Chemotherapeutic Tumor Microparticles Elicit a Neutrophil Response Targeting Malignant Pleural Effusions

Despite adaptive antitumor immunity, the mobilization of the innate immune system in cancer immunotherapies is lagging. Neutrophils, the most abundant innate immune cells, represent 50% to 70% of all leukocytes, and more than 10¹¹ neutrophils undergo death daily in our body (1). Maintaining this innate cell death without activation is presumed to prevent ubiquitous infections. However, this physiologic process presents an opportunity to modulate neutrophils to exert an antitumor effect before their death. Indeed, antitumor responses from neutrophils can be elicited by Coley toxin and Bacillus Calmette-Guérin in certain patients with cancer (2, 3). However, neutrophils can also be polarized to become tumor promoting in particular tumor microenvironments, where factors such as inhibitory cytokines, hypoxia, low pH, and potassium overabundance can redirect neutrophils, thus limiting the treatment efficiency (4). Studies have revealed the double-edged sword role of neutrophils in tumor microenvironments (5, 6) but mobilizing neutrophils in cancer immunotherapy remains a challenge.

Solid tumor cells are likely to metastasize to the pleura through the bloodstream and invade the visceral pleura, resulting in an effusion in the pleural cavity, referred to as malignant pleural effusion (MPE; refs. 7, 8). About 40% of patients with lung cancer develop MPE (9). Patients with refractory MPE commonly suffer dyspnea, followed by chest discomfort, coughing, and poor quality of life and may have a reduced life span. Patients with lung cancer with MPE display a decreased rate of survival compared with patients without pleural effusions (9, 10). Despite the progress in cancer treatment, the current management of MPE remains palliative. Thus, approaches that can efficiently target MPE with minimal side effects are highly desirable. MPE is stemmed from the damaged blood vessels, where a relatively physiologic state with average oxygen pressure, pH value, and electrolyte balance may be present. The increased capillary permeability may facilitate the entry of leukocytes into the effusions. Thus, we asked whether we could use this to mobilize neutrophils into malignant effusions where the neutrophils could avoid malignant reeducation and exert an antitumor effect.

Tumor cells are able to release extracellular vesicles, which are labeled as tumor microparticles (T-MP; ref. 11). We previously reported that T-MPs can stimulate DCs using the cGAS/STING pathway and function as a vaccination platform (12, 13), but such T-MPs also stimulate M2 macrophage polarization to exert a tumor-promoting effect (14). We also found that T-MPs can function as natural carriers to deliver chemotherapeutic drugs or oncolytic viruses to tumor cells, effectively killing malignant cells (15, 16). T-MPs, packaged with chemotherapeutic drugs, can reprogram M2 macrophages toward M1 phenotype (17). Thus, the incorporation of drugs into T-MPs may yield chemo-immunotherapeutic, dual-functional MPs. In this study, we provide evidence that methotrexate (MTX)-packaging T-MPs recruited neutrophils to the pleural cavity where they exerted a therapeutic effect against MPE in patients with cancer.

The clinical trial was registered in the Chinese Clinical Trial Register (no. ChiCTR-ICR-15006304), as a multicenter, randomized, double-blind, parallel-treatment study. The trial was conducted in accordance with the Declaration of Helsinki, the relevant International Conference on Harmonization Good Clinical Practice guidelines, and all applicable regulatory and ethical requirements. The Clinical Trials Ethics Committee of the Huazhong University of Science and Technology approved the clinical trial [NO: (2015) 0702-2]. Sixty-two patients, who completed the whole treatment schedule (Supplementary Table S1), were used for sample collection. The main inclusion criteria were age 18 to 70 years and diagnosed with nonsquamous non–small cell lung cancer (NSCLC) with primary malignant pleural effusion and pathologically confirmed nonsquamous advanced NSCLC with pleural effusion cytology showing tumor cells. Main exclusion criteria included pregnancy, lactation, allergy to multiple drugs, severe comorbidities, psychologic diseases, severe infection including pleural infection, previously received chemotherapy or local intrathoracic administration, and participation in other clinical trials within the past 3 months. All patients were recruited from the Hubei Cancer Hospital or Union Hospital affiliated to Tongji Medical College of Huazhong University of Science and Technology. Patients were received pemetrexed (500 mg/m², Nanjing Pharmaceuticals) and cisplatin (DDP, 75 mg/m², Qilu Pharmaceutical) intravenous administration at day 1 in 21-day cycles according to National Cancer Comprehensive Network guidelines, and treated with 50 mL MTX tumor cell–derived microparticles (MTX-MP; 5 × 10⁷ MPs containing 25 μg MTX preparation discussed below) or 50 mL saline at days 5, 7, 9, 11, 13, and 15 by intrathoracic (i.t.) injection every other day (QOD; see schedule in Fig. 1A). MPE was collected at day 5 (before MTX-MP treatment) and day 7 (after first time of MTX-MP treatment) for cell population analysis. MPE (10 mL) was collected at 2, 4, 8, 12, 24, and 48 hours after MTX-MP or saline treatment (n = 6) for chemokine detection. All MPE samples were transferred to the lab for analysis no more than 1 hour after drainage. All patients signed informed consent forms to participate in the study.

Following informed consent, peripheral blood samples (10 mL) were collected from healthy volunteers (n = 9). Blood samples were transferred to the lab for analysis no more than 15 minutes after blood draw.

Female wild-type C57BL/6 and BALB/c mice (6–8 weeks) were purchased from the Center of Medical Experimental Animals of Hubei Province (Wuhan, China). Female nude mice (6–8 weeks) were purchased from Beijing Vital River Laboratory Animal Technology Co. Ltd. C57BL/6 CD45.1 mice (B6.SJL-PtprcaPepcb/BoyJ) were purchased from Peking University Health Science Center. All animal experiments were conducted in accordance with a protocol approved by the Animal Care and Use Committee of Tongji Medical College.

