Combination therapy with lipid prodrug liposomes reshapes disease-associated neutrophils to promote the cancer-immunity cycle

Abstract

Neutrophils play a critical role in the cancer-immunity cycle and are associated with poor clinical outcomes. Recent research has primarily focused on the targeted delivery, phenotypic reversal, and reprogramming of tumor-associated neutrophils, while the impact of disease-associated neutrophils (DANs) on antitumor therapy remains understudied. Since liposomes, as drug delivery carriers, possess excellent biocompatibility and stability, making them particularly suitable for combination therapy, we optimized the formulation of asymmetrically branched polyethylene glycol-modified mitomycin C lipid prodrug liposomes (PEG_2,5 K@MLP-L) and prepared hyaluronic acid and sialic acid ester stearate-co-modified dexamethasone palmitate liposomes (HA*SAS@DXP-L) to study DANs in normal, obese, aged, and septic mice. An increase in mitochondria and lysosomes in Ly-6G⁺CXCR2^high DANs accelerated drug clearance, reduced CD3⁺CD8⁺ T cell activity in tumor-draining lymph nodes, and decreased CD8⁺ T cell infiltration in tumors. As the proportion of DANs increased, the efficacy of PEG_2,5 K@MLP-L decreased. Combination therapy with PEG_2,5 K@MLP-L and HA*SAS@DXP-L can reshape DANs, promote the cancer-immunity cycle, and enhance treatment efficacy. This study identifies key characteristics and functions of DANs and presents a promising strategy for improving clinical outcomes.

Graphical Abstract

Introduction

Neutrophils play a crucial role in the immune response to infection and injury [1]. As one of the first cell types to accumulate in damaged tissues, the rapid recruitment of neutrophils is vital for protective functions. Notably, this process is co-opted in pathological conditions, such as cancers, where continuous neutrophil infiltration is associated with sepsis and poor prognosis [2, 3]. Cells recruited in pathological states exhibit heterogeneity and accelerate the clearance of therapeutic agents, a phenomenon that is particularly pronounced in obese and aged patients and is likely related to chronic inflammation induced by obesity and aging [4,5,6,7]. Although extensive research has focused on tumor-associated neutrophils, including their phenotype, function, and numbers [8, 9], neutrophils in the bone marrow have a greater impact on antitumor immunity than highly differentiated neutrophils found within tumors. Changes in bone marrow neutrophils may be key factors influencing clinical outcomes.

The heterogeneity of neutrophils in the bone marrow is characterized by different stages of maturation [4]. Numerous studies have shown that granulocyte progenitor cells sequentially differentiate into precursor and immature cells before maturing into neutrophils, with each cell subtype exhibiting unique functional capabilities [10]. Notably, tumor-induced chronic inflammation promotes the premature egress of these precursor cells into circulation, followed by their migration to extramedullary sites such as tumors and the spleen [10, 11]. Furthermore, during their transit through the bloodstream, neutrophils gradually undergo senescence and can even migrate back to the bone marrow by crossing the bone marrow–blood barrier [12, 13]. This indicates that although neutrophils are among the first responders to diseased tissues, they are also among the first to be adversely affected. This effect can extend to newly formed neutrophils in the bone marrow, thereby contributing to disease progression. These affected newborn neutrophils are referred to as disease-associated neutrophils (DANs).

The development of therapeutic strategies targeting DANs holds promise for enhancing the clinical efficacy of nanoparticles. Lipid prodrugs, known for their ability to combine targeted delivery with low toxicity, have emerged as a highly regarded class of nanomedicines for cancer treatment [14]. In this study, we synthesized a mitomycin C lipid prodrug (MLP) that not only improved the encapsulation efficiency (EE) of liposomes but also stabilized the formulation through strong interactions with the lipid bilayer. To optimize liposome formulation, we utilized artificial intelligence (AI) techniques to successfully produce asymmetrically branched polyethylene glycol (PEG)-modified MLP liposomes (PEG_2,5 K@MLP-L, Fig. 1). These liposomes can undergo thiolysis within tumors, effectively releasing mitomycin C (MMC) at levels nearly equivalent to those of the prodrug [15]. Although PEG-modified liposomes can be influenced by anti-PEG antibodies, this effect can be mitigated by employing dual-layer PEG modifications with varying densities, which reduces the recognition by the immune system and prolongs circulation time in vivo [16, 17]. To further enhance the targeted reshaping of DANs, we developed a novel liposomal formulation co-modified with hyaluronic acid (HA) and sialic acid ester stearate (SAS) (HA*SAS@DXP-L, Fig. 1). The incorporation of HA, a natural ligand for CD44 receptors on neutrophils and other immune cells, combined with SAS, enhances specific binding and internalization by neutrophils, which play a critical role in the tumor microenvironment [18,19,20]. Moreover, the use of liposomal drug delivery systems has been shown to significantly improve the therapeutic outcomes of antitumor treatments by enhancing drug stability, increasing targeting efficiency, and reducing systemic toxicity [21, 22]. This dual modification approach offers a promising strategy to optimize both the pharmacokinetics and therapeutic efficacy of liposomal formulations. In antitumor experiments conducted in healthy, obese, aged, and septic mice (Fig. 1), we observed that the efficacy of PEG_2,5 K@MLP-L was diminished and this reduced efficacy was positively correlated with the proportion of DANs present in the bone marrow. However, the combination therapy of PEG_2,5 K@MLP-L and HA*SAS@DXP-L effectively reshaped neutrophils in the bone marrow, thereby promoting the cancer-immune cycle and achieving significant antitumor effects across all mouse models.

Fig. 1

Preparation of PEG_2,5 K@MLP-L and HA*SAS@DXP-L for treating clinically relevant tumor models. In obesity, aging, and sepsis models, tumor-bearing mice exhibit DAN generation, characterized by Ly-6G⁺CXCR2^high. These cells have increased numbers of mitochondria and lysosomes, which in turn affect T cell activation, resulting in reduced intratumoral infiltration of CD3⁺CD8⁺ T cells. Following phagocytosis by DANs, PEG_2,5 K@MLP-L and HA*SAS@DXP-L can be effectively delivered to the tumor and bone marrow via a neutrophil drug delivery system. In the bone marrow, the lipid prodrug liposomes can reshape neutrophils, thereby enhancing the antitumor immune response. Additionally, the MLP undergoes a thiolytic reaction at the tumor site and reduces to MMC, which directly kills tumor cells and induces immunogenic cell death, thereby promoting the cancer-immune cycle

