT). SB431542 was removed on day 6, and cells were maintained in Stemline II medium containing 50 ng/mL SCF and 50 ng/mL FLT3L. Additionally, 25 ng/mL GM-CSF was introduced during half-medium changes on days 9 and 12. On day 15, non-adherent hematopoietic progenitors were collected and further differentiated in HPC Expansion Medium XF (Promocell, Cat#C-28021) supplemented with 1x GlutaMAX (ThermoFisher Scientific, Cat#35050079), 2.5 μM AM580, and 150 ng/mL G-CSF. Half-medium changes were performed every three days, and mature neutrophils were harvested on day 21 for downstream functional assays.

Hoxb8 neutrophil differentiation

Cells were maintained at 37°C in a humidified incubator with 5% CO₂. Hoxb8-conditional murine myeloid progenitors were cultured in basal medium consisting of Opti-MEM with GlutaMax (Gibco, Cat# 59185–034), supplemented with 1x Pen/Strep (Gibco, Cat# 15140–122), 1x non-essential amino acids (NEAA; Sigma, Cat# M-7145), 30 uM β-mercaptoethanol (BME; Sigma, Cat# M-6260), and 10% fetal bovine serum (FBS; Gemini, Cat# 100–106). To maintain cells in an undifferentiated progenitor state, the medium was further supplemented with 50 ng/mL mouse stem cell factor (mSCF; BioLegend, Cat# 579706) and 100 nM Z-4-hydroxytamoxifen (4-OHT; Tocris, Cat# 3412). Progenitors were typically maintained at a density of 2×10⁵ to 2×10⁶ cells/mL in non-adherent flasks to reduce cell adherence. To initiate neutrophil differentiation, progenitors were harvested and washed 3–4 times with sterile PBS to ensure complete removal of 4-OHT. Cells were then resuspended at a density of 2–3×10⁵ cells/mL in differentiation medium, consisting of basal medium supplemented with 20 ng/mL mSCF and 20 ng/mL mouse granulocyte colony-stimulating factor (mG-CSF; BioLegend, Cat# 574604). The medium was refreshed every other day, and differentiation was carried out for 4–5 days in non-adherent culture vessels. Neutrophil maturation and purity were assessed by flow cytometry, monitoring the downregulation of hematopoietic marker CD117 (BioLegend, Cat# 127608) and the upregulation of neutrophil surface marker Ly6G (BioLegend, Cat# 105812).

In vitro cell treatments

To inhibit LD biogenesis, cells were treated with 1 μM DGAT1 inhibitor A922500 (Sigma-Aldrich, Cat#A1737) and 2.5 μM DGAT2 inhibitor PF-06424439 (Sigma-Aldrich, Cat#PZ0233) for 2 days at defined time points in both cell models. To inhibit ATGL-mediated LD catabolism, cells were treated with 10 μM Atglistatin (Cayman Chemical, Cat#15284) for 2 days at specified stages in both cell models. All inhibitors were dissolved in dimethyl sulfoxide (DMSO). Following treatment, cells were washed with PBS and replenished with fresh medium containing the respective inhibitors.

Cytotoxicity assay

Tumor cell killing was assessed using a luciferase-based viability assay. Human U87MG or mouse GL261 glioma cells that express luciferase were plated in a 96-well plate at 10,000 cells per well in 100 μL of MEM or DMEM supplemented with 10% FBS. Neutrophils were then added to each well at the indicated effector-to-target (E:T) ratios in an additional 100 μL of medium. Co-cultures were incubated at 37 °C with 5% CO₂ for 24 hours. Followiing incubation, the medium was aspirated and replaced with α-MEM containing 150 μg/mL D-luciferin (Cayman Chemical, Cat#14681) for 30 minutes at 37°C. Luminescence was subsequently measured using a plate reader (Molecular Devices; San Jose, CA, SpectraMax^® iD3). Tumor cell killing was calculated as the percentage decrease in luminescence relative to control wells containing tumor cells only, after background subtraction.

