143B (CVCL_2270), KHOS (CVCL_2546), HOS (CVCL_0312), MG63 (CVCL_0426), U2OS (CVCL_0042), and SaoS2 (CVCL_0548) cells were human osteosarcoma cell lines, and K7M2 cells were mouse osteosarcoma cell lines; all were purchased from the American Type Culture Collection. The human normal osteoblastic hFOB 1.19 cell line was kindly provided by the Department of Hematology, Peking University People’s Hospital. 143B, U2OS, SaoS2, K7M2, and hFOB 1.19 cells were cultured in DMEM (Gibco, American); KHOS and MG63 cells were cultured in RPMI-1640 (Gibco); and HOS cells were cultured in MEM (Gibco), each supplemented with 10% fetal bovine serum (FBS, Wisent, Canada) and 1% penicillin/streptomycin (Gibco). The cells were all cultured in an incubator at 37°C in humidified air with 5% CO₂. All human cell lines were authenticated using STR (or SNP) profiling within the last 3 years, and all experiments were performed with mycoplasma-free cells.

The RNA sequencing data of osteosarcoma tissues were obtained from the Therapeutically Applicable Research to Generate Effective Treatments on Osteosarcoma (TARGET‐OS, phs000468) at https://portal.gdc.cancer.gov/projects/TARGET‐OS. The gene expression of CCNE1 was analyzed to investigate its correlation with patients’ prognosis.

OS samples were collected from Peking University People’s Hospital, with approval from the Ethics Committee of Peking University People’s Hospital. Informed consent was obtained from all patients. The samples to detect cyclin E1 expression were from 69 primary osteosarcoma in situ cases, and the samples used to detect neutrophil infiltration and ELA2 expression were from 34 primary osteosarcoma tissues and 33 lung metastasis lesions, which were processed as a tumor microarray. The tissues were fixed with 4% paraformaldehyde and made into a tissue microarray as described previously (18).

The animal experiments were approved by the Ethics Committee of Peking University People’s Hospital (2023PHE123) and conducted according to the guidelines for the care and use of laboratory animals.

Four-week-old BALB/c mice were purchased from Beijing Vital River, China to establish the mouse OS model. The BALB/c mice were randomly divided into two groups: 1 × 10⁷ K7M2-lvNC cells and K7M2-lvELA2 were subcutaneously injected into the right side of the mice individually. After 25 days, the mice were killed, and the tumors were isolated for further analysis. For neutrophil exhaustion, 1 × 10⁷ K7M2 cells were subcutaneously injected into the right side of the mice. The mice were divided into two equal groups and treated with anti-Ly6G (127649, BioLegend, United States) and anti-IgG (50 μg/head, intraperitoneal injection, I4131, Sigma-Aldrich, United States) every 4 days. After 25 days, the mice were killed, and the tumors were isolated for further analysis.

Cultured cells were lysed with a mixture of lysis buffer (9803, Cell Signaling Technology, United States), protease inhibitor (B14001, Selleck), and phosphatase inhibitor (B15001, Selleck, United States). The protein concentration was measured using the BCA protein assay kit according to the manufacturer’s protocol (PC0020, Solarbio, China). The whole-cell lysate protein dilution was incubated with cylcin E1 antibody (1:50) overnight at 4°C. Protein A/G agarose beads (sc-2003, Santa Cruz Biotechnology, United States) were washed twice with cold PBS and added to the antigen–antibody dilution for another 1.5-h incubation at 4°C. The sample was centrifuged to collect the beads–antigen–antibody complex.

For Western blot, the proteins were loaded into 10% SDS-PAGE gels for protein separation. Protein bands were incubated with specific primary antibodies, including cyclin E1 (1:500), calpain 1 (1:1,000), calpain 2 (1:1,000), ELA2 (1:300), RB (1:1,000), pRB (1:500), and GAPDH (1:2,000). Proteins were visualized using Image Lab Software (Bio-Rad, United States), and the protein bands were quantitatively analyzed using ImageJ software. Information on the antibodies is provided in Supplementary Table S3 .

Neutrophils were isolated from the peripheral blood of healthy donors. Peripheral blood was mixed in equal volume with erythrocyte sedimentation solution (R1000, Solarbio) and set at room temperature for 30 min. The neutrophils were isolated by density gradient separation using lymphocyte separation medium (LTS1077-1) and centrifuging at 500 × g for 25 min at room temperature. Neutrophils were confirmed to be of > 90% purity by flow cytometric analysis with antihuman CD66b Antibody (E-AB-F1267E, Yojanbio, China).

