OSBPL2-mediated lipid metabolism alteration governs lung cancer stem cells properties

Abstract

Lung cancer is the first leading cause of cancer death worldwide. oxysterol-binding protein-like 2 (OSBPL2), is a lipid transport protein regulating cholesterol homeostasis. Here, we clarified the previously unreported role of OSBPL2 in lung cancer stemness properties. We observed that OSBPL2 reduced cholesterol content by HPLC-MS. It inhibited the accumulation of lipid droplets (LDs) in lung cancer. OSBPL2-mediated lipid transportation significantly suppressed tumor sphere formation, stemness markers expression and in vivo tumorigenesis and tumor metastasis. In clinical specimens, we also demonstrated that OSBPL2 repressed the expression of Lung cancer stem-like cells (LCSCs) markers-ALDH1A1, CD133 and Nanog. The level of OSBPL2 was negatively correlated with malignant of lung cancer, such as tumor stage progression and lymph node metastasis. Taken together, these findings illustrated that OSBPL2-mediated lipid transportation inhibited the stemness and aggressiveness of lung cancer cells. OSBPL2 was a potential therapeutic target to develop novel cancer-preventive compound.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13287-026-04919-4.

Introduction

Lung cancer is the first cause of cancer death worldwide. Approximately 85% patients are diagnosed as non-small cell lung cancer (NSCLC) [1, 2]. Recent evidences showed that cancer stem cell was a key factor in contributing to metastasis and recurrence of lung cancer [3]. Thus, identifying efficient target for cancer stem cell formation in lung cancer is very meaningful to develop therapeutic strategies.

Lung cancer stem-like cells (LCSCs) are a rare subpopulation exhibiting stemness properties in lung cancer. LCSCs are able to differentiate, self-renew and express stemness markers, including ALDH1A1, CD133 and Nanog. LCSCs are the major cause for failure of cancer therapies and progression in lung cancer [3, 4].

Oxysterol-binding protein-like 2 (OSBPL2), also known as oxysterol-binding protein-related protein 2 (ORP2). It localizes to lipid droplets (LDs) and transports excessive lipid outside cells. Thus, it modulates lipid metabolism reprogramming and LDs content. LDs are a kind of dynamic organelles. There functions include lipid storage, energy supplying for proliferation, cell signaling and lipid metabolism reprogramming. Indeed, LDs are metabolic determinants for cancer stemness. Production of fatty acid and accumulation of LDs sustained the self-renew and reproduction capacity of cancer stem cells [5, 6]. Lipid metabolism includes lipid import and export, lipolysis, fatty acid oxidation, de novo lipogenesis, and lipid desaturation. Abnormal lipid metabolism has been documented to promote the cancer stemness. Liu reported that enhanced phospholipid metabolism and synthesis of fatty acid activated Hippo/YAP and Wnt/β-catenin signaling, which contributed to the maintenance of cancer stemness [7]. According to the above, OSBPL2-correlated LDs alteration is a potential target for inhibition of cancer stem cell properties. Revealing the underlying mechanism will provide novel insights on developing individualized therapeutic regimens.

In tumor progression, OSBPL2 was also reported to play an important role. Lin et al. found that OSBPL2 defect supported colorectal cancer (CRC) growth and metastasis. The downregulation of OSBPL2 aggravated CRC progression through PARP1/ZEB1 axis and VCAN-correlated ERK signaling [8]. Holý et al. reported that germline variant in OSBPL2 was associated with poor survival of patients with breast cancer [9]. These evidences confirmed that OSBPL2 was an essential molecule in tumor progression. This was consistent with our findings. We investigated the expression characteristics, prognostic value and function of OSBPL2 by TCGA data mining. We found OSBPL2 was significantly decreased in lung cancer tissues and patients with low OSBPL2 expression had a worse survival. OSBPL2 was a tumor suppressor gene.

Taken all the clues above together, we proposed some interesting problems: recently, researchers proved the function of OSBPL2 in tumor progression. However, did OSBPL2 play roles in lung cancer through its LDs modulation effects? Was it really involved in maintenance of LCSCs properties? These issues were fascinating and worthy of further explorations.

To the best of our knowledge, this research is the first one to reveal the important potentiality of OSBPL2 in LCSCs features regulation and tumor progression for lung cancer (Fig. 1). Identifying the key factor OSBPL2 in cancer stemness, tumor aggressiveness and recurrence of lung cancer will provide a potential target on developing individualized therapeutic regimens.

Fig. 1.