Flow cytometry analysis

Cells were collected by passing through a 70- or 100-μm strainer into a 50 mL conical tube and centrifuged at 200 × g for 5 minutes at room temperature. The supernatant was carefully removed, and the cell pellet was washed three times with 1 mL of PBS. Cells were then stained with the appropriate fluorophore-conjugated antibodies for 25 minutes at room temperature in the dark. Following staining, cells were washed three times with PBS and analyzed using a flow cytometer (Accuri C6 Plus; BD Biosciences). Flow cytometry data were processed and analyzed using FlowJo software.

Measurement of reactive oxygen species (ROS) production in neutrophils

To quantify neutrophil ROS production, supernatants from stimulated neutrophils were analyzed for hydrogen peroxide (H₂O₂) using the Amplex^™ Red Hydrogen Peroxide/Peroxidase Assay Kit (ThermoFisher Scientific, Cat#A22188). Briefly, cell culture supernatants were collected immediately after stimulation and incubated with reaction buffer containing 50 μM Amplex Red reagent and 0.1–0.2 U/mL horseradish peroxidase (HRP), prepared according to the manufacturer’s instructions. Samples, along with appropriate H₂O₂ standards, were incubated at room temperature for 60 minutes in the dark to allow the Amplex Red reagent to react stoichiometrically with H₂O₂, generating the fluorescent product resorufin. Fluorescence was measured using a microplate reader (Molecular Devices, SpectraMax iD3) with excitation at ~530–560 nm and emission at ~590 nm. Background fluorescence from buffer controls was subtracted. ROS production was expressed relative to unstimulated controls and, when applicable, normalized to the total cell number.

Stimulated Raman Scattering (SRS) microscopy and Fluorescence Imaging for label-free lipid droplet imaging

For fluorescence imaging of differentiated neutrophils, cells were stained with PE-conjugated anti-CD66b and anti-CD45 antibodies according to the manufacturer’s instructions. Briefly, cells were incubated with antibodies for 15–20 min at room temperature, followed by two washes with PBS prior to imaging.

Stimulated Raman Scattering (SRS) imaging was performed using a dual-output 80 MHz femtosecond laser (InSight X3+, Spectra-Physics). The tunable beam (680–1300 nm) was used as the pump beam, while the fixed 1045 nm output served as the Stokes beam. Laser power on the sample was ~12 mW for the pump and 25 mW for Stokes. Both lasers have a pulse duration of ~250 fs at the sample. For CH vibration imaging, the pump wavelength was set to 800 nm. The Stokes beam was modulated at 2.5 MHz with a 50% duty cycle using an acousto-optic modulator (AOM, M1205-P80L-1, Isomet). The optical delay was controlled by a motorized stage. The beams were scanned by a 2D galvo mirror system (GVS002, Thorlabs) and focused onto the sample with a 40×/1.2 NA water immersion objective (LUMPLFLN 40X, Olympus). The SRS signal was detected in the transmitted pump beam using a photodiode (S3994, Hamamatsu) and demodulated with a lock-in amplifier (HF2LI, 50 MHz, Zurich Instruments). Data acquisition and imaging were controlled using a high-speed DAQ system (PCIe-6363, National Instruments) and custom LabVIEW software. Pixel dwell time was 10 microseconds for all images acquired.

Two-photon fluorescence images were acquired on the same microscope in the epi-direction under identical scanning conditions. Fluorescence emission from PE-conjugated antibodies was collected using a photomultiplier tube (PMT, H7422P-40, Hamamatsu) equipped with a bandpass filter (FF01–562/40–25, Semrock).

Quantification of LDs in cells

Image analysis was performed using ImageJ/Fiji following the workflow outlined below. For each field of view, paired stimulated Raman scattering (SRS) and two-photon excitation fluorescence (TPEF) images were acquired from the same single imaging plane. TPEF images of PE-conjugated CD66b and CD45 staining were used for cell-type identification within heterogeneous cell populations. Only fluorescence-positive cells were included for subsequent lipid droplet (LD) quantification.

