Deubiquitinase USP48 Promotes Gastric Cancer Peritoneal Metastasis by Stabilizing S100A9 to Drive Neutrophil Extracellular Traps Formation

ABSTRACT

Peritoneal metastasis (PM) represents the primary cause of death in patients with advanced gastric cancer (GC), yet its underlying mechanisms remain poorly understood. Neutrophil extracellular traps (NETs) have emerged as key mediators of tumor progression, but how cancer cells regulate NETs formation to facilitate PM is unknown. This study investigates the role of the deubiquitinating enzyme USP48 in this process. USP48 expression was analyzed in GC tissues from patients with or without PM. To evaluate the functional role of USP48 in NETs formation and metastasis, we employed a range of in vitro and in vivo models, including co‐culture systems, subcutaneous xenografts, and peritoneal dissemination models. Molecular interactions were examined through co‐immunoprecipitation, ubiquitination assays, and protein stability analysis. USP48 was significantly overexpressed in GC tissues, with the highest levels observed in PM samples, correlating with poor prognosis. The proliferation, migration, invasion, and adhesion to peritoneal mesothelial cells of gastric cancer (GC) cells were all inhibited upon USP48 knockdown, the effects rescued by NETs‐conditioned medium. Mechanistically, USP48 directly interacted with and deubiquitinated the alarmin S100A9, enhancing its stability and secretion. The receptor for advanced glycation end‐products (RAGE) and toll‐like receptor 4 (TLR4) in neutrophils were subsequently activated and induced NETs formation by S100A9 in neutrophils. In vivo, knockdown of USP48 suppressed tumor growth, NETs formation, and PM, while S100A9 overexpression reversed these effects. Our findings unveil a novel USP48/S100A9 axis through which GC cells drive NETs formation to promote PM. This pathway may serve as a promising therapeutic strategy for the prevention and treatment of PM in GC.

USP48 is specifically upregulated in GC tissues with PM and acts as a central driver of metastasis by stabilizing the alarmin S100A9. This stabilization initiates a molecular cascade involving TLR4/RAGE activation, which in turn induces NETs formation, thereby creating a pro‐metastatic niche that promotes tumor growth and peritoneal dissemination.

Abbreviations

ANC

absolute neutrophil counts

ANOVA

one‐way analysis of variance

BSA

bovine serum albumin

CCK‐8

Cell Counting Kit‐8

cDNA

complementary DNA

CHX

cycloheximide

CitH3

citrullinated Histone H3

Co‐IP

co‐immunoprecipitation

DUBs

deubiquitinating enzymes

ELISA

enzyme‐linked immunosorbent assay

FBS

fetal bovine serum

GC

gastric cancer

HPMCs

human peritoneal mesothelial cells

IF

immunofluorescence

IHC

immunohistochemistry

MPO

myeloperoxidase‐DNA

NETs

neutrophil extracellular traps

PM

peritoneal metastasis

qRT‐PCR

quantitative real‐time PCR

RAGE

receptor for advanced glycation end‐products

ROS

reactive oxygen species

SD

standard deviation

STAD

stomach adenocarcinoma

TCGA

the cancer genome atlas

TLR4

Toll‐like receptor 4

USP

ubiquitin‐specific protease

WB

western blot

1. Introduction

Gastric cancer (GC) remains a major global health burden, with persistently high incidence and mortality rates worldwide [1, 2]. Despite advances in therapeutic strategies, the prognosis for patients with advanced GC, particularly those presenting with peritoneal metastasis (PM), remains exceedingly poor [3]. PM is a lethal manifestation of GC that significantly diminishes survival outcomes and limits treatment efficacy [4, 5]. The tumor microenvironment serves as a critical determinant in promoting the progression of GC and enabling metastatic dissemination [6]. Tumor‐derived factors polarize neutrophils in the tumor microenvironment into N2 tumor‐associated neutrophils (TANs), which subsequently promote tumor growth, metastasis, angiogenesis, therapy resistance, and immunosuppression [7]. This complex network, composed of diverse cellular components, signaling molecules, and extracellular matrix, actively participates in tumor initiation, local invasion, and the establishment of distant metastases, including PM [4].

In recent years, neutrophil extracellular traps (NETs) have emerged as a critical factor in cancer biology [8]. Although first characterized as a host defense mechanism against pathogens, NETs are now understood to play multifaceted roles in various pathological conditions, including cancer [8, 9]. In the context of malignancy, NETs contribute to tumor proliferation, invasion, and metastatic spread [10, 11]. They also interact with other pathways involved in inflammation and immune regulation, thereby shaping a pro‐tumorigenic milieu [12, 13]. Specifically in GC, recent evidence highlights the involvement of NETs in the process of PM, underscoring their potential as a therapeutic target [14]. Neutrophil recruitment to tumor tissue promotes tumor metastasis [15]. The degree of neutrophil infiltration in peritoneal disseminated lesions is a prognostic factor in advanced ovarian cancer [16]. Neutrophil infiltration of premetastatic niches is involved in seeding formation.

A key mediator linking inflammation, neutrophil activation, and NETs formation is S100A9, a calcium‐binding protein often associated with myeloid cell function [17]. S100A9 is known to regulate neutrophil recruitment and activation, including processes such as degranulation [18]. Through activation of the RAGE/TLR4–ROS (receptor for advanced glycation end‐products/toll‐like receptor 4–reactive oxygen species) signaling axis, S100A9 has been demonstrated to induce NETs formation, consequently establishing a tumor‐supportive microenvironment that facilitates cancer cell proliferation and metastatic spread [19]. This pathway may represent a critical bridge between chronic inflammation, NETs release, and metastatic progression in GC.

The regulation of protein homeostasis by deubiquitinating enzymes (DUBs) represents another crucial layer of control in cancer pathogenesis [20]. DUBs are involved in maintaining cellular protein balance and modulating key processes linked to tumorigenesis and metastasis [21]. Among them, ubiquitin‐specific protease 48 (USP48), a member of the USP subfamily, has drawn increasing attention. Although its functions remain incompletely characterized, USP48 is commonly dysregulated across multiple cancer types and displays functions that vary depending on the context, either promoting or suppressing tumor progression in different cancer types [22, 23, 24]. Its potential interplay with NETs‐related pathways in GC warrants further investigation. USP48 was identified as a deubiquitinating enzyme targeting HMGA2. Overexpression of USP48 significantly increased HMGA2 protein levels by blocking its degradation. Conversely, USP48 depletion accelerated HMGA2 degradation and consequently inhibited colorectal cancer invasion and metastasis [25]. USP48 directly binds to and deubiquitinates Gli1 in glioblastoma cells. Depletion of USP48 consequently suppresses cell proliferation and downregulates Gli1 target gene expression, leading to reduced glioblastoma tumorigenesis [26]. These lines of evidence collectively prompted us to prioritize USP48 for further mechanistic investigation in the context of GC.

Given the established roles of NETs in GC peritoneal metastasis and the regulatory functions of DUBs like USP48 in tumor biology, together with the S100A9‐mediated signaling that drives NETosis, we hypothesize that a mechanistic link exists between USP48, S100A9, and NETs formation in the context of GC PM. Our previous study found that USP48 bound to SA009. We hypothesize that USP48 regulates the deubiquitination of SA009, thereby modulating its stability. This regulation of SA009 by USP48 may further influence NETs formation and ultimately contribute to the PM of GC. This study aims to investigate whether USP48 modulates S100A9 expression or function, thereby influencing NETs formation and subsequent peritoneal dissemination in gastric cancer. Understanding this axis may offer new insights into the molecular mechanisms driving PM and identify potential therapeutic targets to improve outcomes for patients with advanced GC.