Flow diagram of our investigation

Results

The expression pattern of OSBPL2 and it is a potential tumor suppressor

In bioinformatic analysis of TCGA tissue samples, we found that OSBPL2 expressed generally in different cancer (Figure S1A). It was not specifically expressed in one certain cancer. But OSBPL2 mRNA level was significantly lower in lung adenocarcinoma (LUAD) and Lung Squamous Cell Carcinoma (LUSC) samples than that in normal tissues. It suggested the potential tumor suppressor role of OSBPL2 (Figure S1B). By CCLE online analysis, OSBPL2 mRNA expression in different cancer cell lines was showed in Figure S1D. We further analyzed another dataset, CancerSEA. Next, we found that OSBPL2 was negative correlated with stemness in lung cancer (Fig. 3A). Thus, the lowly expression of OSBPL2 in lung cancer suggested that it might have some biological function in stemness.

Fig. 3.

OSBPL2 inhibits the sphere formation and stemness markers expression of LCSCs. A OSBPL2 was negatively correlated with stemness in lung cancer based on CancerSEA database. B The self-renewal activity of cells with different treatment were evaluated by 3D sphere formation assay. The number of spheres was calculated in (C). Scale bar=200 μm. D Determination for stemness marker expression of L-Vector+DMSO, L-OSBPL2 + DMSO, L-OSBPL2 + cholesterol groups using flow cytometry. The data were displayed as means ± s.e.m. Comparisons of four or three groups was statistically analyzed by one-way analysis of variance (ANOVA). p < 0.05 was considered as significant. *p < 0.05; **p < 0.01. Error bars represented s.e.m

OSBPL2 significantly inhibited cholesterol production and LDs accumulation in lung cancer cells

We further analyzed the expression level of OSBPL2 and we found it was significantly lower in lung adenocarcinoma (n = 487) vs. normal tissues (n = 347) (Fig. 2A). Kaplan-Meier curve demonstrated that low expression of OSBPL2 was associated with poor prognosis of patients with lung cancer (overall survival: HR = 0.67, p = 0.0088 < 0.05) (Fig. 2B). These data suggested that OSBPL2 was a potential tumor suppressor of prognostic value. To study the role of OSBPL2 in progression of lung cancer, the stable cell lines were constructed by transducing L-Vector or L-OSBPL2 virus into A549 cells. As OSBPL2 is a lipid transport protein regulating cholesterol homeostasis and LDs content, we tested the function of OSBPL2 in lipid metabolism of lung cancer. Using specific LD probe-Bodipy493/503, we found that the ectopic expression of OSBPL2 markedly inhibited LD accumulation (L-OSBPL2 vs. L-Vector cells: 6367.67 ± 1297.70 vs. 13615.33 ± 1484.80) (Fig. 2C, F). si-OSBPL2 promoted LD accumulation in lung cancer cells (Figure S2E-F). Following, we determined the concentration of cholesterol species by HPLC-MS. The cholesterol abundance was notably decreased from (25.90 ± 0.20) ×10⁷ (L-Vector cells) to (19.87 ± 1.00) ×10⁷ (L-OSBPL2 cells) (Fig. 2D, G). Loss of function assay validated the role of OSBPL2 in inhibiting cholesterol accumulation (Figure S2 C-D). Finally, because OSBPL2 was a pivotal cholesterol binding protein in LD production pathway, revealing the structural basis underlying OSBPL2-substrate affinity was very important. We conducted molecular dynamics simulation by Amber and AutoDock software. The pair of OSBPL2-cholesterol molecular model was constructed applying 3D X-ray crystal structure from PDB database (http://www.rcsb.org/). To evaluate the affinity between OSBPL2 and cholesterol, we calculated the molecular mechanics energies, the MM-GBSA related binding free energy and the contribution of surrounding residues. Free energy of OSBPL2 was − 2.09 kcal/mol. However, the free energy of OSBPL2-cholesterol was much lower, -9.65 kcal/mol. This data demonstrated that OSBPL2 was prone to bind to its substrate-cholesterol. The OSBPL2-cholesterol complex had the best orientation and conformation, thus obtained the lowest free energy. The dominant residue of OSBPL2 favoring metabolite binding was Asn153, forming one hydrogen bond (Fig. 2E). The conformation alteration of OSBPL2 was the main reason for its higher affinity to substrate and transport efficiency of cholesterol.

Fig. 2.