To define cell boundaries, TPEF images were processed using a Gaussian blur followed by automated Otsu thresholding. The resulting binary masks were used to generate Regions of Interest (ROIs) corresponding to individual fluorescence-positive cells. In parallel, raw SRS images were pre-processed by subtracting a Gaussian-blurred version of the image to reduce diffuse cytoplasmic background and enhance punctate LD signals. The ROIs derived from TPEF images were then overlaid onto the processed SRS images. Within each cell ROI, LDs were segmented using an additional automated thresholding step. Quantitative features, including LD count, total LD area, and center of mass, were extracted using the Analyze Particles function in ImageJ/Fiji.

All analyses were performed using single-plane images acquired from individual cells. Three-dimensional image stacks and volumetric LD quantification were not performed in this study.

RESULTS

Quantitative SRS imaging reveals dynamic regulation of lipid droplets during Hoxb8-driven neutrophil differentiation

To investigate the role of LDs in neutrophil differentiation, we first examined LD dynamics over time using SRS imaging in a murine Hoxb8 myeloid progenitor model. In this system, day 0 represents granulocyte-monocyte progenitors (GMPs), which differentiate into mature neutrophils by day 4, as confirmed by Ly6G expression (Fig. 1a). SRS imaging revealed that LD abundance remained relatively less throughout the neutrophil differentiation process, consistent with levels observed in primary human peripheral blood neutrophils³¹ (Fig. 1a–b). Notably, LDs were largely absent in fully differentiated neutrophils, consistent with their terminally differentiated state.

Fig. 1. LD are highly dynamic during Hoxb8-dervied neutrophil differentiation.

(a) Schematic illustration of Hoxb8 neutrophil differentiation and the experimental timeline for LD analysis and representative SRS images of LDs (green) merged with TPEF images of Ly6G⁺ neutrophil (red) at day 4

(b) Quantification of LD number per cell at three differentiation stages (n = 100)

(c) Representative immunofluorescence images of LDs in progenitor-stage cells following pharmacological inhibition with DGATin or Atglistatin for 48 hours. Control denotes untreated cells, and DMSO denotes vehicle-treated cells.

(d) Quantification of LD number per progenitor cell following pharmacological inhibition during the progenitor stage (n = 100)

(e) Representative SRS images of LDs (green) merged with TPEF images of Ly6G⁺ neutrophil (red) at day 4 and quantification of LD per neutrophils with control, DGATin and Atglistatin treatment

(f) Quantification of LD number per neutrophil after DGATin or Atglistatin treatment (n = 100)

(g) Differentiation efficiency (% Ly6G⁺ neutrophils) after LD metabolic perturbation (n = 5)

(h) ROS production of neutrophils with control, DGATin and Atglistatin treatment (n = 3)

(i) Tumor-killing assay of HOXB8 neutrophils (effector:target ratio 1:5) under control, DGATin, or Atglistatin treatment (n = 3)

Statistical significance was determined using one-way ANOVA followed by Tukey’s multiple comparison test; *p < 0.05 was considered statistically significant. Scale bar is 10 μm.

To determine if perturbation of LD metabolism affects neutrophil differentiation or LD accumulation, we next examined the impact of inhibiting LD biosynthesis or degradation. LD biosynthesis was inhibited using DGAT1 and DGAT2 inhibitors, collectively referred to as DGATin, which block the final step of TAG synthesis, while LD catabolism was inhibited using an ATGL inhibitor, targeting the rate-limiting enzyme in LD breakdown. Given the minimal changes in LD levels during differentiation, inhibition of LD degradation was specifically evaluated at the progenitor stage. Pharmacological inhibition of ATGL significantly increased LD accumulation (approximately two-fold) in progenitor cells (Fig. 1c–d). In contrast, inhibition of DGAT-mediated LD biosynthesis did not produce a detectable decrease in LD abundance in differentiated neutrophils, likely reflecting the intrinsically low basal LD content and limited capacity for further reduction in Hoxb8 cells (Fig. 1e–f). However, inhibition of ATGL-mediated catabolism resulted in a pronounced accumulation of LDs compared to control conditions (Fig. 1e–f). Notably, neither DGAT nor ATGL inhibition affected neutrophil differentiation efficiency, as measured by surface marker expression (Ly6G⁺) (Fig. 1g). Together, these results indicate that while LD metabolism can modulate intracellular lipid accumulation, it does not directly regulate neutrophil differentiation in this model.