A 0.4-μm transwell plate (LabSelect) was used to construct the noncontact coculture system. Neutrophils were isolated as previously mentioned; 5 × 10⁵ osteosarcoma cells were seeded into the lower layer of the transwell plate overnight. Subsequently, 5 × 10⁵ neutrophils were added to the upper layer for a 24-h coculture. For the contact co-culture system, 5 × 10⁵ osteosarcoma cells were preseeded into a six-well plate, and 5 × 10⁵ neutrophils were added directly to the same plate.

Anti-ELA2 antibody (1:50), anticyclin E1 antibody (1:100), and antihistone H3 antibody (1:100) were used for cell immunofluorescence. Rhodamine Phalloidin (R415, Invitrogen, United States) was used to visualize F-actin in OS cells. Anti-ELA2 antibody (1:400), anti-CD66b antibody (1:2,000), and anti-Ki67 (1:4,000) were used for immunohistochemistry (IHC) staining.

Neutrophils collected from healthy donors were stained using the Fast Giemsa Stain Kit (40751ES02, Yeasen, China) according to the manufacturer’s protocol. The images were obtained and observed under a microscope (Leica, Germany).

The cells were lysed using TRIzol (Invitrogen), and total RNA was isolated with isopropanol and 75% ethanol. PrimeScript RT Master Mix (TaKaRa Biotechnology, Japan) was used for reverse transcription. qPCR was performed to measure the relative expression levels of target genes using SYBR Green premix Ex Taq (TaKaRa, RR420A). The primers are all listed in Supplementary Table S1 . The relative fold changes in gene expressions were calculated using the 2^−▵▵Ct method.

For the CCK8 assay, 3 × 10³ 143B or U2OS cells were plated into a 96-well plate. To measure the 24-h IC₅₀ values of AZD 9668 (Cayman, United States), different concentrations of AZD 9668 were added to the cell culture medium. Cell Counting Kit-8 (CCK8; Dojindo , Japan) was added to the culture medium (1:10). After 1 h of incubation, the optical density (OD) value was measured at 450 nm using a microplate reader (Bio-Rad).

SiRNA targeted calpain 1, calpain 2, negative control siRNA, and ELA2 overexpression plasmids were designed and synthesized by Hanbio Tech (Shanghai, China), with the sequences listed in Supplementary Table S2 . The 143B and U2OS cells were transfected with Lipofectamine 3000 transfection reagent (L3000015, Thermo Scientific, United States).

To generate stable ELA2 overexpressed K7M2 cells, K7M2 cells were infected with 10 multiplicity of infection (MOI) lentivirus (Lv), followed by 3 mg/ml puromycin selection for 1 week.

The cell culture supernatants were collected and centrifuged to remove cellular fragments, and then used immediately or stored at −80°C. Enzyme-linked immunosorbent assay (ELISA) was performed according to the manufacturer’s instructions, using an enzyme‐linked immunosorbent assay kit (CSB-EL007587HU, CUSABIO, China) for ELA2 and an ELISA assay kit (RX102517H, Rruixin Bio, China) for NETs.

The cell cycle assay was performed using PI staining, and the cell cycle and apoptosis analysis kit (C1052, Beyotime, China) was used according to the operator’s manual. Briefly, the cells were washed with cold PBS and mixed with 70% ethanol for 2 h. PI stain buffer and RNase A were added and incubated at 37°C for 30 min. The cells were then subjected to further flow cytometry.

Data were expressed as mean ± standard deviation (SD). An unpaired two-tailed Student’s t-test was used for the analysis of two groups. One-way ANOVA was used for comparisons among several different groups. The relation between neutrophil infiltration and osteosarcoma lung metastasis was analyzed using Fisher’s exact test. Chi-square test was performed to analyze the distribution of ELA2 expression among different neutrophil infiltration groups. The relationship between gene expression and patient survival was assessed using Kaplan–Meier survival analysis. All statistical analyses were performed with the GraphPad Prism 9.0. The level of significance was chosen as p < 0.05.

As we previously reported, cyclin E1 and its coding gene, CCNE1, were both highly expressed in OS tissues. The expression of cyclin E1 protein was significantly correlated with overall survival, progression-free survival, and sensitivity to neoadjuvant chemotherapy (11). However, CCNE1 mRNA expression was not associated with patients’ prognosis via analyzing the mRNA data from public databases ( Figure 1A ), which suggests there may be a special post-transcriptional regulation mechanism for cyclin E1 in osteosarcoma.