OSBPL2 inhibited cholesterol production and LDs accumulation in lung cancer cells. A OSBPL2 expression in LUAD based on TCGA data. B Low expression of OSBPL2 predicted unfavorable prognosis of LUAD patients (n = 503). C Representative graphs of immunofluorescence staining for lipid droplets in L-Vector or L-OSBPL2 cells and quantitative analysis (F). Scale bar=100 μm. D Determination of cholesterol concentration in L-Vector or L-OSBPL2 groups by HPLC–MS and its quantification analyses (G). E The conformations of OSBPL2 binding to its substrate-cholesterol. The protein-ligand binding complexes in ribbon (a) and surface (b) models. The molecular mechanics energies for OSBPL2 that presented favorable interactions with its substrate-cholesterol were listed in the table. The data were displayed as means ± s.e.m. Significance was calculated by unpaired two-tailed Student’s t-test (two groups). *p < 0.05; **p < 0.01; ns not significant. Error bars represented s.e.m

Taken together, OSBPL2 significantly repressed cholesterol production and LDs accumulation in lung cancer. The tumor suppression effect of OSBPL2 might dependent on OSBPL2-mediated lipid transportation.

OSBPL2 was correlated with lipid metabolism and thus interfered the maintenance of LCSC features in lung cancer

To explore the potential function of OSBPL2 in lung cancer, we conducted bioinformatic analyses applying CancerSEA online tools (http://biocc.hrbmu.edu.cn/CancerSEA/). CancerSEA Providing a cancer single-cell functional state atlas, involving 14 functional states of 41,900 cancer single cells from 25 cancer types. ScRNA-seq provides an unprecedented opportunity to explore the functional heterogeneity of cancer cells. Querying the gene of interest would generate functional states relating to different cancer types. Our analysis data revealed that OSBPL2 negatively modulating tumor stemness (correlation strength: -0.36; P value: 0.012) (Fig. 3A). OSBPL2 was a lipid metabolism regulatory protein. Abnormal lipid metabolism has been documented to promote the cancer stemness of lung cancer. Excessive lipid sustained the self-renew and reproduction capacity of cancer stem cells [5, 6]. Fang et al. reported that targeting lipid synthesis could inhibit tumor stemness and lung cancer progression [10]. Thus, we detected the relationship between OSBPL2-correlated LDs alteration and maintenance of cancer stem cell properties in lung cancer. In the first step, LCSCs were separated and enriched from A549 cells by serum-free suspension culturing. In serum-free medium (SFM), tumorspheres generated from lung cancer cell lines could form suspend spheroids and present an active tumorigenic property [11]. The second step, we examined the impact of OSBPL2 and lipid metabolism on LCSC characteristic. 3D tumorsphere formation assay showed that OSBPL2 markedly inhibited the tumorsphere formation ability of LCSCs (L-OSBPL2 + DMSO group vs. L-Vector+DMSO group: 4 ± 1 vs. 12 ± 2 spheres/10³ cells). si-OSBPL2 facilitated the tumorsphere formation of LCSCs (Figure S2A). However, the addition of cholesterol dose-dependently attenuated the effect of OSBPL2 on tumorsphere growth (L-OSBPL2 + DMSO group vs. L-OSBPL2 + cholesterol 25ng/mL group vs. L-OSBPL2 + cholesterol 50ng/mL group: 4 ± 1 vs. 7 ± 1 vs. 13 ± 2 spheres/10³ cells) (Fig. 3B-C, Figure S1C, E). Moreover, extreme limiting dilution analysis (ELDA) demonstrated that cholesterol treatment promoted OSBPL2-inhibited spere formation ability (Figure S5). Consistently, flow Cytometric analyses suggested that OSBPL2 significantly reduced the expression of LCSC markers ALDH1A1, CD133 and Nanog in lung cancer sphere-forming cells. But treatment of cholesterol obviously rescued the cancer stem cell property in control group (ALDH1A1, for example: L-Vector+DMSO group vs. L-OSBPL2 + DMSO group vs. L-OSBPL2 + cholesterol 50ng/mL group: 15.40 ± 2.41 vs. 9.23 ± 1.05 vs. 21.27 ± 2.38%) (Fig. 3D–G, Figure S3). The experimental data indicated that OSBPL2-related lipid transport could decreased LCSC properties, which further blocked lung cancer progression. While excessive cholesterol could interfere the effect of OSBPL2 inhibiting tumor stemness.