Inhibition of LD breakdown enhances ROS production in Hoxb8-derived neutrophils

Given the close link between lipid metabolism and neutrophil function, we next examined whether perturbing LD synthesis or degradation affects neutrophil activity. To assess functional changes, we measured ROS production, a key effector mechanism used by neutrophils to eliminate target cells (Fig. 1h). Notably, neutrophils treated with the ATGL inhibitor exhibited significantly elevated ROS production compared with both control and DGATin-treated groups. We further evaluated the impact of LD metabolic modulation on anti-tumor cytotoxicity by co-culturing Hoxb8-derived neutrophils with mouse GL261 glioma cells in vitro (Fig 1i). Consistent with the ROS findings, inhibition of LD catabolism significantly enhanced tumor-killing activity compared to controls. Collectively, these results demonstrate that inhibition of ATGL-mediated LD breakdown enhances both ROS production and anti-tumor cytotoxicity in Hoxb8-derived neutrophils, highlighting a critical role for LD metabolism in regulating neutrophil effector function.

LDs are highly dynamic during neutrophil differentiation from hPSCs

To extend our findings on DGATin and ATGL-mediated LD metabolism to human systems, we examined LD dynamics during neutrophil differentiation from hPSCs. In this model, day 0 represents the pluripotent stem cell stage, where high LD abundance was observed (Fig. 2a). By day 9, hPSCs differentiated into homogenous CD34⁺CD31⁺ hemogenic endothelial cells, followed by the emergence of hematopoietic stem and progenitor cells (HSPCs, CD45⁺CD43⁺) at day 12. These cells further progressed into myeloid progenitors by day 15 and ultimately matured into neutrophils by day 21. Cell-surface marker expression was also characterized by flow cytometry to validate lineage progression (CD16⁺CD66b⁺) (Fig. 2b). Throughout this differentiation trajectory, LD abundance progressively decreased from the hematopoietic stage to mature neutrophils, indicating highly dynamic and active LD metabolism during lineage progression (Fig. 2c). Mature hPSC-derived neutrophils contained approximately 1–5 LDs per cell, consistent with levels observed in primary human peripheral blood neutrophils³¹ (Fig. 2d–e). The lack of a statistically significant difference between hPSC-derived and primary neutrophils (Fig. 2e) further supports the physiological relevance and fidelity of our in vitro differentiation system. Collectively, these findings demonstrate that LD abundance is dynamically regulated and progressively downregulated during human myeloid differentiation, suggesting that active lipid turnover is a hallmark of successful neutrophil maturation.

Fig. 2. Lipid droplet (LD) dynamics during neutrophil differentiation.

(a) Schematic illustration of the hPSC-derived neutrophil differentiation protocol and representative SRS images of LDs at indicated stages (day 0, 3, 12, 18, 21)

(b) Representative of flow cytometry analysis of hPSC-derived neutrophil expression (CD16⁺CD66b⁺)

(c) Quantification of LD number per cell comparing HSPCs and mature neutrophils from hPSC (n = 100)

(d) Representative SRS images of LDs in HSPCs, hPSC-derived neutrophil, and human primary neutrophils. LD signals are shown in green and merged with TPEF images of CD45⁺ HSPCs at day 15 and CD66b⁺ hPSC-derived neutrophils at day 21.

(e) Quantitative of LD number per cell across HSPCs (CD45⁺), hPSC-derived neutrophils (CD66b⁺), and primary neutrophil (n = 100)

Statistical significance was determined using one-way ANOVA followed by Tukey’s multiple comparison test; *p < 0.05 was considered statistically significant. Scale bar is 10 μm.