The human tumor cell line A549 was purchased from ATCC. Murine hepatocarcinoma cell line H22, Lewis murine lung carcinoma (LLC) cell line, and human umbilical vein endothelial cell (HUVEC) line CRL-1730 were purchased from the China Center for Type Culture Collection (Beijing, China). A549, H22, and CRL-1730 cells were cultured in RPMI1640 medium (Thermo Fisher Scientific) with 10% FBS (Gibco), and Lewis cells were cultured in MEM (Thermo Fisher Scientific) with 10% FBS (Gibco). Cells were tested for Mycoplasma detection, interspecies cross contamination, and authenticated by isoenzyme and short tandem repeat (STR) analyses in the Cell Resource Centre of Peking Union Medical College before the beginning of the study and spontaneously during the research, and the cells were kept in culture for no more than 10 passages. All cells were cultured in a humidified atmosphere containing 5% CO₂ at 37°C.

MTX-MPs were prepared according to our previously published article (15). In brief, 2 × 10⁸ A549 cells containing 2 mg/mL MTX (Heng Rui Pharmaceutical) were exposed to ultraviolet irradiation (UVB, 300 J/m²) for 1 hour in biosafety cabinet (Thermo Fisher Scientific), and then cultured in RPMI1640 medium for 18 hours. The supernatants were collected for centrifugation: first 5 minutes at 500 × g, and then 2 minutes at 14,000 × g to clear cell debris. The supernatants were then centrifuged for 60 minutes at 14,000 × g to pellet MTX-MPs. The pellets were washed and resuspended in saline (0.8–1.5 × 10⁶/mL) for injection. The size, drug concentration, and sterility of MTX-MPs were analyzed as described previously (18).

For clinical trials, the process was conducted according to the good manufacturing practice (GMP) quality standard. For each preparation, equal amounts of MTX treated the same number of tumor cells under the same conditions. This was done to maintain the stability and reproducibility of MTX-MPs.

CT was carried out at day -1 (before PEM and DDP treatment) and day 43 by Discovery CT750 HD (General Electric Company, GE). MPE volumes were analyzed through three-dimensional imaging by GE Advance Workstation 4.5.

In the ascites model, H22 cells (3 × 10⁵ cells were suspended in 200 μL PBS) were injected into the peritoneal cavity. Three days later, the mice were intraperitoneally injected with MTX-MPs (1 × 10⁷) or neutrophils (1 × 10⁶, bone marrow derived), and the control group was intraperitoneally injected with PBS once per day for 5 days. Part of mice (n = 5) were sacrificed for analysis of ascites volume. The remaining mice (n = 10) were used for long-term survival analysis.

For the establishment of the MPE model, mice were anesthetized using isoflurane before all procedures. LLC (1 × 10⁵ cells suspended in 50 μL PBS) were injected into the pleural cavity through the left 10th or 11th intercostal space at the midaxillary line. The animals were observed until complete recovery. The procedure was not associated with any mortality or morbidity. Three days later, the mice were intrapleurally injected with MTX-MPs (5 × 10⁶) or PBS (50 μL) once per day for 5 days. A portion of mice (n = 5) were sacrificed for analysis of MPE volume. The remaining mice (n = 10) were used for long-term survival analysis.

For neutrophil depletion, mice were intraperitoneally injected with H22 cells, and 3 days later, the mice were intraperitoneally injected with Ly-6G antibody (100 μg per mouse, twice a week, #100763, BioLegend). IgG (#400565, BioLegend) was used as control.

Bromodeoxyuridine (BrdU; 1 mg/20 g) was intravenously injected to BALB/c mice, and 24 hours later, MTX-MPs were intraperitoneally injected. Twenty-four hours later, neutrophils were isolated (as described below) from peritoneal cavity, peripheral blood (retro-orbital collection), and bone marrow (harvested from tibias and femurs) and fixed with 4% polyformaldehyde (PFA). Cells were then washed with Hank balanced salt solution (HBSS), resuspended to 1 × 10⁶/100 μL, and incubated with 0.5% Triton-X 100 (X100PC, Sigma-Aldrich) for 15 minutes at room temperature. After washing with HBSS, cells were labeled with APC-conjugated BrdU (#364114, BioLegend), then analyzed by flow cytometry (BD FACSCanto II), and the data were analyzed by FlowJo (version 7.6.1).

For isolating human neutrophils from MPE and peripheral blood, the MPE or peripheral blood was centrifuged, and the sediments were washed in HBSS three times. Neutrophils were isolated by Ficoll Hypaque solution (#LTS1092P, TBD) according to the manufacturer's instructions. Mouse neutrophils were isolated from mouse bone marrow by using Percoll (#17-0891-01, GE) gradient density centrifugation.

Female wild-type BALB/c mice (6–8 weeks) were sacrificed, and tibias and femurs were harvested. After lysis of red blood cells, bone marrow neutrophils were sorted by MojoSort mouse neutrophil isolation kit (#480058, BioLegend) according to the manufacturer's guidelines.

A total of 2 × 10⁶ neutrophils were isolated from bone marrow of CD45.1 mice and injected to CD45.2 mice via tail vein, followed by intrapleural injection of 1 × 10⁷ MTX-MPs or PBS. Eighteen hours later, the CD45.1 neutrophils in pleural cavity were analyzed by flow cytometry.