Results

Compound validation

To validate the synthesized MMC, we labeled its structural components in the naming order (Fig. 2A) and performed analyses using proton nuclear magnetic resonance (¹H NMR), high-resolution mass spectrometry (HRMS), and Fourier-transform infrared (FTIR) spectroscopy. The electrospray ionization (ESI)–HRMS results indicated the following (m/z): 971.46887 [M-H]⁻ and 1007.44586 [M + Cl]⁻. The ¹H NMR spectrum (600 MHz, CDCl₃) exhibited signals at δ 7.496 (d, J = 7.9 Hz, 1H), 7.382 (d, J = 8.1 Hz, 1H), 5.334 (d, J = 5.0 Hz, 1H, 33), 5.063 (s, 1H, 11), 5.032 (d, J = 12.1 Hz, 1H, 9a), 5.001 (d, J = 12.2 Hz, 1H, 9b), 4.744 (dd, J = 10.9, 4.8 Hz, 1H, 8), 4.729 (s, 1H, 18), 4.718 (dtd, J = 11.7, 8.3, 4.3 Hz, 1H, 27), 4.258 (t, J = 10.9 Hz, 1H, 35), 3.636 (dd, 1H, 14), 3.501 (dd, 1H, 15), 3.43 (d, J = 4.5 Hz, 1H, 32a), 3.396 (dd, J = 4.5, 1H, 32b), 3.132 (s, 1H, 16), 2.98 (t, J = 7.2 Hz, 1H, 25), 2.963 (t, J = 7.2 Hz, 2H, 25), 2.653 (t, J = 7.2 Hz, 2H, 26), 2.247 (d, J = 8.0 Hz, 1H, 13), 2.227 – 1.972 (m, 2H, 34), 1.941 – 1.798 (m, 3H, 28, 40), 1.656 (s, 3H, 7), 1.656 – 1.469 (m, 4H), 1.406 – 1.305 (m, 2H), 1.127 (s, 3H, 44), 1.104 – 0.99 (m, 3H), 0.956 (d, J = 12.3 Hz, 3H, 52), 0.853 (dd, J = 6.8, 2.6 Hz, 6H, 51, 53), and 0.647 (s, 3H, 45). Based on these data (Fig. 2B–D, and Supplementary Fig. 1), we confirmed the successful synthesis of MLP. MLP, with a LogP value of 9.799, was found to have regions of positive and negative charges on the surface, predominantly concentrated in the MMC portion with a greater concentration of negatively charged regions (Fig. 2E). We performed thermogravimetric analysis (TGA) to assess MLP stability. As the temperature increased, the compound underwent several changes, including disulfide bond cleavage, cholesterol (CH) melting, CH vaporization, MMC melting, and MMC vaporization (Fig. 2F). This indicates that the MLP was stable and did not degrade below 100 °C.

Fig. 2

Chemical characteristics and molecular simulation results of MLP and SAS. A Chemical structure of MLP and nomenclature of atoms within its structure. Green numbers indicate the naming sequence. HRMS (B), ¹H NMR spectrum (C), and FTIR spectrum (D) of MLP using the same atom nomenclature as that in panel A. E The LogP value of the MLP and distribution of electrostatic potential on its surface, with red indicating positively charged areas and blue indicating negatively charged areas. F TGA curve of the MLP, with red lines indicating mass loss analysis and black lines indicating the rate of mass loss analysis. G Structural formula of the SAS and nomenclature of atoms within its structure. Green numbers indicate the naming sequence. HRMS (H) and ¹H NMR spectrum (I) of the SAS using the same atom nomenclature as that in panel G. Amino acid residue interactions in the three-dimensional structural pocket of L-selectin before (J) and after (K) the cleavage of the ester bond in SAS

Additionally, we verified the synthesis of SAS, labeled it in the naming order (Fig. 2G), and performed HRMS and ¹H NMR analyses. ESI–HRMS m/z: 604.40454[M + H]⁺ and 605.40747[M + 2H]⁺. ¹H NMR (600 MHz, CDCl3): δ 7.496 (d, J = 8.3 Hz, 1H), 3.938 (m, 1H, 6), 3.858 (m, 1H, 7), 3.797 (m, 2H, 12), 2.327 (t, J = 7.3 Hz, 2H, 15), 2.082 (m, 1H, 3), 1.593 (t, J = 6.9 Hz, 2H, 3), 1.239 (s, 30H, 16—30), and 0.855 (t, J = 6.8 Hz, 3H, 31). The absorption peaks at 3447.01 cm⁻¹ and 3344.10 cm⁻¹ were attributed to the stretching vibrations of the amino NH groups. The peaks at 857.48, 802.31, 799.06, and 759.06 cm⁻¹ were assigned to the out-of-plane bending vibrations of C-H bonds in the phenyl ring or the bending vibrations of C–C single bonds. The absorption peaks at 572.68 cm⁻¹ and 515.20 cm⁻¹ were associated with the stretching vibrations of disulfide bonds. Based on these data (Fig. 2H, I, and Supplementary Fig. 2), we confirmed the successful synthesis of SAS. To validate the targeting ability of SAS toward L-selectin, we conducted molecular docking experiments. The results showed that SAS effectively targeted the L-selectin protein, maintaining its targeting efficacy even after ester bond cleavage (Fig. 2J, K). This indicates that SAS possesses significant stability and exhibits effectiveness in targeting L-selectin.

Characterization of lipid prodrug liposomes

To accelerate the development of lipid prodrug liposomes (Fig. 3A) and enhance the accuracy of performance prediction, we employed an AI-assisted liposome formulation screening approach to determine the final formulation based on practical considerations. Liposomes containing phosphatidylcholine (PC) exhibited smaller particle size and higher EE (Table 1). With the exception of the liposomes containing DOTAP, which had a zeta potential of 7.4 mV, the zeta potentials of other phospholipids ranged from − 7 mV to − 16 mV, with the liposomes containing phosphatidylglycerol (PG) having the lowest zeta potential (Table 1). As the CH ratio increased, the EE of the liposomes decreased, and when the CH ratio reached 50%, the particle size increased by more than 50 nm (Tables 2, 3, and Fig. 3B, C). To control the particle size and polydispersity index (PDI), the CH ratio was maintained below 20%. By adjusting the ratio of 1-palmitoyl-2-oleoylphosphatidylglycerol (POPG) to hydrogenated soy phosphatidylcholine (HSPC), we effectively controlled the zeta potential and EE of the liposomes (Table 3, and Fig. 3D, E). The results of the AI simulations indicate that the loading capacity of PC is superior to that of PG; however, our experimental results are contrary to this finding. Hydrogen bonding between 1,2-Diacyl-sn-Glycerol-3-Phospho-[1-rac-glycerol] (EPG) and MLP altered the phase transition temperature of phospholipids, with hydrogen bonds primarily formed between the amino group of MLP and the oxygen atoms of the POPG head group (Fig. 3F, G). This hydrogen bond formation also enhanced the EE of MLP. Consequently, we prepared PEG_2,5 K@MLP-L using EPG, which appeared light purple, whereas HA*SAS@DXP-L was white, with both exhibiting a milky appearance (Fig. 3H). The particle size, PDI, zeta potential, and EE of these two types of liposomes met the standards for injectable-grade liposomes. The prepared lipid prodrug liposomes possessed a phospholipid bilayer and were spherical or quasi-spherical with a particle size of approximately 100 nm, which is consistent with the data in Table 4 (Fig. 3I, J). As shown in Fig. 3K, CYS can promote the drug release from PEG_2,5 K@MLP-L, and this process does not rely on liposome rupture. This characteristic significantly enhances MLP release efficiency and effectively overcomes the common issue of slow drug release in traditional drug delivery systems. Additionally, Fig. 3K indicates that the slightly acidic tumor microenvironment can facilitate the DXP release from HA*SAS@DXP-L, further improving DXP release efficiency and addressing the slow release problem commonly encountered in conventional drug delivery systems.