Inhibition of LD biosynthesis and degradation does not affect neutrophil differentiation efficiency from hPSCs

We next investigated whether perturbing LD metabolism affects neutrophil differentiation from hPSCs across distinct developmental stages. To assess stage-specific contributions, cells were treated with DGAT inhibitors (DGATin) or an ATGL inhibitor during two differentiation windows: HSPCs to neutrophils (day 9–21) and myeloid progenitors to neutrophils (day 15–21). We first examined the late differentiation window (day 15–21), as this stage most closely corresponds to the maturation stage represented in the murine Hoxb8 differentiation system. Consistent with our observations in the murine Hoxb8 system, ATGL inhibition during the final maturation stage (day 15–21) resulted in a marked accumulation of LDs in mature neutrophils (Fig. 3a). Despite these changes in LD content, overall neutrophil differentiation efficiency remained unaffected (Fig. 3b).

Fig. 3. Manipulating lipid droplet (LD) synthesis or degradation does not alter neutrophil differentiation efficiency.

(a) Schematic of inhibitor treatment during late-stage differentiation (Day 15–21) targeting LD synthesis (DGAT inhibitor, DGATin) or LD degradation (ATGL inhibitor, Atglistatin) and representative SRS images of LDs (green) merged with TPEF images of CD66b⁺ neutrophil (red) treated with control, DGATin, or Atglistatin from day 15

(b) Differentiation efficiency (%CD16⁺CD66⁺ neutrophils) following late-stage inhibition of LD synthesis or degradation (n = 3)

(c) Schematic of extended inhibitor treatment (Day 9–21) and representative SRS images of LDs (green) merged with TPEF images of CD66b⁺ neutrophil (red) treated with control, DGATin, or Atglistatin.

(d) Neutrophil differentiation efficiency (%CD16⁺CD66⁺ neutrophils) after inhibitor treatment (Day 9–21) (n = 3)

(e) HSPC differentiation efficiency (%CD43⁺C45⁺ neutrophils) following early-stage inhibitor treatment on day15 (n = 4)

(f) Quantification of LD per hPSC-derived neutrophil after inhibitor treatment across different time windows (Day 9–21 and Day 15–21) (n = 100)

Statistical significance was determined using one-way ANOVA followed by Tukey’s multiple comparison test; *p < 0.05 was considered statistically significant. Scale bar is 10 μm.

To further examine prolonged effects, we inhibited LD metabolism during the HSPC-to-neutrophil transition (day 9–21). We assessed whether prolonged perturbation (day 9–21) affected LD abundance in mature neutrophils. In contrast to late-stage inhibition, continuous inhibition from day 9 to 21 resulted in a modest increase in LD levels in Atglistatin-treated neutrophils, but again did not affect differentiation efficiency (Fig. 3d). Furthermore, evaluation on day 15 showed that neither DGAT nor ATGL inhibition altered HSC differentiation efficiency (Fig. 3e), indicating that LD metabolism is not essential for early hematopoietic commitment. Quantitative analysis revealed that LD accumulation increased approximately threefold when ATGL inhibition was applied specifically during the myeloid progenitor-to-neutrophil transition (Fig. 3f). Together, these findings demonstrate that while perturbation of LD biosynthesis or degradation does not impair neutrophil differentiation, it selectively modulates LD abundance in a stage-dependent manner, highlighting dynamic regulation of lipid metabolism during myeloid maturation.

Blocking LD catabolism in myeloid progenitors enhances ROS production

We next assessed how perturbations in LD metabolism affect neutrophil function, focusing on ROS production and tumor cytotoxicity. Previous studies have suggested that LD metabolism is closely linked to cellular redox regulation and inflammatory activation in innate immune cells. In macrophages and microglia, perturbation of LD turnover or lipolysis pathways has been associated with increased ROS production and inflammatory phenotypes.^32,33 More recent studies further demonstrated that disruption of triglyceride synthesis or ATGL-mediated lipid catabolism alters inflammatory metabolic states in myeloid cells.^34,35 Cells were treated with the inhibitor either throughout differentiation (day 9–21) or specifically during the late maturation stage (day 15–21), and mature neutrophils were subsequently collected for functional analysis at day 21. To evaluate oxidative burst capacity, differentiated neutrophils were stimulated with phorbol myristate acetate (PMA), and ROS production was monitored over time. Neutrophils differentiated in the presence of the ATGL inhibitor (Atglistatin), either during the prolonged window (day 9–21) or specifically during the late-stage window (day 15–21), exhibited an approximately 3-fold increase in ROS production compared to control groups (Fig. 4a–b). Notably, consistent with observations in the Hoxb8 differentiation system, increasing LD abundance through inhibition of LD catabolism during the myeloid progenitor-to-neutrophil transition enhanced the oxidative burst capacity of differentiated neutrophils. However, co-culture of these neutrophils with human U87MG cells did not result in a significant increase in tumor-killing activity (Fig. 4c–d). These findings suggest that enhanced oxidative burst capacity alone may not be sufficient to improve neutrophil-mediated tumor cytotoxicity under these conditions. Together, these findings suggest that altered LD turnover during neutrophil differentiation influences metabolic or lipid-mediated pathways associated with ROS generation.