To further investigate the cyclin E1 regulation mechanism, we first used immunofluorescence and immunohistochemistry staining to analyze cyclin E1 expression and distribution in osteosarcoma cells and 69 osteosarcoma lesions in situ. The staining results showed that cyclin E1 was located in both the osteosarcoma cellular nucleus and cellular cytoplasm, whereas it was located only in the cellular nucleus of the human normal osteoblastic hFOB 1.19 cell line, which indicated that different cyclin E1 isoforms with individual distributions existed in osteosarcoma ( Figures 1B, C ; Supplementary Figure S1 ). The expression of cyclin E1 in osteosarcoma samples was further investigated using a Western blot assay, and the results revealed that, in addition to FL-cyclin E1 in OS, which was shown at 47 kDa, several distinct low molecular weight protein bands were also observed at 33–47 kDa ( Figure 1D ).

According to the cyclin E1 staining intensity in osteosarcoma tissue microarray, four staining patterns (staining score = 0, 1+, 2+, 3+) were defined separately for the cellular nucleus and cellular cytoplasm ( Figure 1B ). OS samples with staining scores of 0 and 1+ were classified into the low cyclin E1 expression group, whereas those with scores of 2+ and 3+ were classified into the high expression group. Correlation analysis revealed that only the cytoplasmic cyclin E1 expression level was significantly associated with the prognosis of osteosarcoma patients, while the nuclear cyclin E1 expression level was significantly associated with the prognosis of osteosarcoma patients, while the nuclear cyclin E1 expression level was not related to overall survival or progression-free survival ( Figures 2A, B ). These findings suggest that cytoplasmic cyclin E1 represents a distinct cyclin E1 subtype that may be more malignant in driving osteosarcoma progression.

As the presence of a subtype of LMW-cyclin E1 in osteosarcoma was shown at 33–47 kDa in Western blot assay ( Figure 1D ), we analyzed the expression level of LMW-cyclin E1 in relation to patient prognosis, and the results showed that the expression of LMW-cyclin E1 was significantly correlated with the overall survival and progression-free survival of osteosarcoma patients ( Figure 2D ). By integrating the immunohistochemical staining data and Western blot results, we further analyzed the correlation between the expression of the LMW-cyclin E1 subtype and the expression of cyclin E1 in the cytoplasm. The results showed that the expression level of cyclin E1 in the cytoplasm was significantly correlated with the expression of LMW-cyclin E1, whereas the expression level of cyclin E1 in the nucleus was not correlated with the expression of LMW-cyclin E1 ( Figure 2C ). Thus, these results indicated that LMW-cyclin E1 is located in the cytoplasm of osteosarcoma cells and is a specific subtype of cyclin E1 that is significantly associated with the prognosis of osteosarcoma.

Many studies have revealed that several proteases can bind to FL-cyclin E1 to induce the formation of LMW-cyclin E1. Calpain 1 and calpain 2 are classical isoforms of the calpain family of calcium (Ca²⁺)-dependent cysteine proteases, and neutrophil elastase (ELA2) is a serine protease predominantly expressed in primary neutrophils; both have been reported to increase LMW-cyclin E1 expression (4, 19). We first detected the protein expression of cyclin E1, calpain 1, calpain 2, and ELA2 by Western blot assay and analyzed the corresponding mRNA levels (CCNE1, CAPN1, CAPN2, ELANE) via RT-PCR in OS cell lines (KHOS, HOS, MG63, U2OS, 143B, SaoS2). LMW-cyclin E1 was expressed in all OS cell lines, especially in 143B cells and U2OS cells ( Figure 3A ), and CCNE1 was shown with the highest expression in U2OS cells ( Supplementary Figure S2 ). Calpain 1, calpain 2, and ELA2 were also present in all OS cells we tested. Considering the qPCR results, we found that calpain 1 and calpain 2 had significantly higher expression levels than ELA2 in each OS cell line ( Figures 3A, B ; Supplementary Figure S2 ).

To investigate the proteases that promote LMW-cyclin E1 formation, we performed a co-immunoprecipitation (Co-IP) assay to evaluate cyclin E1 binding proteins. The results suggested that using the cyclin E1 antibody can enrich the ELA2 but not calpain 1 or calpain 2. It revealed ELA2 may play a regulatory role for cyclin E through interacting with cyclin E1, but calpain 1 and calpain 2 show no direct interaction with cyclin E1 ( Figure 3C ).