Fig. 3

Characteristics of PEG_2,5 K@MLP-L and HA*SAS@DXP-L. A 3D structural model of PEG_2,5 K@MLP-L and HA*SAS@DXP-L. The characteristics of PEG_2,5 K@MLP-L via AI-assisted design: B particle size, C PDI, D EE, and E zeta potential. Higher abundance indicates a greater drug-to-lipid ratio. The areas highlighted within circles represent preferred criteria. F Differential scanning calorimetry curves of the physical mixture of MLP and EPG (MLP + EPG) and the electrostatic complex (MLP@EPG). G Molecular interactions between MLP and POPG, with blue bonds indicating hydrogen bonds. H Photograph of 1 mL PEG_2,5 K@MLP-L and HA*SAS@DXP-L in a sample vial. I Transmission electron microscopy (TEM) image of PEG_2,5 K@MLP-L and HA*SAS@DXP-L, scale bar = 100 nm. J Cryo-TEM image of the liposomes, scale bar = 100 nm. K Cumulative drug release curve of PEG_2,5 K@MLP-L and HA*SAS@DXP-L under various conditions, n = 3

Table 1 Artificial intelligence–assisted design of PEG-modified MLP liposomes incorporating different phospholipid types

Table 2 Artificial intelligence-assisted design of PEG-modified MLP liposomes incorporating varying phospholipid and cholesterol ratios. The characteristics include particle size, PDI, zeta potential, and encapsulation efficiency

Table 3 Artificial intelligence-assisted design of PEG-modified MLP liposomes incorporating varying hydrogenated soy phosphatidylcholine (HSPC), dioleoylphosphatidylglycerol (DOPG), and cholesterol (CH) ratios

Table 4 Particle size, PDI, zeta potential, and encapsulation efficiency of liposomes

Neutrophils accelerate the clearance of drugs from the blood

We investigated the effect of neutrophils on the in vivo fate of the developed drugs using pharmacokinetic studies (Fig. 4A). Stimulation with type II collagen resulted in significant swelling of the rat paws, with volumes notably larger than those in healthy rats (Fig. 4B). Furthermore, the serum levels of tumor necrosis factor-α (TNF-α) and interleukin-1β (IL-1β) in the inflamed rats were significantly elevated compared to those in healthy rats (Fig. 4C), confirming inflammation. In our pharmacokinetic studies, we observed that following the injection of HA*SAS@DXP-L, the drug was rapidly cleared from the bloodstream of inflammatory rats, with a reduction in half-life of approximately 35% and a decrease in the area under the curve (AUC) of approximately 32% (Table 5, Fig. 4D). Similarly, the injection of PEG_2,5 K@MLP-L led to a 66% reduction in the half-life and a 59% decrease in the AUC (Table 6, Fig. 4E). However, the combined injection of HA*SAS@DXP-L and PEG_2,5 K@MLP-L effectively eliminated the reduction in both half-life and AUC caused by inflammation (Table 5). This suggests that the drug clearance rate is accelerated in an inflammatory environment. However, this phenomenon can be mitigated by combination therapy. In a separate analysis of neutrophils, our findings showed that inflammation induced the rapid maturation of bone marrow neutrophils, characterized by a more pronounced tri-nuclear structure and an increase in the number of lysosomes (Fig. 4F). Additionally, the expression was CXCR2 on the surface of these cells (Fig. 4G, I). Importantly, these neutrophils exhibited extended lifespans, resulting in significantly longer survival (Fig. 4H). These findings indicated that DANs play a key role in accelerating drug clearance in inflammatory environments.

Fig. 4

Inflammation-induced neutrophils accelerate drug clearance in rats. A Establishment of an inflammation model in rats for pharmacokinetic studies. B Changes in paw volume during rearing, n = 6. C TNF-α and IL-1β levels in the serum of rats, n = 3. Changes in D PEG_2,5 K@MLP-L and E HA*SAS@ DXP-L concentration in the blood following injection, n = 3. F Neutrophils in bone marrow were accompanied by laser confocal images, and red fluorescence indicates lysosomes stained with Lyso-Tracker, scale bar = 10 μm. G Expression of CXCR2 protein on the surface of neutrophils in bone marrow. H Survival rate of neutrophils in bone marrow within 24 h, n = 3. I The neutrophils in bone marrow were accompanied by laser confocal images, and green fluorescence indicates lysosomes stained with CXCR2, scale bar = 100 μm. In the laser confocal images, blue fluorescence indicates DAPI-stained nuclei

Table 5 Pharmacokinetic parameters of HA*SAS@DXP-L

Table 6 Pharmacokinetic parameters of PEG_2,5 K@MLP-L

Antitumor treatment study in normal mice

We established a C57BL/6 J tumor-bearing mouse model by inoculating B16 tumor cells and administering a combination therapy of PEG_2,5 K@MLP-L and HA*SAS@DXP-L to evaluate their antitumor efficacy (Fig. 5A). In the in vitro cell experiments, the cell death rate in media containing cysteine (CYS) was significantly higher than that in media containing glutathione (GSH, Fig. 5B). Moreover, cells in CYS-containing medium exhibited stronger fluorescence intensity of the high-mobility group box 1 (HMGB1) protein (Fig. 5C), suggesting an increased occurrence of immunogenic cell death. These findings indicate that compared to GSH, CYS is more effective in reducing MLP to MMC, thereby enhancing the therapeutic efficacy of PEG_2,5 K@MLP-L. In 3D tumor spheroid infiltration assays, CYS facilitated deeper penetration of PEG_2,5 K@MLP-L into the tumor cells (Fig. 5D). Most importantly, tumor homogenate significantly enhanced permeability compared with amino acids. This enhanced infiltration is likely linked to the reduction of MLP to MMC. Following the 22-day antitumor treatment period, there were no significant differences in body weight among groups except for the control group, which showed a slight increase in body weight, indicating that the liposomes exhibited low toxicity (Supplementary Fig. 3). Injection of HA*SAS@DXP-L (N-HA*SAS@DXP-L) did not demonstrate significant antitumor efficacy (Fig. 5E, F). In contrast, mice receiving injections of PEG_2,5 K@MLP-L (N-PEG_2,5 K@MLP-L) and those receiving combined therapy (N-Combination) exhibited significantly reduced tumor volumes, with all mice achieving no tumor growth after reduction in volume. Neutrophil counts in the peripheral blood of all treatment groups showed no significant changes, whereas neutrophil counts in the peripheral blood of the control group began to increase significantly after day 16 (Fig. 5G). This indicates that neutrophils require sufficient stimulation to respond to tumor immune responses. Notably, the N-PEG_2,5 K@MLP-L and N-combination groups exhibited reduced neutrophil counts in the tumor-draining lymph nodes, while CD8⁺ T cells were fully activated and increased in number, facilitating the infiltration of CD8⁺ T cells within the tumor (Fig. 5H). Under the cytotoxic effects of PEG_2,5 K@MLP-L and the infiltration of CD8⁺ T cells, the tumors began to undergo extensive apoptosis, releasing antigens (Fig. 5I). This process contributes to the establishment of the cancer-immune cycle. In contrast, HA*SAS@DXP-L did not show any significant advantages in healthy mice.