Fig. 4. Stage-specific effects of lipid droplet metabolic inhibition on hPSC-derived neutrophil ROS production and tumor-killing function.

(a) ROS production assay showing time-dependent fold change of ROS generation in neutrophils treated with LD synthesis or LD degradation inhibitors (Day 9–21). (n = 3)

(b) ROS production assay showing time-dependent fold change of ROS generation in neutrophils treated with LD synthesis or LD degradation inhibitors (Day 15–21). (n = 3)

(c) Tumor cell killing by neutrophils following inhibitor treatment during indicated differentiation windows (Day 9–21). (n = 3)

(d) Tumor cell killing by neutrophils following inhibitor treatment during indicated differentiation windows (Day 15–21). (n = 3)

Statistical significance was determined using one-way ANOVA followed by Tukey’s multiple comparison test; *p < 0.05 was considered statistically significant.

DISCUSSION

Emerging evidence highlights the crucial role of LDs in regulating immune cell metabolism³⁶, cell fate decisions^12–14, and lipid-associated cellular functions^37,38. However, the role of LDs in neutrophil differentiation has remained largely unexplored. In this study, we provide a comprehensive profile and quantitative analysis of LD metabolism during neutrophil differentiation, revealing that LDs are highly dynamic and closely linked to neutrophil biology.

We found that LD abundance peaks at early, HSPC stages and progressively declines during terminal maturation. The high LD levels observed during early differentiation are consistent with metabolic programs that rely on fatty acid oxidation (FAO) to regulate stem cell self-renewal and lineage commitment.³⁹ During later stages, factors such as G-CSF, known to drive neutrophil differentiation, can promote LD accumulation via peroxisome proliferator-activated receptor gamma (PPARγ) signaling, a key regulator of adipogenesis and lipid storage.⁴⁰ Consistent with these reports, our data reveal stage-dependent LD dynamics that parallel FAO-associated metabolic transitions. Although prior studies suggest that lipid metabolites can influence lineage decisions⁴¹, our results show that pharmacological modulation of LD biogenesis or degradation does not significantly impact differentiation efficiency. These findings suggest that LD abundance alone is unlikely to be a major determinant of neutrophil lineage progression and may instead be more closely associated with downstream cellular functions. However, our results should also be interpreted in the context of the experimental model systems used. Although Hoxb8-derived and hPSC-derived neutrophils provide scalable and reproducible platforms that recapitulate many phenotypic and functional features of primary neutrophils, neither fully reproduces native granulopoiesis.^42,43 The Hoxb8 system relies on conditionally immortalized progenitors followed by experimentally induced differentiation, and functional differences from primary murine neutrophils have been reported^42,43. Likewise, although hPSC-derived neutrophils exhibit many characteristics of peripheral blood neutrophils, their differentiation trajectory, phenotype, and degree of maturation can vary depending on the differentiation protocol^44–46. These model-specific features may influence LD remodeling and cellular responses to metabolic perturbation, and therefore should be considered when interpreting the findings and their generalizability to primary neutrophil biology.

While mature neutrophils primarily rely on glycolysis for ATP production, they retain notable metabolic flexibility. Activated neutrophils exhibit increased glycolysis and pentose phosphate pathway (PPP) activity to support rapid responses.⁴⁷ However, they can also store LDs and utilize fatty acids through lipolysis and autophagy, processes implicated in granulopoiesis and tumor progression.^29,48 Under glucose-limited conditions, neutrophils engage alternative pathways, including gluconeogenesis, glycogenesis, and FAO, to sustain function.^49–51 In tumor microenvironments, metabolic stress can further drive FAO-dependent programs that enhance ROS production, neutrophil extracellular trap (NET) formation, and metastasis.^52,53 Together, these studies underscore lipid metabolism as a key regulator of neutrophil effector function.