2. Materials and Methods

2.1. Clinical Sample Collection

40 GC patients who underwent surgical resection at the hospital were consecutively enrolled in this study. Prior to surgery, none of the patients underwent neoadjuvant chemotherapy or radiotherapy. Patients were further stratified into two groups based on intraoperative and postoperative histopathological confirmation of PM: the PM group (n = 20) and the non‐PM group (n = 20). PM was diagnosed following the Japanese Classification of Gastric Carcinoma (15th edition) and confirmed by cytological or histological examination of peritoneal lavage fluid or metastatic nodules. For the control group, normal gastric mucosal tissues and peripheral blood samples were procured from 40 healthy volunteers matched for both age and sex, who underwent routine gastroscopic examination at our hospital's endoscopy center during the same period. These individuals had no history of gastrointestinal diseases, malignancies, or systemic inflammatory conditions. Normal mucosal tissues were biopsied from the gastric antrum under endoscopic guidance, and blood samples were collected concurrently. All control samples were processed and stored using the same protocols as the patient‐derived samples. This study was approved by the Ethics Committee, and prior to sample collection, written informed consent was obtained from every participant.

2.2. Analysis of Protein Ubiquitination Levels by Co‐Immunoprecipitation (Co‐IP)

The protein concentration of each sample was measured with a BCA protein assay kit (Pierce). For each sample, 500 μg of total protein lysate was pre‐cleared with Protein A/G Agarose beads (Santa Cruz) at 4°C for 1 h. Pre‐cleared lysates were incubated overnight at 4°C with 2 μg of anti‐ubiquitin antibody (Cell Signaling Technology) or normal mouse IgG (negative control). Immune complexes were captured with Protein A/G beads (4°C, 4 h), washed five times with cold lysis buffer, and eluted by boiling in 2× Laemmli buffer. Eluted proteins were separated by SDS‐PAGE, transferred to PVDF membranes, and probed with the same anti‐ubiquitin antibody (1:1000). Input controls were probed with anti‐GAPDH antibody. Bands were visualized using enhanced chemiluminescence (Bio‐Rad) and quantified with ImageJ. Ubiquitination levels were normalized to the corresponding β‐actin signal in the input lysate.

Anti‐USP48 or normal IgG antibody was used to incubate whole cell lysates for Co‐IP experiments overnight at 4°C, followed by incubation with Protein A/G beads. The bound proteins were eluted and analyzed by western blotting using anti‐S100A9 antibody.

For ubiquitination analysis (IP), cells were transfected with HA‐Ubiquitin plasmid and incubated with MG132 (20 μM) for 6 h. To detect ubiquitinated S100A9, cell lysates were subjected to immunoprecipitation with an anti‐S100A9 antibody, followed by immunoblotting of the precipitates with an anti‐HA antibody.

2.3. Quantitative Real‐Time PCR (qRT‐PCR)

Total RNA was extracted from frozen specimens using TRIzol (Invitrogen). RNA concentration and purity were measured with a NanoDrop 2000 (Thermo Fisher). One microgram of total RNA was reverse‐transcribed into cDNA using PrimeScript RT Master Mix (Takara Bio) in a 20 μL reaction. mRNA expression of selected USP family members was quantified by qRT‐PCR using SYBR Green Premix Ex Taq (Takara Bio) on a QuantStudio 5 system (Applied Biosystems). Primer sequences are listed in Table S1. Each reaction was run in triplicate. Cycling conditions: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s, with melting curve analysis for specificity. Expression levels were normalized to β‐actin and calculated using the 2^−ΔΔCt method, with normal gastric mucosa as the baseline.

2.4. The Cancer Genome Atlas (TCGA) Database

Using the TCGA portal (https://portal.gdc.cancer.gov/), information on USP48 expression in GC was retrieved from the TCGA database. We gathered clinicopathological information and expression patterns from patients with stomach cancer. Tumor and normal tissues were evaluated for USP48 expression levels, and the relationship with clinical characteristics such as tumor grade and stage was examined. We obtained a visual graph of the aforementioned data via the UALCAN database. Using the kmplot database, we evaluated the prognostic significance of USP48 expression in patients with gastric cancer. Based on the median expression value, patients were categorized into high and low USP48 expression groups, and overall survival was compared between the two groups using the log‐rank test.

2.5. Immunofluorescence Staining

Following deparaffinization and rehydration, tissue sections (4 μm thick) underwent antigen retrieval in citrate buffer (pH 6.0) within a microwave oven. The sections were then blocked with 5% bovine serum albumin (BSA) for 1 h at room temperature, followed by overnight incubation at 4°C with primary antibodies diluted in blocking buffer. To visualize NETs, a rabbit monoclonal anti‐citrullinated histone H3 (CitH3) antibody (Abcam, ab5103) was used as the primary antibody at a 1:200 dilution, and mouse monoclonal anti‐CD66b (BioLegend, 305102; 1:100 dilution), as well as rabbit monoclonal anti‐Ly‐6G antibody (BioLegend, 16‐9668‐82; 1:100 dilution) was used to identify neutrophils. After washing, sections were incubated for 1 h at room temperature with corresponding secondary antibodies: Alexa Fluor 594‐conjugated goat anti‐rabbit IgG (Invitrogen, A‐11012; 1:500) and Alexa Fluor 488‐conjugated goat anti‐mouse IgG (Invitrogen, A‐11001; 1:500). Nuclei were counterstained with DAPI. Negative controls were performed by omitting the primary antibodies. Fluorescence images were captured using a confocal microscope (Zeiss LSM 880). Five randomly selected, non‐overlapping high‐power fields (HPF, 400× magnification) were analyzed per tissue section. ImageJ software (NIH) was used to quantify the positive rate of CD66b or Ly‐6G (green channel) and CitH3 (red channel) within each field.

2.6. Measurement of Myeloperoxidase‐DNA (MPO‐DNA)

MPO‐DNA level was detected using a commercially available ELISA kit. Serum samples (diluted 1:2) were added to a microplate pre‐coated with an anti‐MPO antibody and incubated. Following the washing step, the DNA component of the captured MPO‐DNA complexes was detected by adding a peroxidase‐conjugated anti‐DNA antibody. A colorimetric substrate was then used to develop the reaction, and the absorbance was measured at 450 nm using a BioTek microplate reader. Using the provided standards, a standard curve was generated, and the MPO‐DNA concentration in each sample was calculated and expressed as ‐OD value.

2.7. Serum Absolute Neutrophil Counts (ANC)

ANC in peripheral blood was obtained from routine complete blood count analysis performed at the hospital's clinical laboratory using an automated hematology analyzer (Sysmex XN‐9000). The ANC data were retrieved from the medical records of the enrolled participants, corresponding to the blood draw time point for serum collection.

2.8. Pearson Correlation Analysis

Pearson correlation analysis was applied to evaluate the linear associations between USP48 expression and each of the four neutrophil/NETs‐related parameters.

2.9. Human Neutrophils Isolation

Neutrophils were isolated from healthy donor peripheral blood by density gradient centrifugation using Ficoll‐Paque PLUS (GE Healthcare), then subjected to dextran sedimentation and hypotonic lysis to remove remaining red blood cells. Wright‐Giemsa staining confirmed neutrophil purity exceeding 95%, while trypan blue exclusion demonstrated viability above 98%. The isolated neutrophils were resuspended in RPMI‐1640 containing 10% fetal bovine serum (FBS, Gibco) and 1% penicillin–streptomycin, and incubated at 37°C under 5% CO₂ for 24 h. After incubation, the culture supernatant was obtained by centrifugation at 300× g for 10 min, passed through a 0.22 μm sterile filter, and stored at −80°C until use. Prior to treating gastric cancer cells, all neutrophil‐conditioned medium (CM‐Neu) was diluted 1:1 with fresh culture medium.