Fig. 5

Combined therapy in normal tumor-bearing mice. A Establishment of a normal tumor-bearing mouse model followed by a 22-day antitumor treatment study. B Cytotoxic effect of PEG_{2, 5 K}@MLP-L on tumor cells and C expression of surface HMGB1 protein induced by GSH and CYS. D Effect of PEG_2,5 K@MLP-L on the expression of surface HMGB1 protein in cells within 3D tumor spheroids induced by GSH, CYS, and tumor homogenate. E Changes in tumor volume of the illustrated mice over time, with black arrows indicating the times of drug administration, n = 6. F Photographs of tumor tissue from healthy mice on the last day of antitumor treatment observation. G Changes in the number of peripheral blood neutrophils in healthy tumor-bearing mice. H Immunofluorescence sections of tumor-draining lymph nodes in the treatment group, where red fluorescence indicates cytotoxic T cells (CD8), green fluorescence marks neutrophils (Ly-6G), with a scale bar of 200 μm. I TUNEL staining sections of tumor tissues, where red fluorescence labels apoptotic cells (TUNEL), with a scale bar of 100 μm. In the laser confocal images, blue fluorescence indicates DAPI-stained nuclei

Neutrophil counts in the peripheral blood of all treatment groups showed no significant changes, whereas neutrophil counts in the peripheral blood of the control group began to increase significantly after day 16 (Fig. 5G). This indicates that neutrophils require sufficient stimulation to respond to tumor immune responses. Notably, the H-PEG_2,5 K@MLP-L and H-combination groups exhibited reduced neutrophil counts in the tumor-draining lymph nodes, while CD8⁺ T cells were fully activated and increased in number, facilitating the infiltration of CD8⁺ T cells within the tumor (Fig. 5H). Under the cytotoxic effects of PEG_2,5 K@MLP-L and the infiltration of CD8⁺ T cells, the tumors began to undergo extensive apoptosis, releasing antigens (Fig. 5I). This process contributes to the establishment of the cancer-immune cycle. In contrast, HA*SAS@ DXP-L did not show any significant advantages in healthy mice.

Antitumor treatment study in obese mice

To investigate the role of DANs in antitumor therapy, we fed mice with a high-fat diet and established tumor models (Fig. 6A). By the 10th week of feeding, the CH levels and body weights of the obese mice were significantly higher than those of healthy mice (Fig. 6B). Obese mice had a greater number of adipocytes, and their diameters were larger (Fig. 6C). We used Ly-6G and CXCR2 antibodies to label neutrophils in the bone marrow and found that the proportion of Ly-6G⁺CXCR2^high neutrophils in the obese mice increased by 16% (Supplementary Fig. 4 and Fig. 6D). This indicates that long-term high-fat feeding leads to significant fat accumulation in obese mice, resulting in adipose tissue expansion. This triggers chronic inflammation and subsequently affects the normal maturation of neutrophils in the bone marrow, leading to the production of DANs. The liposomes did not exhibit any toxic effects in obese mice (Supplementary Fig. 5). After discontinuing the injections of PEG_2,5 K@MLP-L, there was a higher proportion of Ly-6G⁺CXCR2^high neutrophils in the bone marrow of obese mice (O-PEG_2,5 K@MLP-L), and tumor regrowth was observed, showing significant differences compared to the neutrophil proportion and tumor volumes in the combined therapy group (O-Combination, Supplementary Fig. 5 and Fig. 6E, F). Mice treated with PEG_2,5 K@MLP-L exhibited CD8⁺ T cells accumulation surrounding the tumor tissue; however, these T cells were trapped by neutrophils at the tumor periphery (Fig. 6G). In contrast, T cells in the O-combination group infiltrated the tumor more deeply (Fig. 6G). This suggests that obesity-induced DANs inhibit T cell-related tumor immune responses, thereby reducing the antitumor efficacy of PEG_2,5 K@MLP-L, whereas the combined therapy mitigates this effect by reshaping neutrophils in the bone marrow. Finally, we observed organelles in Ly-6G⁺ neutrophils in the bone marrow and found that Ly-6G⁺CXCR2^high neutrophils had significantly higher numbers of mitochondria and lysosomes than Ly-6G⁺CXCR2^low neutrophils (Fig. 6H). In summary, Ly-6G⁺CXCR2^high neutrophils, characterized by a higher abundance of lysosomes and mitochondria, are considered key factors influencing the efficacy of antitumor therapy.

Fig. 6

Combined therapy in obese tumor-bearing mice and impact of obesity-induced DANs on the antitumor efficacy of PEG_2,5 K@MLP-L. A Establishment of an obese tumor-bearing mouse model followed by a 22-day antitumor treatment study. B Changes in body weight and CH concentration in the blood of obese mice during the rearing process. The dashed line represents body mass, and the solid line indicates cholesterol concentration. C Oil Red O staining of fat from obese mice after 10 weeks of feeding, scale bar = 100 μm. D Proportion of Ly-6G⁺CXCR2^high neutrophils in the bone marrow of obese mice. The boxes in the scatter plot indicate Ly-6G⁺ cells, while the black peak in the histogram represents Ly-6G⁺CXCR2^low, and the pink peak represents Ly-6G⁺CXCR2^high. E Changes in tumor volume in obese mice, with black arrows indicating the times of drug administration. F Photographs of tumor tissue from obese mice on the last day of antitumor treatment observation. G Cryo-TEM images of Ly-6G⁺ neutrophils in the bone marrow of obese tumor-bearing mice, scale bar = 2 μm. H Post-treatment immunofluorescence sections of tumors from obese mice, where green fluorescence indicates neutrophils labeled with Ly-6G, and red fluorescence indicates T cells labeled with CD8, scale bar = 1000 μm