For exploring the functions of calpain 1 and calpain 2, small interfering RNA (siRNA) was designed to decrease calpain 1, calpain 2 expression in OS cells, and the knockdown efficiency was validated by PCR assay and western blot ( Supplementary Figures S3A, B ; Supplementary Table S2 ). In U2OS cell lines, we found that knocking down calpain 1 and calpain 2 down-regulated the expression of FL-cyclin E1 and LMW-cyclin E1, so that it also inhibited RB phosphorylation, which is the downstream signal pathway for cyclin E1 ( Supplementary Figure S3C ). Immunofluorescence (IF) staining also showed FL-cyclin E1 and cellular cytoplasmic cyclin E1 (LMW-cyclin E1) were both decreased ( Supplementary Figure S3D ). However, calpain 1 and calpain 2 seem to play a more sophisticated role in osteosarcoma, which is shown with the opposite effects in 143B cell lines ( Supplementary Figures S4A–C ).

At the least, the above results revealed that calpain 1 and calpain 2 were not combined with cyclin E1; they may be the upstream molecules of cyclin E1 to regulate cyclin E1 expression, but not the LMW-cyclin E1 formation promoter. The regulation mechanism of the calpain family in osteosarcoma is worth further investigation.

Recent research has suggested that ELA2 is mainly released by tumor-associated neutrophils, but its expression and release from tumor cells remain uncertain. We found ELA2 was expressed in all OS cell lines ( Figures 3A, B ), and that using the cyclin E1 antibody can enrich the ELA2, indicating that ELA2 may play a regulatory role for cyclin E through direct interaction ( Figure 3C ). Using IF staining, we further validated that ELA2 was present and could be secreted in 143B and U2OS cells ( Figure 3D ). To investigate ELA2 function in OS, we overexpressed ELA2 using plasmid transfection. qPCR results showed ELA2 was significantly upregulated ( Supplementary Figure S4D ), along with the LMW-cyclin E1 upexpression and downstream RB phosphorylation ( Figure 3E ). The immunofluorescence staining results also showed that the cytoplasmic cyclin E1 was up-regulated after ELA2 overexpressed in OS cells ( Figure 3F ).

Cyclin E1 is a cell cycle regulation protein that plays an important role in the G1/S transition. Thus, we further validated whether ELA2 can accelerate osteosarcoma proliferation. The results of the colony formation assay indicated that ELA2 can promote proliferative viability in OS cells ( Figure 4A ). In the mouse osteosarcoma cell line, K7M2-lvCtrl and K7M2-lvELA2 cells were constructed by lentiviral transfection. The transfection efficiency was validated under the fluorescence microscope and qPCR assay to examine the ELA2 mRNA expression, as we can see ELA2 was remarkably overexpressed ( Figures 4B, C ). The cell cycle assay was performed to validate the function of ELA2. The proportion of cells in the G0/G1 phase decreased significantly in the ELA2 overexpression group, indicating accelerated G1/S transition ( Figure 4D ). In vivo, BALB/c mice were used as models of subcutaneous tumor-bearing animals. After 25-day tumor bearing, the mice were sacrificed, and the tumor masses were removed for measurement. The tumor weight and tumor volume were obviously increased in the ELA2-OE group ( Figures 4E, F ). In addition, we performed Ki-67 immunohistochemical staining on xenograft tumor tissue, and the results showed that ELA2 overexpression increased Ki-67⁺ cells compared to the control group ( Figure 4G ), mirroring our in vitro data and confirming potent promotion of tumor cell proliferation.

ELA2 was mainly secreted by tumor-associated neutrophils in various tumor types (20, 21). In osteosarcoma, the function was unclear, and whether neutrophil-derived ELA2 has a parallel effect with osteosarcoma-derived ELA2 was still uncertain. First, we performed IHC staining to observe neutrophil infiltration and ELA2 expression in 34 primary osteosarcoma tissue microarrays and 33 lung metastasis tissue microarrays. The results showed that neutrophil infiltration was slightly higher in primary osteosarcoma, but higher infiltration was seen in lung metastasis osteosarcoma samples ( Figure 5A ). According to the classical IHC staining score, which involved the staining intensity and area, IHC staining of CD66b and ELA2 was scored, and the samples of CD66b staining were all distributed in the 0 and 1 score groups. Considering the few neutrophil infiltrations, more than three neutrophil infiltrations in one field were defined as neutrophil-positive infiltration, and we found that neutrophil-positive infiltration was correlated with osteosarcoma lung metastasis ( Figure 5A ). Remarkably, neutrophil infiltration and ELA2 expression were shown at the same location, and neutrophil-positive infiltration was correlated with more ELA2 expression ( Figure 5B ).