Our findings extend this framework by demonstrating that stage-specific inhibition of ATGL-mediated LD degradation significantly enhances extracellular H₂O₂ production in mature neutrophils. LDs in leukocytes contain neutral lipid cores composed predominantly of triglyceride^48,54,55. ATGL is the principal lipase responsible for hydrolyzing triglycerides to release free fatty acids, thereby regulating lipid substrate availability for lipid mediator synthesis and cellular stress signaling^19,56,57. Consistent with a role for ATGL-dependent lipid mobilization in neutrophil redox regulation, myeloid-specific ATGL deficiency has been reported to impair the neutrophil respiratory burst by disrupting lipid mediator availability⁵⁶. Studies in oxidative tissues further indicate that ATGL-mediated lipolysis regulates fatty acid oxidation through PPARα-dependent transcriptional programs, thereby influencing mitochondrial and peroxisomal metabolism as well as cellular redox homeostasis.^58,59 In macrophages, ATGL deficiency increases ROS production in association with elevated NOX1 expression, whereas activation of the ATGL-PPARα axis enhances antioxidant enzyme expression and suppresses oxidative stress, further linking LD lipolysis to redox regulation in innate immune cells⁵⁹. Together, these findings provide a mechanistic framework for our observation that late-stage ATGL inhibition increases ROS production in both hPSC-derived and murine Hoxb8-derived neutrophils, suggesting that altered ATGL-dependent lipid flux modulates redox-sensitive neutrophil functions. The inhibitors used in this study have been extensively characterized and widely employed to target DGAT1-, DGAT2-, and ATGL-mediated lipid metabolic pathways^60–62. Nevertheless, potential off-target effects or secondary consequences associated with prolonged pharmacological inhibition cannot be completely excluded.

Notably, increased ROS production did not consistently translate into enhanced tumor killing by hPSC-derived neutrophils. In both model systems, extracellular H₂O₂ production was measured as a functional indicator of neutrophil oxidative activity. U87MG cells have been reported to activate antioxidant defense pathways that can buffer extracellular oxidative stress^63–65. Although extracellular H₂O₂ can induce oxidative damage and cell death in glioma cells, these adaptive antioxidant mechanisms may attenuate the cytotoxic effects of neutrophil-derived H₂O₂. In addition, neutrophil-mediated tumor cytotoxicity is multifactorial and is not determined solely by oxidative output. Beyond ROS-dependent injury, neutrophils employ additional effector mechanisms, including degranulation, the release of granule-associated cytotoxic proteins, and direct interactions with tumor cells. In glioblastoma, tumor-infiltrating neutrophils have been shown to promote tumor necrosis by transferring myeloperoxidase-containing granules into tumor cells, triggering iron-dependent lipid peroxide accumulation and ferroptotic cell death, highlighting a lipid peroxidation-driven mode of neutrophil cytotoxicity beyond the overall level of extracellular H₂O₂ detected in the surrounding medium⁶⁶. Collectively, these findings suggest that the relationship among LD metabolism, oxidative activity, and neutrophil-mediated tumor cytotoxicity is highly context-dependent and governed by coordinated regulation of multiple effector pathways rather than by ROS production alone.

CONCLUSION

Our findings established a comprehensive quantitative LD profile during neutrophil differentiation, demonstrating that LD abundance peaks in early HSPCs and progressively declines during terminal maturation. Notably, pharmacological modulation of LD metabolism did not significantly affect differentiation efficiency under the conditions examined. These findings suggest that dynamic LD remodeling accompanies stage-dependent metabolic reprogramming during neutrophil development but is not, by itself, sufficient to determine lineage progression. Collectively, our findings support the concept that LD abundance reflects changes in cellular metabolic state throughout neutrophil differentiation rather than acting as a direct regulator of differentiation fate.