Antitumor treatment study in aged mice

In our study on antitumor therapy in obese mice, we found that DANs significantly influenced the tumor immune response. Notably, obesity is reversible, indicating that the effects of DANs can be eliminated. Based on this, we designed an antitumor study targeting 16-month-old mice (Fig. 7A). As age increases, chronic inflammation in the body generally increases, a phenomenon known as "inflammaging.” This inflammatory process is typically irreversible. Over the feeding period, the body weight of aged mice continued to increase, whereas the number of neutrophils in the peripheral blood slightly decreased and remained within the normal range (Fig. 7B). Importantly, collagen fibers in aged mice became thinner, suggesting that cells in the bone marrow were more susceptible to external disturbances (Fig. 7C). The proportion of Ly-6G⁺CXCR2^high neutrophils in the bone marrow of aged mice was higher than that in normal and obese mice (Supplementary Figs. 7, 8, and 7D), further supporting this observation. Additionally, the number of immune cells in the lymph nodes of aged mice was reduced (Fig. 7E). This indicates that although the number of immune cells decreases with age, the aging rate of DANs is relatively slow.

Fig. 7

Combined therapy in aged tumor-bearing mice and the impact of aging-induced DANs on the antitumor efficacy of PEG_2,5 K@MLP-L. A Establishment of an aged tumor-bearing mouse model followed by a 22-day antitumor treatment study. B Changes in body mass and neutrophil counts in the blood of aged mice during the rearing process. The dashed line represents fitted data, and the solid line corresponds to measured data. C Masson's trichrome staining of femur sections from aged mice, with eosin staining revealing muscle fibers and aniline blue staining revealing collagen fibers. D Proportions of Ly6G⁺CXCR2^high neutrophils in the bone marrow of aged mice, where the black peak in the histogram represents Ly6G⁺CXCR2^low neutrophils and the pink peak corresponds to Ly6G⁺CXCR2^high neutrophils. E Hematoxylin and eosin (H&E) stained sections of lymph nodes from aged mice, scale bar = 200 μm. F Changes in tumor volume of aged mice, with black arrows indicating the times of drug administration. G Photographs of tumor tissue from aged mice on the last day of antitumor treatment observation. H Proportion of CD3⁺CD8⁺ T cells in tumor-draining lymph nodes, with FITC-conjugated CD3 labeling lymphocytes and PE-conjugated CD8 labeling T cells. I Post-treatment immunofluorescence sections of tumors from aged mice, where red fluorescence indicates neutrophils labeled with Ly-6G, scale bar = 500 μm

During the 22-day antitumor treatment, the toxicity of liposomes to the liver, spleen, skin, and fat did not increase with age of the aged mice; liver inflammatory cell infiltration was only observed in the control group (Supplementary Fig. 9 and 10). In aged mice from all treatment groups except the combined therapy (A-Combination) group, the proportion of DANs was relatively high and tumor growth was not significantly inhibited (Supplementary Fig. 11 and Supplementary Fig. 9F and G). In the A-combination group, the proportion of Ly-6G⁺CXCR2^high neutrophils in the bone marrow of aged mice decreased, whereas the number of CD3⁺CD8⁺ T cells in the tumor-draining lymph nodes increased (Fig. 7H), indicating that Ly-6G⁺CXCR2^high neutrophils have an inhibitory effect on CD3⁺CD8⁺ T cells. These neutrophils can infiltrate the tumor, kill tumor cells by releasing neutrophil extracellular traps, and form tertiary lymphoid structures within the tumor, demonstrating promising antitumor therapeutic outcomes (Fig. 7I). In conclusion, reshaping Ly-6G⁺CXCR2^high neutrophils not only activates T cells, but also enables newly recruited neutrophils to directly attack tumor cells.

Antitumor treatment study in septic mice

When evaluating tumor immunity in obese and aged mice, we discovered that chronic inflammation induces DAN production. Notably, reshaping these neutrophils could eliminate their immunosuppressive effects. To investigate the role of acute inflammation-induced DANs, we established a mouse model of sepsis by the intraperitoneal injection of lipopolysaccharides (Fig. 8A). Treatment with HA*SAS@DXP-L resulted in the recovery and stabilization of body weight and temperature in septic mice (Fig. 8B). However, inflammation persisted in these mice, leading to the accumulation of neutrophils in the liver (Fig. 8C). Compared with normal, obese, and aged mice, we found the highest number of Ly-6G⁺CXCR2^high neutrophils in the bone marrow of septic mice (Supplementary Fig. 12 and Figs. 6D, 7D, and 8D). This suggests a close correlation between the generation of DANs and the intensity of inflammation; the stronger and longer the inflammation, the higher the proportion of DANs. Among all experimental groups, only septic mice in the control group showed a significant increase in body weight, which was associated with larger tumor volumes in this group (Supplementary Fig. 13). In contrast, the tumor volume in the combined therapy (S-Combination) group of septic mice remained at zero, demonstrating superior performance compared to the other treatment groups (Fig. 8E, F). Simultaneously, the proportion of Ly-6G⁺CXCR2^high neutrophils in the bone marrow of the S-combination group decreased, while the number of tumor-infiltrating neutrophils increased, and the number of liver-infiltrating neutrophils decreased, alleviating liver inflammation symptoms. In contrast, the other treatment groups exhibited significant liver inflammation (Supplementary Fig. 14 and Fig. 8G). This indicates that reshaping neutrophils is effective against acute inflammation. Through comparative analysis of the proportion of Ly-6G⁺CXCR2^high neutrophils in the bone marrow of different tumor model mice with their respective tumor volumes, we found a positive correlation between the proportion of these neutrophils and tumor volume. Tumor growth was completely inhibited when the proportion fell below 10% (Fig. 8H). In both chronic inflammatory states, such as obesity and aging, and acute inflammatory states, such as sepsis, reshaping DANs not only modulates the inflammatory response but also effectively suppresses tumor growth. These findings provide new insights for future antitumor therapies.