Human peripheral blood mononuclear cells (PBMC) were isolated from human peripheral blood using Ficoll density gradient separation. Monocytes were purified by human CD14 Micro-Beads (#130-50-201, MACS), according to the manufacturer's guidelines (∼98% were monocytes), and then cultured in complete RPMI1640 medium containing recombinant human M-CSF (20 ng/mL; #300-25, PeproTech) for the induction of macrophages. Seven days later, human macrophages were harvested and stimulated with A549 cells derived MTX-MPs at a ratio 1:20 (cell:MP) for 2 hours, the using of cells and supernatant described below.

Total cells were collected from MPE, washed with PBS three times, and CD45⁻ cells were purified using human CD45 Micro-Beads (#130-045-801, MACS) according to the manufacturer's guidelines (∼98% were CD45 negative). Cytology examination by hematoxylin and eosin (H&E) staining was performed as described below.

For a phenotype analysis of cells, human cells were stained with surface antibodies: CD3 (#300318), CD4 (#357404), CD8 (#980904), CD11b (#301308), CD15 (#125606), CD66b (#305116), CD68 (#333806), CD56 (#318304), CD45 (#368508), CD44 (#338806), CD62p (#304904), CD62e (#330012), CD162 (#328806), CD62L (#304810), EpCAM (#324204), CD54 (#353120), CD31 (#303106). Mouse cells were stained with antibodies: Ly6G (#127606), NK1.1 (#108707), CD19 (#152403), F4/80 (#123127), CD3 (#100236), CD138 (#142505), PD-1 (#135214), CD45.1 (#110713), CD45.2 (#109806), Ki67 (#151212), CD11b (#101255), IL12 (#505203), or CD31 (#102510).

For intracellular cytokine staining, lymphocytes prepared from mouse peritoneal cavity using CD4/CD8 (TIL) MicroBeads (#130-117-114, MACS) according to manufacturer's guidelines. Cells were restimulated with phorbol 12-myristate 13-acetate (PMA; 20 ng/mL, #P1585, Sigma-Aldrich) and ionomycin (1 μg/mL, #I3909, Sigma-Aldrich) in vitro for 4 hours with monensin (1 μg/mL, #1445481, Sigma-Aldrich), and then stained for surface CD3 and CD8. After surface staining, cells were treated with Fix/Perm solution (#426803, BioLegend) and stained with anti-IFNγ (#505808) and IL10 (#112706). All antibodies were purchased from BioLegend, and flow cytometric was performed with BD FACSCanto II.

Neutrophils (2 × 10⁶) isolated from MPE were cocultured with 1 × 10⁵ CD44-labeled A549 (described above) tumor cells on glass-bottom cell culture dishes (#801001, NEST Biotechnology) for 2 hours, washed with HBSS. One-hundred microliters of HBSS was added, and cells were then incubated with Sytox green (5 nmol/L; #425303, BioLegend) for 30 minutes at temperature. Cells were visualized by two-photon fluorescent microscopy (ZEISS 780). All groups were assessed with the same threshold and parameters.

A total of 1 × 10⁶ neutrophils from healthy mouse bone marrow were cocultured with 1 × 10⁷ Lewis-derived MTX-MPs for 2 hours in the presence or absence of DPI (10 μmol/L, #D2926, Sigma-Aldrich), and NETs were analyzed using Sytox green (#425303, BioLegend) staining. For endothelial cell assays, 1 × 10⁵ HUVECs were labeled with CD31, and seeded on glass-bottom cell culture dishes for 24 hours in RPMI1640 medium with 10% FBS, and neutrophils from patient MPE after MTX-MPs or saline treatment were seeded with the HUVEC layer for 2 hours. The cells were then stained with Sytox green and visualized by two-photon fluorescent microscopy as described above.

Neutrophils isolated from patient peripheral blood (5 × 10⁵) were seeded in the top chamber of 24-well plate with a Millicell Hanging Cell Culture Insert. The bottom chamber was MPE double diluted by RPMI1640 with or without CXCL1 (#MAB275-SP, R&D Systems), CXCL2 (#ab89324, Abcam), GM-CSF (#502319, BioLegend), or IL1β (#MAB601-SP, R&D Systems) neutralization antibody. After 1 hour, cells from the bottom chamber were collected and counted by flow cytometry (Accuri C6, BD Biosciences).

Total RNA extraction was prepared with TRIzol reagent (#15596026, Invitrogen), and cDNA were generated by ReverTra Ace qPCR RT Kit (FSQ-101, Toyobo). Real-time PCR analysis was performed with 2 μg of cDNA as a template. A SYBR Green Mix (Applied Bioscience) and an Agilent Technologies Stratagene M-x3500 Real-time PCR system were used. Relative quantification was performed using 2^−ΔC_t. The procedure was repeated in at last three biologically independent samples. The primer sequences are shown as follows: Arginase-1, 5′-TCATCTGGG TGGATGCTCACAC-3′ (sense) and 5′-GAGAATCCTGGCACATCGGGAA-3′ (antisense); MPO, 5′-GAGCAGGACAAATACCGCACCA-3′ (sense) and 5′-AGAGAAGCCGTCCTCATACTCC-3′ (antisense); NOS2, 5′-GCTCTACACCTCCAATGTGACC-3′ (sense) and 5′-CTGCCGAGATTTGAGCCTCATG-3′ (antisense); NOX2, 5′-CTCTGAACTTGGAGACAGGCAAA-3′ (sense) and 5′-CACAGCGTGATGACAACTCCAG-3′ (antisense); GAPDH, 5′-GTCTCCTCTGACTTCAACAGCG-3′ (sense) and 5′-ACCA CCCTGTTGCTGTAGCCAA-3′ (antisense). Reduced GAPDH was used for normalization.