OSBPL2 inhibited LDs accumulation and aggressiveness of lung cancer in vivo

To investigate the effect of OSBPL2-induced lipid metabolic change on tumor aggressiveness in vivo, we constructed xenograft models by L-Vector or L-Osbpl2 (mouse gene) LCSCs derived from mice lung cancer cell line-LLC. Twenty male Balb/c mice (8-weeks old) were subcutaneously injected in flank to construct xenograft models. The growth of xenografts from different groups was recorded every 5 days. 5 mice were excluded for overthin, weak or suddenly death in each group. The results showed that Osbpl2 significantly suppressed tumor proliferation (L-Osbpl2 vs. L-Vector group at day 25: 249.50 ± 25.24 vs. 830.20 ± 172.42mm³) (Fig. 4A, C). Of note, the body weights of mice from different groups have no significant change (Fig. 4B). According to the Kaplan-Meier curve, we found that there were 20% tumor-free mice in L-Vector group at the 8th day, while there were still 60% tumor-free mice in L-Osbpl2 group (Fig. 4D). This suggested that Osbpl2 markedly repressed the tumorigenesis of LCSCs. However, malignant phenotype was evaluated not only by tumorigenic and proliferation features, but also tumor metastasis. Thus, we constructed peritoneal metastasis models to detect the impact of Osbpl2 on tumor metastasis. The phenomenon prompted us that ectopic expression of Osbpl2 obviously depressed the metastasis capacity of LCSCs (numbers of intraperitoneal metastatic tumors, L-Osbpl2 vs. L-Vector group: 4 ± 2 vs. 15 ± 3) (Fig. 4E, F). In vitro transwell assay also demonstrated that OSBPL2 inhibited the migration of lung cancer cells (Figure S4). Furthermore, we also determined the metabolic influence of Osblp2 in metastatic tumors generated by LCSCs. In immunofluorescence assay of metastatic tumors, we found that the LDs accumulation in tumors was notably inhibited by Osbpl2 (average integral optical density of LDs, L-Osbpl2 vs. L-Vector group: 5836 ± 991 vs. 14296 ± 1679) (Fig. 4G, H). Consistently, HPLC-MS data indicated that the LD component-cholesterol was significantly accumulated in tumors of L-Vector group but was reduced in L-Osbpl2 group (relative abundance, L-Osbpl2 vs. L-Vector group: 3.22 ± 0.64 vs. 6.36 ± 0.61(×10⁶)) (Fig. 4I, J).

Fig. 4.

OSBPL2 inhibited lipid accumulation of lung cancer in vivo. A Xenograft tumors of Balb/c mice in L-Vector or L-Osbpl2 group (for each group, n = 5). B Line chart showed body weight and C average tumor volume of xenograft mice models for each group. D Kaplan–Meier curves for tumor-free mice in L-Vector or L-Osbpl2 group (n = 10). Censored event was tumor growth. E Representative graphs of peritoneal metastasis tumors of L-Vector or L-Osbpl2 mice and F quantitative analysis. H Metastasis tumors in L-Vector or L-Osbpl2 group were immunofluorescence stained by Bodipy 493/503, a LD labeling probe (for each group, n = 5). Scale bar = 100 μm. G Quantified analysis of (H). I Detection of cholesterol concentration in L-Vector or L-Osbpl2 groups by HPLC–MS and its quantification analysis (J). The data were displayed as means ± s.e.m. Significance was calculated by unpaired two-tailed Student’s t-test (two groups). *p < 0.05; **p < 0.01; ns, not significant. Error bars represented s.e.m

OSBPL2-mediated lipid transportation reduced the stemness and aggressiveness of lung cancer in clinical specimens

To corroborate the correlation between OSBPL2-mediated lipid transportation and stemness/progression of lung cancer, immunohistochemistry, immunofluorescence and flow cytometry were performed in clinical specimens. Firstly, OSBPL2 expression was evaluated by immunohistochemistry in samples from patients with lung cancer. According to the high or low expression of OSBPL2, the patients were classed to two groups: LUAD-OSBPL2^high and LUAD-OSBPL2^low groups (Fig. 5B, G). The computed tomography (CT) scanning graphs of patients from LUAD-OSBPL2^high and LUAD-OSBPL2^low groups were showed in Fig. 5A. Tissues immunofluorescence analyses revealed that LDs were significantly more accumulated in LUAD-OSBPL2^low groups (average integral optical density of LDs, LUAD-OSBPL2^high vs. LUAD-OSBPL2^low: 3950 ± 1031 vs. 15051 ± 2268) (Fig. 5C, H). This finding suggested that OSBPL2 also influenced the lipid metabolism in human tissues of lung cancer. Further, fresh fraction of the clinical specimens was digested to single-cell suspension. Then, cells were examined for expression of LCSCs markers by flow cytometry. We observed that low OSBPL2 significantly elevated ALDH1A1, CD133 and Nanog expression (ALDH1A1, for example: LUAD-OSBPL2^high vs. LUAD-OSBPL2^low: 1.09 ± 0.18 vs. 12.90 ± 1.25%) (Fig. 5D-F, I-K). To further understand the effect of OSBPL2 on the aggressiveness of lung cancer, we analyzed expression profile and patients’ clinical information of lung adenocarcinoma (LUAD) from TCGA. The results showed that OSBPL2 expressed significantly lower in advanced LUAD (III + IV stage), compared to that in early LUAD (I + II stage) (Fig. 5L). Similarly, OSBPL2 expression was markedly lower in LUAD patients with lymph node metastasis N1 + N2, in contrast to that in patients with lymph node metastasis N0 (Fig. 5M). According to these data, we concluded that low OSBPL2 expression are related to LD accumulation and tumor stemness, and thus supported progression of lung cancer to advanced tumor.