Fig. 8

Combined therapy in septic tumor-bearing mice and effect of sepsis-induced DANs on the antitumor efficacy of PEG_2,5 K@MLP-L. A Establishment of a septic mouse model followed by a 22-day antitumor treatment study. B Changes in body weight and temperature in septic mice during treatment, with the red line indicating temperature and the black line representing body weight. C Immunofluorescence staining of liver sections from mice after 10 days of treatment, with Ly-6G labeling neutrophils in red, scale bar = 50 μm. D Proportions of Ly-6G⁺CXCR2^high neutrophils in the bone marrow of mice. The boxes in the scatter plot indicate Ly-6G⁺ cells, while the histogram shows black peaks representing Ly-6G⁺CXCR2^low neutrophils and pink peaks representing Ly-6G⁺CXCR2^high neutrophils. E Changes in tumor volume in mice over the treatment period, with black arrows indicating times of drug administration. F Photographs of tumor tissue from septic mice on the last day of antitumor treatment observation. G Immunofluorescence sections of tumors and livers from mice following treatment, alongside H&E-stained liver sections. Ly-6G-labeled neutrophils appear in red, scale bar = 100 μm. H Pie chart displaying the proportion of DANs in the bone marrow, accompanied by a corresponding heatmap of tumor volumes. In the heatmap, deeper blue indicates larger tumor volumes, while lighter blue denotes smaller tumor volumes. Other cells refer to those in the bone marrow with a density comparable to that of DANs

Discussion

The application of AI technology to screen liposome formulations has significantly improved the efficiency, development costs, and time to market for new drugs [23]. In our study, we used AI to evaluate the formulations of PEG_2,5 K@MLP-L, which indicated that phospholipids of the PC and PG types were more suitable for the preparation of PEG_2,5 K@MLP-L (Tables 1–4 and Fig. 3B–E). While artificial intelligence (AI) has made significant strides in recent years, it still has several limitations. One major drawback is its reliance on large amounts of data for training, which can introduce biases if the data is not representative of diverse populations. To overcome the limitations of AI technology, we optimized the formulation of PEG_2,5 K@MLP-L and selected EPG as the phospholipid for encapsulating MLP. Compared with MMC, MLP exhibits a larger logP value, which enhances their integration into the liposomal membrane. Within this membrane, the MMC moiety of MLP is primarily situated within the glycerol backbone and polar headgroup regions of phospholipids [24]. The negatively charged head group of EPG formed electrostatic interactions with MMC (Fig. 3G), thereby stabilizing it within the phospholipid membrane [24]. Furthermore, the π–π stacking interactions between the planar aromatic rings of MMC contribute to the formation of highly ordered domains enriched with the prodrug in the liposomal membrane. Additionally, the unsaturated alkyl chain of EPG facilitates the insertion of the CH moiety of MLP. Classical linear PEG-modified liposomes can induce the generation of anti-PEG antibodies, resulting in liposomal leakage and drug clearance. However, branched PEG effectively mitigates this issue [17, 18]. Consequently, we synthesized PEG_2,5 K@MLP-L. Additionally, we developed HA*SAS@DXP-L, which specifically targets L-selectin receptors on neutrophils (Fig. 2J and K), thereby enhancing the specificity of anti-inflammatory drugs toward these cells. Anti-inflammatory drugs can prevent or delay cancer onset and enhance the efficacy of conventional treatment methods [25].

Neutrophils migrate from the bone marrow to sites of disease, such as cancer and inflammation, through the bloodstream, and subsequently return to the bone marrow. This process occurs repeatedly, and ultimately induces DAN production. These DANs can suppress the immune system and promote disease progression, thereby creating a disease-immunity cycle. The physiological environment of normal mice does not accurately simulate the conditions required for tumor growth, because the immune cell population in the body exerts an inhibitory effect on tumors. When the healthy immune system recognizes tumor cells, it recruits neutrophils [26]. These healthy neutrophils can directly attack tumor cells and perform antigen presentation, thereby triggering both antigen- and non-specific T-cell responses [27, 28]. Reciprocal recruitment and activation of T cells and neutrophils resulted in the significant accumulation of CD3⁺CD8⁺ T cells in the tumor-draining lymph nodes of mice (Fig. 5H). CD8⁺ T cells are critical for immunotherapy; their abundance supports the cancer-immunity cycle and correlates with favorable patient prognosis [29]. This explains why PEG_2,5 K@MLP-L alone and combination therapy with PEG_2,5 K@MLP-L and HA*SAS@DXP-L effectively inhibited tumor growth in normal mice (Fig. 5E, F). Importantly, interactions between neutrophils and T cells vary across different tissue types and disease states [30]. Differences in neutrophil counts between healthy and diseased mice may be one of the primary reasons for the low clinical translation rates of nanomedicines.

Several cancers are closely linked to infection, chronic stimuli, and inflammation [31, 32]. To better reflect the true pathological processes, we established tumor models that closely resembled those of clinical patients using obese, aged, and septic mice. As living standards have improved globally, greater access to nutrient-rich food has contributed to a growing population of obese individuals. Obesity results in fat accumulation (Fig. 6C), which reduces the activity of immune cell receptors and triggers reversible chronic inflammation [33,34,35]. By contrast, aging is an inevitable process that leads to an imbalance in the ratio of neutrophils to T cells and an increase in inflammatory cytokines, which contribute to irreversible chronic inflammation [36]. In contrast to chronic inflammation, sepsis induces a rapid and intense inflammatory response that can cause significant damage (Fig. 8C). Indeed, many patients with COVID-19 succumbed to sepsis. Sepsis leads to the overexpression of PD-L1 in neutrophils, inhibiting T cell activation, and inducing apoptosis, thereby generating a robust inflammatory response [37, 38]. This acute inflammatory reaction results in the substantial dissemination of neutrophils throughout various tissues, leading to systemic inflammation. Such a pro-tumor inflammatory environment can contribute to poor prognoses [39]. Aging, obesity, and sepsis increased the proportion of DANs, which possessed a greater number of mitochondria and lysosomes (Figs. 4H, 6D, H, 7D and 8D). Neutrophil mitochondria play a crucial role in regulating the formation of neutrophil extracellular traps, as well as processes such as cell adhesion, migration, respiratory burst, development, differentiation, and cell death [40]. Consequently, DANs exhibited longer lifespans, enhanced phagocytic abilities, and improved migratory capacities than conventional neutrophils (Fig. 4C, D).

DANs generated under various disease stimuli suppress T-cell activation and reduce immune cell infiltration into tumors, thereby promoting disease progression. Throughout disease progression, new DANs are continuously created, creating a mutually reinforcing relationship between DANs and the disease. Simply reshaping neutrophils in the bone marrow or inhibiting tumors does not yield optimal therapeutic outcomes. In this study, we utilized a neutrophil drug delivery system to transport both PEG_2,5 K@ MLP-L and HA*SAS@DXP-L to tumors and bone marrow. This combination therapy not only kills tumor cells, activates T cells, and recruits neutrophils but also blocks tumor-induced stimulation of neutrophils in the bone marrow. Additionally, this therapy aims to reshape neutrophils in the bone marrow, enabling these newly formed neutrophils to attack tumor cells and present antigens to T cells, thereby promoting the cancer-immunity cycle. This study elucidated the production, characteristics, roles, and impacts of DANs on antitumor therapy. Furthermore, we developed PEG_2,5 K@ MLP-L and HA*SAS@DXP-L to reshape DANs and induce immunogenic cell death in tumor cells via neutrophil drug delivery system, providing potential targets for future immunotherapy development. These findings will contribute to improving the clinical outcomes of cancer patients and enhancing the clinical translation rate of lipid prodrug liposomes.