A549 cells were stained with carboxyfluorescein diacetate succinimidyl ester (CFSE; Sigma-Aldrich, #21888) and cocultured with neutrophils isolated from patient peripheral blood or MPE (48 hours after MTX-MPs treatment) at a ratio of 1:20 for 16 hours, with or without N-acetylcysteine (NAC; 20 mmol/L, #A0737, Sigma-Aldrich), diphenylene iodonium (DPI; 10 μmol/L, #D2926, Sigma-Aldrich), or Cl-amine (10 μmol/L, #S8141, Selleck). CFSE-labeled cells were collected and stained with propidium iodide (PI) and APC-Annexin V (#640932, BioLegend) to assess apoptosis by flow cytometry (BD FACSCanto II) and analyzed by FlowJo 7.6.1.

Hydrogen peroxide (H₂O₂) in MPE patient was detected by a hydrogen peroxide detection kit (#s0038, Beyotime Biotechnology), and ROS in neutrophils was measured using the CellROX probe (#C10422, Thermo Fisher Scientific). Mitochondria-originated ROS was detected with the MitoSOX probe (#M36008, Thermo Fisher Scientific) according to the manufacturer's instructions. The cells were then stained with Hochest (#94403, Sigma-Aldrich) for 1 hour at room temperature and visualized by two-photon fluorescent microscopy, as described above.

A total of 1 × 10⁵ H22 tumor cells were intraperitoneally injected into C57BL/6 mice, and 3 days later, the mice were intraperitoneally injected with 5 × 10⁶ MTX-MPs, Cl-amidine (CLAM; 1 μg/g), or MTX-MPs/CLAM (1 μg/g) daily for 3 days. Normal saline was used as a control. Twenty-four hours after the last treatment, mice were injected with 1% Evans blue (5 mg/kg, #E2129, Sigma-Aldrich) via tail vein injection. One hour after the Evans blue injection, the mice were sacrificed, and 2 mL PBS was used to flush the peritoneal cavity. The flushed fluid was collected and the supernatant was quantified spectrophotometrically by microplate reader (Biotek) according to the standard curve.

Protein from MPE patient neutrophils was extracted, the protein concentrations were determined by BCA Kit (#23235, Thermo Fisher Scientific). Then, the protein (20 μg) was run on an SDS-PAGE and transferred to nitrocellulose membranes. Nitrocellulose membranes were blocked in 5% BSA and probed with anti-human arginase 1 (1:1,000 dilution, Cell Signaling Technology, #79404), anti-human MPO (1:1,000 dilution, Cell Signaling Technology, #79623), anti-human NOX2 (1:1,000 dilution, Cell Signaling Technology, #4312S), iNOS (1:1,000 dilution, Cell Signaling Technology, #39898) or anti-human β-actin antibodies (1:1,000 dilution, Cell Signaling Technology, #4967) overnight at 4°C, followed by a second horseradish peroxidase-coupled antibody incubation at room temperature for 1 hour. Blots were visualized by enhanced chemiluminescence according to the manufacturer's instructions (ECL Kit, Pierce).

MPs were suspended in diluent C solution, and incubated with PKH26 (MINI26, Sigma-Aldrich) for 30 minutes at room temperature followed by quenching with an equal volume of FBS (Gibco) and washing twice with PBS.

To investigate the cellular internalization of MPs, we carried out uptake assays using cells from patients with MPE as recipient cells. Briefly, T cells, B cells, macrophages, neutrophils, and tumor cells in MPE were cocultured with PKH26-labeled MPs for 30 minutes or 2 hours. Cells were then washed with PBS and analyzed by flow cytometry. For in vivo uptake assay, PKH26-labeled MPs were intrapleural or intraperitoneally injected, and 2 hours later, the cells were collected, and uptake of MP was analyzed by flow cytometry.

Human macrophages were harvested and stimulated with A549 cell-derived MTX-MPs at a ratio of 1:20 (cell:MP) for 2 hours. The supernatant was centrifuged at 14,000 × g for 30 minutes to remove MTX-MPs, and then 50 μL supernatant was used for ELISAs.

MPE was collected before and 2, 4, 8, 24, and 48 hours after MTX-MPs or saline treatment. Supernatants were centrifuged at 3,000 × g for 8 minutes, then 50 μL was used for ELISA. CXCL1 and CXCL2 concentrations were assessed using a human GRO-alpha/MGSA (CXCL1) Mini ELISA kit (#900-M38, PeproTech) and human GRO-beta (CXCL2) ELISA Kit (#900-M120, PeproTech) according to manufacturer's instructions.

Neutrophils were collected and fixed in 4% paraformaldehyde (#158127, Sigma-Aldrich) at room temperature for 15 minutes. A total of 1 × 10⁵ cells were smeared on glass slide and dried at 50°C. Hematoxylin staining for 1 minute and eosin staining for 10 seconds were performed, and samples were analyzed by microscope (Leica, DMi8) at 200× and 400× magnification.

Neutrophils were collected and fixed in 4% glutaric dialdehyde solution (#G7651, Sigma-Aldrich) at room temperature for 2 hours. The pellets were then dehydrated in a graded ethanol series, treated with propylene oxide, and embedded with Spurr epoxy resin. Cut sections were stained with uranyl acetate and lead citrate, and then imaged using JEM1010 electron microscope (JEOL).

For scanning electron microscopy assay, 1 × 10⁶ neutrophils were collected from MPE, and fixed with 4% glutaric dialdehyde solution at room temperature for 2 hours. MPE was then incubated with 1% osmium tetroxide and dehydrated with an ascending ethanol series. After critical-point drying, the samples were coated with gold and analyzed by S4800 scanning electron microscopy (Hitachi).