Fig. 5.

OSBPL2-mediated lipid transportation reduced the stemness and aggressiveness of lung cancer. A Representative CT scanning graphs of OSBPL2^high and OSBPL2^low LUAD patients (n = 5/group). B Immunohistochemical analysis of OSBPL2 expression in LUAD-OSBPL2^high and LUAD-OSBPL2^low specimens (n = 5/group). Scale bar=100 μm. C Bodipy493/503 staining for OSBPL2-high or OSBPL2-low LUAD specimens (n = 5/group). Scale bar=100 μm. D–F Testing of ALDH1A1, CD133 and Nanog expression in LUAD-OSBPL2^high or LUAD-OSBPL2^low tumor tissue by flow cytometry (n = 5/group). G Quantitative analysis of (B). H Quantification of (C). I–K Statistical analysis of (D–F). L OSBPL2 expression were significantly lower in advanced LUAD patients based on TCGA-LUAD data (n = 522). M OSBPL2 expression were decreased in LUAD exhibited more malignant phenotype according to TCGA-LUAD data (n = 522). The data were displayed as means ± s.e.m. Significance was calculated by unpaired two-tailed Student’s t-test (two groups). *p < 0.05; **p < 0.01. Error bars represented s.e.m

Discussion

In our study, we found that OSBPL2 interfered LCSCs properties and inhibited tumor aggressiveness though OSBPL2-mediated lipid transport. To our knowledge, this is the first investigation illustrate the role of OSBPL2 in LDs accumulation and LCSCs properties (Fig. 6).

Fig. 6.

The schematic diagram depicting the role of OSBPL2-mediated lipid transportation in modulating cell stemness and aggressiveness of lung cancer. OSBPL2 is responsible for cell lipid (cholesterol or TG etc.) transportation to extracellular matrix. Consequently, OSBPL2 interfered LDs synthesis and accumulation in cancer cells, which suppressed stemness phenotype formation and thus tumor progression

OSBPL2 is a critical molecule in the cholesterol homeostasis pathway. It localizes to LD and regulates LD accumulation. LDs are a kind of dynamic organelles. There functions include lipid storage, energy supplying for tumor proliferation, cell signaling of aggressiveness [12–14]. Thus, it modulates lipid metabolism reprogramming and thus tumor malignancy [15]. Indeed, LDs are also metabolic determinants for cancer stemness. Accumulation of LDs sustained the self-renew and reproduction capacity of cancer stem cells [5, 6]. Analyses for TCGA data showed that low expression of OSBPL2 in lung cancer was associated with poor prognosis. It was a potential tumor suppressor. Using specific LD probe-Bodipy493/503 and HPLC-MS, we found that OSBPL2 markedly decreased cholesterol abundance and inhibited LD accumulation. To deeply investigate the structural mechanism, we conducted molecular dynamics simulation by Amber and AutoDock software. The data demonstrated that OSBPL2 was prone to bind to its substrate-cholesterol. The OSBPL2-cholesterol complex had the best orientation and conformation, thus obtained the lowest free energy, -9.65 kcal/mol. The conformation alteration of OSBPL2 was the main reason for its higher affinity to substrate and transport efficiency of cholesterol.