2.10. Cell Culture

Human GC cell lines HGC‐27 (undifferentiated carcinoma) and NCI‐N87 (differentiated adenocarcinoma) were purchased from the American Type Culture Collection and maintained in RPMI‐1640 medium containing 10% FBS and 1% penicillin–streptomycin at 37°C in a 5% CO₂ humidified atmosphere. For experimental procedures, cells were seeded at densities of 2 × 10⁵ cells/well in 6‐well plates or 5 × 10³ cells/well in 96‐well plates and incubated overnight to allow adhesion. Once cultures reached approximately 70%–80% confluence, they were treated with CM Neu, CM PMA + Neu, CM Neu + PMA + DNase 1.

2.11. Cell Viability Assay

Cell viability was assessed using the Cell Counting Kit‐8 (CCK‐8, Glpbio). After treatment, following the addition of 10 μL of CCK‐8 reagent to each well, the plates were incubated for 2 h at 37°C. The absorbance was subsequently measured at 450 nm on a BioTek Instruments microplate reader.

2.12. Edu Assay

Proliferation of HGC‐27 and NCI‐N87 cells was assessed with the Click‐iT EdU Alexa Fluor 488 Imaging Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. In brief, cells were seeded in 96‐well plates, exposed to the indicated conditioned media (CM) for 24 h, and then incubated with EdU at a final concentration of 10 μM for 2 h before fixation. Following fixation with 4% paraformaldehyde and permeabilization with 0.5% Triton X‐100, cells were subjected to the Click‐iT reaction cocktail. Hoechst 33342 was used for nuclear counterstaining, and fluorescent images were captured on a fluorescence microscope (e.g., Zeiss Axio Observer). Using ImageJ software, the proportion of EdU‐positive cells (red fluorescence) to total nuclei (blue fluorescence) was determined from a minimum of five randomly selected fields per well. All experiments were performed in triplicate and independently repeated three times.

2.13. Cell Migration and Invasion Assays

Transwell chambers (8.0 μm pore size, Corning) were employed to evaluate the migratory and invasive capacities of the cells. For the migration assay, 5 × 10⁴ HGC‐27 and NCI‐N87 cells in 200 μL serum‐free medium were seeded into the upper chamber. For the invasion assay, the upper chamber was pre‐coated with 50 μL of Matrigel (1:8 dilution in serum‐free medium, BD Biosciences) and incubated at 37°C for 2 h to form a matrix layer. The lower chamber received 600 μL of either control CM, CM (PMA+ Neu), CM (Neu + PMA + DNase 1), all containing 10% FBS to serve as a chemoattractant. Following incubation for 24 h (migration) or 48 h (invasion), non‐migrated or non‐invaded cells remaining on the upper membrane surface were carefully wiped away with a cotton swab. Cells that successfully migrated or invaded to the lower membrane surface were fixed in 4% paraformaldehyde (15 min), stained with 0.1% crystal violet (20 min), and washed with PBS. Images of the stained cells were acquired using an inverted phase‐contrast microscope (Nikon Eclipse Ti2). For each membrane, five random fields were photographed at 100× magnification, and cell numbers were quantified with ImageJ software. All experiments were conducted in triplicate and independently repeated three times.

2.14. Cell Adhesion Assay

A monolayer of human peritoneal mesothelial cells (HPMCs) was generated by seeding them in 24‐well plates and allowing them to grow to confluence. Prior to the adhesion assay, HGC‐27 and NCI‐N87 cells were labeled with 5 μM Calcein‐AM (Beyotime) in serum‐free medium for 30 min at 37°C. After washing, 1 × 10⁵ labeled GC cells in 500 μL of CM, CM (PMA+ Neu), CM (Neu + PMA + DNase 1) were added to each well containing the HPMCs monolayer. Following a 2‐h incubation at 37°C to permit adhesion, non‐adherent GC cells were carefully eliminated by three washes with pre‐warmed PBS. The remaining adherent cells were subsequently lysed with 1% Triton X‐100, and fluorescence intensity (Ex/Em = 490/515 nm) was quantified on a BioTek microplate reader.

2.15. Cell Transfection

HGC‐27 and NCI‐N87 human gastric cancer cells were cultured in RPMI‐1640 containing 10% FBS. To manipulate gene expression, transfections were performed using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer's instructions, with the specified plasmids: a USP48 overexpression plasmid (USP48), a corresponding empty vector (NC), a scramble shRNA control (shNC), and two independent USP48‐targeting shRNA constructs (shUSP48‐1#, shUSP48‐2#). Puromycin selection (2 μg/mL) was applied for two weeks to generate stable cell lines. The shRNA sequences were as follows:

shUSP48‐1#: 5′‐GCTGTCAACAATTTGCTGAAA‐3′;

shUSP48‐2#: 5′‐CCGAATTGCTTGGTTGGCATT‐3′′;

shNC: 5′‐TTCTCCGAACGTGTCACGTAA‐3′.

2.16. Western Blot (WB)

After lysing cells in RIPA buffer containing protease and phosphatase inhibitors, 30 μg of protein from each sample was resolved by SDS‐PAGE, electrotransferred to PVDF membranes, and blocked with 5% non‐fat milk. Overnight incubation of the membranes at 4°C was performed with primary antibodies targeting USP48 (1:1000, Cat No. 12076‐1‐AP, Proteintech), CitH3 (1:1000, Cat No. 07‐596, Sigma), S100A9 (1:1000, Cat No. 26992‐1‐AP, Proteintech), TLR4 (1:1000, Cat No. 19811‐1‐AP, Proteintech), RAGE (1:2000, Cat No. 32053‐1‐AP, Proteintech), and β‐actin (1:5000, Cat No. 20536‐1‐AP, Proteintech) as a loading control. After subsequent incubation with HRP‐conjugated secondary antibodies, protein bands were detected using a Bio‐Rad ECL system and quantified by ImageJ software.

2.17. Immunofluorescence (IF) Staining in vitro

After co‐culture, cells were fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton X‐100, and blocked with 3% BSA. Overnight incubation at 4°C was performed using rabbit anti‐Cit‐H3 (1:500) and mouse anti‐MPO (1:500, Abcam) to stain NET structures and tissue sections. Following washes, samples were treated with Alexa Fluor 488‐conjugated anti‐rabbit and Alexa Fluor 594‐conjugated anti‐mouse secondary antibodies (1:1000, Invitrogen), and nuclei were counterstained with DAPI. Confocal microscopy (Zeiss LSM 880) was used for image acquisition.

2.18. Enzyme‐Linked Immunosorbent Assay (ELISA)

Supernatants from the co‐culture system and mouse serum samples were collected. The concentration of extracellular double‐stranded DNA (dsDNA) was determined using the Quant‐iT PicoGreen dsDNA Assay Kit (Thermo Fisher) as per the manufacturer's protocol. Additionally, the levels of MPO‐DNA complexes—a specific biomarker for NETs—were assessed using a human MPO‐DNA Complex ELISA Kit (Cayman Chemical) following the manufacturer's instructions.