A total of 1 × 10⁵ HUVECs were seeded in the top chamber of 24-well plate with a Millicell Hanging Cell Culture Insert (pore size 0.4 μm). A total of 5 × 10⁵ neutrophils isolated from MPE patient in saline- or MTX-MP–treated groups after were seeded on the HUVEC layer for 2 hours in the presence or absence of 10 μmol/L CLAM. One-hundred microliters of FITC-dextran (1 mg/mL, #FD40, Sigma-Aldrich) was then added into the top chamber. One hour later, FITC-dextran in bottom chamber was detected by Microplate System (Biotek).

A total of 1 × 10⁷ MTX-MPs or PBS were intraperitoneally injected to MPE mice for 18 hours and Sytox green (1:1,000 diluted by PBS, 200 μL) was injected through tail vein for 2 hours. Mice were sacrificed and frozen sections of pleural vessel were stained with PE-anti CD31 (#102407, BioLegend) and analyzed by fluorescence microscope (Leica).

A total of 2 × 10⁵ HUVECs were seeded in 24-well culture plate (Corning) for 48 hours, then 1 × 10⁶ neutrophils were collected from MPE and seeded on HUVEC layer for 24 hours in the presence or absence of 1 × 10⁷ MTX-MPs. Endothelium–neutrophil contact was observed by phase contrast microscope after washing.

All experiments were performed at least three times. Results are expressed as mean ± SEM and were analyzed by Student t test. The survival rates were determined by Kaplan–Meier survival analysis. The P value <0.05 was considered statistically significant. The analysis was conducted using the GraphPad 8.0 software. Sample exclusion was never carried out.

To assess the treatment efficacy of a pleural injection of MTX-MPs, newly diagnosed NSCLC patients with MPE were recruited and received routine systemic chemotherapy in advance. A schematic of the clinical treatment procedure is briefly depicted in Fig. 1A. Because drug-free T-MPs can promote M2 tumor–associated macrophage development (14, 19), we used saline instead of drug-free T-MPs as the control. MTX-MPs were characterized by multiple aspects, including size, plasma membrane origin, mitochondrial and genomic DNA fragments, drug concentration, stability, and sterility (Supplementary Figs. S1 and S2), which was consistent with previous reports (15). The amelioration of the color and turbidity of malignant fluids was observed following the initial treatment (Supplementary Fig. S3). In line with this observation, CD45⁻ tumor cells, which were not mesothelial cells (Supplementary Fig. S4), were efficiently removed from the MPE in the MTX-MP–treated patients, whereas abundant CD45⁺ immune cells were present in the fluids (Fig. 1B and C). However, in the saline-treated patients, both CD45⁻ and CD45⁺ cells were not altered, but CD45⁺ cell numbers slightly decreased after saline treatment (Fig. 1D and E). Four weeks after treatment, analysis of pleural effusions showed that MTX-MP treatment resulted in a significant decrease of MPE compared with the saline treatment (Fig. 1F). Given the presence of overabundant immune cells in the MPE after MTX-MP treatment, we asked which immune cell type(s) were mainly present. The flow cytometric analysis showed that the number of CD4⁺ and CD8⁺ T cells was increased, but the number of CD56⁺ natural killer (NK) cells, CD68⁺ macrophages, and CD11b⁺CD33⁺CD14⁻ myeloid-derived suppressor cells (MDSC) was not altered (Fig. 1G). Robust CD11b⁺CD15b⁺ neutrophils were collected in the effusions following the MTX-MP treatment (Fig. 1H and I). Functionally, CD11b⁺CD15b⁺ neutrophils did not affect T-cell proliferation, whereas CD11b⁺CD33⁺CD14⁻ MDSCs inhibited T-cell proliferation (Supplementary Fig. S5). Forty-eight hours after the initial MTX-MP treatment, the neutrophil number showed a correlation with the reduction of MPE in the patients with lung cancer (Fig. 1J), Together, these results suggested that MTX-MPs efficiently treat MPE, which correlated with CD11b⁺CD15⁺ neutrophil recruitment.

The above treatment efficacy was likely due to direct killing of tumor cells, consistent with our previous report (18). However, recruiting numerous neutrophils to the MPE raised a possible role of neutrophils against MPE. We thus tested this using H22 hepatocellular carcinoma-induced ascites in BALB/c mice to mimic a malignant effusion. Mice were intraperitoneally injected with tumor cells, followed by the intraperitoneal injection of MTX-MPs. We found that a large number of neutrophils were present in the ascites (Fig. 2A). However, we did not find alteration of T cells, B cells, NK cells, and macrophages (Supplementary Fig. S6). Twenty percent of macrophages but not T cells, B cells, NK cells, or neutrophils underwent apoptosis following MTX-MP treatment (Supplementary Fig. S6), which might be due to a high uptake of MTX-MPs by macrophages. BrdU was incorporated into proliferating myeloid progenitors in the bone marrow that subsequently differentiated into neutrophils (Fig. 2B). These recruited neutrophils showed low Ki67 staining (Fig. 2C), suggesting that MTX-MP–recruited neutrophils lost their proliferating capability. We also found that the recruited neutrophils were not MDSCs through H&E staining (Fig. 2D). Also, the recruited neutrophils did not suppress CD8⁺ T-cell proliferation and did not affect the expression of IFNγ, IL10, and PD-1 in the T cells (Supplementary Fig. S7). Using a Ly6G-depleting antibody to deplete neutrophils, we found that the therapeutic effect of MTX-MPs was abrogated, as demonstrated by that the increased volume of ascites and decreased survival (Fig. 2E). We isolated neutrophils from healthy mouse bone marrow and intraperitoneally injected them to the H22 ascites in mice. The ascites volume was 50% reduced and long-term survival was prolonged (Fig. 2F). In line with the Ficoll isolation method, neutrophils isolated with negative magnetic sorting also showed antitumor effects (Supplementary Fig. S8). The intraperitoneally injection of MTX-MPs did not affect the frequency of neutrophils in peripheral blood and bone marrow (Supplementary Fig. S9). This anti-ascites effect of neutrophils was T-cell–independent, because similar results were also obtained in nude mice (Fig. 2G). Consistently, MTX-MP treatment also showed antitumor effects in nude mice (Supplementary Fig. S10). Using an LLC-induced MPE, we further demonstrated that neutrophil depletion impaired the effect of MTX-MPs on the long-term survival of the mice (Fig. 2H). Intravenous injection of CD45.1⁺ neutrophils to CD45.2 mice showed that MTX-MP treatment attracted neutrophils to the pleural effusion (Supplementary Fig. S11). Together, these results suggested that neutrophil recruitment enhanced MTX-MP treatment efficacy.