LCSCs are a small population derived from lung cancer cells with self-renewal and tumorigenic features. These cells are the major cause for failure of cancer therapies and progression in lung cancer [3, 4]. Extensive research has demonstrated that the maintenance of cancer stem cell-like properties relied on lipid metabolism alteration [16–19]. Cancer stem cells exhibit distinct metabolic properties in contrast to differentiated cells, such as vigorous lipid metabolism, LD accumulation [16, 20]. Lipid metabolic alteration not only satisfies the energy for cell growth, biomass generation, but also contributed to the oncogenic signaling for the tumor progression. Kuramoto et al. found that LD-related signal was involved in the maintenance of the properties of CSCs [19]. Menard et al. reported that stress-induced LD accumulation was correlated to increased spheroid-forming capacity during lung metastasis of tumor in vivo [18]. These studies suggested the positive role of LD metabolism in supporting tumor stemness. In our study, we constructed LCSCs with ectopic expression of OSBPL2 to explore the impact of OSBPL2 on lipid metabolism and stemness of LCSCs. It was confirmed that OSBPL2 modulated cholesterol transport and blocked LD accumulation in LCSCs. The self-renewal activity of cells with different treatment were evaluated by 3D tumorsphere formation assay. We observed that OSBPL2 notably suppressed self-renewal of LCSCs. The expression of LCSC specific markers, such as ALDH1A1, CD133 and Nanog, were significantly lower in L-OSBPL2 group. Re-adding of cholesterol enhanced the stemness of LCSCs in L-OSBPL2 group, emphasizing the role of lipid metabolism in maintenance of stemness. Cholesterol homeostasis played a critical role in maintaining stemness. It could be modulated by cell signaling pathways of SREBP2 or MYC and thus trigger development of cancer stem cells [21–23]. Feeding excess cholesterol or increasing endogenous cholesterol production could promote intestine stem cell proliferation [22]. This phenomenon was consistent with our findings. Further, in line with cells from lung cancer, patients with lung cancer who presented high expression of OSBPL2 also showed markedly lower level of ALDH1A1, CD133 and Nanog, suggesting that OSBPL2-mediated lipid metabolism was correlated to the modulation of LCSCs properties in vitro and in vivo. In clinical specimens, the level of OSBPL2 was negatively correlated with malignancy of lung cancer, such as tumor stage progression and lymph node metastasis.

As far as we know, this is the first study to clarify the important potentiality of OSBPL2 on regulating LCSC features and progression for lung cancer. OSBPL2-mediated lipid transportation markedly suppressed tumorsphere formation, stemness markers expression and in vivo tumorigenesis and tumor metastasis. Our findings further demonstrated that targeting OSBPL2 mediated-lipid metabolism in LCSCs could be a promising therapeutic avenue against aggressiveness of lung cancer. For example, OSBPL2-related exosome could be combined with current radiotherapy or chemotherapy treatment paradigms. In brief, preloading milk exosome with OSBPL2 inhibitor, and applying this exosome targeting LCSCs (linking ALDH1A1 antibody in package molecules), will specifically inhibit the LD accumulation and proliferation of LCSCs. Thus, it might enhance the therapeutic effects of radiotherapy or chemotherapy, and repressed the progression of lung cancer.

Limitations of this study: (1) Further investigations using a point mutant should be performed to study the necessity of OSBPL2-Asn153 for cholesterol binding and lipid transport; (2) For time limitation, the knock-down experiments of OSBPL2 using animal models were not conducted; (3) The sample sizes for some assays are small (n = 3); (4) The relationship between OSBPL2 and stemness signaling such as Wnt/β-catenin was failed to explored; (5) The treatment-resistant or metastatic patient samples should be collected and the expression of OSBPL2 should be validated in these samples.

Materials and methods

Data extraction and bioinformatic analyses

The expression profiles and clinical information of 522 LUAD patients were downloaded from TCGA database (https://xenabrowser.net/datapages/). The gene functional analyses of OSBPL2 were performed using CancerSEA online tools (http://biocc.hrbmu.edu.cn/CancerSEA/). CancerSEA is the first dedicated database that aims to comprehensively decode distinct functional states of cancer cells at single-cell resolution.

Cell culture and lentiviral vectors construction

Human lung cancer cell line A549 and mouse lung cancer cell line LLC were purchased from National Collection of Authenticated Cell Culture, Chinese Academy of Science (Shanghai, China). A549 was cultured in RPMI 1640 medium (Hyclone, CA, USA) plus 10%FBS (Gibco, CA, USA) and 1% penicillin and streptomycin (Beyotime, Shanghai, China). LLC was maintained in DMEM medium (Hyclone, CA, USA) supplemented with 10%FBS and 1% antibiotics. All the cells were cultured in 37℃, 5%CO₂.