2.19. Animal Models

Following approval from the Ethics Committee, all animal experiments were carried out in accordance with applicable guidelines. Female BALB/c nude mice, aged four weeks, were purchased from Beijing Vital River Laboratory Animal Technology Co. Ltd. and housed under specific pathogen‐free conditions. For the establishment of the subcutaneous xenograft model, stable HGC‐27 cell lines expressing shNC, shUSP48, or shUSP48 + S100A9 (5 × 10⁶ in 100 μL PBS) were subcutaneously injected into the right flank of each mouse (n = 6 per group). Using a digital caliper, tumor length (L) and width (W) were recorded once per day over an 8‐day period. Tumor volumes were determined using the equation: Volume = (L × W²)/2. At 16 days post‐inoculation, mice were euthanized. The excised tumors were then photographed, weighed, and either preserved in 4% paraformaldehyde or immediately frozen in liquid nitrogen for later analysis. For the establishment of the PM model, stable HGC‐27 cell lines expressing shNC, shUSP48, or shUSP48 + S100A9 (5 × 10⁶ in 100 μL PBS) suspended in 200 μL of sterile PBS were injected intraperitoneally (i.p.) into mice (n = 6 per group). An IVIS Spectrum In Vivo Imaging System (PerkinElmer) was used to monitor PM. Ten minutes after the intraperitoneal injection of D‐luciferin (150 mg/kg), mice were imaged under isoflurane anesthesia. Quantification of bioluminescence signal intensity was performed using Living Image software. 16 days after injection, mice were euthanized. The peritoneal cavity was opened and carefully examined. Macroscopic metastatic nodules on the mesentery, diaphragm, and parietal peritoneum were counted. Representative peritoneal tissues were collected for histopathological analysis.

2.20. Immunohistochemistry (IHC)

Citrate buffer was used for antigen retrieval, after which sections were incubated with anti‐Ki67 primary antibodies (1:4000, Cat No. 27309‐1‐AP, Proteintech), USP48 (1:500, Cat No. 12076‐1‐AP, Proteintech), S100A9 (1:500, Cat No. 26992‐1‐AP, Proteintech), TLR4 (1:100, Cat No. 19811‐1‐AP, Proteintech), and RAGE (1:2000, Cat No. 16346‐1‐AP, Proteintech) overnight at 4°C. After incubation with HRP‐conjugated secondary antibodies, staining was developed with DAB and counterstained with hematoxylin. Two independent pathologists, working in a blinded manner, scored both the staining intensity and the percentage of positive cells.

2.21. Statistical Analysis

Results are presented as mean ± standard deviation (SD) based on a minimum of three independent experiments. Student's t‐test was used to compare two groups, while comparisons involving three or more groups were conducted using one‐way ANOVA followed by Tukey's post hoc test. Statistical significance was set at *p < 0.05, **p < 0.01, ***p < 0.001.

3. Results

3.1. Elevated USP48 Expression in Gastric Cancer With Peritoneal Metastasis Is Closely Linked to NETs Formation

Previous studies have found that the deubiquitinated USP enzyme has carcinogenic potential in gastric cancer [27]. However, the relationship with peritoneal metastasis is unknown. Therefore, we examined the expression of deubiquitinated USP family kinases in peritoneal or non‐peritoneal metastatic samples. To investigate alterations in the ubiquitin‐proteasome system during gastric carcinogenesis and progression, we first compared global protein ubiquitination levels in normal gastric mucosa, non‐PM, and PM using Co‐IP. The results revealed a significant decrease in overall ubiquitination in GC tissues compared to normal gastric mucosa (Figure 1A). Notably, this reduction was more pronounced in tissues from patients with PM. qRT‐PCR analysis indicated that compared to normal gastric mucosa, the expression levels of USP1, USP9, USP14, USP38, USP42, and USP48 were significantly elevated in primary gastric cancer tissues (both non‐PM and PM groups combined), suggesting a general dysregulation of the USP family in gastric tumorigenesis (Figure 1B). Notably, among these upregulated USPs, only USP48 exhibited a further substantial increase in the PM group compared to the non‐PM group (Figure 1B). This specific upregulation of USP48 in tumors associated with peritoneal metastasis implies a unique and potentially critical role for this particular deubiquitinase in promoting the metastatic phenotype of gastric cancer cells. Consistent with qRT‐PCR results, analysis of The Cancer Genome Atlas (TCGA) Stomach Adenocarcinoma (STAD) cohort showed significantly higher USP48 expression in GC tissues than in non‐tumor tissues (Figure 1C). Moreover, high USP48 expression was associated with markedly poorer overall survival compared to low expression. We next sought to investigate tumor‐associated neutrophils and NETs formation by immunofluorescence co‐staining for CD66b (a specific neutrophil marker) and CitH3 (a hallmark of NETs). Gastric cancer tissues displayed significantly higher levels of both CD66b and CitH3 signals relative to normal gastric mucosa. The PM group exhibited a significantly higher level of both neutrophil infiltration (CD66b positive rate) and NETs formation (CitH3‐positive) than the non‐PM group (Figure 1D). The serum concentration of MPO‐DNA complexes and ANC was significantly higher in gastric cancer patients compared to healthy controls. Strikingly, the PM group exhibited the most pronounced elevations in MPO‐DNA levels and ANC, which were significantly higher than those in the non‐PM group (Figure 1E,F). In PM patients (20 cases), USP48 was significantly positively correlated with CD66b, CitH3, MPO‐DNA, and neutrophils (Figure 1G). Collectively, these results reveal that USP48 is markedly upregulated in peritoneal metastatic gastric cancer tissues and is strongly correlated with the formation of NETs.

FIGURE 1.

Elevated USP48 expression correlates with neutrophil extracellular trap formation in gastric cancer peritoneal metastasis. (A) An anti‐ubiquitin antibody was used to immunoprecipitate ubiquitinated proteins, which were then detected via immunoblotting in normal gastric mucosa, non‐PM, and PM specimens. (B) The transcript levels of selected USP family members in normal, non‐PM, and PM tissues were assessed via qRT‐PCR. (C) Compared to adjacent normal tissues (n = 34), gastric tumor tissues (n = 415) from the TCGA‐STAD cohort exhibited significantly higher USP48 mRNA expression. Validation of USP48 as a prognostic biomarker in gastric cancer using TCGA data. (D) Representative immunofluorescence images of gastric tissue sections from Normal, non‐PM, and PM groups stained for CD66b, CitH3, and DAPI for nuclei. (E) Serum MPO‐DNA complex levels among normal individuals, non‐PM cases, and PM cases. (F) ANC in blood across the normal, non‐PM, and PM patients. (G) A scatter plot reveals a strong positive correlation between USP48 and CD66b, CitH3, MPO‐DNA levels and ANC. ***p < 0.001.

As shown in Table 1, no significant correlation was observed between USP48 expression and patient gender (p = 0.3458) or age (p = 0.3391). USP48 expression was significantly correlated with aggressive tumor phenotypes. Among poorly differentiated tumors, 6 out of 12 (50%) showed high USP48 expression, and among undifferentiated tumors, 9 out of 15 (60%) showed high expression. In contrast, only 5 out of 13 (38.5%) moderately differentiated tumors exhibited high USP48 levels. Furthermore, USP48 expression was closely linked to tumor progression. High USP48 expression was slightly more common in patients with advanced T stage (T3–4: 10/18, 55.6% vs. T1‐2: 10/22, 45.5%; p = 0.4238) and advanced TNM stage (stage 3–4: 12/20, 60% vs. stage 1–2: 8/20, 40%; p = 0.3369). Most notably, among PM‐positive patients, 17 out of 20 (85%) exhibited high USP48 expression, whereas only 3 out of 20 (15%) showed low expression. Conversely, among PM‐negative patients, 17 out of 20 (85%) had low USP48 expression, with only 3 out of 20 (15%) showing high expression (p < 0.0001). These findings suggest that elevated USP48 expression is not a random event but is closely linked to peritoneal metastasis, a key indicator of gastric cancer aggressiveness. While associations with other clinicopathological features such as differentiation, T stage, and TNM stage were not statistically significant in this cohort, the strong correlation with PM highlights the potential role of USP48 in promoting metastatic spread in gastric cancer.