Conclusions

In conclusion, this study highlights the pivotal role of DANs in modulating the cancer-immunity cycle and their potential as a target for cancer immunotherapy. The research demonstrates that reshaping neutrophils in the bone marrow, specifically by reducing the proportion of Ly-6G⁺CXCR2^high neutrophils, can significantly enhance the antitumor immune response. The developed lipid prodrug liposome-based combination therapy effectively reprograms neutrophils, promoting tumor-infiltrating neutrophils while alleviating inflammation in non-target organs, such as the liver. These findings provide a promising strategy to improve the clinical outcomes of cancer patients by modulating both the inflammatory response and tumor progression, contributing to the future development of more effective immunotherapeutic approaches.

Methods and materials

Material

We designed and guided Meidi Xisheng Biopharmaceutical (Hangzhou) Co., Ltd. in the synthesis of MLP, Dexamethasone Palmitate (Purity ≥ 98%) was purchased from Shanghai Hekang Biotechnology Co., Ltd., Prednisolone (Purity ≥ 97%), LysoTracker Red DND-99, Coenzyme Q, CCK-8 Assay Kit, Glutathione, Cysteine, penicillin/streptomycin solution, DMEM medium, fetal bovine serum, and TUNEL Apoptosis Detection Kit were purchased from Dalian Meilun Biotechnology Co., Ltd., DSPE-mPEG_2,5 K was purchased from Xi'an Ruixi Biotechnology Co., Ltd. Hyaluronic Acid was purchased from Huaxi Furuida Biopharmaceutical Co., Ltd., Sialic Acid was purchased from Shanghai Macklin Biochemical Technology Co., Ltd. Stearic Acid was purchased from Shanghai Aladdin Biochemical Technology Co., Ltd., (2,3-Dioleoylpropyl)-trimethylammonium Chloride, Hydrogenated Soy Phosphatidylcholine (HSPC), Cholesterol, and EPG were purchased from AVT (Shanghai) Pharmaceutical Tech Co., Ltd., Bovine Type II Collagen was purchased from Chondrex, USA, Mouse and Rat HMGB1 Enzyme-Linked ImmunoSorbent Assay Kit and Calcein/PI cell assay kit were purchased from Biyuntian Biotechnology Co., Ltd., Rat Serum IL-1β and TNF-α ELISA Kit were purchased from Beijing Dakowei Biotechnology Co., Ltd., High-Fat Diet was purchased from Beijing Huafukang Biotechnology Co., Ltd., Lipopolysaccharide was purchased from Sigma, USA, H&E and Oil Red O Staining Solution were purchased from Wuhan Saiweier Biotechnology Co., Ltd., 4',6-Diamidino-2-phenylindole (DAPI), purchased from Sigma, USA. Mouse Peripheral Blood Neutrophil and Lymphocyte Isolation Kit were purchased from Tianjin Haoyang Biological Products Technology Co., Ltd., FITC-labelled Ly-6G Antibody (Cat. 108,405), FITC-labelled CD3 Antibody (Cat. 100,203), APC-labelled CD8 Antibody (Cat. 100,707), and PE-labelled CXCR2 Antibody (Cat. 124,307) were purchased from BioLegend, USA, Wistar Rats (Female) were purchased from Liaoning Changsheng Biotechnology Co., Ltd. C57BL/6 J Mice (Female) were purchased from Beijing VitoLiHua Experimental Animal Technology Co., Ltd.

Synthesis and characterization of compounds

In a 100 mL three-necked flask, 9.2 mmol of sialic acid ethyl ester, 0.92 mmol of DMAP, and 30 mL of anhydrous pyridine were added. In an ice water bath, a solution of 10.2 mmol of octadecanoyl chloride in dichloromethane was gradually added over 10 min. The reaction proceeded at room temperature for 10 h. Once the reaction was complete, 50 mL of distilled water was added, and the mixture was extracted with dichloromethane to obtain the SAS. The resulting samples were analyzed by HRMS (Waters Corporation, USA) and ¹H NMR spectroscopy (Agilent Technologies, USA). The custom MLP were characterized using ¹H NMR spectroscopy, HRMS, FTIR spectroscopy (Bruker Corporation, USA), thermogravimetric analysis (Nippon Instruments Corporation, Japan), and differential scanning calorimetry (Shimadzu Corporation, Japan).

Computer-aided drug design

Molecular docking was performed using AutoDock Vina to analyze the interaction between SAS and L-selectin and identify the optimal conformation, which was then saved in the PDB format. The docking results were visualized using PyMOL software. Subsequently, the surface electrostatic potential and LogP values of the MLP were calculated. Furthermore, molecular interaction simulations between MLP and EPG were performed using AutoDock Vina, followed by visualization with PyMOL.

AI screening of liposomal formulations

AI was used to screen the liposome formulations. The types of phospholipids investigated are listed in Table 1, the ratios of phospholipids to CH are presented in Table 2, and the ratios of HSPC:DOPG:CH as study variables are detailed in Table 3. The following parameters were set for the program: the ratio of phospholipids to CH to mPEG2000-DSPE was maintained at 85:10:5, the drug-to-lipid ratio was set to 3:10, the reaction temperature was 55 °C, ethanol was used as the solvent, and the thin-film hydration technique was used as the liposome preparation method. The reaction time was fixed at 30 min, the ultrasonic treatment was disabled, and a 100 nm microfiltration membrane was selected for the process. The AI was then used to calculate the particle size, PDI, zeta potential, and EE of the liposomes.

Preparation and characterization of liposomes

Prepare liposomes using the modified ethanol injection method. The liposomal membrane materials and drug were weighed according to the formulations shown in Table 4, and 10% (v/v) anhydrous ethanol was added to reach the final volume of the formulation. The mixture was stirred in a water bath at 55 °C until fully dissolved. Stirring was continued to evaporate most of the anhydrous ethanol. Following this, sterilized water for injection at 55 °C was added to the membrane materials, and stirring was maintained for an additional 20 min to obtain the initial liposome product. The initial product was then subjected to ultrasonication and subsequently filtered sequentially through 0.80, 0.45, and 0.22 µm microfiltration membranes to yield the final liposomes. The particle size, PDI, and zeta potential of the liposomes were measured using a Malvern particle size analyzer (PSS Company, USA), and liposome morphology was observed using TEM (ZEISS, USA) and cryo-TEM (JEOL, USA).