To construct the lentiviral vector, a full-length cDNA clone of human OSBPL2 (gene ID 9885) was purchased from Miaolingbio (Wuhan, China). After PCR, the CDS of OSBPL2 was inserted into lentiviral vector pCDH-CMV-MCS-EF1-copGFP (CD511B-1, System Bioscience Inc., CA, USA). The restriction enzyme sites were EcoRI and BamHI. The Forward primer: 5’-GCGAATTCATGAACGGAGAGGAAG-3’; Reverse primer: 5’- GCGGATCCTCAGTAGATATCTGGGC-3’. Followingly, pCDH-CMV-MCS-EF1-copGFP-OSBPL2 (8 µg) or empty plasmid, plus lentiviral package plasmids pMD2.G (4 µg) and psPAX2 (6 µg) were transfected to 293T cells. Next, lentiviral particles in supernatants were purified and condensed using PEG8000. Lentiviral containing pCDH-CMV-MCS-EF1-copGFP-OSBPL2 or empty vector stably transducted A549 cells were called L-OSBPL2 (Lenti-OSBPL2) or L-Vector (Lenti-Vector) cells. L-Osbpl2 (Lengti-Osbpl2) cells were constructed by similar procedure using Osbpl2 cDNA (Osbpl2: gene ID 228983, murine homologous gene of OSBPL2) (Miaolingbio, Wuhan, China).

LCSCs culture and identification

LCSCs were separated and enriched from A549 cells by serum-free suspension culturing. In serum-free medium (SFM), tumorspheres generated from A549 could form suspend spheroids and present an active tumorigenic property [11]. In briefly, A549 cells were cultured in a stem cell specific SFM, including DMEM/F12 medium(Hyclone, CA, USA), 20 µg/L bFGF, 20 µg/L EGF, and 2% B27(Beyotime, Shanghai, China). The new medium was changed every two days. Upon the serum free condition, cells without stemness would go to apoptosis. But LCSCs would grow quickly to a circular spheroid with good refraction. Cells in the spheroid were combined closely to each other. The stemness marker of LCSCs was CD133, ALDH1A1, Nanog.

Tumorsphere formation assay

Cells were maintained in a stem cell specific medium, SFM. There were 4 groups with different treatment: L-Vector virus+DMSO, L-OSBPL2 virus+DMSO, L-OSBPL2 virus+Cholesterol 25ng/mL and L-OSBPL2 virus+Cholesterol 50ng/mL. Cells in different groups were seeded onto the 24-well ultralow attachment plate (2000 cells/well). After 7d treatment, the numbers of spheres (spheres diameter>100 μm) were counted and compared between different groups.

Extreme limiting dilution analysis (ELDA)

A549 cells were seeded on 96-well ultra-low attachment surface plates at the concentration of 64, 32, 16, 8, 4 and 1 cells/well, cultured in 100µL SFM. After 5–7 days, the numbers of tumor spheroid from different groups were calculated and the frequency of sphere-initiating cells was analyzed using ELDA online tools (https://bioinf.wehi.edu.au/software/elda/).

Flow cytometry analysis

Sphere-forming cells or tumors were collected and digested to single cells. Antibodies applied in LCSCs specific markers detection were anti-ALDH1A1-Alexa flour488, anti-CD133-Alexa flour594, anti-Nanog-APC-Cy7 (eBioscience, CA, USA). The experiments were done as previously reported [24]. In briefly, cells in different groups were washed by PBS and fixed with 70% precooled ethanol, 2 h. Next, cells were washed and stained by anti-ALDH1A1-Alexa flour488, anti-CD133-Alexa flour594, anti-Nanog-APC-Cy7, 37℃ at darkness for 30 min. The percentage of positive cells in different groups were tested by Agilent NovoCyte D3000 flow cytometer.

Metabolites extraction and HPLC-MS analysis

Metabolites extraction and HPLC-MS analysis were conducted according to former protocols in our investigation of lipid metabolism [15]. In briefly, cell pellets were resuspended by 480µL solvent (MTBE: MeOH = 5:1) and 200µL H₂O. Next, cells were 35 Hz homogenized for 4 min. This step was repeated thrice. The samples were − 40℃ incubated for 1 h, and centrifuged. 300µL supernatant of each sample was dried at 37℃ in a vacuum concentrator. The dried samples were resuspended in 100µL solution (methanol: dichloromethane = 1:1) with sonication on ice, 10 min. Then, they were centrifuged and 75µL supernatants were pipetted for lipidomic analysis.