TABLE 1.

Relationship between USP48 expression and the clinicopathological characteristics of 40 gastric cancer patients.

Clinico‐pathological features	Cases (n = 40)	USP48 expression		p
		Low (n = 20)	High (n = 20)
Gender
Female	18	11	7	0.3458
Male	22	9	13
Age
≤ 60	18	10	8	0.3391
> 60	22	10	12
Grade
Moderate differentiation	13	8	5	0.1217
Poor differentiation	12	6	6
Undifferentiation	15	6	9
T stage
T1–2	22	12	10	0.4238
T3–4	18	8	10
TNM stage
1–2	20	12	8	0.3369
3–4	20	8	12
PM
Yes	20	17	3	< 0.0001
No	20	3	17

3.2. Neutrophil NETs Promotes Peritoneal Metastasis in GC Cells

To examine the role of neutrophil‐derived factors in regulating GC cell viability, HGC‐27 and NCI‐N87 cells were treated for 24 h with CM (Neu), CM (PMA+ Neu), or CM (Neu + PMA + DNase 1). Notably, treatment with CM (PMA+ Neu) substantially increased viability in both cell types when compared to CM (Neu), whereas CM (Neu + PMA + DNase I) completely abolished the increase in cell viability (Figure 2A). According to the Edu incorporation assay, exposure to NETs‐CM significantly increased the fraction of Edu‐incorporating GC cells relative to the control group maintained in standard medium (Figure 2B). As shown in Figure 2B, the proportion of Edu‐positive cells was substantially reduced in the CM (Neu + PMA + DNase I) group compared to the CM (PMA + Neu) group. Cell migration and invasion were then assessed using Transwell assays. The results (Figure 2C,D) revealed that CM (PMA + Neu) treatment significantly increased both migration and invasion of GC cells relative to the CM (Neu) group. CM (Neu + PMA + DNase I) treatment of GC cells effectively reversed the enhanced migration and invasion induced by CM (PMA+ Neu). GC cells treated with CM (PMA + Neu) significantly increased their adhesion to the HPMCs monolayer by approximately 2‐fold compared to the CM (Neu) group, whereas the adhesion rate of GC cells treated with CM (Neu + PMA + DNase I) was remarkably lower than that of the CM (PMA + Neu) group (Figure 2E). These results revealed that intact NETs function as a potent promoter of GC aggressiveness by enhancing tumor cell proliferation, motility, and adhesion to HPMCs.

FIGURE 2.

NETs promote peritoneal metastasis of GC cells. (A) The CCK‐8 assay was employed to evaluate cell viability. (B) Representative fluorescence micrographs of GC cells after Edu assay. (C–D) GC cells migration (without Matrigel, C) and invasion (with Matrigel coating, D) evaluated by Transwell assay. Scale bar: 100 μm. (E) Quantitative analysis of the adhesion of Calcein‐AM‐labeled GC cells to a monolayer of HPMCs. *p < 0.05, **p < 0.01, ***p < 0.001.

To exclude the possibility that the observed effects on gastric cancer cells were mediated by residual PMA carried over from neutrophil stimulation, gastric cancer cells (HGC‐27 and NCI‐N87) were treated with PMA alone at the same concentration used for neutrophil stimulation, and a series of functional assays were performed. As shown in Figure S1A–D, PMA treatment alone did not significantly affect cell viability (CCK‐8 assay), proliferation (EdU staining), migration, or invasion (Transwell assays) in either HGC‐27 or NCI‐N87 cells compared to the untreated control groups. Furthermore, to evaluate whether PMA‐treated gastric cancer cells exhibited enhanced adhesion to human peritoneal mesothelial cells (HPMCs), HGC‐27 and NCI‐N87 cells pretreated with or without PMA were co‐cultured with HPMCs. The tumor cell adhesion assay revealed no significant difference in adhesion ability between the PMA‐pretreated group and the control group (Figure S1E). Collectively, these results demonstrate that PMA alone, at the concentration used in this study, does not directly promote the malignant phenotypes of gastric cancer cells, thereby ruling out the possibility that the effects observed in subsequent experiments were attributable to residual PMA.

3.3. Knockdown of USP48 Inhibits NETs Formation in GC Cells

We constructed stable cell lines with modulated USP48 expression to investigate the role of this protein in GC. WB analysis confirmed that transfection with the USP48 overexpression plasmid significantly increased USP48 protein levels in both HGC‐27 and NCI‐N87 cells relative to the negative control (NC) group. Conversely, transfection with two shRNAs targeting USP48 effectively knocked down its expression. Notably, shUSP48‐2# exhibited a more potent silencing efficiency than shUSP48‐1# (Figure 3A). Therefore, the shUSP48‐2# construct was selected for all subsequent functional experiments. Using a co‐culture system, we found that neutrophils co‐cultured with USP48‐overexpressing GC cells displayed significantly enhanced NETs formation. This was evidenced by a marked enhancement in the immunofluorescence co‐localization of Cit‐H3 (a recognized indicator of NETs) and MPO (a neutrophil granule protein). In stark contrast, neutrophils co‐cultured with USP48‐knockdown GC cells (shUSP48‐2#) showed a substantial reduction in Cit‐H3/MPO double‐positive NETs structures compared to the shNC group (Figure 3B). ELISA results demonstrated that the levels of both extracellular dsDNA and MPO‐DNA complexes were significantly elevated in the supernatant from neutrophils co‐cultured with USP48‐overexpressing GC cells. Conversely, a pronounced decline in the release of these NETs components was observed following USP48 knockdown in GC cells (Figure 3C). Consistent with the above results, WB analysis revealed that the protein level of Cit‐H3 was significantly increased when neutrophils were co‐cultured with USP48‐overexpressing GC cells. Co‐culture with USP48‐knockdown GC cells resulted in a clear reduction of Cit‐H3 levels (Figure 3D). Collectively, our data provided consistent and compelling evidence that high expression of USP48 in gastric cancer cells promotes NETs formation in tumor‐associated neutrophils. Conversely, genetic knockdown of USP48 in GC cells effectively inhibits this process.

FIGURE 3.

USP48 potentiates NETs formation in GC cells. (A) Western blotting was performed to assess and quantify USP48 protein expression in both HGC‐27 and NCI‐N87 cells stably transfected with NC, USP48, shNC, shUSP48‐1#, shUSP48‐2#. (B) Representative immunofluorescence images showing NETs formation. NETs were labeled using antibodies against Citrullinated Histone H3 (Cit‐H3, green) and Myeloperoxidase (MPO, red), with nuclear counterstaining using DAPI (blue). Yellow arrows indicate characteristic NETs structures. Scale bar: 50 μm. (C) ELISA of NETs markers in the co‐culture supernatant: extracellular dsDNA and MPO‐DNA complexes. (D) Western blot analysis was conducted to assess Cit‐H3 protein expression in both HGC‐27 and NCI‐N87 cells stably transfected with NC, USP48, shNC, shUSP48‐1#, shUSP48‐2#. *p < 0.05, **p < 0.01, ***p < 0.001.