EE of liposomes

A 100 μL aliquot of the liposome was transferred to a 10 mL volumetric flask and diluted to the mark with methanol, followed by thorough mixing. A precise 20 μL aliquot was then taken for injection, and the peak area was recorded as A_before. Subsequently, the liposomes were filtered through a 0.45 μm aqueous microfiltration membrane and subjected to the same procedure. Another 20 μL aliquot was obtained for analysis, and the peak area was recorded as A_after. The EE was calculated as follows:

$${\text{EE}}\% = \,\left( {{{A_{{{\text{before}}}} } \mathord{\left/ {\vphantom {{A_{{{\text{before}}}} } {A_{{{\text{after}}}} }}} \right. \kern-0pt} {A_{{{\text{after}}}} }}} \right) \times 100\%$$

In vitro release of liposomes

An aliquot of the liposome was transferred to a dialysis bag and placed in a dissolution cup containing 40 mL release medium (PBS solution containing 5% Tween 80) [41]. Continuous stirring was performed under light-protected conditions at 37 ± 1 °C. At 1, 2, 4, 6, 12, and 24 h, 1 mL of the dialysis solution was collected, and an equal volume of the release medium at the same temperature was added to maintain the volume. The drug concentration (C) was determined by high-performance liquid chromatography (HPLC, Agilent Technologies, USA) from the peak area using a standard curve equation for calculations. The cumulative degree of release (R_n) was calculated as follows:

$${R}_{n}=\left({C}_{n}{V}_{o}+{\sum }_{n-1}^{n}{C}_{n-1}V\right)/{M}_{t}\times 100\text{\%}$$

Confocal imaging of B16 cells

The tumor cell suspension was added to a 96-well plate and incubated for 24 h. Following this incubation, the cells were placed in a culture medium (this includes 89% DMEM cell culture medium, 10% fetal bovine serum, and 1% penicillin/streptomycin solution) supplemented with GSH and CYS, along with PEG_2,5 K@MLP-L (MLP: 10 μg/mL). After a 12-h treatment, the resulting spheroids were stained with HMGB1 and calcein/PI. Finally, the prepared samples were imaged using confocal laser microscopy (Nikon Corporation, Japan).

3D tumor spheroid experiment

The tumor cell suspension was added to a 96-well plate containing agarose gel, and after a 5-day incubation, tumor spheroids were formed. These spheroids were then placed in a culture medium supplemented with GSH, CYS, and TH, along with PEG_2.5 K@MLP-L. After a 12-h treatment, the spheroids were stained using the HMGB1 kit. Finally, the prepared samples were imaged using confocal laser microscopy.

Establishment of an inflammatory rat model

Eighteen Wistar rats, each weighing 180–220 g, were randomly divided into two groups of nine. One group received subcutaneous injections of type II collagen into the foot, which were maintained for nine days. During this period, foot volumes were measured daily.

Pharmacokinetic experiments

Nine inflamed and nine healthy rats were randomly divided into six groups with three rats in each group, resulting in three inflamed and three healthy groups. Each type of rat was assigned to one of three injection groups: injection of 0.9% saline, injection of lipid prodrug liposomes (HA*SAS@DXP-L or PEG_2,5 K@MLP-L), injection of a mixture of liposomes (HA*SAS@DXP-L and PEG_2,5 K@MLP-L). All injections were administered via intravenous injection at a dose of 10 mg/kg. Blood samples were collected at 2, 5, 15, and 30 min and 2, 6, and 12 h post-administration, and the plasma was separated. Coenzyme Q10 and prednisone solutions were added to plasma as internal standards. The drugs were extracted from the plasma using isopropanol and methanol, and the drug concentrations in the plasma were determined using HPLC.

Establishment of an obese mouse model

Sixty C57BL/6 J mice were randomly divided into two groups of 30 mice each. The mice in the first group were fed a standard diet, whereas those in the second group were fed a high-fat diet. The body weight of each mouse was measured regularly during the feeding period. On day 70, the axillary fat was collected from the mice for processing. Axillary fat was embedded, sectioned, and stained with Oil Red O to observe changes in adipose tissue.

Establishment of an aged mouse model

Thirty eleven-month-old mice were purchased from Beijing Weitong Lihua Co., Ltd. and housed for 5 months. During the housing period, body weights of the mice were recorded daily. When the mice reached 16 months of age, their tibias were harvested for further analysis. The tibias were decalcified, followed by embedding, sectioning, and Masson's trichrome staining.

Establishment of a sepsis mouse model

Thirty mice received an intraperitoneal injection of 10 mg/kg lipopolysaccharide solution. One hour after injection, HA*SAS@DXP-L was administered via the tail vein. The mice were housed until their body weights and temperatures returned to normal. During this period, body weight and temperature of the mice were recorded daily.

Pharmacological efficacy experiments

A total of 24 normal mice, 24 obese mice, 24 aged mice, and 24 septic mice were randomly divided into 4 groups, resulting in 24 total groups with 6 mice in each. Each mouse was assigned to one of four treatment groups that received intravenous injections of 0.9% saline, HA*SAS@DXP-L (dose: 5 mg/kg), PEG_2,5 K@ MLP-L (dose: 20 mg/kg) [42], or a mixture of liposomes (HA*SAS@DXP-L and PEG_2,5 K@MLP-L). The tumor volume and body weight of the tumor-bearing mice were measured every three days. At the end of the experiment, tumor tissues were collected from the mice for imaging and subjected to immunofluorescence and H&E staining.

Isolation of neutrophils from bone marrow

First, blood was collected from mice using an anticoagulant, and the number of neutrophils was determined using a blood cell analyzer (Mindray, China). Next, the femurs and tibias of mice were dissected to collect cell aggregates from the bone marrow. A neutrophil isolation kit was used to extract neutrophils from the bone marrow, were placed in a culture medium (this includes 89% DMEM cell culture medium, 10% fetal bovine serum, and 1% penicillin/streptomycin solution), which were subsequently labeled with phycoerythrin-conjugated anti-CXCR2 and fluorescein isothiocyanate-conjugated anti-Ly-6G monoclonal antibodies. The labeled cells were sorted and analyzed by flow cytometry (BD Company, USA). Finally, images of the neutrophils were captured using cryo-transmission electron microscopy and confocal microscopy. The survival rate of the isolated Ly-6G⁺CXCR2^high neutrophils was assessed after 24 h using the CCK-8 assay kit.

Isolation of T cells from tumor-draining lymph nodes

The tumor-draining lymph nodes of the mice were excised, and cell aggregates were collected. Subsequently, T cells within the tumor-draining lymph nodes were labeled with PE-conjugated anti-CD3 and FITC-conjugated anti-CD8a monoclonal antibodies. The labeled cells were sorted and analyzed using flow cytometry.

Statistical analysis

The statistical analysis was performed using the software Prism 8 and RStudio. Data are presented as mean ± standard deviation. The P value was determined using SPSS 23.0 software (IBM Corp., IBM SPSS Statistics 23.0, Armonk, NY, USA). One-way ANOVA followed by a post hoc test was used for multiple group comparisons. Statistical significance was set at *P < 0.05, **P < 0.01, ***P < 0.001.