The HPLC separation and identification was performed using a ExionLC Infinity series UHPLC System (AB Sciex, USA), equipped with a Kinetex C18 column (2.1 × 100 mm, 1.7 µm; Phenomen). MS/MS spectra detection was carried out by TripleTOF 5600 mass spectrometer with acquisition software (Analyst TF 1.7; AB Sciex) on an information-dependent basis. For every cycle, the Top12 intensive precursorions (intensity > 100) were chosen for MS/MS, 45 eV collision energy (CE). The in-house program was conducted with R for automatic analysis of metabolite data. Raw data files were converted (wiff to mzXML format) through the “msconvert” program from ProteoWizard (version 3.0.19282). Next, the data were analyzed applying LipidAnalyzer. Peak detection for MS1 data was processed via CentWave algorithm in XCMS. The specific lipid determination was completed with a spectral match which was dependent on an in-house MS/MS spectral library.

Peritoneal carcinomatosis models

Male BALB/c mice (8-week-old, ~ 20 g weight) were randomly classified to 2 groups. Completely randomized design was used to generate the randomisation sequence. Mice were injected intraperitoneally by L-Osbpl2 or L-Vector lentiviral transducted LLC cells (5 × 10⁶cells/100µL) (each group: n = 10). The provenance of BALB/c mice, health/immune status, genetic modification status, genotype were reported previously [25]. These mice were usually used in metabolic and tumor investigations [15, 26]. The sample size was decided according to literatures [15, 26]. Mice were monitored for bodyweights and status once/2 days. After 3 weeks, these tumor-vaccinated mice were sacrificed. Animal anaesthesia was not used in this study. The method of euthanizing mice was cervical dislocation. The peritoneal metastasis tumors were counted and compared [15]. Mice were fed with whole value grain feedstuff in SPF level lab, with constant temperature of 25℃. During the experiments (before the time point of 3 weeks), the overthin, weak and suddenly dead mice would be excluded. But there were no exclusions in these experiments.

Statistics

Each experiment was repeated for three times and representative data was showed. All the statistical analyses were performed by GraphPad 9.0 (GraphPad Software Inc., CA, USA). All data were displayed as means ± s.e.m. and calculated with one-way ANOVA (multi-groups) or Student’s t-test (two groups). The percentage of tumor-free mice was evaluated using Kaplan-Meier curve with log-rank test. Statistical tests were two-sided and p < 0.05 was identified as significant. We used statistical tests (such as t-test or chi-square test) to evaluate whether the data met the assumptions. If the assumptions were not met, we would choose a more appropriate statistical techniques or increase the sample size, improving the stability of the results.

Statement

Our work has been performed and reported in line with the ARRIVE guidelines 2.0. To minimise the potential confounders, we used randomly allocation method of mice, cage location and crossover design of treatment/measurement order.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Hongtao Liu, Pei Yin and Ying Wang designed the study and performed the wet-lab experiments. Guangliang Bian conducted the TCGA data analyses. Ying Wang did the bioinformatic analyses for function prediction. Hongtao Liu and Caihua Xu collected the patients’ samples and completed clinical information analyses. Hongtao Liu and Ying Wang wrote the manuscript and graphed the figures. All authors reviewed and proved the final manuscript.

Funding

The protocol (including the research question, key design features, and analysis plan) was prepared before the study in project CKY2022-13. This work is supported by Chongming District Science and Technology Committee (CKY2022-13) to Hongtao Liu; Top Talent Support Program for young and middle-aged people of Wuxi Health Committee (BJ2023062, HB2020042), and the Fundamental Research Funds for the Central Universities (JUSRP221036) to Ying Wang; Wuxi Taihu Lake Talent Plan, Supports for Leading Talents in Medical and Health Profession to Lihua Li; Suzhou Scientific Research Project (SYW2024021), and Natural Science Foundation of Boxi Incubation Program (BXQN202140) to Caihua Xu; The foundations have no roles on experiment design, data analysis, or article writing.

Data availability

The authors declare that all the data supporting the findings of this study are available within the article and its supplementary information files or from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

All the protocols of animal experiments were approved by the Institutional Animal Care and Use Committee of Jiangnan University (No: 20250059). Approval project: The in vivo function of lipid metabolism-related protein in progression of lung cancer, April 26, 2025. The experiments using specimens and clinical data of lung cancer patients were approved by the ethics committee of The First Affiliated Hospital of Soochow University (No: 2024700). Approval project: The potential function of lipid metabolism-related protein in clinical specimens of lung cancer, Apr. 01, 2024. All the patients enrolled in this study were sighed informed consents.

Competing interests

The authors declare no competing interests.