3.4. Knockdown of USP48 Inhibits NETs‐Mediated Peritoneal Metastasis in GC Cells

Knockdown of USP48 in NCI‐N87 and HGC‐27 cells significantly reduced cell viability. Notably, the inhibitory effects of USP48 knockdown on cell viability and proliferation were substantially diminished upon treatment with CM (PMA + Neu) (Figure 4A,B). Knockdown of USP48 obviously decreased the migratory and invasive capacity of GC cells. Importantly, the migratory and invasive abilities of USP48‐knockdown cells were effectively restored by the addition of NETs‐conditioned medium, reaching levels comparable to or surpassing those of control cells (Figure 4C,D). USP48 knockdown impaired the adhesive ability of GC cells to adhere to peritoneal mesothelial cells, while CM (PMA + Neu) could counteract the inhibitory effect of shUSP48 on GC cell peritoneal adhesion (Figure 4E). Collectively, these results show that USP48 drives the proliferation, migration, invasion, and peritoneal adhesion, highlighting a significant role of USP48 in driving peritoneal metastasis of GC.

FIGURE 4.

USP48 knockdown impedes GC cell proliferation, migration, and invasion effects rescued by CM (PMA+ Neu). (A) The cell viability of HGC‐27 and NCI‐N87 cells in shNC, shUSP48‐2#, shNC + CM, and shUSP48‐2# + CM groups. (B) Representative micrographs and corresponding quantification of EdU incorporation assays. (C–D) Representative images and quantification of Transwell migration assay (C) and Matrigel‐coated Transwell invasion assay (D). (E) Calcein AM‐labeled HGC‐27 and NCI‐N87 cells were co‐cultured with HPMC monolayers. *p < 0.05, **p < 0.01, ***p < 0.001.

3.5. USP48 Deubiquitinates and Stabilizes S100A9

IHC analysis revealed that both USP48 and S100A9 protein expression levels were significantly higher in GC tissues compared to adjacent normal tissues (Figure 5A). Notably, their expression was most pronounced in GC tissues from patients with PM (PM group) compared to those without metastasis (non‐PM group). WB of tissue lysates confirmed these findings, demonstrating a consistent upregulation of both proteins in tumor tissues, with the highest levels in the PM group (Figure 5B). Furthermore, analysis of IHC scores from the 40 gastric cancer samples revealed a strong positive correlation between USP48 and S100A9 expression levels (Figure 5C). Furthermore, overexpression of USP48 markedly elevated the protein levels of both USP48 and S100A9, whereas knockdown of USP48 markedly decreased their expression compared to their respective controls (Figure 5D). ELISA results demonstrated that the concentration of S100A9 secreted into the cell culture supernatant was significantly elevated upon USP48 overexpression and reduced upon USP48 knockdown (Figure 5E). Overexpression of USP48 increases the expression of S100A9, while treatment with MG132 has little effect on S100A9 expression, indicating that USP48 regulates S100A9 expression through the ubiquitination system (Figure 5F). Co‐IP experiments verified that endogenous USP48 and S100A9 proteins directly bind to each other in HGC‐27 cells (Figure 5G). IF staining further revealed that USP48 and S100A9 co‐localized predominantly in the cytoplasm (Figure 5H). The poly‐ubiquitination level of S100A9 was significantly lowered upon USP48 overexpression (Figure 5I). To directly assess the impact on protein stability, we performed a cycloheximide (CHX) chase assay. Compared to control cells, USP48‐overexpressing cells exhibited a significantly slower degradation rate of the S100A9 protein (Figure 5J). In summary, our findings demonstrate that USP48 directly binds to S100A9, promotes its deubiquitination, and thereby enhances its protein stability.

FIGURE 5.

USP48 stabilizes S100A9 protein through deubiquitination in GC cells. (A) Representative micrographs (left) and quantification (right) of USP48 and S100A9 level in normal gastric tissues (n = 20), non‐PM GC tissues (n = 20), and PM GC tissues (n = 20). (B) Protein levels of USP48 and S100A9 in normal gastric tissues, non‐PM GC tissues, and PM GC tissues. (C) Scatter plot showed the correlation between USP48 and S100A9. (D) Western blotting was used to assess USP48 and S100A9 protein levels in the NC group, USP48 shNC, and shUSP48‐2# groups. (E) ELISA measurement of S100A9 concentration in the cell supernatant. (F) Western blot analysis of S100A9 expression in shNC and shUSP48‐2# cells treated with DMSO or the proteasome inhibitor MG132 (10 μM). (G) Co‐IP revealed that endogenous USP48 bound to S100A9. (H) IF staining showed the co‐localization of USP48 and S100A9. (I) Ubiquitination assay and WB were used to detect ubiquitinated S100A9. (J) S100A9 protein stability was analyzed by WB. *p < 0.05, **p < 0.01, ***p < 0.001.

3.6. USP48 Facilitates NETs Formation by Stabilizing S100A9 and USP48 Upregulates RAGE and TLR4 Expression, Which Is Associated With Enhanced Signaling Activity

We altered the expression of USP48 and S100A9 in HGC‐27 cells to examine the regulatory axis between these two proteins. As confirmed by Western blotting, forced expression of USP48 (USP48 + shNC group) led to a significant increase in the protein levels of both USP48 and S100A9 compared with the control (NC + shNC) group. Conversely, knockdown of S100A9 (NC + shS100A9 group) effectively reduced S100A9 expression without affecting USP48 levels. In the rescue group (USP48 + shS100A9), USP48 overexpression failed to elevate S100A9 protein levels due to concurrent S100A9 knockdown (Figure 6A). Consistent with intracellular protein levels, ELISA of cell culture supernatants demonstrated that USP48 overexpression markedly increased the secretion of S100A9, which was abolished by S100A9 knockdown (Figure 6B). Given that S100A9 has been reported to activate TLR4 and RAGE signaling in neutrophils [19], we next examined whether USP48 overexpression affects the expression of these receptors. The protein expression levels of TLR4 and RAGE were both significantly elevated upon USP48 overexpression in neutrophils. Importantly, this upregulation was completely reversed when S100A9 was knocked down in USP48‐overexpressing cells (USP48 + shS100A9 group) (Figure 6C). Co‐culture with USP48‐overexpressing GC cells (USP48 + shNC) significantly induced NETs formation, as evidenced by a pronounced increase in extracellular structures co‐staining for Cit‐H3 and MPO. Knockdown of S100A9 in GC cells (both NC + shS100A9 and USP48 + shS100A9 groups) drastically reduced this NETs‐inducing effect (Figure 6D). The extracellular dsDNA and MPO‐DNA complex levels were significantly elevated in the supernatant from the Neu + USP48 group, and this increase was attenuated in the Neu + USP48 + shS100A9 group (Figure 6E). Cit‐H3 levels were markedly higher when neutrophils were co‐cultured with USP48‐overexpressing GC cells. This induction was again dependent on S100A9, as it was reversed by S100A9 knockdown (Figure 6F). To determine whether S100A9 promotes neutrophil extracellular trap (NET) formation via TLR4 and/or RAGE receptors, gastric cancer cells were co‐cultured with neutrophils under the following conditions: control (NC), S100A9 treatment, S100A9 plus TAK‐242 (a TLR4 inhibitor), and S100A9 plus FPS‐ZM1 (a RAGE inhibitor). NET formation in neutrophils was then assessed by multiple approaches. As shown in Figure S2A, immunofluorescence staining revealed that S100A9 treatment markedly increased the colocalization of citrullinated histone H3 (Cit‐H3), myeloperoxidase (MPO), and DAPI, indicating enhanced NETosis. However, this effect was substantially attenuated upon co‐treatment with either TAK‐242 or FPS‐ZM1. Consistently, ELISA measurements demonstrated that the elevated levels of extracellular dsDNA and MPO‐DNA complexes induced by S100A9 were significantly reduced by both inhibitors (Figure S2B). Western blot analysis further confirmed that S100A9‐induced upregulation of Cit‐H3 protein expression in neutrophils was effectively abrogated by TAK‐242 or FPS‐ZM1 treatment (Figure S2C). In summary, our findings demonstrate that USP48 upregulates and stabilizes S100A9, increases the expression of RAGE and TLR4, and USP48 expression, which is associated with enhanced signaling activity.