Using density gradient centrifugation, we isolated neutrophils from the peripheral blood of healthy donors, and the purity was validated by flow cytometry ( Figure 5C ). The results of Giemsa stain also showed that the isolated neutrophils had multinuclei and were dark purple, which is the typical neutrophil feature ( Figure 5D ). To investigate neutrophils’ effect on osteosarcoma, 5 × 10⁵ neutrophils were cocultured with 5 × 10⁵ osteosarcoma cells in contact and noncontact ways for 24 h ( Figure 5E ). After 24-h coculture, the osteosarcoma cells’ total proteins were collected and detected by Western blot assay. Contacting coculture with neutrophils increased osteosarcoma ELA2 expression and promoted LMW-cyclinE1 expression. No obvious expression change was observed in noncontacted coculture with neutrophils ( Figure 5F ). To further validate the effect of neutrophil-derived ELA2 on osteosarcoma, AZD9668, a highly selective inhibitor of ELA2, was used. AZD9668 showed no toxicity for osteosarcoma cells from 1 to 128 nM ( Supplementary Figure S5 ); 50 nM AZD9668 was selected for further experiments. Osteosarcoma cells were cocultured with different concentrations of neutrophils. As the concentration increased, the expressions of ELA2 and LMW-cyclin E1 were upregulated, promoting downstream RB phosphorylation. At the same time, AZD9668 could inhibit ELA2 to reverse neutrophils’ effect on osteosarcoma ( Figure 5G ).

NETs are web-like structures composed of granule proteins and decondensed chromatin, which are induced by surgical stress or inflammation. It is currently assumed that ELA2 is one of the main components of NETs. First, we verified the presence of NETs in osteosarcoma by immunofluorescence staining, showing that histone H3 was released into the extracellular space along with ELA2 ( Figure 6A ). In addition, the concentration of NETs in individually cultured neutrophils was 6.649 ng/ml. After coculturing with OS cells, NET formation was significantly increased, with the concentration rising to 48.90 ng/ml. Inhibiting NET formation with GSK484 also decreased the NET concentration to 44.61 ng/ml in the neutrophil-OS cell coculture system ( Figure 6B ). The results further revealed that OS can induce NET formation, and that this can be reversed by GSK484. To validate whether NETs could promote OS proliferation, we collected the conditioned medium of individually cultured neutrophils, cocultured neutrophils, and cocultured neutrophils treated with GSK484, and added them to the OS cell medium. As the results show, the conditioned medium of individually cultured neutrophils decreased the OS cells. However, after coculturing with OS cells, the neutrophils were educated into tumor-associated neutrophils, and NETs were formed. The conditioned medium of the coculture system could promote OS cell proliferation, and this effect could be inhibited by the NET inhibitor GSK484 ( Figure 6C ).

As for ELA2 expression, we also performed the ELISA assay to test the ELA2 concentrations in individually cultured neutrophils and the cocultured system. As the results show, after coculture with OS cells, ELA2 expression was significantly increased in the coculture system, which also indicated NETs were formed and triggered ELA2 release ( Figure 6D ). At the same time, the results of the ELISA assay also supported our previous findings that neutrophils had higher expression of ELA2 than OS cells, which were the main source of ELA2 in osteosarcoma.

We further investigated the functions of neutrophils in osteosarcoma malignant progression in vivo. Balb/c mice were used to construct a subcutaneous osteosarcoma model, and anti-ly6G was administered via intraperitoneal injection every 4 days to exhaust neutrophils. After 25 days of treatment, the tumor tissues were removed, and the IHC results showed the neutrophils were significantly exhausted after anti-Ly6G treatment ( Figures 6E, F ). Remarkably, neutrophil exhaustion obviously hindered osteosarcoma progression; both tumor weight and volume were significantly inhibited ( Figures 6G, H ). In addition, the lungs of mice were harvested for histological examination. In the anti-ly6G group, there was no lung metastasis. In the Ctrl group, the lung metastasis niche was observed ( Supplementary Figure S6 ). While the results did not reach statistical significance, we observed a trend suggesting that neutrophil depletion may reduce lung metastasis formation. After depleting neutrophils, the expression level of ELA2 was also significantly deregulated ( Figure 6I ). All the results indicated that targeting neutrophils and their ELA2 expression may be a prospective immunotherapy strategy for osteosarcoma ( Figure 6J ).