Next, we investigated how neutrophils were attracted to the MPE. We isolated the cells from the MPE of patients and incubated them with MTX-MPs. We found that MTX-MPs could be taken up by CD68⁺ macrophages, CD11b⁺CD15⁺ neutrophils, and CD45^– tumor cells (Fig. 3A). Macrophages showed the strongest fluorescent intensity (Fig. 3A). These macrophages rather than the neutrophils or tumor cells could take up the MPs within 30 minutes, as demonstrated by the increased mean fluorescence intensity (MFI) of PKH26 (Fig. 3B). Using an LLC-induced MPE or H22-induced ascites model, a consistent uptake by macrophages was observed (Supplementary Fig. S12). We then asked whether the cells following MTX-MP uptake produced neutrophil chemoattractant(s). A transwell assay showed that only the supernatants from a 2-hour coincubation of MTX-MPs with patient MPE macrophages could recruit (Fig. 3C). A 2-hour incubation was based on the observation that the number and proportion of neutrophils in MPEs increased 2 hours after MTX-MP injection (Fig. 3D). GM-CSF and IL1β are known chemoattractants of neutrophils (20, 21). However, the addition of GM-CSF- or IL1β-neutralizing antibody did not affect the above supernatant-mediated attraction of neutrophils (Fig. 3E). Other CXCL chemokines including CXCLs 1–3 and CXCLs 5–8 also attract neutrophils (22, 23). CXCL1 and CXCL2 were upregulated in MTX-MP–treated macrophages (Fig. 3F) and were detected in the supernatants of the 2-hour coincubation (Fig. 3G). In line with these results, clinical MPE samples with a high neutrophil infiltration (n = 6) also showed increased CXCL1 and CXCL2 2 hours after MTX-MP injection (Fig. 3H and I). CXCL1 and CXCL2 were not detected in the macrophages in the MPE before MTX-MP injection, but were induced 2 hours after MTX-MP treatment (Fig. 3J). CXCL1 or CXCL2 neutralization abrogated the attraction of neutrophils to the bottom chamber in the above transwell assay (Fig. 3K). Despite the chemokine-mediated effect, recruiting neutrophils also need endothelial activation, which provides binding molecules for the rolling, tethering, and arrest of neutrophils to the endothelium (23, 24). We found that MTX-MPs not only upregulated CD62e and CD62p in the HUVECs (Fig. 3L), but also upregulated P-selectin glycoprotein ligand 1 (PSGL1) in neutrophils (Fig. 3M). In line with this result, the promotion of neutrophil–endothelial cell contact by MTX-MPs was observed in H22 ascites (Supplementary Fig. S13). Also, MTX-MPs promoted the tight contact of patient MPE-derived neutrophils with endothelial cells in vitro (Fig. 3N). These results suggest that MTX-MPs treatment not only induced macrophages in the MPE to release CXCL1 and CXCL2, but also activated endothelial cells, allowing for neutrophil recruitment.

Neutrophils can have a N1 or N2 phenotype to exert an antitumor or protumor effect (25–27). To investigate this, we analyzed the neutrophils in patients' MPE after MTX-MP treatment. We found that the forward scatter (FSC) of neutrophils was reduced after MTX-MP treatment (Fig. 4A) but was not altered in the saline-treated patients (Fig. 4B). Lower FSC phenotype may reflect an antitumor property of neutrophils (28). In addition to the low FSC, the activated phenotype (CD66b⁺CD62L^low or CD54⁺CD62L^low) might also relate to the antitumor property of neutrophils (29). We found that neutrophils in the MTX-MP–treated MPE upregulated the expression of CD11b, CD66b, and CD54, but downregulated the expression of CD15 compared with the cells before MTX-MP treatment (Fig. 4C). However, neutrophils in the saline group did not show the phenotype alteration (Fig. 4D). In line with the surface markers, oxidative enzymes NADPH oxidase 2 (NOX2), myeloperoxidase (MPO), and inducible nitric oxide synthase (NOS2) were upregulated but arginase 1 was downregulated after MTX-MP treatment (Fig. 4E). Morphologically, a more vacuolated structure and degranulation in MTX-MP–treated neutrophils were observed under transmission electron microscope (TEM; Fig. 4F), and more cytoplasmic azurophilic granules and hypersegmented nuclei were observed (Fig. 4G and H), which are a signature for lysosome activation (28), further suggesting a bias toward an antitumor phenotype (24, 29). Using MTX-MPs to treat neutrophils isolated from the MPE, we found that the surface markers CD11b, CD66b, and CD54 and the enzymes iNOS, NOX2, and MPO were upregulated but CD15 and arginase 1 were downregulated (Fig. 4I and J). Supernatants from the MTX-MP–treated A549 cells could stimulate neutrophils to upregulate CD11b, CD66b, CD54, iNOS, NOX2, and MPO but to downregulate CD15 and arginase 1 (Fig. 4K and L). Together, the injection of MTX-MP not only recruited neutrophils to the MPE of patients but also conferred antitumor properties.