FIGURE 6.

USP48 activates NETs formation through upregulating S100A9 and enhancing TLR4/RAGE signaling. (A) Protein levels of USP48 and S100A9 in HGC‐27 cells were examined by western blot. (B) ELISA of S100A9 levels in the cell culture supernatant from HGC‐27 cells. (C) Protein levels of TLR4 and RAGE in neutrophils were examined by western blot. (D) Representative IF images showed NETs formation. HGC‐27 cells were stained for Cit‐H3, MPO, and DAPI. (E) Quantification of NETs markers (dsDNA and MPO‐DNA complexes) in co‐culture supernatants. (F) Western blot analysis of Cit‐H3 protein levels. *p < 0.05, **p < 0.01, ***p < 0.001.

3.7. USP48 Drives GC Growth and PM by Stabilizing S100A9 In Vivo

We created subcutaneous xenograft models with HGC‐27 cells to verify the in vivo oncogenic role of USP48. Macroscopic examination at the endpoint revealed that tumors derived from shUSP48 cells were substantially smaller in size compared to those from shNC control cells (Figure 7A). The shUSP48 group showed marked reductions in both tumor volume (Figure 7B) and final tumor weight. Importantly, restoring S100A9 expression in USP48‐knockdown cells (shUSP48 + S100A9 group) largely reversed the tumor‐suppressive effect. Tumors from the shUSP48 group exhibited significantly lower expression levels of KI67 (a proliferation marker), USP48, S100A9, TLR4, and RAGE compared to the shNC group (Figure 7D). IF staining showed that tumors from the shUSP48 group had significantly fewer Ly‐6G+ neutrophils and markedly reduced Cit‐H3 signal compared to control tumors (Figure 7E). Correspondingly, the level of circulating MPO‐DNA complexes was also decreased in the serum of shUSP48 tumor‐bearing mice (Figure 7F).

FIGURE 7.

USP48 promotes gastric tumor growth and peritoneal metastasis in vivo via S100A9. (A) Representative macroscopic views of resected subcutaneous tumors. (B) Growth curves of subcutaneous tumor volumes measured once a day. (C) Quantification of final tumor weights. (D) IHC images and semi‐quantitative analysis of KI67, USP48, S100A9, TLR4, and RAGE expression in tumor tissues. (E) IF images and quantification of tumor sections stained for Ly‐6G and Cit‐H3. (F) ELISA quantification of serum MPO‐DNA complex levels in tumor‐bearing mice. (G) Representative IVIS images of mice at 16 days after intraperitoneal injection of luciferase‐labeled HGC‐27 cells. (H) Image of peritoneal cavities from each group at necropsy, showing metastatic nodules. Quantitation of the number of macroscopic peritoneal metastatic nodules per mouse. **p < 0.01, ***p < 0.001.

To directly assess the involvement of USP48 in metastatic spread, we utilized a peritoneal dissemination model using luciferase‐labeled HGC‐27 cells. Mice injected with shUSP48 cells had significantly weaker peritoneal luciferase signals compared to the shNC group, indicating impaired tumor establishment and growth in the peritoneum (Figure 7G). This was corroborated by a gross anatomical examination at the endpoint, showing a dramatic decline in the number of visible peritoneal metastatic nodules in the shUSP48 group (Figure 7H). Crucially, overexpression of S100A9 in USP48‐deficient cells (shUSP48 + S100A9) completely rescued the metastatic potential (Figure 7G,H). Taken together, in vivo studies provide compelling evidence that USP48 is a potent driver of GC growth and PM, and its function is mediated largely through stabilizing S100A9.

4. Discussion

Gastric cancer PM represents a devastating stage of the disease, characterized by therapeutic resistance and poor prognosis [28]. While the ubiquitin‐proteasome system is known to be dysregulated in cancer, its specific role in orchestrating the complex tumor microenvironment that facilitates PM remains poorly understood [29]. We uncovered a novel and critical function of the deubiquitinating enzyme USP48. We demonstrate that USP48 is specifically upregulated in GC tissues with PM and acts as a central driver of metastasis by stabilizing the alarmin S100A9. This stabilization initiates a molecular cascade involving TLR4/RAGE activation, which in turn induces NETs formation, thereby creating a pro‐metastatic niche that promotes tumor growth and peritoneal dissemination. This USP48/S100A9/TLR4/RAGE axis offers a new mechanistic framework for understanding and potentially treating PM.

Emerging evidence highlights the pivotal role of the tumor microenvironment, particularly innate immune components, in shaping the metastatic niche [30]. NETs are web‐like structures of DNA decorated with antimicrobial proteins released by activated neutrophils and have recently been implicated in cancer progression [31]. In the context of PM, NETs are thought to physically trap circulating tumor cells, shield them from immune surveillance, and directly promote their proliferation and invasiveness, thereby facilitating peritoneal seeding and outgrowth [32]. In alignment with this paradigm, our study provides direct experimental evidence linking NETs to GC progression. We demonstrated that CM (PMA + Neu) potently rescued the impaired proliferation, migration, invasion, and peritoneal adhesion capacity of USP48‐knockdown GC cells. Furthermore, we observed increased NETs formation and neutrophil infiltration in GC tissues from patients with PM and in tumor xenografts with high metastatic potential, establishing a strong clinical correlation.

USP48, a deubiquitinating enzyme, has been identified as an oncogene in various malignancies by stabilizing key proteins involved in proliferation, epithelial‐mesenchymal transition, and therapy resistance [26, 33, 34]. However, its role in modulating the immune microenvironment, particularly in NETs‐driven metastasis, remained unexplored. Our work bridges this gap by identifying USP48 as a crucial regulator at the intersection of tumor‐intrinsic signaling and neutrophil‐mediated extrinsic promotion of metastasis. GC tissues exhibited significantly higher USP48 expression compared to normal controls, especially those from patients with PM, and its knockdown profoundly inhibited tumor growth and peritoneal dissemination in vivo. Mechanistically, we discovered that this pro‐metastatic function is not cell‐autonomous but is executed through reshaping the immune landscape, specifically by promoting NETs formation in a S100A9‐dependent manner.

A major discovery of this work is the recognition of S100A9 as a previously unknown target of USP48. USP48, through its deubiquitinating activity, interacts with and stabilizes S100A9 protein by removing its polyubiquitin chains, thereby protecting it from proteasomal degradation. This post‐translational regulation was confirmed by Co‐IP, ubiquitination assays, and CHX chase experiments. Consequently, USP48 overexpression led to a marked accumulation and increased secretion of S100A9 from GC cells. As a well‐characterized damage‐associated molecular pattern, S100A9, often functioning as a heterodimer with S100A8, activates pattern recognition receptors, primarily TLR4 and RAGE, on various immune and stromal cells [35, 36, 37]. The engagement of S100A9 with RAGE and TLR4 is a potent inducer of NF‐κB signaling, resulting in the production of cytokines and chemokines that fuel chronic inflammation and immune cell recruitment [38, 39]. Our results extend this knowledge to the context of NETosis in cancer. We demonstrated that USP48‐upregulated S100A9 activates the TLR4/RAGE pathway within neutrophils, creating a pro‐inflammatory feedback loop. More importantly, secreted S100A9 acts in a paracrine manner on adjacent neutrophils [40]. Co‐culture experiments unequivocally showed that GC cells overexpressing USP48, via secreting S100A9, robustly induced NETs formation in neutrophils, as evidenced by increased Cit‐H3 expression, dsDNA release, and MPO‐DNA complexes. This effect was completely abrogated by S100A9 knockdown.

It is well established that S100A9 lacks a classical signal peptide for endoplasmic reticulum/Golgi‐mediated secretion and is therefore released via unconventional protein secretion pathways. Known mechanisms for S100A9 secretion include extracellular vesicle (exosome)‐mediated release, autophagy‐associated secretion, and microtubule‐dependent alternative pathways. Upon neutrophil activation or cellular stress, S100A9 can be translocated from the cytoplasm to the cytoskeleton and cell membrane, followed by secretion through these non‐classical routes.

In the present study, we demonstrated that USP48‐mediated deubiquitination enhances S100A9 protein stability, leading to increased intracellular S100A9 accumulation and its subsequent secretion into the extracellular space. However, the precise secretory route through which USP48‐stabilized S100A9 is released from gastric cancer cells, whether via exosomes, autophagy‐related secretion, or other unconventional pathways, was not investigated in this work. Given the established role of S100A9 as a key paracrine signal that activates RAGE and TLR4 on neighboring neutrophils to induce NETosis, understanding its secretion mechanism represents an important direction for future research. Elucidating the specific non‐classical secretory machinery engaged by USP48‐upregulated S100A9 may reveal additional therapeutic entry points for disrupting this pro‐metastatic axis.

The in vivo significance of this pathway was validated in two complementary mouse models. In subcutaneous xenografts, USP48 knockdown dramatically suppressed tumor growth, proliferation (Ki67), and the intratumoral levels of S100A9, TLR4, and RAGE. It also markedly reduced neutrophil infiltration and NETs formation within the tumors, along with circulating NETs markers. All these effects were reversed by S100A9 overexpression, proving that S100A9 is the primary effector downstream of USP48. Most decisively, in a luciferase‐labeled peritoneal dissemination model, USP48 knockdown severely impaired the establishment and outgrowth of peritoneal tumors, leading to fewer metastatic nodules. This metastatic defect was again fully rescued by S100A9 restoration. These in vivo results translate our molecular findings into a coherent pathophysiological narrative: USP48 in GC cells stabilizes S100A9, which promotes a local and systemic environment rich in NETs, ultimately fueling tumor growth and facilitating peritoneal colonization. Notably, S100A9 knockdown almost completely reversed all effects induced by USP48 overexpression, including elevated S100A9 secretion, upregulation of TLR4/RAGE expression, enhanced NETs formation, and increased Cit‐H3 levels (Figure 6). This complete reversal carries important mechanistic implications: it demonstrates that S100A9 serves as the sole and indispensable downstream effector of USP48 in this pathway, rather than being one of multiple parallel targets. The pro‐NETs function of USP48 is entirely dependent on its stabilization of S100A9; when S100A9 is knocked down, even USP48 overexpression fails to drive downstream RAGE/TLR4 activation and NETs formation. These findings further consolidate the causal logic of the linear USP48/S100A9/RAGE/TLR4/NETs signaling axis and highlight that targeting S100A9 or its downstream pathways may be sufficient to block USP48‐mediated pro‐metastatic effects, even under conditions of persistent USP48 overexpression.

Several limitations of this study should be acknowledged. First, although we have established the USP48/S100A9 axis as a critical upstream regulator of NETosis through activation of RAGE and TLR4, we did not directly examine the downstream signaling events linking receptor engagement to the execution of NETosis, including PAD4 activation, ROS generation, and neutrophil elastase (NE) release. Nevertheless, these events have been well documented as downstream effectors of S100A9, RAGE, and TLR4 in neutrophils. Further investigations are warranted to determine whether USP48‐mediated S100A9 upregulation promotes NETosis through these canonical pathways, thereby refining the mechanistic model proposed herein. Second, the relatively small sample size represents a major limitation. Future studies with larger clinical cohorts are required to validate the efficacy of USP48 as a reliable biomarker for peritoneal metastasis in gastric cancer.

Taken together, this study identifies a pathway that has not been recognized before in which deubiquitinase USP48 drives gastric cancer peritoneal metastasis by orchestrating a cross‐talk between tumor cells and neutrophils. We elucidate that USP48 stabilizes S100A9, which in turn activates RAGE and TLR4 signaling in neutrophils and, upon secretion, acts as a potent NETs inducer in neutrophils. The resulting NETs‐rich microenvironment facilitates tumor cell survival, proliferation, and adhesion within the peritoneum. Therapeutically, this work highlights the USP48/S100A9 axis as a promising multi‐node target. Inhibiting USP48's deubiquitinase activity, neutralizing S100A9, or blocking TLR4/RAGE signaling could potentially disrupt the vicious cycle of NETs‐driven metastasis. Future studies should explore the efficacy of targeting this axis, potentially through USP48 inhibitors or S100A9‐neutralizing antibodies, in combination with existing therapies to combat this devastating complication of gastric cancer. Future studies should also explore whether this axis is active in other cancers prone to peritoneal or NETs‐associated metastasis, such as ovarian or pancreatic cancer, potentially broadening the impact of these findings.

Author Contributions

Jun Song: conceptualization, investigation, resources, validation, writing – original draft. Kaiyun You: data curation, software, writing – review and editing. Xiangyue Ren: formal analysis, visualization, writing – review and editing. Ning Xie: methodology, project administration, supervision, writing – review and editing. All authors read and approved the final manuscript.

Funding

This work was funded in part by grants from the Natural Science Foundation of Jiangxi Province (Grant 20242BAB20399).

Ethics Statement

This study was approved by the Ethics Committee of The First Affiliated Hospital of Nanchang University (No. CDYFYYLK(01–019)) and written informed consent was obtained from every participant. All procedures adhered to the principles in the Declaration of Helsinki.

All animal experiments received approval from the Ethics Committee of The First Affiliated Hospital of Nanchang University (No. CDYFY‐IACUC‐202605GR031).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: The primer sequences of USP family members.

Figure S1: PMA‐treated gastric cancer cells promote tumor cell proliferation, migration, invasion, and adhesion to HPMCs. (A) Cell viability was assessed by CCK‐8 assay. (B) Cell proliferation was evaluated by EdU incorporation assay. (C) Cell migration was measured by Transwell migration assay. (D) Cell invasion was determined by Transwell invasion assay. (E) Tumor cell adhesion to HPMCs was evaluated by tumor cell adhesion assay.

Figure S2: S100A9 promotes NETs formation in neutrophils via TLR4 and RAGE. (A) Representative immunofluorescence images of NETs formation. Neutrophils were stained for Cit‐H3 (green), MPO (red), and DAPI (blue). (B) Extracellular dsDNA levels and MPO‐DNA complex levels in the culture supernatant were measured by ELISA. (C) Protein expression of Cit‐H3 in neutrophils was determined by Western blotting.

Data Availability Statement

The corresponding author responsible for this manuscript will provide the original data that backs up its conclusions to any qualified researcher upon request.