Next, we investigated the antitumor role of attracted neutrophils. We isolated CD45^– tumor cells and neutrophils from the MPE before MTX-MP treatment. We also isolated neutrophils from the MPE after the treatment in the same patient. Then, we coincubated the isolated neutrophils and tumor cells, respectively, and found that only neutrophils from the treatment setting were able to kill tumor cells (Fig. 5A). We further found that such neutrophils also eliminated A549 cancer cells in vitro (Fig. 5B). We observed that the tumor cells were surrounded by neutrophils (Fig. 5C), suggesting that the neutrophil-mediated killing of tumor cells may be in a cell–cell contact manner (28). Such cell–cell contact was also observed in the MTX-MP–treated MPE in patients, but not from the saline-treated MPE (Fig. 5D). ROS is a mechanism neutrophils use to kill tumor cells. In line with the upregulation of iNOS, NOX2, and MPO, ROS expression in neutrophils was elevated in the MPE of patients treated with MTX-MPs, but not in the MPE of the saline group (Fig. 5E). We also found that the H₂O₂ concentration in the MTX-MP–treated MPE of patients was elevated (Supplementary Fig. S14). Both the mitochondria and lysosomes are the main organelles that generate ROS in a cell. However, we only detected minor ROS in mitochondria of the MTX-MP–treated neutrophils by fluorescent microscopy and flow cytometry (Fig. 5F), suggesting that the ROS was mainly produced from the lysosomes. Consistently, the use of ROS scavenger N-acetylcysteine (NAC) or NOX2 inhibitor diphenylene iodonium (DPI) could lower the ROS levels (Supplementary Fig. S15) and also inhibited the killing of tumor cells by neutrophils isolated from clinical MPE samples (Fig. 5G). We found that lysosomal biogenesis was enhanced in MTX-MP–treated neutrophils (Supplementary Fig. S16).

Besides ROS, neutrophils also use neutrophil extracellular traps (NETs) to induce tumor cell NETosis. This NET formation is regulated by ROS, especially from MPO-processed ROS (30), which, however, relies on the NOX2 system-derived H₂O₂ (31, 32). Using an H22 ascites model, we stained the MTX-MP–treated neutrophils with Sytox green, a DNA binding dye, which showed presence of extracellular traps, and 35.8% of neutrophils released NETs (Supplementary Fig. S17). As expected, the blockade of ROS generation resulted in the inhibition of the NET formation in MTX-MP–treated murine neutrophils (Supplementary Fig. S18). We also found NETs surrounding tumor cells in the MTX-MP–treated MPE of patients but not in saline group (Fig. 5H), which was also confirmed by scanning electron microscopy (SEM; Fig. 5I), suggesting that NETs might be involved in the killing of tumor cells. Indeed, the inhibition of NETs by CLAM inhibited the killing of A549 tumor cells by the MTX-MP–recruited neutrophils in vitro (Fig. 5J). Together, these results suggest that neutrophils may use ROS and NETs to kill tumor cells in the MPE of patients following MTX-MP treatment.

The formation of MPE is due to the damage of pleural capillaries caused by tumor invasion and soluble factors, which leads to excessive plasma leakage through hyperpermeable pleural vasculature networks (33). Histologic analysis of the mouse MPE model showed that some vascular endothelial cells from the pleural vasculature underwent apoptosis, and the structure of endothelium was not integrative (Fig. 6A). Regarding filamentous net-like structure and the stickiness of NETs, we further hypothesized that besides the killing of tumor cells, NETs had a repairing effect on the vascular leakage in pleural cavity. To test this hypothesis, we seeded human endothelial cells in the top chamber of transwells, and MTX-MP– or saline-treated neutrophils were placed on an endothelial cell layer. We found that saline-treated neutrophils could not prevent FITC-dextran (molecular 40,000) from entering the bottom chamber by crossing the endothelial layer (Fig. 6B). However, MTX-MP–treated neutrophils effectively prevented this process (Fig. 6C), which was abrogated by CLAM (Fig. 6D). Using fluorescence staining, we found that NETs intertwined with the membrane of endothelial cells in the MTX-MP–treated neutrophil group but not in the saline control group (Fig. 6E). In the MPE mouse model, we also found that Evans blue intravenously injected could be detected in the MPE of the saline-treated mice, but very little detected in the MTX-MP–treated mice (Fig. 6F). Similarly, the administration of NET inhibitors increased the vascular permeability and allowed Evans blue to enter the pleural cavity (Fig. 6G). In line with these results, the frozen section analysis showed that neutrophils anchored to blood vessels and abundant NET were located between the endothelial gap (Fig. 6H). We further evaluated the therapeutic effect of the NETs on MPE by treating the MPE mice with NET inhibitors instead of neutrophil depletion. Under this condition, we found that the therapeutic effect of MTX-MPs on pleural effusions was also inhibited (Fig. 6I) and the mice survival was reduced (Fig. 6J). Together, these results suggest that MTX-MP–recruited neutrophils may release NETs to seal off the damaged endothelium, thus inhibiting plasma leakage.

A way to effectively mobilize endogenous neutrophils and promote their intrinsic antitumor activities as an immunotherapeutic approach is not available. In this study, we showed that tumor cell–derived MPs packaging MTX mobilize neutrophils to the MPE of patients with NSCLC, where the neutrophils display an antitumor phenotype, not only exerting a tumor cell–killing effect but also releasing NETs to seal off the damaged endothelium, leading to the effective treatment of MPE with low adverse effects. During the treatment process, only some patients developed a fever and did not exhibit other symptoms, consistent with our previous reports (18).

Many factors contribute to the chemoattraction of neutrophils to inflammatory sites (20–22, 34). In this study, we found that CXCL1 and CXCL2, released by MTX-MP–stimulated macrophages, played a role in the attraction of neutrophils to the MPE. Macrophages are commonly polarized to tumor-associated M2 macrophages with anti-inflammatory properties in MPEs. It is unclear how MTX-MP treatment switches the M2 phenotype to a proinflammatory M1 with the release of CXCL1 and CXCL2 for neutrophil recruitment. However, our previous studies have shown that drug-packaging tumor cell–derived MPs are able to reset tumor-associated macrophages toward the M1 phenotype (17). Notably, drug-free tumor cell–derived MPs promote M2 tumor–associated macrophage development (14). The key point lies in the very low amount of chemotherapeutic drug in the MPs. Such little drug may not exert a cytotoxic effect but act as a lysosomal regulator because MPs enter lysosomes following their uptake. How drug-packaging MPs regulate macrophage M1 phenotype is under investigation.

In addition to recruitment, the effect of MTX-MPs also confers neutrophils antitumor properties, which seems not to be mediated by MTX-MP–triggered CXCL1 and CXCL2 (35). MTX-MPs can exert an indirect effect on neutrophil activation through the macrophage pathway. Macrophages upon taking up MTX-MPs may release soluble factors, thus inducing neutrophil polarization toward a N1 phenotype. These MTX-MP–triggered neutrophils are not terminally differentiated but are immature cells, allowing BrdU incorporation. Although we propose that these immature neutrophils are mobilized from the bone marrow by MTX-MPs, an alternative possibility is that MTX-MPs induce the differentiation of MDSCs to become neutrophils with the ability to proliferate. In this regard, whether MTX-MPs induce MDSC differentiation to become neutrophils is worthy of investigation.

We found in this study that MTX-MP–recruited neutrophils released NETs in the MPE. NETs comprise a web of fibers composed of chromatin and serine proteases that trap and kill microbes extracellularly (36) and also serve as a physical barrier, preventing the further spread of pathogens. Besides the antimicrobial properties, NETs may have pathologic effects on organ injury such as acute lung injury (37, 38). Studies have also shown a proposed role for NETs in cancer (39). In preclinical models, NETs show a tumor-promoting effect, including facilitating metastasis. However, in this study we found that NETs have a tumor-inhibiting effect. NETs may mobilize their DNA fiber, histone structure proteins, and the released proteases to exert this antitumor effect, similar to their bactericidal process. In the MPE of patients with NSCLC, NETs were found to surround tumor cells, which may prevent tumor cells from migrating and benefit the killing of tumor cells by nearby neutrophils.

Besides the inhibitory effect on tumor cells, NETs are also able to interfere with the vascular leakage, leading to the effective treatment of malignant fluids. The primary cause of MPE lies in the loss of endothelial integrity such as tight junction alteration and endothelial cell apoptosis. However, MTX-MP treatment may enhance the interaction of endothelial cells and neutrophils by upregulating the expression of endothelial E-selectin and P-selectin. Thus, the anchored neutrophils may release and allow NETs to seal off the damaged endothelium. An important issue about NETs is the manner of how MTX-MPs trigger neutrophils to release NETs in the MPE of patients with NSCLC. MTX-MPs may stimulate macrophages to produce CXCL1 and CXCL2. These two chemokines not only attract neutrophils but also facilitate neutrophils to release NETs.

Malignant pleural effusion poses a significant clinical problem with a poor treatment efficacy. However, based on the present findings and our previous studies (15, 17, 18), we propose that drug-packaging MPs may represent a unique approach, which can efficiently treat MPEs in patients with NSCLC through the processes, as shown in Fig. 7. All in all, our data showed that drug-packaging MPs, by virtue of their ability to recruit neutrophils, lead to a robust antitumor innate immunity. These findings provide insights into the versatile antitumor function of neutrophils, especially in the control of malignant fluids.

B. Huang reports a U.S. patent for No. 9,351,931 and a China patent for No. ZL201110241369.8 licensed. These patents, held by Hubei Soundny Bio-Tech Co. Ltd., cover the pharmaceutical preparation for “Tumor cell–derived microparticles packaging of chemotherapeutic drugs.” No potential conflicts of interest were disclosed by the other authors.

P. Xu: Data curation, software, methodology. K. Tang: Data curation, funding acquisition, methodology, writing–original draft, writing–review and editing. J. Ma: Data curation. H. Zhang: Data curation, software, formal analysis. D. Wang: Data curation, software, formal analysis, validation. L. Zhu: Data curation. J. Chen: Data curation. K. Wei: Data curation. J. Liu: Data curation. H. Fang: Data curation. L. Tang: Data curation. Y. Zhang: Data curation. J. Xie: Data curation, project administration. Y. Liu: Methodology. R. Meng: Resources, investigation. L. Liu: Resources, project administration. X. Dong: Resources, project administration. K. Yang: Resources, project administration. G. Wu: Resources, project administration. F. Ma: Resources, project administration. B. Huang: Conceptualization, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing.

This work was supported by the National Natural Science Foundation of China (81788101, 81530080, and 91742112), CAMS Initiative for Innovative Medicine (2016-I2M-1-007), Young Elite Scientists Sponsorship Program by CAST (YESS20160166), Program for HUST Academic Frontier Youth Team (2018QYTD01), and National Major Science and Technology Projects of China (2019ZX09301001).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.